General internists need to be able to recognize community-acquired pneumonia (CAP) so that diagnostic and therapeutic interventions can be initiated promptly. It is also important to understand the most likely and possible causes of CAP so that appropriate initial antimicrobial therapy can be chosen. Especially during flu season, influenza can present as CAP and should be included in the differential diagnosis.

When managing a patient with CAP, the internist must decide which level of care, diagnostic tests, antimicrobial agents, and follow-up plans are needed. These topics will be reviewed in this article.

TWO TERMS TO REMEMBER

• CAP refers to pneumonia acquired outside a health care facility. It can be either bacterial or viral.
• CABP (community-acquired bacterial pneumonia) refers only to those cases caused by bacterial pathogens.

NUMBERS AND TRENDS

In the United States, CAP is the number-one cause of death from infection and the sixth leading cause of death overall.¹ Each year, it is responsible for about 4.2 million outpatient visits, more than 60,000 deaths, and more than $17 billion in health care expenses.²

Community-acquired bacterial pneumonia: Common, serious

In a population-based US study in 1991, the incidence of CABP requiring hospitalization was 266.8 per 100,000 people.³

Estimates of overall mortality in CABP vary depending on the severity of illness and comorbid conditions. A meta-analysis published in 1996 found the overall mortality rate to be 13.7%, with a range of 5.1% to 36.5% depending on severity.⁴

In hospitalized patients, mortality rates and length of hospital stay appear to be declining over time. Between 1993 and 2005, the age-adjusted mortality rate decreased from 8.9% to 4.1%, and the average length of stay decreased from 7.5 to 5.7 days, with an overall reduction in hospital cost.⁵

CABP is more prevalent in older people than in the general population, and it increases with age from 18.2 cases per 1,000 patient-years in patients 60 to 69 years to 52.3 cases per 1,000 patient-years in those older than 85 years.⁶ Risk factors for pneumonia in the elderly include heart disease, chronic lung disease, immunosuppressive drugs, alcoholism, and increasing age.⁷ Similar to the trend in the general population, the mortality rate in elderly CABP patients appears to be decreasing over time, possibly thanks to rising rates of pneumococcal and influenza vaccination.⁸

Among the general population, risk factors for developing CABP also include smoking, occupational dust exposure, history of childhood pneumonia, unemployment, and single marital status.⁹ The incidence of CABP does not appear to be higher among pregnant women, although it is the most frequent cause of nonobstetric death in this population.¹⁰

The use of proton pump inhibitors may be an emerging risk factor for CABP.¹¹ Also, use of nonsteroidal anti-inflammatory drugs among patients with CABP is associated with a blunted inflammatory response as well as a higher risk of pleuropulmonary complications and a delay in presentation.¹²

Influenza is also common, potentially severe

Influenza is also very common and potentially severe. It can cause a spectrum of disease, from mild upper respiratory tract symptoms to severe viral pneumonia that can be life-threatening and complicated by respiratory failure and the acute respiratory distress syndrome (ARDS).

Influenza infection can also be complicated by subsequent bacterial pneumonia. However, the epidemiology of influenza infection differs from that of CABP in that influenza occurs seasonally.

In the United States, seasonal influenza causes 36,000 deaths and 200,000 hospitalizations annually.^13,14 As with CABP, the risk of death from influenza increases with age: it is 16 times greater in people age 85 and older than in those ages 65 to 69.¹³

During yearly seasonal epidemics, those at the highest risk of hospitalization and death are at the extremes of age. Risk factors for complicated influenza include heart disease, lung disease, diabetes, renal failure, rheumatologic conditions, dementia, and neurologic disease.^15,16 During the 2009 H1N1 influenza pandemic, unexpected severity was seen in previously healthy young adults as well as those with obesity, neurodegenerative disease, pregnancy, and asthma.¹⁷

PATHOGENS: TYPICAL, ATYPICAL, VIRAL

Identifying the etiologic organism in CAP is confounded by limitations in the available diagnostic tests and also by poor-quality specimens that often are contaminated with bacteria that colonize the upper airways. Given these caveats, the primary pathogens responsible for CAP broadly include typical bacterial pathogens, atypical bacterial pathogens, and viruses.

Typical bacterial pathogens include Streptococcus pneumoniae, Haemophilus influenzae, Staphylococcus aureus, Moraxella catarrhalis, and, less commonly, a variety of aerobic and anaerobic gram-negative rods including Pseudomonas aeruginosa, Acinetobacter species, and Klebsiella pneumoniae.

Atypical bacterial pathogens include Mycoplasma pneumoniae, Chlamydia pneumoniae, and Legionella species.¹⁸

Viruses implicated in adult CAP include influenza A and B, parainfluenza viruses, respiratory syncytial virus, and adenovirus.¹⁹ More recently, human metapneumovirus has been described as a cause of adult CAP.²⁰

Clues to uncommon microbes

Specific historic features or coexisting conditions that may suggest an uncommon microbiologic diagnosis include²¹:

• Recent travel to the southwestern United States or Southeast Asia
• Ill contacts
• Exposure to birds, bats, rabbits, or farm animals
• Alcoholism
• Chronic obstructive pulmonary disease
• Human immunodeficiency virus infection
• Structural lung disease
• Prolonged cough with whoop or posttussive vomiting
• Aspiration
• Bioterrorism.

In cases in which one or more of these conditions exist, CAP may also be caused by other agents not listed above, including Mycobacterium tuberculosis, oral anaerobes, atypical mycobacteria, Histoplasma capsulatum, Chlamydophila psittaci, Francisella tularensis, Coxiella burnettii, Pneumocystis jiroveci, Cryptococcus, Aspergillus, Coccidioides, Hantavirus, avian influenza, Burkholderia pseudomallei, severe acute respiratory syndrome virus, Bordetella pertussis, Bacillus anthracis, and Yersinia pestis.

 

HOW BACTERIA INVADE THE LUNGS

The pathophysiology of CABP involves both host defense and microbial virulence factors.

The airways are most commonly exposed to microbes by microaspiration of upper airway flora, although hematogenous seeding of the lungs in a bacteremic patient or contiguous spread of infection from an adjacent site can also occur.

Mucociliary clearance and the cough reflex are important initial defenses against infection and can be inhibited by neurologic diseases and conditions that impair the mucociliary mechanism. Mucosal immune cells, including macrophages and neutrophils, recognize invading pathogens and generate an antibody response.

Regulation of the host inflammatory response to infection depends on a complex interaction between immune cells, inflammatory cytokines (eg, tumor necrosis factor alpha, interleukin 1-beta, and interleukin 6), and anti-inflammatory cytokines such as interleukin 1 receptor antagonist and soluble tumor necrosis factor receptor type I.²²

The interaction and timing of the inflammatory and anti-inflammatory response are essential in manifesting an appropriate host response to infection. An inadequate inflammatory response can lead to sepsis and death, but an excessive, late anti-inflammatory response can lead to a systemic inflammatory response such as ARDS. Polymorphisms within the genes coding for these factors may explain the variation in severity of illness among patients with CABP.²³

HOW INFLUENZA DOES ITS DAMAGE

There are three types of influenza virus: A, B, and C. Type A causes most human infections. The influenza A virus envelope comprises a lipid bilayer that contains the projecting glycoproteins hemagglutinin and neuraminidase. Influenza viruses are named on the basis of these proteins and are designated with an H and an N, respectively, each followed by a number referring to the subtype.

Influenza infection begins when the virus makes contact with the epithelium. Hemagglutinin binds to the host cell and allows viral entry, where it begins replication. Neuraminidase prevents viral aggregation and facilitates the release of virus from infected cells.²⁴

Mature virions are released and spread to neighboring host cells; this process is associated with desquamation and inflammation of the airways, causing cough, rhinorrhea, and sore throat. Systemic symptoms are associated with the induction of interferon, which causes fever and myalgia.²⁵

Recovery and immunity to influenza infection occurs through both humoral and cell-mediated immunity, with antibodies directed against the specific hemagglutinin and neuraminidase antigens of the infecting virus. Immunity wanes over time and with antigenic drift of circulating viruses, making the host susceptible to recurrent influenza infection.²⁴

Influenza is often complicated by bacterial superinfection

The influenza virus acts synergistically with certain bacteria to increase infectivity, and this may explain why influenza is often complicated by bacterial superinfection.

Mechanisms leading to bacterial superinfection include increased binding and invasion of bacteria, increased viral replication, and modification of the host inflammatory response. Some S aureus strains produce a protease that directly activates influenza virus hemagglutinin; other bacteria can activate plasminogen to promote influenza replication. The resulting increase in proteases in host tissues promotes activation of influenza through cleavage of hemagglutinin.²⁶

The influenza virus also causes damage to the airway epithelial layer, leading to increased exposure of the binding sites necessary for adherence of S pneumoniae.²⁷

CLINICAL PRESENTATION OF COMMUNITY-ACQUIRED PNEUMONIA

Although CAP is common, agreement on its essential clinical signs and symptoms is surprisingly limited, due in part to heterogeneous patient presentations and in part to interobserver variability. The reader is referred to two excellent reviews on this topic.^28,29

The diagnosis of CAP is made on clinical grounds, based on a combination of signs and symptoms. Symptoms of pneumonia can include cough, fever, chills, sputum production, dyspnea, and pleuritic pain. Physical findings can include tachypnea, tachycardia, hypoxemia, and consolidation or rales on auscultation. Laboratory data may show leukocytosis or elevated C-reactive protein, and radiographic studies may show evidence of a new infiltrate.^21,30,31

Clinical presentation of influenza

Seasonal influenza as a cause of CAP is difficult to distinguish from bacterial causes. The clinical presentation of seasonal influenza most commonly includes fever or subjective feverishness, cough, myalgia, and weakness.³² In a recent multivariate analysis, five clinical features were shown to be predictive of pandemic H1N1 influenza pneumonia rather than CABP: age younger than 65 years, absence of confusion, white blood cell count less than 12 × 10⁹/L, temperature higher than 38°C (100.4°F), and bilateral opacities on radiography.^32,33

Complicated influenza infection can be either primary viral pneumonia or bacterial superinfection.

During the 1918 influenza pandemic, which predated the ability to isolate viruses, two clinical syndromes emerged: an ARDS associated with the rapid onset of cyanosis, delirium, and frothy blood-tinged sputum; and an acute bronchopneumonia characterized by necrosis, hemorrhage, edema, and vasculitis.^34,35 The first syndrome has subsequently been shown to be associated with primary viral pneumonia, while the second is caused by bacterial superinfection. Modern reexamination of 1918 data has shown that bacterial superinfection was likely the reason for the distinctly fulminant presentation of that pandemic.^36,37

The 2009 H1N1 influenza pandemic caused relatively mild disease in most patients. However, those with severe pneumonia more commonly developed ARDS from primary influenza pneumonia than from bacterial superinfection.¹⁷

A third influenza-associated infection is secondary bacterial pneumonia, which follows influenza infection and mimics the presentation of CABP. A typical patient presents with a recent history of influenza-like illness, followed 4 to 14 days later by a recurrence of fever, dyspnea, productive cough, and consolidation on chest radiographs.³⁸ Leukocytosis with an increased number of immature neutrophil forms, prolonged duration of fever, and elevated erythrocyte sedimentation rate are more likely in patients with secondary bacterial pneumonia.³⁹ Isolates from sputum samples commonly include S pneumoniae, S aureus, H influenzae, and other gram-negative rods.⁴⁰

In recent flu seasons, methicillin-resistant S aureus (MRSA) has emerged as a cause of severe secondary pneumonia. Most of these isolates carry genes for the toxin Panton-Valentine leukocidin; the associated mortality rate is as high as 33%.^41,42 Although community-acquired MRSA pneumonia has only been reported in case series, distinct clinical features that have been described include severe pneumonia with high fever, hypotension, shock, respiratory failure, leukopenia, and multilobar and cavitary infiltrates.⁴³

WHEN TO SUSPECT INFLUENZA

The triad of fever, cough, and abrupt onset are the best predictors of influenza, but no single combination of signs and symptoms predict influenza infection with 100% certainty. Therefore, an understanding of local epidemiologic data regarding circulating influenza is essential to maintain a high index of suspicion.⁴⁴

It is appropriate to suspect influenza in:

• Anyone who is epidemiologically linked to a known outbreak of influenza
• Children, adults, and health care workers who have fever and abrupt onset of respiratory symptoms
• Patients with fever plus exacerbation of underlying pulmonary disease
• Severely ill patients with fever or hypothermia, especially during influenza season.⁴⁵

 

DIAGNOSTIC TESTING

Once the diagnosis of pulmonary infection is suggested by clinical features, the initial evaluation should include measurement of vital signs, physical examination, and radiographic imaging of the chest. Additional diagnostic measures to consider include viral testing, blood culture, sputum culture, urinary antigen testing for Legionella and for S pneumoniae, fungal culture, and mycobacterial smear and culture.

Chest radiography (with posterior-anterior and lateral films) is the study that usually demonstrates the presence of a pulmonary infiltrate. If initial chest radiographs do not show an infiltrate, imaging can be repeated after treatment is started if the patient’s clinical presentation still suggests pneumonia. Chest radiographs are of limited value in predicting the pathogen, but they help to determine the extent of pneumonia and to detect parapneumonic effusion.⁴⁶

A caveat: anterior-posterior, posterior-anterior, and lateral views can miss more than 10% of effusions large enough to warrant thoracentesis, especially when there is lower-lobe consolidation.⁴⁷

Blood cultures are recommended for patients admitted to the intensive care unit and for those with cavitary infiltrates, leukopenia, alcohol abuse, severe liver disease, asplenia, positive pneumococcal urinary antigen testing, or a pleural effusion.²¹ However, blood cultures are positive in only 3% to 14% of hospitalized patients with CABP, and the impact of a positive blood culture on management decisions in CABP appears to be quite small.^48–50

For the highest yield, blood culture results should be obtained before antibiotics are given. Not only is this good practice, but obtaining blood culture results before starting antibiotics is one of the quality measures evaluated by the Center for Medicare and Medicaid Services.⁵¹

Sputum culture is considered optional for outpatients and patients with less-severe pneumonia.²¹ While it can provide a rapid diagnosis in certain cases, a good-quality sputum sample is obtained in only 39% to 54% of patients with CABP, yields a predominant morphotype in only 45% of cases, and provides a useful microbiologic diagnosis in only 14.4%.^52,53 Fungal and mycobacterial cultures are only indicated in certain situations such as cavitary infiltrates or immunosuppression.

Urinary antigen testing for Legionella and S pneumoniae should be done in patients with more severe illness and in those for whom outpatient therapy has failed.²¹S pneumoniae testing has been shown to allow early diagnosis of pneumococcal pneumonia in 26% more patients than with Gram staining, but it fails to identify 22% of the rapid diagnoses initially identified by Gram staining.⁵⁴ Thus, a sequential approach is reasonable, with urinary antigen testing for patients at high risk without useful results from sputum Gram staining. Also, recent data suggest that the pneumococcal urinary antigen test may allow optimization of antimicrobial therapy with good clinical outcomes.⁵⁵

Endotracheal tests. If the patient is intubated, collection of endotracheal aspirates, bronchoscopy, or nonbronchoscopic bronchial lavage (sometimes called “mini-BAL”) should be performed.

Thoracentesis and pleural fluid cultures should be done if a pleural effusion is found. Empyema, large or loculated effusions, and parapneumonic effusions with a pH lower than 7.20, glucose levels less than 3.4 mmol/L (60 mg/dL), or positive results on microbial staining or culture should be drained by chest tube or surgically.⁵⁶

Testing for influenza should be done if it will change the clinical management, such as the choice of antibiotic or infection control practices. Specimens should be obtained with either a nasopharyngeal swab or aspirate and tested with reverse transcriptase polymerase chain reaction, immunofluorescent staining, or rapid antigen detection, depending on local availability.⁴⁵

Inflammatory biomarkers such as C-reactive protein and procalcitonin have been receiving interest as ways to predict the etiology and prognosis of CAP and to guide therapy. Several studies have shown that C-reactive protein can help distinguish between CAP and bronchitis, with higher values suggesting more severe pneumonia and pneumonia caused by S pneumoniae or L pneumophila.⁵⁷ Procalcitonin may help discriminate between severe lower respiratory tract infections of bacterial and 2009 H1N1 origin, although less effectively than C-reactive protein. Low procalcitonin values, particularly when combined with low C-reactive protein levels, suggest that bacterial infection is unlikely.⁵⁸

RISK STRATIFICATION AND SITE-OF-CARE DECISIONS

Following a presumptive diagnosis of CAP, it is important to decide not only what treatment the patient will receive but whether he or she should be hospitalized. If the patient is to be admitted to the hospital, the clinician must also decide if his or her condition warrants intensive care.

Severity-of-illness scores

Several severity-of-illness scores and prognostic models have been validated for use in deciding on inpatient vs outpatient treatment and to aid in the decision of whether a patient with pneumonia should be admitted to an intensive care unit. The most extensively studied and widely used scoring systems are the Pneumonia Severity Index (PSI) (Table 1)⁵⁹ and the CURB-65 (Figure 1).⁶⁰

The PSI is the more complicated of the two, as it is based on 19 variables. Online calculators are available for the PSI (http://pda.ahrq.gov/clinic/psi/psicalc.asp) and the CURB-65 (http://www.mdcalc.com/curb-65-severity-score-community-acquiredpneumonia).

A recent meta-analysis compared the performance characteristics of the PSI and CURB-65 scores for predicting mortality in CAP and found no significant differences in overall test performance.⁶¹

Another meta-analysis found that the PSI was more sensitive than the CURB-65 and had a low false-negative rate, and so was better at showing which patients do not need to be hospitalized. Conversely, the CURB-65 was more specific and had a higher positive predictive value, and thus was more likely to correctly classify high-risk patients.⁶²

Other scoring systems that aid in deciding about hospital admission and level of care include the CRB-65⁶³ (which can be used instead of the CURB-65 if laboratory values are not available), SMART-COP,⁶⁴ and SCAP.^,65

Guidelines on when to admit to the intensive care unit

Guidelines from the Infectious Diseases Society of America (IDSA) and the American Thoracic Society (ATS) also provide guidance on when intensive care admission is advised,²¹ and their criteria were recently validated.⁶⁶ The guidelines advocate direct admission to the intensive care unit for patients requiring vasopressors or mechanical ventilation, and intensive care unit or high-level monitoring for patients with three of the following criteria for severe CAP²¹:

• Respiratory rate ≥ 30
• Pao₂/Fio₂ ratio ≤ 250
• Multilobar infiltrates
• Confusion or disorientation
• Uremia (blood urea nitrogen ≥ 20 mg/dL)
• Leukopenia (white blood cell count < 4.0 × 10⁹/L)
• Thrombocytopenia (platelet count < 100 × 10⁹/L)
• Hypothermia (core temperature < 36.0°C [96.8°F])
• Hypotension requiring aggressive fluid resuscitation.

None of these scoring systems or criteria is meant to replace clinical judgment. A recent study has suggested that an oxygen saturation of less than 92% is an appropriate threshold for hospital admission, in view of higher rates of adverse events in outpatients with saturations below this value.⁶⁷

 

TREATMENT

Multiple studies have shown that treatment of CAP in accordance with guidelines has led to improved clinical outcomes.^21,68–70

How fast must antibiotics be started?

Based on studies that showed a lower mortality rate when antibiotics were started sooner, Medicare and Medicaid adopted a quality measure calling for starting antibiotics within 4 hours in patients being admitted to the hospital.^50,71 However, several subsequent studies showed that the diagnosis of pneumonia is often incorrect and that rapid administration of antibiotics could lead to misdiagnosis, overuse of antibiotics, and a higher risk of Clostridium difficile infection.^72,73

The current IDSA/ATS guidelines²¹ recommend that the first antibiotic dose be given while the patient is still in the emergency department, but do not give a specific time within which it should be given. Medicare and Medicaid later updated their quality measure to antibiotic administration within 6 hours.

Which antibiotics should be used?

The selection of antimicrobial agent depends upon the patient’s severity of illness and comorbid conditions.

Although most studies of combination antibiotic therapy have been retrospective and observational, they suggest that a macrolide (ie, one of the “mycins”) added to a beta-lactam antibiotic is beneficial, possibly by covering atypical organisms or via anti-inflammatory action.^74–76 The choice of one antibiotic over another appears to be less important, and a recent Cochrane review concluded that there was no significant difference in efficacy among five antibiotic pairs studied.⁷⁷

Empiric outpatient treatment of a previously healthy patient with CAP and no risk factors for drug-resistant S pneumoniae should include either a macrolide (azithromycin [Zithromax], clarithromycin [Biaxin], or erythromycin) or doxycycline. If the patient has a chronic comorbid condition such as heart, lung, liver, or renal disease, diabetes mellitus, alcoholism, malignancy, asplenia, or immunosuppression or has received antimicrobials within the preceding 3 months, then treatment should include either a respiratory fluoroquinolone (moxifloxacin [Avelox] or levofloxacin [Levaquin]) or a beta-lactam plus a macrolide.²¹

Overall, published data suggest that the survival rate is about the same with fluoroquinolone monotherapy as with beta-lactam plus macrolide combination therapy, and better than with beta-lactam monotherapy.⁷⁸

Selection of antibiotics for inpatient treatment of CAP is influenced by severity of illness. Inpatients who do not require intensive care should be treated with either a respiratory fluoroquinolone or combination therapy with a beta-lactam (cefotaxime [Claforan], ceftriaxone [Rocephin], ampicillin, or ertapenem [Invanz]) plus a macrolide or doxycycline.^21,76,79

If a specific microbiologic diagnosis is made, then treatment can be narrowed. However in certain cases, such as invasive pneumococcal infection, combination therapy may still be superior.^80,81 For patients who need intensive care, treatment should always include a beta-lactam plus either azithromycin or a respiratory fluoroquinolone.²¹ In certain situations, additional antibiotics may be added as well, such as agents to treat Pseudomonas, community-acquired MRSA, or both.

Switching to oral therapy; short-course therapy

In the interest of avoiding unnecessary antibiotics, numerous studies have addressed the issue of an “early switch” to oral antibiotics and “short-course” therapy for CAP. In general, once clinically stable, patients with CAP, including bacteremic S pneumoniae pneumonia, can be safely switched to oral antibiotics.⁸²

The issue of short-course therapy is more complicated, and the appropriate length of therapy for CAP is not well established. However, 5 days of levofloxacin 750 mg was shown to be as successful as 7 to 10 days of levofloxacin 500 mg.⁸³ In another study, in patients who improved after 3 days of intravenous therapy for CAP, there was no difference in clinical outcome between those who were changed to oral therapy for 5 more days and those who received an oral placebo.⁸⁴

Most patients who achieve clinical stability in the first week do not need prolonged antibiotic therapy. However, certain conditions, such as S aureus bacteremic pneumonia, complicated pneumonia, and pneumonia due to unusual organisms, may require prolonged treatment.

Other therapies

Additional therapies studied in patients with pneumonia include early mobilization, adjunctive corticosteroids, and statin drugs.

Early mobilization was shown in one study to decrease hospital length of stay without increasing adverse effects.⁸⁵

Corticosteroids are not supported as a standard of care for patients with severe CAP according to current available studies.^86,87 Furthermore, a randomized, controlled trial showed that prednisolone daily for a week did not improve outcomes in hospitalized patients with CAP, and it was associated with increased late failure.⁸⁸

Statin trials under way. Several observational studies have suggested that statins might be beneficial in managing sepsis through their effects on endothelial cell function, antioxidant effects, anti-inflammatory effects, and immunomodulatory effects.⁸⁹ However, a recent large prospective multicenter cohort study of hospitalized patients with CAP did not find evidence of a protective effect of statins on clinically meaningful outcomes in CAP or significant differences in circulating biomarkers.⁹⁰ Several randomized trials of statin therapy in patients with both ventilator-associated pneumonia and CAP are under way.

INFLUENZA TREATMENT: MOST EFFECTIVE WITHIN 48 HOURS

Treatment with antiviral drugs is most effective if started within 48 hours after symptom onset, although some patients with confirmed influenza who are either not improving or who are critically ill may still benefit from treatment started later.

Treatment should be considered in patients with laboratory-confirmed or suspected influenza who are at risk of developing complicated influenza and in otherwise healthy patients who wish to reduce the duration of illness or who have close contact with patients who are at high risk of complications.

Antiviral medications are oseltamivir (Tamiflu), zanamivir (Relenza), and the adamantines amantadine (Symmetrel) and rimantadine (Flumadine).

Due to evolving viral resistance patterns, the choice of antiviral drug depends on the strain. Seasonal H1N1 is best treated with zanamivir or an adamantine, while pandemic 2009 H1N1 and H3N2 are best treated with zanamivir or oseltamivir. When strain typing is not available, empiric therapy should be with either zanamivir monotherapy or a combination of oseltamivir plus rimantadine. Influenza B viruses are resistant to adamantines and should be treated only with either zanamivir or oseltamivir.⁴⁵

 

FOLLOW-UP AND PREVENTION

Patients with CAP can generally be expected to improve within 3 to 7 days.⁹¹ However, it may be several weeks before they return to baseline.⁹²

Follow-up plans may be guided by the time to clinical stability. For patients who do not achieve clinical stability until more than 72 hours after admission, more aggressive follow-up on discharge is indicated, since they are more likely to experience early readmission and death.⁹³

Pneumococcal vaccination. Because S pneumoniae remains the most common cause of CAP, efforts should be made to vaccinate patients appropriately. The Advisory Committee on Immunization Practices (ACIP) and the US Centers for Disease Control and Prevention recommend that the pneumococcal polysaccharide vaccine (Pneumovax 23; PPSV23) be given to those over age 65. Those who were vaccinated before age 65 should receive another dose at age 65 or later if at least 5 years have passed since their previous dose. Those who receive it at or after age 65 should receive only a single dose. A second dose is recommended 5 years after the first dose for people age 19 to 64 years with functional or anatomic asplenia and for those who are immunocompromised.

Influenza vaccination for all. Of note, the ACIP updated its guidelines on influenza vaccination beginning with the 2010–2011 influenza season. It no longer advocates a risk-stratified approach. Instead, it recommends universal influenza vaccination for everybody more than 6 months old.⁹⁴

Smoking cessation should be addressed. Smoking cessation is a Medicare and Medicaid quality measure and should be encouraged after an episode of CAP because quitting smoking reduces the risk of pneumococcal disease by approximately 14% each year thereafter.⁹⁵

General internists need to be able to recognize community-acquired pneumonia (CAP) so that diagnostic and therapeutic interventions can be initiated promptly. It is also important to understand the most likely and possible causes of CAP so that appropriate initial antimicrobial therapy can be chosen. Especially during flu season, influenza can present as CAP and should be included in the differential diagnosis.

When managing a patient with CAP, the internist must decide which level of care, diagnostic tests, antimicrobial agents, and follow-up plans are needed. These topics will be reviewed in this article.

TWO TERMS TO REMEMBER

• CAP refers to pneumonia acquired outside a health care facility. It can be either bacterial or viral.
• CABP (community-acquired bacterial pneumonia) refers only to those cases caused by bacterial pathogens.

NUMBERS AND TRENDS

In the United States, CAP is the number-one cause of death from infection and the sixth leading cause of death overall.¹ Each year, it is responsible for about 4.2 million outpatient visits, more than 60,000 deaths, and more than $17 billion in health care expenses.²

Community-acquired bacterial pneumonia: Common, serious

In a population-based US study in 1991, the incidence of CABP requiring hospitalization was 266.8 per 100,000 people.³

Estimates of overall mortality in CABP vary depending on the severity of illness and comorbid conditions. A meta-analysis published in 1996 found the overall mortality rate to be 13.7%, with a range of 5.1% to 36.5% depending on severity.⁴

In hospitalized patients, mortality rates and length of hospital stay appear to be declining over time. Between 1993 and 2005, the age-adjusted mortality rate decreased from 8.9% to 4.1%, and the average length of stay decreased from 7.5 to 5.7 days, with an overall reduction in hospital cost.⁵

CABP is more prevalent in older people than in the general population, and it increases with age from 18.2 cases per 1,000 patient-years in patients 60 to 69 years to 52.3 cases per 1,000 patient-years in those older than 85 years.⁶ Risk factors for pneumonia in the elderly include heart disease, chronic lung disease, immunosuppressive drugs, alcoholism, and increasing age.⁷ Similar to the trend in the general population, the mortality rate in elderly CABP patients appears to be decreasing over time, possibly thanks to rising rates of pneumococcal and influenza vaccination.⁸

Among the general population, risk factors for developing CABP also include smoking, occupational dust exposure, history of childhood pneumonia, unemployment, and single marital status.⁹ The incidence of CABP does not appear to be higher among pregnant women, although it is the most frequent cause of nonobstetric death in this population.¹⁰

The use of proton pump inhibitors may be an emerging risk factor for CABP.¹¹ Also, use of nonsteroidal anti-inflammatory drugs among patients with CABP is associated with a blunted inflammatory response as well as a higher risk of pleuropulmonary complications and a delay in presentation.¹²

Influenza is also common, potentially severe

Influenza is also very common and potentially severe. It can cause a spectrum of disease, from mild upper respiratory tract symptoms to severe viral pneumonia that can be life-threatening and complicated by respiratory failure and the acute respiratory distress syndrome (ARDS).

Influenza infection can also be complicated by subsequent bacterial pneumonia. However, the epidemiology of influenza infection differs from that of CABP in that influenza occurs seasonally.

In the United States, seasonal influenza causes 36,000 deaths and 200,000 hospitalizations annually.^13,14 As with CABP, the risk of death from influenza increases with age: it is 16 times greater in people age 85 and older than in those ages 65 to 69.¹³

During yearly seasonal epidemics, those at the highest risk of hospitalization and death are at the extremes of age. Risk factors for complicated influenza include heart disease, lung disease, diabetes, renal failure, rheumatologic conditions, dementia, and neurologic disease.^15,16 During the 2009 H1N1 influenza pandemic, unexpected severity was seen in previously healthy young adults as well as those with obesity, neurodegenerative disease, pregnancy, and asthma.¹⁷

PATHOGENS: TYPICAL, ATYPICAL, VIRAL

Identifying the etiologic organism in CAP is confounded by limitations in the available diagnostic tests and also by poor-quality specimens that often are contaminated with bacteria that colonize the upper airways. Given these caveats, the primary pathogens responsible for CAP broadly include typical bacterial pathogens, atypical bacterial pathogens, and viruses.

Typical bacterial pathogens include Streptococcus pneumoniae, Haemophilus influenzae, Staphylococcus aureus, Moraxella catarrhalis, and, less commonly, a variety of aerobic and anaerobic gram-negative rods including Pseudomonas aeruginosa, Acinetobacter species, and Klebsiella pneumoniae.

Atypical bacterial pathogens include Mycoplasma pneumoniae, Chlamydia pneumoniae, and Legionella species.¹⁸

Viruses implicated in adult CAP include influenza A and B, parainfluenza viruses, respiratory syncytial virus, and adenovirus.¹⁹ More recently, human metapneumovirus has been described as a cause of adult CAP.²⁰

Clues to uncommon microbes

Specific historic features or coexisting conditions that may suggest an uncommon microbiologic diagnosis include²¹:

• Recent travel to the southwestern United States or Southeast Asia
• Ill contacts
• Exposure to birds, bats, rabbits, or farm animals
• Alcoholism
• Chronic obstructive pulmonary disease
• Human immunodeficiency virus infection
• Structural lung disease
• Prolonged cough with whoop or posttussive vomiting
• Aspiration
• Bioterrorism.

In cases in which one or more of these conditions exist, CAP may also be caused by other agents not listed above, including Mycobacterium tuberculosis, oral anaerobes, atypical mycobacteria, Histoplasma capsulatum, Chlamydophila psittaci, Francisella tularensis, Coxiella burnettii, Pneumocystis jiroveci, Cryptococcus, Aspergillus, Coccidioides, Hantavirus, avian influenza, Burkholderia pseudomallei, severe acute respiratory syndrome virus, Bordetella pertussis, Bacillus anthracis, and Yersinia pestis.

 

HOW BACTERIA INVADE THE LUNGS

The pathophysiology of CABP involves both host defense and microbial virulence factors.

The airways are most commonly exposed to microbes by microaspiration of upper airway flora, although hematogenous seeding of the lungs in a bacteremic patient or contiguous spread of infection from an adjacent site can also occur.

Mucociliary clearance and the cough reflex are important initial defenses against infection and can be inhibited by neurologic diseases and conditions that impair the mucociliary mechanism. Mucosal immune cells, including macrophages and neutrophils, recognize invading pathogens and generate an antibody response.

Regulation of the host inflammatory response to infection depends on a complex interaction between immune cells, inflammatory cytokines (eg, tumor necrosis factor alpha, interleukin 1-beta, and interleukin 6), and anti-inflammatory cytokines such as interleukin 1 receptor antagonist and soluble tumor necrosis factor receptor type I.²²

The interaction and timing of the inflammatory and anti-inflammatory response are essential in manifesting an appropriate host response to infection. An inadequate inflammatory response can lead to sepsis and death, but an excessive, late anti-inflammatory response can lead to a systemic inflammatory response such as ARDS. Polymorphisms within the genes coding for these factors may explain the variation in severity of illness among patients with CABP.²³

HOW INFLUENZA DOES ITS DAMAGE

There are three types of influenza virus: A, B, and C. Type A causes most human infections. The influenza A virus envelope comprises a lipid bilayer that contains the projecting glycoproteins hemagglutinin and neuraminidase. Influenza viruses are named on the basis of these proteins and are designated with an H and an N, respectively, each followed by a number referring to the subtype.

Influenza infection begins when the virus makes contact with the epithelium. Hemagglutinin binds to the host cell and allows viral entry, where it begins replication. Neuraminidase prevents viral aggregation and facilitates the release of virus from infected cells.²⁴

Mature virions are released and spread to neighboring host cells; this process is associated with desquamation and inflammation of the airways, causing cough, rhinorrhea, and sore throat. Systemic symptoms are associated with the induction of interferon, which causes fever and myalgia.²⁵

Recovery and immunity to influenza infection occurs through both humoral and cell-mediated immunity, with antibodies directed against the specific hemagglutinin and neuraminidase antigens of the infecting virus. Immunity wanes over time and with antigenic drift of circulating viruses, making the host susceptible to recurrent influenza infection.²⁴

Influenza is often complicated by bacterial superinfection

The influenza virus acts synergistically with certain bacteria to increase infectivity, and this may explain why influenza is often complicated by bacterial superinfection.

Mechanisms leading to bacterial superinfection include increased binding and invasion of bacteria, increased viral replication, and modification of the host inflammatory response. Some S aureus strains produce a protease that directly activates influenza virus hemagglutinin; other bacteria can activate plasminogen to promote influenza replication. The resulting increase in proteases in host tissues promotes activation of influenza through cleavage of hemagglutinin.²⁶

The influenza virus also causes damage to the airway epithelial layer, leading to increased exposure of the binding sites necessary for adherence of S pneumoniae.²⁷

CLINICAL PRESENTATION OF COMMUNITY-ACQUIRED PNEUMONIA

Although CAP is common, agreement on its essential clinical signs and symptoms is surprisingly limited, due in part to heterogeneous patient presentations and in part to interobserver variability. The reader is referred to two excellent reviews on this topic.^28,29

The diagnosis of CAP is made on clinical grounds, based on a combination of signs and symptoms. Symptoms of pneumonia can include cough, fever, chills, sputum production, dyspnea, and pleuritic pain. Physical findings can include tachypnea, tachycardia, hypoxemia, and consolidation or rales on auscultation. Laboratory data may show leukocytosis or elevated C-reactive protein, and radiographic studies may show evidence of a new infiltrate.^21,30,31

Clinical presentation of influenza

Seasonal influenza as a cause of CAP is difficult to distinguish from bacterial causes. The clinical presentation of seasonal influenza most commonly includes fever or subjective feverishness, cough, myalgia, and weakness.³² In a recent multivariate analysis, five clinical features were shown to be predictive of pandemic H1N1 influenza pneumonia rather than CABP: age younger than 65 years, absence of confusion, white blood cell count less than 12 × 10⁹/L, temperature higher than 38°C (100.4°F), and bilateral opacities on radiography.^32,33

Complicated influenza infection can be either primary viral pneumonia or bacterial superinfection.

During the 1918 influenza pandemic, which predated the ability to isolate viruses, two clinical syndromes emerged: an ARDS associated with the rapid onset of cyanosis, delirium, and frothy blood-tinged sputum; and an acute bronchopneumonia characterized by necrosis, hemorrhage, edema, and vasculitis.^34,35 The first syndrome has subsequently been shown to be associated with primary viral pneumonia, while the second is caused by bacterial superinfection. Modern reexamination of 1918 data has shown that bacterial superinfection was likely the reason for the distinctly fulminant presentation of that pandemic.^36,37

The 2009 H1N1 influenza pandemic caused relatively mild disease in most patients. However, those with severe pneumonia more commonly developed ARDS from primary influenza pneumonia than from bacterial superinfection.¹⁷

A third influenza-associated infection is secondary bacterial pneumonia, which follows influenza infection and mimics the presentation of CABP. A typical patient presents with a recent history of influenza-like illness, followed 4 to 14 days later by a recurrence of fever, dyspnea, productive cough, and consolidation on chest radiographs.³⁸ Leukocytosis with an increased number of immature neutrophil forms, prolonged duration of fever, and elevated erythrocyte sedimentation rate are more likely in patients with secondary bacterial pneumonia.³⁹ Isolates from sputum samples commonly include S pneumoniae, S aureus, H influenzae, and other gram-negative rods.⁴⁰

In recent flu seasons, methicillin-resistant S aureus (MRSA) has emerged as a cause of severe secondary pneumonia. Most of these isolates carry genes for the toxin Panton-Valentine leukocidin; the associated mortality rate is as high as 33%.^41,42 Although community-acquired MRSA pneumonia has only been reported in case series, distinct clinical features that have been described include severe pneumonia with high fever, hypotension, shock, respiratory failure, leukopenia, and multilobar and cavitary infiltrates.⁴³

WHEN TO SUSPECT INFLUENZA

The triad of fever, cough, and abrupt onset are the best predictors of influenza, but no single combination of signs and symptoms predict influenza infection with 100% certainty. Therefore, an understanding of local epidemiologic data regarding circulating influenza is essential to maintain a high index of suspicion.⁴⁴

It is appropriate to suspect influenza in:

• Anyone who is epidemiologically linked to a known outbreak of influenza
• Children, adults, and health care workers who have fever and abrupt onset of respiratory symptoms
• Patients with fever plus exacerbation of underlying pulmonary disease
• Severely ill patients with fever or hypothermia, especially during influenza season.⁴⁵

 

DIAGNOSTIC TESTING

Once the diagnosis of pulmonary infection is suggested by clinical features, the initial evaluation should include measurement of vital signs, physical examination, and radiographic imaging of the chest. Additional diagnostic measures to consider include viral testing, blood culture, sputum culture, urinary antigen testing for Legionella and for S pneumoniae, fungal culture, and mycobacterial smear and culture.

Chest radiography (with posterior-anterior and lateral films) is the study that usually demonstrates the presence of a pulmonary infiltrate. If initial chest radiographs do not show an infiltrate, imaging can be repeated after treatment is started if the patient’s clinical presentation still suggests pneumonia. Chest radiographs are of limited value in predicting the pathogen, but they help to determine the extent of pneumonia and to detect parapneumonic effusion.⁴⁶

A caveat: anterior-posterior, posterior-anterior, and lateral views can miss more than 10% of effusions large enough to warrant thoracentesis, especially when there is lower-lobe consolidation.⁴⁷

Blood cultures are recommended for patients admitted to the intensive care unit and for those with cavitary infiltrates, leukopenia, alcohol abuse, severe liver disease, asplenia, positive pneumococcal urinary antigen testing, or a pleural effusion.²¹ However, blood cultures are positive in only 3% to 14% of hospitalized patients with CABP, and the impact of a positive blood culture on management decisions in CABP appears to be quite small.^48–50

For the highest yield, blood culture results should be obtained before antibiotics are given. Not only is this good practice, but obtaining blood culture results before starting antibiotics is one of the quality measures evaluated by the Center for Medicare and Medicaid Services.⁵¹

Sputum culture is considered optional for outpatients and patients with less-severe pneumonia.²¹ While it can provide a rapid diagnosis in certain cases, a good-quality sputum sample is obtained in only 39% to 54% of patients with CABP, yields a predominant morphotype in only 45% of cases, and provides a useful microbiologic diagnosis in only 14.4%.^52,53 Fungal and mycobacterial cultures are only indicated in certain situations such as cavitary infiltrates or immunosuppression.

Urinary antigen testing for Legionella and S pneumoniae should be done in patients with more severe illness and in those for whom outpatient therapy has failed.²¹S pneumoniae testing has been shown to allow early diagnosis of pneumococcal pneumonia in 26% more patients than with Gram staining, but it fails to identify 22% of the rapid diagnoses initially identified by Gram staining.⁵⁴ Thus, a sequential approach is reasonable, with urinary antigen testing for patients at high risk without useful results from sputum Gram staining. Also, recent data suggest that the pneumococcal urinary antigen test may allow optimization of antimicrobial therapy with good clinical outcomes.⁵⁵

Endotracheal tests. If the patient is intubated, collection of endotracheal aspirates, bronchoscopy, or nonbronchoscopic bronchial lavage (sometimes called “mini-BAL”) should be performed.

Thoracentesis and pleural fluid cultures should be done if a pleural effusion is found. Empyema, large or loculated effusions, and parapneumonic effusions with a pH lower than 7.20, glucose levels less than 3.4 mmol/L (60 mg/dL), or positive results on microbial staining or culture should be drained by chest tube or surgically.⁵⁶

Testing for influenza should be done if it will change the clinical management, such as the choice of antibiotic or infection control practices. Specimens should be obtained with either a nasopharyngeal swab or aspirate and tested with reverse transcriptase polymerase chain reaction, immunofluorescent staining, or rapid antigen detection, depending on local availability.⁴⁵

Inflammatory biomarkers such as C-reactive protein and procalcitonin have been receiving interest as ways to predict the etiology and prognosis of CAP and to guide therapy. Several studies have shown that C-reactive protein can help distinguish between CAP and bronchitis, with higher values suggesting more severe pneumonia and pneumonia caused by S pneumoniae or L pneumophila.⁵⁷ Procalcitonin may help discriminate between severe lower respiratory tract infections of bacterial and 2009 H1N1 origin, although less effectively than C-reactive protein. Low procalcitonin values, particularly when combined with low C-reactive protein levels, suggest that bacterial infection is unlikely.⁵⁸

RISK STRATIFICATION AND SITE-OF-CARE DECISIONS

Following a presumptive diagnosis of CAP, it is important to decide not only what treatment the patient will receive but whether he or she should be hospitalized. If the patient is to be admitted to the hospital, the clinician must also decide if his or her condition warrants intensive care.

Severity-of-illness scores

Several severity-of-illness scores and prognostic models have been validated for use in deciding on inpatient vs outpatient treatment and to aid in the decision of whether a patient with pneumonia should be admitted to an intensive care unit. The most extensively studied and widely used scoring systems are the Pneumonia Severity Index (PSI) (Table 1)⁵⁹ and the CURB-65 (Figure 1).⁶⁰

The PSI is the more complicated of the two, as it is based on 19 variables. Online calculators are available for the PSI (http://pda.ahrq.gov/clinic/psi/psicalc.asp) and the CURB-65 (http://www.mdcalc.com/curb-65-severity-score-community-acquiredpneumonia).

A recent meta-analysis compared the performance characteristics of the PSI and CURB-65 scores for predicting mortality in CAP and found no significant differences in overall test performance.⁶¹

Another meta-analysis found that the PSI was more sensitive than the CURB-65 and had a low false-negative rate, and so was better at showing which patients do not need to be hospitalized. Conversely, the CURB-65 was more specific and had a higher positive predictive value, and thus was more likely to correctly classify high-risk patients.⁶²

Other scoring systems that aid in deciding about hospital admission and level of care include the CRB-65⁶³ (which can be used instead of the CURB-65 if laboratory values are not available), SMART-COP,⁶⁴ and SCAP.^,65

Guidelines on when to admit to the intensive care unit

Guidelines from the Infectious Diseases Society of America (IDSA) and the American Thoracic Society (ATS) also provide guidance on when intensive care admission is advised,²¹ and their criteria were recently validated.⁶⁶ The guidelines advocate direct admission to the intensive care unit for patients requiring vasopressors or mechanical ventilation, and intensive care unit or high-level monitoring for patients with three of the following criteria for severe CAP²¹:

• Respiratory rate ≥ 30
• Pao₂/Fio₂ ratio ≤ 250
• Multilobar infiltrates
• Confusion or disorientation
• Uremia (blood urea nitrogen ≥ 20 mg/dL)
• Leukopenia (white blood cell count < 4.0 × 10⁹/L)
• Thrombocytopenia (platelet count < 100 × 10⁹/L)
• Hypothermia (core temperature < 36.0°C [96.8°F])
• Hypotension requiring aggressive fluid resuscitation.

None of these scoring systems or criteria is meant to replace clinical judgment. A recent study has suggested that an oxygen saturation of less than 92% is an appropriate threshold for hospital admission, in view of higher rates of adverse events in outpatients with saturations below this value.⁶⁷

 

TREATMENT

Multiple studies have shown that treatment of CAP in accordance with guidelines has led to improved clinical outcomes.^21,68–70

How fast must antibiotics be started?

Based on studies that showed a lower mortality rate when antibiotics were started sooner, Medicare and Medicaid adopted a quality measure calling for starting antibiotics within 4 hours in patients being admitted to the hospital.^50,71 However, several subsequent studies showed that the diagnosis of pneumonia is often incorrect and that rapid administration of antibiotics could lead to misdiagnosis, overuse of antibiotics, and a higher risk of Clostridium difficile infection.^72,73

The current IDSA/ATS guidelines²¹ recommend that the first antibiotic dose be given while the patient is still in the emergency department, but do not give a specific time within which it should be given. Medicare and Medicaid later updated their quality measure to antibiotic administration within 6 hours.

Which antibiotics should be used?

The selection of antimicrobial agent depends upon the patient’s severity of illness and comorbid conditions.

Although most studies of combination antibiotic therapy have been retrospective and observational, they suggest that a macrolide (ie, one of the “mycins”) added to a beta-lactam antibiotic is beneficial, possibly by covering atypical organisms or via anti-inflammatory action.^74–76 The choice of one antibiotic over another appears to be less important, and a recent Cochrane review concluded that there was no significant difference in efficacy among five antibiotic pairs studied.⁷⁷

Empiric outpatient treatment of a previously healthy patient with CAP and no risk factors for drug-resistant S pneumoniae should include either a macrolide (azithromycin [Zithromax], clarithromycin [Biaxin], or erythromycin) or doxycycline. If the patient has a chronic comorbid condition such as heart, lung, liver, or renal disease, diabetes mellitus, alcoholism, malignancy, asplenia, or immunosuppression or has received antimicrobials within the preceding 3 months, then treatment should include either a respiratory fluoroquinolone (moxifloxacin [Avelox] or levofloxacin [Levaquin]) or a beta-lactam plus a macrolide.²¹

Overall, published data suggest that the survival rate is about the same with fluoroquinolone monotherapy as with beta-lactam plus macrolide combination therapy, and better than with beta-lactam monotherapy.⁷⁸

Selection of antibiotics for inpatient treatment of CAP is influenced by severity of illness. Inpatients who do not require intensive care should be treated with either a respiratory fluoroquinolone or combination therapy with a beta-lactam (cefotaxime [Claforan], ceftriaxone [Rocephin], ampicillin, or ertapenem [Invanz]) plus a macrolide or doxycycline.^21,76,79

If a specific microbiologic diagnosis is made, then treatment can be narrowed. However in certain cases, such as invasive pneumococcal infection, combination therapy may still be superior.^80,81 For patients who need intensive care, treatment should always include a beta-lactam plus either azithromycin or a respiratory fluoroquinolone.²¹ In certain situations, additional antibiotics may be added as well, such as agents to treat Pseudomonas, community-acquired MRSA, or both.

Switching to oral therapy; short-course therapy

In the interest of avoiding unnecessary antibiotics, numerous studies have addressed the issue of an “early switch” to oral antibiotics and “short-course” therapy for CAP. In general, once clinically stable, patients with CAP, including bacteremic S pneumoniae pneumonia, can be safely switched to oral antibiotics.⁸²

The issue of short-course therapy is more complicated, and the appropriate length of therapy for CAP is not well established. However, 5 days of levofloxacin 750 mg was shown to be as successful as 7 to 10 days of levofloxacin 500 mg.⁸³ In another study, in patients who improved after 3 days of intravenous therapy for CAP, there was no difference in clinical outcome between those who were changed to oral therapy for 5 more days and those who received an oral placebo.⁸⁴

Most patients who achieve clinical stability in the first week do not need prolonged antibiotic therapy. However, certain conditions, such as S aureus bacteremic pneumonia, complicated pneumonia, and pneumonia due to unusual organisms, may require prolonged treatment.

Other therapies

Additional therapies studied in patients with pneumonia include early mobilization, adjunctive corticosteroids, and statin drugs.

Early mobilization was shown in one study to decrease hospital length of stay without increasing adverse effects.⁸⁵

Corticosteroids are not supported as a standard of care for patients with severe CAP according to current available studies.^86,87 Furthermore, a randomized, controlled trial showed that prednisolone daily for a week did not improve outcomes in hospitalized patients with CAP, and it was associated with increased late failure.⁸⁸

Statin trials under way. Several observational studies have suggested that statins might be beneficial in managing sepsis through their effects on endothelial cell function, antioxidant effects, anti-inflammatory effects, and immunomodulatory effects.⁸⁹ However, a recent large prospective multicenter cohort study of hospitalized patients with CAP did not find evidence of a protective effect of statins on clinically meaningful outcomes in CAP or significant differences in circulating biomarkers.⁹⁰ Several randomized trials of statin therapy in patients with both ventilator-associated pneumonia and CAP are under way.

INFLUENZA TREATMENT: MOST EFFECTIVE WITHIN 48 HOURS

Treatment with antiviral drugs is most effective if started within 48 hours after symptom onset, although some patients with confirmed influenza who are either not improving or who are critically ill may still benefit from treatment started later.

Treatment should be considered in patients with laboratory-confirmed or suspected influenza who are at risk of developing complicated influenza and in otherwise healthy patients who wish to reduce the duration of illness or who have close contact with patients who are at high risk of complications.

Antiviral medications are oseltamivir (Tamiflu), zanamivir (Relenza), and the adamantines amantadine (Symmetrel) and rimantadine (Flumadine).

Due to evolving viral resistance patterns, the choice of antiviral drug depends on the strain. Seasonal H1N1 is best treated with zanamivir or an adamantine, while pandemic 2009 H1N1 and H3N2 are best treated with zanamivir or oseltamivir. When strain typing is not available, empiric therapy should be with either zanamivir monotherapy or a combination of oseltamivir plus rimantadine. Influenza B viruses are resistant to adamantines and should be treated only with either zanamivir or oseltamivir.⁴⁵

 

FOLLOW-UP AND PREVENTION

Patients with CAP can generally be expected to improve within 3 to 7 days.⁹¹ However, it may be several weeks before they return to baseline.⁹²

Follow-up plans may be guided by the time to clinical stability. For patients who do not achieve clinical stability until more than 72 hours after admission, more aggressive follow-up on discharge is indicated, since they are more likely to experience early readmission and death.⁹³

Pneumococcal vaccination. Because S pneumoniae remains the most common cause of CAP, efforts should be made to vaccinate patients appropriately. The Advisory Committee on Immunization Practices (ACIP) and the US Centers for Disease Control and Prevention recommend that the pneumococcal polysaccharide vaccine (Pneumovax 23; PPSV23) be given to those over age 65. Those who were vaccinated before age 65 should receive another dose at age 65 or later if at least 5 years have passed since their previous dose. Those who receive it at or after age 65 should receive only a single dose. A second dose is recommended 5 years after the first dose for people age 19 to 64 years with functional or anatomic asplenia and for those who are immunocompromised.

Influenza vaccination for all. Of note, the ACIP updated its guidelines on influenza vaccination beginning with the 2010–2011 influenza season. It no longer advocates a risk-stratified approach. Instead, it recommends universal influenza vaccination for everybody more than 6 months old.⁹⁴

Smoking cessation should be addressed. Smoking cessation is a Medicare and Medicaid quality measure and should be encouraged after an episode of CAP because quitting smoking reduces the risk of pneumococcal disease by approximately 14% each year thereafter.⁹⁵

General internists need to be able to recognize community-acquired pneumonia (CAP) so that diagnostic and therapeutic interventions can be initiated promptly. It is also important to understand the most likely and possible causes of CAP so that appropriate initial antimicrobial therapy can be chosen. Especially during flu season, influenza can present as CAP and should be included in the differential diagnosis.

When managing a patient with CAP, the internist must decide which level of care, diagnostic tests, antimicrobial agents, and follow-up plans are needed. These topics will be reviewed in this article.

TWO TERMS TO REMEMBER

• CAP refers to pneumonia acquired outside a health care facility. It can be either bacterial or viral.
• CABP (community-acquired bacterial pneumonia) refers only to those cases caused by bacterial pathogens.

NUMBERS AND TRENDS

In the United States, CAP is the number-one cause of death from infection and the sixth leading cause of death overall.¹ Each year, it is responsible for about 4.2 million outpatient visits, more than 60,000 deaths, and more than $17 billion in health care expenses.²

Community-acquired bacterial pneumonia: Common, serious

In a population-based US study in 1991, the incidence of CABP requiring hospitalization was 266.8 per 100,000 people.³

Estimates of overall mortality in CABP vary depending on the severity of illness and comorbid conditions. A meta-analysis published in 1996 found the overall mortality rate to be 13.7%, with a range of 5.1% to 36.5% depending on severity.⁴

In hospitalized patients, mortality rates and length of hospital stay appear to be declining over time. Between 1993 and 2005, the age-adjusted mortality rate decreased from 8.9% to 4.1%, and the average length of stay decreased from 7.5 to 5.7 days, with an overall reduction in hospital cost.⁵

CABP is more prevalent in older people than in the general population, and it increases with age from 18.2 cases per 1,000 patient-years in patients 60 to 69 years to 52.3 cases per 1,000 patient-years in those older than 85 years.⁶ Risk factors for pneumonia in the elderly include heart disease, chronic lung disease, immunosuppressive drugs, alcoholism, and increasing age.⁷ Similar to the trend in the general population, the mortality rate in elderly CABP patients appears to be decreasing over time, possibly thanks to rising rates of pneumococcal and influenza vaccination.⁸

Among the general population, risk factors for developing CABP also include smoking, occupational dust exposure, history of childhood pneumonia, unemployment, and single marital status.⁹ The incidence of CABP does not appear to be higher among pregnant women, although it is the most frequent cause of nonobstetric death in this population.¹⁰

The use of proton pump inhibitors may be an emerging risk factor for CABP.¹¹ Also, use of nonsteroidal anti-inflammatory drugs among patients with CABP is associated with a blunted inflammatory response as well as a higher risk of pleuropulmonary complications and a delay in presentation.¹²

Influenza is also common, potentially severe

Influenza is also very common and potentially severe. It can cause a spectrum of disease, from mild upper respiratory tract symptoms to severe viral pneumonia that can be life-threatening and complicated by respiratory failure and the acute respiratory distress syndrome (ARDS).

Influenza infection can also be complicated by subsequent bacterial pneumonia. However, the epidemiology of influenza infection differs from that of CABP in that influenza occurs seasonally.

In the United States, seasonal influenza causes 36,000 deaths and 200,000 hospitalizations annually.^13,14 As with CABP, the risk of death from influenza increases with age: it is 16 times greater in people age 85 and older than in those ages 65 to 69.¹³

During yearly seasonal epidemics, those at the highest risk of hospitalization and death are at the extremes of age. Risk factors for complicated influenza include heart disease, lung disease, diabetes, renal failure, rheumatologic conditions, dementia, and neurologic disease.^15,16 During the 2009 H1N1 influenza pandemic, unexpected severity was seen in previously healthy young adults as well as those with obesity, neurodegenerative disease, pregnancy, and asthma.¹⁷

PATHOGENS: TYPICAL, ATYPICAL, VIRAL

Identifying the etiologic organism in CAP is confounded by limitations in the available diagnostic tests and also by poor-quality specimens that often are contaminated with bacteria that colonize the upper airways. Given these caveats, the primary pathogens responsible for CAP broadly include typical bacterial pathogens, atypical bacterial pathogens, and viruses.

Typical bacterial pathogens include Streptococcus pneumoniae, Haemophilus influenzae, Staphylococcus aureus, Moraxella catarrhalis, and, less commonly, a variety of aerobic and anaerobic gram-negative rods including Pseudomonas aeruginosa, Acinetobacter species, and Klebsiella pneumoniae.

Atypical bacterial pathogens include Mycoplasma pneumoniae, Chlamydia pneumoniae, and Legionella species.¹⁸

Viruses implicated in adult CAP include influenza A and B, parainfluenza viruses, respiratory syncytial virus, and adenovirus.¹⁹ More recently, human metapneumovirus has been described as a cause of adult CAP.²⁰

Clues to uncommon microbes

Specific historic features or coexisting conditions that may suggest an uncommon microbiologic diagnosis include²¹:

• Recent travel to the southwestern United States or Southeast Asia
• Ill contacts
• Exposure to birds, bats, rabbits, or farm animals
• Alcoholism
• Chronic obstructive pulmonary disease
• Human immunodeficiency virus infection
• Structural lung disease
• Prolonged cough with whoop or posttussive vomiting
• Aspiration
• Bioterrorism.

In cases in which one or more of these conditions exist, CAP may also be caused by other agents not listed above, including Mycobacterium tuberculosis, oral anaerobes, atypical mycobacteria, Histoplasma capsulatum, Chlamydophila psittaci, Francisella tularensis, Coxiella burnettii, Pneumocystis jiroveci, Cryptococcus, Aspergillus, Coccidioides, Hantavirus, avian influenza, Burkholderia pseudomallei, severe acute respiratory syndrome virus, Bordetella pertussis, Bacillus anthracis, and Yersinia pestis.

 

HOW BACTERIA INVADE THE LUNGS

The pathophysiology of CABP involves both host defense and microbial virulence factors.

The airways are most commonly exposed to microbes by microaspiration of upper airway flora, although hematogenous seeding of the lungs in a bacteremic patient or contiguous spread of infection from an adjacent site can also occur.

Mucociliary clearance and the cough reflex are important initial defenses against infection and can be inhibited by neurologic diseases and conditions that impair the mucociliary mechanism. Mucosal immune cells, including macrophages and neutrophils, recognize invading pathogens and generate an antibody response.

Regulation of the host inflammatory response to infection depends on a complex interaction between immune cells, inflammatory cytokines (eg, tumor necrosis factor alpha, interleukin 1-beta, and interleukin 6), and anti-inflammatory cytokines such as interleukin 1 receptor antagonist and soluble tumor necrosis factor receptor type I.²²

The interaction and timing of the inflammatory and anti-inflammatory response are essential in manifesting an appropriate host response to infection. An inadequate inflammatory response can lead to sepsis and death, but an excessive, late anti-inflammatory response can lead to a systemic inflammatory response such as ARDS. Polymorphisms within the genes coding for these factors may explain the variation in severity of illness among patients with CABP.²³

HOW INFLUENZA DOES ITS DAMAGE

There are three types of influenza virus: A, B, and C. Type A causes most human infections. The influenza A virus envelope comprises a lipid bilayer that contains the projecting glycoproteins hemagglutinin and neuraminidase. Influenza viruses are named on the basis of these proteins and are designated with an H and an N, respectively, each followed by a number referring to the subtype.

Influenza infection begins when the virus makes contact with the epithelium. Hemagglutinin binds to the host cell and allows viral entry, where it begins replication. Neuraminidase prevents viral aggregation and facilitates the release of virus from infected cells.²⁴

Mature virions are released and spread to neighboring host cells; this process is associated with desquamation and inflammation of the airways, causing cough, rhinorrhea, and sore throat. Systemic symptoms are associated with the induction of interferon, which causes fever and myalgia.²⁵

Recovery and immunity to influenza infection occurs through both humoral and cell-mediated immunity, with antibodies directed against the specific hemagglutinin and neuraminidase antigens of the infecting virus. Immunity wanes over time and with antigenic drift of circulating viruses, making the host susceptible to recurrent influenza infection.²⁴

Influenza is often complicated by bacterial superinfection

The influenza virus acts synergistically with certain bacteria to increase infectivity, and this may explain why influenza is often complicated by bacterial superinfection.

Mechanisms leading to bacterial superinfection include increased binding and invasion of bacteria, increased viral replication, and modification of the host inflammatory response. Some S aureus strains produce a protease that directly activates influenza virus hemagglutinin; other bacteria can activate plasminogen to promote influenza replication. The resulting increase in proteases in host tissues promotes activation of influenza through cleavage of hemagglutinin.²⁶

The influenza virus also causes damage to the airway epithelial layer, leading to increased exposure of the binding sites necessary for adherence of S pneumoniae.²⁷

CLINICAL PRESENTATION OF COMMUNITY-ACQUIRED PNEUMONIA

Although CAP is common, agreement on its essential clinical signs and symptoms is surprisingly limited, due in part to heterogeneous patient presentations and in part to interobserver variability. The reader is referred to two excellent reviews on this topic.^28,29

The diagnosis of CAP is made on clinical grounds, based on a combination of signs and symptoms. Symptoms of pneumonia can include cough, fever, chills, sputum production, dyspnea, and pleuritic pain. Physical findings can include tachypnea, tachycardia, hypoxemia, and consolidation or rales on auscultation. Laboratory data may show leukocytosis or elevated C-reactive protein, and radiographic studies may show evidence of a new infiltrate.^21,30,31

Clinical presentation of influenza

Seasonal influenza as a cause of CAP is difficult to distinguish from bacterial causes. The clinical presentation of seasonal influenza most commonly includes fever or subjective feverishness, cough, myalgia, and weakness.³² In a recent multivariate analysis, five clinical features were shown to be predictive of pandemic H1N1 influenza pneumonia rather than CABP: age younger than 65 years, absence of confusion, white blood cell count less than 12 × 10⁹/L, temperature higher than 38°C (100.4°F), and bilateral opacities on radiography.^32,33

Complicated influenza infection can be either primary viral pneumonia or bacterial superinfection.

During the 1918 influenza pandemic, which predated the ability to isolate viruses, two clinical syndromes emerged: an ARDS associated with the rapid onset of cyanosis, delirium, and frothy blood-tinged sputum; and an acute bronchopneumonia characterized by necrosis, hemorrhage, edema, and vasculitis.^34,35 The first syndrome has subsequently been shown to be associated with primary viral pneumonia, while the second is caused by bacterial superinfection. Modern reexamination of 1918 data has shown that bacterial superinfection was likely the reason for the distinctly fulminant presentation of that pandemic.^36,37

The 2009 H1N1 influenza pandemic caused relatively mild disease in most patients. However, those with severe pneumonia more commonly developed ARDS from primary influenza pneumonia than from bacterial superinfection.¹⁷

A third influenza-associated infection is secondary bacterial pneumonia, which follows influenza infection and mimics the presentation of CABP. A typical patient presents with a recent history of influenza-like illness, followed 4 to 14 days later by a recurrence of fever, dyspnea, productive cough, and consolidation on chest radiographs.³⁸ Leukocytosis with an increased number of immature neutrophil forms, prolonged duration of fever, and elevated erythrocyte sedimentation rate are more likely in patients with secondary bacterial pneumonia.³⁹ Isolates from sputum samples commonly include S pneumoniae, S aureus, H influenzae, and other gram-negative rods.⁴⁰

In recent flu seasons, methicillin-resistant S aureus (MRSA) has emerged as a cause of severe secondary pneumonia. Most of these isolates carry genes for the toxin Panton-Valentine leukocidin; the associated mortality rate is as high as 33%.^41,42 Although community-acquired MRSA pneumonia has only been reported in case series, distinct clinical features that have been described include severe pneumonia with high fever, hypotension, shock, respiratory failure, leukopenia, and multilobar and cavitary infiltrates.⁴³

WHEN TO SUSPECT INFLUENZA

The triad of fever, cough, and abrupt onset are the best predictors of influenza, but no single combination of signs and symptoms predict influenza infection with 100% certainty. Therefore, an understanding of local epidemiologic data regarding circulating influenza is essential to maintain a high index of suspicion.⁴⁴

It is appropriate to suspect influenza in:

• Anyone who is epidemiologically linked to a known outbreak of influenza
• Children, adults, and health care workers who have fever and abrupt onset of respiratory symptoms
• Patients with fever plus exacerbation of underlying pulmonary disease
• Severely ill patients with fever or hypothermia, especially during influenza season.⁴⁵

 

DIAGNOSTIC TESTING

Once the diagnosis of pulmonary infection is suggested by clinical features, the initial evaluation should include measurement of vital signs, physical examination, and radiographic imaging of the chest. Additional diagnostic measures to consider include viral testing, blood culture, sputum culture, urinary antigen testing for Legionella and for S pneumoniae, fungal culture, and mycobacterial smear and culture.

Chest radiography (with posterior-anterior and lateral films) is the study that usually demonstrates the presence of a pulmonary infiltrate. If initial chest radiographs do not show an infiltrate, imaging can be repeated after treatment is started if the patient’s clinical presentation still suggests pneumonia. Chest radiographs are of limited value in predicting the pathogen, but they help to determine the extent of pneumonia and to detect parapneumonic effusion.⁴⁶

A caveat: anterior-posterior, posterior-anterior, and lateral views can miss more than 10% of effusions large enough to warrant thoracentesis, especially when there is lower-lobe consolidation.⁴⁷

Blood cultures are recommended for patients admitted to the intensive care unit and for those with cavitary infiltrates, leukopenia, alcohol abuse, severe liver disease, asplenia, positive pneumococcal urinary antigen testing, or a pleural effusion.²¹ However, blood cultures are positive in only 3% to 14% of hospitalized patients with CABP, and the impact of a positive blood culture on management decisions in CABP appears to be quite small.^48–50

For the highest yield, blood culture results should be obtained before antibiotics are given. Not only is this good practice, but obtaining blood culture results before starting antibiotics is one of the quality measures evaluated by the Center for Medicare and Medicaid Services.⁵¹

Sputum culture is considered optional for outpatients and patients with less-severe pneumonia.²¹ While it can provide a rapid diagnosis in certain cases, a good-quality sputum sample is obtained in only 39% to 54% of patients with CABP, yields a predominant morphotype in only 45% of cases, and provides a useful microbiologic diagnosis in only 14.4%.^52,53 Fungal and mycobacterial cultures are only indicated in certain situations such as cavitary infiltrates or immunosuppression.

Urinary antigen testing for Legionella and S pneumoniae should be done in patients with more severe illness and in those for whom outpatient therapy has failed.²¹S pneumoniae testing has been shown to allow early diagnosis of pneumococcal pneumonia in 26% more patients than with Gram staining, but it fails to identify 22% of the rapid diagnoses initially identified by Gram staining.⁵⁴ Thus, a sequential approach is reasonable, with urinary antigen testing for patients at high risk without useful results from sputum Gram staining. Also, recent data suggest that the pneumococcal urinary antigen test may allow optimization of antimicrobial therapy with good clinical outcomes.⁵⁵

Endotracheal tests. If the patient is intubated, collection of endotracheal aspirates, bronchoscopy, or nonbronchoscopic bronchial lavage (sometimes called “mini-BAL”) should be performed.

Thoracentesis and pleural fluid cultures should be done if a pleural effusion is found. Empyema, large or loculated effusions, and parapneumonic effusions with a pH lower than 7.20, glucose levels less than 3.4 mmol/L (60 mg/dL), or positive results on microbial staining or culture should be drained by chest tube or surgically.⁵⁶

Testing for influenza should be done if it will change the clinical management, such as the choice of antibiotic or infection control practices. Specimens should be obtained with either a nasopharyngeal swab or aspirate and tested with reverse transcriptase polymerase chain reaction, immunofluorescent staining, or rapid antigen detection, depending on local availability.⁴⁵

Inflammatory biomarkers such as C-reactive protein and procalcitonin have been receiving interest as ways to predict the etiology and prognosis of CAP and to guide therapy. Several studies have shown that C-reactive protein can help distinguish between CAP and bronchitis, with higher values suggesting more severe pneumonia and pneumonia caused by S pneumoniae or L pneumophila.⁵⁷ Procalcitonin may help discriminate between severe lower respiratory tract infections of bacterial and 2009 H1N1 origin, although less effectively than C-reactive protein. Low procalcitonin values, particularly when combined with low C-reactive protein levels, suggest that bacterial infection is unlikely.⁵⁸

RISK STRATIFICATION AND SITE-OF-CARE DECISIONS

Following a presumptive diagnosis of CAP, it is important to decide not only what treatment the patient will receive but whether he or she should be hospitalized. If the patient is to be admitted to the hospital, the clinician must also decide if his or her condition warrants intensive care.

Severity-of-illness scores

Several severity-of-illness scores and prognostic models have been validated for use in deciding on inpatient vs outpatient treatment and to aid in the decision of whether a patient with pneumonia should be admitted to an intensive care unit. The most extensively studied and widely used scoring systems are the Pneumonia Severity Index (PSI) (Table 1)⁵⁹ and the CURB-65 (Figure 1).⁶⁰

The PSI is the more complicated of the two, as it is based on 19 variables. Online calculators are available for the PSI (http://pda.ahrq.gov/clinic/psi/psicalc.asp) and the CURB-65 (http://www.mdcalc.com/curb-65-severity-score-community-acquiredpneumonia).

A recent meta-analysis compared the performance characteristics of the PSI and CURB-65 scores for predicting mortality in CAP and found no significant differences in overall test performance.⁶¹

Another meta-analysis found that the PSI was more sensitive than the CURB-65 and had a low false-negative rate, and so was better at showing which patients do not need to be hospitalized. Conversely, the CURB-65 was more specific and had a higher positive predictive value, and thus was more likely to correctly classify high-risk patients.⁶²

Other scoring systems that aid in deciding about hospital admission and level of care include the CRB-65⁶³ (which can be used instead of the CURB-65 if laboratory values are not available), SMART-COP,⁶⁴ and SCAP.^,65

Guidelines on when to admit to the intensive care unit

Guidelines from the Infectious Diseases Society of America (IDSA) and the American Thoracic Society (ATS) also provide guidance on when intensive care admission is advised,²¹ and their criteria were recently validated.⁶⁶ The guidelines advocate direct admission to the intensive care unit for patients requiring vasopressors or mechanical ventilation, and intensive care unit or high-level monitoring for patients with three of the following criteria for severe CAP²¹:

• Respiratory rate ≥ 30
• Pao₂/Fio₂ ratio ≤ 250
• Multilobar infiltrates
• Confusion or disorientation
• Uremia (blood urea nitrogen ≥ 20 mg/dL)
• Leukopenia (white blood cell count < 4.0 × 10⁹/L)
• Thrombocytopenia (platelet count < 100 × 10⁹/L)
• Hypothermia (core temperature < 36.0°C [96.8°F])
• Hypotension requiring aggressive fluid resuscitation.

None of these scoring systems or criteria is meant to replace clinical judgment. A recent study has suggested that an oxygen saturation of less than 92% is an appropriate threshold for hospital admission, in view of higher rates of adverse events in outpatients with saturations below this value.⁶⁷

 

TREATMENT

Multiple studies have shown that treatment of CAP in accordance with guidelines has led to improved clinical outcomes.^21,68–70

How fast must antibiotics be started?

Based on studies that showed a lower mortality rate when antibiotics were started sooner, Medicare and Medicaid adopted a quality measure calling for starting antibiotics within 4 hours in patients being admitted to the hospital.^50,71 However, several subsequent studies showed that the diagnosis of pneumonia is often incorrect and that rapid administration of antibiotics could lead to misdiagnosis, overuse of antibiotics, and a higher risk of Clostridium difficile infection.^72,73

The current IDSA/ATS guidelines²¹ recommend that the first antibiotic dose be given while the patient is still in the emergency department, but do not give a specific time within which it should be given. Medicare and Medicaid later updated their quality measure to antibiotic administration within 6 hours.

Which antibiotics should be used?

The selection of antimicrobial agent depends upon the patient’s severity of illness and comorbid conditions.

Although most studies of combination antibiotic therapy have been retrospective and observational, they suggest that a macrolide (ie, one of the “mycins”) added to a beta-lactam antibiotic is beneficial, possibly by covering atypical organisms or via anti-inflammatory action.^74–76 The choice of one antibiotic over another appears to be less important, and a recent Cochrane review concluded that there was no significant difference in efficacy among five antibiotic pairs studied.⁷⁷

Empiric outpatient treatment of a previously healthy patient with CAP and no risk factors for drug-resistant S pneumoniae should include either a macrolide (azithromycin [Zithromax], clarithromycin [Biaxin], or erythromycin) or doxycycline. If the patient has a chronic comorbid condition such as heart, lung, liver, or renal disease, diabetes mellitus, alcoholism, malignancy, asplenia, or immunosuppression or has received antimicrobials within the preceding 3 months, then treatment should include either a respiratory fluoroquinolone (moxifloxacin [Avelox] or levofloxacin [Levaquin]) or a beta-lactam plus a macrolide.²¹

Overall, published data suggest that the survival rate is about the same with fluoroquinolone monotherapy as with beta-lactam plus macrolide combination therapy, and better than with beta-lactam monotherapy.⁷⁸

Selection of antibiotics for inpatient treatment of CAP is influenced by severity of illness. Inpatients who do not require intensive care should be treated with either a respiratory fluoroquinolone or combination therapy with a beta-lactam (cefotaxime [Claforan], ceftriaxone [Rocephin], ampicillin, or ertapenem [Invanz]) plus a macrolide or doxycycline.^21,76,79

If a specific microbiologic diagnosis is made, then treatment can be narrowed. However in certain cases, such as invasive pneumococcal infection, combination therapy may still be superior.^80,81 For patients who need intensive care, treatment should always include a beta-lactam plus either azithromycin or a respiratory fluoroquinolone.²¹ In certain situations, additional antibiotics may be added as well, such as agents to treat Pseudomonas, community-acquired MRSA, or both.

Switching to oral therapy; short-course therapy

In the interest of avoiding unnecessary antibiotics, numerous studies have addressed the issue of an “early switch” to oral antibiotics and “short-course” therapy for CAP. In general, once clinically stable, patients with CAP, including bacteremic S pneumoniae pneumonia, can be safely switched to oral antibiotics.⁸²

The issue of short-course therapy is more complicated, and the appropriate length of therapy for CAP is not well established. However, 5 days of levofloxacin 750 mg was shown to be as successful as 7 to 10 days of levofloxacin 500 mg.⁸³ In another study, in patients who improved after 3 days of intravenous therapy for CAP, there was no difference in clinical outcome between those who were changed to oral therapy for 5 more days and those who received an oral placebo.⁸⁴

Most patients who achieve clinical stability in the first week do not need prolonged antibiotic therapy. However, certain conditions, such as S aureus bacteremic pneumonia, complicated pneumonia, and pneumonia due to unusual organisms, may require prolonged treatment.

Other therapies

Additional therapies studied in patients with pneumonia include early mobilization, adjunctive corticosteroids, and statin drugs.

Early mobilization was shown in one study to decrease hospital length of stay without increasing adverse effects.⁸⁵

Corticosteroids are not supported as a standard of care for patients with severe CAP according to current available studies.^86,87 Furthermore, a randomized, controlled trial showed that prednisolone daily for a week did not improve outcomes in hospitalized patients with CAP, and it was associated with increased late failure.⁸⁸

Statin trials under way. Several observational studies have suggested that statins might be beneficial in managing sepsis through their effects on endothelial cell function, antioxidant effects, anti-inflammatory effects, and immunomodulatory effects.⁸⁹ However, a recent large prospective multicenter cohort study of hospitalized patients with CAP did not find evidence of a protective effect of statins on clinically meaningful outcomes in CAP or significant differences in circulating biomarkers.⁹⁰ Several randomized trials of statin therapy in patients with both ventilator-associated pneumonia and CAP are under way.

INFLUENZA TREATMENT: MOST EFFECTIVE WITHIN 48 HOURS

Treatment with antiviral drugs is most effective if started within 48 hours after symptom onset, although some patients with confirmed influenza who are either not improving or who are critically ill may still benefit from treatment started later.

Treatment should be considered in patients with laboratory-confirmed or suspected influenza who are at risk of developing complicated influenza and in otherwise healthy patients who wish to reduce the duration of illness or who have close contact with patients who are at high risk of complications.

Antiviral medications are oseltamivir (Tamiflu), zanamivir (Relenza), and the adamantines amantadine (Symmetrel) and rimantadine (Flumadine).

Due to evolving viral resistance patterns, the choice of antiviral drug depends on the strain. Seasonal H1N1 is best treated with zanamivir or an adamantine, while pandemic 2009 H1N1 and H3N2 are best treated with zanamivir or oseltamivir. When strain typing is not available, empiric therapy should be with either zanamivir monotherapy or a combination of oseltamivir plus rimantadine. Influenza B viruses are resistant to adamantines and should be treated only with either zanamivir or oseltamivir.⁴⁵

 

FOLLOW-UP AND PREVENTION

Patients with CAP can generally be expected to improve within 3 to 7 days.⁹¹ However, it may be several weeks before they return to baseline.⁹²

Follow-up plans may be guided by the time to clinical stability. For patients who do not achieve clinical stability until more than 72 hours after admission, more aggressive follow-up on discharge is indicated, since they are more likely to experience early readmission and death.⁹³

Pneumococcal vaccination. Because S pneumoniae remains the most common cause of CAP, efforts should be made to vaccinate patients appropriately. The Advisory Committee on Immunization Practices (ACIP) and the US Centers for Disease Control and Prevention recommend that the pneumococcal polysaccharide vaccine (Pneumovax 23; PPSV23) be given to those over age 65. Those who were vaccinated before age 65 should receive another dose at age 65 or later if at least 5 years have passed since their previous dose. Those who receive it at or after age 65 should receive only a single dose. A second dose is recommended 5 years after the first dose for people age 19 to 64 years with functional or anatomic asplenia and for those who are immunocompromised.

Influenza vaccination for all. Of note, the ACIP updated its guidelines on influenza vaccination beginning with the 2010–2011 influenza season. It no longer advocates a risk-stratified approach. Instead, it recommends universal influenza vaccination for everybody more than 6 months old.⁹⁴

Smoking cessation should be addressed. Smoking cessation is a Medicare and Medicaid quality measure and should be encouraged after an episode of CAP because quitting smoking reduces the risk of pneumococcal disease by approximately 14% each year thereafter.⁹⁵

High blood pressure takes a devastating toll on African Americans. Better control can go a long way to closing the “mortality gap” between African Americans and white Americans. But which strategies are best to address this complex problem?

In this report, we review the evidence on practice-based approaches to improving blood pressure control, from new styles of patient education to home blood pressure monitoring, focusing on studies in African Americans (Table 1).^1–11

BETTER CONTROL IS NEEDED

Better control of hypertension is certainly needed. In the United States, African Americans have disparately high rates of cardiovascular disease and death from cardiovascular disease.¹² (In this review, “African American” refers to non-Hispanic blacks, and “whites” refers to non-Hispanic whites.) According to the National Health and Nutrition Examination Survey (NHANES), from 1988 to 2008 the overall age-adjusted prevalence of hypertension in African Americans was 40%, vs 30% in whites.¹³ Partly because of this, African Americans have worse hypertension-related outcomes, including higher rates of fatal stroke, heart disease, end-stage kidney disease, and death compared with whites.^14–18 Thus, hypertension is the single most common contributor to the mortality gap between African Americans and white Americans.¹⁹

Fortunately, clinical research has shown that better control of blood pressure produces cardiovascular benefits in African Americans.²⁰ To date, however, the primary care treatment of hypertension in African Americans is suboptimal due to patient-related factors, to physician practice factors, and also to barriers in the health care system (Table 2).^21–23

PATIENT-RELATED BARRIERS

Patient-related barriers^24–40 include:

• Poor knowledge about hypertension and its consequences^31,32
• Poor adherence to drug therapy (a major factor,^24–26 as African Americans have poorer adherence rates than whites,^27–29 which may explain some of the racial disparity in blood pressure control³⁰)
• False health beliefs^34–37
• Inability to change one’s lifestyle
• Side effects of antihypertensive drugs³²
• Unrealistic expectations of treatment (eg, a cure³³)
• Demographic factors (eg, socioeconomic status, educational level, age, sex).^24,38–40

Perhaps the most salient and easily modifiable of these factors are patients’ reluctance to modify their lifestyle and their misconceptions about the causes, treatment, and prevention of hypertension. Patients whose beliefs are discordant with traditional biomedical concepts of hypertension have poorer blood pressure control than those whose beliefs are concordant.⁴¹ This may be more relevant to African Americans, since they are known to have cultural health beliefs that differ from those of Western culture (eg, that hypertension is a curable rather than a chronic illness, and that hypertension is a disease of nerves that often affects the blood and clogs the arteries).⁴²

PHYSICIAN-RELATED BARRIERS

Barriers to effective blood pressure control at the physician level^43–48 include:

• Nonadherence to treatment guidelines⁴⁴
• Failure to intensify the regimen if goals are not met⁴⁵
• Failure to emphasize therapeutic lifestyle changes.^43,46–48

When primary care physicians do not follow evidence-based guidelines, the reason may be that they are not aware of them or that they do not understand them. In a national survey of 1,029 physicians that was designed to explore how well physicians know the indications for specific antihypertensive drugs and how closely their opinions and practice agreed with national guidelines, only 37.3% correctly answered all of the knowledge-related questions.⁴⁹

Other reasons for nonadherence are that physicians may disagree with the guidelines, may not be able to follow the guidelines, may not believe that following them will achieve the desired effect, or may have no motivation to change their practice.⁵⁰

Whatever the reason, Hyman et al⁵¹ reported that as many as 30% of physicians did not recommend treatment for patients with diastolic blood pressures of 90 to 100 mm Hg, and a higher proportion did not treat patients with systolic blood pressures of 140 to 160 mm Hg.

BARRIERS IN HEALTH CARE SYSTEMS

Although health care systems present barriers to optimal blood pressure control,^20,27,31,52 there is evidence that most cases of uncontrolled hypertension occur in patients with good access to care.^32,53,54 For example, an NHANES study⁵³ suggested that most patients with uncontrolled hypertension had in fact seen a physician on average at least three times in the previous year. And this may be more pervasive in African Americans: one survey found hypertension was uncontrolled in 75% of hypertensive African American patients despite free access to care, free medications, and regular follow-up visits.⁴¹

Thus, the most significant barriers to blood pressure control appear to be patient-related and physician-related.

INTERVENTIONS AIMED AT PATIENTS

The most common approaches to improving blood pressure control at the patient level, regardless of race, are patient education,^55–61 home blood pressure monitoring,^62–67 and behavioral counseling to address misconceptions about hypertension,⁶⁸ to improve adherence to drug therapy,^69–73 and to encourage lifestyle modifications.^74–78

 

Patient education

Patient education can improve blood pressure control.^58,79–82 Its aims are to increase patients’ understanding of the disease⁸³ and to encourage them to be more active in their own care.^80,84,85

Patient education has a moderate effect on blood pressure control. The average proportion of patients whose hypertension was under control in community-based trials of various interventions ranged from 60% to 70%, compared with 38% to 46% with usual care.^56,80,81

However, these strategies largely did not address misconceptions patients have about hypertension. This issue is especially critical in African Americans, who may have different perceptions of hypertension and different expectations for care⁴¹: beliefs that hypertension is “curable,” not chronic, and that medication is needed only for hypertension-related symptoms may translate to poorer rates of medication adherence.

Levine et al¹ evaluated the efficacy of home visits by trained community health advisory board workers in a neighborhood in Baltimore, MD, with a high prevalence of hypertension. Participants were randomized to receive either one visit or five visits during the 40-month study period. Both groups had a statistically significant reduction in blood pressure, and in both groups the proportion of patients with adequate blood pressure control increased significantly. The results support the use of a practice- and community-based partnership to improve blood pressure control in African American patients.

Ogedegbe et al² randomized 190 hypertensive African American patients to receive usual care or quarterly counseling sessions that used motivational interviewing focused on medication adherence. The counseled patients stayed adherent to their medications, whereas adherence declined significantly in those receiving usual care. This effect was associated with a modest, nonsignificant trend toward a net reduction in systolic blood pressure with motivational interviewing.

A novel method of health education is the use of narrative communication—ie, storytelling. It has a good amount of evidence to support it, as culturally appropriate storytelling may allow patients to identify with a story as it relates to their own lives.^86–89 Examples of educational storytelling include:

• A woman with hypertension discussing what it means to have high blood pressure, and the benefits of controlling it, such as living long enough to see her grandchildren grow up
• A man discussing the importance of involving family and friends to help control blood pressure, and how dietary modifications can be made to ensure that salt alternatives are used when the family does the cooking.

Storytelling should be done in a culturally appropriate context. For example, storytellers should have the same background as the patient (ie, similar socioeconomic status and ethnic background): patients are more likely to be influenced if they identify with the storyteller and imagine themselves in a similar situation.

Houston et al³ randomized 299 hypertensive African Americans to view either three DVDs that featured patients with hypertension or three “attention-control DVDs” on topics not related to hypertension. The intervention group’s DVDs focused on storytelling and “learning more.” In the storytelling section, patients told personal stories about what it meant to have hypertension and gave advice on how to best interact with health care providers and methods to improve medication adherence. A “learning more” section focused on what high blood pressure is, addressed therapeutic lifestyle changes, and encouraged patients to communicate with their health care providers. The patients who viewed the patient narratives had significantly lower blood pressure at 3 months than those assigned to usual care. Although blood pressure subsequently increased in both groups, the benefits of the intervention still existed at the end of follow-up.

Important to note about two of the above three studies^1,3 is that the interventions were done by people other than physicians, thus emphasizing the importance of a team approach to blood pressure control.

Behavioral counseling

The effectiveness of lifestyle modifications such as diet, weight loss, and physical activity in preventing and treating hypertension is well established.^74–78 For example:

• In the Dietary Approaches to Stop Hypertension (DASH) trial,⁷⁶ a healthy diet lowered blood pressure about as much as single drugs do, particularly in African Americans.
• The Trial of Nonpharmacologic Interventions in the Elderly (TONE)⁷⁴ showed that exercise can lower blood pressure in obese hypertensive patients.
• The PREMIER trial (Lifestyle Interventions for Blood Pressure Control)⁷⁵ showed that a single brief counseling session could produce substantial decreases in blood pressure in patients with stage 1 hypertension or high-normal blood pressure.

Unfortunately, these results have been hard to translate into primary care practice, especially for African American patients. Several studies have evaluated the impact of lifestyle interventions on blood pressure control in primary care practices with a large population of African American patients.

Bosworth et al,⁴ in a study of a practice in which almost half the patients were African American, randomized patients to receive usual care, nurse-administered tailored behavioral telephone counseling, home blood pressure monitoring, or home monitoring plus tailored behavioral telephone counseling. The combination of home monitoring and tailored behavioral telephone counseling led to a statistically significant improvement at 24 months compared with baseline.

Home blood pressure monitoring

The effectiveness of self-monitoring in improving blood pressure control is also well documented.^{62,63,65–67,90–95}

Pickering et al⁶² studied patients with poorly controlled hypertension in a managed-care setting and found a reduction of 7 mm Hg systolic and 5 mm Hg diastolic pressure after 3 to 6 months of home monitoring compared with usual care.

Mengden et al,⁹⁴ in a similar study, found average blood pressure reductions at 6 months of 19.3/11.9 mm Hg in the home-monitoring group vs 10.6/8.8 mm Hg in the usual-care group.

The effect of home blood pressure monitoring may be greater in African Americans.

Rogers et al⁹³ found it to be more effective at lowering blood pressure than usual care in a group of 121 patients with poorly controlled hypertension followed in primary care practices, and these reductions were twice as large in African American patients than in white patients.⁹³

Bondmass,⁹² in a study of 33 African American patients with poorly controlled hypertension, reported a 53% control rate within 4 weeks of home monitoring. All patients in the study had uncontrolled blood pressure at baseline (> 140/90 mm Hg).

Artinian et al⁵ evaluated the effect of nurse-managed telemonitoring on blood pressure control vs enhanced usual care. All participants were African American. The monitored group had a significantly greater reduction in systolic pressure at 12 months compared with those who received enhanced usual care.

 

PHYSICIAN-LEVEL INTERVENTIONS

Most interventions to improve how physicians manage patients with hypertension are designed to improve adherence to treatment guidelines. In most cases, these interventions are based on continuous quality improvement and disease management concepts such as physician education and academic detailing, reminders, feedback on performance measures, and risk-assessment tools.^96,97

Physician education

Interest is increasing in physician educational interventions for blood pressure control.^24,98

Inui et al,⁹⁹ in an early study in a primary care practice, found that patients of physicians who received tutorials on hypertension management were more compliant with their drug regimens and had better blood pressure control than patients of physicians in the control group.

Jennett et al,¹⁰⁰ in a similar randomized clinical trial, found that physicians who participated in an education activity were more adherent to treatment guidelines at 6 and 12 months compared with those who did not participate.

Maue et al¹⁰¹ showed that rates of blood pressure control improved from 41% to 52% after a 6-month educational intervention for physicians in a managed-care setting.

Tu et al¹⁰² reviewed 12 studies in which seven different physician educational interventions were used either alone or in combination and concluded that physician education improves compliance with guidelines for managing hypertension.

Unfortunately, these studies did not report outcomes separately for African American and white patients.

Hicks et al⁶ found that disease management approaches that target physicians whose patients with hypertension are mostly African American did not yield clinically relevant improvement in these patients, and that minority patients were significantly less likely to have their blood pressure controlled at the end of the study compared with their non-Hispanic white counterparts.

Feedback to providers

Several studies have shown that, given reminders and feedback systems, physicians will change their practice.^103–106

Mashru and Lant¹⁰⁴ combined chart audits and physician education in primary care practices and found they improved physician performance measures such as accuracy of diagnosis, number of patients who received cardiovascular risk assessment, and number of patients whose treatment was based on clinical laboratory assessments.

Feedback takes many forms but consists mostly of computerized information¹⁰⁷ or peer-to-peer academic detailing with opinion leaders.^108–110

Dickinson et al,¹⁰⁶ for instance, showed that computer-generated listings of patients’ blood pressures combined with a physician education program on clinical management of hypertension led to increased knowledge and better follow-up on their patients.

Again, however, these studies did not distinguish between African American and white patients, which makes it difficult to judge whether or not these approaches work differently for physicians with a large proportion of African American patients.

Computerized decision-support systems

Computerized decision-support systems have proliferated in primary care practices.¹¹¹

McAlister et al¹⁰³ found that general practitioners randomized to manage hypertension with the assistance of a computer obtained better outcomes than with usual care.

Montgomery and Fahey,¹⁰⁷ in a systematic review, found improved blood pressure control in two of the three trials that compared computer-generated feedback reports and reminders to usual care. Specifically, 51% of patients whose physicians received reminders either had controlled blood pressure or were at least receiving treatment vs 33% in the control group at 12 months. This difference was even higher at 24 months.

Montgomery et al⁷ later randomized primary care practices to use a computer-based decision-support system and a cardiovascular risk chart, the risk chart alone, or to continue as usual. Results indicated no reduction in cardiovascular risk in the computer-system or the chart-only group, whereas patients in the chart-only group had a significant reduction in systolic pressure and were prescribed more cardiovascular drugs. This study indicates that use of a computerized decision-support system is not superior to chart review and audit feedback alone.

Evidence that computerized decision systems improve blood pressure control in African Americans is scant. However, when one looks at the evidence from studies of African Americans, the outcomes do not seem to differ between African American and white patients.

Hicks et al⁶ examined the effectiveness of computerized decision support in improving hypertension care in a racially diverse population. Physicians were randomized to receive computerized decision support or to provide usual care without computerized support. Both groups improved significantly in prescribing appropriate drugs but not in overall blood pressure control. Furthermore, the study showed no reduction in racial disparities of care and blood pressure control.

A potential explanation for the lack of improvement in blood pressure was that the intervention dealt with making sure the appropriate drugs were prescribed rather than making sure physicians also appropriately intensified antihypertensive management when necessary.

 

INTERVENTIONS TARGETING PATIENTS AND PHYSICIANS

Several studies have targeted both patient and physician-level barriers to blood pressure control in practice-based settings.

Roumie et al⁸ randomized physicians to one of three intervention groups:

• “Provider education” consisting of an email message with a Web-based link to the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC-7)
• Provider education plus a computer alert with information about their patient’s blood pressure
• Provider education, a computer alert, and patient education (ie, patients received a letter encouraging adherence to drug therapy, changing their lifestyle, and talking with their doctor about their blood pressure).

Patients whose providers were randomized to the third group had better blood pressure control. The report did not differentiate African American vs white patients. The data, however, did show the effectiveness of adding patient education to provider education to improve blood pressure control.

Bosworth et al,¹¹² in a study in which 40% of patients were African American, randomized patients to usual care or to bimonthly nurse-delivered behavioral telephone counseling. They also randomized providers either to receive computer-generated decision support designed to improve adherence to guidelines or to receive no support.

There were no significant differences in rates of blood pressure control in the intervention groups compared with a control group. Although differences in blood pressure control between groups were not significant, patients randomized to behavioral intervention had significantly better blood pressure control at the 24-month follow-up than at baseline.

Svetkey et al⁹ evaluated the effects of physician intervention, patient intervention, and physician intervention plus patient intervention compared with control on systolic blood pressure at 6 months. They found that an intensive behavioral lifestyle intervention led to a significant reduction in systolic pressure at 6 months. By itself, the physician intervention did not have a meaningful effect, but patients in the combined physician-and-patient-intervention group experienced the greatest reduction (9.7 ± 12.7 mm Hg).

It takes a team

Physicians should not be the only focus in helping patients achieve blood pressure control. Although physician and patient factors need to be addressed to improve blood pressure control in African Americans, emphasis should also be placed on interdisciplinary, team-based care utilizing health care providers such as nurses, physician assistants, and pharmacists. Team-based care has been shown to have the greatest impact of all the strategies for improving blood pressure control.¹¹³ There is a good amount of evidence involving interventions with a focus on health care providers other than physicians, although the data lack a sufficient focus on African Americans.

Carter et al,¹⁰ in a randomized controlled trial in which 26.3% of the patients were African American, found that an intervention consisting of clinical pharmacists giving physicians drug therapy recommendations based on national guidelines resulted in a significantly lower blood pressure compared with a control group: the mean reduction was 20.7/9.7 mm Hg in the intervention group vs 6.8/4.5 mm Hg in the control group.

Carter et al¹¹⁴ performed a meta-analysis of 37 studies and found that two strategies led to a significant reduction in blood pressure: a pharmacist-led intervention with treatment recommendations to physicians resulted in a systolic pressure reduction of 9.30 mm Hg; and nurse-led interventions resulted in a systolic pressure reduction of 4.80 mm Hg. Again, many of the studies cited in this meta-analysis lacked a focus on African Americans.

Hunt et al¹¹ conducted a randomized controlled trial in which pharmacists actively participated in the management of blood pressure. They were involved with every aspect of care, including reviewing medications and adverse drug reactions, assessing lifestyle behaviors and barriers to adherence, making dosing adjustments, and adding medications. Patients randomized to the intervention group achieved significantly lower systolic and diastolic pressures (137/75 vs 143/78 mm Hg in the control group). However, information about race was not included.

The above studies are just a few out of a large body of evidence demonstrating the value of team-based care to improve blood pressure control. It has yet to be determined whether these models can improve blood pressure control specifically in African Americans, since so many of these trials lacked a focus on this group. Promising is an ongoing randomized prospective trial by Carter et al¹¹⁵ evaluating a model of collaboration between physicians and pharmacists, with a focus on patients in underrepresented minorities.

SO WHAT WORKS?

Although there is a growing body of literature on interventions to try to reduce disparities in hypertension and blood pressure control between African Americans and whites, only a few randomized controlled trials have focused on African Americans, and several have not reported their results.¹¹⁶ So the question remains: How should we interpret the available data, which are aggregated across racial groups, and put it into practice when caring for hypertensive African American patients?

Patient education. In trying to overcome patient-related barriers, emphasis should be on patient education, in particular addressing misconceptions about hypertension and promoting adherence to antihypertensive therapy. This is evident from the narrative storytelling intervention by Houston et al.³ Although this is the first study of its kind, this strategy may be something to consider if future studies replicate these findings. Culturally appropriate storytelling may allow patients to identify with the stories as they relate to their own personal lives. It can be an effective way to address patient education and change behaviors.

Self-monitoring with a home blood pressure monitor has also proven effective in the management of hypertension in African Americans. Indeed, the few studies that reported findings in African Americans showed impressive reductions in blood pressure. The benefits of home monitoring are well documented, and the effect on physician-related barriers such as clinical inertia are also quite impressive.¹¹⁷ However, most of these studies did not assess the long-term impact or cost-effectiveness of home monitoring on blood pressure control.

Behavioral counseling. Although we have good evidence of the effectiveness of behavioral counseling, whether this is sustained long-term has been less studied in African Americans. Thus, while interventions that targeted African Americans have reported impressive reductions in blood pressure, the effect tends to be greatest during the first few months of implementation, with the benefits disappearing over time.

Physician-related interventions. With regard to physician-level interventions, research has focused on physician education, utilizing alerts and computerized clinical decision-support systems. Evidence is scant on whether the use of computerized systems results in improves hypertension care in African Americans. However, a closer look at the data from studies that report outcomes in African American and white patients shows that the results do not seem to differ between these groups. Still, there is insufficient information about the impact on hypertensive African Americans.⁶

Strategies that address both patient- and physician-related barriers can improve overall blood pressure control; however, there is a lack of data comparing outcomes in hypertensive African Americans with those of whites, making it difficult to know if this would be an effective strategy in African American patients alone.

More studies needed that focus on African Americans

Developing interventions to improve blood pressure control in African Americans should be an ongoing priority for research if we intend to address racial disparities in cardiovascular disease. Although it is reassuring that there is a growing body of evidence and research with this focus,^118–121 more research is needed to determine effective strategies that address barriers related to physician practice and to the health care system overall as they relate to blood pressure control in African Americans. More importantly, these strategies should also emphasize a team-based approach that includes nurses, pharmacists, and physician assistants. Developing targeted interventions for hypertensive African Americans will help reduce disparities in the rates of cardiovascular illness and death in this patient population.

High blood pressure takes a devastating toll on African Americans. Better control can go a long way to closing the “mortality gap” between African Americans and white Americans. But which strategies are best to address this complex problem?

In this report, we review the evidence on practice-based approaches to improving blood pressure control, from new styles of patient education to home blood pressure monitoring, focusing on studies in African Americans (Table 1).^1–11

BETTER CONTROL IS NEEDED

Better control of hypertension is certainly needed. In the United States, African Americans have disparately high rates of cardiovascular disease and death from cardiovascular disease.¹² (In this review, “African American” refers to non-Hispanic blacks, and “whites” refers to non-Hispanic whites.) According to the National Health and Nutrition Examination Survey (NHANES), from 1988 to 2008 the overall age-adjusted prevalence of hypertension in African Americans was 40%, vs 30% in whites.¹³ Partly because of this, African Americans have worse hypertension-related outcomes, including higher rates of fatal stroke, heart disease, end-stage kidney disease, and death compared with whites.^14–18 Thus, hypertension is the single most common contributor to the mortality gap between African Americans and white Americans.¹⁹

Fortunately, clinical research has shown that better control of blood pressure produces cardiovascular benefits in African Americans.²⁰ To date, however, the primary care treatment of hypertension in African Americans is suboptimal due to patient-related factors, to physician practice factors, and also to barriers in the health care system (Table 2).^21–23

PATIENT-RELATED BARRIERS

Patient-related barriers^24–40 include:

• Poor knowledge about hypertension and its consequences^31,32
• Poor adherence to drug therapy (a major factor,^24–26 as African Americans have poorer adherence rates than whites,^27–29 which may explain some of the racial disparity in blood pressure control³⁰)
• False health beliefs^34–37
• Inability to change one’s lifestyle
• Side effects of antihypertensive drugs³²
• Unrealistic expectations of treatment (eg, a cure³³)
• Demographic factors (eg, socioeconomic status, educational level, age, sex).^24,38–40

Perhaps the most salient and easily modifiable of these factors are patients’ reluctance to modify their lifestyle and their misconceptions about the causes, treatment, and prevention of hypertension. Patients whose beliefs are discordant with traditional biomedical concepts of hypertension have poorer blood pressure control than those whose beliefs are concordant.⁴¹ This may be more relevant to African Americans, since they are known to have cultural health beliefs that differ from those of Western culture (eg, that hypertension is a curable rather than a chronic illness, and that hypertension is a disease of nerves that often affects the blood and clogs the arteries).⁴²

PHYSICIAN-RELATED BARRIERS

Barriers to effective blood pressure control at the physician level^43–48 include:

• Nonadherence to treatment guidelines⁴⁴
• Failure to intensify the regimen if goals are not met⁴⁵
• Failure to emphasize therapeutic lifestyle changes.^43,46–48

When primary care physicians do not follow evidence-based guidelines, the reason may be that they are not aware of them or that they do not understand them. In a national survey of 1,029 physicians that was designed to explore how well physicians know the indications for specific antihypertensive drugs and how closely their opinions and practice agreed with national guidelines, only 37.3% correctly answered all of the knowledge-related questions.⁴⁹

Other reasons for nonadherence are that physicians may disagree with the guidelines, may not be able to follow the guidelines, may not believe that following them will achieve the desired effect, or may have no motivation to change their practice.⁵⁰

Whatever the reason, Hyman et al⁵¹ reported that as many as 30% of physicians did not recommend treatment for patients with diastolic blood pressures of 90 to 100 mm Hg, and a higher proportion did not treat patients with systolic blood pressures of 140 to 160 mm Hg.

BARRIERS IN HEALTH CARE SYSTEMS

Although health care systems present barriers to optimal blood pressure control,^20,27,31,52 there is evidence that most cases of uncontrolled hypertension occur in patients with good access to care.^32,53,54 For example, an NHANES study⁵³ suggested that most patients with uncontrolled hypertension had in fact seen a physician on average at least three times in the previous year. And this may be more pervasive in African Americans: one survey found hypertension was uncontrolled in 75% of hypertensive African American patients despite free access to care, free medications, and regular follow-up visits.⁴¹

Thus, the most significant barriers to blood pressure control appear to be patient-related and physician-related.

INTERVENTIONS AIMED AT PATIENTS

The most common approaches to improving blood pressure control at the patient level, regardless of race, are patient education,^55–61 home blood pressure monitoring,^62–67 and behavioral counseling to address misconceptions about hypertension,⁶⁸ to improve adherence to drug therapy,^69–73 and to encourage lifestyle modifications.^74–78

 

Patient education

Patient education can improve blood pressure control.^58,79–82 Its aims are to increase patients’ understanding of the disease⁸³ and to encourage them to be more active in their own care.^80,84,85

Patient education has a moderate effect on blood pressure control. The average proportion of patients whose hypertension was under control in community-based trials of various interventions ranged from 60% to 70%, compared with 38% to 46% with usual care.^56,80,81

However, these strategies largely did not address misconceptions patients have about hypertension. This issue is especially critical in African Americans, who may have different perceptions of hypertension and different expectations for care⁴¹: beliefs that hypertension is “curable,” not chronic, and that medication is needed only for hypertension-related symptoms may translate to poorer rates of medication adherence.

Levine et al¹ evaluated the efficacy of home visits by trained community health advisory board workers in a neighborhood in Baltimore, MD, with a high prevalence of hypertension. Participants were randomized to receive either one visit or five visits during the 40-month study period. Both groups had a statistically significant reduction in blood pressure, and in both groups the proportion of patients with adequate blood pressure control increased significantly. The results support the use of a practice- and community-based partnership to improve blood pressure control in African American patients.

Ogedegbe et al² randomized 190 hypertensive African American patients to receive usual care or quarterly counseling sessions that used motivational interviewing focused on medication adherence. The counseled patients stayed adherent to their medications, whereas adherence declined significantly in those receiving usual care. This effect was associated with a modest, nonsignificant trend toward a net reduction in systolic blood pressure with motivational interviewing.

A novel method of health education is the use of narrative communication—ie, storytelling. It has a good amount of evidence to support it, as culturally appropriate storytelling may allow patients to identify with a story as it relates to their own lives.^86–89 Examples of educational storytelling include:

• A woman with hypertension discussing what it means to have high blood pressure, and the benefits of controlling it, such as living long enough to see her grandchildren grow up
• A man discussing the importance of involving family and friends to help control blood pressure, and how dietary modifications can be made to ensure that salt alternatives are used when the family does the cooking.

Storytelling should be done in a culturally appropriate context. For example, storytellers should have the same background as the patient (ie, similar socioeconomic status and ethnic background): patients are more likely to be influenced if they identify with the storyteller and imagine themselves in a similar situation.

Houston et al³ randomized 299 hypertensive African Americans to view either three DVDs that featured patients with hypertension or three “attention-control DVDs” on topics not related to hypertension. The intervention group’s DVDs focused on storytelling and “learning more.” In the storytelling section, patients told personal stories about what it meant to have hypertension and gave advice on how to best interact with health care providers and methods to improve medication adherence. A “learning more” section focused on what high blood pressure is, addressed therapeutic lifestyle changes, and encouraged patients to communicate with their health care providers. The patients who viewed the patient narratives had significantly lower blood pressure at 3 months than those assigned to usual care. Although blood pressure subsequently increased in both groups, the benefits of the intervention still existed at the end of follow-up.

Important to note about two of the above three studies^1,3 is that the interventions were done by people other than physicians, thus emphasizing the importance of a team approach to blood pressure control.

Behavioral counseling

The effectiveness of lifestyle modifications such as diet, weight loss, and physical activity in preventing and treating hypertension is well established.^74–78 For example:

• In the Dietary Approaches to Stop Hypertension (DASH) trial,⁷⁶ a healthy diet lowered blood pressure about as much as single drugs do, particularly in African Americans.
• The Trial of Nonpharmacologic Interventions in the Elderly (TONE)⁷⁴ showed that exercise can lower blood pressure in obese hypertensive patients.
• The PREMIER trial (Lifestyle Interventions for Blood Pressure Control)⁷⁵ showed that a single brief counseling session could produce substantial decreases in blood pressure in patients with stage 1 hypertension or high-normal blood pressure.

Unfortunately, these results have been hard to translate into primary care practice, especially for African American patients. Several studies have evaluated the impact of lifestyle interventions on blood pressure control in primary care practices with a large population of African American patients.

Bosworth et al,⁴ in a study of a practice in which almost half the patients were African American, randomized patients to receive usual care, nurse-administered tailored behavioral telephone counseling, home blood pressure monitoring, or home monitoring plus tailored behavioral telephone counseling. The combination of home monitoring and tailored behavioral telephone counseling led to a statistically significant improvement at 24 months compared with baseline.

Home blood pressure monitoring

The effectiveness of self-monitoring in improving blood pressure control is also well documented.^{62,63,65–67,90–95}

Pickering et al⁶² studied patients with poorly controlled hypertension in a managed-care setting and found a reduction of 7 mm Hg systolic and 5 mm Hg diastolic pressure after 3 to 6 months of home monitoring compared with usual care.

Mengden et al,⁹⁴ in a similar study, found average blood pressure reductions at 6 months of 19.3/11.9 mm Hg in the home-monitoring group vs 10.6/8.8 mm Hg in the usual-care group.

The effect of home blood pressure monitoring may be greater in African Americans.

Rogers et al⁹³ found it to be more effective at lowering blood pressure than usual care in a group of 121 patients with poorly controlled hypertension followed in primary care practices, and these reductions were twice as large in African American patients than in white patients.⁹³

Bondmass,⁹² in a study of 33 African American patients with poorly controlled hypertension, reported a 53% control rate within 4 weeks of home monitoring. All patients in the study had uncontrolled blood pressure at baseline (> 140/90 mm Hg).

Artinian et al⁵ evaluated the effect of nurse-managed telemonitoring on blood pressure control vs enhanced usual care. All participants were African American. The monitored group had a significantly greater reduction in systolic pressure at 12 months compared with those who received enhanced usual care.

 

PHYSICIAN-LEVEL INTERVENTIONS

Most interventions to improve how physicians manage patients with hypertension are designed to improve adherence to treatment guidelines. In most cases, these interventions are based on continuous quality improvement and disease management concepts such as physician education and academic detailing, reminders, feedback on performance measures, and risk-assessment tools.^96,97

Physician education

Interest is increasing in physician educational interventions for blood pressure control.^24,98

Inui et al,⁹⁹ in an early study in a primary care practice, found that patients of physicians who received tutorials on hypertension management were more compliant with their drug regimens and had better blood pressure control than patients of physicians in the control group.

Jennett et al,¹⁰⁰ in a similar randomized clinical trial, found that physicians who participated in an education activity were more adherent to treatment guidelines at 6 and 12 months compared with those who did not participate.

Maue et al¹⁰¹ showed that rates of blood pressure control improved from 41% to 52% after a 6-month educational intervention for physicians in a managed-care setting.

Tu et al¹⁰² reviewed 12 studies in which seven different physician educational interventions were used either alone or in combination and concluded that physician education improves compliance with guidelines for managing hypertension.

Unfortunately, these studies did not report outcomes separately for African American and white patients.

Hicks et al⁶ found that disease management approaches that target physicians whose patients with hypertension are mostly African American did not yield clinically relevant improvement in these patients, and that minority patients were significantly less likely to have their blood pressure controlled at the end of the study compared with their non-Hispanic white counterparts.

Feedback to providers

Several studies have shown that, given reminders and feedback systems, physicians will change their practice.^103–106

Mashru and Lant¹⁰⁴ combined chart audits and physician education in primary care practices and found they improved physician performance measures such as accuracy of diagnosis, number of patients who received cardiovascular risk assessment, and number of patients whose treatment was based on clinical laboratory assessments.

Feedback takes many forms but consists mostly of computerized information¹⁰⁷ or peer-to-peer academic detailing with opinion leaders.^108–110

Dickinson et al,¹⁰⁶ for instance, showed that computer-generated listings of patients’ blood pressures combined with a physician education program on clinical management of hypertension led to increased knowledge and better follow-up on their patients.

Again, however, these studies did not distinguish between African American and white patients, which makes it difficult to judge whether or not these approaches work differently for physicians with a large proportion of African American patients.

Computerized decision-support systems

Computerized decision-support systems have proliferated in primary care practices.¹¹¹

McAlister et al¹⁰³ found that general practitioners randomized to manage hypertension with the assistance of a computer obtained better outcomes than with usual care.

Montgomery and Fahey,¹⁰⁷ in a systematic review, found improved blood pressure control in two of the three trials that compared computer-generated feedback reports and reminders to usual care. Specifically, 51% of patients whose physicians received reminders either had controlled blood pressure or were at least receiving treatment vs 33% in the control group at 12 months. This difference was even higher at 24 months.

Montgomery et al⁷ later randomized primary care practices to use a computer-based decision-support system and a cardiovascular risk chart, the risk chart alone, or to continue as usual. Results indicated no reduction in cardiovascular risk in the computer-system or the chart-only group, whereas patients in the chart-only group had a significant reduction in systolic pressure and were prescribed more cardiovascular drugs. This study indicates that use of a computerized decision-support system is not superior to chart review and audit feedback alone.

Evidence that computerized decision systems improve blood pressure control in African Americans is scant. However, when one looks at the evidence from studies of African Americans, the outcomes do not seem to differ between African American and white patients.

Hicks et al⁶ examined the effectiveness of computerized decision support in improving hypertension care in a racially diverse population. Physicians were randomized to receive computerized decision support or to provide usual care without computerized support. Both groups improved significantly in prescribing appropriate drugs but not in overall blood pressure control. Furthermore, the study showed no reduction in racial disparities of care and blood pressure control.

A potential explanation for the lack of improvement in blood pressure was that the intervention dealt with making sure the appropriate drugs were prescribed rather than making sure physicians also appropriately intensified antihypertensive management when necessary.

 

INTERVENTIONS TARGETING PATIENTS AND PHYSICIANS

Several studies have targeted both patient and physician-level barriers to blood pressure control in practice-based settings.

Roumie et al⁸ randomized physicians to one of three intervention groups:

• “Provider education” consisting of an email message with a Web-based link to the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC-7)
• Provider education plus a computer alert with information about their patient’s blood pressure
• Provider education, a computer alert, and patient education (ie, patients received a letter encouraging adherence to drug therapy, changing their lifestyle, and talking with their doctor about their blood pressure).

Patients whose providers were randomized to the third group had better blood pressure control. The report did not differentiate African American vs white patients. The data, however, did show the effectiveness of adding patient education to provider education to improve blood pressure control.

Bosworth et al,¹¹² in a study in which 40% of patients were African American, randomized patients to usual care or to bimonthly nurse-delivered behavioral telephone counseling. They also randomized providers either to receive computer-generated decision support designed to improve adherence to guidelines or to receive no support.

There were no significant differences in rates of blood pressure control in the intervention groups compared with a control group. Although differences in blood pressure control between groups were not significant, patients randomized to behavioral intervention had significantly better blood pressure control at the 24-month follow-up than at baseline.

Svetkey et al⁹ evaluated the effects of physician intervention, patient intervention, and physician intervention plus patient intervention compared with control on systolic blood pressure at 6 months. They found that an intensive behavioral lifestyle intervention led to a significant reduction in systolic pressure at 6 months. By itself, the physician intervention did not have a meaningful effect, but patients in the combined physician-and-patient-intervention group experienced the greatest reduction (9.7 ± 12.7 mm Hg).

It takes a team

Physicians should not be the only focus in helping patients achieve blood pressure control. Although physician and patient factors need to be addressed to improve blood pressure control in African Americans, emphasis should also be placed on interdisciplinary, team-based care utilizing health care providers such as nurses, physician assistants, and pharmacists. Team-based care has been shown to have the greatest impact of all the strategies for improving blood pressure control.¹¹³ There is a good amount of evidence involving interventions with a focus on health care providers other than physicians, although the data lack a sufficient focus on African Americans.

Carter et al,¹⁰ in a randomized controlled trial in which 26.3% of the patients were African American, found that an intervention consisting of clinical pharmacists giving physicians drug therapy recommendations based on national guidelines resulted in a significantly lower blood pressure compared with a control group: the mean reduction was 20.7/9.7 mm Hg in the intervention group vs 6.8/4.5 mm Hg in the control group.

Carter et al¹¹⁴ performed a meta-analysis of 37 studies and found that two strategies led to a significant reduction in blood pressure: a pharmacist-led intervention with treatment recommendations to physicians resulted in a systolic pressure reduction of 9.30 mm Hg; and nurse-led interventions resulted in a systolic pressure reduction of 4.80 mm Hg. Again, many of the studies cited in this meta-analysis lacked a focus on African Americans.

Hunt et al¹¹ conducted a randomized controlled trial in which pharmacists actively participated in the management of blood pressure. They were involved with every aspect of care, including reviewing medications and adverse drug reactions, assessing lifestyle behaviors and barriers to adherence, making dosing adjustments, and adding medications. Patients randomized to the intervention group achieved significantly lower systolic and diastolic pressures (137/75 vs 143/78 mm Hg in the control group). However, information about race was not included.

The above studies are just a few out of a large body of evidence demonstrating the value of team-based care to improve blood pressure control. It has yet to be determined whether these models can improve blood pressure control specifically in African Americans, since so many of these trials lacked a focus on this group. Promising is an ongoing randomized prospective trial by Carter et al¹¹⁵ evaluating a model of collaboration between physicians and pharmacists, with a focus on patients in underrepresented minorities.

SO WHAT WORKS?

Although there is a growing body of literature on interventions to try to reduce disparities in hypertension and blood pressure control between African Americans and whites, only a few randomized controlled trials have focused on African Americans, and several have not reported their results.¹¹⁶ So the question remains: How should we interpret the available data, which are aggregated across racial groups, and put it into practice when caring for hypertensive African American patients?

Patient education. In trying to overcome patient-related barriers, emphasis should be on patient education, in particular addressing misconceptions about hypertension and promoting adherence to antihypertensive therapy. This is evident from the narrative storytelling intervention by Houston et al.³ Although this is the first study of its kind, this strategy may be something to consider if future studies replicate these findings. Culturally appropriate storytelling may allow patients to identify with the stories as they relate to their own personal lives. It can be an effective way to address patient education and change behaviors.

Self-monitoring with a home blood pressure monitor has also proven effective in the management of hypertension in African Americans. Indeed, the few studies that reported findings in African Americans showed impressive reductions in blood pressure. The benefits of home monitoring are well documented, and the effect on physician-related barriers such as clinical inertia are also quite impressive.¹¹⁷ However, most of these studies did not assess the long-term impact or cost-effectiveness of home monitoring on blood pressure control.

Behavioral counseling. Although we have good evidence of the effectiveness of behavioral counseling, whether this is sustained long-term has been less studied in African Americans. Thus, while interventions that targeted African Americans have reported impressive reductions in blood pressure, the effect tends to be greatest during the first few months of implementation, with the benefits disappearing over time.

Physician-related interventions. With regard to physician-level interventions, research has focused on physician education, utilizing alerts and computerized clinical decision-support systems. Evidence is scant on whether the use of computerized systems results in improves hypertension care in African Americans. However, a closer look at the data from studies that report outcomes in African American and white patients shows that the results do not seem to differ between these groups. Still, there is insufficient information about the impact on hypertensive African Americans.⁶

Strategies that address both patient- and physician-related barriers can improve overall blood pressure control; however, there is a lack of data comparing outcomes in hypertensive African Americans with those of whites, making it difficult to know if this would be an effective strategy in African American patients alone.

More studies needed that focus on African Americans

Developing interventions to improve blood pressure control in African Americans should be an ongoing priority for research if we intend to address racial disparities in cardiovascular disease. Although it is reassuring that there is a growing body of evidence and research with this focus,^118–121 more research is needed to determine effective strategies that address barriers related to physician practice and to the health care system overall as they relate to blood pressure control in African Americans. More importantly, these strategies should also emphasize a team-based approach that includes nurses, pharmacists, and physician assistants. Developing targeted interventions for hypertensive African Americans will help reduce disparities in the rates of cardiovascular illness and death in this patient population.

High blood pressure takes a devastating toll on African Americans. Better control can go a long way to closing the “mortality gap” between African Americans and white Americans. But which strategies are best to address this complex problem?

In this report, we review the evidence on practice-based approaches to improving blood pressure control, from new styles of patient education to home blood pressure monitoring, focusing on studies in African Americans (Table 1).^1–11

BETTER CONTROL IS NEEDED

Better control of hypertension is certainly needed. In the United States, African Americans have disparately high rates of cardiovascular disease and death from cardiovascular disease.¹² (In this review, “African American” refers to non-Hispanic blacks, and “whites” refers to non-Hispanic whites.) According to the National Health and Nutrition Examination Survey (NHANES), from 1988 to 2008 the overall age-adjusted prevalence of hypertension in African Americans was 40%, vs 30% in whites.¹³ Partly because of this, African Americans have worse hypertension-related outcomes, including higher rates of fatal stroke, heart disease, end-stage kidney disease, and death compared with whites.^14–18 Thus, hypertension is the single most common contributor to the mortality gap between African Americans and white Americans.¹⁹

Fortunately, clinical research has shown that better control of blood pressure produces cardiovascular benefits in African Americans.²⁰ To date, however, the primary care treatment of hypertension in African Americans is suboptimal due to patient-related factors, to physician practice factors, and also to barriers in the health care system (Table 2).^21–23

PATIENT-RELATED BARRIERS

Patient-related barriers^24–40 include:

• Poor knowledge about hypertension and its consequences^31,32
• Poor adherence to drug therapy (a major factor,^24–26 as African Americans have poorer adherence rates than whites,^27–29 which may explain some of the racial disparity in blood pressure control³⁰)
• False health beliefs^34–37
• Inability to change one’s lifestyle
• Side effects of antihypertensive drugs³²
• Unrealistic expectations of treatment (eg, a cure³³)
• Demographic factors (eg, socioeconomic status, educational level, age, sex).^24,38–40

Perhaps the most salient and easily modifiable of these factors are patients’ reluctance to modify their lifestyle and their misconceptions about the causes, treatment, and prevention of hypertension. Patients whose beliefs are discordant with traditional biomedical concepts of hypertension have poorer blood pressure control than those whose beliefs are concordant.⁴¹ This may be more relevant to African Americans, since they are known to have cultural health beliefs that differ from those of Western culture (eg, that hypertension is a curable rather than a chronic illness, and that hypertension is a disease of nerves that often affects the blood and clogs the arteries).⁴²

PHYSICIAN-RELATED BARRIERS

Barriers to effective blood pressure control at the physician level^43–48 include:

• Nonadherence to treatment guidelines⁴⁴
• Failure to intensify the regimen if goals are not met⁴⁵
• Failure to emphasize therapeutic lifestyle changes.^43,46–48

When primary care physicians do not follow evidence-based guidelines, the reason may be that they are not aware of them or that they do not understand them. In a national survey of 1,029 physicians that was designed to explore how well physicians know the indications for specific antihypertensive drugs and how closely their opinions and practice agreed with national guidelines, only 37.3% correctly answered all of the knowledge-related questions.⁴⁹

Other reasons for nonadherence are that physicians may disagree with the guidelines, may not be able to follow the guidelines, may not believe that following them will achieve the desired effect, or may have no motivation to change their practice.⁵⁰

Whatever the reason, Hyman et al⁵¹ reported that as many as 30% of physicians did not recommend treatment for patients with diastolic blood pressures of 90 to 100 mm Hg, and a higher proportion did not treat patients with systolic blood pressures of 140 to 160 mm Hg.

BARRIERS IN HEALTH CARE SYSTEMS

Although health care systems present barriers to optimal blood pressure control,^20,27,31,52 there is evidence that most cases of uncontrolled hypertension occur in patients with good access to care.^32,53,54 For example, an NHANES study⁵³ suggested that most patients with uncontrolled hypertension had in fact seen a physician on average at least three times in the previous year. And this may be more pervasive in African Americans: one survey found hypertension was uncontrolled in 75% of hypertensive African American patients despite free access to care, free medications, and regular follow-up visits.⁴¹

Thus, the most significant barriers to blood pressure control appear to be patient-related and physician-related.

INTERVENTIONS AIMED AT PATIENTS

The most common approaches to improving blood pressure control at the patient level, regardless of race, are patient education,^55–61 home blood pressure monitoring,^62–67 and behavioral counseling to address misconceptions about hypertension,⁶⁸ to improve adherence to drug therapy,^69–73 and to encourage lifestyle modifications.^74–78

 

Patient education

Patient education can improve blood pressure control.^58,79–82 Its aims are to increase patients’ understanding of the disease⁸³ and to encourage them to be more active in their own care.^80,84,85

Patient education has a moderate effect on blood pressure control. The average proportion of patients whose hypertension was under control in community-based trials of various interventions ranged from 60% to 70%, compared with 38% to 46% with usual care.^56,80,81

However, these strategies largely did not address misconceptions patients have about hypertension. This issue is especially critical in African Americans, who may have different perceptions of hypertension and different expectations for care⁴¹: beliefs that hypertension is “curable,” not chronic, and that medication is needed only for hypertension-related symptoms may translate to poorer rates of medication adherence.

Levine et al¹ evaluated the efficacy of home visits by trained community health advisory board workers in a neighborhood in Baltimore, MD, with a high prevalence of hypertension. Participants were randomized to receive either one visit or five visits during the 40-month study period. Both groups had a statistically significant reduction in blood pressure, and in both groups the proportion of patients with adequate blood pressure control increased significantly. The results support the use of a practice- and community-based partnership to improve blood pressure control in African American patients.

Ogedegbe et al² randomized 190 hypertensive African American patients to receive usual care or quarterly counseling sessions that used motivational interviewing focused on medication adherence. The counseled patients stayed adherent to their medications, whereas adherence declined significantly in those receiving usual care. This effect was associated with a modest, nonsignificant trend toward a net reduction in systolic blood pressure with motivational interviewing.

A novel method of health education is the use of narrative communication—ie, storytelling. It has a good amount of evidence to support it, as culturally appropriate storytelling may allow patients to identify with a story as it relates to their own lives.^86–89 Examples of educational storytelling include:

• A woman with hypertension discussing what it means to have high blood pressure, and the benefits of controlling it, such as living long enough to see her grandchildren grow up
• A man discussing the importance of involving family and friends to help control blood pressure, and how dietary modifications can be made to ensure that salt alternatives are used when the family does the cooking.

Storytelling should be done in a culturally appropriate context. For example, storytellers should have the same background as the patient (ie, similar socioeconomic status and ethnic background): patients are more likely to be influenced if they identify with the storyteller and imagine themselves in a similar situation.

Houston et al³ randomized 299 hypertensive African Americans to view either three DVDs that featured patients with hypertension or three “attention-control DVDs” on topics not related to hypertension. The intervention group’s DVDs focused on storytelling and “learning more.” In the storytelling section, patients told personal stories about what it meant to have hypertension and gave advice on how to best interact with health care providers and methods to improve medication adherence. A “learning more” section focused on what high blood pressure is, addressed therapeutic lifestyle changes, and encouraged patients to communicate with their health care providers. The patients who viewed the patient narratives had significantly lower blood pressure at 3 months than those assigned to usual care. Although blood pressure subsequently increased in both groups, the benefits of the intervention still existed at the end of follow-up.

Important to note about two of the above three studies^1,3 is that the interventions were done by people other than physicians, thus emphasizing the importance of a team approach to blood pressure control.

Behavioral counseling

The effectiveness of lifestyle modifications such as diet, weight loss, and physical activity in preventing and treating hypertension is well established.^74–78 For example:

• In the Dietary Approaches to Stop Hypertension (DASH) trial,⁷⁶ a healthy diet lowered blood pressure about as much as single drugs do, particularly in African Americans.
• The Trial of Nonpharmacologic Interventions in the Elderly (TONE)⁷⁴ showed that exercise can lower blood pressure in obese hypertensive patients.
• The PREMIER trial (Lifestyle Interventions for Blood Pressure Control)⁷⁵ showed that a single brief counseling session could produce substantial decreases in blood pressure in patients with stage 1 hypertension or high-normal blood pressure.

Unfortunately, these results have been hard to translate into primary care practice, especially for African American patients. Several studies have evaluated the impact of lifestyle interventions on blood pressure control in primary care practices with a large population of African American patients.

Bosworth et al,⁴ in a study of a practice in which almost half the patients were African American, randomized patients to receive usual care, nurse-administered tailored behavioral telephone counseling, home blood pressure monitoring, or home monitoring plus tailored behavioral telephone counseling. The combination of home monitoring and tailored behavioral telephone counseling led to a statistically significant improvement at 24 months compared with baseline.

Home blood pressure monitoring

The effectiveness of self-monitoring in improving blood pressure control is also well documented.^{62,63,65–67,90–95}

Pickering et al⁶² studied patients with poorly controlled hypertension in a managed-care setting and found a reduction of 7 mm Hg systolic and 5 mm Hg diastolic pressure after 3 to 6 months of home monitoring compared with usual care.

Mengden et al,⁹⁴ in a similar study, found average blood pressure reductions at 6 months of 19.3/11.9 mm Hg in the home-monitoring group vs 10.6/8.8 mm Hg in the usual-care group.

The effect of home blood pressure monitoring may be greater in African Americans.

Rogers et al⁹³ found it to be more effective at lowering blood pressure than usual care in a group of 121 patients with poorly controlled hypertension followed in primary care practices, and these reductions were twice as large in African American patients than in white patients.⁹³

Bondmass,⁹² in a study of 33 African American patients with poorly controlled hypertension, reported a 53% control rate within 4 weeks of home monitoring. All patients in the study had uncontrolled blood pressure at baseline (> 140/90 mm Hg).

Artinian et al⁵ evaluated the effect of nurse-managed telemonitoring on blood pressure control vs enhanced usual care. All participants were African American. The monitored group had a significantly greater reduction in systolic pressure at 12 months compared with those who received enhanced usual care.

 

PHYSICIAN-LEVEL INTERVENTIONS

Most interventions to improve how physicians manage patients with hypertension are designed to improve adherence to treatment guidelines. In most cases, these interventions are based on continuous quality improvement and disease management concepts such as physician education and academic detailing, reminders, feedback on performance measures, and risk-assessment tools.^96,97

Physician education

Interest is increasing in physician educational interventions for blood pressure control.^24,98

Inui et al,⁹⁹ in an early study in a primary care practice, found that patients of physicians who received tutorials on hypertension management were more compliant with their drug regimens and had better blood pressure control than patients of physicians in the control group.

Jennett et al,¹⁰⁰ in a similar randomized clinical trial, found that physicians who participated in an education activity were more adherent to treatment guidelines at 6 and 12 months compared with those who did not participate.

Maue et al¹⁰¹ showed that rates of blood pressure control improved from 41% to 52% after a 6-month educational intervention for physicians in a managed-care setting.

Tu et al¹⁰² reviewed 12 studies in which seven different physician educational interventions were used either alone or in combination and concluded that physician education improves compliance with guidelines for managing hypertension.

Unfortunately, these studies did not report outcomes separately for African American and white patients.

Hicks et al⁶ found that disease management approaches that target physicians whose patients with hypertension are mostly African American did not yield clinically relevant improvement in these patients, and that minority patients were significantly less likely to have their blood pressure controlled at the end of the study compared with their non-Hispanic white counterparts.

Feedback to providers

Several studies have shown that, given reminders and feedback systems, physicians will change their practice.^103–106

Mashru and Lant¹⁰⁴ combined chart audits and physician education in primary care practices and found they improved physician performance measures such as accuracy of diagnosis, number of patients who received cardiovascular risk assessment, and number of patients whose treatment was based on clinical laboratory assessments.

Feedback takes many forms but consists mostly of computerized information¹⁰⁷ or peer-to-peer academic detailing with opinion leaders.^108–110

Dickinson et al,¹⁰⁶ for instance, showed that computer-generated listings of patients’ blood pressures combined with a physician education program on clinical management of hypertension led to increased knowledge and better follow-up on their patients.

Again, however, these studies did not distinguish between African American and white patients, which makes it difficult to judge whether or not these approaches work differently for physicians with a large proportion of African American patients.

Computerized decision-support systems

Computerized decision-support systems have proliferated in primary care practices.¹¹¹

McAlister et al¹⁰³ found that general practitioners randomized to manage hypertension with the assistance of a computer obtained better outcomes than with usual care.

Montgomery and Fahey,¹⁰⁷ in a systematic review, found improved blood pressure control in two of the three trials that compared computer-generated feedback reports and reminders to usual care. Specifically, 51% of patients whose physicians received reminders either had controlled blood pressure or were at least receiving treatment vs 33% in the control group at 12 months. This difference was even higher at 24 months.

Montgomery et al⁷ later randomized primary care practices to use a computer-based decision-support system and a cardiovascular risk chart, the risk chart alone, or to continue as usual. Results indicated no reduction in cardiovascular risk in the computer-system or the chart-only group, whereas patients in the chart-only group had a significant reduction in systolic pressure and were prescribed more cardiovascular drugs. This study indicates that use of a computerized decision-support system is not superior to chart review and audit feedback alone.

Evidence that computerized decision systems improve blood pressure control in African Americans is scant. However, when one looks at the evidence from studies of African Americans, the outcomes do not seem to differ between African American and white patients.

Hicks et al⁶ examined the effectiveness of computerized decision support in improving hypertension care in a racially diverse population. Physicians were randomized to receive computerized decision support or to provide usual care without computerized support. Both groups improved significantly in prescribing appropriate drugs but not in overall blood pressure control. Furthermore, the study showed no reduction in racial disparities of care and blood pressure control.

A potential explanation for the lack of improvement in blood pressure was that the intervention dealt with making sure the appropriate drugs were prescribed rather than making sure physicians also appropriately intensified antihypertensive management when necessary.

 

INTERVENTIONS TARGETING PATIENTS AND PHYSICIANS

Several studies have targeted both patient and physician-level barriers to blood pressure control in practice-based settings.

Roumie et al⁸ randomized physicians to one of three intervention groups:

• “Provider education” consisting of an email message with a Web-based link to the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC-7)
• Provider education plus a computer alert with information about their patient’s blood pressure
• Provider education, a computer alert, and patient education (ie, patients received a letter encouraging adherence to drug therapy, changing their lifestyle, and talking with their doctor about their blood pressure).

Patients whose providers were randomized to the third group had better blood pressure control. The report did not differentiate African American vs white patients. The data, however, did show the effectiveness of adding patient education to provider education to improve blood pressure control.

Bosworth et al,¹¹² in a study in which 40% of patients were African American, randomized patients to usual care or to bimonthly nurse-delivered behavioral telephone counseling. They also randomized providers either to receive computer-generated decision support designed to improve adherence to guidelines or to receive no support.

There were no significant differences in rates of blood pressure control in the intervention groups compared with a control group. Although differences in blood pressure control between groups were not significant, patients randomized to behavioral intervention had significantly better blood pressure control at the 24-month follow-up than at baseline.

Svetkey et al⁹ evaluated the effects of physician intervention, patient intervention, and physician intervention plus patient intervention compared with control on systolic blood pressure at 6 months. They found that an intensive behavioral lifestyle intervention led to a significant reduction in systolic pressure at 6 months. By itself, the physician intervention did not have a meaningful effect, but patients in the combined physician-and-patient-intervention group experienced the greatest reduction (9.7 ± 12.7 mm Hg).

It takes a team

Physicians should not be the only focus in helping patients achieve blood pressure control. Although physician and patient factors need to be addressed to improve blood pressure control in African Americans, emphasis should also be placed on interdisciplinary, team-based care utilizing health care providers such as nurses, physician assistants, and pharmacists. Team-based care has been shown to have the greatest impact of all the strategies for improving blood pressure control.¹¹³ There is a good amount of evidence involving interventions with a focus on health care providers other than physicians, although the data lack a sufficient focus on African Americans.

Carter et al,¹⁰ in a randomized controlled trial in which 26.3% of the patients were African American, found that an intervention consisting of clinical pharmacists giving physicians drug therapy recommendations based on national guidelines resulted in a significantly lower blood pressure compared with a control group: the mean reduction was 20.7/9.7 mm Hg in the intervention group vs 6.8/4.5 mm Hg in the control group.

Carter et al¹¹⁴ performed a meta-analysis of 37 studies and found that two strategies led to a significant reduction in blood pressure: a pharmacist-led intervention with treatment recommendations to physicians resulted in a systolic pressure reduction of 9.30 mm Hg; and nurse-led interventions resulted in a systolic pressure reduction of 4.80 mm Hg. Again, many of the studies cited in this meta-analysis lacked a focus on African Americans.

Hunt et al¹¹ conducted a randomized controlled trial in which pharmacists actively participated in the management of blood pressure. They were involved with every aspect of care, including reviewing medications and adverse drug reactions, assessing lifestyle behaviors and barriers to adherence, making dosing adjustments, and adding medications. Patients randomized to the intervention group achieved significantly lower systolic and diastolic pressures (137/75 vs 143/78 mm Hg in the control group). However, information about race was not included.

The above studies are just a few out of a large body of evidence demonstrating the value of team-based care to improve blood pressure control. It has yet to be determined whether these models can improve blood pressure control specifically in African Americans, since so many of these trials lacked a focus on this group. Promising is an ongoing randomized prospective trial by Carter et al¹¹⁵ evaluating a model of collaboration between physicians and pharmacists, with a focus on patients in underrepresented minorities.

SO WHAT WORKS?

Although there is a growing body of literature on interventions to try to reduce disparities in hypertension and blood pressure control between African Americans and whites, only a few randomized controlled trials have focused on African Americans, and several have not reported their results.¹¹⁶ So the question remains: How should we interpret the available data, which are aggregated across racial groups, and put it into practice when caring for hypertensive African American patients?

Patient education. In trying to overcome patient-related barriers, emphasis should be on patient education, in particular addressing misconceptions about hypertension and promoting adherence to antihypertensive therapy. This is evident from the narrative storytelling intervention by Houston et al.³ Although this is the first study of its kind, this strategy may be something to consider if future studies replicate these findings. Culturally appropriate storytelling may allow patients to identify with the stories as they relate to their own personal lives. It can be an effective way to address patient education and change behaviors.

Self-monitoring with a home blood pressure monitor has also proven effective in the management of hypertension in African Americans. Indeed, the few studies that reported findings in African Americans showed impressive reductions in blood pressure. The benefits of home monitoring are well documented, and the effect on physician-related barriers such as clinical inertia are also quite impressive.¹¹⁷ However, most of these studies did not assess the long-term impact or cost-effectiveness of home monitoring on blood pressure control.

Behavioral counseling. Although we have good evidence of the effectiveness of behavioral counseling, whether this is sustained long-term has been less studied in African Americans. Thus, while interventions that targeted African Americans have reported impressive reductions in blood pressure, the effect tends to be greatest during the first few months of implementation, with the benefits disappearing over time.

Physician-related interventions. With regard to physician-level interventions, research has focused on physician education, utilizing alerts and computerized clinical decision-support systems. Evidence is scant on whether the use of computerized systems results in improves hypertension care in African Americans. However, a closer look at the data from studies that report outcomes in African American and white patients shows that the results do not seem to differ between these groups. Still, there is insufficient information about the impact on hypertensive African Americans.⁶

Strategies that address both patient- and physician-related barriers can improve overall blood pressure control; however, there is a lack of data comparing outcomes in hypertensive African Americans with those of whites, making it difficult to know if this would be an effective strategy in African American patients alone.

More studies needed that focus on African Americans

Developing interventions to improve blood pressure control in African Americans should be an ongoing priority for research if we intend to address racial disparities in cardiovascular disease. Although it is reassuring that there is a growing body of evidence and research with this focus,^118–121 more research is needed to determine effective strategies that address barriers related to physician practice and to the health care system overall as they relate to blood pressure control in African Americans. More importantly, these strategies should also emphasize a team-based approach that includes nurses, pharmacists, and physician assistants. Developing targeted interventions for hypertensive African Americans will help reduce disparities in the rates of cardiovascular illness and death in this patient population.