Cleveland Clinic Journal of Medicine

Screening can help prevent colorectal cancer. The United States has seen a steady decline in colorectal cancer incidence and mortality, thanks in large part to screening. Screening rates can be increased with good patient-physician dialogue and by choosing a method the patient prefers and is most likely to complete.

In this article, we review a general approach to screening, focusing on the most commonly used methods in the United States, ie, the guaiac-based fecal occult blood test (FOBT), the fecal immunochemical test (FIT), and colonoscopy. We discuss current colorectal cancer incidence rates, screening recommendations, and how to choose the appropriate screening test.

This article does not discuss patients at high risk of polyps or cancer due to hereditary colon cancer syndromes, a personal history of colorectal neoplasia, inflammatory bowel disease, or primary sclerosing cholangitis.

TRENDS IN INCIDENCE

Colorectal cancer is the second most common type of cancer and cause of cancer-related deaths in the United States, responsible for an estimated 50,000 deaths in 2017. The lifetime risk of its occurrence is estimated to be 1 in 21 men and 1 in 23 women.¹ Encouragingly, the incidence has declined by 24% over the last 30 years,² and by 3% per year from 2004 to 2013.¹ Also, as a result of screening and advances in treatment, 5-year survival rates for patients with colorectal cancer have increased, from 48.6% in 1975 to 66.4% in 2009.²

When detected at a localized stage, the 5-year survival rate in colorectal cancer is greater than 90%. Unfortunately, it is diagnosed early in only 39% of patients. And despite advances in treatment and a doubling of the 5-year survival rate in patients with advanced cancers since 1990,³ the latter is only 14%. In most patients, cancer is diagnosed when it has spread to the lymph nodes (36%) or to distant organs (22%), and the survival rate declines to 71% after lymph-node spread, and 14% after metastasis to distant organs.

It is essential to screen people who have no symptoms, as symptoms such as gastrointestinal bleeding, unexplained abdominal pain or weight loss, a persistent change in bowel movements, and bowel obstruction typically do not arise until the disease is advanced and less amenable to cure.

Increasing prevalence in younger adults

Curiously, the incidence of colorectal cancer is increasing in white US adults under age 50. Over the last 30 years, incidence rates have increased from 1.0% to 2.4% annually in adults ages 20 to 39.⁴ Based on current trends, colon cancer rates are expected to increase by 90% for patients ages 20 to 34 and by 28% for patients 35 to 49 by 2030.⁵

Although recommendations vary for colorectal cancer screening in patients under age 50, clinicians should investigate symptoms such as rectal bleeding, unexplained iron deficiency anemia, progressive abdominal pain, and persistent changes in bowel movements.

Other challenges

Despite the benefits of screening, it is underutilized. Although rates of compliance with screening recommendations have increased 10% over the last 10 years, only 65% of eligible adults currently comply.^1,6

Additionally, certain areas of the country such as Appalachia and the Mississippi Delta have not benefited from the decline in the national rate of colorectal cancer.⁷

SCREENING GUIDELINES

Most guidelines say that colorectal cancer screening should begin at age 50 in people at average risk with no symptoms. However, the American College of Gastroenterology (ACG) recommends beginning screening at age 45 in African Americans, as this group has higher incidence and mortality rates of colorectal cancer.⁸ Also, the American Cancer Society recently recommended beginning screening at age 45 for all individuals.⁹

Screening can stop at age 75 for most patients, according to the ACG,⁸ the US Multi-Society Task Force on Colorectal Cancer,¹⁰ and the US Preventive Services Task Force  (USPSTF).¹¹ However, the decision should be individualized for patients ages 76 to 85. Patients within that age group who are in good health and have not previously been screened are more likely to benefit than those who have previously been screened and had a negative screening test. Patients over age 85 should not begin or continue screening, because of diminished benefit of screening in this age group, shorter life expectancy, advanced comorbid conditions, and the risks of colonoscopy and cancer treatment.

Patients and clinicians are encouraged to collaborate in deciding which screening method is appropriate. Patients adhere better when they are given a choice in the matter.^12–14 And adherence is the key to effective colorectal cancer screening.

Familiarity with the key characteristics of currently available colorectal cancer screening tests will facilitate discussion with patients.

Opportunistic vs programmatic screening

Screening can be classified according to the approach to the patient or population and the intent of the test. Most screening in the United States is opportunistic rather than programmatic—that is, the physician offers the patient screening at the point of service without systematic follow-up or patient re-engagement.

In a programmatic approach, the patient is offered screening through an organized program that streamlines services, reduces overscreening, and provides systematic follow-up of testing.

Stool studies such as FOBT and FIT do not reliably detect cancer precursors such as adenomas and serrated neoplasms. If an FOBT is positive, follow-up diagnostic colonoscopy is required. Unlike screening colonoscopy, diagnostic colonoscopy requires a copayment for Medicare patients, and this should be explained to the patient.

FIT and FOBT detect hemolyzed blood within a stool sample, FOBT by a chemical reaction, and FIT by detecting a globin-specific antibody. Colorectal cancer and some large adenomatous polyps may intermittently bleed and result in occult blood in the stool, iron deficiency anemia, or hematochezia.¹⁵

Fecal occult blood testing

Historically, FOBT was the stool test of choice for screening. It uses an indirect enzymatic reaction to detect hemolyzed blood in the stool. When a specimen containing hemoglobin is added to guaiac paper and a drop of hydrogen peroxide is added to “develop” it, the peroxidase activity of hemoglobin turns the guaiac blue.

Screening with FOBT involves annual testing of 3 consecutively passed stools from different days; FOBT should not be performed at the time of digital rectal examination or if the patient is having overt rectal, urinary, or menstrual bleeding.

Dietary and medication restrictions before and during the testing period are critical, as red meat contains hemoglobin, and certain vegetables (eg, radishes, turnips, cauliflower, cucumbers) contain peroxidase, all of which can cause a false-positive result. Waiting 3 days after the stool sample is collected to develop it can mitigate the peroxidase activity of vegetables.¹⁶ Vitamin C inhibits heme peroxidase activity and leads to false-negative results. Aspirin and high-dose nonsteroidal anti-inflammatory drugs can promote bleeding throughout the intestinal tract.¹⁷

In randomized controlled trials,^18–21 screening with FOBT reduced colorectal cancer mortality rates by 15% to 33%. The 30-year follow-up of a large US trial²² found a 32% relative reduction in mortality rates in patients randomized to annual screening, and a 22% relative reduction in those randomized to screening every 2 years. Despite the many possibilities for false-positive results, the specificity for detecting cancer has ranged from 86.7% to 97.3%, and the sensitivity from 37.1% to 79.4%, highlighting the benefit of colorectal cancer screening programs in unscreened populations.^23–26

FIT vs FOBT in current practice

FIT should replace FOBT as the preferred stool screening method. Instead of an enzymatic reaction that can be altered by food or medication, FIT utilizes an antibody specific to human globin to directly detect hemolyzed blood, thus eliminating the need to modify the diet or medications.²⁷ Additionally, only 1 stool specimen is needed, which may explain why the adherence rate was about 20% higher with FIT than with FOBT in most studies.^28–30

FIT has a sensitivity of 69% to 86% for colorectal cancer and a specificity of 92% to 95%.³¹ The sensitivity can be improved by lowering the threshold value for a positive test, but this is associated with a decrease in specificity. A single FIT has the same sensitivity and specificity as several samples.³²

In a large retrospective US cohort study of programmatic screening with FIT, Jensen et al³³ reported that 48% of 670,841 people who were offered testing actually did the test. Of the 48% who participated in the first round and remained eligible, 75% to 86% participated in subsequent rounds over 4 years. Those who had a positive result on FIT were supposed to undergo colonoscopy, but 22% did not.

The US Multi-Society Task Force on Colorectal Cancer³⁴ suggests that FIT-based screening programs aim for a target FIT completion rate of more than 60% and a target colonoscopy completion rate of more than 80% of patients with positive FITs. These benchmarks were derived from adherence rates in international FIT screening studies in average-risk populations.^35–39 (Note that the large US cohort described above³³ did not meet these goals.) Ideally, every patient with a positive FIT should undergo diagnostic colonoscopy, but in reality only 50% to 83% actually do. Methods shown to improve adherence include structured screening programs with routine performance reports, provider feedback, and involvement of patient navigators.^40–42

Accordingly, several aspects of stool-based testing need to be stressed with patients. Understanding that FOBT is recommended yearly is integral for optimal impact on colorectal cancer incidence and mortality rates.

Additionally, patients should be advised to undergo colonoscopy soon after a positive FIT, because delaying colonoscopy could give precancerous lesions time to progress in stage. The acceptable time between a positive FIT and colonoscopy has yet to be determined. However, a retrospective cohort study of 1.26 million screened patients with 107,000 positive FIT results demonstrated that the rates of cancer discovered on colonoscopy were similar when performed within 30 days or up to 10 months after a positive test. Detection rates increased from 3% to 4.8% at 10 months and to 7.9% at 12 months.⁴³

In modeling studies, Meester et al⁴⁴ showed the estimated lifetime risk and mortality rates from colorectal cancer and life-years gained from screening are significantly better when colonoscopy is completed within 2 weeks rather than 1 year after a positive FIT. Each additional month after 2 weeks incrementally affected these outcomes, with a 1.4% increase in cancer mortality. These data suggest that colonoscopy should be done soon after a positive FIT result and at a maximum of 10 months.^43,44

Screening with FOBT is a multistep process for patients that includes receiving the test kit, collecting the sample, preparing it, returning it, undergoing colonoscopy after a positive test, and repeating in 1 year if negative. The screening program should identify patients at average risk in whom screening is appropriate, ensure delivery of the test, verify the quality of collected samples for laboratory testing against the manufacturer’s recommendations, and report results. Report of a positive FOBT result should provide recommendations for follow-up.

Though evidence clearly supports screening annually or biennially (every 2 years) with FOBT, the ideal interval for FIT is undetermined. Modeling studies utilized by the USPSTF and Multi-Society Task Force demonstrate that colonoscopy and annual FIT result in similar life-years gained, while 2 population-based screening programs have demonstrated that a 2- or 3-year interval may be equally efficacious by lowering the threshold for a positive test.^38,45

Randomized controlled trials of screening colonoscopy vs annual and biennial FIT are currently under way. Cost-effectiveness analysis has shown that offering single-sample FITs at more frequent (annual) intervals performs better than multisample testing at less frequent intervals.^45–47

Compared with stool-based screening, colonoscopy has advantages, including a 10-year screening interval if bowel preparation is adequate and the examination shows no neoplasia, the ability to inspect the entire colon, and the ability to diagnose and treat lesions in the same session.

Screening colonoscopy visualizes the entire colon in more than 98% of cases, although it requires adequate bowel preparation for maximal polyp detection. It can be done safely with or without sedation.⁴⁸

While there are no available randomized controlled trial data on the impact of screening colonoscopy on cancer incidence or mortality, extensive case-control and cohort studies consistently show that screening colonoscopy reduces cancer incidence and mortality rates.^49–54 A US Veterans Administration study of more than 32,000 patients reported a 50% reduction in overall colorectal cancer mortality.⁵⁵ In a microsimulation modeling study that assumed 100% adherence, colonoscopy every 10 years and annual FIT in individuals ages 50 to 75 provided similar life-years gained per 1,000 people screened (270 for colonoscopy, 244 for FIT).⁵⁶

Well-established metrics for maximizing the effectiveness and quality of colonoscopy have been established (Table 2). The most important include the mucosa inspection time (withdrawal time) and adenoma detection rate.⁵⁷ Withdrawal time is directly correlated with adenoma detection, and a 6-minute minimum withdrawal time is recommended in screening colonoscopy examinations of patients at average risk when no polyps are found.⁵⁸ The adenoma detection rate is the strongest evidence-based metric, as each 1% increase in the adenoma detection rate over 19% is associated with a 3% decrease in the risk of colorectal cancer and a 5% decrease in death rate.⁵⁹ The average-risk screening adenoma detection rate differs based on sex: the rate is greater than 20% for women and greater than 30% for men.

Complications from screening, diagnostic, or therapeutic colonoscopy are infrequent but include perforation (4/10,000) and significant intestinal bleeding (8/10,000).^56–62

Patients with a first-degree relative under age 60 with advanced adenomas or colorectal cancer are considered at high risk and should begin screening colonoscopy at age 40, with repeat colonoscopy at 5-year intervals, given a trend toward advanced neoplasia detection compared with FIT.⁶³

Guidelines recently published by the Canadian Association of Gastroenterology and endorsed by the American Gastroenterological Association also support starting screening in high-risk individuals at age 40, with a surveillance interval of 5 to 10 years based on the number of first-degree relatives with colorectal cancer or adenomas.⁶⁴ Consensus statements were based on retrospective cohort, prospective case-controlled, and cross-sectional studies comparing the risk of colorectal cancer in individuals with a family history against those without a family history.

Randomized clinical trials comparing colonoscopy and FIT are under way. Interim analysis of a European trial in which asymptomatic adults ages 50 to 69 were randomized to 1-time colonoscopy (26,703 patients) vs FIT every 2 years (26,599 patients) found significantly higher participation rates in the FIT arm (34.2% vs 24.6%) but higher rates of nonadvanced adenomas (4.2% vs 0.4%) and advanced neoplasia (1.9% vs 0.9%) in the colonoscopy arm.⁶⁵ Cancer was detected in 0.1% in each arm. These findings correlate with those of another study showing higher participation with FIT but higher advanced neoplasia detection rates with colonoscopy.⁶⁶

Detection of precursor lesions is vital, as removing neoplasms is the main strategy to reduce colorectal cancer incidence. Accordingly, the advantage of colonoscopy was illustrated by a study that determined that 53 patients would need to undergo screening colonoscopy to detect 1 advanced adenoma or cancerous lesion, compared with 264 for FIT.⁶⁷

STARTING SCREEING AT AGE 45

The American Cancer Society recently provided a qualified recommendation to start colorectal cancer screening in all individuals at age 45 rather than 50.⁹ This recommendation was based on modeling studies demonstrating that starting screening at age 45 with colonoscopy every 10 years resulted in 25 life-years gained at the cost of 610 colonoscopies per 1,000 individuals. Alternative strategies included FIT, which resulted in an additional 26 life-years gained per 1,000 individuals screened, flexible sigmoidoscopy (23 life-years gained), and computed tomographic colonoscopy (22 life-years gained).

Rates of colorectal cancer are rising in adults under age 50, and 10,000 new cases are anticipated this year.^2,3 Currently, 22 million US adults are between the ages of 45 and 50. The system and support needed to perform screening in all adults over age 45 and a lack of direct evidence to support its benefits in the young population need to be considered before widespread acceptance of the American Cancer Society recommendations. However, if screening is considered, FIT with or without sigmoidoscopy may be appropriate, given that most cancers diagnosed in individuals under age 50 are left-sided.^4,5

Screening has not been proven to reduce all-cause mortality. Randomized controlled trials of FOBT and observational studies of colonoscopy show that screening reduces cancer incidence and mortality. Until the currently ongoing randomized controlled trials comparing colonoscopy with FIT are completed, their comparative impact on colorectal cancer end points is unknown.

PATIENT ADHERENCE IS KEY

FIT and colonoscopy are the most prevalent screening methods in the United States. Careful attention should be given to offer the screening option the patient is most likely to complete, as adherence is key to the benefit from colorectal cancer screening.

The National Colorectal Cancer Roundtable (nccrt.org), established in 1997 by the American Cancer Society and the US Centers for Disease Control and Prevention, is a national coalition of public and private organizations dedicated to reducing colorectal cancer incidence and mortality. The Roundtable waged a national campaign to achieve a colorectal cancer screening rate of 80% in eligible adults by 2018, a goal that was not met. Still, the potential for a substantial impact is a compelling reason to endorse adherence to colorectal cancer screening. The Roundtable provides many resources for physicians to enhance screening in their practice.

The United States has seen a steady decline in colorectal cancer incidence and mortality, mainly as a result of screening. Colorectal cancer is preventable with ensuring patients’ adherence to screening. Screening rates have been shown to increase with patient-provider dialogue and with selection of a screening program the patient prefers and is most likely to complete.

Screening can help prevent colorectal cancer. The United States has seen a steady decline in colorectal cancer incidence and mortality, thanks in large part to screening. Screening rates can be increased with good patient-physician dialogue and by choosing a method the patient prefers and is most likely to complete.

In this article, we review a general approach to screening, focusing on the most commonly used methods in the United States, ie, the guaiac-based fecal occult blood test (FOBT), the fecal immunochemical test (FIT), and colonoscopy. We discuss current colorectal cancer incidence rates, screening recommendations, and how to choose the appropriate screening test.

This article does not discuss patients at high risk of polyps or cancer due to hereditary colon cancer syndromes, a personal history of colorectal neoplasia, inflammatory bowel disease, or primary sclerosing cholangitis.

TRENDS IN INCIDENCE

Colorectal cancer is the second most common type of cancer and cause of cancer-related deaths in the United States, responsible for an estimated 50,000 deaths in 2017. The lifetime risk of its occurrence is estimated to be 1 in 21 men and 1 in 23 women.¹ Encouragingly, the incidence has declined by 24% over the last 30 years,² and by 3% per year from 2004 to 2013.¹ Also, as a result of screening and advances in treatment, 5-year survival rates for patients with colorectal cancer have increased, from 48.6% in 1975 to 66.4% in 2009.²

When detected at a localized stage, the 5-year survival rate in colorectal cancer is greater than 90%. Unfortunately, it is diagnosed early in only 39% of patients. And despite advances in treatment and a doubling of the 5-year survival rate in patients with advanced cancers since 1990,³ the latter is only 14%. In most patients, cancer is diagnosed when it has spread to the lymph nodes (36%) or to distant organs (22%), and the survival rate declines to 71% after lymph-node spread, and 14% after metastasis to distant organs.

It is essential to screen people who have no symptoms, as symptoms such as gastrointestinal bleeding, unexplained abdominal pain or weight loss, a persistent change in bowel movements, and bowel obstruction typically do not arise until the disease is advanced and less amenable to cure.

Increasing prevalence in younger adults

Curiously, the incidence of colorectal cancer is increasing in white US adults under age 50. Over the last 30 years, incidence rates have increased from 1.0% to 2.4% annually in adults ages 20 to 39.⁴ Based on current trends, colon cancer rates are expected to increase by 90% for patients ages 20 to 34 and by 28% for patients 35 to 49 by 2030.⁵

Although recommendations vary for colorectal cancer screening in patients under age 50, clinicians should investigate symptoms such as rectal bleeding, unexplained iron deficiency anemia, progressive abdominal pain, and persistent changes in bowel movements.

Other challenges

Despite the benefits of screening, it is underutilized. Although rates of compliance with screening recommendations have increased 10% over the last 10 years, only 65% of eligible adults currently comply.^1,6

Additionally, certain areas of the country such as Appalachia and the Mississippi Delta have not benefited from the decline in the national rate of colorectal cancer.⁷

SCREENING GUIDELINES

Most guidelines say that colorectal cancer screening should begin at age 50 in people at average risk with no symptoms. However, the American College of Gastroenterology (ACG) recommends beginning screening at age 45 in African Americans, as this group has higher incidence and mortality rates of colorectal cancer.⁸ Also, the American Cancer Society recently recommended beginning screening at age 45 for all individuals.⁹

Screening can stop at age 75 for most patients, according to the ACG,⁸ the US Multi-Society Task Force on Colorectal Cancer,¹⁰ and the US Preventive Services Task Force  (USPSTF).¹¹ However, the decision should be individualized for patients ages 76 to 85. Patients within that age group who are in good health and have not previously been screened are more likely to benefit than those who have previously been screened and had a negative screening test. Patients over age 85 should not begin or continue screening, because of diminished benefit of screening in this age group, shorter life expectancy, advanced comorbid conditions, and the risks of colonoscopy and cancer treatment.

Patients and clinicians are encouraged to collaborate in deciding which screening method is appropriate. Patients adhere better when they are given a choice in the matter.^12–14 And adherence is the key to effective colorectal cancer screening.

Familiarity with the key characteristics of currently available colorectal cancer screening tests will facilitate discussion with patients.

Opportunistic vs programmatic screening

Screening can be classified according to the approach to the patient or population and the intent of the test. Most screening in the United States is opportunistic rather than programmatic—that is, the physician offers the patient screening at the point of service without systematic follow-up or patient re-engagement.

In a programmatic approach, the patient is offered screening through an organized program that streamlines services, reduces overscreening, and provides systematic follow-up of testing.

Stool studies such as FOBT and FIT do not reliably detect cancer precursors such as adenomas and serrated neoplasms. If an FOBT is positive, follow-up diagnostic colonoscopy is required. Unlike screening colonoscopy, diagnostic colonoscopy requires a copayment for Medicare patients, and this should be explained to the patient.

FIT and FOBT detect hemolyzed blood within a stool sample, FOBT by a chemical reaction, and FIT by detecting a globin-specific antibody. Colorectal cancer and some large adenomatous polyps may intermittently bleed and result in occult blood in the stool, iron deficiency anemia, or hematochezia.¹⁵

Fecal occult blood testing

Historically, FOBT was the stool test of choice for screening. It uses an indirect enzymatic reaction to detect hemolyzed blood in the stool. When a specimen containing hemoglobin is added to guaiac paper and a drop of hydrogen peroxide is added to “develop” it, the peroxidase activity of hemoglobin turns the guaiac blue.

Screening with FOBT involves annual testing of 3 consecutively passed stools from different days; FOBT should not be performed at the time of digital rectal examination or if the patient is having overt rectal, urinary, or menstrual bleeding.

Dietary and medication restrictions before and during the testing period are critical, as red meat contains hemoglobin, and certain vegetables (eg, radishes, turnips, cauliflower, cucumbers) contain peroxidase, all of which can cause a false-positive result. Waiting 3 days after the stool sample is collected to develop it can mitigate the peroxidase activity of vegetables.¹⁶ Vitamin C inhibits heme peroxidase activity and leads to false-negative results. Aspirin and high-dose nonsteroidal anti-inflammatory drugs can promote bleeding throughout the intestinal tract.¹⁷

In randomized controlled trials,^18–21 screening with FOBT reduced colorectal cancer mortality rates by 15% to 33%. The 30-year follow-up of a large US trial²² found a 32% relative reduction in mortality rates in patients randomized to annual screening, and a 22% relative reduction in those randomized to screening every 2 years. Despite the many possibilities for false-positive results, the specificity for detecting cancer has ranged from 86.7% to 97.3%, and the sensitivity from 37.1% to 79.4%, highlighting the benefit of colorectal cancer screening programs in unscreened populations.^23–26

FIT vs FOBT in current practice

FIT should replace FOBT as the preferred stool screening method. Instead of an enzymatic reaction that can be altered by food or medication, FIT utilizes an antibody specific to human globin to directly detect hemolyzed blood, thus eliminating the need to modify the diet or medications.²⁷ Additionally, only 1 stool specimen is needed, which may explain why the adherence rate was about 20% higher with FIT than with FOBT in most studies.^28–30

FIT has a sensitivity of 69% to 86% for colorectal cancer and a specificity of 92% to 95%.³¹ The sensitivity can be improved by lowering the threshold value for a positive test, but this is associated with a decrease in specificity. A single FIT has the same sensitivity and specificity as several samples.³²

In a large retrospective US cohort study of programmatic screening with FIT, Jensen et al³³ reported that 48% of 670,841 people who were offered testing actually did the test. Of the 48% who participated in the first round and remained eligible, 75% to 86% participated in subsequent rounds over 4 years. Those who had a positive result on FIT were supposed to undergo colonoscopy, but 22% did not.

The US Multi-Society Task Force on Colorectal Cancer³⁴ suggests that FIT-based screening programs aim for a target FIT completion rate of more than 60% and a target colonoscopy completion rate of more than 80% of patients with positive FITs. These benchmarks were derived from adherence rates in international FIT screening studies in average-risk populations.^35–39 (Note that the large US cohort described above³³ did not meet these goals.) Ideally, every patient with a positive FIT should undergo diagnostic colonoscopy, but in reality only 50% to 83% actually do. Methods shown to improve adherence include structured screening programs with routine performance reports, provider feedback, and involvement of patient navigators.^40–42

Accordingly, several aspects of stool-based testing need to be stressed with patients. Understanding that FOBT is recommended yearly is integral for optimal impact on colorectal cancer incidence and mortality rates.

Additionally, patients should be advised to undergo colonoscopy soon after a positive FIT, because delaying colonoscopy could give precancerous lesions time to progress in stage. The acceptable time between a positive FIT and colonoscopy has yet to be determined. However, a retrospective cohort study of 1.26 million screened patients with 107,000 positive FIT results demonstrated that the rates of cancer discovered on colonoscopy were similar when performed within 30 days or up to 10 months after a positive test. Detection rates increased from 3% to 4.8% at 10 months and to 7.9% at 12 months.⁴³

In modeling studies, Meester et al⁴⁴ showed the estimated lifetime risk and mortality rates from colorectal cancer and life-years gained from screening are significantly better when colonoscopy is completed within 2 weeks rather than 1 year after a positive FIT. Each additional month after 2 weeks incrementally affected these outcomes, with a 1.4% increase in cancer mortality. These data suggest that colonoscopy should be done soon after a positive FIT result and at a maximum of 10 months.^43,44

Screening with FOBT is a multistep process for patients that includes receiving the test kit, collecting the sample, preparing it, returning it, undergoing colonoscopy after a positive test, and repeating in 1 year if negative. The screening program should identify patients at average risk in whom screening is appropriate, ensure delivery of the test, verify the quality of collected samples for laboratory testing against the manufacturer’s recommendations, and report results. Report of a positive FOBT result should provide recommendations for follow-up.

Though evidence clearly supports screening annually or biennially (every 2 years) with FOBT, the ideal interval for FIT is undetermined. Modeling studies utilized by the USPSTF and Multi-Society Task Force demonstrate that colonoscopy and annual FIT result in similar life-years gained, while 2 population-based screening programs have demonstrated that a 2- or 3-year interval may be equally efficacious by lowering the threshold for a positive test.^38,45

Randomized controlled trials of screening colonoscopy vs annual and biennial FIT are currently under way. Cost-effectiveness analysis has shown that offering single-sample FITs at more frequent (annual) intervals performs better than multisample testing at less frequent intervals.^45–47

Compared with stool-based screening, colonoscopy has advantages, including a 10-year screening interval if bowel preparation is adequate and the examination shows no neoplasia, the ability to inspect the entire colon, and the ability to diagnose and treat lesions in the same session.

Screening colonoscopy visualizes the entire colon in more than 98% of cases, although it requires adequate bowel preparation for maximal polyp detection. It can be done safely with or without sedation.⁴⁸

While there are no available randomized controlled trial data on the impact of screening colonoscopy on cancer incidence or mortality, extensive case-control and cohort studies consistently show that screening colonoscopy reduces cancer incidence and mortality rates.^49–54 A US Veterans Administration study of more than 32,000 patients reported a 50% reduction in overall colorectal cancer mortality.⁵⁵ In a microsimulation modeling study that assumed 100% adherence, colonoscopy every 10 years and annual FIT in individuals ages 50 to 75 provided similar life-years gained per 1,000 people screened (270 for colonoscopy, 244 for FIT).⁵⁶

Well-established metrics for maximizing the effectiveness and quality of colonoscopy have been established (Table 2). The most important include the mucosa inspection time (withdrawal time) and adenoma detection rate.⁵⁷ Withdrawal time is directly correlated with adenoma detection, and a 6-minute minimum withdrawal time is recommended in screening colonoscopy examinations of patients at average risk when no polyps are found.⁵⁸ The adenoma detection rate is the strongest evidence-based metric, as each 1% increase in the adenoma detection rate over 19% is associated with a 3% decrease in the risk of colorectal cancer and a 5% decrease in death rate.⁵⁹ The average-risk screening adenoma detection rate differs based on sex: the rate is greater than 20% for women and greater than 30% for men.

Complications from screening, diagnostic, or therapeutic colonoscopy are infrequent but include perforation (4/10,000) and significant intestinal bleeding (8/10,000).^56–62

Patients with a first-degree relative under age 60 with advanced adenomas or colorectal cancer are considered at high risk and should begin screening colonoscopy at age 40, with repeat colonoscopy at 5-year intervals, given a trend toward advanced neoplasia detection compared with FIT.⁶³

Guidelines recently published by the Canadian Association of Gastroenterology and endorsed by the American Gastroenterological Association also support starting screening in high-risk individuals at age 40, with a surveillance interval of 5 to 10 years based on the number of first-degree relatives with colorectal cancer or adenomas.⁶⁴ Consensus statements were based on retrospective cohort, prospective case-controlled, and cross-sectional studies comparing the risk of colorectal cancer in individuals with a family history against those without a family history.

Randomized clinical trials comparing colonoscopy and FIT are under way. Interim analysis of a European trial in which asymptomatic adults ages 50 to 69 were randomized to 1-time colonoscopy (26,703 patients) vs FIT every 2 years (26,599 patients) found significantly higher participation rates in the FIT arm (34.2% vs 24.6%) but higher rates of nonadvanced adenomas (4.2% vs 0.4%) and advanced neoplasia (1.9% vs 0.9%) in the colonoscopy arm.⁶⁵ Cancer was detected in 0.1% in each arm. These findings correlate with those of another study showing higher participation with FIT but higher advanced neoplasia detection rates with colonoscopy.⁶⁶

Detection of precursor lesions is vital, as removing neoplasms is the main strategy to reduce colorectal cancer incidence. Accordingly, the advantage of colonoscopy was illustrated by a study that determined that 53 patients would need to undergo screening colonoscopy to detect 1 advanced adenoma or cancerous lesion, compared with 264 for FIT.⁶⁷

STARTING SCREEING AT AGE 45

The American Cancer Society recently provided a qualified recommendation to start colorectal cancer screening in all individuals at age 45 rather than 50.⁹ This recommendation was based on modeling studies demonstrating that starting screening at age 45 with colonoscopy every 10 years resulted in 25 life-years gained at the cost of 610 colonoscopies per 1,000 individuals. Alternative strategies included FIT, which resulted in an additional 26 life-years gained per 1,000 individuals screened, flexible sigmoidoscopy (23 life-years gained), and computed tomographic colonoscopy (22 life-years gained).

Rates of colorectal cancer are rising in adults under age 50, and 10,000 new cases are anticipated this year.^2,3 Currently, 22 million US adults are between the ages of 45 and 50. The system and support needed to perform screening in all adults over age 45 and a lack of direct evidence to support its benefits in the young population need to be considered before widespread acceptance of the American Cancer Society recommendations. However, if screening is considered, FIT with or without sigmoidoscopy may be appropriate, given that most cancers diagnosed in individuals under age 50 are left-sided.^4,5

Screening has not been proven to reduce all-cause mortality. Randomized controlled trials of FOBT and observational studies of colonoscopy show that screening reduces cancer incidence and mortality. Until the currently ongoing randomized controlled trials comparing colonoscopy with FIT are completed, their comparative impact on colorectal cancer end points is unknown.

PATIENT ADHERENCE IS KEY

FIT and colonoscopy are the most prevalent screening methods in the United States. Careful attention should be given to offer the screening option the patient is most likely to complete, as adherence is key to the benefit from colorectal cancer screening.

The National Colorectal Cancer Roundtable (nccrt.org), established in 1997 by the American Cancer Society and the US Centers for Disease Control and Prevention, is a national coalition of public and private organizations dedicated to reducing colorectal cancer incidence and mortality. The Roundtable waged a national campaign to achieve a colorectal cancer screening rate of 80% in eligible adults by 2018, a goal that was not met. Still, the potential for a substantial impact is a compelling reason to endorse adherence to colorectal cancer screening. The Roundtable provides many resources for physicians to enhance screening in their practice.

The United States has seen a steady decline in colorectal cancer incidence and mortality, mainly as a result of screening. Colorectal cancer is preventable with ensuring patients’ adherence to screening. Screening rates have been shown to increase with patient-provider dialogue and with selection of a screening program the patient prefers and is most likely to complete.

Screening can help prevent colorectal cancer. The United States has seen a steady decline in colorectal cancer incidence and mortality, thanks in large part to screening. Screening rates can be increased with good patient-physician dialogue and by choosing a method the patient prefers and is most likely to complete.

In this article, we review a general approach to screening, focusing on the most commonly used methods in the United States, ie, the guaiac-based fecal occult blood test (FOBT), the fecal immunochemical test (FIT), and colonoscopy. We discuss current colorectal cancer incidence rates, screening recommendations, and how to choose the appropriate screening test.

This article does not discuss patients at high risk of polyps or cancer due to hereditary colon cancer syndromes, a personal history of colorectal neoplasia, inflammatory bowel disease, or primary sclerosing cholangitis.

TRENDS IN INCIDENCE

Colorectal cancer is the second most common type of cancer and cause of cancer-related deaths in the United States, responsible for an estimated 50,000 deaths in 2017. The lifetime risk of its occurrence is estimated to be 1 in 21 men and 1 in 23 women.¹ Encouragingly, the incidence has declined by 24% over the last 30 years,² and by 3% per year from 2004 to 2013.¹ Also, as a result of screening and advances in treatment, 5-year survival rates for patients with colorectal cancer have increased, from 48.6% in 1975 to 66.4% in 2009.²

When detected at a localized stage, the 5-year survival rate in colorectal cancer is greater than 90%. Unfortunately, it is diagnosed early in only 39% of patients. And despite advances in treatment and a doubling of the 5-year survival rate in patients with advanced cancers since 1990,³ the latter is only 14%. In most patients, cancer is diagnosed when it has spread to the lymph nodes (36%) or to distant organs (22%), and the survival rate declines to 71% after lymph-node spread, and 14% after metastasis to distant organs.

It is essential to screen people who have no symptoms, as symptoms such as gastrointestinal bleeding, unexplained abdominal pain or weight loss, a persistent change in bowel movements, and bowel obstruction typically do not arise until the disease is advanced and less amenable to cure.

Increasing prevalence in younger adults

Curiously, the incidence of colorectal cancer is increasing in white US adults under age 50. Over the last 30 years, incidence rates have increased from 1.0% to 2.4% annually in adults ages 20 to 39.⁴ Based on current trends, colon cancer rates are expected to increase by 90% for patients ages 20 to 34 and by 28% for patients 35 to 49 by 2030.⁵

Although recommendations vary for colorectal cancer screening in patients under age 50, clinicians should investigate symptoms such as rectal bleeding, unexplained iron deficiency anemia, progressive abdominal pain, and persistent changes in bowel movements.

Other challenges

Despite the benefits of screening, it is underutilized. Although rates of compliance with screening recommendations have increased 10% over the last 10 years, only 65% of eligible adults currently comply.^1,6

Additionally, certain areas of the country such as Appalachia and the Mississippi Delta have not benefited from the decline in the national rate of colorectal cancer.⁷

SCREENING GUIDELINES

Most guidelines say that colorectal cancer screening should begin at age 50 in people at average risk with no symptoms. However, the American College of Gastroenterology (ACG) recommends beginning screening at age 45 in African Americans, as this group has higher incidence and mortality rates of colorectal cancer.⁸ Also, the American Cancer Society recently recommended beginning screening at age 45 for all individuals.⁹

Screening can stop at age 75 for most patients, according to the ACG,⁸ the US Multi-Society Task Force on Colorectal Cancer,¹⁰ and the US Preventive Services Task Force  (USPSTF).¹¹ However, the decision should be individualized for patients ages 76 to 85. Patients within that age group who are in good health and have not previously been screened are more likely to benefit than those who have previously been screened and had a negative screening test. Patients over age 85 should not begin or continue screening, because of diminished benefit of screening in this age group, shorter life expectancy, advanced comorbid conditions, and the risks of colonoscopy and cancer treatment.

Patients and clinicians are encouraged to collaborate in deciding which screening method is appropriate. Patients adhere better when they are given a choice in the matter.^12–14 And adherence is the key to effective colorectal cancer screening.

Familiarity with the key characteristics of currently available colorectal cancer screening tests will facilitate discussion with patients.

Opportunistic vs programmatic screening

Screening can be classified according to the approach to the patient or population and the intent of the test. Most screening in the United States is opportunistic rather than programmatic—that is, the physician offers the patient screening at the point of service without systematic follow-up or patient re-engagement.

In a programmatic approach, the patient is offered screening through an organized program that streamlines services, reduces overscreening, and provides systematic follow-up of testing.

Stool studies such as FOBT and FIT do not reliably detect cancer precursors such as adenomas and serrated neoplasms. If an FOBT is positive, follow-up diagnostic colonoscopy is required. Unlike screening colonoscopy, diagnostic colonoscopy requires a copayment for Medicare patients, and this should be explained to the patient.

FIT and FOBT detect hemolyzed blood within a stool sample, FOBT by a chemical reaction, and FIT by detecting a globin-specific antibody. Colorectal cancer and some large adenomatous polyps may intermittently bleed and result in occult blood in the stool, iron deficiency anemia, or hematochezia.¹⁵

Fecal occult blood testing

Historically, FOBT was the stool test of choice for screening. It uses an indirect enzymatic reaction to detect hemolyzed blood in the stool. When a specimen containing hemoglobin is added to guaiac paper and a drop of hydrogen peroxide is added to “develop” it, the peroxidase activity of hemoglobin turns the guaiac blue.

Screening with FOBT involves annual testing of 3 consecutively passed stools from different days; FOBT should not be performed at the time of digital rectal examination or if the patient is having overt rectal, urinary, or menstrual bleeding.

Dietary and medication restrictions before and during the testing period are critical, as red meat contains hemoglobin, and certain vegetables (eg, radishes, turnips, cauliflower, cucumbers) contain peroxidase, all of which can cause a false-positive result. Waiting 3 days after the stool sample is collected to develop it can mitigate the peroxidase activity of vegetables.¹⁶ Vitamin C inhibits heme peroxidase activity and leads to false-negative results. Aspirin and high-dose nonsteroidal anti-inflammatory drugs can promote bleeding throughout the intestinal tract.¹⁷

In randomized controlled trials,^18–21 screening with FOBT reduced colorectal cancer mortality rates by 15% to 33%. The 30-year follow-up of a large US trial²² found a 32% relative reduction in mortality rates in patients randomized to annual screening, and a 22% relative reduction in those randomized to screening every 2 years. Despite the many possibilities for false-positive results, the specificity for detecting cancer has ranged from 86.7% to 97.3%, and the sensitivity from 37.1% to 79.4%, highlighting the benefit of colorectal cancer screening programs in unscreened populations.^23–26

FIT vs FOBT in current practice

FIT should replace FOBT as the preferred stool screening method. Instead of an enzymatic reaction that can be altered by food or medication, FIT utilizes an antibody specific to human globin to directly detect hemolyzed blood, thus eliminating the need to modify the diet or medications.²⁷ Additionally, only 1 stool specimen is needed, which may explain why the adherence rate was about 20% higher with FIT than with FOBT in most studies.^28–30

FIT has a sensitivity of 69% to 86% for colorectal cancer and a specificity of 92% to 95%.³¹ The sensitivity can be improved by lowering the threshold value for a positive test, but this is associated with a decrease in specificity. A single FIT has the same sensitivity and specificity as several samples.³²

In a large retrospective US cohort study of programmatic screening with FIT, Jensen et al³³ reported that 48% of 670,841 people who were offered testing actually did the test. Of the 48% who participated in the first round and remained eligible, 75% to 86% participated in subsequent rounds over 4 years. Those who had a positive result on FIT were supposed to undergo colonoscopy, but 22% did not.

The US Multi-Society Task Force on Colorectal Cancer³⁴ suggests that FIT-based screening programs aim for a target FIT completion rate of more than 60% and a target colonoscopy completion rate of more than 80% of patients with positive FITs. These benchmarks were derived from adherence rates in international FIT screening studies in average-risk populations.^35–39 (Note that the large US cohort described above³³ did not meet these goals.) Ideally, every patient with a positive FIT should undergo diagnostic colonoscopy, but in reality only 50% to 83% actually do. Methods shown to improve adherence include structured screening programs with routine performance reports, provider feedback, and involvement of patient navigators.^40–42

Accordingly, several aspects of stool-based testing need to be stressed with patients. Understanding that FOBT is recommended yearly is integral for optimal impact on colorectal cancer incidence and mortality rates.

Additionally, patients should be advised to undergo colonoscopy soon after a positive FIT, because delaying colonoscopy could give precancerous lesions time to progress in stage. The acceptable time between a positive FIT and colonoscopy has yet to be determined. However, a retrospective cohort study of 1.26 million screened patients with 107,000 positive FIT results demonstrated that the rates of cancer discovered on colonoscopy were similar when performed within 30 days or up to 10 months after a positive test. Detection rates increased from 3% to 4.8% at 10 months and to 7.9% at 12 months.⁴³

In modeling studies, Meester et al⁴⁴ showed the estimated lifetime risk and mortality rates from colorectal cancer and life-years gained from screening are significantly better when colonoscopy is completed within 2 weeks rather than 1 year after a positive FIT. Each additional month after 2 weeks incrementally affected these outcomes, with a 1.4% increase in cancer mortality. These data suggest that colonoscopy should be done soon after a positive FIT result and at a maximum of 10 months.^43,44

Screening with FOBT is a multistep process for patients that includes receiving the test kit, collecting the sample, preparing it, returning it, undergoing colonoscopy after a positive test, and repeating in 1 year if negative. The screening program should identify patients at average risk in whom screening is appropriate, ensure delivery of the test, verify the quality of collected samples for laboratory testing against the manufacturer’s recommendations, and report results. Report of a positive FOBT result should provide recommendations for follow-up.

Though evidence clearly supports screening annually or biennially (every 2 years) with FOBT, the ideal interval for FIT is undetermined. Modeling studies utilized by the USPSTF and Multi-Society Task Force demonstrate that colonoscopy and annual FIT result in similar life-years gained, while 2 population-based screening programs have demonstrated that a 2- or 3-year interval may be equally efficacious by lowering the threshold for a positive test.^38,45

Randomized controlled trials of screening colonoscopy vs annual and biennial FIT are currently under way. Cost-effectiveness analysis has shown that offering single-sample FITs at more frequent (annual) intervals performs better than multisample testing at less frequent intervals.^45–47

Compared with stool-based screening, colonoscopy has advantages, including a 10-year screening interval if bowel preparation is adequate and the examination shows no neoplasia, the ability to inspect the entire colon, and the ability to diagnose and treat lesions in the same session.

Screening colonoscopy visualizes the entire colon in more than 98% of cases, although it requires adequate bowel preparation for maximal polyp detection. It can be done safely with or without sedation.⁴⁸

While there are no available randomized controlled trial data on the impact of screening colonoscopy on cancer incidence or mortality, extensive case-control and cohort studies consistently show that screening colonoscopy reduces cancer incidence and mortality rates.^49–54 A US Veterans Administration study of more than 32,000 patients reported a 50% reduction in overall colorectal cancer mortality.⁵⁵ In a microsimulation modeling study that assumed 100% adherence, colonoscopy every 10 years and annual FIT in individuals ages 50 to 75 provided similar life-years gained per 1,000 people screened (270 for colonoscopy, 244 for FIT).⁵⁶

Well-established metrics for maximizing the effectiveness and quality of colonoscopy have been established (Table 2). The most important include the mucosa inspection time (withdrawal time) and adenoma detection rate.⁵⁷ Withdrawal time is directly correlated with adenoma detection, and a 6-minute minimum withdrawal time is recommended in screening colonoscopy examinations of patients at average risk when no polyps are found.⁵⁸ The adenoma detection rate is the strongest evidence-based metric, as each 1% increase in the adenoma detection rate over 19% is associated with a 3% decrease in the risk of colorectal cancer and a 5% decrease in death rate.⁵⁹ The average-risk screening adenoma detection rate differs based on sex: the rate is greater than 20% for women and greater than 30% for men.

Complications from screening, diagnostic, or therapeutic colonoscopy are infrequent but include perforation (4/10,000) and significant intestinal bleeding (8/10,000).^56–62

Patients with a first-degree relative under age 60 with advanced adenomas or colorectal cancer are considered at high risk and should begin screening colonoscopy at age 40, with repeat colonoscopy at 5-year intervals, given a trend toward advanced neoplasia detection compared with FIT.⁶³

Guidelines recently published by the Canadian Association of Gastroenterology and endorsed by the American Gastroenterological Association also support starting screening in high-risk individuals at age 40, with a surveillance interval of 5 to 10 years based on the number of first-degree relatives with colorectal cancer or adenomas.⁶⁴ Consensus statements were based on retrospective cohort, prospective case-controlled, and cross-sectional studies comparing the risk of colorectal cancer in individuals with a family history against those without a family history.

Randomized clinical trials comparing colonoscopy and FIT are under way. Interim analysis of a European trial in which asymptomatic adults ages 50 to 69 were randomized to 1-time colonoscopy (26,703 patients) vs FIT every 2 years (26,599 patients) found significantly higher participation rates in the FIT arm (34.2% vs 24.6%) but higher rates of nonadvanced adenomas (4.2% vs 0.4%) and advanced neoplasia (1.9% vs 0.9%) in the colonoscopy arm.⁶⁵ Cancer was detected in 0.1% in each arm. These findings correlate with those of another study showing higher participation with FIT but higher advanced neoplasia detection rates with colonoscopy.⁶⁶

Detection of precursor lesions is vital, as removing neoplasms is the main strategy to reduce colorectal cancer incidence. Accordingly, the advantage of colonoscopy was illustrated by a study that determined that 53 patients would need to undergo screening colonoscopy to detect 1 advanced adenoma or cancerous lesion, compared with 264 for FIT.⁶⁷

STARTING SCREEING AT AGE 45

The American Cancer Society recently provided a qualified recommendation to start colorectal cancer screening in all individuals at age 45 rather than 50.⁹ This recommendation was based on modeling studies demonstrating that starting screening at age 45 with colonoscopy every 10 years resulted in 25 life-years gained at the cost of 610 colonoscopies per 1,000 individuals. Alternative strategies included FIT, which resulted in an additional 26 life-years gained per 1,000 individuals screened, flexible sigmoidoscopy (23 life-years gained), and computed tomographic colonoscopy (22 life-years gained).

Rates of colorectal cancer are rising in adults under age 50, and 10,000 new cases are anticipated this year.^2,3 Currently, 22 million US adults are between the ages of 45 and 50. The system and support needed to perform screening in all adults over age 45 and a lack of direct evidence to support its benefits in the young population need to be considered before widespread acceptance of the American Cancer Society recommendations. However, if screening is considered, FIT with or without sigmoidoscopy may be appropriate, given that most cancers diagnosed in individuals under age 50 are left-sided.^4,5

Screening has not been proven to reduce all-cause mortality. Randomized controlled trials of FOBT and observational studies of colonoscopy show that screening reduces cancer incidence and mortality. Until the currently ongoing randomized controlled trials comparing colonoscopy with FIT are completed, their comparative impact on colorectal cancer end points is unknown.

PATIENT ADHERENCE IS KEY

FIT and colonoscopy are the most prevalent screening methods in the United States. Careful attention should be given to offer the screening option the patient is most likely to complete, as adherence is key to the benefit from colorectal cancer screening.

The National Colorectal Cancer Roundtable (nccrt.org), established in 1997 by the American Cancer Society and the US Centers for Disease Control and Prevention, is a national coalition of public and private organizations dedicated to reducing colorectal cancer incidence and mortality. The Roundtable waged a national campaign to achieve a colorectal cancer screening rate of 80% in eligible adults by 2018, a goal that was not met. Still, the potential for a substantial impact is a compelling reason to endorse adherence to colorectal cancer screening. The Roundtable provides many resources for physicians to enhance screening in their practice.

The United States has seen a steady decline in colorectal cancer incidence and mortality, mainly as a result of screening. Colorectal cancer is preventable with ensuring patients’ adherence to screening. Screening rates have been shown to increase with patient-provider dialogue and with selection of a screening program the patient prefers and is most likely to complete.

Keeping up with current evidence-based healthcare practices is key to providing good clinical care to patients. This review presents 5 vignettes that highlight key issues in women’s health: osteoporosis screening, hormonal contraceptive interactions with antibiotics, hormone replacement therapy in carriers of the BRCA1 gene mutation, risks associated with hormonal contraception, and breast cancer diagnosis using digital tomosynthesis in addition to digital mammography. Supporting articles, all published in 2017 and 2018, were selected from high-impact medical and women’s health journals.

OSTEOPOROSIS SCREENING FOR FRACTURE PREVENTION

A 60-year-old woman reports that her last menstrual period was 7 years ago. She has no history of falls or fractures, and she takes no medications. She smokes 10 cigarettes per day and drinks 3 to 4 alcoholic beverages on most days of the week. She is 5 feet 6 inches (170 cm) tall and weighs 107 lb. Should she be screened for osteoporosis?

Osteoporosis is underdiagnosed

It is estimated that, in the United States, 12.3 million individuals older than 50 will develop osteoporosis by 2020. Missed opportunities to screen high-risk individuals can lead to fractures, including fractures of the hip.¹

Updated screening recommendations

In 2018, the US Preventive Services Task Force (USPSTF) developed and published evidence-based recommendations for osteoporosis screening to help providers identify and treat osteoporosis early to prevent fractures.² Available evidence on screening and treatment in women and men were reviewed with the intention of updating the 2011 USPSTF recommendations. The review also evaluated risk assessment tools, screening intervals, and efficacy of screening and treatment in various subpopulations.

Since the 2011 recommendations, more data have become available on fracture risk assessment with or without bone mineral density measurements. In its 2018 report, the USPSTF recommends that postmenopausal women younger than 65 should undergo screening with a bone density test if their 10-year risk of major osteoporotic fracture is more than 8.4%. This is equivalent to the fracture risk of a 65-year-old white woman with no major risk factors for fracture (grade B recommendation—high certainty that the benefit is moderate, or moderate certainty that the benefit is moderate to substantial).²

Assessment of fracture risk

For postmenopausal women who are under age 65 and who have at least 1 risk factor for fracture, it is reasonable to use a clinical risk assessment tool to determine who should undergo screening with bone mineral density measurement. Risk factors associated with an increased risk of osteoporotic fractures include a parental history of hip fracture, smoking, intake of 3 or more alcoholic drinks per day, low body weight, malabsorption, rheumatoid arthritis, diabetes, and postmenopausal status (not using estrogen replacement). Medications should be carefully reviewed for those that can increase the risk of fractures, including steroids and antiestrogen treatments.

The 10-year risk of a major osteoporotic or hip fracture can be assessed using the Fractional Risk Assessment Tool (FRAX), available at www.sheffield.ac.uk/FRAX/. Other acceptable tools that perform similarly to FRAX include the Osteoporosis Risk Assessment Instrument (ORAI) (10 studies; N = 16,780), Osteoporosis Index of Risk (OSIRIS) (5 studies; N = 5,649), Osteoporosis Self-Assessment Tool (OST) (13 studies; N = 44,323), and Simple Calculated Osteoporosis Risk Estimation (SCORE) (8 studies; N = 15,362).

Should this patient be screened for osteoporosis?

Based on the FRAX, this patient’s 10-year risk of major osteoporosis fracture is 9.2%. She would benefit from osteoporosis screening with a bone density test.

DO ANTIBIOTICS REDUCE EFFECTIVENESS OF HORMONAL CONTRACEPTION?

A 27-year-old woman presents with a dog bite on her right hand and is started on oral antibiotics. She takes an oral contraceptive that contains 35 µg of ethinyl estradiol and 0.25 mg of norgestimate. She asks if she should use condoms while taking antibiotics.

The antibiotics rifampin and rifabutin are known inducers of the hepatic enzymes required for contraceptive steroid metabolism, whereas other antibiotics are not. Despite the lack of compelling evidence that broad-spectrum antibiotics interfere with the efficacy of hormonal contraception, most pharmacists recommend backup contraception for women who use concomitant antibiotics.³ This practice could lead to poor compliance with the contraceptive regimen, the antibiotic regimen, or both.³

Simmons et al³ conducted a systematic review of randomized and nonrandomized studies that assessed pregnancy rates, breakthrough bleeding, ovulation suppression, and hormone pharmacokinetics in women taking oral or vaginal hormonal contraceptives in combination with nonrifamycin antibiotics, including oral, intramuscular, and intravenous forms. Oral contraceptives used in the studies included a range of doses and progestins, but lowest-dose pills, such as those containing less than 30 µg ethinyl estradiol or less than 150 µg levonorgestrel, were not included.

The contraceptive formulations in this systematic review³ included oral contraceptive pills, emergency contraception pills, and the contraceptive vaginal ring. The effect of antibiotics on other nonoral contraceptives, such as the transdermal patch, injectables, and progestin implants was not studied.

Four observational studies³ evaluated pregnancy rates or hormonal contraception failure with any antibiotic use. In 2 of these 4 studies, there was no difference in pregnancy rates in women who used oral contraceptives with and without nonrifamycin antibiotics. However, ethinyl estradiol was shown to have increased clearance when administered with dirithromycin (a macrolide).³ Twenty-five of the studies reported measures of contraceptive effectiveness (ovulation) and pharmacokinetic outcomes.

There were no observed differences in ovulation suppression or breakthrough bleeding in any study that combined hormonal contraceptives with an antibiotic. Furthermore, there was no significant decrease in progestin pharmacokinetic parameters during coadministration with an antibiotic.³ Study limitations included small sample sizes and the observational nature of the data.

How would you counsel this patient?

Available evidence suggests that nonrifamycin antibiotics do not diminish the effectiveness of the vaginal contraceptive ring or an oral hormonal contraceptive that contains at least 30 µg of ethinyl estradiol or 150 µg of levonorgestrel. Current guidelines do not recommend the use of additional backup contraception, regardless of hormonal contraception dose or formulation.⁴ Likewise, the most recent guidance for dental practitioners (ie, from 2012) no longer advises women to use additional contraceptive protection when taking nonrifamycin antibiotics.⁵

In our practice, we discuss the option of additional protection when prescribing formulations with lower estrogen doses (< 30 µg), not only because of the limitations of the available data, but also because of the high rates of unintended pregnancy with typical use of combined hormonal contraceptives (9% per year, unrelated to use of antibiotics).⁴ However, if our patient would rather not use additional barrier methods, she can be reassured that concomitant nonrifamycin antibiotic use is unlikely to affect contraceptive effectiveness.

A 41-year-old healthy mother of 3 was recently found to be a carrier of the BRCA1 mutation. She is planning to undergo prophylactic bilateral salpingo-oophorectomy for ovarian cancer prevention. However, she is apprehensive about undergoing surgical menopause. Should she be started on hormone replacement therapy after oophorectomy? How would hormone replacement therapy affect her risk of breast cancer?

In females who carry the BRCA1 mutation, the cumulative risk of both ovarian and breast cancer approaches 44% (95% confidence interval [CI] 36%–53%) and 72% (95% CI 65%–79%) by age 80.⁶ Prophylactic salpingo-oophorectomy reduces the risk of breast cancer by 50% and the risk of ovarian cancer by 90%. Unfortunately, premature withdrawal of ovarian hormones has been associated with long-term adverse effects including significant vasomotor symptoms, decreased quality of life, sexual dysfunction, early mortality, bone loss, decline in mood and cognition, and poor cardiovascular outcomes.⁷ Many of these effects can be avoided or lessened with hormone replacement therapy.

Kotsopoulos et al⁸ conducted a longitudinal, prospective analysis of BRCA1 mutation carriers in a multicenter study between 1995 and 2017. The mean follow-up period was 7.6 years (range 0.4–22.1). The study assessed associations between the use of hormone replacement therapy and breast cancer risk in carriers of the BRCA1 mutation who underwent prophylactic salpingo-oophorectomy. Study participants did not have a personal history of cancer. Those with a history of prophylactic mastectomy were excluded.

Participants completed a series of questionnaires every 2 years, disclosing updates in personal medical, cancer, and reproductive history. The questionnaires also inquired about the use of hormone replacement therapy, including the type used (estrogen only, progestin only, estrogen plus progestin, other), brand name, duration of use, and dose and route of administration (pill, patch, suppository).

Of the 13,087 BRCA1 mutation carriers identified, 872 met the study criteria. Of those, 377 (43%) reported using some form of hormone replacement therapy after salpingo-oophorectomy, and 495 (57%) did not. The average duration of use was 3.9 years (range 0.5–19), with most (69%) using estrogen alone; 18% used other regimens, including estrogen plus progestin and progestin only. A small percentage of participants did not indicate which formulation they used. On average, women using hormone replacement therapy underwent prophylactic oophorectomy earlier than nonusers (age 43.0 vs 48.4; absolute difference 5.5 years, P < .001).

During follow-up, there was no significant difference noted in the proportion of women diagnosed with breast cancer between hormone replacement therapy users and nonusers (10.3 vs 10.7%; absolute difference 0.4%; P = .86). In fact, for each year of estrogen-containing hormone replacement therapy, there was an 18% reduction in breast cancer risk when oophorectomy was performed before age 45 (95% CI 0.69–0.97). The authors also noted a nonsignificant 14% trend toward an increase in breast cancer risk for each year of progestin use after oophorectomy when surgery was performed before age 45 (95% CI 0.9–1.46).

Although prophylactic hysterectomy was not recommended, the authors noted that hysterectomy would eliminate the need for progestin-containing hormone replacement therapy. For those who underwent oophorectomy after age 45, hormone replacement therapy did not increase or decrease the risk of breast cancer.⁷

A meta-analysis by Marchetti et al⁹ also supports the safety of hormone replacement therapy after risk-reducing salpingo-oophorectomy. Three studies that included 1,100 patients were analyzed (including the Kotsopoulos study⁸ noted above). There was a nonsignificant decrease in breast cancer risk in women on estrogen-only hormone replacement therapy compared with women on estrogen-plus-progestin therapy (odds ratio 0.53, 95% CI 0.25–1.15). Overall, the authors regarded hormone replacement therapy as a safe therapeutic option after prophylactic salpingo-oophorectomy in carriers of the BRCA1 and BRCA2 mutations.⁹

In a case-control study published in 2016,¹⁰ hormone replacement therapy was assessed in 432 postmenopausal BRCA1 mutation carriers with invasive breast cancer (cases) and in 432 BRCA1 mutation carriers without a history of breast cancer (controls). Results showed no difference in breast cancer risk between hormone replacement therapy users and nonusers.¹⁰

Rebbeck et al¹¹ evaluated short-term hormone replacement therapy in BRCA1 and BRCA2 gene-mutation carriers after they underwent prophylactic salpingo-oophorectomy. The results showed that hormone replacement did not affect the breast cancer risk-reduction conferred with prophylactic bilateral salpingo-oophorectomy.

Johansen et al¹² evaluated hormone replacement therapy in premenopausal women after prophylactic salpingo-oophorectomy. They studied 324 carriers of BRCA gene mutations after they underwent prophylactic salpingo-oophorectomy and a subset of 950 controls who had bilateral salpingo-oophorectomy for reasons unrelated to cancer. In both groups, hormone replacement therapy was underutilized. The authors recommended using it when clinically indicated.

Should your patient start hormone replacement therapy?

This patient is healthy, and in the absence of contraindications, systemic hormone replacement therapy after prophylactic oophorectomy could mitigate the potential adverse effects of surgically induced menopause. The patient can be reassured that estrogen-containing short-term hormone replacement therapy is unlikely to increase her breast cancer risk.

A 44-year-old woman presents to your office for an annual visit. She is sexually active but does not wish to become pregnant. She has a family history of breast cancer: her mother was diagnosed at age 53. She is interested in an oral contraceptive to prevent pregnancy and acne. However, she is nervous about being on any contraceptive that may increase her risk of breast cancer.

To date, studies assessing the effect of hormonal contraception on the risk of breast cancer have produced inconsistent results. Although most studies have shown no associated risk, a few have shown a temporary 20% to 30% increased risk of breast cancer during use.^13,14 Case-controlled studies that reported an association between hormonal contraception and breast cancer included populations taking higher-dose combination pills, which are no longer prescribed. Most studies do not evaluate specific formulations of hormonal contraception, and little is known about effects associated with intrauterine devices or progestin-only contraception.

A prospective study performed by Mørch et al¹³ followed more than 1 million reproductive-aged women for a mean of 10.9 years. The Danish Cancer Registry was used to identify cases of invasive breast cancer. Women who used hormonal contraceptives had a relative risk of breast cancer of 1.20 compared with women not on hormonal contraception (95% CI 1.14–1.26). The study suggested that those who had been on contraceptive agents for more than 5 years had an increased risk and that this risk remained for 5 years after the agents were discontinued. Conversely, no increased risk of cancer was noted in those who used hormonal contraception for less than 5 years. No notable differences were seen among various formulations.

For women using the levonorgestrel-containing intrauterine device, the relative risk of breast cancer was 1.21 (95% CI 1.11–1.33). A few cancers were noted in those who used the progestin-only implant or those using depot medroxyprogesterone acetate. While the study showed an increased relative risk of breast cancer, the absolute risk was low—13 cases per 100,000, or approximately 1 additional case of breast cancer per 7,690 per year.¹³

This study had several important limitations. The authors did not adjust for common breast cancer risk factors including age at menarche, alcohol use, or breastfeeding. Additionally, the study did not account for the use of hormonal contraception before the study period and conversely, did not account for women who may have stopped taking their contraceptive despite their prescribed duration. The frequency of mammography was not explicitly noted, which could have shifted results for women who had more aggressive screening.

It is also noteworthy that the use of high-dose systemic progestins was not associated with an increased risk, whereas the levonorgestrel intrauterine device, which contains only 1/20th the dose of a low-dose oral contraceptive pill, was associated with an increased risk. This discrepancy in risk warrants further investigation, and clinicians should be aware that this inconsistency needs validation before changing clinical practice.

In an observational cohort study,¹⁵ more than 100,000 women ages 50 to 71 were followed prospectively for 15 years to evaluate the association between hormonal contraceptive use and the risk of gynecologic and breast cancers. In this study, the duration of hormonal contraceptive use, smoking status, alcohol use, body mass index, physical activity, and family history of cancer were recorded. Long-term hormonal contraceptive use reduced ovarian and endometrial cancer risks by 40% and 34%, respectively, with no increase in breast cancer risk regardless of family history.

How would you counsel the patient?

The patient should be educated on the benefits of hormonal contraception that extend beyond pregnancy prevention, including regulation of menses, improved acne, decreased risk of endometrial and ovarian cancer, and likely reductions in colorectal cancer and overall mortality risk.^13–16 Further, after their own systematic review of the data assessing risk of breast cancer with hormonal contraception, the US Centers for Disease Control and Prevention state in their guidelines that all contraceptives may be used without limitation in those who have a family history of breast cancer.⁴ Any potential increased risk of breast cancer in women using hormonal contraception is small and would not outweigh the benefits associated with use.

One must consider the impact of an unintended pregnancy in such women, including effects on the health of the fetus and mother. Recent reports on the increasing rates of maternal death in the US (23.8 of 100,000 live births) serve as a reminder of the complications that can arise with pregnancy, especially if a mother’s health is not optimized before conception.¹⁷

The same 44-year-old patient now inquires about screening for breast cancer. She is curious about 3-dimensional mammography and whether it would be a better screening test for her.

Digital breast tomosynthesis (DBT) is a newer imaging modality that provides a 3-dimensional reconstruction of the breast using low-dose x-ray imaging. Some studies have shown that combining DBT with digital mammography may be superior to digital mammography alone in detecting cancers.¹⁸ However, digital mammography is currently the gold standard for breast cancer screening and is the only test proven to reduce mortality.^18,19

In a retrospective US study of 13 medical centers,²⁰ breast cancer detection rates increased by 41% the year after DBT was introduced, from 2.9 to 4.1 per 1,000 cases. DBT was associated with 16 fewer patients recalled for repeat imaging out of 1,000 women screened (as opposed to mammography alone). Two European studies similarly suggested an increase in cancer detection with lower recall rates.^21,22

Is 3-D mammography a better option?

In a 2-arm study by Pattacini et al,¹⁸ nearly 20,000 women ages 45 to 70 were randomized to undergo either digital mammography or digital mammography plus DBT for primary breast cancer screening. Women were enrolled over a 2-year period and were followed for 4.5 years, and the development of a primary invasive cancer was the primary end point. Recall rates, reading times, and radiation doses were also compared between the 2 groups.

Overall, the cancer detection rate was higher in the digital mammography plus DBT arm compared with digital mammography alone (8.6 vs 4.5 per 1,000). The detection rates were higher in the combined screening group among all age subgroups, with relative risks ranging from 1.83 to 2.04 (P = .93). The recall rate was 3.5% in the 2 arms, with relative risks ranging from 0.93 to 1.11 (P = .52). There was a reduction in the number of false positives seen in women undergoing digital mammography plus DBT when compared with digital mammography alone, from 30 per 1,000 to 27 per 1,000.

Detection of ductal carcinoma in situ increased in the experimental arm (relative detection 2.80, 95% CI 1.01–7.65) compared with invasive cancers. Comparing radiation, the dose was 2.3 times higher in those who underwent digital mammography plus DBT. The average reading times for digital mammography alone were 20 to 85 seconds; adding DBT added 35 to 81 seconds.¹⁹

Should you advise 3-D mammography?

The patient should be educated on the benefits of both digital mammography alone and digital mammography plus DBT. The use of digital mammography plus DBT has been supported in various studies and has been shown to increase cancer detection rates, although data are still conflicting regarding recall rates.^19,20 More studies are needed to determine its effect on breast cancer morality.

Routine use of DBT in women with or without dense breast tissue has not been recommended by organizations such as the USPSTF and the American College of Obstetricians and Gynecologists.^23,24 While there is an increased dose of radiation, it still falls below the US Food and Drug Administration limits and should not be the sole barrier to use.

Keeping up with current evidence-based healthcare practices is key to providing good clinical care to patients. This review presents 5 vignettes that highlight key issues in women’s health: osteoporosis screening, hormonal contraceptive interactions with antibiotics, hormone replacement therapy in carriers of the BRCA1 gene mutation, risks associated with hormonal contraception, and breast cancer diagnosis using digital tomosynthesis in addition to digital mammography. Supporting articles, all published in 2017 and 2018, were selected from high-impact medical and women’s health journals.

OSTEOPOROSIS SCREENING FOR FRACTURE PREVENTION

A 60-year-old woman reports that her last menstrual period was 7 years ago. She has no history of falls or fractures, and she takes no medications. She smokes 10 cigarettes per day and drinks 3 to 4 alcoholic beverages on most days of the week. She is 5 feet 6 inches (170 cm) tall and weighs 107 lb. Should she be screened for osteoporosis?

Osteoporosis is underdiagnosed

It is estimated that, in the United States, 12.3 million individuals older than 50 will develop osteoporosis by 2020. Missed opportunities to screen high-risk individuals can lead to fractures, including fractures of the hip.¹

Updated screening recommendations

In 2018, the US Preventive Services Task Force (USPSTF) developed and published evidence-based recommendations for osteoporosis screening to help providers identify and treat osteoporosis early to prevent fractures.² Available evidence on screening and treatment in women and men were reviewed with the intention of updating the 2011 USPSTF recommendations. The review also evaluated risk assessment tools, screening intervals, and efficacy of screening and treatment in various subpopulations.

Since the 2011 recommendations, more data have become available on fracture risk assessment with or without bone mineral density measurements. In its 2018 report, the USPSTF recommends that postmenopausal women younger than 65 should undergo screening with a bone density test if their 10-year risk of major osteoporotic fracture is more than 8.4%. This is equivalent to the fracture risk of a 65-year-old white woman with no major risk factors for fracture (grade B recommendation—high certainty that the benefit is moderate, or moderate certainty that the benefit is moderate to substantial).²

Assessment of fracture risk

For postmenopausal women who are under age 65 and who have at least 1 risk factor for fracture, it is reasonable to use a clinical risk assessment tool to determine who should undergo screening with bone mineral density measurement. Risk factors associated with an increased risk of osteoporotic fractures include a parental history of hip fracture, smoking, intake of 3 or more alcoholic drinks per day, low body weight, malabsorption, rheumatoid arthritis, diabetes, and postmenopausal status (not using estrogen replacement). Medications should be carefully reviewed for those that can increase the risk of fractures, including steroids and antiestrogen treatments.

The 10-year risk of a major osteoporotic or hip fracture can be assessed using the Fractional Risk Assessment Tool (FRAX), available at www.sheffield.ac.uk/FRAX/. Other acceptable tools that perform similarly to FRAX include the Osteoporosis Risk Assessment Instrument (ORAI) (10 studies; N = 16,780), Osteoporosis Index of Risk (OSIRIS) (5 studies; N = 5,649), Osteoporosis Self-Assessment Tool (OST) (13 studies; N = 44,323), and Simple Calculated Osteoporosis Risk Estimation (SCORE) (8 studies; N = 15,362).

Should this patient be screened for osteoporosis?

Based on the FRAX, this patient’s 10-year risk of major osteoporosis fracture is 9.2%. She would benefit from osteoporosis screening with a bone density test.

DO ANTIBIOTICS REDUCE EFFECTIVENESS OF HORMONAL CONTRACEPTION?

A 27-year-old woman presents with a dog bite on her right hand and is started on oral antibiotics. She takes an oral contraceptive that contains 35 µg of ethinyl estradiol and 0.25 mg of norgestimate. She asks if she should use condoms while taking antibiotics.

The antibiotics rifampin and rifabutin are known inducers of the hepatic enzymes required for contraceptive steroid metabolism, whereas other antibiotics are not. Despite the lack of compelling evidence that broad-spectrum antibiotics interfere with the efficacy of hormonal contraception, most pharmacists recommend backup contraception for women who use concomitant antibiotics.³ This practice could lead to poor compliance with the contraceptive regimen, the antibiotic regimen, or both.³

Simmons et al³ conducted a systematic review of randomized and nonrandomized studies that assessed pregnancy rates, breakthrough bleeding, ovulation suppression, and hormone pharmacokinetics in women taking oral or vaginal hormonal contraceptives in combination with nonrifamycin antibiotics, including oral, intramuscular, and intravenous forms. Oral contraceptives used in the studies included a range of doses and progestins, but lowest-dose pills, such as those containing less than 30 µg ethinyl estradiol or less than 150 µg levonorgestrel, were not included.

The contraceptive formulations in this systematic review³ included oral contraceptive pills, emergency contraception pills, and the contraceptive vaginal ring. The effect of antibiotics on other nonoral contraceptives, such as the transdermal patch, injectables, and progestin implants was not studied.

Four observational studies³ evaluated pregnancy rates or hormonal contraception failure with any antibiotic use. In 2 of these 4 studies, there was no difference in pregnancy rates in women who used oral contraceptives with and without nonrifamycin antibiotics. However, ethinyl estradiol was shown to have increased clearance when administered with dirithromycin (a macrolide).³ Twenty-five of the studies reported measures of contraceptive effectiveness (ovulation) and pharmacokinetic outcomes.

There were no observed differences in ovulation suppression or breakthrough bleeding in any study that combined hormonal contraceptives with an antibiotic. Furthermore, there was no significant decrease in progestin pharmacokinetic parameters during coadministration with an antibiotic.³ Study limitations included small sample sizes and the observational nature of the data.

How would you counsel this patient?

Available evidence suggests that nonrifamycin antibiotics do not diminish the effectiveness of the vaginal contraceptive ring or an oral hormonal contraceptive that contains at least 30 µg of ethinyl estradiol or 150 µg of levonorgestrel. Current guidelines do not recommend the use of additional backup contraception, regardless of hormonal contraception dose or formulation.⁴ Likewise, the most recent guidance for dental practitioners (ie, from 2012) no longer advises women to use additional contraceptive protection when taking nonrifamycin antibiotics.⁵

In our practice, we discuss the option of additional protection when prescribing formulations with lower estrogen doses (< 30 µg), not only because of the limitations of the available data, but also because of the high rates of unintended pregnancy with typical use of combined hormonal contraceptives (9% per year, unrelated to use of antibiotics).⁴ However, if our patient would rather not use additional barrier methods, she can be reassured that concomitant nonrifamycin antibiotic use is unlikely to affect contraceptive effectiveness.

A 41-year-old healthy mother of 3 was recently found to be a carrier of the BRCA1 mutation. She is planning to undergo prophylactic bilateral salpingo-oophorectomy for ovarian cancer prevention. However, she is apprehensive about undergoing surgical menopause. Should she be started on hormone replacement therapy after oophorectomy? How would hormone replacement therapy affect her risk of breast cancer?

In females who carry the BRCA1 mutation, the cumulative risk of both ovarian and breast cancer approaches 44% (95% confidence interval [CI] 36%–53%) and 72% (95% CI 65%–79%) by age 80.⁶ Prophylactic salpingo-oophorectomy reduces the risk of breast cancer by 50% and the risk of ovarian cancer by 90%. Unfortunately, premature withdrawal of ovarian hormones has been associated with long-term adverse effects including significant vasomotor symptoms, decreased quality of life, sexual dysfunction, early mortality, bone loss, decline in mood and cognition, and poor cardiovascular outcomes.⁷ Many of these effects can be avoided or lessened with hormone replacement therapy.

Kotsopoulos et al⁸ conducted a longitudinal, prospective analysis of BRCA1 mutation carriers in a multicenter study between 1995 and 2017. The mean follow-up period was 7.6 years (range 0.4–22.1). The study assessed associations between the use of hormone replacement therapy and breast cancer risk in carriers of the BRCA1 mutation who underwent prophylactic salpingo-oophorectomy. Study participants did not have a personal history of cancer. Those with a history of prophylactic mastectomy were excluded.

Participants completed a series of questionnaires every 2 years, disclosing updates in personal medical, cancer, and reproductive history. The questionnaires also inquired about the use of hormone replacement therapy, including the type used (estrogen only, progestin only, estrogen plus progestin, other), brand name, duration of use, and dose and route of administration (pill, patch, suppository).

Of the 13,087 BRCA1 mutation carriers identified, 872 met the study criteria. Of those, 377 (43%) reported using some form of hormone replacement therapy after salpingo-oophorectomy, and 495 (57%) did not. The average duration of use was 3.9 years (range 0.5–19), with most (69%) using estrogen alone; 18% used other regimens, including estrogen plus progestin and progestin only. A small percentage of participants did not indicate which formulation they used. On average, women using hormone replacement therapy underwent prophylactic oophorectomy earlier than nonusers (age 43.0 vs 48.4; absolute difference 5.5 years, P < .001).

During follow-up, there was no significant difference noted in the proportion of women diagnosed with breast cancer between hormone replacement therapy users and nonusers (10.3 vs 10.7%; absolute difference 0.4%; P = .86). In fact, for each year of estrogen-containing hormone replacement therapy, there was an 18% reduction in breast cancer risk when oophorectomy was performed before age 45 (95% CI 0.69–0.97). The authors also noted a nonsignificant 14% trend toward an increase in breast cancer risk for each year of progestin use after oophorectomy when surgery was performed before age 45 (95% CI 0.9–1.46).

Although prophylactic hysterectomy was not recommended, the authors noted that hysterectomy would eliminate the need for progestin-containing hormone replacement therapy. For those who underwent oophorectomy after age 45, hormone replacement therapy did not increase or decrease the risk of breast cancer.⁷

A meta-analysis by Marchetti et al⁹ also supports the safety of hormone replacement therapy after risk-reducing salpingo-oophorectomy. Three studies that included 1,100 patients were analyzed (including the Kotsopoulos study⁸ noted above). There was a nonsignificant decrease in breast cancer risk in women on estrogen-only hormone replacement therapy compared with women on estrogen-plus-progestin therapy (odds ratio 0.53, 95% CI 0.25–1.15). Overall, the authors regarded hormone replacement therapy as a safe therapeutic option after prophylactic salpingo-oophorectomy in carriers of the BRCA1 and BRCA2 mutations.⁹

In a case-control study published in 2016,¹⁰ hormone replacement therapy was assessed in 432 postmenopausal BRCA1 mutation carriers with invasive breast cancer (cases) and in 432 BRCA1 mutation carriers without a history of breast cancer (controls). Results showed no difference in breast cancer risk between hormone replacement therapy users and nonusers.¹⁰

Rebbeck et al¹¹ evaluated short-term hormone replacement therapy in BRCA1 and BRCA2 gene-mutation carriers after they underwent prophylactic salpingo-oophorectomy. The results showed that hormone replacement did not affect the breast cancer risk-reduction conferred with prophylactic bilateral salpingo-oophorectomy.

Johansen et al¹² evaluated hormone replacement therapy in premenopausal women after prophylactic salpingo-oophorectomy. They studied 324 carriers of BRCA gene mutations after they underwent prophylactic salpingo-oophorectomy and a subset of 950 controls who had bilateral salpingo-oophorectomy for reasons unrelated to cancer. In both groups, hormone replacement therapy was underutilized. The authors recommended using it when clinically indicated.

Should your patient start hormone replacement therapy?

This patient is healthy, and in the absence of contraindications, systemic hormone replacement therapy after prophylactic oophorectomy could mitigate the potential adverse effects of surgically induced menopause. The patient can be reassured that estrogen-containing short-term hormone replacement therapy is unlikely to increase her breast cancer risk.

A 44-year-old woman presents to your office for an annual visit. She is sexually active but does not wish to become pregnant. She has a family history of breast cancer: her mother was diagnosed at age 53. She is interested in an oral contraceptive to prevent pregnancy and acne. However, she is nervous about being on any contraceptive that may increase her risk of breast cancer.

To date, studies assessing the effect of hormonal contraception on the risk of breast cancer have produced inconsistent results. Although most studies have shown no associated risk, a few have shown a temporary 20% to 30% increased risk of breast cancer during use.^13,14 Case-controlled studies that reported an association between hormonal contraception and breast cancer included populations taking higher-dose combination pills, which are no longer prescribed. Most studies do not evaluate specific formulations of hormonal contraception, and little is known about effects associated with intrauterine devices or progestin-only contraception.

A prospective study performed by Mørch et al¹³ followed more than 1 million reproductive-aged women for a mean of 10.9 years. The Danish Cancer Registry was used to identify cases of invasive breast cancer. Women who used hormonal contraceptives had a relative risk of breast cancer of 1.20 compared with women not on hormonal contraception (95% CI 1.14–1.26). The study suggested that those who had been on contraceptive agents for more than 5 years had an increased risk and that this risk remained for 5 years after the agents were discontinued. Conversely, no increased risk of cancer was noted in those who used hormonal contraception for less than 5 years. No notable differences were seen among various formulations.

For women using the levonorgestrel-containing intrauterine device, the relative risk of breast cancer was 1.21 (95% CI 1.11–1.33). A few cancers were noted in those who used the progestin-only implant or those using depot medroxyprogesterone acetate. While the study showed an increased relative risk of breast cancer, the absolute risk was low—13 cases per 100,000, or approximately 1 additional case of breast cancer per 7,690 per year.¹³

This study had several important limitations. The authors did not adjust for common breast cancer risk factors including age at menarche, alcohol use, or breastfeeding. Additionally, the study did not account for the use of hormonal contraception before the study period and conversely, did not account for women who may have stopped taking their contraceptive despite their prescribed duration. The frequency of mammography was not explicitly noted, which could have shifted results for women who had more aggressive screening.

It is also noteworthy that the use of high-dose systemic progestins was not associated with an increased risk, whereas the levonorgestrel intrauterine device, which contains only 1/20th the dose of a low-dose oral contraceptive pill, was associated with an increased risk. This discrepancy in risk warrants further investigation, and clinicians should be aware that this inconsistency needs validation before changing clinical practice.

In an observational cohort study,¹⁵ more than 100,000 women ages 50 to 71 were followed prospectively for 15 years to evaluate the association between hormonal contraceptive use and the risk of gynecologic and breast cancers. In this study, the duration of hormonal contraceptive use, smoking status, alcohol use, body mass index, physical activity, and family history of cancer were recorded. Long-term hormonal contraceptive use reduced ovarian and endometrial cancer risks by 40% and 34%, respectively, with no increase in breast cancer risk regardless of family history.

How would you counsel the patient?

The patient should be educated on the benefits of hormonal contraception that extend beyond pregnancy prevention, including regulation of menses, improved acne, decreased risk of endometrial and ovarian cancer, and likely reductions in colorectal cancer and overall mortality risk.^13–16 Further, after their own systematic review of the data assessing risk of breast cancer with hormonal contraception, the US Centers for Disease Control and Prevention state in their guidelines that all contraceptives may be used without limitation in those who have a family history of breast cancer.⁴ Any potential increased risk of breast cancer in women using hormonal contraception is small and would not outweigh the benefits associated with use.

One must consider the impact of an unintended pregnancy in such women, including effects on the health of the fetus and mother. Recent reports on the increasing rates of maternal death in the US (23.8 of 100,000 live births) serve as a reminder of the complications that can arise with pregnancy, especially if a mother’s health is not optimized before conception.¹⁷

The same 44-year-old patient now inquires about screening for breast cancer. She is curious about 3-dimensional mammography and whether it would be a better screening test for her.

Digital breast tomosynthesis (DBT) is a newer imaging modality that provides a 3-dimensional reconstruction of the breast using low-dose x-ray imaging. Some studies have shown that combining DBT with digital mammography may be superior to digital mammography alone in detecting cancers.¹⁸ However, digital mammography is currently the gold standard for breast cancer screening and is the only test proven to reduce mortality.^18,19

In a retrospective US study of 13 medical centers,²⁰ breast cancer detection rates increased by 41% the year after DBT was introduced, from 2.9 to 4.1 per 1,000 cases. DBT was associated with 16 fewer patients recalled for repeat imaging out of 1,000 women screened (as opposed to mammography alone). Two European studies similarly suggested an increase in cancer detection with lower recall rates.^21,22

Is 3-D mammography a better option?

In a 2-arm study by Pattacini et al,¹⁸ nearly 20,000 women ages 45 to 70 were randomized to undergo either digital mammography or digital mammography plus DBT for primary breast cancer screening. Women were enrolled over a 2-year period and were followed for 4.5 years, and the development of a primary invasive cancer was the primary end point. Recall rates, reading times, and radiation doses were also compared between the 2 groups.

Overall, the cancer detection rate was higher in the digital mammography plus DBT arm compared with digital mammography alone (8.6 vs 4.5 per 1,000). The detection rates were higher in the combined screening group among all age subgroups, with relative risks ranging from 1.83 to 2.04 (P = .93). The recall rate was 3.5% in the 2 arms, with relative risks ranging from 0.93 to 1.11 (P = .52). There was a reduction in the number of false positives seen in women undergoing digital mammography plus DBT when compared with digital mammography alone, from 30 per 1,000 to 27 per 1,000.

Detection of ductal carcinoma in situ increased in the experimental arm (relative detection 2.80, 95% CI 1.01–7.65) compared with invasive cancers. Comparing radiation, the dose was 2.3 times higher in those who underwent digital mammography plus DBT. The average reading times for digital mammography alone were 20 to 85 seconds; adding DBT added 35 to 81 seconds.¹⁹

Should you advise 3-D mammography?

The patient should be educated on the benefits of both digital mammography alone and digital mammography plus DBT. The use of digital mammography plus DBT has been supported in various studies and has been shown to increase cancer detection rates, although data are still conflicting regarding recall rates.^19,20 More studies are needed to determine its effect on breast cancer morality.

Routine use of DBT in women with or without dense breast tissue has not been recommended by organizations such as the USPSTF and the American College of Obstetricians and Gynecologists.^23,24 While there is an increased dose of radiation, it still falls below the US Food and Drug Administration limits and should not be the sole barrier to use.

Keeping up with current evidence-based healthcare practices is key to providing good clinical care to patients. This review presents 5 vignettes that highlight key issues in women’s health: osteoporosis screening, hormonal contraceptive interactions with antibiotics, hormone replacement therapy in carriers of the BRCA1 gene mutation, risks associated with hormonal contraception, and breast cancer diagnosis using digital tomosynthesis in addition to digital mammography. Supporting articles, all published in 2017 and 2018, were selected from high-impact medical and women’s health journals.

OSTEOPOROSIS SCREENING FOR FRACTURE PREVENTION

A 60-year-old woman reports that her last menstrual period was 7 years ago. She has no history of falls or fractures, and she takes no medications. She smokes 10 cigarettes per day and drinks 3 to 4 alcoholic beverages on most days of the week. She is 5 feet 6 inches (170 cm) tall and weighs 107 lb. Should she be screened for osteoporosis?

Osteoporosis is underdiagnosed

It is estimated that, in the United States, 12.3 million individuals older than 50 will develop osteoporosis by 2020. Missed opportunities to screen high-risk individuals can lead to fractures, including fractures of the hip.¹

Updated screening recommendations

In 2018, the US Preventive Services Task Force (USPSTF) developed and published evidence-based recommendations for osteoporosis screening to help providers identify and treat osteoporosis early to prevent fractures.² Available evidence on screening and treatment in women and men were reviewed with the intention of updating the 2011 USPSTF recommendations. The review also evaluated risk assessment tools, screening intervals, and efficacy of screening and treatment in various subpopulations.

Since the 2011 recommendations, more data have become available on fracture risk assessment with or without bone mineral density measurements. In its 2018 report, the USPSTF recommends that postmenopausal women younger than 65 should undergo screening with a bone density test if their 10-year risk of major osteoporotic fracture is more than 8.4%. This is equivalent to the fracture risk of a 65-year-old white woman with no major risk factors for fracture (grade B recommendation—high certainty that the benefit is moderate, or moderate certainty that the benefit is moderate to substantial).²

Assessment of fracture risk

For postmenopausal women who are under age 65 and who have at least 1 risk factor for fracture, it is reasonable to use a clinical risk assessment tool to determine who should undergo screening with bone mineral density measurement. Risk factors associated with an increased risk of osteoporotic fractures include a parental history of hip fracture, smoking, intake of 3 or more alcoholic drinks per day, low body weight, malabsorption, rheumatoid arthritis, diabetes, and postmenopausal status (not using estrogen replacement). Medications should be carefully reviewed for those that can increase the risk of fractures, including steroids and antiestrogen treatments.

The 10-year risk of a major osteoporotic or hip fracture can be assessed using the Fractional Risk Assessment Tool (FRAX), available at www.sheffield.ac.uk/FRAX/. Other acceptable tools that perform similarly to FRAX include the Osteoporosis Risk Assessment Instrument (ORAI) (10 studies; N = 16,780), Osteoporosis Index of Risk (OSIRIS) (5 studies; N = 5,649), Osteoporosis Self-Assessment Tool (OST) (13 studies; N = 44,323), and Simple Calculated Osteoporosis Risk Estimation (SCORE) (8 studies; N = 15,362).

Should this patient be screened for osteoporosis?

Based on the FRAX, this patient’s 10-year risk of major osteoporosis fracture is 9.2%. She would benefit from osteoporosis screening with a bone density test.

DO ANTIBIOTICS REDUCE EFFECTIVENESS OF HORMONAL CONTRACEPTION?

A 27-year-old woman presents with a dog bite on her right hand and is started on oral antibiotics. She takes an oral contraceptive that contains 35 µg of ethinyl estradiol and 0.25 mg of norgestimate. She asks if she should use condoms while taking antibiotics.

The antibiotics rifampin and rifabutin are known inducers of the hepatic enzymes required for contraceptive steroid metabolism, whereas other antibiotics are not. Despite the lack of compelling evidence that broad-spectrum antibiotics interfere with the efficacy of hormonal contraception, most pharmacists recommend backup contraception for women who use concomitant antibiotics.³ This practice could lead to poor compliance with the contraceptive regimen, the antibiotic regimen, or both.³

Simmons et al³ conducted a systematic review of randomized and nonrandomized studies that assessed pregnancy rates, breakthrough bleeding, ovulation suppression, and hormone pharmacokinetics in women taking oral or vaginal hormonal contraceptives in combination with nonrifamycin antibiotics, including oral, intramuscular, and intravenous forms. Oral contraceptives used in the studies included a range of doses and progestins, but lowest-dose pills, such as those containing less than 30 µg ethinyl estradiol or less than 150 µg levonorgestrel, were not included.

The contraceptive formulations in this systematic review³ included oral contraceptive pills, emergency contraception pills, and the contraceptive vaginal ring. The effect of antibiotics on other nonoral contraceptives, such as the transdermal patch, injectables, and progestin implants was not studied.

Four observational studies³ evaluated pregnancy rates or hormonal contraception failure with any antibiotic use. In 2 of these 4 studies, there was no difference in pregnancy rates in women who used oral contraceptives with and without nonrifamycin antibiotics. However, ethinyl estradiol was shown to have increased clearance when administered with dirithromycin (a macrolide).³ Twenty-five of the studies reported measures of contraceptive effectiveness (ovulation) and pharmacokinetic outcomes.

There were no observed differences in ovulation suppression or breakthrough bleeding in any study that combined hormonal contraceptives with an antibiotic. Furthermore, there was no significant decrease in progestin pharmacokinetic parameters during coadministration with an antibiotic.³ Study limitations included small sample sizes and the observational nature of the data.

How would you counsel this patient?

Available evidence suggests that nonrifamycin antibiotics do not diminish the effectiveness of the vaginal contraceptive ring or an oral hormonal contraceptive that contains at least 30 µg of ethinyl estradiol or 150 µg of levonorgestrel. Current guidelines do not recommend the use of additional backup contraception, regardless of hormonal contraception dose or formulation.⁴ Likewise, the most recent guidance for dental practitioners (ie, from 2012) no longer advises women to use additional contraceptive protection when taking nonrifamycin antibiotics.⁵

In our practice, we discuss the option of additional protection when prescribing formulations with lower estrogen doses (< 30 µg), not only because of the limitations of the available data, but also because of the high rates of unintended pregnancy with typical use of combined hormonal contraceptives (9% per year, unrelated to use of antibiotics).⁴ However, if our patient would rather not use additional barrier methods, she can be reassured that concomitant nonrifamycin antibiotic use is unlikely to affect contraceptive effectiveness.

A 41-year-old healthy mother of 3 was recently found to be a carrier of the BRCA1 mutation. She is planning to undergo prophylactic bilateral salpingo-oophorectomy for ovarian cancer prevention. However, she is apprehensive about undergoing surgical menopause. Should she be started on hormone replacement therapy after oophorectomy? How would hormone replacement therapy affect her risk of breast cancer?

In females who carry the BRCA1 mutation, the cumulative risk of both ovarian and breast cancer approaches 44% (95% confidence interval [CI] 36%–53%) and 72% (95% CI 65%–79%) by age 80.⁶ Prophylactic salpingo-oophorectomy reduces the risk of breast cancer by 50% and the risk of ovarian cancer by 90%. Unfortunately, premature withdrawal of ovarian hormones has been associated with long-term adverse effects including significant vasomotor symptoms, decreased quality of life, sexual dysfunction, early mortality, bone loss, decline in mood and cognition, and poor cardiovascular outcomes.⁷ Many of these effects can be avoided or lessened with hormone replacement therapy.

Kotsopoulos et al⁸ conducted a longitudinal, prospective analysis of BRCA1 mutation carriers in a multicenter study between 1995 and 2017. The mean follow-up period was 7.6 years (range 0.4–22.1). The study assessed associations between the use of hormone replacement therapy and breast cancer risk in carriers of the BRCA1 mutation who underwent prophylactic salpingo-oophorectomy. Study participants did not have a personal history of cancer. Those with a history of prophylactic mastectomy were excluded.

Participants completed a series of questionnaires every 2 years, disclosing updates in personal medical, cancer, and reproductive history. The questionnaires also inquired about the use of hormone replacement therapy, including the type used (estrogen only, progestin only, estrogen plus progestin, other), brand name, duration of use, and dose and route of administration (pill, patch, suppository).

Of the 13,087 BRCA1 mutation carriers identified, 872 met the study criteria. Of those, 377 (43%) reported using some form of hormone replacement therapy after salpingo-oophorectomy, and 495 (57%) did not. The average duration of use was 3.9 years (range 0.5–19), with most (69%) using estrogen alone; 18% used other regimens, including estrogen plus progestin and progestin only. A small percentage of participants did not indicate which formulation they used. On average, women using hormone replacement therapy underwent prophylactic oophorectomy earlier than nonusers (age 43.0 vs 48.4; absolute difference 5.5 years, P < .001).

During follow-up, there was no significant difference noted in the proportion of women diagnosed with breast cancer between hormone replacement therapy users and nonusers (10.3 vs 10.7%; absolute difference 0.4%; P = .86). In fact, for each year of estrogen-containing hormone replacement therapy, there was an 18% reduction in breast cancer risk when oophorectomy was performed before age 45 (95% CI 0.69–0.97). The authors also noted a nonsignificant 14% trend toward an increase in breast cancer risk for each year of progestin use after oophorectomy when surgery was performed before age 45 (95% CI 0.9–1.46).

Although prophylactic hysterectomy was not recommended, the authors noted that hysterectomy would eliminate the need for progestin-containing hormone replacement therapy. For those who underwent oophorectomy after age 45, hormone replacement therapy did not increase or decrease the risk of breast cancer.⁷

A meta-analysis by Marchetti et al⁹ also supports the safety of hormone replacement therapy after risk-reducing salpingo-oophorectomy. Three studies that included 1,100 patients were analyzed (including the Kotsopoulos study⁸ noted above). There was a nonsignificant decrease in breast cancer risk in women on estrogen-only hormone replacement therapy compared with women on estrogen-plus-progestin therapy (odds ratio 0.53, 95% CI 0.25–1.15). Overall, the authors regarded hormone replacement therapy as a safe therapeutic option after prophylactic salpingo-oophorectomy in carriers of the BRCA1 and BRCA2 mutations.⁹

In a case-control study published in 2016,¹⁰ hormone replacement therapy was assessed in 432 postmenopausal BRCA1 mutation carriers with invasive breast cancer (cases) and in 432 BRCA1 mutation carriers without a history of breast cancer (controls). Results showed no difference in breast cancer risk between hormone replacement therapy users and nonusers.¹⁰

Rebbeck et al¹¹ evaluated short-term hormone replacement therapy in BRCA1 and BRCA2 gene-mutation carriers after they underwent prophylactic salpingo-oophorectomy. The results showed that hormone replacement did not affect the breast cancer risk-reduction conferred with prophylactic bilateral salpingo-oophorectomy.

Johansen et al¹² evaluated hormone replacement therapy in premenopausal women after prophylactic salpingo-oophorectomy. They studied 324 carriers of BRCA gene mutations after they underwent prophylactic salpingo-oophorectomy and a subset of 950 controls who had bilateral salpingo-oophorectomy for reasons unrelated to cancer. In both groups, hormone replacement therapy was underutilized. The authors recommended using it when clinically indicated.

Should your patient start hormone replacement therapy?

This patient is healthy, and in the absence of contraindications, systemic hormone replacement therapy after prophylactic oophorectomy could mitigate the potential adverse effects of surgically induced menopause. The patient can be reassured that estrogen-containing short-term hormone replacement therapy is unlikely to increase her breast cancer risk.

A 44-year-old woman presents to your office for an annual visit. She is sexually active but does not wish to become pregnant. She has a family history of breast cancer: her mother was diagnosed at age 53. She is interested in an oral contraceptive to prevent pregnancy and acne. However, she is nervous about being on any contraceptive that may increase her risk of breast cancer.

To date, studies assessing the effect of hormonal contraception on the risk of breast cancer have produced inconsistent results. Although most studies have shown no associated risk, a few have shown a temporary 20% to 30% increased risk of breast cancer during use.^13,14 Case-controlled studies that reported an association between hormonal contraception and breast cancer included populations taking higher-dose combination pills, which are no longer prescribed. Most studies do not evaluate specific formulations of hormonal contraception, and little is known about effects associated with intrauterine devices or progestin-only contraception.

A prospective study performed by Mørch et al¹³ followed more than 1 million reproductive-aged women for a mean of 10.9 years. The Danish Cancer Registry was used to identify cases of invasive breast cancer. Women who used hormonal contraceptives had a relative risk of breast cancer of 1.20 compared with women not on hormonal contraception (95% CI 1.14–1.26). The study suggested that those who had been on contraceptive agents for more than 5 years had an increased risk and that this risk remained for 5 years after the agents were discontinued. Conversely, no increased risk of cancer was noted in those who used hormonal contraception for less than 5 years. No notable differences were seen among various formulations.

For women using the levonorgestrel-containing intrauterine device, the relative risk of breast cancer was 1.21 (95% CI 1.11–1.33). A few cancers were noted in those who used the progestin-only implant or those using depot medroxyprogesterone acetate. While the study showed an increased relative risk of breast cancer, the absolute risk was low—13 cases per 100,000, or approximately 1 additional case of breast cancer per 7,690 per year.¹³

This study had several important limitations. The authors did not adjust for common breast cancer risk factors including age at menarche, alcohol use, or breastfeeding. Additionally, the study did not account for the use of hormonal contraception before the study period and conversely, did not account for women who may have stopped taking their contraceptive despite their prescribed duration. The frequency of mammography was not explicitly noted, which could have shifted results for women who had more aggressive screening.

It is also noteworthy that the use of high-dose systemic progestins was not associated with an increased risk, whereas the levonorgestrel intrauterine device, which contains only 1/20th the dose of a low-dose oral contraceptive pill, was associated with an increased risk. This discrepancy in risk warrants further investigation, and clinicians should be aware that this inconsistency needs validation before changing clinical practice.

In an observational cohort study,¹⁵ more than 100,000 women ages 50 to 71 were followed prospectively for 15 years to evaluate the association between hormonal contraceptive use and the risk of gynecologic and breast cancers. In this study, the duration of hormonal contraceptive use, smoking status, alcohol use, body mass index, physical activity, and family history of cancer were recorded. Long-term hormonal contraceptive use reduced ovarian and endometrial cancer risks by 40% and 34%, respectively, with no increase in breast cancer risk regardless of family history.

How would you counsel the patient?

The patient should be educated on the benefits of hormonal contraception that extend beyond pregnancy prevention, including regulation of menses, improved acne, decreased risk of endometrial and ovarian cancer, and likely reductions in colorectal cancer and overall mortality risk.^13–16 Further, after their own systematic review of the data assessing risk of breast cancer with hormonal contraception, the US Centers for Disease Control and Prevention state in their guidelines that all contraceptives may be used without limitation in those who have a family history of breast cancer.⁴ Any potential increased risk of breast cancer in women using hormonal contraception is small and would not outweigh the benefits associated with use.

One must consider the impact of an unintended pregnancy in such women, including effects on the health of the fetus and mother. Recent reports on the increasing rates of maternal death in the US (23.8 of 100,000 live births) serve as a reminder of the complications that can arise with pregnancy, especially if a mother’s health is not optimized before conception.¹⁷

The same 44-year-old patient now inquires about screening for breast cancer. She is curious about 3-dimensional mammography and whether it would be a better screening test for her.

Digital breast tomosynthesis (DBT) is a newer imaging modality that provides a 3-dimensional reconstruction of the breast using low-dose x-ray imaging. Some studies have shown that combining DBT with digital mammography may be superior to digital mammography alone in detecting cancers.¹⁸ However, digital mammography is currently the gold standard for breast cancer screening and is the only test proven to reduce mortality.^18,19

In a retrospective US study of 13 medical centers,²⁰ breast cancer detection rates increased by 41% the year after DBT was introduced, from 2.9 to 4.1 per 1,000 cases. DBT was associated with 16 fewer patients recalled for repeat imaging out of 1,000 women screened (as opposed to mammography alone). Two European studies similarly suggested an increase in cancer detection with lower recall rates.^21,22

Is 3-D mammography a better option?

In a 2-arm study by Pattacini et al,¹⁸ nearly 20,000 women ages 45 to 70 were randomized to undergo either digital mammography or digital mammography plus DBT for primary breast cancer screening. Women were enrolled over a 2-year period and were followed for 4.5 years, and the development of a primary invasive cancer was the primary end point. Recall rates, reading times, and radiation doses were also compared between the 2 groups.

Overall, the cancer detection rate was higher in the digital mammography plus DBT arm compared with digital mammography alone (8.6 vs 4.5 per 1,000). The detection rates were higher in the combined screening group among all age subgroups, with relative risks ranging from 1.83 to 2.04 (P = .93). The recall rate was 3.5% in the 2 arms, with relative risks ranging from 0.93 to 1.11 (P = .52). There was a reduction in the number of false positives seen in women undergoing digital mammography plus DBT when compared with digital mammography alone, from 30 per 1,000 to 27 per 1,000.

Detection of ductal carcinoma in situ increased in the experimental arm (relative detection 2.80, 95% CI 1.01–7.65) compared with invasive cancers. Comparing radiation, the dose was 2.3 times higher in those who underwent digital mammography plus DBT. The average reading times for digital mammography alone were 20 to 85 seconds; adding DBT added 35 to 81 seconds.¹⁹

Should you advise 3-D mammography?

The patient should be educated on the benefits of both digital mammography alone and digital mammography plus DBT. The use of digital mammography plus DBT has been supported in various studies and has been shown to increase cancer detection rates, although data are still conflicting regarding recall rates.^19,20 More studies are needed to determine its effect on breast cancer morality.

Routine use of DBT in women with or without dense breast tissue has not been recommended by organizations such as the USPSTF and the American College of Obstetricians and Gynecologists.^23,24 While there is an increased dose of radiation, it still falls below the US Food and Drug Administration limits and should not be the sole barrier to use.

Obstructive sleep apnea (OSA) is an independent risk factor for ischemic stroke and may also, infrequently, be a consequence of stroke. It is significantly underdiagnosed in the general population and is highly prevalent in patients who have had a stroke. Many patients likely had their stroke because of this chronic untreated condition.

This review focuses on OSA and its prevalence, consequences, and treatment in patients after a stroke.

DEFINING AND QUANTIFYING OSA

OSA is the most common type of sleep-disordered breathing.^1,2 It involves repeated narrowing or complete collapse of the upper airway despite ongoing respiratory effort.^3,4 Apneic episodes are terminated by arousals from hypoxemia or efforts to breathe.⁵ In contrast, central sleep apnea is characterized by a patent airway but lack of airflow due to absent respiratory effort.⁵

In OSA, the number of episodes of apnea (absent airflow) and hypopnea (reduced airflow) are added together and divided by hours of sleep to calculate the apnea-hypopnea index (AHI). OSA is diagnosed by either of the following^3,4:

• AHI of 5 or higher, with clinical symptoms related to OSA (described below)
• AHI of 15 or higher, regardless of symptoms.

The AHI also defines OSA severity, as follows³:

• Mild: AHI 5 to 15
• Moderate: AHI 15 to 30
• Severe: AHI greater than 30.

Diagnostic criteria (eg, definition of hypopnea, testing methods, and AHI thresholds) have varied over time, an important consideration when reviewing the literature.

OSA IS MORE COMMON THAN EXPECTED AFTER STROKE

In the most methodologically sound and generalizable study of this topic to date, the Wisconsin Sleep Cohort Study⁶ reported in 2013 that about 14% of men and 5% of women ages 30 to 70 have an AHI greater than 5 (using 4% desaturation to score hypopneic episodes) with daytime sleepiness. Other studies suggest that 80% to 90% of people with OSA are undiagnosed and untreated.^1,7

The prevalence of OSA in patients who have had a stroke is much higher, ranging from 30% to 96% depending on the study methods and population.^1,8–12 A 2010 meta-analysis¹¹ of 29 studies reported that 72% of patients who had a stroke had an AHI greater than 5, and 29% had severe OSA. In this analysis, 7% of those with sleep-disordered breathing had central sleep apnea; still, these data indicate that the prevalence of OSA in these patients is about 5 times higher than in the general population.

RISK FACTORS MAY DIFFER IN STROKE POPULATION

Several risk factors for OSA have been identified.

Obesity is one of the strongest risk factors, with increasing body mass index (BMI) associated with increased OSA prevalence.^4,6,13 However, obesity appears to be a less significant risk factor in patients who have had a stroke than in the general population. In the 2010 meta-analysis¹¹ of OSA after stroke, the average BMI was only 26.4 kg/m² (with obesity defined as a BMI > 30.0 kg/m²), and increasing BMI was not associated with increasing AHI.

Male sex and advanced age are also OSA risk factors.^4,5 They remain significant in patients after a stroke; about 65% of poststroke patients who have OSA are men, and the older the patient, the more likely the AHI is greater than 10.¹¹

Ethnicity and genetics may also play important roles in OSA risk, with roughly 25% of OSA prevalence estimated to have a genetic basis.^14,15 Some risk factors for OSA such as craniofacial shape, upper airway anatomy, upper airway muscle dysfunction, increased respiratory chemosensitivity, and poor arousal threshold during sleep are likely determined by genetics and ethnicity.^14,15 Compared with people of European origin, Asians have a similar prevalence of OSA, but at a much lower average BMI, suggesting that other factors are significant.¹⁴ Possible genetically determined anatomic risk factors have not been specifically studied in the poststroke population, but it can be assumed they remain relevant.

Several studies have tried to find an association between OSA and type, location, etiology, or pattern of stroke.^{10,11,16–19} Although some suggest links between cardioembolic stroke and OSA,^16,20 or thrombolysis and OSA,¹⁰ most have found no association between OSA and stroke features.^11,12,21,22

HOW DOES OSA INCREASE STROKE RISK?

Untreated severe OSA is associated with increased cardiovascular mortality,^21,22 and OSA is an independent risk factor for incident stroke.²³ A number of mechanisms may explain these relationships.

Intermittent hypoxemia and recurrent sympathetic arousals resulting from OSA are thought to lead to many of the comorbid conditions with which it is associated: hypertension, coronary artery disease, heart failure, arrhythmias, pulmonary hypertension, and stroke. Repetitive decreases in ventilation lead to oxygen desaturations that result in cycles of increased sympathetic outflow and eventual sustained nocturnal hypertension and daytime chronic hypertension.^1,5,9,13 Also implicated are various changes in vasodilator and vasoconstrictor substances due to endothelial dysfunction and inflammation, which are thought to play a role in the atherogenic and prothrombotic states induced by OSA.^1,5,13

Cerebral circulation is altered primarily by the changes in partial pressure of carbon dioxide (Pco₂). During apnea, the Pco₂ rises, causing vasodilation and increased blood flow. After the apnea resolves, there is hyperpnea with resultant decreased Pco₂, and vasoconstriction. In a patient who already has vascular disease, the enhanced vasoconstriction could lead to ischemia.^1,5

Changes in intrathoracic pressure result in distortion of cardiac architecture. When the patient tries to breathe against an occluded airway, the intrathoracic pressure becomes more and more negative, increasing preload and afterload. When this happens repeatedly every night for years, it leads to remodeling of the heart such as left and right ventricular hypertrophy, with reduced stroke volume, myocardial ischemia, and increased risk of arrhythmia.^1,5,13

Untreated OSA is believed to predispose patients to develop atrial fibrillation through sympathetic overactivity, vascular inflammation, heart rate variability, and cardiac remodeling.²⁴ As atrial fibrillation is a major risk factor for stroke, particularly cardioembolic stroke, it may be another pathway of increased stroke risk in OSA.^16,20,25

OSA typically causes both daytime symptoms (excessive sleepiness, poor concentration, morning headache, depressive symptoms) and nighttime signs and symptoms (snoring, choking, gasping, night sweats, insomnia, nocturia, witnessed episodes of apnea).^3,4,26 Unfortunately, because these are nonspecific, OSA is often underdiagnosed.^4,26

Identifying OSA after a stroke may be a particular challenge, as patients often do not report classic symptoms, and the typical picture of OSA may have less predictive validity in these patients.^1,27,28 Within the first 24 hours after a stroke, hypersomnia, snoring history, and age are not predictive of OSA.¹ Patients found to have OSA after a stroke frequently do not have the traditional symptoms (sleepiness, snoring) seen in usual OSA patients. And they have higher rates of OSA at a younger age than the usual OSA patients, so age is not a predictive risk factor. In addition, daytime sleepiness and obesity are often absent or less prominent.^1,9,27,28  Finally, typical OSA signs and symptoms may be attributed to the stroke itself or to comorbidities affecting the patient, lowering suspicion for OSA.

OSA MAY HINDER STROKE RECOVERY, WORSEN OUTCOMES

OSA, particularly when moderate to severe, is linked to pathophysiologic changes that can hinder recovery from a stroke.

Intermittent hypoxemia during sleep can worsen vascular damage of at-risk tissue: nocturnal hypoxemia correlates with white matter hyperintensities on magnetic resonance imaging, a marker of ischemic demyelination.²⁹ Oxidative stress and release of inflammatory mediators associated with intermittent hypoxemia may impair vascular blood flow to brain tissue attempting to repair itself.³⁰ In addition, sympathetic overactivity and Pco₂ fluctuations associated with OSA may impede cerebral circulation.

Taken together, such ongoing nocturnal insults can lead to clinical consequences during this vulnerable period.

A 1996 study³¹ of patients recovering from a stroke found that an oxygen desaturation index (number of times that the blood oxygen level drops below a certain threshold, as measured by overnight oximetry) of more than 10 per hour was associated with worse functional recovery at discharge and at 3 and 12 months after discharge. This study also noted an association between time spent with oxygen saturations below 90% and the rate of death at 1 year.

A 2003 study³² reported that patients with an AHI greater than 10 by polysomnography spent an average of 13 days longer on the rehabilitation service and had worse functional and cognitive status on discharge, even after controlling for multiple confounders. Several subsequent studies have confirmed these and similar findings.^8,33,34

OSA has also been linked to depression,³⁵ which is common after stroke and may worsen outcomes.³⁶ The interaction between OSA, depression, and poststroke outcomes warrants further study.

In the general population, OSA has been independently associated with increased risk of stroke or death from any cause.^21,22,37 These associations have also been reported in the poststroke population: a 2014 meta-analysis found that OSA increased the risk of a repeat stroke (relative risk [RR] 1.8, 95% confidence interval [CI] 1.2–2.6) and all-cause mortality (RR 1.69, 95% CI 1.4–2.1).³⁸

TESTING FOR OSA AFTER STROKE

Because of the high prevalence of OSA in patients who have had a stroke and the potential for worse outcomes associated with untreated OSA, there should be a low threshold for evaluating for OSA soon after stroke. Objective testing is required to qualify for therapy,  and the gold standard for diagnosis of OSA is formal polysomnography conducted in a sleep laboratory.^2–4 Unfortunately, polysomnography may be unacceptable to some patients, is costly, and is resource-intensive, particularly in an inpatient or rehabilitation setting.²⁸ Ideally, to optimize testing efficiency, patients should be screened for the likelihood of OSA before polysomnography is ordered.

Questionnaires can help determine the need for further testing

Questionnaires developed to assess OSA risk³⁹ include the following:

The Berlin questionnaire, developed in 1999, has 10 questions assessing daytime and nighttime signs and symptoms and presence of hypertension.

The STOP questionnaire, developed in 2008, assesses snoring, tiredness, observed apneic episodes, and elevated blood pressure.

The STOP-BANG questionnaire, published in 2010, includes the STOP questions plus BMI over 35 kg/m², age over 50, neck circumference over 41 cm, and male gender.

A 2017 meta-analysis³⁹ of 108 studies with nearly 50,000 people found that the STOP-BANG questionnaire performed best with regard to sensitivity and diagnostic odds ratio, but with poor specificity.

These screening tools and modified versions of them have also been evaluated in patients who have had a stroke.

In 2015, Boulos et al²⁸ found that the STOP-BAG (a version of STOP-BANG that excludes neck circumference) and the 4-variable (4V) questionnaire (sex, BMI, blood pressure, snoring) had moderate predictive value for OSA within 6 months after sroke.

In 2016, Katzan et al⁴⁰ found that the STOP-BAG2 (STOP-BAG criteria plus continuous variables for BMI and age) had a high sensitivity for polysomnographically diagnosed OSA within the first year after a stroke. The specificity was significantly better than the STOP-BANG or the STOP-BAG questionnaire, although it remained suboptimal at 60.5%.

In 2017, Sico et al⁴¹ developed and assessed the SLEEP Inventory (sex, left heart failure, Epworth Sleepiness Scale, enlarged neck, weight in pounds, insulin resistance or diabetes, and National Institutes of Health Stroke Scale) and found that it outperformed the Berlin and STOP-BANG questionnaires in the poststroke setting. The SLEEP Inventory had the best specificity and negative predictive value, and a slightly better ability to correctly classify patients as having OSA or not, classifying 80% of patients correctly.

These newer screening tools (eg, STOP-BAG, STOP-BAG2, SLEEP) can be used to identify with reasonable accuracy which patients need definitive testing after stroke.

Pulse oximetry is another possible screening tool          

Overnight pulse oximetry may also help screen for sleep apnea and stratify risk after a stroke. A 2012 study⁴² of overnight oximetry to screen patients before surgery found that the oxygen desaturation index was significantly associated with the AHI measured by polysomnography. However, oximetry testing cannot distinguish between OSA and central sleep apnea, so it is insufficient to diagnose OSA or qualify patients for therapy. Further study is needed to examine the ability of overnight pulse oximetry to screen or to stratify risk for OSA after stroke.

Polysomnography vs home testing

Polysomnography is the gold standard for diagnosing OSA. Benefits include technical support and trouble-shooting, determining relationships between OSA, body position, and sleep stage, and the ability to intervene with treatment.² However, polysomnography can be cumbersome, costly, and resource-intensive.

A home sleep apnea test, ie, an unattended, limited-channel sleep study, may be an acceptable alternative.^2–4,43,44 Home testing does not require a sleep technologist to be present during testing, uses fewer sensors, and is less expensive than overnight polysomnography, but its utility can be limited: it fails to accurately discriminate between episodes of OSA and central sleep apnea, there is potential for false-negative results, and it can underestimate sleep apnea burden because it does not measure sleep.²

Institutional resources and logistics may influence the choice of diagnostic modality. No data exist on outcomes from different diagnostic testing methods in poststroke patients. Further research is needed.

The first-line treatment for OSA is positive airway pressure (PAP).³ For most patients, this is continuous PAP (CPAP) or autoadjusting PAP (APAP). In some instances, particularly for those who cannot tolerate CPAP or who have comorbid hypoventilation, bilevel PAP (BPAP) may be indicated. More advanced PAP therapies are unlikely to be used after stroke.

PAP therapy is associated with reduced daytime sleepiness, improved mood, normalization of sleep architecture, improved systemic and pulmonary artery blood pressure, reduced rates of atrial fibrillation after ablation, and improved insulin sensitivity.^45–49 Whether it reduces the risk of cardiovascular events, including stroke, remains controversial; most data suggest that it does not.^50,51 However, when adherence to PAP therapy is considered rather than intention to treat, treatment has been found to lead to improved cardiovascular outcomes.⁵²

Mixed evidence of benefits after stroke

Observational studies provide evidence that CPAP may help patients with OSA after stroke, although results are mixed.^53–58 The studies ranged in size from 14 to 105 patients, enrolled patients with mostly moderate to severe OSA, and followed patients from 10 days to 7 years. Adherence to therapy was generally good in the short term (50%–70%), but only  15% to 30% of patients remained adherent at 5 to 7 years. Variable outcomes were reported, with some studies finding improved symptoms in the near term and mixed evidence of cardiovascular benefit in the longer ones. However, as these studies lacked randomization, drawing definitive conclusions on CPAP efficacy is difficult.

Patients were enrolled in the index admission or when starting a rehabilitation service—generally 2 to 3 weeks after their stroke. No clear association was found between the timing of initiating PAP therapy and outcomes. All patients had ischemic strokes, but few details were provided regarding stroke location, size, and severity. Exclusion criteria included severe underlying cardiopulmonary disease, confusion, severe stroke with marked impairment, and inability to cooperate. Almost all patients had moderate to severe OSA, and patients with central sleep apnea were excluded.

The major outcomes examined were drop-out rates, PAP adherence, and neurologic improvement based on neurologic functional scales (National Institutes of Health Stroke Scale and Canadian Neurologic Scale). As expected, dropout rates were higher in patients randomized to CPAP (OR 1.83, 95% CI 1.05–3.21, P = .03), although overall adherence was better than anticipated, with mean CPAP use across trials of 4.5 hours per night (95% CI 3.97–5.08) and with about 50% to 60% of patients adhering to therapy for at least 4 hours nightly.

Improvement in neurologic outcomes favored CPAP (standard mean difference 0.54, 95% CI 0.026–1.05), although considerable heterogeneity was seen. Improved sleepiness outcomes were inconsistent. Major cardiovascular outcomes were reported in only 2 studies (using the same data set) and showed delayed time to the next cardiovascular event for those treated with CPAP but no difference in cardiovascular event-free survival.

PAP poses more challenges after stroke

The primary limitation to PAP therapy is poor acceptance and adherence to therapy.⁵⁹ High rates of refusal of therapy and difficulty complying with treatment have been noted in the poststroke population, although recent studies have reported better adherence rates. How rates of adherence play out in real-world settings, outside of the controlled environment of a research study, has yet to be determined.

In general, CPAP adherence is affected by claustrophobia, difficulty tolerating a mask, problems with pressure intolerance, irritating air leaks, nasal congestion, and naso-oral dryness. Many such barriers can be overcome with use of a properly fitted mask, an appropriate pressure setting, heated humidification, nasal sprays (eg, saline, inhaled steroids), and education, encouragement, and reassurance.

After a stroke, additional obstacles may impede the ability to use PAP therapy.⁶⁸ Facial paresis (hemi- or bifacial) may make fitting of the mask problematic. Paralysis or weakness of the extremities may limit the ability to adjust or remove a mask. Aphasia can impair communication and understanding of the need to use PAP therapy, and upper-airway problems related to stroke, including dysphagia, may lead to pressure intolerance or risk of aspiration. Finally, a lack of perceived benefit, particularly if the patient does not have daytime sleepiness, may limit motivation.

Consider alternatives

For patients unlikely to succeed with PAP therapy, there are alternatives. Surgery and oral appliances are not usually realistic options in the setting of recent stroke, but positional therapy, including the use of body positioners to prevent supine sleep, as well as elevating the head of the bed, may be of some benefit.^69,70 A nasopharyngeal airway stenting device (nasal trumpet) may also be tolerated by some patients.

A proposed algorithm for screening, diagnosing, and treating OSA in patients after stroke is presented in Figure 1.

Obstructive sleep apnea (OSA) is an independent risk factor for ischemic stroke and may also, infrequently, be a consequence of stroke. It is significantly underdiagnosed in the general population and is highly prevalent in patients who have had a stroke. Many patients likely had their stroke because of this chronic untreated condition.

This review focuses on OSA and its prevalence, consequences, and treatment in patients after a stroke.

DEFINING AND QUANTIFYING OSA

OSA is the most common type of sleep-disordered breathing.^1,2 It involves repeated narrowing or complete collapse of the upper airway despite ongoing respiratory effort.^3,4 Apneic episodes are terminated by arousals from hypoxemia or efforts to breathe.⁵ In contrast, central sleep apnea is characterized by a patent airway but lack of airflow due to absent respiratory effort.⁵

In OSA, the number of episodes of apnea (absent airflow) and hypopnea (reduced airflow) are added together and divided by hours of sleep to calculate the apnea-hypopnea index (AHI). OSA is diagnosed by either of the following^3,4:

• AHI of 5 or higher, with clinical symptoms related to OSA (described below)
• AHI of 15 or higher, regardless of symptoms.

The AHI also defines OSA severity, as follows³:

• Mild: AHI 5 to 15
• Moderate: AHI 15 to 30
• Severe: AHI greater than 30.

Diagnostic criteria (eg, definition of hypopnea, testing methods, and AHI thresholds) have varied over time, an important consideration when reviewing the literature.

OSA IS MORE COMMON THAN EXPECTED AFTER STROKE

In the most methodologically sound and generalizable study of this topic to date, the Wisconsin Sleep Cohort Study⁶ reported in 2013 that about 14% of men and 5% of women ages 30 to 70 have an AHI greater than 5 (using 4% desaturation to score hypopneic episodes) with daytime sleepiness. Other studies suggest that 80% to 90% of people with OSA are undiagnosed and untreated.^1,7

The prevalence of OSA in patients who have had a stroke is much higher, ranging from 30% to 96% depending on the study methods and population.^1,8–12 A 2010 meta-analysis¹¹ of 29 studies reported that 72% of patients who had a stroke had an AHI greater than 5, and 29% had severe OSA. In this analysis, 7% of those with sleep-disordered breathing had central sleep apnea; still, these data indicate that the prevalence of OSA in these patients is about 5 times higher than in the general population.

RISK FACTORS MAY DIFFER IN STROKE POPULATION

Several risk factors for OSA have been identified.

Obesity is one of the strongest risk factors, with increasing body mass index (BMI) associated with increased OSA prevalence.^4,6,13 However, obesity appears to be a less significant risk factor in patients who have had a stroke than in the general population. In the 2010 meta-analysis¹¹ of OSA after stroke, the average BMI was only 26.4 kg/m² (with obesity defined as a BMI > 30.0 kg/m²), and increasing BMI was not associated with increasing AHI.

Male sex and advanced age are also OSA risk factors.^4,5 They remain significant in patients after a stroke; about 65% of poststroke patients who have OSA are men, and the older the patient, the more likely the AHI is greater than 10.¹¹

Ethnicity and genetics may also play important roles in OSA risk, with roughly 25% of OSA prevalence estimated to have a genetic basis.^14,15 Some risk factors for OSA such as craniofacial shape, upper airway anatomy, upper airway muscle dysfunction, increased respiratory chemosensitivity, and poor arousal threshold during sleep are likely determined by genetics and ethnicity.^14,15 Compared with people of European origin, Asians have a similar prevalence of OSA, but at a much lower average BMI, suggesting that other factors are significant.¹⁴ Possible genetically determined anatomic risk factors have not been specifically studied in the poststroke population, but it can be assumed they remain relevant.

Several studies have tried to find an association between OSA and type, location, etiology, or pattern of stroke.^{10,11,16–19} Although some suggest links between cardioembolic stroke and OSA,^16,20 or thrombolysis and OSA,¹⁰ most have found no association between OSA and stroke features.^11,12,21,22

HOW DOES OSA INCREASE STROKE RISK?

Untreated severe OSA is associated with increased cardiovascular mortality,^21,22 and OSA is an independent risk factor for incident stroke.²³ A number of mechanisms may explain these relationships.

Intermittent hypoxemia and recurrent sympathetic arousals resulting from OSA are thought to lead to many of the comorbid conditions with which it is associated: hypertension, coronary artery disease, heart failure, arrhythmias, pulmonary hypertension, and stroke. Repetitive decreases in ventilation lead to oxygen desaturations that result in cycles of increased sympathetic outflow and eventual sustained nocturnal hypertension and daytime chronic hypertension.^1,5,9,13 Also implicated are various changes in vasodilator and vasoconstrictor substances due to endothelial dysfunction and inflammation, which are thought to play a role in the atherogenic and prothrombotic states induced by OSA.^1,5,13

Cerebral circulation is altered primarily by the changes in partial pressure of carbon dioxide (Pco₂). During apnea, the Pco₂ rises, causing vasodilation and increased blood flow. After the apnea resolves, there is hyperpnea with resultant decreased Pco₂, and vasoconstriction. In a patient who already has vascular disease, the enhanced vasoconstriction could lead to ischemia.^1,5

Changes in intrathoracic pressure result in distortion of cardiac architecture. When the patient tries to breathe against an occluded airway, the intrathoracic pressure becomes more and more negative, increasing preload and afterload. When this happens repeatedly every night for years, it leads to remodeling of the heart such as left and right ventricular hypertrophy, with reduced stroke volume, myocardial ischemia, and increased risk of arrhythmia.^1,5,13

Untreated OSA is believed to predispose patients to develop atrial fibrillation through sympathetic overactivity, vascular inflammation, heart rate variability, and cardiac remodeling.²⁴ As atrial fibrillation is a major risk factor for stroke, particularly cardioembolic stroke, it may be another pathway of increased stroke risk in OSA.^16,20,25

OSA typically causes both daytime symptoms (excessive sleepiness, poor concentration, morning headache, depressive symptoms) and nighttime signs and symptoms (snoring, choking, gasping, night sweats, insomnia, nocturia, witnessed episodes of apnea).^3,4,26 Unfortunately, because these are nonspecific, OSA is often underdiagnosed.^4,26

Identifying OSA after a stroke may be a particular challenge, as patients often do not report classic symptoms, and the typical picture of OSA may have less predictive validity in these patients.^1,27,28 Within the first 24 hours after a stroke, hypersomnia, snoring history, and age are not predictive of OSA.¹ Patients found to have OSA after a stroke frequently do not have the traditional symptoms (sleepiness, snoring) seen in usual OSA patients. And they have higher rates of OSA at a younger age than the usual OSA patients, so age is not a predictive risk factor. In addition, daytime sleepiness and obesity are often absent or less prominent.^1,9,27,28  Finally, typical OSA signs and symptoms may be attributed to the stroke itself or to comorbidities affecting the patient, lowering suspicion for OSA.

OSA MAY HINDER STROKE RECOVERY, WORSEN OUTCOMES

OSA, particularly when moderate to severe, is linked to pathophysiologic changes that can hinder recovery from a stroke.

Intermittent hypoxemia during sleep can worsen vascular damage of at-risk tissue: nocturnal hypoxemia correlates with white matter hyperintensities on magnetic resonance imaging, a marker of ischemic demyelination.²⁹ Oxidative stress and release of inflammatory mediators associated with intermittent hypoxemia may impair vascular blood flow to brain tissue attempting to repair itself.³⁰ In addition, sympathetic overactivity and Pco₂ fluctuations associated with OSA may impede cerebral circulation.

Taken together, such ongoing nocturnal insults can lead to clinical consequences during this vulnerable period.

A 1996 study³¹ of patients recovering from a stroke found that an oxygen desaturation index (number of times that the blood oxygen level drops below a certain threshold, as measured by overnight oximetry) of more than 10 per hour was associated with worse functional recovery at discharge and at 3 and 12 months after discharge. This study also noted an association between time spent with oxygen saturations below 90% and the rate of death at 1 year.

A 2003 study³² reported that patients with an AHI greater than 10 by polysomnography spent an average of 13 days longer on the rehabilitation service and had worse functional and cognitive status on discharge, even after controlling for multiple confounders. Several subsequent studies have confirmed these and similar findings.^8,33,34

OSA has also been linked to depression,³⁵ which is common after stroke and may worsen outcomes.³⁶ The interaction between OSA, depression, and poststroke outcomes warrants further study.

In the general population, OSA has been independently associated with increased risk of stroke or death from any cause.^21,22,37 These associations have also been reported in the poststroke population: a 2014 meta-analysis found that OSA increased the risk of a repeat stroke (relative risk [RR] 1.8, 95% confidence interval [CI] 1.2–2.6) and all-cause mortality (RR 1.69, 95% CI 1.4–2.1).³⁸

TESTING FOR OSA AFTER STROKE

Because of the high prevalence of OSA in patients who have had a stroke and the potential for worse outcomes associated with untreated OSA, there should be a low threshold for evaluating for OSA soon after stroke. Objective testing is required to qualify for therapy,  and the gold standard for diagnosis of OSA is formal polysomnography conducted in a sleep laboratory.^2–4 Unfortunately, polysomnography may be unacceptable to some patients, is costly, and is resource-intensive, particularly in an inpatient or rehabilitation setting.²⁸ Ideally, to optimize testing efficiency, patients should be screened for the likelihood of OSA before polysomnography is ordered.

Questionnaires can help determine the need for further testing

Questionnaires developed to assess OSA risk³⁹ include the following:

The Berlin questionnaire, developed in 1999, has 10 questions assessing daytime and nighttime signs and symptoms and presence of hypertension.

The STOP questionnaire, developed in 2008, assesses snoring, tiredness, observed apneic episodes, and elevated blood pressure.

The STOP-BANG questionnaire, published in 2010, includes the STOP questions plus BMI over 35 kg/m², age over 50, neck circumference over 41 cm, and male gender.

A 2017 meta-analysis³⁹ of 108 studies with nearly 50,000 people found that the STOP-BANG questionnaire performed best with regard to sensitivity and diagnostic odds ratio, but with poor specificity.

These screening tools and modified versions of them have also been evaluated in patients who have had a stroke.

In 2015, Boulos et al²⁸ found that the STOP-BAG (a version of STOP-BANG that excludes neck circumference) and the 4-variable (4V) questionnaire (sex, BMI, blood pressure, snoring) had moderate predictive value for OSA within 6 months after sroke.

In 2016, Katzan et al⁴⁰ found that the STOP-BAG2 (STOP-BAG criteria plus continuous variables for BMI and age) had a high sensitivity for polysomnographically diagnosed OSA within the first year after a stroke. The specificity was significantly better than the STOP-BANG or the STOP-BAG questionnaire, although it remained suboptimal at 60.5%.

In 2017, Sico et al⁴¹ developed and assessed the SLEEP Inventory (sex, left heart failure, Epworth Sleepiness Scale, enlarged neck, weight in pounds, insulin resistance or diabetes, and National Institutes of Health Stroke Scale) and found that it outperformed the Berlin and STOP-BANG questionnaires in the poststroke setting. The SLEEP Inventory had the best specificity and negative predictive value, and a slightly better ability to correctly classify patients as having OSA or not, classifying 80% of patients correctly.

These newer screening tools (eg, STOP-BAG, STOP-BAG2, SLEEP) can be used to identify with reasonable accuracy which patients need definitive testing after stroke.

Pulse oximetry is another possible screening tool          

Overnight pulse oximetry may also help screen for sleep apnea and stratify risk after a stroke. A 2012 study⁴² of overnight oximetry to screen patients before surgery found that the oxygen desaturation index was significantly associated with the AHI measured by polysomnography. However, oximetry testing cannot distinguish between OSA and central sleep apnea, so it is insufficient to diagnose OSA or qualify patients for therapy. Further study is needed to examine the ability of overnight pulse oximetry to screen or to stratify risk for OSA after stroke.

Polysomnography vs home testing

Polysomnography is the gold standard for diagnosing OSA. Benefits include technical support and trouble-shooting, determining relationships between OSA, body position, and sleep stage, and the ability to intervene with treatment.² However, polysomnography can be cumbersome, costly, and resource-intensive.

A home sleep apnea test, ie, an unattended, limited-channel sleep study, may be an acceptable alternative.^2–4,43,44 Home testing does not require a sleep technologist to be present during testing, uses fewer sensors, and is less expensive than overnight polysomnography, but its utility can be limited: it fails to accurately discriminate between episodes of OSA and central sleep apnea, there is potential for false-negative results, and it can underestimate sleep apnea burden because it does not measure sleep.²

Institutional resources and logistics may influence the choice of diagnostic modality. No data exist on outcomes from different diagnostic testing methods in poststroke patients. Further research is needed.

The first-line treatment for OSA is positive airway pressure (PAP).³ For most patients, this is continuous PAP (CPAP) or autoadjusting PAP (APAP). In some instances, particularly for those who cannot tolerate CPAP or who have comorbid hypoventilation, bilevel PAP (BPAP) may be indicated. More advanced PAP therapies are unlikely to be used after stroke.

PAP therapy is associated with reduced daytime sleepiness, improved mood, normalization of sleep architecture, improved systemic and pulmonary artery blood pressure, reduced rates of atrial fibrillation after ablation, and improved insulin sensitivity.^45–49 Whether it reduces the risk of cardiovascular events, including stroke, remains controversial; most data suggest that it does not.^50,51 However, when adherence to PAP therapy is considered rather than intention to treat, treatment has been found to lead to improved cardiovascular outcomes.⁵²

Mixed evidence of benefits after stroke

Observational studies provide evidence that CPAP may help patients with OSA after stroke, although results are mixed.^53–58 The studies ranged in size from 14 to 105 patients, enrolled patients with mostly moderate to severe OSA, and followed patients from 10 days to 7 years. Adherence to therapy was generally good in the short term (50%–70%), but only  15% to 30% of patients remained adherent at 5 to 7 years. Variable outcomes were reported, with some studies finding improved symptoms in the near term and mixed evidence of cardiovascular benefit in the longer ones. However, as these studies lacked randomization, drawing definitive conclusions on CPAP efficacy is difficult.

Patients were enrolled in the index admission or when starting a rehabilitation service—generally 2 to 3 weeks after their stroke. No clear association was found between the timing of initiating PAP therapy and outcomes. All patients had ischemic strokes, but few details were provided regarding stroke location, size, and severity. Exclusion criteria included severe underlying cardiopulmonary disease, confusion, severe stroke with marked impairment, and inability to cooperate. Almost all patients had moderate to severe OSA, and patients with central sleep apnea were excluded.

The major outcomes examined were drop-out rates, PAP adherence, and neurologic improvement based on neurologic functional scales (National Institutes of Health Stroke Scale and Canadian Neurologic Scale). As expected, dropout rates were higher in patients randomized to CPAP (OR 1.83, 95% CI 1.05–3.21, P = .03), although overall adherence was better than anticipated, with mean CPAP use across trials of 4.5 hours per night (95% CI 3.97–5.08) and with about 50% to 60% of patients adhering to therapy for at least 4 hours nightly.

Improvement in neurologic outcomes favored CPAP (standard mean difference 0.54, 95% CI 0.026–1.05), although considerable heterogeneity was seen. Improved sleepiness outcomes were inconsistent. Major cardiovascular outcomes were reported in only 2 studies (using the same data set) and showed delayed time to the next cardiovascular event for those treated with CPAP but no difference in cardiovascular event-free survival.

PAP poses more challenges after stroke

The primary limitation to PAP therapy is poor acceptance and adherence to therapy.⁵⁹ High rates of refusal of therapy and difficulty complying with treatment have been noted in the poststroke population, although recent studies have reported better adherence rates. How rates of adherence play out in real-world settings, outside of the controlled environment of a research study, has yet to be determined.

In general, CPAP adherence is affected by claustrophobia, difficulty tolerating a mask, problems with pressure intolerance, irritating air leaks, nasal congestion, and naso-oral dryness. Many such barriers can be overcome with use of a properly fitted mask, an appropriate pressure setting, heated humidification, nasal sprays (eg, saline, inhaled steroids), and education, encouragement, and reassurance.

After a stroke, additional obstacles may impede the ability to use PAP therapy.⁶⁸ Facial paresis (hemi- or bifacial) may make fitting of the mask problematic. Paralysis or weakness of the extremities may limit the ability to adjust or remove a mask. Aphasia can impair communication and understanding of the need to use PAP therapy, and upper-airway problems related to stroke, including dysphagia, may lead to pressure intolerance or risk of aspiration. Finally, a lack of perceived benefit, particularly if the patient does not have daytime sleepiness, may limit motivation.

Consider alternatives

For patients unlikely to succeed with PAP therapy, there are alternatives. Surgery and oral appliances are not usually realistic options in the setting of recent stroke, but positional therapy, including the use of body positioners to prevent supine sleep, as well as elevating the head of the bed, may be of some benefit.^69,70 A nasopharyngeal airway stenting device (nasal trumpet) may also be tolerated by some patients.

A proposed algorithm for screening, diagnosing, and treating OSA in patients after stroke is presented in Figure 1.

Obstructive sleep apnea (OSA) is an independent risk factor for ischemic stroke and may also, infrequently, be a consequence of stroke. It is significantly underdiagnosed in the general population and is highly prevalent in patients who have had a stroke. Many patients likely had their stroke because of this chronic untreated condition.

This review focuses on OSA and its prevalence, consequences, and treatment in patients after a stroke.

DEFINING AND QUANTIFYING OSA

OSA is the most common type of sleep-disordered breathing.^1,2 It involves repeated narrowing or complete collapse of the upper airway despite ongoing respiratory effort.^3,4 Apneic episodes are terminated by arousals from hypoxemia or efforts to breathe.⁵ In contrast, central sleep apnea is characterized by a patent airway but lack of airflow due to absent respiratory effort.⁵

In OSA, the number of episodes of apnea (absent airflow) and hypopnea (reduced airflow) are added together and divided by hours of sleep to calculate the apnea-hypopnea index (AHI). OSA is diagnosed by either of the following^3,4:

• AHI of 5 or higher, with clinical symptoms related to OSA (described below)
• AHI of 15 or higher, regardless of symptoms.

The AHI also defines OSA severity, as follows³:

• Mild: AHI 5 to 15
• Moderate: AHI 15 to 30
• Severe: AHI greater than 30.

Diagnostic criteria (eg, definition of hypopnea, testing methods, and AHI thresholds) have varied over time, an important consideration when reviewing the literature.

OSA IS MORE COMMON THAN EXPECTED AFTER STROKE

In the most methodologically sound and generalizable study of this topic to date, the Wisconsin Sleep Cohort Study⁶ reported in 2013 that about 14% of men and 5% of women ages 30 to 70 have an AHI greater than 5 (using 4% desaturation to score hypopneic episodes) with daytime sleepiness. Other studies suggest that 80% to 90% of people with OSA are undiagnosed and untreated.^1,7

The prevalence of OSA in patients who have had a stroke is much higher, ranging from 30% to 96% depending on the study methods and population.^1,8–12 A 2010 meta-analysis¹¹ of 29 studies reported that 72% of patients who had a stroke had an AHI greater than 5, and 29% had severe OSA. In this analysis, 7% of those with sleep-disordered breathing had central sleep apnea; still, these data indicate that the prevalence of OSA in these patients is about 5 times higher than in the general population.

RISK FACTORS MAY DIFFER IN STROKE POPULATION

Several risk factors for OSA have been identified.

Obesity is one of the strongest risk factors, with increasing body mass index (BMI) associated with increased OSA prevalence.^4,6,13 However, obesity appears to be a less significant risk factor in patients who have had a stroke than in the general population. In the 2010 meta-analysis¹¹ of OSA after stroke, the average BMI was only 26.4 kg/m² (with obesity defined as a BMI > 30.0 kg/m²), and increasing BMI was not associated with increasing AHI.

Male sex and advanced age are also OSA risk factors.^4,5 They remain significant in patients after a stroke; about 65% of poststroke patients who have OSA are men, and the older the patient, the more likely the AHI is greater than 10.¹¹

Ethnicity and genetics may also play important roles in OSA risk, with roughly 25% of OSA prevalence estimated to have a genetic basis.^14,15 Some risk factors for OSA such as craniofacial shape, upper airway anatomy, upper airway muscle dysfunction, increased respiratory chemosensitivity, and poor arousal threshold during sleep are likely determined by genetics and ethnicity.^14,15 Compared with people of European origin, Asians have a similar prevalence of OSA, but at a much lower average BMI, suggesting that other factors are significant.¹⁴ Possible genetically determined anatomic risk factors have not been specifically studied in the poststroke population, but it can be assumed they remain relevant.

Several studies have tried to find an association between OSA and type, location, etiology, or pattern of stroke.^{10,11,16–19} Although some suggest links between cardioembolic stroke and OSA,^16,20 or thrombolysis and OSA,¹⁰ most have found no association between OSA and stroke features.^11,12,21,22

HOW DOES OSA INCREASE STROKE RISK?

Untreated severe OSA is associated with increased cardiovascular mortality,^21,22 and OSA is an independent risk factor for incident stroke.²³ A number of mechanisms may explain these relationships.

Intermittent hypoxemia and recurrent sympathetic arousals resulting from OSA are thought to lead to many of the comorbid conditions with which it is associated: hypertension, coronary artery disease, heart failure, arrhythmias, pulmonary hypertension, and stroke. Repetitive decreases in ventilation lead to oxygen desaturations that result in cycles of increased sympathetic outflow and eventual sustained nocturnal hypertension and daytime chronic hypertension.^1,5,9,13 Also implicated are various changes in vasodilator and vasoconstrictor substances due to endothelial dysfunction and inflammation, which are thought to play a role in the atherogenic and prothrombotic states induced by OSA.^1,5,13

Cerebral circulation is altered primarily by the changes in partial pressure of carbon dioxide (Pco₂). During apnea, the Pco₂ rises, causing vasodilation and increased blood flow. After the apnea resolves, there is hyperpnea with resultant decreased Pco₂, and vasoconstriction. In a patient who already has vascular disease, the enhanced vasoconstriction could lead to ischemia.^1,5

Changes in intrathoracic pressure result in distortion of cardiac architecture. When the patient tries to breathe against an occluded airway, the intrathoracic pressure becomes more and more negative, increasing preload and afterload. When this happens repeatedly every night for years, it leads to remodeling of the heart such as left and right ventricular hypertrophy, with reduced stroke volume, myocardial ischemia, and increased risk of arrhythmia.^1,5,13

Untreated OSA is believed to predispose patients to develop atrial fibrillation through sympathetic overactivity, vascular inflammation, heart rate variability, and cardiac remodeling.²⁴ As atrial fibrillation is a major risk factor for stroke, particularly cardioembolic stroke, it may be another pathway of increased stroke risk in OSA.^16,20,25

OSA typically causes both daytime symptoms (excessive sleepiness, poor concentration, morning headache, depressive symptoms) and nighttime signs and symptoms (snoring, choking, gasping, night sweats, insomnia, nocturia, witnessed episodes of apnea).^3,4,26 Unfortunately, because these are nonspecific, OSA is often underdiagnosed.^4,26

Identifying OSA after a stroke may be a particular challenge, as patients often do not report classic symptoms, and the typical picture of OSA may have less predictive validity in these patients.^1,27,28 Within the first 24 hours after a stroke, hypersomnia, snoring history, and age are not predictive of OSA.¹ Patients found to have OSA after a stroke frequently do not have the traditional symptoms (sleepiness, snoring) seen in usual OSA patients. And they have higher rates of OSA at a younger age than the usual OSA patients, so age is not a predictive risk factor. In addition, daytime sleepiness and obesity are often absent or less prominent.^1,9,27,28  Finally, typical OSA signs and symptoms may be attributed to the stroke itself or to comorbidities affecting the patient, lowering suspicion for OSA.

OSA MAY HINDER STROKE RECOVERY, WORSEN OUTCOMES

OSA, particularly when moderate to severe, is linked to pathophysiologic changes that can hinder recovery from a stroke.

Intermittent hypoxemia during sleep can worsen vascular damage of at-risk tissue: nocturnal hypoxemia correlates with white matter hyperintensities on magnetic resonance imaging, a marker of ischemic demyelination.²⁹ Oxidative stress and release of inflammatory mediators associated with intermittent hypoxemia may impair vascular blood flow to brain tissue attempting to repair itself.³⁰ In addition, sympathetic overactivity and Pco₂ fluctuations associated with OSA may impede cerebral circulation.

Taken together, such ongoing nocturnal insults can lead to clinical consequences during this vulnerable period.

A 1996 study³¹ of patients recovering from a stroke found that an oxygen desaturation index (number of times that the blood oxygen level drops below a certain threshold, as measured by overnight oximetry) of more than 10 per hour was associated with worse functional recovery at discharge and at 3 and 12 months after discharge. This study also noted an association between time spent with oxygen saturations below 90% and the rate of death at 1 year.

A 2003 study³² reported that patients with an AHI greater than 10 by polysomnography spent an average of 13 days longer on the rehabilitation service and had worse functional and cognitive status on discharge, even after controlling for multiple confounders. Several subsequent studies have confirmed these and similar findings.^8,33,34

OSA has also been linked to depression,³⁵ which is common after stroke and may worsen outcomes.³⁶ The interaction between OSA, depression, and poststroke outcomes warrants further study.

In the general population, OSA has been independently associated with increased risk of stroke or death from any cause.^21,22,37 These associations have also been reported in the poststroke population: a 2014 meta-analysis found that OSA increased the risk of a repeat stroke (relative risk [RR] 1.8, 95% confidence interval [CI] 1.2–2.6) and all-cause mortality (RR 1.69, 95% CI 1.4–2.1).³⁸

TESTING FOR OSA AFTER STROKE

Because of the high prevalence of OSA in patients who have had a stroke and the potential for worse outcomes associated with untreated OSA, there should be a low threshold for evaluating for OSA soon after stroke. Objective testing is required to qualify for therapy,  and the gold standard for diagnosis of OSA is formal polysomnography conducted in a sleep laboratory.^2–4 Unfortunately, polysomnography may be unacceptable to some patients, is costly, and is resource-intensive, particularly in an inpatient or rehabilitation setting.²⁸ Ideally, to optimize testing efficiency, patients should be screened for the likelihood of OSA before polysomnography is ordered.

Questionnaires can help determine the need for further testing

Questionnaires developed to assess OSA risk³⁹ include the following:

The Berlin questionnaire, developed in 1999, has 10 questions assessing daytime and nighttime signs and symptoms and presence of hypertension.

The STOP questionnaire, developed in 2008, assesses snoring, tiredness, observed apneic episodes, and elevated blood pressure.

The STOP-BANG questionnaire, published in 2010, includes the STOP questions plus BMI over 35 kg/m², age over 50, neck circumference over 41 cm, and male gender.

A 2017 meta-analysis³⁹ of 108 studies with nearly 50,000 people found that the STOP-BANG questionnaire performed best with regard to sensitivity and diagnostic odds ratio, but with poor specificity.

These screening tools and modified versions of them have also been evaluated in patients who have had a stroke.

In 2015, Boulos et al²⁸ found that the STOP-BAG (a version of STOP-BANG that excludes neck circumference) and the 4-variable (4V) questionnaire (sex, BMI, blood pressure, snoring) had moderate predictive value for OSA within 6 months after sroke.

In 2016, Katzan et al⁴⁰ found that the STOP-BAG2 (STOP-BAG criteria plus continuous variables for BMI and age) had a high sensitivity for polysomnographically diagnosed OSA within the first year after a stroke. The specificity was significantly better than the STOP-BANG or the STOP-BAG questionnaire, although it remained suboptimal at 60.5%.

In 2017, Sico et al⁴¹ developed and assessed the SLEEP Inventory (sex, left heart failure, Epworth Sleepiness Scale, enlarged neck, weight in pounds, insulin resistance or diabetes, and National Institutes of Health Stroke Scale) and found that it outperformed the Berlin and STOP-BANG questionnaires in the poststroke setting. The SLEEP Inventory had the best specificity and negative predictive value, and a slightly better ability to correctly classify patients as having OSA or not, classifying 80% of patients correctly.

These newer screening tools (eg, STOP-BAG, STOP-BAG2, SLEEP) can be used to identify with reasonable accuracy which patients need definitive testing after stroke.

Pulse oximetry is another possible screening tool          

Overnight pulse oximetry may also help screen for sleep apnea and stratify risk after a stroke. A 2012 study⁴² of overnight oximetry to screen patients before surgery found that the oxygen desaturation index was significantly associated with the AHI measured by polysomnography. However, oximetry testing cannot distinguish between OSA and central sleep apnea, so it is insufficient to diagnose OSA or qualify patients for therapy. Further study is needed to examine the ability of overnight pulse oximetry to screen or to stratify risk for OSA after stroke.

Polysomnography vs home testing

Polysomnography is the gold standard for diagnosing OSA. Benefits include technical support and trouble-shooting, determining relationships between OSA, body position, and sleep stage, and the ability to intervene with treatment.² However, polysomnography can be cumbersome, costly, and resource-intensive.

A home sleep apnea test, ie, an unattended, limited-channel sleep study, may be an acceptable alternative.^2–4,43,44 Home testing does not require a sleep technologist to be present during testing, uses fewer sensors, and is less expensive than overnight polysomnography, but its utility can be limited: it fails to accurately discriminate between episodes of OSA and central sleep apnea, there is potential for false-negative results, and it can underestimate sleep apnea burden because it does not measure sleep.²

Institutional resources and logistics may influence the choice of diagnostic modality. No data exist on outcomes from different diagnostic testing methods in poststroke patients. Further research is needed.

The first-line treatment for OSA is positive airway pressure (PAP).³ For most patients, this is continuous PAP (CPAP) or autoadjusting PAP (APAP). In some instances, particularly for those who cannot tolerate CPAP or who have comorbid hypoventilation, bilevel PAP (BPAP) may be indicated. More advanced PAP therapies are unlikely to be used after stroke.

PAP therapy is associated with reduced daytime sleepiness, improved mood, normalization of sleep architecture, improved systemic and pulmonary artery blood pressure, reduced rates of atrial fibrillation after ablation, and improved insulin sensitivity.^45–49 Whether it reduces the risk of cardiovascular events, including stroke, remains controversial; most data suggest that it does not.^50,51 However, when adherence to PAP therapy is considered rather than intention to treat, treatment has been found to lead to improved cardiovascular outcomes.⁵²

Mixed evidence of benefits after stroke

Observational studies provide evidence that CPAP may help patients with OSA after stroke, although results are mixed.^53–58 The studies ranged in size from 14 to 105 patients, enrolled patients with mostly moderate to severe OSA, and followed patients from 10 days to 7 years. Adherence to therapy was generally good in the short term (50%–70%), but only  15% to 30% of patients remained adherent at 5 to 7 years. Variable outcomes were reported, with some studies finding improved symptoms in the near term and mixed evidence of cardiovascular benefit in the longer ones. However, as these studies lacked randomization, drawing definitive conclusions on CPAP efficacy is difficult.

Patients were enrolled in the index admission or when starting a rehabilitation service—generally 2 to 3 weeks after their stroke. No clear association was found between the timing of initiating PAP therapy and outcomes. All patients had ischemic strokes, but few details were provided regarding stroke location, size, and severity. Exclusion criteria included severe underlying cardiopulmonary disease, confusion, severe stroke with marked impairment, and inability to cooperate. Almost all patients had moderate to severe OSA, and patients with central sleep apnea were excluded.

The major outcomes examined were drop-out rates, PAP adherence, and neurologic improvement based on neurologic functional scales (National Institutes of Health Stroke Scale and Canadian Neurologic Scale). As expected, dropout rates were higher in patients randomized to CPAP (OR 1.83, 95% CI 1.05–3.21, P = .03), although overall adherence was better than anticipated, with mean CPAP use across trials of 4.5 hours per night (95% CI 3.97–5.08) and with about 50% to 60% of patients adhering to therapy for at least 4 hours nightly.

Improvement in neurologic outcomes favored CPAP (standard mean difference 0.54, 95% CI 0.026–1.05), although considerable heterogeneity was seen. Improved sleepiness outcomes were inconsistent. Major cardiovascular outcomes were reported in only 2 studies (using the same data set) and showed delayed time to the next cardiovascular event for those treated with CPAP but no difference in cardiovascular event-free survival.

PAP poses more challenges after stroke

The primary limitation to PAP therapy is poor acceptance and adherence to therapy.⁵⁹ High rates of refusal of therapy and difficulty complying with treatment have been noted in the poststroke population, although recent studies have reported better adherence rates. How rates of adherence play out in real-world settings, outside of the controlled environment of a research study, has yet to be determined.

In general, CPAP adherence is affected by claustrophobia, difficulty tolerating a mask, problems with pressure intolerance, irritating air leaks, nasal congestion, and naso-oral dryness. Many such barriers can be overcome with use of a properly fitted mask, an appropriate pressure setting, heated humidification, nasal sprays (eg, saline, inhaled steroids), and education, encouragement, and reassurance.

After a stroke, additional obstacles may impede the ability to use PAP therapy.⁶⁸ Facial paresis (hemi- or bifacial) may make fitting of the mask problematic. Paralysis or weakness of the extremities may limit the ability to adjust or remove a mask. Aphasia can impair communication and understanding of the need to use PAP therapy, and upper-airway problems related to stroke, including dysphagia, may lead to pressure intolerance or risk of aspiration. Finally, a lack of perceived benefit, particularly if the patient does not have daytime sleepiness, may limit motivation.

Consider alternatives

For patients unlikely to succeed with PAP therapy, there are alternatives. Surgery and oral appliances are not usually realistic options in the setting of recent stroke, but positional therapy, including the use of body positioners to prevent supine sleep, as well as elevating the head of the bed, may be of some benefit.^69,70 A nasopharyngeal airway stenting device (nasal trumpet) may also be tolerated by some patients.

A proposed algorithm for screening, diagnosing, and treating OSA in patients after stroke is presented in Figure 1.

Should clinicians abandon the activated partial thromboplastin time (aPTT) for monitoring heparin therapy in favor of tests that measure the activity of the patient’s plasma against activated factor X (anti-Xa assays)?

Although other anticoagulants are now available for preventing and treating arterial and venous thromboembolism, unfractionated heparin—which requires laboratory monitoring of therapy—is still widely used. And this monitoring can be challenging. Despite its wide use, the aPTT lacks standardization, and the role of alternative monitoring assays such as the anti-Xa assay is not well defined.

This article reviews the advantages, limitations, and clinical applicability of anti-Xa assays for monitoring therapy with unfractionated heparin and other anticoagulants.

UNFRACTIONATED HEPARIN AND WARFARIN ARE STILL WIDELY USED

Until the mid-1990s, unfractionated heparin and oral vitamin K antagonists (eg, warfarin) were the only anticoagulants widely available for clinical use. These agents have complex pharmacokinetic and pharmacodynamic properties, resulting in highly variable dosing requirements (both between patients and in individual patients) and narrow therapeutic windows, making frequent laboratory monitoring and dose adjustments mandatory.

Over the past 3 decades, other anticoagulants have been approved, including low-molecular-weight heparins, fondaparinux, parenteral direct thrombin inhibitors, and direct oral anticoagulants. While these agents have expanded the options for preventing and treating thromboembolism, unfractionated heparin and warfarin are still the most appropriate choices for many patients, eg, those with stage 4 chronic kidney disease and end-stage renal disease on dialysis, and those with mechanical heart valves.

In addition, unfractionated heparin remains the anticoagulant of choice during procedures such as hemodialysis, percutaneous transluminal angioplasty, and cardiopulmonary bypass, as well as in hospitalized and critically ill patients, who often have acute kidney injury or require frequent interruptions of therapy for invasive procedures. In these scenarios, unfractionated heparin is typically preferred because of its short plasma half-life, complete reversibility by protamine, safety regardless of renal function, and low cost compared with parenteral direct thrombin inhibitors.

As long as unfractionated heparin and warfarin remain important therapies, the need for their laboratory monitoring continues. For warfarin monitoring, the prothrombin time and international normalized ratio are validated and widely reproducible methods. But monitoring unfractionated heparin therapy remains a challenge.

UNFRACTIONATED HEPARIN’S EFFECT IS UNPREDICTABLE

Unfractionated heparin, a negatively charged mucopolysaccharide, inhibits coagulation by binding to antithrombin through the high-affinity pentasaccharide sequence.^1–6 Such binding induces a conformational change in the antithrombin molecule, converting it to a rapid inhibitor of several coagulation proteins, especially factors IIa and Xa.^2–4

Unfractionated heparin inhibits factors IIa and Xa in a 1:1 ratio, but low-molecular-weight heparins inhibit factor Xa more than factor IIa, with IIa-Xa inhibition ratios ranging from 1:2 to 1:4, owing to their smaller molecular size.⁷

One of the most important reasons for the unpredictable and highly variable individual responses to unfractionated heparin is that, infused into the blood, the large and negatively charged unfractionated heparin molecules bind nonspecifically to positively charged plasma proteins.⁷ In patients who are critically ill, have acute infections or inflammatory states, or have undergone major surgery, unfractionated heparin binds to acute-phase proteins that are elevated, particularly factor VIII. This results in fewer free heparin molecules and a variable anticoagulant effect.⁸

In contrast, low-molecular-weight heparins have longer half-lives and bind less to plasma proteins, resulting in more predictable plasma levels following subcutaneous injection.⁹

In 1960, Barritt and Jordan¹⁰ conducted a small but landmark trial that established the clinical importance of unfractionated heparin for treating venous thromboembolism. None of the patients who received unfractionated heparin for acute pulmonary embolism developed a recurrence during the subsequent 2 weeks, while 50% of those who did not receive it had recurrent pulmonary embolism, fatal in half of the cases.

The importance of achieving a specific aPTT therapeutic target was not demonstrated until a 1972 study by Basu et al,¹¹ in which 162 patients with venous thromboembolism were treated with heparin with a target aPTT of 1.5 to 2.5 times the control value. Patients who suffered recurrent events had subtherapeutic aPTT values on 71% of treatment days, while the rest of the patients, with no recurrences, had subtherapeutic aPTT values only 28% of treatment days. The different outcomes could not be explained by the average daily dose of unfractionated heparin, which was similar in the patients regardless of recurrence.

Subsequent studies showed that the best outcomes occur when unfractionated heparin is given in doses high enough to rapidly achieve a therapeutic prolongation of the aPTT,^12–14 and that the total daily dose is also important in preventing recurrences.^15,16 Failure to achieve a target aPTT within 24 hours of starting unfractionated heparin is associated with increased risk of recurrent venous thromboembolism.^13,17

Raschke et al¹⁷ found that patients prospectively randomized to weight-based doses of intravenous unfractionated heparin (bolus plus infusion) achieved significantly higher rates of therapeutic aPTT within 6 hours and 24 hours after starting the infusion, and had significantly lower rates of recurrent venous thromboembolism than those randomized to a fixed unfractionated heparin protocol, without an increase in major bleeding.

Smith et al,¹⁸ in a study of 400 consecutive patients with acute pulmonary embolism treated with unfractionated heparin, found that patients who achieved a therapeutic aPTT within 24 hours had lower in-hospital and 30-day mortality rates than those who did not achieve the first therapeutic aPTT until more than 24 hours after starting unfractionated heparin infusion.

Such data lend support to the widely accepted practice and current guideline recommendation⁸ of using laboratory assays to adjust the dose of unfractionated heparin to achieve and maintain a therapeutic target. The use of dosing nomograms significantly reduces the time to achieve a therapeutic aPTT while minimizing subtherapeutic and supratherapeutic unfractionated heparin levels.^19,20

THE aPTT REFLECTS THROMBIN INHIBITION

The aPTT has a log-linear relationship with plasma concentrations of unfractionated heparin,²¹ but it was not developed specifically for monitoring unfractionated heparin therapy. Originally described in 1953 as a screening tool for hemophilia,^22–24 the aPTT is prolonged in the setting of factor deficiencies (typically with levels < 45%, except for factors VII and XIII), as well as lupus anticoagulants and therapy with parenteral direct thrombin inhibitors.^8,25,26

Because thrombin (factor IIa) is 10 times more sensitive than factor Xa to inhibition by the heparin-antithrombin complex,^4,7 thrombin inhibition appears to be the most likely mechanism by which unfractionated heparin prolongs the aPTT. In contrast, aPTT is minimally or not at all prolonged by low-molecular-weight heparins, which are predominantly factor Xa inhibitors.⁷

HEPARIN ASSAYS MEASURE UNFRACTIONATED HEPARIN ACTIVITY

While the aPTT is a surrogate marker of unfractionated heparin activity in plasma, unfractionated heparin activity can be measured more precisely by so-called heparin assays, which are typically not direct measures of the plasma concentration of heparins, but rather functional assays that provide indirect estimates. They include protamine sulfate titration assays and anti-Xa assays.

Protamine sulfate titration assays measure the amount of protamine sulfate required to neutralize heparin: the more protamine required, the greater the estimated concentration of unfractionated heparin in plasma.^8,27–29 Protamine titration assays are technically demanding, so they are rarely used clinically.

Anti-Xa assays provide a measure of the functional level of heparins in plasma.^29–33 Chromogenic anti-Xa assays are available on automated analyzers with standardized kits^29,33,34 and may be faster to perform than the aPTT.³⁵

Experiments in rabbits show that unfractionated heparin inhibits thrombus formation and extension at concentrations of 0.2 to 0.4 U/mL as measured by the protamine titration assay,²⁷ which correlated with an anti-Xa activity of 0.35 to 0.67 U/mL in a randomized controlled trial.³²

Assays that directly measure the plasma concentration of heparin exist but are not clinically relevant because they also measure heparin molecules lacking the pentasaccharide sequence, which have no anticoagulant activity.³⁶

Anti-Xa assays are more expensive than the aPTT and are not available in all hospitals. For these reasons, the aPTT remains the most commonly used laboratory assay for monitoring unfractionated heparin therapy.

However, the aPTT correlates poorly with the activity level of unfractionated heparin in plasma. In one study, an anti-Xa level of 0.3 U/mL corresponded to aPTT results ranging from 47 to 108 seconds.³¹ Furthermore, in studies that used a heparin therapeutic target based on an aPTT ratio 1.5 to 2.5 times the control aPTT value, the lower end of that target range was often associated with subtherapeutic plasma unfractionated heparin activity measured by anti-Xa and protamine titration assays.^28,31

Because of these limitations, individual laboratories should determine their own aPTT therapeutic target ranges for unfractionated heparin based on the response curves obtained with the reagent and coagulometer used. The optimal therapeutic aPTT range for treating acute venous thromboembolism should be defined as the aPTT range (in seconds) that correlates with a plasma activity level of unfractionated heparin of 0.3 to 0.7 U/mL based on a chromogenic anti-Xa assay, or 0.2 to 0.4 U/mL based on a protamine titration assay.^32,34–36

Nevertheless, the anticoagulant effect of unfractionated heparin as measured by the aPTT can be unpredictable and can vary widely among individuals and in the same patient.⁷ This wide variability can be explained by a number of technical and biologic variables. Different commercial aPTT reagents, different lots of the same reagent, and different reagent and instrument combinations have different sensitivities to unfractionated heparin, which can lead to variable aPTT results.³⁷ Moreover, high plasma levels of acute-phase proteins, low plasma antithrombin levels, consumptive coagulopathies, liver failure, and lupus anticoagulants may also affect the aPTT.^{7,25,32,36–41} These variables account for the poor correlation—ranging from 25% to 66%—reported between aPTT and anti-Xa assays.^32,42–48

Such discrepancies may have serious clinical implications: if a patient’s aPTT is low (subtherapeutic) or high (supratherapeutic) but the anti-Xa assay result is within the therapeutic range (0.3–0.7 units/mL), changing the dose of unfractionated heparin (guided by an aPTT nomogram) may increase the risk of bleeding or of recurrent thromboembolism.

CLINICAL APPLICABILITY OF THE ANTI-Xa ASSAY

Neither anti-Xa nor protamine titration assays are standardized across reference laboratories, but chromogenic anti-Xa assays have better interlaboratory correlation than the aPTT^49,50 and can be calibrated specifically for unfractionated or low-molecular-weight heparins.^29,33

Although reagent costs are higher for chromogenic anti-Xa assays than for the aPTT, some technical variables (described below) may partially offset the cost difference.^29,33,41 In addition, unlike the aPTT, anti-Xa assays do not need local calibration; the therapeutic range for unfractionated heparin is the same (0.3–0.7 U/mL) regardless of instrument or reagent.^33,41

Most important, studies have found that patients monitored by anti-Xa assay achieve significantly higher rates of therapeutic anticoagulation within 24 and 48 hours after starting unfractionated heparin infusion than those monitored by the aPTT. Fewer dose adjustments and repeat tests are required, which may also result in lower cost.^32,51–55

While these studies found chromogenic anti-Xa assays better for achieving laboratory end points, data regarding relevant clinical outcomes are more limited. In a retrospective, observational cohort study,⁵¹ the rate of venous thromboembolism or bleeding-related death was 2% in patients receiving unfractionated heparin therapy monitored by anti-Xa assay and 6% in patients monitored by aPTT (P = .62). Rates of major hemorrhage were also not significantly different.

In a randomized controlled trial³² in 131 patients with acute venous thromboembolism and heparin resistance, rates of recurrent venous thromboembolism were 4.6% and 6.1% in the groups randomized to anti-Xa and aPTT monitoring, respectively, whereas overall bleeding rates were 1.5% and 6.1%, respectively. Again, the differences were not statistically significant.

Heparin resistance. Some patients require unusually high doses of unfractionated heparin to achieve a therapeutic aPTT: typically, more than 35,000 U over 24 hours,^7,8,32 or total daily doses that exceed their estimated weight-based requirements. Heparin resistance has been observed in various clinical settings.^{7,8,32,37–40,59–61} Patients with heparin resistance monitored by anti-Xa had similar rates of recurrent venous thromboembolism while receiving significantly lower doses of unfractionated heparin than those monitored by the aPTT.³²

Lupus anticoagulant. Patients with the specific antiphospholipid antibody known as lupus anticoagulant frequently have a prolonged baseline aPTT,²⁵ making it an unreliable marker of anticoagulant effect for intravenous unfractionated heparin therapy.

Critically ill infants and children. Arachchillage et al³⁵ found that infants (< 1 year old) treated with intravenous unfractionated heparin in an intensive care department had only a 32.4% correlation between aPTT and anti-Xa levels, which was lower than that found in children ages 1 to 15 (66%) and adults (52%). In two-thirds of cases of discordant aPTT and anti-Xa levels, the aPTT was elevated (supratherapeutic) while the anti-Xa assay was within the therapeutic range (0.3–0.7 U/mL). Despite the lack of data on clinical outcomes (eg, rates of thrombosis and bleeding) with the use of an anti-Xa assay, it has been considered the method of choice for unfractionated heparin monitoring in critically ill children, and especially in those under age 1.^{41,44,62–64}

While anti-Xa assays may also be better for unfractionated heparin monitoring in critically ill adults, the lack of clinical outcome data from large-scale randomized trials has precluded evidence-based recommendations favoring them over the aPTT.^8,34

Anti-Xa assays are hampered by some technical limitations:

Samples must be processed within 1 hour to avoid heparin neutralization.³⁴

Samples must be clear. Hemolyzed or opaque samples (eg, due to bilirubin levels > 6.6 mg/dL or triglyceride levels > 360 mg/dL) cannot be processed, as they can cause falsely low levels.

Exposure to other anticoagulants can interfere with the results. The anti-Xa assay may be unreliable for unfractionated heparin monitoring in patients who are transitioned from low-molecular-weight heparins, fondaparinux, or an oral factor Xa inhibitor (apixaban, betrixaban, edoxaban, rivaroxaban) to intravenous unfractionated heparin, eg, due to hospitalization or acute kidney injury.^65,66 Different reports have found that anti-Xa assays may be elevated for as long as 63 to 96 hours after the last dose of oral Xa inhibitors,^67–69 potentially resulting in underdosing of unfractionated heparin. In such settings, unfractionated heparin therapy should be monitored by the aPTT.

ANTI-Xa ASSAYS AND LOW-MOLECULAR-WEIGHT HEPARINS

Most patients receiving low-molecular-weight heparins do not need laboratory monitoring.⁸ Alhenc-Gelas et al⁷⁰ randomized patients to receive dalteparin in doses either based on weight or guided by anti-Xa assay results, and found that dose adjustments were rare and lacked clinical benefit.

The suggested therapeutic anti-Xa levels for low-molecular-weight heparins are:

• 0.5–1.2 U/mL for twice-daily enoxaparin
• 1.0–2.0 U/mL for once-daily enoxaparin or dalteparin.

Levels should be measured at peak plasma level (ie, 3–4 hours after subcutaneous injection, except during pregnancy, when it is 4–6 hours), and only after at least 3 doses of low-molecular-weight heparin.^8,71 Unlike the anti-Xa therapeutic range recommended for unfractionated heparin therapy, these ranges are not based on prospective data, and if the assay result is outside the suggested therapeutic target range, current guidelines offer no advice on safely adjusting the dose.^8,71

Measuring anti-Xa activity is particularly important for pregnant women with a mechanical prosthetic heart valve who are treated with low-molecular-weight heparins. In this setting, valve thrombosis and cardioembolic events have been reported in patients with peak low-molecular-weight heparin anti-Xa assay levels below or even at the lower end of the therapeutic range, and increased bleeding risk has been reported with elevated anti-Xa levels.^71–74 Measuring trough low-molecular-weight heparin anti-Xa levels has been suggested to guide dose adjustments during pregnancy.⁷⁵

Clearance of low-molecular-weight heparins as measured by the anti-Xa assay is highly correlated with creatinine clearance.^76,77 A strong linear correlation has been demonstrated between creatine clearance and anti-Xa levels of enoxaparin after multiple therapeutic doses, and low-molecular-weight heparins accumulate in the plasma, especially in patients with creatine clearance less than 30 mL/min.⁷⁸ The risk of major bleeding is significantly increased in patients with severe renal insufficiency (creatinine clearance < 30 mL/min) not on dialysis who are treated with either prophylactic or therapeutic doses of low-molecular-weight heparin.^79–81 In a meta-analysis, the risk of bleeding with therapeutic-intensity doses of enoxaparin was 4 times higher than with prophylactic-intensity doses.⁷⁹ Although bleeding risk appears to be reduced when the enoxaparin dose is reduced by 50%,⁸ the efficacy and safety of this strategy has not been determined by prospective trials.

ANTI-Xa ASSAYS IN PATIENTS RECEIVING DIRECT ORAL ANTICOAGULANTS

Direct oral factor Xa inhibitors cannot be measured accurately by heparin anti-Xa assays. Nevertheless, such assays may be useful to assess whether clinically relevant plasma levels are present in cases of major bleeding, suspected anticoagulant failure, or patient noncompliance.⁸²

Intense research has focused on developing drug-specific chromogenic anti-Xa assays using calibrators and standards for apixaban, edoxaban, and rivaroxaban,^82,83 and good linear correlation has been shown with some assays.^82,84 In patients treated with oral factor Xa inhibitors who need to undergo an urgent invasive procedure associated with high bleeding risk, use of a specific reversal agent may be considered with drug concentrations more than 30 ng/mL measured by a drug-specific anti-Xa assay. A similar suggestion has been made for drug concentrations more than 50 ng/mL in the setting of major bleeding.⁸⁵ Unfortunately, such assays are not widely available at this time.^82,86

While drug-specific anti-Xa assays could become clinically important to guide reversal strategies, their relevance for drug monitoring remains uncertain. This is because no therapeutic target ranges have been established for any of the direct oral anticoagulants, which were approved on the basis of favorable clinical trial outcomes that neither measured nor were correlated with specific drug levels in plasma. Therefore, a specific anti-Xa level cannot yet be used as a marker of clinical efficacy for any specific oral direct Xa inhibitor.