John R. McConaghy, MD

Heart failure (HF) affects more than 2 million adults in the United States.¹ This common, costly, and disabling disorder mainly affects the elderly, with prevalence rates of up to 10% in patients older than 65 years.^2,3 The management of HF is responsible for millions of outpatient visits per year,⁴ is the most common discharge diagnosis for Medicare beneficiaries,⁵ and accounts for more than 5% of total health care dollars spent.⁶

Treatment

Major advances in the pharmacologic treatment of heart failure (HF) have emerged in recent years. An approach to the diagnosis and evaluation of HF is described elsewhere.⁷ This article summarizes the evidence for outpatient treatment of HF. Current intervention trials do not distinguish between systolic and diastolic heart failure; it is therefore unknown whether or how drug therapy should be tailored according to the type of HF. The treatment of cardiac dysrhythmias in the setting of HF is beyond the scope of this article and is presented elsewhere.⁸Table 1 compares the available outpatient treatments of HF and includes the levels of evidence, numbers needed to treat, and appropriate situations for use. In the remainder of this article, we will discuss pharmacologic and nonpharmacologic management, including identification of ineffective treatments.

TABLE 1
Treatment options in heart failure

Pharmacologic treatment

Angiotensin-converting enzyme inhibitors. A systematic review⁹ of 32 trials with a total of 7105 patients demonstrated that mortality rates were lower in patients taking an angiotensin-converting enzyme (ACE) inhibitor than in those not taking one (number needed to treat [NNT] = 24 for > 90 days, meaning that 1 fewer death occurs for every 24 patients who take an ACE inhibitor for more than 90 days). In addition, there is a reduction in the combined endpoints of death and hospitalization because of HF (NNT = 11). Although most of this benefit was realized in the first 90 days of therapy, benefits lasted for 4 to 5 years and were more pronounced in patients categorized in more severe New York Heart Association (NYHA) HF classes¹⁰ (class I: no limitation of activities; class II: slight limitation of activity; class III: marked limitation of activity and comfortable only at rest; class IV: symptoms at rest).

Dosage comparison studies demonstrate that HF patients can benefit from even moderate doses of ACE inhibitors. A recent multicenter trial comparing moderate dose enalapril (10 mg twice a day) with a higher dose (30 mg twice a day) in patients with a left ventricular ejection fraction (LVEF) of less than 20% found no differences in mortality at 1 year between the 2 groups.¹¹ In addition, both groups achieved similar increases in functional status and LVEF.

Several trials have demonstrated good tolerability of ACE inhibitors.^12-14 Dropout rates of 15% to 30% were similar between patients in the ACE inhibitor and placebo groups, mainly because of side effects, including dizziness, altered taste, hypotension, hyperkalemia, and cough.

Angiotensin-receptor blockers. Angiotensin-receptor blockers (ARBs) reduce all-cause mortality and HF-related hospitalizations in patients with NYHA class II and III HF at rates comparable with those of ACE inhibitors.^15,16 Cough is not a side effect of ARBs. Although they are more expensive, ARBs offer a reasonable alternative for patients who do not tolerate ACE inhibitors.

Beta-blockers. The beta-blockers carvedilol, metoprolol, and bisoprolol have a proven mortality benefit for patients with HF.^17-19 Pooled results of 6 randomized controlled trials (RCTs), including more than 9000 patients already taking ACE inhibitors, showed a significant reduction in total mortality (NNT = 24 over 1–2 years) and sudden death (NNT = 35), regardless of NYHA classification.²⁰ The average dropout rate of 16% was similar in the betablocker and placebo groups.

Early beta-blocker studies included few NYHA class IV patients until a recent study of the use of carvedilol in severe chronic HF.²¹ In this study, all patients were taking diuretics plus either an ACE inhibitor or ARB and were permitted to take digoxin, nitrates, hydralazine, spironolactone, or amiodarone. Carvedilol at an average dose of 37 mg per day decreased mortality (NNT = 18 for 10 months) and lowered combined mortality and hospitalization for worsening HF (NNT = 13). Study patients taking carvedilol withdrew from the study at a lower rate (approximately 15%) than placebo.

Because the pharmacologic properties of betablockers vary, clinicians have wondered which are most beneficial. The investigators in a study comparing metoprolol (a beta-1 antagonist) with carvedilol (a beta-1, beta-2, and alpha-1 antagonist) in NYHA class II or III patients found no differences in quality-of-life measures or changes in NYHA classification.²²

Spironolactone. The addition of spironolactone to standard care can help patients with severe HF.²³ In NYHA class III and IV HF patients, spironolactone at doses ranging from 25 mg every other day to 50 mg per day reduces mortality (NNT = 9 for 2 years), reduces hospitalization from all cardiac causes (NNT = 4), and reduces hospitalization for worsening HF (NNT = 3). The most common serious adverse event in the spironolactone group was severe hyperkalemia (number needed to harm [NNH] = 195). Ten percent of men taking spironolactone experienced breast pain and gynecomastia.

Hydralazine and isosorbide dinitrate. The combination of hydralazine and isosorbide dinitrate (ISDN) reduces mortality in HF patients, but tolerability is an issue. In earlier trials, men with HF symptoms that were optimally controlled with digoxin and diuretics and treated with hydralazine (average dose = 270 mg/day) plus ISDN (average dose = 136 mg/day) had a decrease in all-cause mortality of 28% (NNT = 19 for 6 years).²⁴ A more recent trial comparing hydralazine plus ISDN with enalapril²⁵ (average daily doses of hydralazine = 300 mg/day; ISDN = 160 mg/day; enalapril = 20 mg/day) in NYHA class II–III patients showed no differences in mortality between the 2 groups over 3 years. Tolerability was a problem in these trials; more than 30% of patients stopped taking hydralazine, nitrate, or both.

Digoxin. Digoxin is effective for treating the symptoms of HF in the absence of dysrhythmias but there are no data demonstrating a mortality benefit. Digoxin increases functional capacity in NYHA class II–III patients and heart failure symptoms worsen if digoxin is withdrawn.²⁶ Although there are no differences in all-cause mortality with the use of digoxin, there are fewer hospitalizations due to worsening HF (NNT = 27–114 over 3 years) and a lower rate of clinical deterioration (NNT = 4–75).²⁷ In a randomized trial comparing digoxin and placebo, patients taking digoxin were twice as likely to be hospitalized for suspected digoxin toxicity (2.0% vs 0.9%; P < .001; NNH = 52).²⁸

Diuretics. Diuretics are a mainstay of the symptomatic treatment of heart failure. Short-term studies have shown that diuretics improve the symptoms of sodium and fluid retention and increase exercise tolerance and cardiac function regardless of NYHA classification.^29-32 No studies that examine their effects on morbidity and mortality are available.

Antiplatelet therapy and anticoagulation. Patients with HF have an increased risk for thromboembolic events of 1.6% to 3.2% per year.³³ One systematic review concluded that antiplatelet therapy is not useful in preventing thromboembolism in patients with HF in sinus rhythm and may even be harmful.³⁴ Another systematic review also concluded that the data do not support the routine use of anticoagulants (eg, warfarin) in patients with HF and sinus rhythm.³⁵ Anticoagulation may be beneficial, however, if there is echocardiographic visualization of a left ventricular thrombus or in cases of “severe” HF or concomitant atrial fibrillation.³⁵

Nonpharmacologic management

Dietary sodium restriction. There is consensus that dietary sodium restriction is important in the treatment of HF³⁶ and is recommended in published guidelines.^8,37 Sodium restriction assists with fluid volume control and minimizes the dosages of HF drugs used. These recommendations are based on the retention of sodium and water in symptomatic HF. No studies, however, have examined the effect of dietary sodium restriction on morbidity or mortality, either alone or in combination with pharmacologic treatments.

Exercise training. Moderate exercise training improves quality of life and decreases mortality in patients with stable chronic HF. A recent RCT demonstrated a decrease in mortality (NNT = 4 for 14 months) and hospital readmission for HF (NNT = 5) with only moderate exercise on a stationary bicycle (60% of maximum exercise capacity) for 2 to 3 hours per week.³⁸ Other studies have demonstrated improvements in physiologic markers³⁹ and in quality-of-life ratings with short-term, symptom-limited exercise.⁴⁰

Multidisciplinary or case-management approach. A case-based or disease-management approach to patients with HF decreases the frequency of unplanned and repeat hospitalizations, increases functional status, and increases quality of life.⁴¹ Even a single in-home visit by a clinical pharmacist and a nurse results in fewer unplanned readmissions and fewer days of hospitalization up to 18 months after discharge.^42,43 A small study of 27 patients in a Veterans Affairs hospital demonstrated that patient instruction in the self-monitoring of weight and blood pressure, combined with frequent telephone follow-up from a nurse, lowered repeat hospitalizations over 1 year, with the effect more pronounced in patients with more severe NYHA classifications.⁴⁴ A large RCT demonstrated that a multidisciplinary management approach (intensive patient education about HF and its treatment, dietary assessment and instruction, medication analysis and elimination of unnecessary medications, and telephone and home visit follow-up) results in fewer hospitalizations (NNT = 5 for 3 months) and reduced costs of care.⁴⁵

Treatments that have no benefit or are harmful

Calcium-channel blockers. Although some of the newer, longer-acting calcium-channel blockers (CCBs) appear to be safe in the treatment of heart failure,^46-49 no trials are available demonstrating that they lower mortality, decrease hospitalizations, or improve quality of life in patients with a failing heart. Older, short-acting CCBs can worsen HF.⁵⁰

Positive inotropic therapy. Intermittent positive inotropic therapy, either orally (milrinone) or intravenously (dobutamine), should be avoided. Although short-term studies have shown some increase in cardiac function and symptoms,⁵¹ long-term studies demonstrate no mortality benefit.⁵² One RCT of milrinone demonstrated an increase in mortality (NNH = 17 for 5 months), an increased rate of hospitalization for worsening HF (NNH = 20), and more serious side effects (NNH = 25).⁵³

Prognosis

Despite the increased longevity in Western developed nations and increased survival from coronary artery disease over recent decades, the overall prognosis of HF has improved very little.^6,54 Mortality data derived from several different sources, the largest being the Framingham Heart Study,^2,55 have shown that HF remains highly lethal, with a 5-year survival rate of 25% in men and 38% in women with NYHA II–IV heart failure. Mortality data from the placebo arms of intervention trials show an average 1-year mortality of 18%.^{9,17,19,20,56} A recent population-based study of patients with a new diagnosis of HF showed survival rates of only 62% at 12 months and 57% at 18 months.⁵⁷ Despite these dismal population-based data, predicting the likelihood of survival in individuals with HF is largely unreliable.⁸ Estimating individual prognosis is only somewhat useful in making end-of-life care and hospice decisions for patients with very advanced HF. Table 2 summarizes specific prognostic factors for patients with HF.

TABLE 2
Factors that affect prognosis in patients with heart failure (HF)

Suggested management of patients with heart failure

Although the optimal sequence of pharmacologic interventions for treating HF has not been examined in RCTs, recommendations can be made on the basis of existing evidence in HF management (Figure). This approach can be divided into 4 steps performed simultaneously: (1) control risks for the development and progression of HF (treat concomitant diseases); (2) HF treatment; (3) symptom control; (4) close follow-up.

Control risks. Risks for the development and worsening of HF should be addressed as described else-where.⁸ Steps include longitudinal surveillance; identification and treatment of hypertension, diabetes and thyroid diseases; management of atherosclerotic and coronary artery disease and myocardial ischemia; and the elimination of alcohol and tobacco use.

Heart failure treatment. All patients with HF should take a drug or a combination of drugs that affects the disease process. Drugs shown by the preponderance of evidence to decrease morbidity and mortality include ACE inhibitors, beta-blockers, and ARBs. For most HF patients, regardless of NYHA class, ACE inhibitors should be the initial baseline treatment because of their proven track record and the observation that most recent HF trials include patients who are already taking these medications. ARBs are similar in efficacy to ACE inhibitors and, therefore, are an adequate alternative when ACE inhibitors are not tolerated. Beta-blockers (metoprolol and bisoprolol) added to ACE inhibitors are also useful as a baseline treatment in most HF patients and may be especially useful in the case of tachydysrhythmias and in the postmyocardial infarction period.

For severe HF (NYHA III–IV), spironolactone and carvedilol are useful additions to baseline drug therapy. Carvedilol may be added if a beta-blocker is not currently used. If the patient is currently taking a beta-blocker, the drug should be discontinued before the patient is switched to carvedilol.

The hydralazine–nitrate combination has been proved effective, but tolerability and ease-of-use issues limit its usefulness. No data are available to support the use of nitrates other than isosorbide dinitrate. Nitrates may be useful, however, for concomitant chronic myocardial ischemia.

Patients with stable HF should be encouraged to begin and maintain a regular aerobic exercise program. The level of exercise can range from brief, symptom-limited exercise to moderate exercise (60% capacity) for 3 or more hours per week.

The use of antiplatelet therapy or the routine use of anticoagulation in patients with HF who are in sinus rhythm provides no benefit. Anticoagulation may be useful if the patient has severe HF or has a known mural thrombus. HF patients with atrial fibrillation should be considered for antiplatelet or anticoagulation therapy as described elsewhere.⁵⁸

Short-acting CCBs may worsen HF. No data support the use of any CCB in the primary treatment of HF. Similarly, intermittent use of milrinone or dobutamine is not indicated.

Symptom control. The symptomatic treatment of HF includes the use of diuretics and dietary sodium restriction to control sodium levels and volume status. Symptom control should be accomplished along with the pharmacologic disease management outlined above.

The role of digoxin in the failing heart without dysrhythmias is unclear. Digoxin may be most useful in symptom control, as it reduces hospitalizations attributed to worsening HF. This benefit must be balanced against an increased risk of hospitalization caused by digoxin toxicity. Patients who are already taking digoxin should probably continue to do so. The role of digoxin in newly diagnosed HF patients is unknown.

Close follow-up. Comprehensive follow-up, with the patient as a more active participant and in which care is extended beyond the hospital or office to the home, is a key strategy in the long-term care of HF patients. This aspect of HF management should include educating patients about their disease process and their dietary and pharmacologic treatments; teaching them how to monitor their weight, symptoms, and blood pressure and to understand when to seek care; and following up periodically by telephone between scheduled office visits.

FIGURE
Management of adults with heart failure

Heart failure (HF) affects more than 2 million adults in the United States.¹ This common, costly, and disabling disorder mainly affects the elderly, with prevalence rates of up to 10% in patients older than 65 years.^2,3 The management of HF is responsible for millions of outpatient visits per year,⁴ is the most common discharge diagnosis for Medicare beneficiaries,⁵ and accounts for more than 5% of total health care dollars spent.⁶

Treatment

Major advances in the pharmacologic treatment of heart failure (HF) have emerged in recent years. An approach to the diagnosis and evaluation of HF is described elsewhere.⁷ This article summarizes the evidence for outpatient treatment of HF. Current intervention trials do not distinguish between systolic and diastolic heart failure; it is therefore unknown whether or how drug therapy should be tailored according to the type of HF. The treatment of cardiac dysrhythmias in the setting of HF is beyond the scope of this article and is presented elsewhere.⁸Table 1 compares the available outpatient treatments of HF and includes the levels of evidence, numbers needed to treat, and appropriate situations for use. In the remainder of this article, we will discuss pharmacologic and nonpharmacologic management, including identification of ineffective treatments.

TABLE 1
Treatment options in heart failure

Pharmacologic treatment

Angiotensin-converting enzyme inhibitors. A systematic review⁹ of 32 trials with a total of 7105 patients demonstrated that mortality rates were lower in patients taking an angiotensin-converting enzyme (ACE) inhibitor than in those not taking one (number needed to treat [NNT] = 24 for > 90 days, meaning that 1 fewer death occurs for every 24 patients who take an ACE inhibitor for more than 90 days). In addition, there is a reduction in the combined endpoints of death and hospitalization because of HF (NNT = 11). Although most of this benefit was realized in the first 90 days of therapy, benefits lasted for 4 to 5 years and were more pronounced in patients categorized in more severe New York Heart Association (NYHA) HF classes¹⁰ (class I: no limitation of activities; class II: slight limitation of activity; class III: marked limitation of activity and comfortable only at rest; class IV: symptoms at rest).

Dosage comparison studies demonstrate that HF patients can benefit from even moderate doses of ACE inhibitors. A recent multicenter trial comparing moderate dose enalapril (10 mg twice a day) with a higher dose (30 mg twice a day) in patients with a left ventricular ejection fraction (LVEF) of less than 20% found no differences in mortality at 1 year between the 2 groups.¹¹ In addition, both groups achieved similar increases in functional status and LVEF.

Several trials have demonstrated good tolerability of ACE inhibitors.^12-14 Dropout rates of 15% to 30% were similar between patients in the ACE inhibitor and placebo groups, mainly because of side effects, including dizziness, altered taste, hypotension, hyperkalemia, and cough.

Angiotensin-receptor blockers. Angiotensin-receptor blockers (ARBs) reduce all-cause mortality and HF-related hospitalizations in patients with NYHA class II and III HF at rates comparable with those of ACE inhibitors.^15,16 Cough is not a side effect of ARBs. Although they are more expensive, ARBs offer a reasonable alternative for patients who do not tolerate ACE inhibitors.

Beta-blockers. The beta-blockers carvedilol, metoprolol, and bisoprolol have a proven mortality benefit for patients with HF.^17-19 Pooled results of 6 randomized controlled trials (RCTs), including more than 9000 patients already taking ACE inhibitors, showed a significant reduction in total mortality (NNT = 24 over 1–2 years) and sudden death (NNT = 35), regardless of NYHA classification.²⁰ The average dropout rate of 16% was similar in the betablocker and placebo groups.

Early beta-blocker studies included few NYHA class IV patients until a recent study of the use of carvedilol in severe chronic HF.²¹ In this study, all patients were taking diuretics plus either an ACE inhibitor or ARB and were permitted to take digoxin, nitrates, hydralazine, spironolactone, or amiodarone. Carvedilol at an average dose of 37 mg per day decreased mortality (NNT = 18 for 10 months) and lowered combined mortality and hospitalization for worsening HF (NNT = 13). Study patients taking carvedilol withdrew from the study at a lower rate (approximately 15%) than placebo.

Because the pharmacologic properties of betablockers vary, clinicians have wondered which are most beneficial. The investigators in a study comparing metoprolol (a beta-1 antagonist) with carvedilol (a beta-1, beta-2, and alpha-1 antagonist) in NYHA class II or III patients found no differences in quality-of-life measures or changes in NYHA classification.²²

Spironolactone. The addition of spironolactone to standard care can help patients with severe HF.²³ In NYHA class III and IV HF patients, spironolactone at doses ranging from 25 mg every other day to 50 mg per day reduces mortality (NNT = 9 for 2 years), reduces hospitalization from all cardiac causes (NNT = 4), and reduces hospitalization for worsening HF (NNT = 3). The most common serious adverse event in the spironolactone group was severe hyperkalemia (number needed to harm [NNH] = 195). Ten percent of men taking spironolactone experienced breast pain and gynecomastia.

Hydralazine and isosorbide dinitrate. The combination of hydralazine and isosorbide dinitrate (ISDN) reduces mortality in HF patients, but tolerability is an issue. In earlier trials, men with HF symptoms that were optimally controlled with digoxin and diuretics and treated with hydralazine (average dose = 270 mg/day) plus ISDN (average dose = 136 mg/day) had a decrease in all-cause mortality of 28% (NNT = 19 for 6 years).²⁴ A more recent trial comparing hydralazine plus ISDN with enalapril²⁵ (average daily doses of hydralazine = 300 mg/day; ISDN = 160 mg/day; enalapril = 20 mg/day) in NYHA class II–III patients showed no differences in mortality between the 2 groups over 3 years. Tolerability was a problem in these trials; more than 30% of patients stopped taking hydralazine, nitrate, or both.

Digoxin. Digoxin is effective for treating the symptoms of HF in the absence of dysrhythmias but there are no data demonstrating a mortality benefit. Digoxin increases functional capacity in NYHA class II–III patients and heart failure symptoms worsen if digoxin is withdrawn.²⁶ Although there are no differences in all-cause mortality with the use of digoxin, there are fewer hospitalizations due to worsening HF (NNT = 27–114 over 3 years) and a lower rate of clinical deterioration (NNT = 4–75).²⁷ In a randomized trial comparing digoxin and placebo, patients taking digoxin were twice as likely to be hospitalized for suspected digoxin toxicity (2.0% vs 0.9%; P < .001; NNH = 52).²⁸

Diuretics. Diuretics are a mainstay of the symptomatic treatment of heart failure. Short-term studies have shown that diuretics improve the symptoms of sodium and fluid retention and increase exercise tolerance and cardiac function regardless of NYHA classification.^29-32 No studies that examine their effects on morbidity and mortality are available.

Antiplatelet therapy and anticoagulation. Patients with HF have an increased risk for thromboembolic events of 1.6% to 3.2% per year.³³ One systematic review concluded that antiplatelet therapy is not useful in preventing thromboembolism in patients with HF in sinus rhythm and may even be harmful.³⁴ Another systematic review also concluded that the data do not support the routine use of anticoagulants (eg, warfarin) in patients with HF and sinus rhythm.³⁵ Anticoagulation may be beneficial, however, if there is echocardiographic visualization of a left ventricular thrombus or in cases of “severe” HF or concomitant atrial fibrillation.³⁵

Nonpharmacologic management

Dietary sodium restriction. There is consensus that dietary sodium restriction is important in the treatment of HF³⁶ and is recommended in published guidelines.^8,37 Sodium restriction assists with fluid volume control and minimizes the dosages of HF drugs used. These recommendations are based on the retention of sodium and water in symptomatic HF. No studies, however, have examined the effect of dietary sodium restriction on morbidity or mortality, either alone or in combination with pharmacologic treatments.

Exercise training. Moderate exercise training improves quality of life and decreases mortality in patients with stable chronic HF. A recent RCT demonstrated a decrease in mortality (NNT = 4 for 14 months) and hospital readmission for HF (NNT = 5) with only moderate exercise on a stationary bicycle (60% of maximum exercise capacity) for 2 to 3 hours per week.³⁸ Other studies have demonstrated improvements in physiologic markers³⁹ and in quality-of-life ratings with short-term, symptom-limited exercise.⁴⁰

Multidisciplinary or case-management approach. A case-based or disease-management approach to patients with HF decreases the frequency of unplanned and repeat hospitalizations, increases functional status, and increases quality of life.⁴¹ Even a single in-home visit by a clinical pharmacist and a nurse results in fewer unplanned readmissions and fewer days of hospitalization up to 18 months after discharge.^42,43 A small study of 27 patients in a Veterans Affairs hospital demonstrated that patient instruction in the self-monitoring of weight and blood pressure, combined with frequent telephone follow-up from a nurse, lowered repeat hospitalizations over 1 year, with the effect more pronounced in patients with more severe NYHA classifications.⁴⁴ A large RCT demonstrated that a multidisciplinary management approach (intensive patient education about HF and its treatment, dietary assessment and instruction, medication analysis and elimination of unnecessary medications, and telephone and home visit follow-up) results in fewer hospitalizations (NNT = 5 for 3 months) and reduced costs of care.⁴⁵

Treatments that have no benefit or are harmful

Calcium-channel blockers. Although some of the newer, longer-acting calcium-channel blockers (CCBs) appear to be safe in the treatment of heart failure,^46-49 no trials are available demonstrating that they lower mortality, decrease hospitalizations, or improve quality of life in patients with a failing heart. Older, short-acting CCBs can worsen HF.⁵⁰

Positive inotropic therapy. Intermittent positive inotropic therapy, either orally (milrinone) or intravenously (dobutamine), should be avoided. Although short-term studies have shown some increase in cardiac function and symptoms,⁵¹ long-term studies demonstrate no mortality benefit.⁵² One RCT of milrinone demonstrated an increase in mortality (NNH = 17 for 5 months), an increased rate of hospitalization for worsening HF (NNH = 20), and more serious side effects (NNH = 25).⁵³

Prognosis

Despite the increased longevity in Western developed nations and increased survival from coronary artery disease over recent decades, the overall prognosis of HF has improved very little.^6,54 Mortality data derived from several different sources, the largest being the Framingham Heart Study,^2,55 have shown that HF remains highly lethal, with a 5-year survival rate of 25% in men and 38% in women with NYHA II–IV heart failure. Mortality data from the placebo arms of intervention trials show an average 1-year mortality of 18%.^{9,17,19,20,56} A recent population-based study of patients with a new diagnosis of HF showed survival rates of only 62% at 12 months and 57% at 18 months.⁵⁷ Despite these dismal population-based data, predicting the likelihood of survival in individuals with HF is largely unreliable.⁸ Estimating individual prognosis is only somewhat useful in making end-of-life care and hospice decisions for patients with very advanced HF. Table 2 summarizes specific prognostic factors for patients with HF.

TABLE 2
Factors that affect prognosis in patients with heart failure (HF)

Suggested management of patients with heart failure

Although the optimal sequence of pharmacologic interventions for treating HF has not been examined in RCTs, recommendations can be made on the basis of existing evidence in HF management (Figure). This approach can be divided into 4 steps performed simultaneously: (1) control risks for the development and progression of HF (treat concomitant diseases); (2) HF treatment; (3) symptom control; (4) close follow-up.

Control risks. Risks for the development and worsening of HF should be addressed as described else-where.⁸ Steps include longitudinal surveillance; identification and treatment of hypertension, diabetes and thyroid diseases; management of atherosclerotic and coronary artery disease and myocardial ischemia; and the elimination of alcohol and tobacco use.

Heart failure treatment. All patients with HF should take a drug or a combination of drugs that affects the disease process. Drugs shown by the preponderance of evidence to decrease morbidity and mortality include ACE inhibitors, beta-blockers, and ARBs. For most HF patients, regardless of NYHA class, ACE inhibitors should be the initial baseline treatment because of their proven track record and the observation that most recent HF trials include patients who are already taking these medications. ARBs are similar in efficacy to ACE inhibitors and, therefore, are an adequate alternative when ACE inhibitors are not tolerated. Beta-blockers (metoprolol and bisoprolol) added to ACE inhibitors are also useful as a baseline treatment in most HF patients and may be especially useful in the case of tachydysrhythmias and in the postmyocardial infarction period.

For severe HF (NYHA III–IV), spironolactone and carvedilol are useful additions to baseline drug therapy. Carvedilol may be added if a beta-blocker is not currently used. If the patient is currently taking a beta-blocker, the drug should be discontinued before the patient is switched to carvedilol.

The hydralazine–nitrate combination has been proved effective, but tolerability and ease-of-use issues limit its usefulness. No data are available to support the use of nitrates other than isosorbide dinitrate. Nitrates may be useful, however, for concomitant chronic myocardial ischemia.

Patients with stable HF should be encouraged to begin and maintain a regular aerobic exercise program. The level of exercise can range from brief, symptom-limited exercise to moderate exercise (60% capacity) for 3 or more hours per week.

The use of antiplatelet therapy or the routine use of anticoagulation in patients with HF who are in sinus rhythm provides no benefit. Anticoagulation may be useful if the patient has severe HF or has a known mural thrombus. HF patients with atrial fibrillation should be considered for antiplatelet or anticoagulation therapy as described elsewhere.⁵⁸

Short-acting CCBs may worsen HF. No data support the use of any CCB in the primary treatment of HF. Similarly, intermittent use of milrinone or dobutamine is not indicated.

Symptom control. The symptomatic treatment of HF includes the use of diuretics and dietary sodium restriction to control sodium levels and volume status. Symptom control should be accomplished along with the pharmacologic disease management outlined above.

The role of digoxin in the failing heart without dysrhythmias is unclear. Digoxin may be most useful in symptom control, as it reduces hospitalizations attributed to worsening HF. This benefit must be balanced against an increased risk of hospitalization caused by digoxin toxicity. Patients who are already taking digoxin should probably continue to do so. The role of digoxin in newly diagnosed HF patients is unknown.

Close follow-up. Comprehensive follow-up, with the patient as a more active participant and in which care is extended beyond the hospital or office to the home, is a key strategy in the long-term care of HF patients. This aspect of HF management should include educating patients about their disease process and their dietary and pharmacologic treatments; teaching them how to monitor their weight, symptoms, and blood pressure and to understand when to seek care; and following up periodically by telephone between scheduled office visits.

FIGURE
Management of adults with heart failure

Heart failure (HF) affects more than 2 million adults in the United States.¹ This common, costly, and disabling disorder mainly affects the elderly, with prevalence rates of up to 10% in patients older than 65 years.^2,3 The management of HF is responsible for millions of outpatient visits per year,⁴ is the most common discharge diagnosis for Medicare beneficiaries,⁵ and accounts for more than 5% of total health care dollars spent.⁶

Treatment

Major advances in the pharmacologic treatment of heart failure (HF) have emerged in recent years. An approach to the diagnosis and evaluation of HF is described elsewhere.⁷ This article summarizes the evidence for outpatient treatment of HF. Current intervention trials do not distinguish between systolic and diastolic heart failure; it is therefore unknown whether or how drug therapy should be tailored according to the type of HF. The treatment of cardiac dysrhythmias in the setting of HF is beyond the scope of this article and is presented elsewhere.⁸Table 1 compares the available outpatient treatments of HF and includes the levels of evidence, numbers needed to treat, and appropriate situations for use. In the remainder of this article, we will discuss pharmacologic and nonpharmacologic management, including identification of ineffective treatments.

TABLE 1
Treatment options in heart failure

Pharmacologic treatment

Angiotensin-converting enzyme inhibitors. A systematic review⁹ of 32 trials with a total of 7105 patients demonstrated that mortality rates were lower in patients taking an angiotensin-converting enzyme (ACE) inhibitor than in those not taking one (number needed to treat [NNT] = 24 for > 90 days, meaning that 1 fewer death occurs for every 24 patients who take an ACE inhibitor for more than 90 days). In addition, there is a reduction in the combined endpoints of death and hospitalization because of HF (NNT = 11). Although most of this benefit was realized in the first 90 days of therapy, benefits lasted for 4 to 5 years and were more pronounced in patients categorized in more severe New York Heart Association (NYHA) HF classes¹⁰ (class I: no limitation of activities; class II: slight limitation of activity; class III: marked limitation of activity and comfortable only at rest; class IV: symptoms at rest).

Dosage comparison studies demonstrate that HF patients can benefit from even moderate doses of ACE inhibitors. A recent multicenter trial comparing moderate dose enalapril (10 mg twice a day) with a higher dose (30 mg twice a day) in patients with a left ventricular ejection fraction (LVEF) of less than 20% found no differences in mortality at 1 year between the 2 groups.¹¹ In addition, both groups achieved similar increases in functional status and LVEF.

Several trials have demonstrated good tolerability of ACE inhibitors.^12-14 Dropout rates of 15% to 30% were similar between patients in the ACE inhibitor and placebo groups, mainly because of side effects, including dizziness, altered taste, hypotension, hyperkalemia, and cough.

Angiotensin-receptor blockers. Angiotensin-receptor blockers (ARBs) reduce all-cause mortality and HF-related hospitalizations in patients with NYHA class II and III HF at rates comparable with those of ACE inhibitors.^15,16 Cough is not a side effect of ARBs. Although they are more expensive, ARBs offer a reasonable alternative for patients who do not tolerate ACE inhibitors.

Beta-blockers. The beta-blockers carvedilol, metoprolol, and bisoprolol have a proven mortality benefit for patients with HF.^17-19 Pooled results of 6 randomized controlled trials (RCTs), including more than 9000 patients already taking ACE inhibitors, showed a significant reduction in total mortality (NNT = 24 over 1–2 years) and sudden death (NNT = 35), regardless of NYHA classification.²⁰ The average dropout rate of 16% was similar in the betablocker and placebo groups.

Early beta-blocker studies included few NYHA class IV patients until a recent study of the use of carvedilol in severe chronic HF.²¹ In this study, all patients were taking diuretics plus either an ACE inhibitor or ARB and were permitted to take digoxin, nitrates, hydralazine, spironolactone, or amiodarone. Carvedilol at an average dose of 37 mg per day decreased mortality (NNT = 18 for 10 months) and lowered combined mortality and hospitalization for worsening HF (NNT = 13). Study patients taking carvedilol withdrew from the study at a lower rate (approximately 15%) than placebo.

Because the pharmacologic properties of betablockers vary, clinicians have wondered which are most beneficial. The investigators in a study comparing metoprolol (a beta-1 antagonist) with carvedilol (a beta-1, beta-2, and alpha-1 antagonist) in NYHA class II or III patients found no differences in quality-of-life measures or changes in NYHA classification.²²

Spironolactone. The addition of spironolactone to standard care can help patients with severe HF.²³ In NYHA class III and IV HF patients, spironolactone at doses ranging from 25 mg every other day to 50 mg per day reduces mortality (NNT = 9 for 2 years), reduces hospitalization from all cardiac causes (NNT = 4), and reduces hospitalization for worsening HF (NNT = 3). The most common serious adverse event in the spironolactone group was severe hyperkalemia (number needed to harm [NNH] = 195). Ten percent of men taking spironolactone experienced breast pain and gynecomastia.

Hydralazine and isosorbide dinitrate. The combination of hydralazine and isosorbide dinitrate (ISDN) reduces mortality in HF patients, but tolerability is an issue. In earlier trials, men with HF symptoms that were optimally controlled with digoxin and diuretics and treated with hydralazine (average dose = 270 mg/day) plus ISDN (average dose = 136 mg/day) had a decrease in all-cause mortality of 28% (NNT = 19 for 6 years).²⁴ A more recent trial comparing hydralazine plus ISDN with enalapril²⁵ (average daily doses of hydralazine = 300 mg/day; ISDN = 160 mg/day; enalapril = 20 mg/day) in NYHA class II–III patients showed no differences in mortality between the 2 groups over 3 years. Tolerability was a problem in these trials; more than 30% of patients stopped taking hydralazine, nitrate, or both.

Digoxin. Digoxin is effective for treating the symptoms of HF in the absence of dysrhythmias but there are no data demonstrating a mortality benefit. Digoxin increases functional capacity in NYHA class II–III patients and heart failure symptoms worsen if digoxin is withdrawn.²⁶ Although there are no differences in all-cause mortality with the use of digoxin, there are fewer hospitalizations due to worsening HF (NNT = 27–114 over 3 years) and a lower rate of clinical deterioration (NNT = 4–75).²⁷ In a randomized trial comparing digoxin and placebo, patients taking digoxin were twice as likely to be hospitalized for suspected digoxin toxicity (2.0% vs 0.9%; P < .001; NNH = 52).²⁸

Diuretics. Diuretics are a mainstay of the symptomatic treatment of heart failure. Short-term studies have shown that diuretics improve the symptoms of sodium and fluid retention and increase exercise tolerance and cardiac function regardless of NYHA classification.^29-32 No studies that examine their effects on morbidity and mortality are available.

Antiplatelet therapy and anticoagulation. Patients with HF have an increased risk for thromboembolic events of 1.6% to 3.2% per year.³³ One systematic review concluded that antiplatelet therapy is not useful in preventing thromboembolism in patients with HF in sinus rhythm and may even be harmful.³⁴ Another systematic review also concluded that the data do not support the routine use of anticoagulants (eg, warfarin) in patients with HF and sinus rhythm.³⁵ Anticoagulation may be beneficial, however, if there is echocardiographic visualization of a left ventricular thrombus or in cases of “severe” HF or concomitant atrial fibrillation.³⁵

Nonpharmacologic management

Dietary sodium restriction. There is consensus that dietary sodium restriction is important in the treatment of HF³⁶ and is recommended in published guidelines.^8,37 Sodium restriction assists with fluid volume control and minimizes the dosages of HF drugs used. These recommendations are based on the retention of sodium and water in symptomatic HF. No studies, however, have examined the effect of dietary sodium restriction on morbidity or mortality, either alone or in combination with pharmacologic treatments.

Exercise training. Moderate exercise training improves quality of life and decreases mortality in patients with stable chronic HF. A recent RCT demonstrated a decrease in mortality (NNT = 4 for 14 months) and hospital readmission for HF (NNT = 5) with only moderate exercise on a stationary bicycle (60% of maximum exercise capacity) for 2 to 3 hours per week.³⁸ Other studies have demonstrated improvements in physiologic markers³⁹ and in quality-of-life ratings with short-term, symptom-limited exercise.⁴⁰

Multidisciplinary or case-management approach. A case-based or disease-management approach to patients with HF decreases the frequency of unplanned and repeat hospitalizations, increases functional status, and increases quality of life.⁴¹ Even a single in-home visit by a clinical pharmacist and a nurse results in fewer unplanned readmissions and fewer days of hospitalization up to 18 months after discharge.^42,43 A small study of 27 patients in a Veterans Affairs hospital demonstrated that patient instruction in the self-monitoring of weight and blood pressure, combined with frequent telephone follow-up from a nurse, lowered repeat hospitalizations over 1 year, with the effect more pronounced in patients with more severe NYHA classifications.⁴⁴ A large RCT demonstrated that a multidisciplinary management approach (intensive patient education about HF and its treatment, dietary assessment and instruction, medication analysis and elimination of unnecessary medications, and telephone and home visit follow-up) results in fewer hospitalizations (NNT = 5 for 3 months) and reduced costs of care.⁴⁵

Treatments that have no benefit or are harmful

Calcium-channel blockers. Although some of the newer, longer-acting calcium-channel blockers (CCBs) appear to be safe in the treatment of heart failure,^46-49 no trials are available demonstrating that they lower mortality, decrease hospitalizations, or improve quality of life in patients with a failing heart. Older, short-acting CCBs can worsen HF.⁵⁰

Positive inotropic therapy. Intermittent positive inotropic therapy, either orally (milrinone) or intravenously (dobutamine), should be avoided. Although short-term studies have shown some increase in cardiac function and symptoms,⁵¹ long-term studies demonstrate no mortality benefit.⁵² One RCT of milrinone demonstrated an increase in mortality (NNH = 17 for 5 months), an increased rate of hospitalization for worsening HF (NNH = 20), and more serious side effects (NNH = 25).⁵³

Prognosis

Despite the increased longevity in Western developed nations and increased survival from coronary artery disease over recent decades, the overall prognosis of HF has improved very little.^6,54 Mortality data derived from several different sources, the largest being the Framingham Heart Study,^2,55 have shown that HF remains highly lethal, with a 5-year survival rate of 25% in men and 38% in women with NYHA II–IV heart failure. Mortality data from the placebo arms of intervention trials show an average 1-year mortality of 18%.^{9,17,19,20,56} A recent population-based study of patients with a new diagnosis of HF showed survival rates of only 62% at 12 months and 57% at 18 months.⁵⁷ Despite these dismal population-based data, predicting the likelihood of survival in individuals with HF is largely unreliable.⁸ Estimating individual prognosis is only somewhat useful in making end-of-life care and hospice decisions for patients with very advanced HF. Table 2 summarizes specific prognostic factors for patients with HF.

TABLE 2
Factors that affect prognosis in patients with heart failure (HF)

Suggested management of patients with heart failure

Although the optimal sequence of pharmacologic interventions for treating HF has not been examined in RCTs, recommendations can be made on the basis of existing evidence in HF management (Figure). This approach can be divided into 4 steps performed simultaneously: (1) control risks for the development and progression of HF (treat concomitant diseases); (2) HF treatment; (3) symptom control; (4) close follow-up.

Control risks. Risks for the development and worsening of HF should be addressed as described else-where.⁸ Steps include longitudinal surveillance; identification and treatment of hypertension, diabetes and thyroid diseases; management of atherosclerotic and coronary artery disease and myocardial ischemia; and the elimination of alcohol and tobacco use.

Heart failure treatment. All patients with HF should take a drug or a combination of drugs that affects the disease process. Drugs shown by the preponderance of evidence to decrease morbidity and mortality include ACE inhibitors, beta-blockers, and ARBs. For most HF patients, regardless of NYHA class, ACE inhibitors should be the initial baseline treatment because of their proven track record and the observation that most recent HF trials include patients who are already taking these medications. ARBs are similar in efficacy to ACE inhibitors and, therefore, are an adequate alternative when ACE inhibitors are not tolerated. Beta-blockers (metoprolol and bisoprolol) added to ACE inhibitors are also useful as a baseline treatment in most HF patients and may be especially useful in the case of tachydysrhythmias and in the postmyocardial infarction period.

For severe HF (NYHA III–IV), spironolactone and carvedilol are useful additions to baseline drug therapy. Carvedilol may be added if a beta-blocker is not currently used. If the patient is currently taking a beta-blocker, the drug should be discontinued before the patient is switched to carvedilol.

The hydralazine–nitrate combination has been proved effective, but tolerability and ease-of-use issues limit its usefulness. No data are available to support the use of nitrates other than isosorbide dinitrate. Nitrates may be useful, however, for concomitant chronic myocardial ischemia.

Patients with stable HF should be encouraged to begin and maintain a regular aerobic exercise program. The level of exercise can range from brief, symptom-limited exercise to moderate exercise (60% capacity) for 3 or more hours per week.

The use of antiplatelet therapy or the routine use of anticoagulation in patients with HF who are in sinus rhythm provides no benefit. Anticoagulation may be useful if the patient has severe HF or has a known mural thrombus. HF patients with atrial fibrillation should be considered for antiplatelet or anticoagulation therapy as described elsewhere.⁵⁸

Short-acting CCBs may worsen HF. No data support the use of any CCB in the primary treatment of HF. Similarly, intermittent use of milrinone or dobutamine is not indicated.

Symptom control. The symptomatic treatment of HF includes the use of diuretics and dietary sodium restriction to control sodium levels and volume status. Symptom control should be accomplished along with the pharmacologic disease management outlined above.

The role of digoxin in the failing heart without dysrhythmias is unclear. Digoxin may be most useful in symptom control, as it reduces hospitalizations attributed to worsening HF. This benefit must be balanced against an increased risk of hospitalization caused by digoxin toxicity. Patients who are already taking digoxin should probably continue to do so. The role of digoxin in newly diagnosed HF patients is unknown.

Close follow-up. Comprehensive follow-up, with the patient as a more active participant and in which care is extended beyond the hospital or office to the home, is a key strategy in the long-term care of HF patients. This aspect of HF management should include educating patients about their disease process and their dietary and pharmacologic treatments; teaching them how to monitor their weight, symptoms, and blood pressure and to understand when to seek care; and following up periodically by telephone between scheduled office visits.

FIGURE
Management of adults with heart failure

Between 1993 and 1995 more than 20 million visits were provided to children younger than 15 years for otitis media; 77% of these were for children aged 4 years and younger.¹ Acute otitis media (AOM) is the most frequent primary diagnosis in preschool children and accounts for almost 20% of ambulatory care visits in this age group. By the age of 3 months, 10% of children will be given at least 1 diagnosis of otitis media and more than 90% by the age of 2 years.² Peak incidence occurs between the ages 6 and 15 months, although there is a second peak at approximately age 5 years that is thought to be associated with entrance into school.³ In the mid-1990s, treatment of otitis media cost $3.8 billion per year4; 20% of the more than 110 million prescriptions for oral antibiotics are for otitis media.⁵

The most important contributor to AOM is a dysfunction of the eustachian tube, allowing reflux of fluid and bacteria into the middle ear space from the nasopharynx.⁶ This dysfunction is usually multifactorial and is likely a combination of anatomy (shorter and more flexible eustachian tubes) and function (inefficiency at clearing secretions and equilibrating negative intratympanic pressures) in younger children.⁷ Acute viral upper respiratory infections create inflammation and secretions that magnify this eustachian tube dysfunction and predispose to or induce AOM.

The 3 most common bacteria in AOM are Streptococcus pneumoniae, Haemophilus species, and Branhamella catarrhalis. A recent study showed that both bacteria and viruses were isolated in the middle ear fluid of 65% of children with otitis media. In fact, 35% had viruses isolated as the sole middle ear pathogen.⁸ Other studies have failed to identify a specific infectious agent in a significant number of middle ear fluid aspirates.⁹

Diagnosis

The diagnosis of AOM in children is often based on a combination of symptoms and physical findings. It is usually defined as bulging or opacification of the tympanic membrane with or without erythema, middle ear effusion, marked decrease or absence of tympanic membrane mobility, and accompanied by at least one of the following signs and symptoms of acute infection: fever, otalgia, irritability, otorrhea, lethargy, anorexia, vomiting, or diarrhea.10 Although the best reference standard for the diagnosis of AOM is myringotomy or tympanocentesis, no published studies were identified in which this reference was performed in all study patients. Most published studies use pneumatic otoscopy combined with tympanocentesis or myringotomy as the reference standard when middle ear effusion is suspected.

Most symptoms are not specific for AOM. In a survey of patients seeing general practitioners in Finland,11 the symptoms that most increased the likelihood of diagnosing AOM were earache (relative risk [RR]=5.4; 95% confidence interval [CI], 3.3-8.9), rubbing of the ear (RR=5.0; 95% CI, 2.9-8.6), and excessive crying (RR=2.8; 95% CI, 1.8-4.3). Although fever, earache, or excessive crying were present in 90% of children considered to have AOM, one or more of these symptoms were also present in 72% of children who did not have otitis media. In another survey of children aged younger than 4 years with a diagnosis of AOM, 40% did not have otalgia, and 30% did not have fever.¹²

Physical examination findings are of variable reliability. Findings that physicians rely on to diagnose AOM, such as a bulging or erythematous tympanic membrane, may occasionally be found in normal ears.13 One study compared the 3 most commonly documented otoscopic findings (color, position, and mobility of the tympanic membrane) with those of pneumatic otoscopy and myringotomy.14 Only 65% of patients with a distinctly red tympanic membrane and 16% with a slightly red tympanic membrane had AOM. A cloudy tympanic membrane was the most predictive color change, with an 80% positive predictive value (PPV), the percentage of patients with the symptom who have AOM. A bulging tympanic membrane was most predictive of AOM (PPV=89%), despite its 27% false-negative rate. Retraction was found in only 19% of children with AOM and had a PPV of only 50%. Distinctly impaired mobility was predictive of otitis media (PPV=78%), but diminished mobility was also found in 30% of children without middle ear effusion. These data and those presented in Table 1 show that each of these otoscopic examination elements (color, position, mobility) alone is inadequate for discriminating between cases in which the diagnosis of AOM is uncertain.

However, combining these signs can be useful.¹⁵ A cloudy (opaque), bulging, and immobile tympanic membrane on pneumatic otoscopy was nearly 100% predictive of otitis media in children with acute symptoms. In addition, 94% of children with the combination of distinctly red erythema, bulging, and immobility had AOM. No combination of findings that included only slight redness was more than 53% predictive of AOM.

In children aged younger than 1 year in whom otoscopy can be quite difficult, tympanometry can be a useful tool for detecting a middle ear effusion.¹⁶ An abnormal type B tympanogram (flat curve with no distinct peak) in infants presenting with acute symptoms is strong evidence in favor of AOM, although a normal test is not helpful in ruling out the diagnosis.

The diagnosis of AOM can be influenced by the physician’s perception of parental expectations for antibiotics.¹⁷ A study of children presenting with ear pain or other upper respiratory symptoms found that physicians diagnosed AOM 49% of the time when they perceived parents wanted antibiotics and only 13% of the time when they thought parents did not want antibiotics. Physicians were 23 times more likely to prescribe an antibiotic for an upper respiratory illness if they perceived that parents expected antimicrobials.

Treatment

A recent systematic review¹⁸ found that the symptoms of AOM (mainly otalgia) spontaneously resolved in two thirds of children by 24 hours and in 80% at 2 to 7 days. This was also observed in 2 earlier meta-analyses.^10,19 Seventeen children would need to be treated with antibiotics (vs placebo) for 1 child to have less pain at 2 to 7 days (number needed to treat [NNT]=17). There were no differences between antibiotic and placebo groups in other clinical outcomes, such as tympanometry findings, perforation, and recurrences. Also, children treated with antibiotics were almost twice as likely to have vomiting, diarrhea, or a rash.

Initially not treating uncomplicated AOM with antibiotics is an acceptable alternative. In a study by van Buchem and colleagues,²⁰ 90% of children with AOM recovered (symptoms resolved) in the first 4 days with nose drops and oral analgesics and without the use of antibiotics. Only 3% of the 4860 children in this study had a clinical course that required further treatment with antibiotics or myringotomy. A recent randomized controlled trial of 315 children demonstrated that children treated immediately with antibiotics had 1 less day of symptoms, but that 1 in 5 of these had diarrhea.²¹ The group not treated with antibiotics had no serious sequellae and used more analgesics. There were no differences in the number of missed school days, and more than 75% of the parents were satisfied with this “wait and see” approach. These studies also emphasize that antibiotics have a very modest effect on the clinical course of AOM and seem to decrease the duration of symptoms only to a small degree.

Physicians often recommend other symptomatic treatments for ear infections. Non–aspirin analgesics are effective in relieving pain,²¹ as are ibuprofen²² and Auralgan.²³ Antihistamine-decongestant preparations offer no added benefit in resolution of symptoms and have no effect on clinical outcomes when given with antibiotics.^24,25

If antibiotics are used, the meta-analysis by Rosenfeld and colleagues¹⁰ showed there were no differences in outcomes between treatment with cefaclor, cefixime, erythromycin, trimethoprim-sulfamethoxazole, amoxicillin clavulanate, or erythromycin sulfisoxazole, and treatment with amoxicillin or ampicillin. Broader spectrum and/or more expensive antibiotics therefore offer no advantage over amoxicillin for the initial treatment of AOM. In another meta-analysis, Kozyrskyj and coworkers26 showed that a 5-day antibiotic course was an acceptable alternative to a 8- to 14-day treatment course. There was a slightly increased risk of treatment failure at 1-month follow-up with the shortened course of antibiotics; the NNT to prevent 1 excess failure at 30 days was 17. There were no differences in long-term outcomes (2-3 months) or medication side effects (vomiting, diarrhea, rash) between the short and long antibiotic courses. The broader spectrum azithromycin and intramuscular ceftriaxone offered no advantage over amoxicillin. Children aged younger than 2 years deserve special mention, since they are at higher risk for treatment failures,^27,28 persistent symptoms,²⁹ and recurrent otitis media.³⁰ Few well-designed studies exist to guide treatment in this age group. Although children aged younger than 2 years were not excluded, neither review by Glasziou and colleagues¹⁸ or Rosenfeld and coworkers¹⁰ specifically examined this age group.

Another review demonstrated that (as with older children) routinely using antibiotics initially does not seem to add any clinical benefit.³¹ A recent randomized controlled trial of 240 children demonstrated that 8 children in this age group would have to be treated with amoxicillin for 1 child to have fewer symptoms (fever, crying, irritability) at 4 days (NNT=8).³² The major benefit of amoxicillin in this study was 1 day less of fever (P=.004). Adverse effects were almost twice as likely in the amoxicillin group, although this difference was not statistically significant. There were also no differences between the groups in clinical failure rates at 11 days or in the likelihood of recurrent otitis media, antibiotic use, specialist referrals, or surgery at 6 weeks. Effects on hearing were not measured. The authors conclude that “this modest effect does not justify prescription of antibiotics at the first visit, provided close surveillance can be guaranteed.” Also, Kozyrskyj and colleagues²⁶ demonstrated in their meta-analysis that as a subgroup there were no differences in clinical failures between 5 and 10 days of antibiotics in children aged younger than 2 years. However, there were only 118 children in this age group. Table 2 shows treatment options for AOM.

In recent years, there has been an increasing concern about worldwide bacterial resistance to antimicrobial drugs^33-35 by the World Health Organization³⁶ and the Centers for Disease Control and Prevention (CDC).³⁷ In response to the increasing antimicrobial resistance patterns seen in the common middle ear pathogens, especially S pneumoniae, the CDC Drug-resistant Streptococcus pneumoniae Therapeutic Working Group recommends doubling the dosage of amoxicillin to 80 to 90 mg per kg per day in the empiric treatment of AOM.³⁸ These recommendations are based on in vitro mean inhibitory concentration data of S pneumoniae cultures from middle ear fluid and nasopharyngeal swabs. However, there currently is no patient-oriented evidence to suggest that increasing the amoxicillin dosage actually decreases suppurative or invasive complications of AOM (meningitis, mastoiditis, and so forth), affects recurrence rates or treatment success, affects long-term outcomes of AOM, or even decreases the rate of drug-resistant S pneumoniae. Also, bacteriologic outcomes do not correlate with clinical outcomes. In the meta-analysis by Rosenfeld and coworkers,¹⁰ 89% of middle ear pathogens from treatment failures were susceptible in vitro to the antibiotic prescribed, and 13% of isolates from clinical cures were resistant in vitro to the prescribed antibiotic. We must use extreme caution in extrapolating the microbiologic findings to the clinical care of the child with AOM.

Prognosis

There are insufficient data to suggest that routine antibiotic use in AOM results in fewer cases of mastoiditis or meningitis. In the systematic reviews cited, the incidence of these suppurative complications was rare. In the Cochrane review by Glasziou and colleagues,¹⁸ only 1 case of mastoiditis developed in 2202 children, and this was in a child treated with penicillin. In the Netherlands among 4860 consecutive children with AOM, 2 experienced mastoiditis (both responded to outpatient antibiotic therapy), and there were no cases of meningitis.²⁰ In the meta-analysis by Rosenfeld and coworkers,¹⁰ there were no suppurative complications in the 5400 children studied. Although mastoiditis has been quoted as being more common in the preantibiotic era, it is unclear if the current rarity of this condition is due to antibiotic treatment, changes in organism virulence or host defenses, or the assertion that uncomplicated otitis media was often not reported, thus increasing the relative rate of mastoiditis.²⁷ In a recent review,³⁹ antibiotics did not seem to have an appreciable effect on complication rates, leading the authors to conclude that “antibiotic treatment for AOM cannot be considered as a safeguard against the development of complications.” Even in developing countries where the burden of otitis media is great, mastoiditis is quite rare, with a prevalence rate of much less than 1%.⁴⁰

Antibiotic use does influence bacterial resistance rates. In children previously treated with antibiotics for AOM, there is a 3-fold increased risk of isolating drug-resistant organisms from middle ear effusions with subsequent bouts of otitis media.^41,42 In the Netherlands and Iceland, routinely not treating AOM with antibiotics has resulted in a reduction in antibiotic resistance.^33,43 The most important risk factors for a poor outcome are age younger than 2 years and attendance at a daycare center.40 Children in daycare have a higher risk of requiring a hospital admission and up to a 50% increased risk of repeated or recurrent ear infections.^44,45 Children with chronic underlying illnesses or chronic otitis media with effusion have added risks of poor outcomes that are beyond the scope of this review.

Conclusions

Thus, the natural course of AOM in children is quite favorable. If left untreated, 80% will recover spontaneously within 2 weeks. The addition of antibiotics provides at best a modest reduction in symptoms, while adding cost, adverse drug effects, and increasing bacterial resistance in the patient and community. Minimizing the use of antibiotics in patients with AOM does not increase the risks of perforation, deafness, or contralateral or recurrent AOM.

Between 1993 and 1995 more than 20 million visits were provided to children younger than 15 years for otitis media; 77% of these were for children aged 4 years and younger.¹ Acute otitis media (AOM) is the most frequent primary diagnosis in preschool children and accounts for almost 20% of ambulatory care visits in this age group. By the age of 3 months, 10% of children will be given at least 1 diagnosis of otitis media and more than 90% by the age of 2 years.² Peak incidence occurs between the ages 6 and 15 months, although there is a second peak at approximately age 5 years that is thought to be associated with entrance into school.³ In the mid-1990s, treatment of otitis media cost $3.8 billion per year4; 20% of the more than 110 million prescriptions for oral antibiotics are for otitis media.⁵

The most important contributor to AOM is a dysfunction of the eustachian tube, allowing reflux of fluid and bacteria into the middle ear space from the nasopharynx.⁶ This dysfunction is usually multifactorial and is likely a combination of anatomy (shorter and more flexible eustachian tubes) and function (inefficiency at clearing secretions and equilibrating negative intratympanic pressures) in younger children.⁷ Acute viral upper respiratory infections create inflammation and secretions that magnify this eustachian tube dysfunction and predispose to or induce AOM.

The 3 most common bacteria in AOM are Streptococcus pneumoniae, Haemophilus species, and Branhamella catarrhalis. A recent study showed that both bacteria and viruses were isolated in the middle ear fluid of 65% of children with otitis media. In fact, 35% had viruses isolated as the sole middle ear pathogen.⁸ Other studies have failed to identify a specific infectious agent in a significant number of middle ear fluid aspirates.⁹

Diagnosis

The diagnosis of AOM in children is often based on a combination of symptoms and physical findings. It is usually defined as bulging or opacification of the tympanic membrane with or without erythema, middle ear effusion, marked decrease or absence of tympanic membrane mobility, and accompanied by at least one of the following signs and symptoms of acute infection: fever, otalgia, irritability, otorrhea, lethargy, anorexia, vomiting, or diarrhea.10 Although the best reference standard for the diagnosis of AOM is myringotomy or tympanocentesis, no published studies were identified in which this reference was performed in all study patients. Most published studies use pneumatic otoscopy combined with tympanocentesis or myringotomy as the reference standard when middle ear effusion is suspected.

Most symptoms are not specific for AOM. In a survey of patients seeing general practitioners in Finland,11 the symptoms that most increased the likelihood of diagnosing AOM were earache (relative risk [RR]=5.4; 95% confidence interval [CI], 3.3-8.9), rubbing of the ear (RR=5.0; 95% CI, 2.9-8.6), and excessive crying (RR=2.8; 95% CI, 1.8-4.3). Although fever, earache, or excessive crying were present in 90% of children considered to have AOM, one or more of these symptoms were also present in 72% of children who did not have otitis media. In another survey of children aged younger than 4 years with a diagnosis of AOM, 40% did not have otalgia, and 30% did not have fever.¹²

Physical examination findings are of variable reliability. Findings that physicians rely on to diagnose AOM, such as a bulging or erythematous tympanic membrane, may occasionally be found in normal ears.13 One study compared the 3 most commonly documented otoscopic findings (color, position, and mobility of the tympanic membrane) with those of pneumatic otoscopy and myringotomy.14 Only 65% of patients with a distinctly red tympanic membrane and 16% with a slightly red tympanic membrane had AOM. A cloudy tympanic membrane was the most predictive color change, with an 80% positive predictive value (PPV), the percentage of patients with the symptom who have AOM. A bulging tympanic membrane was most predictive of AOM (PPV=89%), despite its 27% false-negative rate. Retraction was found in only 19% of children with AOM and had a PPV of only 50%. Distinctly impaired mobility was predictive of otitis media (PPV=78%), but diminished mobility was also found in 30% of children without middle ear effusion. These data and those presented in Table 1 show that each of these otoscopic examination elements (color, position, mobility) alone is inadequate for discriminating between cases in which the diagnosis of AOM is uncertain.

However, combining these signs can be useful.¹⁵ A cloudy (opaque), bulging, and immobile tympanic membrane on pneumatic otoscopy was nearly 100% predictive of otitis media in children with acute symptoms. In addition, 94% of children with the combination of distinctly red erythema, bulging, and immobility had AOM. No combination of findings that included only slight redness was more than 53% predictive of AOM.

In children aged younger than 1 year in whom otoscopy can be quite difficult, tympanometry can be a useful tool for detecting a middle ear effusion.¹⁶ An abnormal type B tympanogram (flat curve with no distinct peak) in infants presenting with acute symptoms is strong evidence in favor of AOM, although a normal test is not helpful in ruling out the diagnosis.

The diagnosis of AOM can be influenced by the physician’s perception of parental expectations for antibiotics.¹⁷ A study of children presenting with ear pain or other upper respiratory symptoms found that physicians diagnosed AOM 49% of the time when they perceived parents wanted antibiotics and only 13% of the time when they thought parents did not want antibiotics. Physicians were 23 times more likely to prescribe an antibiotic for an upper respiratory illness if they perceived that parents expected antimicrobials.

Treatment

A recent systematic review¹⁸ found that the symptoms of AOM (mainly otalgia) spontaneously resolved in two thirds of children by 24 hours and in 80% at 2 to 7 days. This was also observed in 2 earlier meta-analyses.^10,19 Seventeen children would need to be treated with antibiotics (vs placebo) for 1 child to have less pain at 2 to 7 days (number needed to treat [NNT]=17). There were no differences between antibiotic and placebo groups in other clinical outcomes, such as tympanometry findings, perforation, and recurrences. Also, children treated with antibiotics were almost twice as likely to have vomiting, diarrhea, or a rash.

Initially not treating uncomplicated AOM with antibiotics is an acceptable alternative. In a study by van Buchem and colleagues,²⁰ 90% of children with AOM recovered (symptoms resolved) in the first 4 days with nose drops and oral analgesics and without the use of antibiotics. Only 3% of the 4860 children in this study had a clinical course that required further treatment with antibiotics or myringotomy. A recent randomized controlled trial of 315 children demonstrated that children treated immediately with antibiotics had 1 less day of symptoms, but that 1 in 5 of these had diarrhea.²¹ The group not treated with antibiotics had no serious sequellae and used more analgesics. There were no differences in the number of missed school days, and more than 75% of the parents were satisfied with this “wait and see” approach. These studies also emphasize that antibiotics have a very modest effect on the clinical course of AOM and seem to decrease the duration of symptoms only to a small degree.

Physicians often recommend other symptomatic treatments for ear infections. Non–aspirin analgesics are effective in relieving pain,²¹ as are ibuprofen²² and Auralgan.²³ Antihistamine-decongestant preparations offer no added benefit in resolution of symptoms and have no effect on clinical outcomes when given with antibiotics.^24,25

If antibiotics are used, the meta-analysis by Rosenfeld and colleagues¹⁰ showed there were no differences in outcomes between treatment with cefaclor, cefixime, erythromycin, trimethoprim-sulfamethoxazole, amoxicillin clavulanate, or erythromycin sulfisoxazole, and treatment with amoxicillin or ampicillin. Broader spectrum and/or more expensive antibiotics therefore offer no advantage over amoxicillin for the initial treatment of AOM. In another meta-analysis, Kozyrskyj and coworkers26 showed that a 5-day antibiotic course was an acceptable alternative to a 8- to 14-day treatment course. There was a slightly increased risk of treatment failure at 1-month follow-up with the shortened course of antibiotics; the NNT to prevent 1 excess failure at 30 days was 17. There were no differences in long-term outcomes (2-3 months) or medication side effects (vomiting, diarrhea, rash) between the short and long antibiotic courses. The broader spectrum azithromycin and intramuscular ceftriaxone offered no advantage over amoxicillin. Children aged younger than 2 years deserve special mention, since they are at higher risk for treatment failures,^27,28 persistent symptoms,²⁹ and recurrent otitis media.³⁰ Few well-designed studies exist to guide treatment in this age group. Although children aged younger than 2 years were not excluded, neither review by Glasziou and colleagues¹⁸ or Rosenfeld and coworkers¹⁰ specifically examined this age group.

Another review demonstrated that (as with older children) routinely using antibiotics initially does not seem to add any clinical benefit.³¹ A recent randomized controlled trial of 240 children demonstrated that 8 children in this age group would have to be treated with amoxicillin for 1 child to have fewer symptoms (fever, crying, irritability) at 4 days (NNT=8).³² The major benefit of amoxicillin in this study was 1 day less of fever (P=.004). Adverse effects were almost twice as likely in the amoxicillin group, although this difference was not statistically significant. There were also no differences between the groups in clinical failure rates at 11 days or in the likelihood of recurrent otitis media, antibiotic use, specialist referrals, or surgery at 6 weeks. Effects on hearing were not measured. The authors conclude that “this modest effect does not justify prescription of antibiotics at the first visit, provided close surveillance can be guaranteed.” Also, Kozyrskyj and colleagues²⁶ demonstrated in their meta-analysis that as a subgroup there were no differences in clinical failures between 5 and 10 days of antibiotics in children aged younger than 2 years. However, there were only 118 children in this age group. Table 2 shows treatment options for AOM.

In recent years, there has been an increasing concern about worldwide bacterial resistance to antimicrobial drugs^33-35 by the World Health Organization³⁶ and the Centers for Disease Control and Prevention (CDC).³⁷ In response to the increasing antimicrobial resistance patterns seen in the common middle ear pathogens, especially S pneumoniae, the CDC Drug-resistant Streptococcus pneumoniae Therapeutic Working Group recommends doubling the dosage of amoxicillin to 80 to 90 mg per kg per day in the empiric treatment of AOM.³⁸ These recommendations are based on in vitro mean inhibitory concentration data of S pneumoniae cultures from middle ear fluid and nasopharyngeal swabs. However, there currently is no patient-oriented evidence to suggest that increasing the amoxicillin dosage actually decreases suppurative or invasive complications of AOM (meningitis, mastoiditis, and so forth), affects recurrence rates or treatment success, affects long-term outcomes of AOM, or even decreases the rate of drug-resistant S pneumoniae. Also, bacteriologic outcomes do not correlate with clinical outcomes. In the meta-analysis by Rosenfeld and coworkers,¹⁰ 89% of middle ear pathogens from treatment failures were susceptible in vitro to the antibiotic prescribed, and 13% of isolates from clinical cures were resistant in vitro to the prescribed antibiotic. We must use extreme caution in extrapolating the microbiologic findings to the clinical care of the child with AOM.

Prognosis

There are insufficient data to suggest that routine antibiotic use in AOM results in fewer cases of mastoiditis or meningitis. In the systematic reviews cited, the incidence of these suppurative complications was rare. In the Cochrane review by Glasziou and colleagues,¹⁸ only 1 case of mastoiditis developed in 2202 children, and this was in a child treated with penicillin. In the Netherlands among 4860 consecutive children with AOM, 2 experienced mastoiditis (both responded to outpatient antibiotic therapy), and there were no cases of meningitis.²⁰ In the meta-analysis by Rosenfeld and coworkers,¹⁰ there were no suppurative complications in the 5400 children studied. Although mastoiditis has been quoted as being more common in the preantibiotic era, it is unclear if the current rarity of this condition is due to antibiotic treatment, changes in organism virulence or host defenses, or the assertion that uncomplicated otitis media was often not reported, thus increasing the relative rate of mastoiditis.²⁷ In a recent review,³⁹ antibiotics did not seem to have an appreciable effect on complication rates, leading the authors to conclude that “antibiotic treatment for AOM cannot be considered as a safeguard against the development of complications.” Even in developing countries where the burden of otitis media is great, mastoiditis is quite rare, with a prevalence rate of much less than 1%.⁴⁰

Antibiotic use does influence bacterial resistance rates. In children previously treated with antibiotics for AOM, there is a 3-fold increased risk of isolating drug-resistant organisms from middle ear effusions with subsequent bouts of otitis media.^41,42 In the Netherlands and Iceland, routinely not treating AOM with antibiotics has resulted in a reduction in antibiotic resistance.^33,43 The most important risk factors for a poor outcome are age younger than 2 years and attendance at a daycare center.40 Children in daycare have a higher risk of requiring a hospital admission and up to a 50% increased risk of repeated or recurrent ear infections.^44,45 Children with chronic underlying illnesses or chronic otitis media with effusion have added risks of poor outcomes that are beyond the scope of this review.

Conclusions

Thus, the natural course of AOM in children is quite favorable. If left untreated, 80% will recover spontaneously within 2 weeks. The addition of antibiotics provides at best a modest reduction in symptoms, while adding cost, adverse drug effects, and increasing bacterial resistance in the patient and community. Minimizing the use of antibiotics in patients with AOM does not increase the risks of perforation, deafness, or contralateral or recurrent AOM.

Between 1993 and 1995 more than 20 million visits were provided to children younger than 15 years for otitis media; 77% of these were for children aged 4 years and younger.¹ Acute otitis media (AOM) is the most frequent primary diagnosis in preschool children and accounts for almost 20% of ambulatory care visits in this age group. By the age of 3 months, 10% of children will be given at least 1 diagnosis of otitis media and more than 90% by the age of 2 years.² Peak incidence occurs between the ages 6 and 15 months, although there is a second peak at approximately age 5 years that is thought to be associated with entrance into school.³ In the mid-1990s, treatment of otitis media cost $3.8 billion per year4; 20% of the more than 110 million prescriptions for oral antibiotics are for otitis media.⁵

The most important contributor to AOM is a dysfunction of the eustachian tube, allowing reflux of fluid and bacteria into the middle ear space from the nasopharynx.⁶ This dysfunction is usually multifactorial and is likely a combination of anatomy (shorter and more flexible eustachian tubes) and function (inefficiency at clearing secretions and equilibrating negative intratympanic pressures) in younger children.⁷ Acute viral upper respiratory infections create inflammation and secretions that magnify this eustachian tube dysfunction and predispose to or induce AOM.

The 3 most common bacteria in AOM are Streptococcus pneumoniae, Haemophilus species, and Branhamella catarrhalis. A recent study showed that both bacteria and viruses were isolated in the middle ear fluid of 65% of children with otitis media. In fact, 35% had viruses isolated as the sole middle ear pathogen.⁸ Other studies have failed to identify a specific infectious agent in a significant number of middle ear fluid aspirates.⁹

Diagnosis

The diagnosis of AOM in children is often based on a combination of symptoms and physical findings. It is usually defined as bulging or opacification of the tympanic membrane with or without erythema, middle ear effusion, marked decrease or absence of tympanic membrane mobility, and accompanied by at least one of the following signs and symptoms of acute infection: fever, otalgia, irritability, otorrhea, lethargy, anorexia, vomiting, or diarrhea.10 Although the best reference standard for the diagnosis of AOM is myringotomy or tympanocentesis, no published studies were identified in which this reference was performed in all study patients. Most published studies use pneumatic otoscopy combined with tympanocentesis or myringotomy as the reference standard when middle ear effusion is suspected.

Most symptoms are not specific for AOM. In a survey of patients seeing general practitioners in Finland,11 the symptoms that most increased the likelihood of diagnosing AOM were earache (relative risk [RR]=5.4; 95% confidence interval [CI], 3.3-8.9), rubbing of the ear (RR=5.0; 95% CI, 2.9-8.6), and excessive crying (RR=2.8; 95% CI, 1.8-4.3). Although fever, earache, or excessive crying were present in 90% of children considered to have AOM, one or more of these symptoms were also present in 72% of children who did not have otitis media. In another survey of children aged younger than 4 years with a diagnosis of AOM, 40% did not have otalgia, and 30% did not have fever.¹²

Physical examination findings are of variable reliability. Findings that physicians rely on to diagnose AOM, such as a bulging or erythematous tympanic membrane, may occasionally be found in normal ears.13 One study compared the 3 most commonly documented otoscopic findings (color, position, and mobility of the tympanic membrane) with those of pneumatic otoscopy and myringotomy.14 Only 65% of patients with a distinctly red tympanic membrane and 16% with a slightly red tympanic membrane had AOM. A cloudy tympanic membrane was the most predictive color change, with an 80% positive predictive value (PPV), the percentage of patients with the symptom who have AOM. A bulging tympanic membrane was most predictive of AOM (PPV=89%), despite its 27% false-negative rate. Retraction was found in only 19% of children with AOM and had a PPV of only 50%. Distinctly impaired mobility was predictive of otitis media (PPV=78%), but diminished mobility was also found in 30% of children without middle ear effusion. These data and those presented in Table 1 show that each of these otoscopic examination elements (color, position, mobility) alone is inadequate for discriminating between cases in which the diagnosis of AOM is uncertain.

However, combining these signs can be useful.¹⁵ A cloudy (opaque), bulging, and immobile tympanic membrane on pneumatic otoscopy was nearly 100% predictive of otitis media in children with acute symptoms. In addition, 94% of children with the combination of distinctly red erythema, bulging, and immobility had AOM. No combination of findings that included only slight redness was more than 53% predictive of AOM.

In children aged younger than 1 year in whom otoscopy can be quite difficult, tympanometry can be a useful tool for detecting a middle ear effusion.¹⁶ An abnormal type B tympanogram (flat curve with no distinct peak) in infants presenting with acute symptoms is strong evidence in favor of AOM, although a normal test is not helpful in ruling out the diagnosis.

The diagnosis of AOM can be influenced by the physician’s perception of parental expectations for antibiotics.¹⁷ A study of children presenting with ear pain or other upper respiratory symptoms found that physicians diagnosed AOM 49% of the time when they perceived parents wanted antibiotics and only 13% of the time when they thought parents did not want antibiotics. Physicians were 23 times more likely to prescribe an antibiotic for an upper respiratory illness if they perceived that parents expected antimicrobials.

Treatment

A recent systematic review¹⁸ found that the symptoms of AOM (mainly otalgia) spontaneously resolved in two thirds of children by 24 hours and in 80% at 2 to 7 days. This was also observed in 2 earlier meta-analyses.^10,19 Seventeen children would need to be treated with antibiotics (vs placebo) for 1 child to have less pain at 2 to 7 days (number needed to treat [NNT]=17). There were no differences between antibiotic and placebo groups in other clinical outcomes, such as tympanometry findings, perforation, and recurrences. Also, children treated with antibiotics were almost twice as likely to have vomiting, diarrhea, or a rash.

Initially not treating uncomplicated AOM with antibiotics is an acceptable alternative. In a study by van Buchem and colleagues,²⁰ 90% of children with AOM recovered (symptoms resolved) in the first 4 days with nose drops and oral analgesics and without the use of antibiotics. Only 3% of the 4860 children in this study had a clinical course that required further treatment with antibiotics or myringotomy. A recent randomized controlled trial of 315 children demonstrated that children treated immediately with antibiotics had 1 less day of symptoms, but that 1 in 5 of these had diarrhea.²¹ The group not treated with antibiotics had no serious sequellae and used more analgesics. There were no differences in the number of missed school days, and more than 75% of the parents were satisfied with this “wait and see” approach. These studies also emphasize that antibiotics have a very modest effect on the clinical course of AOM and seem to decrease the duration of symptoms only to a small degree.

Physicians often recommend other symptomatic treatments for ear infections. Non–aspirin analgesics are effective in relieving pain,²¹ as are ibuprofen²² and Auralgan.²³ Antihistamine-decongestant preparations offer no added benefit in resolution of symptoms and have no effect on clinical outcomes when given with antibiotics.^24,25

If antibiotics are used, the meta-analysis by Rosenfeld and colleagues¹⁰ showed there were no differences in outcomes between treatment with cefaclor, cefixime, erythromycin, trimethoprim-sulfamethoxazole, amoxicillin clavulanate, or erythromycin sulfisoxazole, and treatment with amoxicillin or ampicillin. Broader spectrum and/or more expensive antibiotics therefore offer no advantage over amoxicillin for the initial treatment of AOM. In another meta-analysis, Kozyrskyj and coworkers26 showed that a 5-day antibiotic course was an acceptable alternative to a 8- to 14-day treatment course. There was a slightly increased risk of treatment failure at 1-month follow-up with the shortened course of antibiotics; the NNT to prevent 1 excess failure at 30 days was 17. There were no differences in long-term outcomes (2-3 months) or medication side effects (vomiting, diarrhea, rash) between the short and long antibiotic courses. The broader spectrum azithromycin and intramuscular ceftriaxone offered no advantage over amoxicillin. Children aged younger than 2 years deserve special mention, since they are at higher risk for treatment failures,^27,28 persistent symptoms,²⁹ and recurrent otitis media.³⁰ Few well-designed studies exist to guide treatment in this age group. Although children aged younger than 2 years were not excluded, neither review by Glasziou and colleagues¹⁸ or Rosenfeld and coworkers¹⁰ specifically examined this age group.

Another review demonstrated that (as with older children) routinely using antibiotics initially does not seem to add any clinical benefit.³¹ A recent randomized controlled trial of 240 children demonstrated that 8 children in this age group would have to be treated with amoxicillin for 1 child to have fewer symptoms (fever, crying, irritability) at 4 days (NNT=8).³² The major benefit of amoxicillin in this study was 1 day less of fever (P=.004). Adverse effects were almost twice as likely in the amoxicillin group, although this difference was not statistically significant. There were also no differences between the groups in clinical failure rates at 11 days or in the likelihood of recurrent otitis media, antibiotic use, specialist referrals, or surgery at 6 weeks. Effects on hearing were not measured. The authors conclude that “this modest effect does not justify prescription of antibiotics at the first visit, provided close surveillance can be guaranteed.” Also, Kozyrskyj and colleagues²⁶ demonstrated in their meta-analysis that as a subgroup there were no differences in clinical failures between 5 and 10 days of antibiotics in children aged younger than 2 years. However, there were only 118 children in this age group. Table 2 shows treatment options for AOM.

In recent years, there has been an increasing concern about worldwide bacterial resistance to antimicrobial drugs^33-35 by the World Health Organization³⁶ and the Centers for Disease Control and Prevention (CDC).³⁷ In response to the increasing antimicrobial resistance patterns seen in the common middle ear pathogens, especially S pneumoniae, the CDC Drug-resistant Streptococcus pneumoniae Therapeutic Working Group recommends doubling the dosage of amoxicillin to 80 to 90 mg per kg per day in the empiric treatment of AOM.³⁸ These recommendations are based on in vitro mean inhibitory concentration data of S pneumoniae cultures from middle ear fluid and nasopharyngeal swabs. However, there currently is no patient-oriented evidence to suggest that increasing the amoxicillin dosage actually decreases suppurative or invasive complications of AOM (meningitis, mastoiditis, and so forth), affects recurrence rates or treatment success, affects long-term outcomes of AOM, or even decreases the rate of drug-resistant S pneumoniae. Also, bacteriologic outcomes do not correlate with clinical outcomes. In the meta-analysis by Rosenfeld and coworkers,¹⁰ 89% of middle ear pathogens from treatment failures were susceptible in vitro to the antibiotic prescribed, and 13% of isolates from clinical cures were resistant in vitro to the prescribed antibiotic. We must use extreme caution in extrapolating the microbiologic findings to the clinical care of the child with AOM.

Prognosis

There are insufficient data to suggest that routine antibiotic use in AOM results in fewer cases of mastoiditis or meningitis. In the systematic reviews cited, the incidence of these suppurative complications was rare. In the Cochrane review by Glasziou and colleagues,¹⁸ only 1 case of mastoiditis developed in 2202 children, and this was in a child treated with penicillin. In the Netherlands among 4860 consecutive children with AOM, 2 experienced mastoiditis (both responded to outpatient antibiotic therapy), and there were no cases of meningitis.²⁰ In the meta-analysis by Rosenfeld and coworkers,¹⁰ there were no suppurative complications in the 5400 children studied. Although mastoiditis has been quoted as being more common in the preantibiotic era, it is unclear if the current rarity of this condition is due to antibiotic treatment, changes in organism virulence or host defenses, or the assertion that uncomplicated otitis media was often not reported, thus increasing the relative rate of mastoiditis.²⁷ In a recent review,³⁹ antibiotics did not seem to have an appreciable effect on complication rates, leading the authors to conclude that “antibiotic treatment for AOM cannot be considered as a safeguard against the development of complications.” Even in developing countries where the burden of otitis media is great, mastoiditis is quite rare, with a prevalence rate of much less than 1%.⁴⁰

Antibiotic use does influence bacterial resistance rates. In children previously treated with antibiotics for AOM, there is a 3-fold increased risk of isolating drug-resistant organisms from middle ear effusions with subsequent bouts of otitis media.^41,42 In the Netherlands and Iceland, routinely not treating AOM with antibiotics has resulted in a reduction in antibiotic resistance.^33,43 The most important risk factors for a poor outcome are age younger than 2 years and attendance at a daycare center.40 Children in daycare have a higher risk of requiring a hospital admission and up to a 50% increased risk of repeated or recurrent ear infections.^44,45 Children with chronic underlying illnesses or chronic otitis media with effusion have added risks of poor outcomes that are beyond the scope of this review.

Conclusions

Thus, the natural course of AOM in children is quite favorable. If left untreated, 80% will recover spontaneously within 2 weeks. The addition of antibiotics provides at best a modest reduction in symptoms, while adding cost, adverse drug effects, and increasing bacterial resistance in the patient and community. Minimizing the use of antibiotics in patients with AOM does not increase the risks of perforation, deafness, or contralateral or recurrent AOM.

BACKGROUND: Strong evidence exists that the greater the blood pressure reduction in patients with type 2 diabetes, the greater the clinical benefit in reducing significant cardiovascular events and death.1 It is not known whether these benefits are actually due to the lowering of blood pressure or if there is a drug class effect, particularly among ACEIs.

POPULATION STUDIED: The average age of participants in the 4 studies was 58 years, and 60% were men. The study settings (primary care or referral) were not reported. Baseline blood pressures ranged from 155 to 170 systolic and 93 to 98 diastolic. The duration of diabetes was 8.5 to 10.7 years in 2 studies, 2.7 years in 1 study, and was not reported in 1 study.

STUDY DESIGN AND VALIDITY: This was a meta-analysis. The authors performed a MEDLINE search for studies pertaining to ACEIs, diabetes, and hypertension but did not search any other databases such as the Cochrane Controlled Trials Register. Studies were included if they were randomized controlled trials of patients with type 2 diabetes that evaluated an ACEI versus another antihypertensive medication with a minimum follow-up of 2 years. There was no indication whether retrieved studies were reviewed independently by 2 or more authors for inclusion. Tests for heterogeneity were appropriately applied to assure that only data from similar studies were combined. A fixed effects model, which assumes homogeneity, was used to combine data.

OUTCOMES MEASURED: The outcomes measured were acute myocardial infarction (AMI), stroke, combined cardiovascular events, and all-cause mortality. The combined cardiovascular events varied depending on the particular trial and could include cardiovascular death, fatal and nonfatal MI, fatal and nonfatal stroke, angina, congestive heart failure, and pulmonary infarction.

RESULTS: The initial search identified 195 articles of which 4 studies with a total of 2180 patients met the inclusion criteria. After applying tests for heterogeneity, only data from 3 trials (which studied enalopril vs nisoldipine, captopril vs diuretic/b-blocker, and fosinopril vs amlodipine) were used in the meta-analysis. In one of these trials (captopril vs diuretic/b-blocker), the number of events in persons with diabetes was not reported; it was therefore estimated by the authors of this review. This is a limitation of this meta-analysis. The relative risk calculations for these 3 studies showed that using ACEIs versus other antihypertensives lowered the number of AMIs (summary risk ratio [RR]=0.37; 95% confidence interval [CI], 0.24-0.57), had no affect on the incidence of stroke (RR=0.76; 95% CI, 0.48-1.22), lowered the number of combined cardiovascular events (RR=0.49; 95% CI, 0.36-0.67), and lowered overall cardiovascular mortality (RR=0.57; 95% CI, 0.38-0.87). On the basis of these data, to prevent an additional AMI, cardiovascular event, or death one would have to treat 17, 13, and 40 patients respectively with an ACEI instead of another agent for approximately 4.5 years. The study excluded from the meta-analysis compared outcomes in patients taking an ACEI versus atenolol and found no statistically significant differences between the groups for the 4 clinical outcomes.¹

BACKGROUND: Strong evidence exists that the greater the blood pressure reduction in patients with type 2 diabetes, the greater the clinical benefit in reducing significant cardiovascular events and death.1 It is not known whether these benefits are actually due to the lowering of blood pressure or if there is a drug class effect, particularly among ACEIs.

POPULATION STUDIED: The average age of participants in the 4 studies was 58 years, and 60% were men. The study settings (primary care or referral) were not reported. Baseline blood pressures ranged from 155 to 170 systolic and 93 to 98 diastolic. The duration of diabetes was 8.5 to 10.7 years in 2 studies, 2.7 years in 1 study, and was not reported in 1 study.

STUDY DESIGN AND VALIDITY: This was a meta-analysis. The authors performed a MEDLINE search for studies pertaining to ACEIs, diabetes, and hypertension but did not search any other databases such as the Cochrane Controlled Trials Register. Studies were included if they were randomized controlled trials of patients with type 2 diabetes that evaluated an ACEI versus another antihypertensive medication with a minimum follow-up of 2 years. There was no indication whether retrieved studies were reviewed independently by 2 or more authors for inclusion. Tests for heterogeneity were appropriately applied to assure that only data from similar studies were combined. A fixed effects model, which assumes homogeneity, was used to combine data.

OUTCOMES MEASURED: The outcomes measured were acute myocardial infarction (AMI), stroke, combined cardiovascular events, and all-cause mortality. The combined cardiovascular events varied depending on the particular trial and could include cardiovascular death, fatal and nonfatal MI, fatal and nonfatal stroke, angina, congestive heart failure, and pulmonary infarction.

RESULTS: The initial search identified 195 articles of which 4 studies with a total of 2180 patients met the inclusion criteria. After applying tests for heterogeneity, only data from 3 trials (which studied enalopril vs nisoldipine, captopril vs diuretic/b-blocker, and fosinopril vs amlodipine) were used in the meta-analysis. In one of these trials (captopril vs diuretic/b-blocker), the number of events in persons with diabetes was not reported; it was therefore estimated by the authors of this review. This is a limitation of this meta-analysis. The relative risk calculations for these 3 studies showed that using ACEIs versus other antihypertensives lowered the number of AMIs (summary risk ratio [RR]=0.37; 95% confidence interval [CI], 0.24-0.57), had no affect on the incidence of stroke (RR=0.76; 95% CI, 0.48-1.22), lowered the number of combined cardiovascular events (RR=0.49; 95% CI, 0.36-0.67), and lowered overall cardiovascular mortality (RR=0.57; 95% CI, 0.38-0.87). On the basis of these data, to prevent an additional AMI, cardiovascular event, or death one would have to treat 17, 13, and 40 patients respectively with an ACEI instead of another agent for approximately 4.5 years. The study excluded from the meta-analysis compared outcomes in patients taking an ACEI versus atenolol and found no statistically significant differences between the groups for the 4 clinical outcomes.¹

BACKGROUND: Strong evidence exists that the greater the blood pressure reduction in patients with type 2 diabetes, the greater the clinical benefit in reducing significant cardiovascular events and death.1 It is not known whether these benefits are actually due to the lowering of blood pressure or if there is a drug class effect, particularly among ACEIs.

POPULATION STUDIED: The average age of participants in the 4 studies was 58 years, and 60% were men. The study settings (primary care or referral) were not reported. Baseline blood pressures ranged from 155 to 170 systolic and 93 to 98 diastolic. The duration of diabetes was 8.5 to 10.7 years in 2 studies, 2.7 years in 1 study, and was not reported in 1 study.

STUDY DESIGN AND VALIDITY: This was a meta-analysis. The authors performed a MEDLINE search for studies pertaining to ACEIs, diabetes, and hypertension but did not search any other databases such as the Cochrane Controlled Trials Register. Studies were included if they were randomized controlled trials of patients with type 2 diabetes that evaluated an ACEI versus another antihypertensive medication with a minimum follow-up of 2 years. There was no indication whether retrieved studies were reviewed independently by 2 or more authors for inclusion. Tests for heterogeneity were appropriately applied to assure that only data from similar studies were combined. A fixed effects model, which assumes homogeneity, was used to combine data.

OUTCOMES MEASURED: The outcomes measured were acute myocardial infarction (AMI), stroke, combined cardiovascular events, and all-cause mortality. The combined cardiovascular events varied depending on the particular trial and could include cardiovascular death, fatal and nonfatal MI, fatal and nonfatal stroke, angina, congestive heart failure, and pulmonary infarction.

RESULTS: The initial search identified 195 articles of which 4 studies with a total of 2180 patients met the inclusion criteria. After applying tests for heterogeneity, only data from 3 trials (which studied enalopril vs nisoldipine, captopril vs diuretic/b-blocker, and fosinopril vs amlodipine) were used in the meta-analysis. In one of these trials (captopril vs diuretic/b-blocker), the number of events in persons with diabetes was not reported; it was therefore estimated by the authors of this review. This is a limitation of this meta-analysis. The relative risk calculations for these 3 studies showed that using ACEIs versus other antihypertensives lowered the number of AMIs (summary risk ratio [RR]=0.37; 95% confidence interval [CI], 0.24-0.57), had no affect on the incidence of stroke (RR=0.76; 95% CI, 0.48-1.22), lowered the number of combined cardiovascular events (RR=0.49; 95% CI, 0.36-0.67), and lowered overall cardiovascular mortality (RR=0.57; 95% CI, 0.38-0.87). On the basis of these data, to prevent an additional AMI, cardiovascular event, or death one would have to treat 17, 13, and 40 patients respectively with an ACEI instead of another agent for approximately 4.5 years. The study excluded from the meta-analysis compared outcomes in patients taking an ACEI versus atenolol and found no statistically significant differences between the groups for the 4 clinical outcomes.¹