Applied Evidence

In the long-term care of patients with acute coronary syndrome, recently published w that prognostic benefits improve with more aggressive antiplatelet therapy for those at high risk for recurrent events. Moreover, long-term care should include aggressive LDL cholesterol-lowering therapy and use of beta-blockers and angiotensin-converting enzyme (ACE) inhibitors, in addition to diet modification and exercise.

STEMI and NSTEMI: The new nomenclature

Coronary artery disease, the leading cause of death in the United States,¹ can manifest in many ways involving a constellation of symptoms, electrocardiogram changes, and serum markers. These acute coronary syndromes result from decreased coronary blood flow and cause chest discomfort, usually at rest, with or without characteristic radiation, or such comparable anginal equivalents as weakness, dyspnea, and diaphoresis.

STEMI. An elevated ST-segment with elevated levels of such cardiac markers as creatine kinase myocardial band or troponin I or troponin T are consistent with a diagnosis of ST-elevation myocardial infarction (STEMI). The old term, acute myocardial infarction, was defined by the presence of pathological Q waves (Q wave MI).

NSTEMI. Patients with elevated serum levels of creatine kinase myocardial band or troponin I or troponin T, but no ST-segment elevation, are said to have non-ST-elevation myocardial infarction (NSTEMI).

Unstable angina. Normal levels of serum cardiac markers and an absence of ST-elevation are consistent with a diagnosis of unstable angina (Figure). Of patients with STEMI, most will ultimately experience a Q-wave MI; a minority will have a non–Q-wave MI. Of patients with NSTEMI, most will sustain a non–Q-wave MI, while a minority will sustain a Q-wave MI.

Unstable angina and NSTEMI are urgent and life-threatening problems. Chest pain and related symptoms account for 5.3 million visits to US emergency departments per year² and account for 1.4 million hospitalizations annually.³ Approximately 15% of those presenting with unstable angina and NSTEMI go on to (re)infarct or die within 30 days.⁴

FIGURE
Unstable angina and NSTEMI

FIGURE
Nomenclature of acute coronary syndrome

ACC/AHA Guidelines

In 2000 the American College of Cardiology and the American Heart Association (ACC/AHA) Task Force on Practice Guidelines published their evidence-based recommendations for treatment of unstable angina/NSTEMI following an exhaustive review of the literature.

New trials in acute coronary syndromes

Since 2000, knowledge of acute coronary syndromes advanced considerably with results of large pivotal randomized controlled studies, which necessitated updating the ACC/AHA guidelines just 21 months after their completion. The most notable of these studies were the Clopidogrel in Unstable Angina to Prevent Recurrent Ischemic Events (CURE) trial,⁶ the Percutaneous Coronary Intervention (PCI)-CURE⁷ subset analysis, and the Myocardial Ischemia Reduction with Aggressive Cholesterol Lowering (MIRACL) trial.⁸

Although incorporation of antiplatelet therapy is the focus of this article, the ultimate objective of the new revised guidelines is improved clinical outcomes for patients with acute coronary syndromes by improving early risk assessment, using revascularization procedures aggressively when risk for future cardiac events is high, using short- and long-term antiplatelet and antithrombotic agents, and modifying long-term risk.

2002 revised guidelines

The studies reviewed in the 2002 revised guidelines assessed multiple therapies in the reduction of recurrent MI, stroke, and other cardiovascular events following patients’ presentation with unstable angina and NSTEMI. The studies also evaluated traditional therapies for acute coronary syndromes, such as unfractionated heparin, beta-blockers, and aspirin, as well as more recent therapies, including low-molecular-weight heparin, antiplatelet therapy (parenteral glycoprotein IIb/IIIa [GP IIb/IIIa] antagonists and ADP-receptor antagonists [thienopyridines]), lipid-lowering therapy (HMG-CoA reductase inhibitors, or statins), and antihypertensive agents.²

The results of these trials have lead to changes in the initial management of patients following an new ischemic event, as well as the choice of medical therapy begun in the hospital and continued following discharge. Patients are given an individualized medical regimen based on specific needs that include in-hospital findings relating to the type of recent procedure, risk factors for subsequent ischemic events and drug tolerability. Such a regimen, although beyond the scope of this discussion, can be recalled as “ABCDE,” for Anti-platelet agents/ACE inhibitors, Beta-blockers, Blood pressure control, Cholesterol (lipid)-lowering agents, Cigarette cessation, Diet modification, Diabetes control, Exercise, and Education.

Important changes in long-term management

For most primary care physicians—with the possible exception of those in rural areas, who are faced with the burden of emergent care—the most important aspect of the revised guidelines is the changes in recommendations for long-term management of patients with unstable angina and NSTEMI. The goal of these new recommendations is to reduce the risk of subsequent cardiovascular events, such as death, recurrent MI, congestive heart failure, and stroke. Based on the results of the groundbreaking studies reviewed by the ACC/AHA panel, major changes to the original guidelines were necessary for the areas of long-term antiplatelet therapy and risk reduction.

TABLE
ACC/AHA task force classifications on patient evaluation and therapy

Antiplatelet therapy

Antiplatelet therapy—aspirin, GP IIb/IIIa antagonists, or an ADP-receptor antagonist (eg, clopidogrel)—is critical in the treatment and management of patients with unstable angina/NSTEMI.² Based on the ACC/AHA Task Force classification (Table), aspirin (81–325 mg) should be initiated as quickly as possible after the condition is recognized, and continued indefinitely (SOR: A).

Clopidogrel (300-mg loading dose followed by 75 mg/d) should be administered to patients unable to take aspirin (SOR: A). In addition, clopidogrel 75 mg (once daily) should be added to aspirin therapy as quickly as possible and continued for up to 9 months in patients for whom an early noninterventional approach is planned (SOR: A), or for patients with a planned PCI who are not at high risk of bleeding (SOR: B).²

The CURE trial: clopidogrel and aspirin

The principal studies underlying the new recommendations for long-term antiplatelet therapy are the CURE trial⁶ and the PCI-CURE⁷ subset analysis. In the CURE trial, 12,562 patients who presented with unstable angina/NSTEMI within 24 hours following the onset of symptoms were randomized to receive clopidogrel (loading dose of 300 mg followed by 75 mg/d) with aspirin or placebo with aspirin for 3 to 12 months (mean follow-up, 9 months).

The relative risk (RR) of the primary composite outcome including incidence of cardiovascular death, nonfatal MI, or stroke was lower by 20% (RR=0.80; 95% confidence interval [CI], 0.72–0.90; P<.001) in the clopidogrel arm. Similarly, a composite outcome also including revascularization was lower by 14% (RR=0.86; P<.001) for those who received clopidogrel. Benefits were seen in all risk groups. A significant increase in bleeding events was observed in the group that received clopidogrel plus aspirin compared with aspirin alone (major bleeding, P=.001; minor bleeding, P<.001).⁶

CURE patients were followed for up to 1 year, with a mean follow-up period of 9 months. In addition to the early benefits seen, from day 31 up to 1 year there was a highly significant incremental reduction of 18% in the primary outcome (P<.001) with clopidogrel.⁷

The PCI-CURE subset: Percutaneous coronary interventions

In the PCI-CURE trial, a subset of patients recruited for CURE who underwent PCI was pretreated with aspirin 325 mg and clopidogrel 75 mg for a median of 10 days. These patients received either clopidogrel or ticlopidine for 4 weeks, and then were restarted on either clopidogrel 75 mg (80%) or placebo in addition to aspirin for an additional mean of 8 months with up to 1 year of follow-up.

Long-term administration of clopidogrel 75 mg following PCI resulted in a 30% reduction (RR=0.70; 95% CI, 0.50–0.70; P=.03) in cardiovascular death, MI, and any revascularization, with a 31% reduction in cardiovascular death or MI (P=.002) compared with placebo. Major bleeding rates were similar between groups (P=.64).⁹

The CREDO study

The new guidelines for long-term antiplatelet therapy are further supported by the subsequently published Clopidogrel for the Reduction of Events During Observation (CREDO) study.¹⁰ CREDO demonstrated a 26.9% relative risk reduction in the combined risk of death, MI, or stroke (95% CI, 3.9%–44.4%; P=.02) in patients with long-term (12 months) aspirin plus clopidogrel 75 mg therapy compared with aspirin plus placebo in 2116 patients undergoing elective PCI or deemed at high likelihood of undergoing PCI. There was no significant increase in the risk of major bleeding (P=.07) between the placebo and clopidogrel arms.

Risk reduction

The revised guidelines also included updated recommendations for risk reduction. It is now recommended that a fibrate or niacin be administered if the HDL cholesterol level is <40 mg/dL (SOR: B).² Further, statins and a heart-healthy diet should be started during admission and continued after discharge for patients with LDL cholesterol >100 mg/dL (SOR: B).²

This recommendation is based in part on the MIRACL trial, in which 3086 acute coronary syndrome patients treated with atorvastatin, 24 to 96 hours after hospital admission, demonstrated a significant reduction in the composite rate of death, nonfatal MI, resuscitated cardiac arrest, or recurrent ischemia compared with those who received placebo (14.8% vs 17.4%) (RR=0.84; 95% CI, 0.70–1.00; P=.048). Patients were followed for up to 16 weeks after starting therapy. Abnormal liver transaminases (>3 times upper limit of normal) occurred more often in the atorvastatin group than the placebo group (2.5% vs 0.6%; P<.001).⁸

Correspondence
John S. Banas, MD, FACC, Morristown Memorial Hospital, 100 Madison Ave, Morristown, NJ 07960. E-mail: [email protected].

In the long-term care of patients with acute coronary syndrome, recently published w that prognostic benefits improve with more aggressive antiplatelet therapy for those at high risk for recurrent events. Moreover, long-term care should include aggressive LDL cholesterol-lowering therapy and use of beta-blockers and angiotensin-converting enzyme (ACE) inhibitors, in addition to diet modification and exercise.

STEMI and NSTEMI: The new nomenclature

Coronary artery disease, the leading cause of death in the United States,¹ can manifest in many ways involving a constellation of symptoms, electrocardiogram changes, and serum markers. These acute coronary syndromes result from decreased coronary blood flow and cause chest discomfort, usually at rest, with or without characteristic radiation, or such comparable anginal equivalents as weakness, dyspnea, and diaphoresis.

STEMI. An elevated ST-segment with elevated levels of such cardiac markers as creatine kinase myocardial band or troponin I or troponin T are consistent with a diagnosis of ST-elevation myocardial infarction (STEMI). The old term, acute myocardial infarction, was defined by the presence of pathological Q waves (Q wave MI).

NSTEMI. Patients with elevated serum levels of creatine kinase myocardial band or troponin I or troponin T, but no ST-segment elevation, are said to have non-ST-elevation myocardial infarction (NSTEMI).

Unstable angina. Normal levels of serum cardiac markers and an absence of ST-elevation are consistent with a diagnosis of unstable angina (Figure). Of patients with STEMI, most will ultimately experience a Q-wave MI; a minority will have a non–Q-wave MI. Of patients with NSTEMI, most will sustain a non–Q-wave MI, while a minority will sustain a Q-wave MI.

Unstable angina and NSTEMI are urgent and life-threatening problems. Chest pain and related symptoms account for 5.3 million visits to US emergency departments per year² and account for 1.4 million hospitalizations annually.³ Approximately 15% of those presenting with unstable angina and NSTEMI go on to (re)infarct or die within 30 days.⁴

FIGURE
Unstable angina and NSTEMI

FIGURE
Nomenclature of acute coronary syndrome

ACC/AHA Guidelines

In 2000 the American College of Cardiology and the American Heart Association (ACC/AHA) Task Force on Practice Guidelines published their evidence-based recommendations for treatment of unstable angina/NSTEMI following an exhaustive review of the literature.

New trials in acute coronary syndromes

Since 2000, knowledge of acute coronary syndromes advanced considerably with results of large pivotal randomized controlled studies, which necessitated updating the ACC/AHA guidelines just 21 months after their completion. The most notable of these studies were the Clopidogrel in Unstable Angina to Prevent Recurrent Ischemic Events (CURE) trial,⁶ the Percutaneous Coronary Intervention (PCI)-CURE⁷ subset analysis, and the Myocardial Ischemia Reduction with Aggressive Cholesterol Lowering (MIRACL) trial.⁸

Although incorporation of antiplatelet therapy is the focus of this article, the ultimate objective of the new revised guidelines is improved clinical outcomes for patients with acute coronary syndromes by improving early risk assessment, using revascularization procedures aggressively when risk for future cardiac events is high, using short- and long-term antiplatelet and antithrombotic agents, and modifying long-term risk.

2002 revised guidelines

The studies reviewed in the 2002 revised guidelines assessed multiple therapies in the reduction of recurrent MI, stroke, and other cardiovascular events following patients’ presentation with unstable angina and NSTEMI. The studies also evaluated traditional therapies for acute coronary syndromes, such as unfractionated heparin, beta-blockers, and aspirin, as well as more recent therapies, including low-molecular-weight heparin, antiplatelet therapy (parenteral glycoprotein IIb/IIIa [GP IIb/IIIa] antagonists and ADP-receptor antagonists [thienopyridines]), lipid-lowering therapy (HMG-CoA reductase inhibitors, or statins), and antihypertensive agents.²

The results of these trials have lead to changes in the initial management of patients following an new ischemic event, as well as the choice of medical therapy begun in the hospital and continued following discharge. Patients are given an individualized medical regimen based on specific needs that include in-hospital findings relating to the type of recent procedure, risk factors for subsequent ischemic events and drug tolerability. Such a regimen, although beyond the scope of this discussion, can be recalled as “ABCDE,” for Anti-platelet agents/ACE inhibitors, Beta-blockers, Blood pressure control, Cholesterol (lipid)-lowering agents, Cigarette cessation, Diet modification, Diabetes control, Exercise, and Education.

Important changes in long-term management

For most primary care physicians—with the possible exception of those in rural areas, who are faced with the burden of emergent care—the most important aspect of the revised guidelines is the changes in recommendations for long-term management of patients with unstable angina and NSTEMI. The goal of these new recommendations is to reduce the risk of subsequent cardiovascular events, such as death, recurrent MI, congestive heart failure, and stroke. Based on the results of the groundbreaking studies reviewed by the ACC/AHA panel, major changes to the original guidelines were necessary for the areas of long-term antiplatelet therapy and risk reduction.

TABLE
ACC/AHA task force classifications on patient evaluation and therapy

Antiplatelet therapy

Antiplatelet therapy—aspirin, GP IIb/IIIa antagonists, or an ADP-receptor antagonist (eg, clopidogrel)—is critical in the treatment and management of patients with unstable angina/NSTEMI.² Based on the ACC/AHA Task Force classification (Table), aspirin (81–325 mg) should be initiated as quickly as possible after the condition is recognized, and continued indefinitely (SOR: A).

Clopidogrel (300-mg loading dose followed by 75 mg/d) should be administered to patients unable to take aspirin (SOR: A). In addition, clopidogrel 75 mg (once daily) should be added to aspirin therapy as quickly as possible and continued for up to 9 months in patients for whom an early noninterventional approach is planned (SOR: A), or for patients with a planned PCI who are not at high risk of bleeding (SOR: B).²

The CURE trial: clopidogrel and aspirin

The principal studies underlying the new recommendations for long-term antiplatelet therapy are the CURE trial⁶ and the PCI-CURE⁷ subset analysis. In the CURE trial, 12,562 patients who presented with unstable angina/NSTEMI within 24 hours following the onset of symptoms were randomized to receive clopidogrel (loading dose of 300 mg followed by 75 mg/d) with aspirin or placebo with aspirin for 3 to 12 months (mean follow-up, 9 months).

The relative risk (RR) of the primary composite outcome including incidence of cardiovascular death, nonfatal MI, or stroke was lower by 20% (RR=0.80; 95% confidence interval [CI], 0.72–0.90; P<.001) in the clopidogrel arm. Similarly, a composite outcome also including revascularization was lower by 14% (RR=0.86; P<.001) for those who received clopidogrel. Benefits were seen in all risk groups. A significant increase in bleeding events was observed in the group that received clopidogrel plus aspirin compared with aspirin alone (major bleeding, P=.001; minor bleeding, P<.001).⁶

CURE patients were followed for up to 1 year, with a mean follow-up period of 9 months. In addition to the early benefits seen, from day 31 up to 1 year there was a highly significant incremental reduction of 18% in the primary outcome (P<.001) with clopidogrel.⁷

The PCI-CURE subset: Percutaneous coronary interventions

In the PCI-CURE trial, a subset of patients recruited for CURE who underwent PCI was pretreated with aspirin 325 mg and clopidogrel 75 mg for a median of 10 days. These patients received either clopidogrel or ticlopidine for 4 weeks, and then were restarted on either clopidogrel 75 mg (80%) or placebo in addition to aspirin for an additional mean of 8 months with up to 1 year of follow-up.

Long-term administration of clopidogrel 75 mg following PCI resulted in a 30% reduction (RR=0.70; 95% CI, 0.50–0.70; P=.03) in cardiovascular death, MI, and any revascularization, with a 31% reduction in cardiovascular death or MI (P=.002) compared with placebo. Major bleeding rates were similar between groups (P=.64).⁹

The CREDO study

The new guidelines for long-term antiplatelet therapy are further supported by the subsequently published Clopidogrel for the Reduction of Events During Observation (CREDO) study.¹⁰ CREDO demonstrated a 26.9% relative risk reduction in the combined risk of death, MI, or stroke (95% CI, 3.9%–44.4%; P=.02) in patients with long-term (12 months) aspirin plus clopidogrel 75 mg therapy compared with aspirin plus placebo in 2116 patients undergoing elective PCI or deemed at high likelihood of undergoing PCI. There was no significant increase in the risk of major bleeding (P=.07) between the placebo and clopidogrel arms.

Risk reduction

The revised guidelines also included updated recommendations for risk reduction. It is now recommended that a fibrate or niacin be administered if the HDL cholesterol level is <40 mg/dL (SOR: B).² Further, statins and a heart-healthy diet should be started during admission and continued after discharge for patients with LDL cholesterol >100 mg/dL (SOR: B).²

This recommendation is based in part on the MIRACL trial, in which 3086 acute coronary syndrome patients treated with atorvastatin, 24 to 96 hours after hospital admission, demonstrated a significant reduction in the composite rate of death, nonfatal MI, resuscitated cardiac arrest, or recurrent ischemia compared with those who received placebo (14.8% vs 17.4%) (RR=0.84; 95% CI, 0.70–1.00; P=.048). Patients were followed for up to 16 weeks after starting therapy. Abnormal liver transaminases (>3 times upper limit of normal) occurred more often in the atorvastatin group than the placebo group (2.5% vs 0.6%; P<.001).⁸

Correspondence
John S. Banas, MD, FACC, Morristown Memorial Hospital, 100 Madison Ave, Morristown, NJ 07960. E-mail: [email protected].

In the long-term care of patients with acute coronary syndrome, recently published w that prognostic benefits improve with more aggressive antiplatelet therapy for those at high risk for recurrent events. Moreover, long-term care should include aggressive LDL cholesterol-lowering therapy and use of beta-blockers and angiotensin-converting enzyme (ACE) inhibitors, in addition to diet modification and exercise.

STEMI and NSTEMI: The new nomenclature

Coronary artery disease, the leading cause of death in the United States,¹ can manifest in many ways involving a constellation of symptoms, electrocardiogram changes, and serum markers. These acute coronary syndromes result from decreased coronary blood flow and cause chest discomfort, usually at rest, with or without characteristic radiation, or such comparable anginal equivalents as weakness, dyspnea, and diaphoresis.

STEMI. An elevated ST-segment with elevated levels of such cardiac markers as creatine kinase myocardial band or troponin I or troponin T are consistent with a diagnosis of ST-elevation myocardial infarction (STEMI). The old term, acute myocardial infarction, was defined by the presence of pathological Q waves (Q wave MI).

NSTEMI. Patients with elevated serum levels of creatine kinase myocardial band or troponin I or troponin T, but no ST-segment elevation, are said to have non-ST-elevation myocardial infarction (NSTEMI).

Unstable angina. Normal levels of serum cardiac markers and an absence of ST-elevation are consistent with a diagnosis of unstable angina (Figure). Of patients with STEMI, most will ultimately experience a Q-wave MI; a minority will have a non–Q-wave MI. Of patients with NSTEMI, most will sustain a non–Q-wave MI, while a minority will sustain a Q-wave MI.

Unstable angina and NSTEMI are urgent and life-threatening problems. Chest pain and related symptoms account for 5.3 million visits to US emergency departments per year² and account for 1.4 million hospitalizations annually.³ Approximately 15% of those presenting with unstable angina and NSTEMI go on to (re)infarct or die within 30 days.⁴

FIGURE
Unstable angina and NSTEMI

FIGURE
Nomenclature of acute coronary syndrome

ACC/AHA Guidelines

In 2000 the American College of Cardiology and the American Heart Association (ACC/AHA) Task Force on Practice Guidelines published their evidence-based recommendations for treatment of unstable angina/NSTEMI following an exhaustive review of the literature.

New trials in acute coronary syndromes

Since 2000, knowledge of acute coronary syndromes advanced considerably with results of large pivotal randomized controlled studies, which necessitated updating the ACC/AHA guidelines just 21 months after their completion. The most notable of these studies were the Clopidogrel in Unstable Angina to Prevent Recurrent Ischemic Events (CURE) trial,⁶ the Percutaneous Coronary Intervention (PCI)-CURE⁷ subset analysis, and the Myocardial Ischemia Reduction with Aggressive Cholesterol Lowering (MIRACL) trial.⁸

Although incorporation of antiplatelet therapy is the focus of this article, the ultimate objective of the new revised guidelines is improved clinical outcomes for patients with acute coronary syndromes by improving early risk assessment, using revascularization procedures aggressively when risk for future cardiac events is high, using short- and long-term antiplatelet and antithrombotic agents, and modifying long-term risk.

2002 revised guidelines

The studies reviewed in the 2002 revised guidelines assessed multiple therapies in the reduction of recurrent MI, stroke, and other cardiovascular events following patients’ presentation with unstable angina and NSTEMI. The studies also evaluated traditional therapies for acute coronary syndromes, such as unfractionated heparin, beta-blockers, and aspirin, as well as more recent therapies, including low-molecular-weight heparin, antiplatelet therapy (parenteral glycoprotein IIb/IIIa [GP IIb/IIIa] antagonists and ADP-receptor antagonists [thienopyridines]), lipid-lowering therapy (HMG-CoA reductase inhibitors, or statins), and antihypertensive agents.²

The results of these trials have lead to changes in the initial management of patients following an new ischemic event, as well as the choice of medical therapy begun in the hospital and continued following discharge. Patients are given an individualized medical regimen based on specific needs that include in-hospital findings relating to the type of recent procedure, risk factors for subsequent ischemic events and drug tolerability. Such a regimen, although beyond the scope of this discussion, can be recalled as “ABCDE,” for Anti-platelet agents/ACE inhibitors, Beta-blockers, Blood pressure control, Cholesterol (lipid)-lowering agents, Cigarette cessation, Diet modification, Diabetes control, Exercise, and Education.

Important changes in long-term management

For most primary care physicians—with the possible exception of those in rural areas, who are faced with the burden of emergent care—the most important aspect of the revised guidelines is the changes in recommendations for long-term management of patients with unstable angina and NSTEMI. The goal of these new recommendations is to reduce the risk of subsequent cardiovascular events, such as death, recurrent MI, congestive heart failure, and stroke. Based on the results of the groundbreaking studies reviewed by the ACC/AHA panel, major changes to the original guidelines were necessary for the areas of long-term antiplatelet therapy and risk reduction.

TABLE
ACC/AHA task force classifications on patient evaluation and therapy

Antiplatelet therapy

Antiplatelet therapy—aspirin, GP IIb/IIIa antagonists, or an ADP-receptor antagonist (eg, clopidogrel)—is critical in the treatment and management of patients with unstable angina/NSTEMI.² Based on the ACC/AHA Task Force classification (Table), aspirin (81–325 mg) should be initiated as quickly as possible after the condition is recognized, and continued indefinitely (SOR: A).

Clopidogrel (300-mg loading dose followed by 75 mg/d) should be administered to patients unable to take aspirin (SOR: A). In addition, clopidogrel 75 mg (once daily) should be added to aspirin therapy as quickly as possible and continued for up to 9 months in patients for whom an early noninterventional approach is planned (SOR: A), or for patients with a planned PCI who are not at high risk of bleeding (SOR: B).²

The CURE trial: clopidogrel and aspirin

The principal studies underlying the new recommendations for long-term antiplatelet therapy are the CURE trial⁶ and the PCI-CURE⁷ subset analysis. In the CURE trial, 12,562 patients who presented with unstable angina/NSTEMI within 24 hours following the onset of symptoms were randomized to receive clopidogrel (loading dose of 300 mg followed by 75 mg/d) with aspirin or placebo with aspirin for 3 to 12 months (mean follow-up, 9 months).

The relative risk (RR) of the primary composite outcome including incidence of cardiovascular death, nonfatal MI, or stroke was lower by 20% (RR=0.80; 95% confidence interval [CI], 0.72–0.90; P<.001) in the clopidogrel arm. Similarly, a composite outcome also including revascularization was lower by 14% (RR=0.86; P<.001) for those who received clopidogrel. Benefits were seen in all risk groups. A significant increase in bleeding events was observed in the group that received clopidogrel plus aspirin compared with aspirin alone (major bleeding, P=.001; minor bleeding, P<.001).⁶

CURE patients were followed for up to 1 year, with a mean follow-up period of 9 months. In addition to the early benefits seen, from day 31 up to 1 year there was a highly significant incremental reduction of 18% in the primary outcome (P<.001) with clopidogrel.⁷

The PCI-CURE subset: Percutaneous coronary interventions

In the PCI-CURE trial, a subset of patients recruited for CURE who underwent PCI was pretreated with aspirin 325 mg and clopidogrel 75 mg for a median of 10 days. These patients received either clopidogrel or ticlopidine for 4 weeks, and then were restarted on either clopidogrel 75 mg (80%) or placebo in addition to aspirin for an additional mean of 8 months with up to 1 year of follow-up.

Long-term administration of clopidogrel 75 mg following PCI resulted in a 30% reduction (RR=0.70; 95% CI, 0.50–0.70; P=.03) in cardiovascular death, MI, and any revascularization, with a 31% reduction in cardiovascular death or MI (P=.002) compared with placebo. Major bleeding rates were similar between groups (P=.64).⁹

The CREDO study

The new guidelines for long-term antiplatelet therapy are further supported by the subsequently published Clopidogrel for the Reduction of Events During Observation (CREDO) study.¹⁰ CREDO demonstrated a 26.9% relative risk reduction in the combined risk of death, MI, or stroke (95% CI, 3.9%–44.4%; P=.02) in patients with long-term (12 months) aspirin plus clopidogrel 75 mg therapy compared with aspirin plus placebo in 2116 patients undergoing elective PCI or deemed at high likelihood of undergoing PCI. There was no significant increase in the risk of major bleeding (P=.07) between the placebo and clopidogrel arms.

Risk reduction

The revised guidelines also included updated recommendations for risk reduction. It is now recommended that a fibrate or niacin be administered if the HDL cholesterol level is <40 mg/dL (SOR: B).² Further, statins and a heart-healthy diet should be started during admission and continued after discharge for patients with LDL cholesterol >100 mg/dL (SOR: B).²

This recommendation is based in part on the MIRACL trial, in which 3086 acute coronary syndrome patients treated with atorvastatin, 24 to 96 hours after hospital admission, demonstrated a significant reduction in the composite rate of death, nonfatal MI, resuscitated cardiac arrest, or recurrent ischemia compared with those who received placebo (14.8% vs 17.4%) (RR=0.84; 95% CI, 0.70–1.00; P=.048). Patients were followed for up to 16 weeks after starting therapy. Abnormal liver transaminases (>3 times upper limit of normal) occurred more often in the atorvastatin group than the placebo group (2.5% vs 0.6%; P<.001).⁸

Correspondence
John S. Banas, MD, FACC, Morristown Memorial Hospital, 100 Madison Ave, Morristown, NJ 07960. E-mail: [email protected].

Glycemic control in diabetes begins with a patient’s adherence to several nonpharmacologic measures. Without such a commitment, success in controlling the disease will be difficult to achieve, and otherwise appropriate drug therapy will be hindered.

Most antidiabetic agents comparably reduce glycosylated hemoglobin (A_1c) levels. However, a particular agent may be preferred depending on a patient’s characteristics. And some circumstances call for combination therapy. This article reviews the advantages and disadvantages of the many pharmacologic treatments for glucose control and hyperglycemia in type 2 diabetes.

Benefits Of Diabetes Control

The benefits of diabetes control are detailed in this issue of THE JOURNAL OF FAMILY PRACTICE (“Strategies to reduce complications in type 2 diabetes,” pages 366–374). For every percentage-point reduction in hemoglobin A_1c, it is possible to achieve a 22% to 35% reduction in microvascular complications.^1,2 Cardiovascular disease can be reduced in patients with diabetes by treating hypertension^3,4 and hyperlipidemia, prescribing aspirin therapy, using angiotensin-converting enzyme (ACE) inhibitors, and with smoking cessation.^5,6

Targets For Glycemic Control

The American Diabetes Association’s (ADA) recommended targets for glycemic control are a preprandial blood glucose level of 80–120 mg/dL, a bedtime blood glucose level of 100–140 mg/dL, and a hemoglobin A_1c level of <7% (with a level of >8% requiring additional measures). Hemoglobin A_1c is the best determinant of glycemic exposure, and its mean value is a nationally recognized indicator of how well diabetes is being managed.⁷ The American College of Endocrinology has adopted a more aggressive approach by designating an A_1c level of 6.5% as both a target and action level.⁸

Self-monitoring of blood glucose

Self-monitoring of blood glucose (SMBG) is an integral component of diabetes therapy (strength of recommendation [SOR]: C) and should always be included in the management plan (SOR: C). The optimal frequency and timing of SMBG for type 2 diabetes is not known, but they should be sufficient to facilitate reaching glucose goals. The A_1c test should be performed at least semi-annually for patients with stable glycemic control, and quarterly for patients not meeting glycemic goals or those who are changing therapy. A_1c levels and mean plasma glucose levels can be approximately correlated (Table 1).⁷

TABLE 1
Correlation between hemoglobin A_1clevels and mean plasma glucose levels

Nonpharmacologic Therapy

Nonpharmacologic measures remain the cornerstone of managing type 2 diabetes. Hyperglycemia adversely and reversibly affects both insulin resistance and insulin secretion. Improvement in glycemic control can occur through dietary modification and regular exercise.

A recent meta-analysis of randomized controlled trials of diabetes patient education observed a net decrease in HbA_1c of 0.32% in intervention groups vs control.⁹ Interventions that included a face-to-face delivery, cognitive reframing teaching method, and exercise content were more likely to improve glycemic control.

Education

Lifestyle changes involving diet, exercise, and usually weight loss are key to effective management of diabetes. If patients are to change their behavior, they must be given detailed training.⁶ Self-management also necessitates that patients engage in problem solving. This requires that each aspect of the management plan is understood and agreed upon by the patient and providers, and that the goals and treatment plan are individualized and reasonable.

Diet: recommend soluble fiber, reduce calories

Medical nutrition therapy should be individualized and preferably provided by a registered dietitian familiar with diabetes (SOR: B). The goals of nutrition therapy, according to the ADA, are to attain recommended body weight and prevent or reverse obesity. The means of achieving these goals are nutrition assessment and modification of nutrient intake and lifestyle through healthy food choices and physical activity.⁷

A high intake of dietary fiber (particularly the soluble type) above the level recommended by the ADA improves glycemic control, decreases hyperinsulinemia, and lowers plasma lipid concentrations.¹⁰

Hypocaloric diets cause glucose plasma levels to fall, in some cases to a normal level with a weight loss of even 5 to 10 pounds.^7,11 Hypoglycemic medications are of course most effective in nonobese persons. But effectiveness is also improved if weight that is gained can be limited. Despite the clear benefit of weight loss, only a few patients are able to attain and maintain substantial weight loss. Maintenance of a reduced or elevated body weight is associated with compensatory changes in energy expenditure, which oppose the maintenance of body weight that is different from the usual weight.¹³ Part of the individualization of therapy is respect of personal and cultural preferences, lifestyles, and financial considerations.

Physical activity: a little goes a long way

A regular physical activity program is recommended for all patients with diabetes who are capable of participating (SOR: B).⁷ It improves blood glucose control, reduces cardiovascular risk factors, aids weight loss, and enhances well being.⁷ A recently published prospective cohort study showed that walking at least 2 hours a week was associated with a 39% lower all-cause mortality (hazard rate ratio [HRR], 0.61; 95% CI, 0.48–0.78) and a 34% lower cardiovascular mortality (HRR, 0.66; 95% CI, 0.45–0.96) across a diverse spectrum of adults with diabetes. The NNT (to prevent 1 death per year) is 61 for patients who walk at least 2 hours/week.¹⁴

In prescribing a physical activity plan for a patient, consider cardiovascular disease risk factors or complications to minimize the risk of untoward events. Micro- and macrovascular disease are of course prevalent among persons with diabetes, often resulting in functional limitations that make exercise more difficult.

Other priorities

Other recommended components of care include daily aspirin use, foot care exams, tobacco cessation, pneumococcal and influenza vaccinations, and an annual dilated retinal exam.

Pharmacologic Therapy

The coexisting defects in type 2 diabetes mellitus are as follows:

• resistance to insulin action in muscle
• defective pancreatic insulin secretion
• unrestrained hepatic glucose production, aggravated by increased lipolysis in adipose tissue.

Drug therapy is aimed at each of these defects, and also at reducing carbohydrate absorption in the small intestine (Figure 1). As far as antihyperglycemic effect is concerned, no one category of antidiabetic agent is favored over another.¹⁵ Except for nateglinide and α-glucosidase inhibitors (AGIs), each of the drug categories leads to a similar reduction in A_1c.¹⁶ However, patient characteristics may lead to selection of a particular agent. Table 2 summarizes oral treatment options, their relative advantages and costs.

FIGURE 1
Drug therapies for coexisting defects in type 2 diabetes

TABLE 2
Pharmacologic treatments for type 2 diabetes: monotherapies

Sulfonylureas

Sulfonylureas directly increase insulin secretion by binding to the sulfonylurea receptor on pancreatic beta cells; they provide a relatively quick onset of action. First-generation sulfonylureas (eg, tolbutamide, chlorpropamide) and second-generation sulfonylureas (eg, glyburide, glipizide, glimepiride) are equivalent in their maximum hypoglycemic effect.¹⁷

Second-generation agents are used more commonly than first-generation. They all contain the sulfonylurea moiety, but different chemical substitutions in the basic molecule change pharmacokinetics, resulting in different durations of action.¹⁷ Second-generation agents are probably safer than first-generation drugs, being less likely to cause hyponatremia, disulfiram-like reactions, or prolonged hypoglycemia.¹⁸

At maximal doses, sulfonylureas lower A_1c levels by 1–2 percentage points and fasting plasma glucose concentrations by 60–70 mg/dL;¹⁵ however, the glucose lowering effect typically plateaus after half the maximal recommended dose is reached. Sulfonylureas have no consistent effect on dyslipidemia. In UK Prospective Diabetes Study (UKPDS) 33, though improved glycemic control with sulfonylureas (or insulin) led to a 25% reduction in microvascular endpoints (mostly less retinal photocoagulation) (P<.01), sulfonylureas (or insulin) did not significantly reduce death or all-cause mortality compared with diet treatment.²

Adverse effects. The primary adverse effects of sulfonylureas are weight gain and hypoglycemia. In UKPDS 33, weight gain at 10 years was 2.6 kg (99% confidence interval [CI], 1.6–3.6) with chlorpropamide and 1.7 kg (99% CI, 0.7–2.7) with glyburide, compared with patients receiving diet therapy (each P<.001).² In the same study, the rate of major hypoglycemic episodes (third-party help or medical intervention necessary) while on therapy was 0.4%/year for chlorpropamide and 0.6%/year for glyburide, compared with 0.1%/year for diet.

Glyburide and chlorpropamide have active metabolites with renal elimination, and they should therefore be used with caution in patients with renal insufficiency. In 1971, the University Group Diabetes Project (UGDP) observed a twofold increase in the rate of cardiovascular death among patients receiving tolbutamide compared with those receiving insulin or placebo.¹⁸ This led to a decades long debate on the validity of this conclusion.¹⁹ More recently, UKPDS 33 did not demonstrate any increased cardiovascular mortality among patients receiving glyburide or chlorpropamide, and has largely negated this earlier concern.²

Cost. Sulfonylureas are the least expensive oral agents used to treat type 2 diabetes.

TABLE 3
Pharmacological treatments for type 2 diabetes: combination therapies

Non-sulfonylurea secretagogues

Like sulfonylureas, the non-sulfonylurea secretagogues (non-SU), repaglinide and nateglinide, stimulate beta cells to increase insulin secretion. However, the non-SU agents mediate their action through a different, adjacent site on the “sulfonylurea receptor.” Comparatively, the non-SU agents have a faster onset of action (20 minutes), shorter half-life (about 1.0–1.5 hours), and greater effects on postprandial glucose excursions than do sulfonylureas.²⁰ In contrast to the sulfonylureas, the extent of insulin release with non-SU agents is glucose dependent, and therefore they may have less risk of hypoglycemia several hours after meals.¹⁵

Repaglinide lowers the A_1c level by 1.7–1.9 percentage points, similar in efficacy to sulfonylureas. Nateglinide appears somewhat less efficacious and lowers A_1c by 0.6–1.0 percentage points.¹⁵ Nateglinide was significantly less effective than glyburide at lowering A_1c levels and the fasting plasma glucose in one 24-week study. Non-SUs added to sulfonylureas produce no additional benefit in glycemic control. The effect of non-SUs on microvascular or macrovascular endpoints is unknown.

Adverse effects. Hypoglycemia is the primary adverse effect of non-SUs. Confirmed hypoglycemia (plasma glucose <60 mg/dL) was observed in 2.4% of patients taking nateglinide compared with 0.4% of those receiving placebo. Mild or moderate hypoglycemia occurred in 16% of repaglinide patients, 20% of glyburide patients, and 19% of glipizide patients in one-year comparative studies. Further comparative studies are needed to determine if non-SUs produce significantly less hypoglycemia and weight gain than sulfonylureas.

Cost. Non-SUs must be dosed 3 times daily at the start of meals. One relative disadvantage is their increased cost compared with sulfonylureas.

Biguanides

The only biguanide marketed in the US is metformin. Its primary action is to inhibit hepatic glucose production and, to a much lesser extent, enhance insulin sensitivity in peripheral tissues.²¹ Metformin does not stimulate insulin secretion and does not cause hypoglycemia when used as monotherapy, but it can potentiate hypoglycemia in combination with insulin or insulin secretagogues.

Metformin is similar in efficacy to the sulfonylureas. It lowers A_1c by 1.5–2.0 percentage points and fasting plasma glucose by 60–80 mg/dL. Its antihyperglycemic efficacy is independent of patient age, duration of diabetes, or BMI.²²

In the UKPDS 34 study, a subgroup of obese patients was randomized to receive intensive control (group 1, metformin; group 2, a sulfonylurea or insulin) or conventional diet therapy (group 3). Despite a similar reduction in the A_1c level between the 2 intensive-treatment groups, patients treated with metformin had a 32% reduction for any diabetes-related endpoint (95% CI, 13–47; P=.002), 43% fewer diabetes-related deaths (95% CI, 9–63; P=.017), and a 36% reduction in all cause mortality, compared with the diet therapy group (95% CI, 9–55; P=.011).²³

Metformin also showed significant benefit when compared with patients receiving sulfonylurea or insulin (group 2). The absolute risk of any diabetes endpoint was 29.8 vs. 40.1 (events per 1000 patient-years; P=.0034), all-cause mortality (13.5 vs 18.9; P=.021), and stroke (3.3 vs 6.2; P=.032), respectively, for metformin vs sulfonylurea or insulin (group 2). Thus, metformin is the only oral hypoglycemic agent proven to reduce macrovascular risk in overweight patients with type 2 diabetes. For perspective, in overweight patients, metformin significantly reduced all-cause mortality (NNT per year=141; 95% CI, 115–183; P=.011), and any diabetes-related outcome (NNT per year=74; 95% CI, 63–90; P=.0023), compared with diet alone.^23,24

Metformin induces weight loss (2–3 kg), preferentially involving adipose tissue in obese patients with type 2 diabetes over 4 to 6 months.^22,25 In UKPDS 34, weight gain was similar among those treated with metformin and diet (approximately 2 kg); weight gain over 10 years was less with metformin, however, than with sulfonylurea (approximately 4 kg) or insulin (approximately 6 kg).²³ Metformin also significantly improved levels of total cholesterol, low-density lipoprotein (LDL) cholesterol, and triglycerides when compared with glyburide or placebo.²²

Risk of lactic acidosis. Lactic acidosis associated with metformin is a rare but serious adverse event, with an estimated prevalence of 3 cases per 100,000.²⁶ The product labeling notes most of these cases have occurred among patients with significant renal insufficiency, including both intrinsic renal disease and renal hypoperfusion. Absolute contraindications include renal disease (serum creatinine ≥1.5 mg/dL [males] and ≥1.4 mg/dL [females]), congestive heart failure requiring pharmacological treatment, and acute or chronic metabolic acidosis. It should also be discontinued at the time of radiologic studies using intravascular iodinated contrast materials.

Additional “precautionary conditions” include age ≥80 years (unless measurement of creatinine clearance demonstrates that renal function is not reduced), hepatic disease, cationic drug use, conditions associated with hypoxia (eg, chronic obstructive pulmonary disease [COPD], acute myocardial infarction, dehydration, sepsis), excessive alcohol intake, and surgery, until patient’s oral intake is resumed.

Is the risk overstated? Despite these extensive precautions, published studies show that metformin is commonly prescribed to patients with absolute contraindications.^27,28 One recent study observed that 11.2% of Medicare beneficiaries hospitalized with congestive heart failure and concomitant diabetes were treated with metformin.²⁸ In the absence of advanced renal dysfunction, metformin rarely accumulates in the body,²⁹ and accumulation of metformin is rarely reported as a cause of lactic acidosis.^30,31 Rather, tissue hypoxia acts as a trigger in most cases. Metformin should therefore be discontinued whenever tissue hypoxia is suspected.³¹

A recent systematic review and meta analysis found no evidence that metformin was associated with an increased risk of lactic acidosis if the drug was prescribed under study conditions, taking into account contraindications.³² Refinement and clarification of the risk for lactic acidosis in these various populations is needed, to ensure optimal patient safety and to further assess this highly effective medication.

Common adverse effects associated with metformin are diarrhea and nausea, which can be minimized by administering the drug with meals and slowly titrating the dose, or perhaps by using the extended-release formulation.

Thiazolidinediones

Thiazolidinediones (TZDs) include rosiglitazone and pioglitazone. These agents, like metformin, do not increase insulin secretion but depend on the presence of insulin for their activity. TZDs are agonists at peroxisome-proliferator-activated receptor gamma (PPAR-γ) receptors in peripheral tissues such as skeletal muscle, where they increase glucose uptake.¹⁵ Thus, their predominant effect is to decrease insulin resistance.

TZDs have similar antihyperglycemic efficacy as sulfonylureas or metformin. They decrease A_1c levels by 0.6–1.9 percentage points and lower fasting plasma glucose levels by 50–80 mg/dL.¹⁵ They have a slower onset of action compared with other hypoglycemic drugs, and intervals of 3 to 4 weeks should be allowed between doses before increasing the dosage. TZDs also have favorable effects on lipid levels: HDL concentrations increase and triglyceride concentrations decrease with their use.³³ It is not known whether they decrease macrovascular or microvascular complications, although such studies are underway.

Adverse effects. TZDs are typically well tolerated, though weight gain of 1–3 kg, edema (4%–5%) and anemia (1%–2%) can occur. Weight gain and edema are more pronounced when TZDs are used in combination with insulin. Anemia is likely due to increased plasma volume rather than any significant hematological effect.

Due to adverse events related to volume expansion, TZDs are not recommended for patients with New York Heart Association class III or IV heart failure. A recent consensus statement from the American Heart Association and the ADA stresses that before administering TZD treatment, the physician should explore the possible presence of cardiac disease, use of other drugs that cause fluid retention, and the pathogenesis of any existing edema or dyspnea.³⁴

Although troglitazone was removed from the market due to its association with hepatocellular injury, pioglitazone and rosiglitazone are not as convincingly associated with liver injury.¹⁵ In preapproval clinical studies, less than 0.5% of patients treated with rosiglitzone and pioglitazone had elevations in alanine transaminase (ALT) >3 times the upper limit of normal.

The incidence of hepatitis or acute liver failure from troglitazone was compared with rosiglitazone, pioglitazone, metformin, and glyburide, by analysis of spontaneously reported adverse events to the Food and Drug Administration (FDA) MEDWATCH database during the first 15 months of marketing of each drug.^35,36 The incidence of hepatitis per million prescriptions was 21.5, 14.7, 9.4, 2.9, and 4.1, respectively, while the incidence of acute liver failure per 100,000 prescriptions was 4.6, 0.9, 0.8, 0.2, and 0. It appears that postmarketing data support preclinical studies, in that the incidence of acute liver failure is an order of magnitude higher for troglitazone vs. other TZDs.³⁵ However, the FDA recommends avoiding their use in patients with baseline ALT levels >2.5 times the upper limit of normal. The FDA recently reduced the recommended frequency for ALT monitoring for pioglitazone (and is currently considering the same for rosiglitazone). Serum ALT is recommended prior to initiation and then periodically thereafter.

Cost. TZDs are expensive relative to other hypoglycemic agents.

α-glucosidase inhibitors

The α-glucosidase inhibitors (AGIs), acarbose and miglitol, act through competitive, reversible inhibition of membrane-bound intestinal α-glucosidase, which hydrolyzes complex carbohydrates to glucose and other monosaccharides. This inhibition delays glucose absorption and decreases postprandial hyperglycemia.³⁷ Thus, they have a nonsystemic mechanism of action.

These agents cause a modest reduction in the A_1c level (0.5–1.0 percentage points) and are thus less effective than sulfonylureas, metformin, or TZDs. They do not reduce fasting plasma glucose levels, but reduce postprandial hyperglycemia by 50 mg/dL.³⁸ No long-term studies have evaluated whether AGIs reduce diabetes-related macrovascular or microvascular outcomes.

Adverse effects. While AGIs are virtually free of serious toxicities, patient tolerability can be a problem due to adverse gastrointestinal effects. In indirect comparisons from placebo-controlled trials, patients treated with miglitol and acarbose commonly reported abdominal pain (11.7%, 19%), diarrhea (28.7%, 31%), and flatulence (41.5%, 74%), respectively. Systemic accumulation of AGIs has been shown to increase in proportion to the degree of renal insufficiency, and their use is not recommended for patients with serum creatinine >2.0 mg/dL. However, whether such patients are at greater risk of any toxicity is unknown. Acarbose at doses above 100 mg 3 times daily has been associated with elevated serum transaminase levels; however, this risk appears negligible at standard doses.

Insulin

Insulin is the oldest therapy for diabetes, and it has no upper dose limit.³⁹ It increases insulin levels and can reduce A_1c levels by 1.5 to 2.5 percentage points. Though half of diabetes patients need insulin eventually for optimal control, historically it has been introduced late in the disease process unless patients have severe hyperglycemia (fasting blood sugar >350 mg/dL) or ketonuria.³⁸ However, it is effective in gaining initial control, decreasing gluconeogenesis and increasing glucose uptake. Disadvantages are weight gain, hypoglycemia, and patient reluctance to give injections.

When insulin is indicated. Patients who exhibit persistent hyperglycemia despite oral hypoglycemic therapy may stop the oral drug(s) and begin insulin. By combining insulin with oral therapy, lower insulin doses may be used to achieve desired control vs using insulin alone.⁴⁰ For some patients a basal supplement of insulin may be sufficient and can be given as a single dose at bedtime, without an oral hypoglycemic drug.⁴¹

Insulin regimens. Various insulin regimens are available: very rapid acting (lispro and aspart), rapid acting (regular), intermediate acting (isophane insulin [NPH] and lente) and very long acting (ultralente and glargine). Glargine insulin (Lantus) has more predictable absorption than NPH, lente, and ultralente. Lantus, compared with NPH, has been associated with less nocturnal and postprandial hypoglycemia.^38,42,43 This is consistent with the peakless and longer duration of glargine compared with NPH.⁴⁴ A recent randomized controlled trial demonstrated that morning insulin glargine lowered A_1c levels more than a bedtime dose of NPH (–1.24 vs –0.84; 95% CI, 0.23%–0.58%) or a bedtime dose of glargine (–1.24 vs –0.96%; 95% CI, 0.11%–0.46%).⁴⁵ Glargine’s only relative disadvantage is increased cost.

Combination products. Combination insulin options are 70 NPH/30 regular, 50 NPH/50 regular, and 75 lispro protamine/25 lispro. Many combinations of insulin regimens have been used successfully. The typical range of insulin needed for monotherapy is 0.4–1 U/kg/d. Once-daily injection of intermediate acting or long acting insulins at bedtime or before breakfast, once-daily or twice-daily combinations of intermediate and rapid acting insulins, and more complex regimens have been used to good effect.

Using prandial insulin at each meal with separate basal insulin adds flexibility to meal times and doses administered.⁴³ With multiple-dose intensive insulin therapy, a basal dose suppresses hepatic glucose output and the bolus doses enhance postprandial glucose uptake. This intensive insulin treatment reduces mortality among critically ill patients in surgical intensive care units and for those with acute myocardial infarction.^46,47 An algorithm for using progressive therapy in diabetes mellitus is shown in Figure 2.⁴⁸

FIGURE 2
ADA recommendations for the treatment of type 2 diabetes

Combination Therapy

Over time glycemic control becomes more difficult, even with maximum monotherapy for patients with healthy lifestyles. It was shown in UKPDS 49 that monotherapy with sulfonylurea, metformin, or insulin eventually fails in most cases—by 3 years after diagnosis, about 50% of patients need more than monotherapy; 75% by 9 years.⁴⁹ In UKPDS 33, the median A_1c level increased steadily over 10 years with both conventional therapy and intensive therapy (Figure 3).²

Several options are available when monotherapy fails. Based on expert opinion, the principle is to combine drugs with different mechanisms of action to achieve an additive effect for glycemic control. Combination products may simplify the treatment regimen and improve adherence. In many instances, they may also cost less.⁵⁰

Successful combinations. The combination of sulfonylurea and metformin has proven effective in many studies.^22,51,52 One study showed that initial treatment with glyburide/metformin improved glycemic control better than either glyburide or metformin monotherapy (SOR: A).^53,54 The addition of the non-SU secretagogues repaglinide and nateglinide to metformin significantly improved glycemic control, with repaglinide showing superiority over nateglinide.⁵⁵ A TZD added to a sulfonylurea has also significantly improved A_1c and fasting blood sugar results.⁵⁶ Patients whose diabetes was inadequately controlled with diet alone or diet plus a sulfonylurea showed improvement with the addition of the AGI miglitol, compared with addition of placebo.⁵⁷ The AGI acarbose has shown to be an effective addition to diet, metformin, sulfonylurea, and insulin.⁵⁸ A TZD added to metformin has also been shown to improve glycemic control.⁵⁹ A non-SU added to patients inadequately controlled with a TZD has also been effective.⁶⁰

The early addition of insulin when maximal sulfonylurea therapy is inadequate has been effective.^61-63 When introducing insulin, a nighttime regimen of NPH or glargine, 10 units at bedtime, is an appropriate dose (SOR: C). This is easier and less costly than often assumed, and helps improve glycemic control.⁶⁴ Most patients require combination therapy as their disease progresses.³⁹

FIGURE 3
Glycemic control in type 2 DM

Improving Outcomes

Cumulative survey data reveal a wide gap between guideline recommendations and the care patients receive.⁶⁵ One study showed that physicians initiated treatment changes only after the A_1c level had reached 9.0% or higher instead of the 8.0% level recommended by ADA.⁶⁶ How can the quality of management be improved?

In private practices and institutions, many interventions have been shown to improve outcomes in diabetes mellitus. Education measures work, and they include chart audits, reminder cards, pharmacist collaboration, flow sheets, and nursing initiatives.^67,68 Effective disease-management programs have also used clinical guidelines, outcomes reporting, coverage of glucose meters and strips, and the support of clinical leadership.⁶⁹

Computerized systems that track patients and recommended laboratory tests have improved screening rates and glycemic and blood pressure control.⁷⁰ Monitoring patients’ readiness to change has allowed targeted education to improve A_1c levels.⁷¹ Continuity of care has also improved the quality of disease control by increasing adherence to recommended tests and exams.⁷²

Acknowlegments

The authors thank Marie Hamer, RN, for her continuous diabetes quality improvement efforts and Jean Camarata for her editorial and reference acquisition assistance.

Corresponding author
John E. Sutherland, MD, Northeast Iowa Family Practice Residency Program, University of Iowa College of Medicine, 2055 Kimball Avenue, Waterloo, Iowa 50702. E-mail: [email protected].

Glycemic control in diabetes begins with a patient’s adherence to several nonpharmacologic measures. Without such a commitment, success in controlling the disease will be difficult to achieve, and otherwise appropriate drug therapy will be hindered.

Most antidiabetic agents comparably reduce glycosylated hemoglobin (A_1c) levels. However, a particular agent may be preferred depending on a patient’s characteristics. And some circumstances call for combination therapy. This article reviews the advantages and disadvantages of the many pharmacologic treatments for glucose control and hyperglycemia in type 2 diabetes.

Benefits Of Diabetes Control

The benefits of diabetes control are detailed in this issue of THE JOURNAL OF FAMILY PRACTICE (“Strategies to reduce complications in type 2 diabetes,” pages 366–374). For every percentage-point reduction in hemoglobin A_1c, it is possible to achieve a 22% to 35% reduction in microvascular complications.^1,2 Cardiovascular disease can be reduced in patients with diabetes by treating hypertension^3,4 and hyperlipidemia, prescribing aspirin therapy, using angiotensin-converting enzyme (ACE) inhibitors, and with smoking cessation.^5,6

Targets For Glycemic Control

The American Diabetes Association’s (ADA) recommended targets for glycemic control are a preprandial blood glucose level of 80–120 mg/dL, a bedtime blood glucose level of 100–140 mg/dL, and a hemoglobin A_1c level of <7% (with a level of >8% requiring additional measures). Hemoglobin A_1c is the best determinant of glycemic exposure, and its mean value is a nationally recognized indicator of how well diabetes is being managed.⁷ The American College of Endocrinology has adopted a more aggressive approach by designating an A_1c level of 6.5% as both a target and action level.⁸

Self-monitoring of blood glucose

Self-monitoring of blood glucose (SMBG) is an integral component of diabetes therapy (strength of recommendation [SOR]: C) and should always be included in the management plan (SOR: C). The optimal frequency and timing of SMBG for type 2 diabetes is not known, but they should be sufficient to facilitate reaching glucose goals. The A_1c test should be performed at least semi-annually for patients with stable glycemic control, and quarterly for patients not meeting glycemic goals or those who are changing therapy. A_1c levels and mean plasma glucose levels can be approximately correlated (Table 1).⁷

TABLE 1
Correlation between hemoglobin A_1clevels and mean plasma glucose levels

Nonpharmacologic Therapy

Nonpharmacologic measures remain the cornerstone of managing type 2 diabetes. Hyperglycemia adversely and reversibly affects both insulin resistance and insulin secretion. Improvement in glycemic control can occur through dietary modification and regular exercise.

A recent meta-analysis of randomized controlled trials of diabetes patient education observed a net decrease in HbA_1c of 0.32% in intervention groups vs control.⁹ Interventions that included a face-to-face delivery, cognitive reframing teaching method, and exercise content were more likely to improve glycemic control.

Education

Lifestyle changes involving diet, exercise, and usually weight loss are key to effective management of diabetes. If patients are to change their behavior, they must be given detailed training.⁶ Self-management also necessitates that patients engage in problem solving. This requires that each aspect of the management plan is understood and agreed upon by the patient and providers, and that the goals and treatment plan are individualized and reasonable.

Diet: recommend soluble fiber, reduce calories

Medical nutrition therapy should be individualized and preferably provided by a registered dietitian familiar with diabetes (SOR: B). The goals of nutrition therapy, according to the ADA, are to attain recommended body weight and prevent or reverse obesity. The means of achieving these goals are nutrition assessment and modification of nutrient intake and lifestyle through healthy food choices and physical activity.⁷

A high intake of dietary fiber (particularly the soluble type) above the level recommended by the ADA improves glycemic control, decreases hyperinsulinemia, and lowers plasma lipid concentrations.¹⁰

Hypocaloric diets cause glucose plasma levels to fall, in some cases to a normal level with a weight loss of even 5 to 10 pounds.^7,11 Hypoglycemic medications are of course most effective in nonobese persons. But effectiveness is also improved if weight that is gained can be limited. Despite the clear benefit of weight loss, only a few patients are able to attain and maintain substantial weight loss. Maintenance of a reduced or elevated body weight is associated with compensatory changes in energy expenditure, which oppose the maintenance of body weight that is different from the usual weight.¹³ Part of the individualization of therapy is respect of personal and cultural preferences, lifestyles, and financial considerations.

Physical activity: a little goes a long way

A regular physical activity program is recommended for all patients with diabetes who are capable of participating (SOR: B).⁷ It improves blood glucose control, reduces cardiovascular risk factors, aids weight loss, and enhances well being.⁷ A recently published prospective cohort study showed that walking at least 2 hours a week was associated with a 39% lower all-cause mortality (hazard rate ratio [HRR], 0.61; 95% CI, 0.48–0.78) and a 34% lower cardiovascular mortality (HRR, 0.66; 95% CI, 0.45–0.96) across a diverse spectrum of adults with diabetes. The NNT (to prevent 1 death per year) is 61 for patients who walk at least 2 hours/week.¹⁴

In prescribing a physical activity plan for a patient, consider cardiovascular disease risk factors or complications to minimize the risk of untoward events. Micro- and macrovascular disease are of course prevalent among persons with diabetes, often resulting in functional limitations that make exercise more difficult.

Other priorities

Other recommended components of care include daily aspirin use, foot care exams, tobacco cessation, pneumococcal and influenza vaccinations, and an annual dilated retinal exam.

Pharmacologic Therapy

The coexisting defects in type 2 diabetes mellitus are as follows:

• resistance to insulin action in muscle
• defective pancreatic insulin secretion
• unrestrained hepatic glucose production, aggravated by increased lipolysis in adipose tissue.

Drug therapy is aimed at each of these defects, and also at reducing carbohydrate absorption in the small intestine (Figure 1). As far as antihyperglycemic effect is concerned, no one category of antidiabetic agent is favored over another.¹⁵ Except for nateglinide and α-glucosidase inhibitors (AGIs), each of the drug categories leads to a similar reduction in A_1c.¹⁶ However, patient characteristics may lead to selection of a particular agent. Table 2 summarizes oral treatment options, their relative advantages and costs.

FIGURE 1
Drug therapies for coexisting defects in type 2 diabetes

TABLE 2
Pharmacologic treatments for type 2 diabetes: monotherapies

Sulfonylureas

Sulfonylureas directly increase insulin secretion by binding to the sulfonylurea receptor on pancreatic beta cells; they provide a relatively quick onset of action. First-generation sulfonylureas (eg, tolbutamide, chlorpropamide) and second-generation sulfonylureas (eg, glyburide, glipizide, glimepiride) are equivalent in their maximum hypoglycemic effect.¹⁷

Second-generation agents are used more commonly than first-generation. They all contain the sulfonylurea moiety, but different chemical substitutions in the basic molecule change pharmacokinetics, resulting in different durations of action.¹⁷ Second-generation agents are probably safer than first-generation drugs, being less likely to cause hyponatremia, disulfiram-like reactions, or prolonged hypoglycemia.¹⁸

At maximal doses, sulfonylureas lower A_1c levels by 1–2 percentage points and fasting plasma glucose concentrations by 60–70 mg/dL;¹⁵ however, the glucose lowering effect typically plateaus after half the maximal recommended dose is reached. Sulfonylureas have no consistent effect on dyslipidemia. In UK Prospective Diabetes Study (UKPDS) 33, though improved glycemic control with sulfonylureas (or insulin) led to a 25% reduction in microvascular endpoints (mostly less retinal photocoagulation) (P<.01), sulfonylureas (or insulin) did not significantly reduce death or all-cause mortality compared with diet treatment.²

Adverse effects. The primary adverse effects of sulfonylureas are weight gain and hypoglycemia. In UKPDS 33, weight gain at 10 years was 2.6 kg (99% confidence interval [CI], 1.6–3.6) with chlorpropamide and 1.7 kg (99% CI, 0.7–2.7) with glyburide, compared with patients receiving diet therapy (each P<.001).² In the same study, the rate of major hypoglycemic episodes (third-party help or medical intervention necessary) while on therapy was 0.4%/year for chlorpropamide and 0.6%/year for glyburide, compared with 0.1%/year for diet.

Glyburide and chlorpropamide have active metabolites with renal elimination, and they should therefore be used with caution in patients with renal insufficiency. In 1971, the University Group Diabetes Project (UGDP) observed a twofold increase in the rate of cardiovascular death among patients receiving tolbutamide compared with those receiving insulin or placebo.¹⁸ This led to a decades long debate on the validity of this conclusion.¹⁹ More recently, UKPDS 33 did not demonstrate any increased cardiovascular mortality among patients receiving glyburide or chlorpropamide, and has largely negated this earlier concern.²

Cost. Sulfonylureas are the least expensive oral agents used to treat type 2 diabetes.

TABLE 3
Pharmacological treatments for type 2 diabetes: combination therapies

Non-sulfonylurea secretagogues

Like sulfonylureas, the non-sulfonylurea secretagogues (non-SU), repaglinide and nateglinide, stimulate beta cells to increase insulin secretion. However, the non-SU agents mediate their action through a different, adjacent site on the “sulfonylurea receptor.” Comparatively, the non-SU agents have a faster onset of action (20 minutes), shorter half-life (about 1.0–1.5 hours), and greater effects on postprandial glucose excursions than do sulfonylureas.²⁰ In contrast to the sulfonylureas, the extent of insulin release with non-SU agents is glucose dependent, and therefore they may have less risk of hypoglycemia several hours after meals.¹⁵

Repaglinide lowers the A_1c level by 1.7–1.9 percentage points, similar in efficacy to sulfonylureas. Nateglinide appears somewhat less efficacious and lowers A_1c by 0.6–1.0 percentage points.¹⁵ Nateglinide was significantly less effective than glyburide at lowering A_1c levels and the fasting plasma glucose in one 24-week study. Non-SUs added to sulfonylureas produce no additional benefit in glycemic control. The effect of non-SUs on microvascular or macrovascular endpoints is unknown.

Adverse effects. Hypoglycemia is the primary adverse effect of non-SUs. Confirmed hypoglycemia (plasma glucose <60 mg/dL) was observed in 2.4% of patients taking nateglinide compared with 0.4% of those receiving placebo. Mild or moderate hypoglycemia occurred in 16% of repaglinide patients, 20% of glyburide patients, and 19% of glipizide patients in one-year comparative studies. Further comparative studies are needed to determine if non-SUs produce significantly less hypoglycemia and weight gain than sulfonylureas.

Cost. Non-SUs must be dosed 3 times daily at the start of meals. One relative disadvantage is their increased cost compared with sulfonylureas.

Biguanides

The only biguanide marketed in the US is metformin. Its primary action is to inhibit hepatic glucose production and, to a much lesser extent, enhance insulin sensitivity in peripheral tissues.²¹ Metformin does not stimulate insulin secretion and does not cause hypoglycemia when used as monotherapy, but it can potentiate hypoglycemia in combination with insulin or insulin secretagogues.

Metformin is similar in efficacy to the sulfonylureas. It lowers A_1c by 1.5–2.0 percentage points and fasting plasma glucose by 60–80 mg/dL. Its antihyperglycemic efficacy is independent of patient age, duration of diabetes, or BMI.²²

In the UKPDS 34 study, a subgroup of obese patients was randomized to receive intensive control (group 1, metformin; group 2, a sulfonylurea or insulin) or conventional diet therapy (group 3). Despite a similar reduction in the A_1c level between the 2 intensive-treatment groups, patients treated with metformin had a 32% reduction for any diabetes-related endpoint (95% CI, 13–47; P=.002), 43% fewer diabetes-related deaths (95% CI, 9–63; P=.017), and a 36% reduction in all cause mortality, compared with the diet therapy group (95% CI, 9–55; P=.011).²³

Metformin also showed significant benefit when compared with patients receiving sulfonylurea or insulin (group 2). The absolute risk of any diabetes endpoint was 29.8 vs. 40.1 (events per 1000 patient-years; P=.0034), all-cause mortality (13.5 vs 18.9; P=.021), and stroke (3.3 vs 6.2; P=.032), respectively, for metformin vs sulfonylurea or insulin (group 2). Thus, metformin is the only oral hypoglycemic agent proven to reduce macrovascular risk in overweight patients with type 2 diabetes. For perspective, in overweight patients, metformin significantly reduced all-cause mortality (NNT per year=141; 95% CI, 115–183; P=.011), and any diabetes-related outcome (NNT per year=74; 95% CI, 63–90; P=.0023), compared with diet alone.^23,24

Metformin induces weight loss (2–3 kg), preferentially involving adipose tissue in obese patients with type 2 diabetes over 4 to 6 months.^22,25 In UKPDS 34, weight gain was similar among those treated with metformin and diet (approximately 2 kg); weight gain over 10 years was less with metformin, however, than with sulfonylurea (approximately 4 kg) or insulin (approximately 6 kg).²³ Metformin also significantly improved levels of total cholesterol, low-density lipoprotein (LDL) cholesterol, and triglycerides when compared with glyburide or placebo.²²

Risk of lactic acidosis. Lactic acidosis associated with metformin is a rare but serious adverse event, with an estimated prevalence of 3 cases per 100,000.²⁶ The product labeling notes most of these cases have occurred among patients with significant renal insufficiency, including both intrinsic renal disease and renal hypoperfusion. Absolute contraindications include renal disease (serum creatinine ≥1.5 mg/dL [males] and ≥1.4 mg/dL [females]), congestive heart failure requiring pharmacological treatment, and acute or chronic metabolic acidosis. It should also be discontinued at the time of radiologic studies using intravascular iodinated contrast materials.

Additional “precautionary conditions” include age ≥80 years (unless measurement of creatinine clearance demonstrates that renal function is not reduced), hepatic disease, cationic drug use, conditions associated with hypoxia (eg, chronic obstructive pulmonary disease [COPD], acute myocardial infarction, dehydration, sepsis), excessive alcohol intake, and surgery, until patient’s oral intake is resumed.

Is the risk overstated? Despite these extensive precautions, published studies show that metformin is commonly prescribed to patients with absolute contraindications.^27,28 One recent study observed that 11.2% of Medicare beneficiaries hospitalized with congestive heart failure and concomitant diabetes were treated with metformin.²⁸ In the absence of advanced renal dysfunction, metformin rarely accumulates in the body,²⁹ and accumulation of metformin is rarely reported as a cause of lactic acidosis.^30,31 Rather, tissue hypoxia acts as a trigger in most cases. Metformin should therefore be discontinued whenever tissue hypoxia is suspected.³¹

A recent systematic review and meta analysis found no evidence that metformin was associated with an increased risk of lactic acidosis if the drug was prescribed under study conditions, taking into account contraindications.³² Refinement and clarification of the risk for lactic acidosis in these various populations is needed, to ensure optimal patient safety and to further assess this highly effective medication.

Common adverse effects associated with metformin are diarrhea and nausea, which can be minimized by administering the drug with meals and slowly titrating the dose, or perhaps by using the extended-release formulation.

Thiazolidinediones

Thiazolidinediones (TZDs) include rosiglitazone and pioglitazone. These agents, like metformin, do not increase insulin secretion but depend on the presence of insulin for their activity. TZDs are agonists at peroxisome-proliferator-activated receptor gamma (PPAR-γ) receptors in peripheral tissues such as skeletal muscle, where they increase glucose uptake.¹⁵ Thus, their predominant effect is to decrease insulin resistance.

TZDs have similar antihyperglycemic efficacy as sulfonylureas or metformin. They decrease A_1c levels by 0.6–1.9 percentage points and lower fasting plasma glucose levels by 50–80 mg/dL.¹⁵ They have a slower onset of action compared with other hypoglycemic drugs, and intervals of 3 to 4 weeks should be allowed between doses before increasing the dosage. TZDs also have favorable effects on lipid levels: HDL concentrations increase and triglyceride concentrations decrease with their use.³³ It is not known whether they decrease macrovascular or microvascular complications, although such studies are underway.

Adverse effects. TZDs are typically well tolerated, though weight gain of 1–3 kg, edema (4%–5%) and anemia (1%–2%) can occur. Weight gain and edema are more pronounced when TZDs are used in combination with insulin. Anemia is likely due to increased plasma volume rather than any significant hematological effect.

Due to adverse events related to volume expansion, TZDs are not recommended for patients with New York Heart Association class III or IV heart failure. A recent consensus statement from the American Heart Association and the ADA stresses that before administering TZD treatment, the physician should explore the possible presence of cardiac disease, use of other drugs that cause fluid retention, and the pathogenesis of any existing edema or dyspnea.³⁴

Although troglitazone was removed from the market due to its association with hepatocellular injury, pioglitazone and rosiglitazone are not as convincingly associated with liver injury.¹⁵ In preapproval clinical studies, less than 0.5% of patients treated with rosiglitzone and pioglitazone had elevations in alanine transaminase (ALT) >3 times the upper limit of normal.

The incidence of hepatitis or acute liver failure from troglitazone was compared with rosiglitazone, pioglitazone, metformin, and glyburide, by analysis of spontaneously reported adverse events to the Food and Drug Administration (FDA) MEDWATCH database during the first 15 months of marketing of each drug.^35,36 The incidence of hepatitis per million prescriptions was 21.5, 14.7, 9.4, 2.9, and 4.1, respectively, while the incidence of acute liver failure per 100,000 prescriptions was 4.6, 0.9, 0.8, 0.2, and 0. It appears that postmarketing data support preclinical studies, in that the incidence of acute liver failure is an order of magnitude higher for troglitazone vs. other TZDs.³⁵ However, the FDA recommends avoiding their use in patients with baseline ALT levels >2.5 times the upper limit of normal. The FDA recently reduced the recommended frequency for ALT monitoring for pioglitazone (and is currently considering the same for rosiglitazone). Serum ALT is recommended prior to initiation and then periodically thereafter.

Cost. TZDs are expensive relative to other hypoglycemic agents.

α-glucosidase inhibitors

The α-glucosidase inhibitors (AGIs), acarbose and miglitol, act through competitive, reversible inhibition of membrane-bound intestinal α-glucosidase, which hydrolyzes complex carbohydrates to glucose and other monosaccharides. This inhibition delays glucose absorption and decreases postprandial hyperglycemia.³⁷ Thus, they have a nonsystemic mechanism of action.

These agents cause a modest reduction in the A_1c level (0.5–1.0 percentage points) and are thus less effective than sulfonylureas, metformin, or TZDs. They do not reduce fasting plasma glucose levels, but reduce postprandial hyperglycemia by 50 mg/dL.³⁸ No long-term studies have evaluated whether AGIs reduce diabetes-related macrovascular or microvascular outcomes.

Adverse effects. While AGIs are virtually free of serious toxicities, patient tolerability can be a problem due to adverse gastrointestinal effects. In indirect comparisons from placebo-controlled trials, patients treated with miglitol and acarbose commonly reported abdominal pain (11.7%, 19%), diarrhea (28.7%, 31%), and flatulence (41.5%, 74%), respectively. Systemic accumulation of AGIs has been shown to increase in proportion to the degree of renal insufficiency, and their use is not recommended for patients with serum creatinine >2.0 mg/dL. However, whether such patients are at greater risk of any toxicity is unknown. Acarbose at doses above 100 mg 3 times daily has been associated with elevated serum transaminase levels; however, this risk appears negligible at standard doses.

Insulin

Insulin is the oldest therapy for diabetes, and it has no upper dose limit.³⁹ It increases insulin levels and can reduce A_1c levels by 1.5 to 2.5 percentage points. Though half of diabetes patients need insulin eventually for optimal control, historically it has been introduced late in the disease process unless patients have severe hyperglycemia (fasting blood sugar >350 mg/dL) or ketonuria.³⁸ However, it is effective in gaining initial control, decreasing gluconeogenesis and increasing glucose uptake. Disadvantages are weight gain, hypoglycemia, and patient reluctance to give injections.

When insulin is indicated. Patients who exhibit persistent hyperglycemia despite oral hypoglycemic therapy may stop the oral drug(s) and begin insulin. By combining insulin with oral therapy, lower insulin doses may be used to achieve desired control vs using insulin alone.⁴⁰ For some patients a basal supplement of insulin may be sufficient and can be given as a single dose at bedtime, without an oral hypoglycemic drug.⁴¹

Insulin regimens. Various insulin regimens are available: very rapid acting (lispro and aspart), rapid acting (regular), intermediate acting (isophane insulin [NPH] and lente) and very long acting (ultralente and glargine). Glargine insulin (Lantus) has more predictable absorption than NPH, lente, and ultralente. Lantus, compared with NPH, has been associated with less nocturnal and postprandial hypoglycemia.^38,42,43 This is consistent with the peakless and longer duration of glargine compared with NPH.⁴⁴ A recent randomized controlled trial demonstrated that morning insulin glargine lowered A_1c levels more than a bedtime dose of NPH (–1.24 vs –0.84; 95% CI, 0.23%–0.58%) or a bedtime dose of glargine (–1.24 vs –0.96%; 95% CI, 0.11%–0.46%).⁴⁵ Glargine’s only relative disadvantage is increased cost.

Combination products. Combination insulin options are 70 NPH/30 regular, 50 NPH/50 regular, and 75 lispro protamine/25 lispro. Many combinations of insulin regimens have been used successfully. The typical range of insulin needed for monotherapy is 0.4–1 U/kg/d. Once-daily injection of intermediate acting or long acting insulins at bedtime or before breakfast, once-daily or twice-daily combinations of intermediate and rapid acting insulins, and more complex regimens have been used to good effect.

Using prandial insulin at each meal with separate basal insulin adds flexibility to meal times and doses administered.⁴³ With multiple-dose intensive insulin therapy, a basal dose suppresses hepatic glucose output and the bolus doses enhance postprandial glucose uptake. This intensive insulin treatment reduces mortality among critically ill patients in surgical intensive care units and for those with acute myocardial infarction.^46,47 An algorithm for using progressive therapy in diabetes mellitus is shown in Figure 2.⁴⁸

FIGURE 2
ADA recommendations for the treatment of type 2 diabetes

Combination Therapy

Over time glycemic control becomes more difficult, even with maximum monotherapy for patients with healthy lifestyles. It was shown in UKPDS 49 that monotherapy with sulfonylurea, metformin, or insulin eventually fails in most cases—by 3 years after diagnosis, about 50% of patients need more than monotherapy; 75% by 9 years.⁴⁹ In UKPDS 33, the median A_1c level increased steadily over 10 years with both conventional therapy and intensive therapy (Figure 3).²

Several options are available when monotherapy fails. Based on expert opinion, the principle is to combine drugs with different mechanisms of action to achieve an additive effect for glycemic control. Combination products may simplify the treatment regimen and improve adherence. In many instances, they may also cost less.⁵⁰

Successful combinations. The combination of sulfonylurea and metformin has proven effective in many studies.^22,51,52 One study showed that initial treatment with glyburide/metformin improved glycemic control better than either glyburide or metformin monotherapy (SOR: A).^53,54 The addition of the non-SU secretagogues repaglinide and nateglinide to metformin significantly improved glycemic control, with repaglinide showing superiority over nateglinide.⁵⁵ A TZD added to a sulfonylurea has also significantly improved A_1c and fasting blood sugar results.⁵⁶ Patients whose diabetes was inadequately controlled with diet alone or diet plus a sulfonylurea showed improvement with the addition of the AGI miglitol, compared with addition of placebo.⁵⁷ The AGI acarbose has shown to be an effective addition to diet, metformin, sulfonylurea, and insulin.⁵⁸ A TZD added to metformin has also been shown to improve glycemic control.⁵⁹ A non-SU added to patients inadequately controlled with a TZD has also been effective.⁶⁰

The early addition of insulin when maximal sulfonylurea therapy is inadequate has been effective.^61-63 When introducing insulin, a nighttime regimen of NPH or glargine, 10 units at bedtime, is an appropriate dose (SOR: C). This is easier and less costly than often assumed, and helps improve glycemic control.⁶⁴ Most patients require combination therapy as their disease progresses.³⁹

FIGURE 3
Glycemic control in type 2 DM

Improving Outcomes

Cumulative survey data reveal a wide gap between guideline recommendations and the care patients receive.⁶⁵ One study showed that physicians initiated treatment changes only after the A_1c level had reached 9.0% or higher instead of the 8.0% level recommended by ADA.⁶⁶ How can the quality of management be improved?

In private practices and institutions, many interventions have been shown to improve outcomes in diabetes mellitus. Education measures work, and they include chart audits, reminder cards, pharmacist collaboration, flow sheets, and nursing initiatives.^67,68 Effective disease-management programs have also used clinical guidelines, outcomes reporting, coverage of glucose meters and strips, and the support of clinical leadership.⁶⁹

Computerized systems that track patients and recommended laboratory tests have improved screening rates and glycemic and blood pressure control.⁷⁰ Monitoring patients’ readiness to change has allowed targeted education to improve A_1c levels.⁷¹ Continuity of care has also improved the quality of disease control by increasing adherence to recommended tests and exams.⁷²

Acknowlegments

The authors thank Marie Hamer, RN, for her continuous diabetes quality improvement efforts and Jean Camarata for her editorial and reference acquisition assistance.

Corresponding author
John E. Sutherland, MD, Northeast Iowa Family Practice Residency Program, University of Iowa College of Medicine, 2055 Kimball Avenue, Waterloo, Iowa 50702. E-mail: [email protected].

Glycemic control in diabetes begins with a patient’s adherence to several nonpharmacologic measures. Without such a commitment, success in controlling the disease will be difficult to achieve, and otherwise appropriate drug therapy will be hindered.

Most antidiabetic agents comparably reduce glycosylated hemoglobin (A_1c) levels. However, a particular agent may be preferred depending on a patient’s characteristics. And some circumstances call for combination therapy. This article reviews the advantages and disadvantages of the many pharmacologic treatments for glucose control and hyperglycemia in type 2 diabetes.

Benefits Of Diabetes Control

The benefits of diabetes control are detailed in this issue of THE JOURNAL OF FAMILY PRACTICE (“Strategies to reduce complications in type 2 diabetes,” pages 366–374). For every percentage-point reduction in hemoglobin A_1c, it is possible to achieve a 22% to 35% reduction in microvascular complications.^1,2 Cardiovascular disease can be reduced in patients with diabetes by treating hypertension^3,4 and hyperlipidemia, prescribing aspirin therapy, using angiotensin-converting enzyme (ACE) inhibitors, and with smoking cessation.^5,6

Targets For Glycemic Control

The American Diabetes Association’s (ADA) recommended targets for glycemic control are a preprandial blood glucose level of 80–120 mg/dL, a bedtime blood glucose level of 100–140 mg/dL, and a hemoglobin A_1c level of <7% (with a level of >8% requiring additional measures). Hemoglobin A_1c is the best determinant of glycemic exposure, and its mean value is a nationally recognized indicator of how well diabetes is being managed.⁷ The American College of Endocrinology has adopted a more aggressive approach by designating an A_1c level of 6.5% as both a target and action level.⁸

Self-monitoring of blood glucose

Self-monitoring of blood glucose (SMBG) is an integral component of diabetes therapy (strength of recommendation [SOR]: C) and should always be included in the management plan (SOR: C). The optimal frequency and timing of SMBG for type 2 diabetes is not known, but they should be sufficient to facilitate reaching glucose goals. The A_1c test should be performed at least semi-annually for patients with stable glycemic control, and quarterly for patients not meeting glycemic goals or those who are changing therapy. A_1c levels and mean plasma glucose levels can be approximately correlated (Table 1).⁷

TABLE 1
Correlation between hemoglobin A_1clevels and mean plasma glucose levels

Nonpharmacologic Therapy

Nonpharmacologic measures remain the cornerstone of managing type 2 diabetes. Hyperglycemia adversely and reversibly affects both insulin resistance and insulin secretion. Improvement in glycemic control can occur through dietary modification and regular exercise.

A recent meta-analysis of randomized controlled trials of diabetes patient education observed a net decrease in HbA_1c of 0.32% in intervention groups vs control.⁹ Interventions that included a face-to-face delivery, cognitive reframing teaching method, and exercise content were more likely to improve glycemic control.

Education

Lifestyle changes involving diet, exercise, and usually weight loss are key to effective management of diabetes. If patients are to change their behavior, they must be given detailed training.⁶ Self-management also necessitates that patients engage in problem solving. This requires that each aspect of the management plan is understood and agreed upon by the patient and providers, and that the goals and treatment plan are individualized and reasonable.

Diet: recommend soluble fiber, reduce calories

Medical nutrition therapy should be individualized and preferably provided by a registered dietitian familiar with diabetes (SOR: B). The goals of nutrition therapy, according to the ADA, are to attain recommended body weight and prevent or reverse obesity. The means of achieving these goals are nutrition assessment and modification of nutrient intake and lifestyle through healthy food choices and physical activity.⁷

A high intake of dietary fiber (particularly the soluble type) above the level recommended by the ADA improves glycemic control, decreases hyperinsulinemia, and lowers plasma lipid concentrations.¹⁰

Hypocaloric diets cause glucose plasma levels to fall, in some cases to a normal level with a weight loss of even 5 to 10 pounds.^7,11 Hypoglycemic medications are of course most effective in nonobese persons. But effectiveness is also improved if weight that is gained can be limited. Despite the clear benefit of weight loss, only a few patients are able to attain and maintain substantial weight loss. Maintenance of a reduced or elevated body weight is associated with compensatory changes in energy expenditure, which oppose the maintenance of body weight that is different from the usual weight.¹³ Part of the individualization of therapy is respect of personal and cultural preferences, lifestyles, and financial considerations.

Physical activity: a little goes a long way

A regular physical activity program is recommended for all patients with diabetes who are capable of participating (SOR: B).⁷ It improves blood glucose control, reduces cardiovascular risk factors, aids weight loss, and enhances well being.⁷ A recently published prospective cohort study showed that walking at least 2 hours a week was associated with a 39% lower all-cause mortality (hazard rate ratio [HRR], 0.61; 95% CI, 0.48–0.78) and a 34% lower cardiovascular mortality (HRR, 0.66; 95% CI, 0.45–0.96) across a diverse spectrum of adults with diabetes. The NNT (to prevent 1 death per year) is 61 for patients who walk at least 2 hours/week.¹⁴

In prescribing a physical activity plan for a patient, consider cardiovascular disease risk factors or complications to minimize the risk of untoward events. Micro- and macrovascular disease are of course prevalent among persons with diabetes, often resulting in functional limitations that make exercise more difficult.

Other priorities

Other recommended components of care include daily aspirin use, foot care exams, tobacco cessation, pneumococcal and influenza vaccinations, and an annual dilated retinal exam.

Pharmacologic Therapy

The coexisting defects in type 2 diabetes mellitus are as follows:

• resistance to insulin action in muscle
• defective pancreatic insulin secretion
• unrestrained hepatic glucose production, aggravated by increased lipolysis in adipose tissue.

Drug therapy is aimed at each of these defects, and also at reducing carbohydrate absorption in the small intestine (Figure 1). As far as antihyperglycemic effect is concerned, no one category of antidiabetic agent is favored over another.¹⁵ Except for nateglinide and α-glucosidase inhibitors (AGIs), each of the drug categories leads to a similar reduction in A_1c.¹⁶ However, patient characteristics may lead to selection of a particular agent. Table 2 summarizes oral treatment options, their relative advantages and costs.

FIGURE 1
Drug therapies for coexisting defects in type 2 diabetes

TABLE 2
Pharmacologic treatments for type 2 diabetes: monotherapies

Sulfonylureas

Sulfonylureas directly increase insulin secretion by binding to the sulfonylurea receptor on pancreatic beta cells; they provide a relatively quick onset of action. First-generation sulfonylureas (eg, tolbutamide, chlorpropamide) and second-generation sulfonylureas (eg, glyburide, glipizide, glimepiride) are equivalent in their maximum hypoglycemic effect.¹⁷

Second-generation agents are used more commonly than first-generation. They all contain the sulfonylurea moiety, but different chemical substitutions in the basic molecule change pharmacokinetics, resulting in different durations of action.¹⁷ Second-generation agents are probably safer than first-generation drugs, being less likely to cause hyponatremia, disulfiram-like reactions, or prolonged hypoglycemia.¹⁸

At maximal doses, sulfonylureas lower A_1c levels by 1–2 percentage points and fasting plasma glucose concentrations by 60–70 mg/dL;¹⁵ however, the glucose lowering effect typically plateaus after half the maximal recommended dose is reached. Sulfonylureas have no consistent effect on dyslipidemia. In UK Prospective Diabetes Study (UKPDS) 33, though improved glycemic control with sulfonylureas (or insulin) led to a 25% reduction in microvascular endpoints (mostly less retinal photocoagulation) (P<.01), sulfonylureas (or insulin) did not significantly reduce death or all-cause mortality compared with diet treatment.²

Adverse effects. The primary adverse effects of sulfonylureas are weight gain and hypoglycemia. In UKPDS 33, weight gain at 10 years was 2.6 kg (99% confidence interval [CI], 1.6–3.6) with chlorpropamide and 1.7 kg (99% CI, 0.7–2.7) with glyburide, compared with patients receiving diet therapy (each P<.001).² In the same study, the rate of major hypoglycemic episodes (third-party help or medical intervention necessary) while on therapy was 0.4%/year for chlorpropamide and 0.6%/year for glyburide, compared with 0.1%/year for diet.

Glyburide and chlorpropamide have active metabolites with renal elimination, and they should therefore be used with caution in patients with renal insufficiency. In 1971, the University Group Diabetes Project (UGDP) observed a twofold increase in the rate of cardiovascular death among patients receiving tolbutamide compared with those receiving insulin or placebo.¹⁸ This led to a decades long debate on the validity of this conclusion.¹⁹ More recently, UKPDS 33 did not demonstrate any increased cardiovascular mortality among patients receiving glyburide or chlorpropamide, and has largely negated this earlier concern.²

Cost. Sulfonylureas are the least expensive oral agents used to treat type 2 diabetes.

TABLE 3
Pharmacological treatments for type 2 diabetes: combination therapies

Non-sulfonylurea secretagogues

Like sulfonylureas, the non-sulfonylurea secretagogues (non-SU), repaglinide and nateglinide, stimulate beta cells to increase insulin secretion. However, the non-SU agents mediate their action through a different, adjacent site on the “sulfonylurea receptor.” Comparatively, the non-SU agents have a faster onset of action (20 minutes), shorter half-life (about 1.0–1.5 hours), and greater effects on postprandial glucose excursions than do sulfonylureas.²⁰ In contrast to the sulfonylureas, the extent of insulin release with non-SU agents is glucose dependent, and therefore they may have less risk of hypoglycemia several hours after meals.¹⁵

Repaglinide lowers the A_1c level by 1.7–1.9 percentage points, similar in efficacy to sulfonylureas. Nateglinide appears somewhat less efficacious and lowers A_1c by 0.6–1.0 percentage points.¹⁵ Nateglinide was significantly less effective than glyburide at lowering A_1c levels and the fasting plasma glucose in one 24-week study. Non-SUs added to sulfonylureas produce no additional benefit in glycemic control. The effect of non-SUs on microvascular or macrovascular endpoints is unknown.

Adverse effects. Hypoglycemia is the primary adverse effect of non-SUs. Confirmed hypoglycemia (plasma glucose <60 mg/dL) was observed in 2.4% of patients taking nateglinide compared with 0.4% of those receiving placebo. Mild or moderate hypoglycemia occurred in 16% of repaglinide patients, 20% of glyburide patients, and 19% of glipizide patients in one-year comparative studies. Further comparative studies are needed to determine if non-SUs produce significantly less hypoglycemia and weight gain than sulfonylureas.

Cost. Non-SUs must be dosed 3 times daily at the start of meals. One relative disadvantage is their increased cost compared with sulfonylureas.

Biguanides

The only biguanide marketed in the US is metformin. Its primary action is to inhibit hepatic glucose production and, to a much lesser extent, enhance insulin sensitivity in peripheral tissues.²¹ Metformin does not stimulate insulin secretion and does not cause hypoglycemia when used as monotherapy, but it can potentiate hypoglycemia in combination with insulin or insulin secretagogues.

Metformin is similar in efficacy to the sulfonylureas. It lowers A_1c by 1.5–2.0 percentage points and fasting plasma glucose by 60–80 mg/dL. Its antihyperglycemic efficacy is independent of patient age, duration of diabetes, or BMI.²²

In the UKPDS 34 study, a subgroup of obese patients was randomized to receive intensive control (group 1, metformin; group 2, a sulfonylurea or insulin) or conventional diet therapy (group 3). Despite a similar reduction in the A_1c level between the 2 intensive-treatment groups, patients treated with metformin had a 32% reduction for any diabetes-related endpoint (95% CI, 13–47; P=.002), 43% fewer diabetes-related deaths (95% CI, 9–63; P=.017), and a 36% reduction in all cause mortality, compared with the diet therapy group (95% CI, 9–55; P=.011).²³

Metformin also showed significant benefit when compared with patients receiving sulfonylurea or insulin (group 2). The absolute risk of any diabetes endpoint was 29.8 vs. 40.1 (events per 1000 patient-years; P=.0034), all-cause mortality (13.5 vs 18.9; P=.021), and stroke (3.3 vs 6.2; P=.032), respectively, for metformin vs sulfonylurea or insulin (group 2). Thus, metformin is the only oral hypoglycemic agent proven to reduce macrovascular risk in overweight patients with type 2 diabetes. For perspective, in overweight patients, metformin significantly reduced all-cause mortality (NNT per year=141; 95% CI, 115–183; P=.011), and any diabetes-related outcome (NNT per year=74; 95% CI, 63–90; P=.0023), compared with diet alone.^23,24

Metformin induces weight loss (2–3 kg), preferentially involving adipose tissue in obese patients with type 2 diabetes over 4 to 6 months.^22,25 In UKPDS 34, weight gain was similar among those treated with metformin and diet (approximately 2 kg); weight gain over 10 years was less with metformin, however, than with sulfonylurea (approximately 4 kg) or insulin (approximately 6 kg).²³ Metformin also significantly improved levels of total cholesterol, low-density lipoprotein (LDL) cholesterol, and triglycerides when compared with glyburide or placebo.²²

Risk of lactic acidosis. Lactic acidosis associated with metformin is a rare but serious adverse event, with an estimated prevalence of 3 cases per 100,000.²⁶ The product labeling notes most of these cases have occurred among patients with significant renal insufficiency, including both intrinsic renal disease and renal hypoperfusion. Absolute contraindications include renal disease (serum creatinine ≥1.5 mg/dL [males] and ≥1.4 mg/dL [females]), congestive heart failure requiring pharmacological treatment, and acute or chronic metabolic acidosis. It should also be discontinued at the time of radiologic studies using intravascular iodinated contrast materials.

Additional “precautionary conditions” include age ≥80 years (unless measurement of creatinine clearance demonstrates that renal function is not reduced), hepatic disease, cationic drug use, conditions associated with hypoxia (eg, chronic obstructive pulmonary disease [COPD], acute myocardial infarction, dehydration, sepsis), excessive alcohol intake, and surgery, until patient’s oral intake is resumed.

Is the risk overstated? Despite these extensive precautions, published studies show that metformin is commonly prescribed to patients with absolute contraindications.^27,28 One recent study observed that 11.2% of Medicare beneficiaries hospitalized with congestive heart failure and concomitant diabetes were treated with metformin.²⁸ In the absence of advanced renal dysfunction, metformin rarely accumulates in the body,²⁹ and accumulation of metformin is rarely reported as a cause of lactic acidosis.^30,31 Rather, tissue hypoxia acts as a trigger in most cases. Metformin should therefore be discontinued whenever tissue hypoxia is suspected.³¹

A recent systematic review and meta analysis found no evidence that metformin was associated with an increased risk of lactic acidosis if the drug was prescribed under study conditions, taking into account contraindications.³² Refinement and clarification of the risk for lactic acidosis in these various populations is needed, to ensure optimal patient safety and to further assess this highly effective medication.

Common adverse effects associated with metformin are diarrhea and nausea, which can be minimized by administering the drug with meals and slowly titrating the dose, or perhaps by using the extended-release formulation.

Thiazolidinediones

Thiazolidinediones (TZDs) include rosiglitazone and pioglitazone. These agents, like metformin, do not increase insulin secretion but depend on the presence of insulin for their activity. TZDs are agonists at peroxisome-proliferator-activated receptor gamma (PPAR-γ) receptors in peripheral tissues such as skeletal muscle, where they increase glucose uptake.¹⁵ Thus, their predominant effect is to decrease insulin resistance.

TZDs have similar antihyperglycemic efficacy as sulfonylureas or metformin. They decrease A_1c levels by 0.6–1.9 percentage points and lower fasting plasma glucose levels by 50–80 mg/dL.¹⁵ They have a slower onset of action compared with other hypoglycemic drugs, and intervals of 3 to 4 weeks should be allowed between doses before increasing the dosage. TZDs also have favorable effects on lipid levels: HDL concentrations increase and triglyceride concentrations decrease with their use.³³ It is not known whether they decrease macrovascular or microvascular complications, although such studies are underway.

Adverse effects. TZDs are typically well tolerated, though weight gain of 1–3 kg, edema (4%–5%) and anemia (1%–2%) can occur. Weight gain and edema are more pronounced when TZDs are used in combination with insulin. Anemia is likely due to increased plasma volume rather than any significant hematological effect.

Due to adverse events related to volume expansion, TZDs are not recommended for patients with New York Heart Association class III or IV heart failure. A recent consensus statement from the American Heart Association and the ADA stresses that before administering TZD treatment, the physician should explore the possible presence of cardiac disease, use of other drugs that cause fluid retention, and the pathogenesis of any existing edema or dyspnea.³⁴

Although troglitazone was removed from the market due to its association with hepatocellular injury, pioglitazone and rosiglitazone are not as convincingly associated with liver injury.¹⁵ In preapproval clinical studies, less than 0.5% of patients treated with rosiglitzone and pioglitazone had elevations in alanine transaminase (ALT) >3 times the upper limit of normal.

The incidence of hepatitis or acute liver failure from troglitazone was compared with rosiglitazone, pioglitazone, metformin, and glyburide, by analysis of spontaneously reported adverse events to the Food and Drug Administration (FDA) MEDWATCH database during the first 15 months of marketing of each drug.^35,36 The incidence of hepatitis per million prescriptions was 21.5, 14.7, 9.4, 2.9, and 4.1, respectively, while the incidence of acute liver failure per 100,000 prescriptions was 4.6, 0.9, 0.8, 0.2, and 0. It appears that postmarketing data support preclinical studies, in that the incidence of acute liver failure is an order of magnitude higher for troglitazone vs. other TZDs.³⁵ However, the FDA recommends avoiding their use in patients with baseline ALT levels >2.5 times the upper limit of normal. The FDA recently reduced the recommended frequency for ALT monitoring for pioglitazone (and is currently considering the same for rosiglitazone). Serum ALT is recommended prior to initiation and then periodically thereafter.

Cost. TZDs are expensive relative to other hypoglycemic agents.

α-glucosidase inhibitors

The α-glucosidase inhibitors (AGIs), acarbose and miglitol, act through competitive, reversible inhibition of membrane-bound intestinal α-glucosidase, which hydrolyzes complex carbohydrates to glucose and other monosaccharides. This inhibition delays glucose absorption and decreases postprandial hyperglycemia.³⁷ Thus, they have a nonsystemic mechanism of action.

These agents cause a modest reduction in the A_1c level (0.5–1.0 percentage points) and are thus less effective than sulfonylureas, metformin, or TZDs. They do not reduce fasting plasma glucose levels, but reduce postprandial hyperglycemia by 50 mg/dL.³⁸ No long-term studies have evaluated whether AGIs reduce diabetes-related macrovascular or microvascular outcomes.

Adverse effects. While AGIs are virtually free of serious toxicities, patient tolerability can be a problem due to adverse gastrointestinal effects. In indirect comparisons from placebo-controlled trials, patients treated with miglitol and acarbose commonly reported abdominal pain (11.7%, 19%), diarrhea (28.7%, 31%), and flatulence (41.5%, 74%), respectively. Systemic accumulation of AGIs has been shown to increase in proportion to the degree of renal insufficiency, and their use is not recommended for patients with serum creatinine >2.0 mg/dL. However, whether such patients are at greater risk of any toxicity is unknown. Acarbose at doses above 100 mg 3 times daily has been associated with elevated serum transaminase levels; however, this risk appears negligible at standard doses.

Insulin

Insulin is the oldest therapy for diabetes, and it has no upper dose limit.³⁹ It increases insulin levels and can reduce A_1c levels by 1.5 to 2.5 percentage points. Though half of diabetes patients need insulin eventually for optimal control, historically it has been introduced late in the disease process unless patients have severe hyperglycemia (fasting blood sugar >350 mg/dL) or ketonuria.³⁸ However, it is effective in gaining initial control, decreasing gluconeogenesis and increasing glucose uptake. Disadvantages are weight gain, hypoglycemia, and patient reluctance to give injections.

When insulin is indicated. Patients who exhibit persistent hyperglycemia despite oral hypoglycemic therapy may stop the oral drug(s) and begin insulin. By combining insulin with oral therapy, lower insulin doses may be used to achieve desired control vs using insulin alone.⁴⁰ For some patients a basal supplement of insulin may be sufficient and can be given as a single dose at bedtime, without an oral hypoglycemic drug.⁴¹

Insulin regimens. Various insulin regimens are available: very rapid acting (lispro and aspart), rapid acting (regular), intermediate acting (isophane insulin [NPH] and lente) and very long acting (ultralente and glargine). Glargine insulin (Lantus) has more predictable absorption than NPH, lente, and ultralente. Lantus, compared with NPH, has been associated with less nocturnal and postprandial hypoglycemia.^38,42,43 This is consistent with the peakless and longer duration of glargine compared with NPH.⁴⁴ A recent randomized controlled trial demonstrated that morning insulin glargine lowered A_1c levels more than a bedtime dose of NPH (–1.24 vs –0.84; 95% CI, 0.23%–0.58%) or a bedtime dose of glargine (–1.24 vs –0.96%; 95% CI, 0.11%–0.46%).⁴⁵ Glargine’s only relative disadvantage is increased cost.

Combination products. Combination insulin options are 70 NPH/30 regular, 50 NPH/50 regular, and 75 lispro protamine/25 lispro. Many combinations of insulin regimens have been used successfully. The typical range of insulin needed for monotherapy is 0.4–1 U/kg/d. Once-daily injection of intermediate acting or long acting insulins at bedtime or before breakfast, once-daily or twice-daily combinations of intermediate and rapid acting insulins, and more complex regimens have been used to good effect.

Using prandial insulin at each meal with separate basal insulin adds flexibility to meal times and doses administered.⁴³ With multiple-dose intensive insulin therapy, a basal dose suppresses hepatic glucose output and the bolus doses enhance postprandial glucose uptake. This intensive insulin treatment reduces mortality among critically ill patients in surgical intensive care units and for those with acute myocardial infarction.^46,47 An algorithm for using progressive therapy in diabetes mellitus is shown in Figure 2.⁴⁸

FIGURE 2
ADA recommendations for the treatment of type 2 diabetes

Combination Therapy

Over time glycemic control becomes more difficult, even with maximum monotherapy for patients with healthy lifestyles. It was shown in UKPDS 49 that monotherapy with sulfonylurea, metformin, or insulin eventually fails in most cases—by 3 years after diagnosis, about 50% of patients need more than monotherapy; 75% by 9 years.⁴⁹ In UKPDS 33, the median A_1c level increased steadily over 10 years with both conventional therapy and intensive therapy (Figure 3).²

Several options are available when monotherapy fails. Based on expert opinion, the principle is to combine drugs with different mechanisms of action to achieve an additive effect for glycemic control. Combination products may simplify the treatment regimen and improve adherence. In many instances, they may also cost less.⁵⁰

Successful combinations. The combination of sulfonylurea and metformin has proven effective in many studies.^22,51,52 One study showed that initial treatment with glyburide/metformin improved glycemic control better than either glyburide or metformin monotherapy (SOR: A).^53,54 The addition of the non-SU secretagogues repaglinide and nateglinide to metformin significantly improved glycemic control, with repaglinide showing superiority over nateglinide.⁵⁵ A TZD added to a sulfonylurea has also significantly improved A_1c and fasting blood sugar results.⁵⁶ Patients whose diabetes was inadequately controlled with diet alone or diet plus a sulfonylurea showed improvement with the addition of the AGI miglitol, compared with addition of placebo.⁵⁷ The AGI acarbose has shown to be an effective addition to diet, metformin, sulfonylurea, and insulin.⁵⁸ A TZD added to metformin has also been shown to improve glycemic control.⁵⁹ A non-SU added to patients inadequately controlled with a TZD has also been effective.⁶⁰

The early addition of insulin when maximal sulfonylurea therapy is inadequate has been effective.^61-63 When introducing insulin, a nighttime regimen of NPH or glargine, 10 units at bedtime, is an appropriate dose (SOR: C). This is easier and less costly than often assumed, and helps improve glycemic control.⁶⁴ Most patients require combination therapy as their disease progresses.³⁹

FIGURE 3
Glycemic control in type 2 DM

Improving Outcomes

Cumulative survey data reveal a wide gap between guideline recommendations and the care patients receive.⁶⁵ One study showed that physicians initiated treatment changes only after the A_1c level had reached 9.0% or higher instead of the 8.0% level recommended by ADA.⁶⁶ How can the quality of management be improved?

In private practices and institutions, many interventions have been shown to improve outcomes in diabetes mellitus. Education measures work, and they include chart audits, reminder cards, pharmacist collaboration, flow sheets, and nursing initiatives.^67,68 Effective disease-management programs have also used clinical guidelines, outcomes reporting, coverage of glucose meters and strips, and the support of clinical leadership.⁶⁹

Computerized systems that track patients and recommended laboratory tests have improved screening rates and glycemic and blood pressure control.⁷⁰ Monitoring patients’ readiness to change has allowed targeted education to improve A_1c levels.⁷¹ Continuity of care has also improved the quality of disease control by increasing adherence to recommended tests and exams.⁷²

Acknowlegments

The authors thank Marie Hamer, RN, for her continuous diabetes quality improvement efforts and Jean Camarata for her editorial and reference acquisition assistance.

Corresponding author
John E. Sutherland, MD, Northeast Iowa Family Practice Residency Program, University of Iowa College of Medicine, 2055 Kimball Avenue, Waterloo, Iowa 50702. E-mail: [email protected].

Diabetes need not automatically sentence patients to the well-known ravages of the disease. Evidence-supported preventive strategies can forestall complications.

Increasing evidence suggests diabetes can be prevented by a combination of lifestyle changes and medications. Key interventions for patients with established type 2 diabetes include tight control of blood glucose levels, reduction of blood pressure and lipid levels, and early identification of diabetes-related neuropathy, nephropathy, and retinopathy. Antiplatelet therapy may also be beneficial.

Primary prevention: should intervention begin with prediabetes states?

Treatment of risk factors for diabetes (eg, obesity) and management of prediabetes (eg, impaired glucose tolerance) has suggested early intervention can forestall the development of type 2 diabetes (Table 1). Three trials addressed intensive lifestyle/diet modification. Though the onset of clinically diagnosed diabetes was delayed while patients adhered to these strategies, long-term studies of lifestylemodification have not been performed.^1–3

Several other investigations have looked at whether prescription medications achieve similar benefit. Metformin,³ acarbose,⁴ and orlistat⁵ have been studied in populations with prediabetes. These interventions appear to delay the onset of diabetes (number needed to treat [NNT]=33–88 patient-years) and to reduce blood sugar levels by 5% to 10%. A post-hoc analysis of the Heart Outcomes Prevention Evaluation (HOPE) trial⁶ also suggests a favorable effect with the ACE inhibitor, ramipril (NNT=250 patient-years). However, patient-oriented outcomes—development of microvascular or macrovascular disease—were not assessed in these trials. Consequently, the ultimate benefit of treating prediabetes states remains uncertain.

SECONDARY PREVENTION: DOES EARLY DETECTION OF DIABETES HELP DELAY COMPLICATIONS?

No randomized trial of screening has reported any patient-oriented benefits. However, based on consensus opinion, the American Diabetes Association (ADA) recommends screening every person aged 45 years and older every 3 years (strength of recommendation [SOR]: C).⁷ The characteristics of a clinically recognized disease like diabetes, however, may differ significantly from the characteristics of the subclinical states that would be recognized with screening. Therefore, though the US Preventive Health Services Task Force⁸ has concluded there is no evidence to recommend screening average risk individuals for diabetes, it does recommend screening individuals at increased risk of macrovascular changes (eg, those with hypertension) (SOR: B). This is based in part on indirect evidence that tighter blood pressure targets may be beneficial for patients with diabetes. (See the Clinical Inquiry, “Does screening for diabetes in at-risk patients improve long-term outcomes?”)

Tertiary prevention: preventing complications of existing diabetes

Patient-oriented outcomes in diabetes can be significantly improved with numerous interventions (Table 2).

Tight glycemic control warranted
The United Kingdom Prospective Diabetes Study¹⁴ (UKPDS) randomized participants to usual diabetic care or intensive glycemic control with insulin or sulfonylureas over 10 years. Intensive control reduced average hemoglobin A_1c from 7.9% to 7.0%; it also reduced aggregate microvascular complications, mainly a relative 39% decreased need for photocoagulation for diabetic retinopathy (NNT=320 patient-years). No other individual endpoint was independently affected.

The UKPDS suggested a trend toward 16% fewer relative MIs in the intensive control group; however, the results did not reach statistical significance (P=.052). In addition an increase in major hypoglycemic episodes was noted, worse with insulin than sulfonylureas (relative risk [RR]=257%; number needed to harm [NNH]=1110 patient-years).

Approach to obese patients. The UKPDS also studied intensive control with metformin among a subgroup of obese patients with diabetes (>120% ideal body weight).⁹ Metformin lowered average hemoglobin A_1c only from 8.0% to 7.4%, but reduced relative total mortality by 36% (NNT=142 patient-years). Metformin also reduced the relative chance of MI by 39% (NNT=143 patient-years).

These benefits were reversed, however, in a separate UKPDS subgroup placed first on a sulfonylurea, then receiving metformin if glycemic control was inadequate. Total mortality was relatively increased by 60% (NNH=89 patient-years).⁹ This adverse outcome disappeared when adjustment was made for age, sex, ethnic group, and glycemic control.

Other potential risks of metformin include lactic acidosis, but 1 recent systematic review found no associated cases of this condition with metformin prescribed in published studies.¹⁷

Applying the evidence. Initial treatment of patients with diabetes over 120% of ideal body weight should include tight glucose control with metformin, unless contraindicated (SOR: A). For leaner patients, therapy with a sulfonylurea or insulin is supported by evidence (SOR: B).

Control blood pressure to below 150/80 mm Hg
The UKPDS also compared tight blood pressure control (aiming at systolic <150 mm Hg and diastolic <80) with usual treatment. Tight control reduced relative deaths attributed to diabetes by 32% (NNT=150 patient-years), and demonstrated a trend toward reduced total mortality that was not statistically significant (relative risk reduction [RRR]=18%; 95% confidence interval [CI], –0.8% to 37%). Both stroke (NNT=196 patient-years) and the aggregate endpoint of “all microvascular disease” (NNT=139 patient-years) were significantly reduced by tight blood pressure control.10

The Hypertension Optimal Treatment (HOT) Study Group11 compared various levels of blood pressure control. A reduction in cardiovascular mortality was significant even between those treated with a goal of 80 and 85 mm Hg diastolic (RRR=67%; NNT=133 patient-years). The trend toward reduction in total mortality in this trial also approached statistical significance (P=.068; RRR=42%).

The Syst-Eur trial13 compared tight blood pressure control on nitrendipine with looser control with a variety of agents. Cardiovascular mortality was reduced by 30% with tight control (NNT=100 patient-years).

The Heart Outcomes Prevention Evaluation (HOPE) study6 included patients with diabetes, of which only 56% had a diagnosis of hypertension. It randomized participants to receive ramipril or placebo, in addition to any off-study antihypertensive agents they were already using. The intervention group had both lower average blood pressures as well as lower total mortality by 24% (NNT=140 patient-years).

Choice of agent important. Lower blood pressure goals have consistently demonstrated benefit across multiple studies. The choice of antihypertensive agent may also affect outcomes (Table 3). The UKPDS blood pressure analysis was also stratified to evaluate whether the results of blood pressure treatment differed between captopril and atenolol. Though compliance was slightly better with captopril, there were no differences in patient-oriented outcomes between the groups.²⁶

The Captopril Prevention Project (CAPPP)²⁷ compared captopril with diuretics and betablockers alone or in combination. While blood pressure control was about the same in all study groups, the captopril group realized a 66% reduction in MI (NNT=96 patient-years) and a total mortality 46% less than that seen with the other agents (NNT=96 patient-years).

The Appropriate Blood Pressure Control in Diabetes (ABCD) trial²⁸ compared nisoldipine and enalapril. Participants randomized to receive the ACE inhibitor had an 80% decreased risk of MI (NNT=25 patient-years).

In the Fosinopril versus Amlodipine Cardiovascular Events Trial (FACET),²⁹ blood pressure was better controlled with amlodipine, but major vascular events were 51% fewer with the ACE inhibitor (NNT=146 patient-years), again supporting the superior performance of ACE inhibitors.

Angiotensin receptor blockers (ARBs) may have comparable effects to ACE inhibitors. The Losartan Intervention For Endpoint Reduction in Hypertension (LIFE) trial³⁰ studied patients with left ventricular hypertrophy, diabetes, and hypertension, comparing losartan with atenolol. Total mortality with losartan was reduced by a relative 40% compared with atenolol (NNT=167 patientyears). Another study³¹ of patients with diabetes, hypertension, and nephropathy compared irbesartan with amlodipine. This trial demonstrated no differences in patient-oriented outcomes between the calcium-channel blocker and the ARB.

Applying the evidence. The goals for blood pressure control in type 2 diabetes should be less than 150 mm Hg systolic and 80 mm Hg diastolic (SOR: A). Evidence also strongly supports the use of an ACE inhibitor, or possibly ARB, as first-linetreatment for hypertension in diabetes (SOR: A). Many authorities recommend even more aggressive blood pressure goals.

Lipid management: statins improve outcome
Lowering elevated triglycerides has not been independently associated with an improvement in patient-oriented outcomes. In the Helsinki Heart Study²² and the St. Mary’s, Ealing, Northwick Park Diabetes Cardiovascular Disease Prevention (SENDCAP) trial,³² a fibric acid derivative was compared with placebo. The average triglyceride concentration was decreased, but no significant effect on coronary events was noted. However, elevated triglycerides are associated with the metabolic syndrome, which may warrant lifestyle changes or medication (based on expert opinion).

Treatment with hydroxymethyl glutaryl coenzyme A reductase inhibitors (statins) has better supporting evidence. In the Scandinavia Simvastatin Survival Study (4S),¹⁵ 202 patients with diabetes, coronary artery disease, and elevated cholesterol were randomized to receive simvastatin or placebo. Major coronary events were reduced by 55% with simvastatin (NNT=22.5 patient-years). Total mortality was reduced, but not to a statistically significant extent (P=.087; RRR=43%).

The Air Force/Texas Coronary Atherosclerosis Prevention Study³³ (AFCAPS/TexCAPS) compared lovastatin with placebo for patients with diabetes and healthy individuals with normal cholesterol levels. Though the overall population demonstrated a 37% reduction in first coronary events, the diabetes subgroup had insufficient power to confirm this trend independently.

The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial³⁴ (ALLHAT-LLT) randomized 3638 patients with diabetes to receive pravastatin 40 mg or placebo. No changes were seen in total mortality or cardiovascular events. These results may have been confounded by off-study prescription for statins, given to 32% of those randomized to the placebo group.

The best independent evidence for use of statins in diabetes comes from the Heart Protection Study¹⁶ of 3982 patients with diabetes with a total cholesterol level >135 mg/dL and no evidence of coronary disease. They were randomized to receive 40 mg simvastatin or placebo for 5 years. First major vascular events (MI or stroke) were decreased in the simvastatin group by 26% over 5 years (NNT=104 patient-years).

Applying the evidence. Reducing elevated triglycerides in type 2 diabetes is not supported by clear evidence, although elevated triglycerides may be associated with the metabolic syndrome and may warrant lifestyle change (SOR: C).

However, statins—even for those with a normal cholesterol level—reduce macrovascular outcomes (SOR: A), a measure now endorsed by the American College of Physicians.

Anti-platelet therapy
Aspirin prophylaxis in diabetes has weaker support. The Early Treatment Diabetic Retinopathy Study³⁵ (ETDRS) randomized 3711 patients with diabetes to receive aspirin 650 mg or placebo. No benefit in mortality or cardiovascular event rates was documented after 5 years. However, the results did suggest a nonsignificant trend toward reduction of MI (RRR=17%; 95% CI, –4% to 34%; NNT=333).

In the Physician’s Health Study,³⁶ 22,071 participants (most of whom did not have diabetes) were randomized to receive aspirin 325 mg every other day or placebo. MI was reduced by 44% with aspirin (NNT=500). The risk of MI in the subgroup of 533 individuals with diabetes paralleled this reduction (RRR=39%), though, independently, the reduction in the subgroup did not reach statistical significance. Bleeding problems were the most common adverse effect of the aspirin, and were increased by 32% (NNH=78 person-years).

In the HOT study,¹¹ a subgroup of 1501 patients with diabetes was randomized to receive aspirin 75 mg or placebo daily. The rate of MI trended downward with aspirin but was not statistically significant. Among the 18,790 patients in this study (most of whom did not have diabetes), MI was 36% less likely to occur with aspirin (NNT=770).

Applying the evidence. Overall, there is some suggestion that persons with cardiac risk factors, like type 2 diabetes, benefit by taking low-dose aspirin to avoid macrovascular complications. No study has confirmed this in a population of individuals with diabetes, but the trends suggest a possible benefit (SOR: C). The decision to use aspirin should be made in consultation with an informed patient.

SCREENING FOR NEUROPATHY

One trial¹⁷ has been conducted on screening for neuropathy. It compared monofilament testing and palpation of pedal pulses with “no special care.” Patients with an original positive screen result received a calculation of their ankle-brachial index, foot x-rays, and other measurements, and were referred to a high-risk podiatry clinic if the second level testing showed abnormal results. Major amputations were decreased in the screened group by 92% over 2 years (NNT=180 patient-years). This evidence of benefit is sufficient to recommend screening patients with diabetes for peripheral neuropathy or peripheral vascular disease, and appropriate referral (SOR: B). (See the Clinical Inquiry, “What is the best treatment for diabetic neuropathy?”)

Screening for nephropathy

One systematic review³⁷ found no randomized trials of screening for urinary microalbumin and how it might affect overt nephropathy. However, several studies have looked at treatment of gross proteinuria, and have demonstrated some benefit in patient-oriented outcomes. Improved blood pressure control has reduced progression of nephropathy to end-stage renal disease.^38,39 Other studies have suggested that an ACE inhibitor or ARB might provide benefit.^30,39

The ADA recommends annual screening for microalbumin based on expert consensus (SOR: C).⁷ Though early detection of microalbumin might optimize the treatment of nephropathy, it is also possible that screening may detect a population whose disease would have remained subclinical indefinitely. No clear evidence suggests that screening for microalbumin reduces the incidence of patient-oriented outcomes.

Screening for retinopathy

Several trials have evaluated interventions in the diagnosis and prevention of visual loss. As discussed previously, intensive glucose control with sulfonylureas or insulin reduced the need for retinal photocoagulation by 39% in the UKPDS (NNT=320 patient-years).¹⁴ Similarly in this trial, tight blood pressure control reduced the progression of retinopathy, compared with usual care. ¹⁰

Specific treatments for retinopathy include photocoagulation, which has been demonstrated to significantly reduce severe visual loss by 58% (NNT=30 patient-years).⁴⁰ One cohort study suggests that screening for retinopathy coupled with appropriate treatment may reduce the onset of visual loss.⁴¹ The ADA recommends annual screening for retinopathy with a dilated eye exam based on indirect evidence of its benefit (SOR: B).⁷

Correspondence
Paul Ullom-Minnich, MD, Partners in Family Care, PO Box 640, Moundridge, KS 67107. E-mail: [email protected].

Diabetes need not automatically sentence patients to the well-known ravages of the disease. Evidence-supported preventive strategies can forestall complications.

Increasing evidence suggests diabetes can be prevented by a combination of lifestyle changes and medications. Key interventions for patients with established type 2 diabetes include tight control of blood glucose levels, reduction of blood pressure and lipid levels, and early identification of diabetes-related neuropathy, nephropathy, and retinopathy. Antiplatelet therapy may also be beneficial.

Primary prevention: should intervention begin with prediabetes states?

Treatment of risk factors for diabetes (eg, obesity) and management of prediabetes (eg, impaired glucose tolerance) has suggested early intervention can forestall the development of type 2 diabetes (Table 1). Three trials addressed intensive lifestyle/diet modification. Though the onset of clinically diagnosed diabetes was delayed while patients adhered to these strategies, long-term studies of lifestylemodification have not been performed.^1–3

Several other investigations have looked at whether prescription medications achieve similar benefit. Metformin,³ acarbose,⁴ and orlistat⁵ have been studied in populations with prediabetes. These interventions appear to delay the onset of diabetes (number needed to treat [NNT]=33–88 patient-years) and to reduce blood sugar levels by 5% to 10%. A post-hoc analysis of the Heart Outcomes Prevention Evaluation (HOPE) trial⁶ also suggests a favorable effect with the ACE inhibitor, ramipril (NNT=250 patient-years). However, patient-oriented outcomes—development of microvascular or macrovascular disease—were not assessed in these trials. Consequently, the ultimate benefit of treating prediabetes states remains uncertain.

SECONDARY PREVENTION: DOES EARLY DETECTION OF DIABETES HELP DELAY COMPLICATIONS?

No randomized trial of screening has reported any patient-oriented benefits. However, based on consensus opinion, the American Diabetes Association (ADA) recommends screening every person aged 45 years and older every 3 years (strength of recommendation [SOR]: C).⁷ The characteristics of a clinically recognized disease like diabetes, however, may differ significantly from the characteristics of the subclinical states that would be recognized with screening. Therefore, though the US Preventive Health Services Task Force⁸ has concluded there is no evidence to recommend screening average risk individuals for diabetes, it does recommend screening individuals at increased risk of macrovascular changes (eg, those with hypertension) (SOR: B). This is based in part on indirect evidence that tighter blood pressure targets may be beneficial for patients with diabetes. (See the Clinical Inquiry, “Does screening for diabetes in at-risk patients improve long-term outcomes?”)

Tertiary prevention: preventing complications of existing diabetes

Patient-oriented outcomes in diabetes can be significantly improved with numerous interventions (Table 2).

Tight glycemic control warranted
The United Kingdom Prospective Diabetes Study¹⁴ (UKPDS) randomized participants to usual diabetic care or intensive glycemic control with insulin or sulfonylureas over 10 years. Intensive control reduced average hemoglobin A_1c from 7.9% to 7.0%; it also reduced aggregate microvascular complications, mainly a relative 39% decreased need for photocoagulation for diabetic retinopathy (NNT=320 patient-years). No other individual endpoint was independently affected.

The UKPDS suggested a trend toward 16% fewer relative MIs in the intensive control group; however, the results did not reach statistical significance (P=.052). In addition an increase in major hypoglycemic episodes was noted, worse with insulin than sulfonylureas (relative risk [RR]=257%; number needed to harm [NNH]=1110 patient-years).

Approach to obese patients. The UKPDS also studied intensive control with metformin among a subgroup of obese patients with diabetes (>120% ideal body weight).⁹ Metformin lowered average hemoglobin A_1c only from 8.0% to 7.4%, but reduced relative total mortality by 36% (NNT=142 patient-years). Metformin also reduced the relative chance of MI by 39% (NNT=143 patient-years).

These benefits were reversed, however, in a separate UKPDS subgroup placed first on a sulfonylurea, then receiving metformin if glycemic control was inadequate. Total mortality was relatively increased by 60% (NNH=89 patient-years).⁹ This adverse outcome disappeared when adjustment was made for age, sex, ethnic group, and glycemic control.

Other potential risks of metformin include lactic acidosis, but 1 recent systematic review found no associated cases of this condition with metformin prescribed in published studies.¹⁷

Applying the evidence. Initial treatment of patients with diabetes over 120% of ideal body weight should include tight glucose control with metformin, unless contraindicated (SOR: A). For leaner patients, therapy with a sulfonylurea or insulin is supported by evidence (SOR: B).

Control blood pressure to below 150/80 mm Hg
The UKPDS also compared tight blood pressure control (aiming at systolic <150 mm Hg and diastolic <80) with usual treatment. Tight control reduced relative deaths attributed to diabetes by 32% (NNT=150 patient-years), and demonstrated a trend toward reduced total mortality that was not statistically significant (relative risk reduction [RRR]=18%; 95% confidence interval [CI], –0.8% to 37%). Both stroke (NNT=196 patient-years) and the aggregate endpoint of “all microvascular disease” (NNT=139 patient-years) were significantly reduced by tight blood pressure control.10

The Hypertension Optimal Treatment (HOT) Study Group11 compared various levels of blood pressure control. A reduction in cardiovascular mortality was significant even between those treated with a goal of 80 and 85 mm Hg diastolic (RRR=67%; NNT=133 patient-years). The trend toward reduction in total mortality in this trial also approached statistical significance (P=.068; RRR=42%).

The Syst-Eur trial13 compared tight blood pressure control on nitrendipine with looser control with a variety of agents. Cardiovascular mortality was reduced by 30% with tight control (NNT=100 patient-years).

The Heart Outcomes Prevention Evaluation (HOPE) study6 included patients with diabetes, of which only 56% had a diagnosis of hypertension. It randomized participants to receive ramipril or placebo, in addition to any off-study antihypertensive agents they were already using. The intervention group had both lower average blood pressures as well as lower total mortality by 24% (NNT=140 patient-years).

Choice of agent important. Lower blood pressure goals have consistently demonstrated benefit across multiple studies. The choice of antihypertensive agent may also affect outcomes (Table 3). The UKPDS blood pressure analysis was also stratified to evaluate whether the results of blood pressure treatment differed between captopril and atenolol. Though compliance was slightly better with captopril, there were no differences in patient-oriented outcomes between the groups.²⁶

The Captopril Prevention Project (CAPPP)²⁷ compared captopril with diuretics and betablockers alone or in combination. While blood pressure control was about the same in all study groups, the captopril group realized a 66% reduction in MI (NNT=96 patient-years) and a total mortality 46% less than that seen with the other agents (NNT=96 patient-years).

The Appropriate Blood Pressure Control in Diabetes (ABCD) trial²⁸ compared nisoldipine and enalapril. Participants randomized to receive the ACE inhibitor had an 80% decreased risk of MI (NNT=25 patient-years).

In the Fosinopril versus Amlodipine Cardiovascular Events Trial (FACET),²⁹ blood pressure was better controlled with amlodipine, but major vascular events were 51% fewer with the ACE inhibitor (NNT=146 patient-years), again supporting the superior performance of ACE inhibitors.

Angiotensin receptor blockers (ARBs) may have comparable effects to ACE inhibitors. The Losartan Intervention For Endpoint Reduction in Hypertension (LIFE) trial³⁰ studied patients with left ventricular hypertrophy, diabetes, and hypertension, comparing losartan with atenolol. Total mortality with losartan was reduced by a relative 40% compared with atenolol (NNT=167 patientyears). Another study³¹ of patients with diabetes, hypertension, and nephropathy compared irbesartan with amlodipine. This trial demonstrated no differences in patient-oriented outcomes between the calcium-channel blocker and the ARB.

Applying the evidence. The goals for blood pressure control in type 2 diabetes should be less than 150 mm Hg systolic and 80 mm Hg diastolic (SOR: A). Evidence also strongly supports the use of an ACE inhibitor, or possibly ARB, as first-linetreatment for hypertension in diabetes (SOR: A). Many authorities recommend even more aggressive blood pressure goals.

Lipid management: statins improve outcome
Lowering elevated triglycerides has not been independently associated with an improvement in patient-oriented outcomes. In the Helsinki Heart Study²² and the St. Mary’s, Ealing, Northwick Park Diabetes Cardiovascular Disease Prevention (SENDCAP) trial,³² a fibric acid derivative was compared with placebo. The average triglyceride concentration was decreased, but no significant effect on coronary events was noted. However, elevated triglycerides are associated with the metabolic syndrome, which may warrant lifestyle changes or medication (based on expert opinion).

Treatment with hydroxymethyl glutaryl coenzyme A reductase inhibitors (statins) has better supporting evidence. In the Scandinavia Simvastatin Survival Study (4S),¹⁵ 202 patients with diabetes, coronary artery disease, and elevated cholesterol were randomized to receive simvastatin or placebo. Major coronary events were reduced by 55% with simvastatin (NNT=22.5 patient-years). Total mortality was reduced, but not to a statistically significant extent (P=.087; RRR=43%).

The Air Force/Texas Coronary Atherosclerosis Prevention Study³³ (AFCAPS/TexCAPS) compared lovastatin with placebo for patients with diabetes and healthy individuals with normal cholesterol levels. Though the overall population demonstrated a 37% reduction in first coronary events, the diabetes subgroup had insufficient power to confirm this trend independently.

The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial³⁴ (ALLHAT-LLT) randomized 3638 patients with diabetes to receive pravastatin 40 mg or placebo. No changes were seen in total mortality or cardiovascular events. These results may have been confounded by off-study prescription for statins, given to 32% of those randomized to the placebo group.

The best independent evidence for use of statins in diabetes comes from the Heart Protection Study¹⁶ of 3982 patients with diabetes with a total cholesterol level >135 mg/dL and no evidence of coronary disease. They were randomized to receive 40 mg simvastatin or placebo for 5 years. First major vascular events (MI or stroke) were decreased in the simvastatin group by 26% over 5 years (NNT=104 patient-years).

Applying the evidence. Reducing elevated triglycerides in type 2 diabetes is not supported by clear evidence, although elevated triglycerides may be associated with the metabolic syndrome and may warrant lifestyle change (SOR: C).

However, statins—even for those with a normal cholesterol level—reduce macrovascular outcomes (SOR: A), a measure now endorsed by the American College of Physicians.

Anti-platelet therapy
Aspirin prophylaxis in diabetes has weaker support. The Early Treatment Diabetic Retinopathy Study³⁵ (ETDRS) randomized 3711 patients with diabetes to receive aspirin 650 mg or placebo. No benefit in mortality or cardiovascular event rates was documented after 5 years. However, the results did suggest a nonsignificant trend toward reduction of MI (RRR=17%; 95% CI, –4% to 34%; NNT=333).

In the Physician’s Health Study,³⁶ 22,071 participants (most of whom did not have diabetes) were randomized to receive aspirin 325 mg every other day or placebo. MI was reduced by 44% with aspirin (NNT=500). The risk of MI in the subgroup of 533 individuals with diabetes paralleled this reduction (RRR=39%), though, independently, the reduction in the subgroup did not reach statistical significance. Bleeding problems were the most common adverse effect of the aspirin, and were increased by 32% (NNH=78 person-years).

In the HOT study,¹¹ a subgroup of 1501 patients with diabetes was randomized to receive aspirin 75 mg or placebo daily. The rate of MI trended downward with aspirin but was not statistically significant. Among the 18,790 patients in this study (most of whom did not have diabetes), MI was 36% less likely to occur with aspirin (NNT=770).

Applying the evidence. Overall, there is some suggestion that persons with cardiac risk factors, like type 2 diabetes, benefit by taking low-dose aspirin to avoid macrovascular complications. No study has confirmed this in a population of individuals with diabetes, but the trends suggest a possible benefit (SOR: C). The decision to use aspirin should be made in consultation with an informed patient.

SCREENING FOR NEUROPATHY

One trial¹⁷ has been conducted on screening for neuropathy. It compared monofilament testing and palpation of pedal pulses with “no special care.” Patients with an original positive screen result received a calculation of their ankle-brachial index, foot x-rays, and other measurements, and were referred to a high-risk podiatry clinic if the second level testing showed abnormal results. Major amputations were decreased in the screened group by 92% over 2 years (NNT=180 patient-years). This evidence of benefit is sufficient to recommend screening patients with diabetes for peripheral neuropathy or peripheral vascular disease, and appropriate referral (SOR: B). (See the Clinical Inquiry, “What is the best treatment for diabetic neuropathy?”)

Screening for nephropathy

One systematic review³⁷ found no randomized trials of screening for urinary microalbumin and how it might affect overt nephropathy. However, several studies have looked at treatment of gross proteinuria, and have demonstrated some benefit in patient-oriented outcomes. Improved blood pressure control has reduced progression of nephropathy to end-stage renal disease.^38,39 Other studies have suggested that an ACE inhibitor or ARB might provide benefit.^30,39

The ADA recommends annual screening for microalbumin based on expert consensus (SOR: C).⁷ Though early detection of microalbumin might optimize the treatment of nephropathy, it is also possible that screening may detect a population whose disease would have remained subclinical indefinitely. No clear evidence suggests that screening for microalbumin reduces the incidence of patient-oriented outcomes.

Screening for retinopathy

Several trials have evaluated interventions in the diagnosis and prevention of visual loss. As discussed previously, intensive glucose control with sulfonylureas or insulin reduced the need for retinal photocoagulation by 39% in the UKPDS (NNT=320 patient-years).¹⁴ Similarly in this trial, tight blood pressure control reduced the progression of retinopathy, compared with usual care. ¹⁰

Specific treatments for retinopathy include photocoagulation, which has been demonstrated to significantly reduce severe visual loss by 58% (NNT=30 patient-years).⁴⁰ One cohort study suggests that screening for retinopathy coupled with appropriate treatment may reduce the onset of visual loss.⁴¹ The ADA recommends annual screening for retinopathy with a dilated eye exam based on indirect evidence of its benefit (SOR: B).⁷

Correspondence
Paul Ullom-Minnich, MD, Partners in Family Care, PO Box 640, Moundridge, KS 67107. E-mail: [email protected].

Diabetes need not automatically sentence patients to the well-known ravages of the disease. Evidence-supported preventive strategies can forestall complications.

Increasing evidence suggests diabetes can be prevented by a combination of lifestyle changes and medications. Key interventions for patients with established type 2 diabetes include tight control of blood glucose levels, reduction of blood pressure and lipid levels, and early identification of diabetes-related neuropathy, nephropathy, and retinopathy. Antiplatelet therapy may also be beneficial.

Primary prevention: should intervention begin with prediabetes states?

Treatment of risk factors for diabetes (eg, obesity) and management of prediabetes (eg, impaired glucose tolerance) has suggested early intervention can forestall the development of type 2 diabetes (Table 1). Three trials addressed intensive lifestyle/diet modification. Though the onset of clinically diagnosed diabetes was delayed while patients adhered to these strategies, long-term studies of lifestylemodification have not been performed.^1–3

Several other investigations have looked at whether prescription medications achieve similar benefit. Metformin,³ acarbose,⁴ and orlistat⁵ have been studied in populations with prediabetes. These interventions appear to delay the onset of diabetes (number needed to treat [NNT]=33–88 patient-years) and to reduce blood sugar levels by 5% to 10%. A post-hoc analysis of the Heart Outcomes Prevention Evaluation (HOPE) trial⁶ also suggests a favorable effect with the ACE inhibitor, ramipril (NNT=250 patient-years). However, patient-oriented outcomes—development of microvascular or macrovascular disease—were not assessed in these trials. Consequently, the ultimate benefit of treating prediabetes states remains uncertain.

SECONDARY PREVENTION: DOES EARLY DETECTION OF DIABETES HELP DELAY COMPLICATIONS?

No randomized trial of screening has reported any patient-oriented benefits. However, based on consensus opinion, the American Diabetes Association (ADA) recommends screening every person aged 45 years and older every 3 years (strength of recommendation [SOR]: C).⁷ The characteristics of a clinically recognized disease like diabetes, however, may differ significantly from the characteristics of the subclinical states that would be recognized with screening. Therefore, though the US Preventive Health Services Task Force⁸ has concluded there is no evidence to recommend screening average risk individuals for diabetes, it does recommend screening individuals at increased risk of macrovascular changes (eg, those with hypertension) (SOR: B). This is based in part on indirect evidence that tighter blood pressure targets may be beneficial for patients with diabetes. (See the Clinical Inquiry, “Does screening for diabetes in at-risk patients improve long-term outcomes?”)

Tertiary prevention: preventing complications of existing diabetes

Patient-oriented outcomes in diabetes can be significantly improved with numerous interventions (Table 2).

Tight glycemic control warranted
The United Kingdom Prospective Diabetes Study¹⁴ (UKPDS) randomized participants to usual diabetic care or intensive glycemic control with insulin or sulfonylureas over 10 years. Intensive control reduced average hemoglobin A_1c from 7.9% to 7.0%; it also reduced aggregate microvascular complications, mainly a relative 39% decreased need for photocoagulation for diabetic retinopathy (NNT=320 patient-years). No other individual endpoint was independently affected.

The UKPDS suggested a trend toward 16% fewer relative MIs in the intensive control group; however, the results did not reach statistical significance (P=.052). In addition an increase in major hypoglycemic episodes was noted, worse with insulin than sulfonylureas (relative risk [RR]=257%; number needed to harm [NNH]=1110 patient-years).

Approach to obese patients. The UKPDS also studied intensive control with metformin among a subgroup of obese patients with diabetes (>120% ideal body weight).⁹ Metformin lowered average hemoglobin A_1c only from 8.0% to 7.4%, but reduced relative total mortality by 36% (NNT=142 patient-years). Metformin also reduced the relative chance of MI by 39% (NNT=143 patient-years).

These benefits were reversed, however, in a separate UKPDS subgroup placed first on a sulfonylurea, then receiving metformin if glycemic control was inadequate. Total mortality was relatively increased by 60% (NNH=89 patient-years).⁹ This adverse outcome disappeared when adjustment was made for age, sex, ethnic group, and glycemic control.

Other potential risks of metformin include lactic acidosis, but 1 recent systematic review found no associated cases of this condition with metformin prescribed in published studies.¹⁷

Applying the evidence. Initial treatment of patients with diabetes over 120% of ideal body weight should include tight glucose control with metformin, unless contraindicated (SOR: A). For leaner patients, therapy with a sulfonylurea or insulin is supported by evidence (SOR: B).

Control blood pressure to below 150/80 mm Hg
The UKPDS also compared tight blood pressure control (aiming at systolic <150 mm Hg and diastolic <80) with usual treatment. Tight control reduced relative deaths attributed to diabetes by 32% (NNT=150 patient-years), and demonstrated a trend toward reduced total mortality that was not statistically significant (relative risk reduction [RRR]=18%; 95% confidence interval [CI], –0.8% to 37%). Both stroke (NNT=196 patient-years) and the aggregate endpoint of “all microvascular disease” (NNT=139 patient-years) were significantly reduced by tight blood pressure control.10

The Hypertension Optimal Treatment (HOT) Study Group11 compared various levels of blood pressure control. A reduction in cardiovascular mortality was significant even between those treated with a goal of 80 and 85 mm Hg diastolic (RRR=67%; NNT=133 patient-years). The trend toward reduction in total mortality in this trial also approached statistical significance (P=.068; RRR=42%).

The Syst-Eur trial13 compared tight blood pressure control on nitrendipine with looser control with a variety of agents. Cardiovascular mortality was reduced by 30% with tight control (NNT=100 patient-years).

The Heart Outcomes Prevention Evaluation (HOPE) study6 included patients with diabetes, of which only 56% had a diagnosis of hypertension. It randomized participants to receive ramipril or placebo, in addition to any off-study antihypertensive agents they were already using. The intervention group had both lower average blood pressures as well as lower total mortality by 24% (NNT=140 patient-years).

Choice of agent important. Lower blood pressure goals have consistently demonstrated benefit across multiple studies. The choice of antihypertensive agent may also affect outcomes (Table 3). The UKPDS blood pressure analysis was also stratified to evaluate whether the results of blood pressure treatment differed between captopril and atenolol. Though compliance was slightly better with captopril, there were no differences in patient-oriented outcomes between the groups.²⁶

The Captopril Prevention Project (CAPPP)²⁷ compared captopril with diuretics and betablockers alone or in combination. While blood pressure control was about the same in all study groups, the captopril group realized a 66% reduction in MI (NNT=96 patient-years) and a total mortality 46% less than that seen with the other agents (NNT=96 patient-years).

The Appropriate Blood Pressure Control in Diabetes (ABCD) trial²⁸ compared nisoldipine and enalapril. Participants randomized to receive the ACE inhibitor had an 80% decreased risk of MI (NNT=25 patient-years).

In the Fosinopril versus Amlodipine Cardiovascular Events Trial (FACET),²⁹ blood pressure was better controlled with amlodipine, but major vascular events were 51% fewer with the ACE inhibitor (NNT=146 patient-years), again supporting the superior performance of ACE inhibitors.

Angiotensin receptor blockers (ARBs) may have comparable effects to ACE inhibitors. The Losartan Intervention For Endpoint Reduction in Hypertension (LIFE) trial³⁰ studied patients with left ventricular hypertrophy, diabetes, and hypertension, comparing losartan with atenolol. Total mortality with losartan was reduced by a relative 40% compared with atenolol (NNT=167 patientyears). Another study³¹ of patients with diabetes, hypertension, and nephropathy compared irbesartan with amlodipine. This trial demonstrated no differences in patient-oriented outcomes between the calcium-channel blocker and the ARB.

Applying the evidence. The goals for blood pressure control in type 2 diabetes should be less than 150 mm Hg systolic and 80 mm Hg diastolic (SOR: A). Evidence also strongly supports the use of an ACE inhibitor, or possibly ARB, as first-linetreatment for hypertension in diabetes (SOR: A). Many authorities recommend even more aggressive blood pressure goals.

Lipid management: statins improve outcome
Lowering elevated triglycerides has not been independently associated with an improvement in patient-oriented outcomes. In the Helsinki Heart Study²² and the St. Mary’s, Ealing, Northwick Park Diabetes Cardiovascular Disease Prevention (SENDCAP) trial,³² a fibric acid derivative was compared with placebo. The average triglyceride concentration was decreased, but no significant effect on coronary events was noted. However, elevated triglycerides are associated with the metabolic syndrome, which may warrant lifestyle changes or medication (based on expert opinion).

Treatment with hydroxymethyl glutaryl coenzyme A reductase inhibitors (statins) has better supporting evidence. In the Scandinavia Simvastatin Survival Study (4S),¹⁵ 202 patients with diabetes, coronary artery disease, and elevated cholesterol were randomized to receive simvastatin or placebo. Major coronary events were reduced by 55% with simvastatin (NNT=22.5 patient-years). Total mortality was reduced, but not to a statistically significant extent (P=.087; RRR=43%).

The Air Force/Texas Coronary Atherosclerosis Prevention Study³³ (AFCAPS/TexCAPS) compared lovastatin with placebo for patients with diabetes and healthy individuals with normal cholesterol levels. Though the overall population demonstrated a 37% reduction in first coronary events, the diabetes subgroup had insufficient power to confirm this trend independently.

The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial³⁴ (ALLHAT-LLT) randomized 3638 patients with diabetes to receive pravastatin 40 mg or placebo. No changes were seen in total mortality or cardiovascular events. These results may have been confounded by off-study prescription for statins, given to 32% of those randomized to the placebo group.

The best independent evidence for use of statins in diabetes comes from the Heart Protection Study¹⁶ of 3982 patients with diabetes with a total cholesterol level >135 mg/dL and no evidence of coronary disease. They were randomized to receive 40 mg simvastatin or placebo for 5 years. First major vascular events (MI or stroke) were decreased in the simvastatin group by 26% over 5 years (NNT=104 patient-years).

Applying the evidence. Reducing elevated triglycerides in type 2 diabetes is not supported by clear evidence, although elevated triglycerides may be associated with the metabolic syndrome and may warrant lifestyle change (SOR: C).

However, statins—even for those with a normal cholesterol level—reduce macrovascular outcomes (SOR: A), a measure now endorsed by the American College of Physicians.

Anti-platelet therapy
Aspirin prophylaxis in diabetes has weaker support. The Early Treatment Diabetic Retinopathy Study³⁵ (ETDRS) randomized 3711 patients with diabetes to receive aspirin 650 mg or placebo. No benefit in mortality or cardiovascular event rates was documented after 5 years. However, the results did suggest a nonsignificant trend toward reduction of MI (RRR=17%; 95% CI, –4% to 34%; NNT=333).

In the Physician’s Health Study,³⁶ 22,071 participants (most of whom did not have diabetes) were randomized to receive aspirin 325 mg every other day or placebo. MI was reduced by 44% with aspirin (NNT=500). The risk of MI in the subgroup of 533 individuals with diabetes paralleled this reduction (RRR=39%), though, independently, the reduction in the subgroup did not reach statistical significance. Bleeding problems were the most common adverse effect of the aspirin, and were increased by 32% (NNH=78 person-years).

In the HOT study,¹¹ a subgroup of 1501 patients with diabetes was randomized to receive aspirin 75 mg or placebo daily. The rate of MI trended downward with aspirin but was not statistically significant. Among the 18,790 patients in this study (most of whom did not have diabetes), MI was 36% less likely to occur with aspirin (NNT=770).

Applying the evidence. Overall, there is some suggestion that persons with cardiac risk factors, like type 2 diabetes, benefit by taking low-dose aspirin to avoid macrovascular complications. No study has confirmed this in a population of individuals with diabetes, but the trends suggest a possible benefit (SOR: C). The decision to use aspirin should be made in consultation with an informed patient.

SCREENING FOR NEUROPATHY

One trial¹⁷ has been conducted on screening for neuropathy. It compared monofilament testing and palpation of pedal pulses with “no special care.” Patients with an original positive screen result received a calculation of their ankle-brachial index, foot x-rays, and other measurements, and were referred to a high-risk podiatry clinic if the second level testing showed abnormal results. Major amputations were decreased in the screened group by 92% over 2 years (NNT=180 patient-years). This evidence of benefit is sufficient to recommend screening patients with diabetes for peripheral neuropathy or peripheral vascular disease, and appropriate referral (SOR: B). (See the Clinical Inquiry, “What is the best treatment for diabetic neuropathy?”)

Screening for nephropathy

One systematic review³⁷ found no randomized trials of screening for urinary microalbumin and how it might affect overt nephropathy. However, several studies have looked at treatment of gross proteinuria, and have demonstrated some benefit in patient-oriented outcomes. Improved blood pressure control has reduced progression of nephropathy to end-stage renal disease.^38,39 Other studies have suggested that an ACE inhibitor or ARB might provide benefit.^30,39

The ADA recommends annual screening for microalbumin based on expert consensus (SOR: C).⁷ Though early detection of microalbumin might optimize the treatment of nephropathy, it is also possible that screening may detect a population whose disease would have remained subclinical indefinitely. No clear evidence suggests that screening for microalbumin reduces the incidence of patient-oriented outcomes.

Screening for retinopathy

Several trials have evaluated interventions in the diagnosis and prevention of visual loss. As discussed previously, intensive glucose control with sulfonylureas or insulin reduced the need for retinal photocoagulation by 39% in the UKPDS (NNT=320 patient-years).¹⁴ Similarly in this trial, tight blood pressure control reduced the progression of retinopathy, compared with usual care. ¹⁰

Specific treatments for retinopathy include photocoagulation, which has been demonstrated to significantly reduce severe visual loss by 58% (NNT=30 patient-years).⁴⁰ One cohort study suggests that screening for retinopathy coupled with appropriate treatment may reduce the onset of visual loss.⁴¹ The ADA recommends annual screening for retinopathy with a dilated eye exam based on indirect evidence of its benefit (SOR: B).⁷

Correspondence
Paul Ullom-Minnich, MD, Partners in Family Care, PO Box 640, Moundridge, KS 67107. E-mail: [email protected].

When a patient complains of pain “all over,” consider fibromyalgia, which typically causes a well-documented pattern of pain and characteristic points of tenderness observable on physical exam. Once alternative diagnoses have been ruled out, offer the patient a 2-pronged therapeutic regimen that has proven successful at moderately relieving symptoms.

First rule out concomitant or mimicking disorders

Consider the differential diagnosis carefully.¹ A person who meets the criteria for fibromyalgia may have yet another cause of chronic pain, such as rheumatoid arthritis, or may instead have a different treatable condition that mimics fibromyalgia.

Drug-induced myopathy. Pain suggestive of fibromyalgia should prompt a review of the patient’s medicines. Drug-induced myopathy may occur in persons taking colchicine, statins, corticosteroids, or antimalarial drugs.

Connective tissue, autoimmune, and rheumatologic disorders. Consider this group of disorders next. In 1 study, one fourth of persons referred to a rheumatology clinic with presumed fibromyalgia instead had a spondyloarthropathy.²

Dermatomyositis and polymyositis may present with muscle pain and tenderness but, unlike fibromyalgia, cause proximal muscle weakness.

Systemic lupus erythematosus, rheumatoid arthritis, and polymyalgia rheumatica can also lead to widespread pain.

Blood tests such as antinuclear antibody (ANA), C-reactive protein, or erythrocyte sedimentation rate (ESR) may prove helpful when a patient has a history of unexplained rashes, fever, weight loss, joint swelling, iritis, hepatitis, nephritis, or inflammatory back pain (onset before age 40, insidious onset, present for more than 3 months, associated with morning stiffness, improvement with exercise).³ In the absence of these signs, ANA, rheumatoid factor, and ESR testing in persons with fatigue and diffuse musculoskeletal pain have low positive predictive value.⁴ The rate of false-positive ANA results may be as high as 8% to 11%, especially at low titers.^5,6

Hypothyroidism. Widespread musculoskeletal pain has also been associated with hypothyroidism (level of evidence [LOE]: 2, case-control design),^7,8 supporting the inclusion of a thyroidstimulating hormone in the work-up of fibromyalgia (strength of recommendation [SOR]: B). More recent research suggests that musculoskeletal pain is more related to thyroid microsomal antibodies than to hypothyroidism,⁹ but there has been no further evaluation of antithyroid antibodies in persons with fibromyalgia.

Diagnosis: mostly by clinical judgment

Persons with fibromyalgia have widespread pain, often worst in the neck and trunk.¹ Additional symptoms include fatigue, morning stiffness, waking unrefreshed, paresthesias, and headache.^1,10-15 (See “The toll of fibromyalgia.”)

Accepted criteria

The diagnosis of fibromyalgia is based on 2 criteria:

1. A patient’s report of widespread pain (right and left sides of the body, above and below the waist, and including the axial skeleton) persisting for at least 3 months

2. The clinician’s identification of at least 11 of 18 potential tender points as specified in the American College of Rheumatology (ACR) 1990 Criteria for the Classification of Fibromyalgia (Figure) (LOE: 3, case-control design, nonindependent reference standard).¹

These criteria do not exclude persons with rheumatic diseases or other chronic pain conditions.^1,37-39

Caveats with the criteria

Despite these well-defined criteria, the diagnosis is not as clear-cut as it may appear. In 1990, the ACR convened a panel of 24 experts to define and standardize the diagnosis of fibromyalgia. The basis for this consensus was a group of 293 patients with fibromyalgia, each of whom had been assessed by one of the expert investigators according to “his or her usual method of diagnosis.” ¹

The investigators determined the unique characteristics of fibromyalgia by comparing the 293 cases to 265 controls who had other chronic pain conditions (eg, low back pain syndromes, neck pain syndromes, regional tendonitis, possible systemic lupus erythematosus, rheumatoid arthritis). The investigators considered a multitude of symptoms and signs including sleep disturbance, morning stiffness, paresthesias, irritable bowel syndrome, fatigue, and anxiety. Their conclusion was that widespread pain and tender points were the most sensitive (88.4%) and specific (81.1%) distinguishing criteria for fibromyalgia.¹

No reference standard. However, these calculations of sensitivity and specificity are less meaningful than in studies where an independent reference standard or gold standard is available. The ACR expert panel derived the criteria in a circular way using a nonindependent reference standard—ie, patients thought to have fibromyalgia compared with control patients thought not to have fibromyalgia. The expert panel essentially set the specificity of the criteria at 100%, since the specificity is based on the rate of false positives.

Furthermore, because there was no objective gold standard for determining who truly had fibromyalgia (and we do not yet have an independent biologic “test” for this condition), this panel could not determine whether additional symptoms or signs that should be considered in the diagnosis of fibromyalgia.

Biases, dubious representation? Unknown elements in this analysis are 1) how closely the reference population used to develop these criteria represents the true population of persons with fibromyalgia, and 2) the biases of the ACR experts. Finally, the positive and negative predictive values of these criteria will depend on the prevalence of fibromyalgia and other similar conditions.

Morbidity not predicted by criteria. In addition, the 1990 ACR criteria assume the number of tender points and degree of pain are directly proportional to overall morbidity; however, a person with fewer than 11 tender points may experience significant morbidity, indicating that the sensitivity of the criteria may be low.^40-42 As suggested by Wolfe in 1997, “the tender point count functions as a sedimentation rate for distress” in persons with chronic pain.⁴² Thus, the authors of the 1990 ACR study stated that ACR criteria should not be applied rigidly in diagnosing and treating fibromyalgia,⁴² leaving a large role for clinical judgment.

Subjective factors. A final difficulty with the diagnosis of fibromyalgia is its dependence on patient report and examiner technique.¹ In the 1990 ACR criteria, tender points were defined as a complaint of pain (or any more dramatic response) when an examiner applied 4 kg of pressure with the pulp of the thumb or first two or three fingers, calibrated with a dolorimeter (a device that can measure the amount and rate of pressure applied over a specified surface area).¹ It has been shown that practitioners require training to apply 4 kg of force with regularity.⁴³

However, applying exactly 4 kg of pressure may not be clinically important. Other studies have shown that finger palpation or dolorimetry identifies tender points with equal accuracy (LOE: 3, case-control design with non-independent reference standard).^44,45

Manual palpation

A controlled study of manual palpation was conducted to standardize the tender point survey described in the Figure. ⁴⁶ This method compares well with the ACR 1990 method, with a sensitivity of 88.6% and a specificity of 71.4%.⁴⁶

To speed up the examination, a particular sequence of palpating survey points was established, with the patient positioned as outlined in the Figure. Using the thumb pad of his/her dominant hand, each examiner applied 4 kg of pressure, at a rate of 1 kg per second, just once at each survey point. Examiners learned to apply the proper amount of pressure by standing a patient on a scale and watching the scale while pressing down perpendicularly on the trapezius survey point.

The examinee was seated throughout the exam, except when lying on the side for palpation of the trochanter and lying supine for palpation of the knee. A tender point was identified when the patient rated the pain resulting from palpation at least 2 out of 10 (0, no pain; 10 worst pain) (LOE: 3, case-control design, nonindependent reference standard).⁴⁶

Until a firm biologic basis for fibromyalgia is discovered and a true gold standard for testing is developed, the diagnosis of fibromyalgia will remain a matter of clinical judgment and convention (SOR: C).

Treatment

A diagnosis of fibromyalgia alone may result in health benefits. In a year-long study published in 1986, Cathey et al reported that among 81 persons diagnosed with fibromyalgia, hospitalization rates decreased in the year following diagnosis (LOE: 2, case-control design).⁴⁷

Treatments for fibromyalgia are numerous, ranging from balneotherapy (bathing) to low-energy laser therapy, and studies of interventions are often poorly designed, based on small numbers of patients, report nonstandardized outcomes, and yield conflicting results.⁴⁸

Two interventions—aerobic exercise and antidepressant therapy—appear to improve fibromyalgia.

Aerobic exercise

Though pain relief is insignificant with aerobic exercise, other positive effects are significant (SOR: A). A 2003 Cochrane review identified 7 high-quality studies of aerobic training, defined as: 1) frequency of 2 days per week; 2) intensity sufficient to achieve 40% to 85% of heart rate reserve, or 55% to 90% of predicted maximum heart rate; 3) duration of sessions 20 to 60 minutes, either continuously or intermittently throughout the day, using any mode of aerobic exercise; and 4) a total exercise period of at least 6 weeks (Table 1).⁴⁹

Improved functioning, tender-point threshold. Study subjects engaged in aerobic dancing, whole-body aerobics, stationary cycling, and walking. Persons who exercised improved in global well-being, physical function, and aerobic fitness (by about 17%), and raised the pain threshold of tender points (by about 35%).⁴⁹ Four of the studies were similar enough to be combined for meta-analysis, showing a statistically robust but modest reduction in tender-point threshold (LOE: 1).

Although it seems likely that pain or fatigue might increase at least initially with exercise, participants in the exercise groups were not deterred; the researchers pointed out that reporting of adverse effects of aerobic exercise appeared incomplete, but there was no significant difference in drop-out rates between the exercise (25.1%) and control groups (12.5%).⁴⁹

In the long-term studies (>6 months), improvements were noted up to 1 year after treatment ended but not after 4.5 years.⁴⁹ This Cochrane review further supports aerobic exercise as bring beneficial for persons with fibromyalgia.^50,51

Additional improvement measures. A similar systematic review concluded that although studies were too heterogeneous to draw final conclusions, preliminary data supported aerobic exercise (LOE: 2, with heterogeneous studies).⁵⁰ In another comprehensive meta-analysis of all treatments for fibromyalgia, heterogeneous treatment studies ranging from exercise to physical therapy were identified as physically-based treatments. The analysis revealed a positive effect on physical status (including tender-point index, grip strength, and physician global rating of pain symptoms), fibromyalgia symptoms (including self-reported fatigue and pain using visual analog scales), and psychological status (including measurements of the Hamilton Depression and Anxiety Scales), with no effect on daily functioning (including outcome measures such as the Fibromyalgia Impact Questionnaire [FIQ]) (LOE: 2, with heterogeneous studies).⁵¹

The authors noted that the magnitude of the positive effects of physically-based treatments on fibromyalgia were comparable with drug treatment judged effective for arthritis.⁵¹

TABLE 1
Aerobic exercise for fibromyalgia: the evidence

Less certain nonpharmacologic therapies

Other nonpharmacologic treatments for fibromyalgia are educational interventions, relaxation therapy, cognitive-behavioral therapy, and acupuncture. These therapies have undergone rigorous analysis, but studies have been too heterogeneous to allow for strong conclusions across the studies.⁵⁰

A recent Cochrane review concluded that although physical training plus education had a positive effect at long-term follow up, evidence is insufficient to recommend multidisciplinary rehabilitation, defined as the care of a physician plus psychological, social, and vocational interventions (SOR: C).⁵²

In contrast, other investigators have concluded that multidisciplinary treatment incorporating physically and psychologically based treatments was more successful than treatment with a single modality.⁵¹ A systematic review of acupuncture identified only 1 high-quality randomized controlled trial (Table 2), which did show some improvement in symptoms (SOR: C).⁵³

TABLE 2
Alternative nonpharmacologic therapies for fibromyalgia: the evidence

Therapy with antidepressants

Of all pharmacologic treatments, antidepressants have undergone the most thorough study. Although the optimal role of medications in fibromyalgia has not been delineated, 3 metaanalyses have reported that antidepressants, most commonly amitriptyline, reduce symptoms during treatment of a few months duration (SOR: A) (Table 3).^54,55

Any antidepressants. Pooled results from 13 studies (8 of tricyclics, 3 of selective serotonin reuptake inhibitors, 2 of s-adenosylmethionine) revealed a moderate effect on pain, sleep, and global well-being, and a mild effect on fatigue and number of trigger points.⁵⁴ The authors calculated that persons with fibromyalgia treated with antidepressants were 4 times more likely to improve than persons treated with placebo (number need to treat [NNT]=4). Adverse effects appeared insignificant but were poorly reported in the individual studies.

Tricyclics only. In another meta-analysis, 9 high-quality studies of tricyclic antidepressants (amitriptyline, dothiepin, clomipramine, maprotiline and cyclobenzaprine—classified by the authors as a tricyclic antidepressant) were analyzed for 7 outcomes (patient self-rating of pain, fatigue, stiffness, and sleep; the patient and physician global assessment of improvement; and tenderness of tender points). Significant responses were observed in 25% to 37% of patients. On meta-analysis, outcome measures improved moderately overall with tricyclic treatment, mostly in sleep and global assessment, least in stiffness and tenderness. Long-term safety (more than 8 weeks) and efficacy of tricyclic therapy for fibromyalgia have not been demonstrated.⁵⁵

Combined trials.A third meta-analysis demonstrated improvement when trials of different antidepressants were combined.⁵¹ By pooling studies of antidepressants (amitriptyline, dothiepin, fluoxetine, citalopram, and S-adenosylmethionine) improvements in physical status, fibromyalgia symptoms, and psychological status were found, with no improvement in daily functioning.⁵¹ Although the effect was smaller than physicallybased treatments, the effect size was still comparable to drug treatment for arthritis.⁵¹

Muscle relaxants (primarily cyclobenzaprine) and nonsteroidal anti-inflammatories have been studied, with no evidence of a positive effect.⁵¹ Thus, the best evidence currently supports the use of aerobic exercise and antidepressants, particularly tricyclics, for the treatment of fibromyalgia.

TABLE 3
Antidepressant therapy for fibromyalgia: the evidence

Instructions to patients, management follow-up

Persons with fibromyalgia should know that although specific symptoms, particularly pain, may be not be dramatically reduced with treatment, aerobic exercise and tricyclic antidepressants alleviate some symptoms with minimal adverse effects (SOR: A). Emphasize that these treatments have been shown to improve one’s ability to cope with fibromyalgia symptoms. The best-studied antidepressant for treating fibromyalgia is amitriptyline, usually given at 25 to 50 mg, nightly.

Exercise. Prescribe aerobic exercise, at least twice per week for 20 to 60 minutes, targeting a heart rate of 55% to 90% of the predicted maximum (180 beats per minute-age) (SOR: A). One caveat: aerobic exercise in the literature was usually supervised, so the ideal exercise regimen might be a fibromyalgia-specific program.

Medication. Consider a trial of amitriptyline, 25 to 50 mg every night, for up to 6 weeks (SOR A). A caveat: tricyclic antidepressants may also have significant side effects, which could outweigh moderate benefits. Moreover, treatment effectiveness beyond 2 months has not been proven. Therefore, longitudinal measurement of outcomes should be part of ongoing care.

Follow-up. Studies have not determined which measures are best to follow (see “Assessing treatment efficacy”), but they might include the following (SOR: C):

1. Pain (eg, visual analogue scale, pain drawings)
2. Number of tender points, and tenderness
3. Physical function (eg, cardiorespiratory fitness, self-reported physical function measured by the physical-impairment subscale of the FIQ,⁵⁶ strength)
4. Global well-being or perceived improvement (eg, physician-rated change, FIQ total score)
5. Self-efficacy (eg, Arthritis Self-efficacy Questionnaire)
6. Fatigue and sleep (eg, FIQ fatigsubscale, sleep visual analogue scale)
7. Psychological function (eg, FIQ subscales for depression and anxiety)
8. Ability to work
9. Health care consumption and costs.^47,50

Education or psychological coping strategies may also contribute positively to overall patient and family well-being. Consider education/psychological counseling (SOR: C) and acupuncture (SOR: B).

Acknowledgments

The authors would like to express their appreciation to Cheryl Mongillo, Peggy Lardear, and Brian Pellini for their assistance in preparing the manuscript as well as Dolores Moran and Diane Wolf for their library assistance, and to James Newman, MD, rheumatologist, for his expert suggestions. Funding for this project was provided by a grant from the Delaware Department of Health and Social Services, Division of Public Health.of Health and Social Services, Division of Public Health.

Corresponding author
Anna Quisel, MD, c/o Cheryl Mongillo, Department of Family and Community Medicine, Christiana Care Health Systems, 1401 Foulk Road, Wilmington, DE 19803. E-mail: [email protected].

When a patient complains of pain “all over,” consider fibromyalgia, which typically causes a well-documented pattern of pain and characteristic points of tenderness observable on physical exam. Once alternative diagnoses have been ruled out, offer the patient a 2-pronged therapeutic regimen that has proven successful at moderately relieving symptoms.

First rule out concomitant or mimicking disorders

Consider the differential diagnosis carefully.¹ A person who meets the criteria for fibromyalgia may have yet another cause of chronic pain, such as rheumatoid arthritis, or may instead have a different treatable condition that mimics fibromyalgia.

Drug-induced myopathy. Pain suggestive of fibromyalgia should prompt a review of the patient’s medicines. Drug-induced myopathy may occur in persons taking colchicine, statins, corticosteroids, or antimalarial drugs.

Connective tissue, autoimmune, and rheumatologic disorders. Consider this group of disorders next. In 1 study, one fourth of persons referred to a rheumatology clinic with presumed fibromyalgia instead had a spondyloarthropathy.²

Dermatomyositis and polymyositis may present with muscle pain and tenderness but, unlike fibromyalgia, cause proximal muscle weakness.

Systemic lupus erythematosus, rheumatoid arthritis, and polymyalgia rheumatica can also lead to widespread pain.

Blood tests such as antinuclear antibody (ANA), C-reactive protein, or erythrocyte sedimentation rate (ESR) may prove helpful when a patient has a history of unexplained rashes, fever, weight loss, joint swelling, iritis, hepatitis, nephritis, or inflammatory back pain (onset before age 40, insidious onset, present for more than 3 months, associated with morning stiffness, improvement with exercise).³ In the absence of these signs, ANA, rheumatoid factor, and ESR testing in persons with fatigue and diffuse musculoskeletal pain have low positive predictive value.⁴ The rate of false-positive ANA results may be as high as 8% to 11%, especially at low titers.^5,6

Hypothyroidism. Widespread musculoskeletal pain has also been associated with hypothyroidism (level of evidence [LOE]: 2, case-control design),^7,8 supporting the inclusion of a thyroidstimulating hormone in the work-up of fibromyalgia (strength of recommendation [SOR]: B). More recent research suggests that musculoskeletal pain is more related to thyroid microsomal antibodies than to hypothyroidism,⁹ but there has been no further evaluation of antithyroid antibodies in persons with fibromyalgia.

Diagnosis: mostly by clinical judgment

Persons with fibromyalgia have widespread pain, often worst in the neck and trunk.¹ Additional symptoms include fatigue, morning stiffness, waking unrefreshed, paresthesias, and headache.^1,10-15 (See “The toll of fibromyalgia.”)

Accepted criteria

The diagnosis of fibromyalgia is based on 2 criteria:

1. A patient’s report of widespread pain (right and left sides of the body, above and below the waist, and including the axial skeleton) persisting for at least 3 months

2. The clinician’s identification of at least 11 of 18 potential tender points as specified in the American College of Rheumatology (ACR) 1990 Criteria for the Classification of Fibromyalgia (Figure) (LOE: 3, case-control design, nonindependent reference standard).¹

These criteria do not exclude persons with rheumatic diseases or other chronic pain conditions.^1,37-39

Caveats with the criteria

Despite these well-defined criteria, the diagnosis is not as clear-cut as it may appear. In 1990, the ACR convened a panel of 24 experts to define and standardize the diagnosis of fibromyalgia. The basis for this consensus was a group of 293 patients with fibromyalgia, each of whom had been assessed by one of the expert investigators according to “his or her usual method of diagnosis.” ¹

The investigators determined the unique characteristics of fibromyalgia by comparing the 293 cases to 265 controls who had other chronic pain conditions (eg, low back pain syndromes, neck pain syndromes, regional tendonitis, possible systemic lupus erythematosus, rheumatoid arthritis). The investigators considered a multitude of symptoms and signs including sleep disturbance, morning stiffness, paresthesias, irritable bowel syndrome, fatigue, and anxiety. Their conclusion was that widespread pain and tender points were the most sensitive (88.4%) and specific (81.1%) distinguishing criteria for fibromyalgia.¹

No reference standard. However, these calculations of sensitivity and specificity are less meaningful than in studies where an independent reference standard or gold standard is available. The ACR expert panel derived the criteria in a circular way using a nonindependent reference standard—ie, patients thought to have fibromyalgia compared with control patients thought not to have fibromyalgia. The expert panel essentially set the specificity of the criteria at 100%, since the specificity is based on the rate of false positives.

Furthermore, because there was no objective gold standard for determining who truly had fibromyalgia (and we do not yet have an independent biologic “test” for this condition), this panel could not determine whether additional symptoms or signs that should be considered in the diagnosis of fibromyalgia.

Biases, dubious representation? Unknown elements in this analysis are 1) how closely the reference population used to develop these criteria represents the true population of persons with fibromyalgia, and 2) the biases of the ACR experts. Finally, the positive and negative predictive values of these criteria will depend on the prevalence of fibromyalgia and other similar conditions.

Morbidity not predicted by criteria. In addition, the 1990 ACR criteria assume the number of tender points and degree of pain are directly proportional to overall morbidity; however, a person with fewer than 11 tender points may experience significant morbidity, indicating that the sensitivity of the criteria may be low.^40-42 As suggested by Wolfe in 1997, “the tender point count functions as a sedimentation rate for distress” in persons with chronic pain.⁴² Thus, the authors of the 1990 ACR study stated that ACR criteria should not be applied rigidly in diagnosing and treating fibromyalgia,⁴² leaving a large role for clinical judgment.

Subjective factors. A final difficulty with the diagnosis of fibromyalgia is its dependence on patient report and examiner technique.¹ In the 1990 ACR criteria, tender points were defined as a complaint of pain (or any more dramatic response) when an examiner applied 4 kg of pressure with the pulp of the thumb or first two or three fingers, calibrated with a dolorimeter (a device that can measure the amount and rate of pressure applied over a specified surface area).¹ It has been shown that practitioners require training to apply 4 kg of force with regularity.⁴³

However, applying exactly 4 kg of pressure may not be clinically important. Other studies have shown that finger palpation or dolorimetry identifies tender points with equal accuracy (LOE: 3, case-control design with non-independent reference standard).^44,45

Manual palpation

A controlled study of manual palpation was conducted to standardize the tender point survey described in the Figure. ⁴⁶ This method compares well with the ACR 1990 method, with a sensitivity of 88.6% and a specificity of 71.4%.⁴⁶

To speed up the examination, a particular sequence of palpating survey points was established, with the patient positioned as outlined in the Figure. Using the thumb pad of his/her dominant hand, each examiner applied 4 kg of pressure, at a rate of 1 kg per second, just once at each survey point. Examiners learned to apply the proper amount of pressure by standing a patient on a scale and watching the scale while pressing down perpendicularly on the trapezius survey point.

The examinee was seated throughout the exam, except when lying on the side for palpation of the trochanter and lying supine for palpation of the knee. A tender point was identified when the patient rated the pain resulting from palpation at least 2 out of 10 (0, no pain; 10 worst pain) (LOE: 3, case-control design, nonindependent reference standard).⁴⁶

Until a firm biologic basis for fibromyalgia is discovered and a true gold standard for testing is developed, the diagnosis of fibromyalgia will remain a matter of clinical judgment and convention (SOR: C).

Treatment

A diagnosis of fibromyalgia alone may result in health benefits. In a year-long study published in 1986, Cathey et al reported that among 81 persons diagnosed with fibromyalgia, hospitalization rates decreased in the year following diagnosis (LOE: 2, case-control design).⁴⁷

Treatments for fibromyalgia are numerous, ranging from balneotherapy (bathing) to low-energy laser therapy, and studies of interventions are often poorly designed, based on small numbers of patients, report nonstandardized outcomes, and yield conflicting results.⁴⁸

Two interventions—aerobic exercise and antidepressant therapy—appear to improve fibromyalgia.

Aerobic exercise

Though pain relief is insignificant with aerobic exercise, other positive effects are significant (SOR: A). A 2003 Cochrane review identified 7 high-quality studies of aerobic training, defined as: 1) frequency of 2 days per week; 2) intensity sufficient to achieve 40% to 85% of heart rate reserve, or 55% to 90% of predicted maximum heart rate; 3) duration of sessions 20 to 60 minutes, either continuously or intermittently throughout the day, using any mode of aerobic exercise; and 4) a total exercise period of at least 6 weeks (Table 1).⁴⁹

Improved functioning, tender-point threshold. Study subjects engaged in aerobic dancing, whole-body aerobics, stationary cycling, and walking. Persons who exercised improved in global well-being, physical function, and aerobic fitness (by about 17%), and raised the pain threshold of tender points (by about 35%).⁴⁹ Four of the studies were similar enough to be combined for meta-analysis, showing a statistically robust but modest reduction in tender-point threshold (LOE: 1).

Although it seems likely that pain or fatigue might increase at least initially with exercise, participants in the exercise groups were not deterred; the researchers pointed out that reporting of adverse effects of aerobic exercise appeared incomplete, but there was no significant difference in drop-out rates between the exercise (25.1%) and control groups (12.5%).⁴⁹

In the long-term studies (>6 months), improvements were noted up to 1 year after treatment ended but not after 4.5 years.⁴⁹ This Cochrane review further supports aerobic exercise as bring beneficial for persons with fibromyalgia.^50,51

Additional improvement measures. A similar systematic review concluded that although studies were too heterogeneous to draw final conclusions, preliminary data supported aerobic exercise (LOE: 2, with heterogeneous studies).⁵⁰ In another comprehensive meta-analysis of all treatments for fibromyalgia, heterogeneous treatment studies ranging from exercise to physical therapy were identified as physically-based treatments. The analysis revealed a positive effect on physical status (including tender-point index, grip strength, and physician global rating of pain symptoms), fibromyalgia symptoms (including self-reported fatigue and pain using visual analog scales), and psychological status (including measurements of the Hamilton Depression and Anxiety Scales), with no effect on daily functioning (including outcome measures such as the Fibromyalgia Impact Questionnaire [FIQ]) (LOE: 2, with heterogeneous studies).⁵¹

The authors noted that the magnitude of the positive effects of physically-based treatments on fibromyalgia were comparable with drug treatment judged effective for arthritis.⁵¹

TABLE 1
Aerobic exercise for fibromyalgia: the evidence

Less certain nonpharmacologic therapies

Other nonpharmacologic treatments for fibromyalgia are educational interventions, relaxation therapy, cognitive-behavioral therapy, and acupuncture. These therapies have undergone rigorous analysis, but studies have been too heterogeneous to allow for strong conclusions across the studies.⁵⁰

A recent Cochrane review concluded that although physical training plus education had a positive effect at long-term follow up, evidence is insufficient to recommend multidisciplinary rehabilitation, defined as the care of a physician plus psychological, social, and vocational interventions (SOR: C).⁵²

In contrast, other investigators have concluded that multidisciplinary treatment incorporating physically and psychologically based treatments was more successful than treatment with a single modality.⁵¹ A systematic review of acupuncture identified only 1 high-quality randomized controlled trial (Table 2), which did show some improvement in symptoms (SOR: C).⁵³

TABLE 2
Alternative nonpharmacologic therapies for fibromyalgia: the evidence

Therapy with antidepressants

Of all pharmacologic treatments, antidepressants have undergone the most thorough study. Although the optimal role of medications in fibromyalgia has not been delineated, 3 metaanalyses have reported that antidepressants, most commonly amitriptyline, reduce symptoms during treatment of a few months duration (SOR: A) (Table 3).^54,55

Any antidepressants. Pooled results from 13 studies (8 of tricyclics, 3 of selective serotonin reuptake inhibitors, 2 of s-adenosylmethionine) revealed a moderate effect on pain, sleep, and global well-being, and a mild effect on fatigue and number of trigger points.⁵⁴ The authors calculated that persons with fibromyalgia treated with antidepressants were 4 times more likely to improve than persons treated with placebo (number need to treat [NNT]=4). Adverse effects appeared insignificant but were poorly reported in the individual studies.

Tricyclics only. In another meta-analysis, 9 high-quality studies of tricyclic antidepressants (amitriptyline, dothiepin, clomipramine, maprotiline and cyclobenzaprine—classified by the authors as a tricyclic antidepressant) were analyzed for 7 outcomes (patient self-rating of pain, fatigue, stiffness, and sleep; the patient and physician global assessment of improvement; and tenderness of tender points). Significant responses were observed in 25% to 37% of patients. On meta-analysis, outcome measures improved moderately overall with tricyclic treatment, mostly in sleep and global assessment, least in stiffness and tenderness. Long-term safety (more than 8 weeks) and efficacy of tricyclic therapy for fibromyalgia have not been demonstrated.⁵⁵

Combined trials.A third meta-analysis demonstrated improvement when trials of different antidepressants were combined.⁵¹ By pooling studies of antidepressants (amitriptyline, dothiepin, fluoxetine, citalopram, and S-adenosylmethionine) improvements in physical status, fibromyalgia symptoms, and psychological status were found, with no improvement in daily functioning.⁵¹ Although the effect was smaller than physicallybased treatments, the effect size was still comparable to drug treatment for arthritis.⁵¹

Muscle relaxants (primarily cyclobenzaprine) and nonsteroidal anti-inflammatories have been studied, with no evidence of a positive effect.⁵¹ Thus, the best evidence currently supports the use of aerobic exercise and antidepressants, particularly tricyclics, for the treatment of fibromyalgia.

TABLE 3
Antidepressant therapy for fibromyalgia: the evidence

Instructions to patients, management follow-up

Persons with fibromyalgia should know that although specific symptoms, particularly pain, may be not be dramatically reduced with treatment, aerobic exercise and tricyclic antidepressants alleviate some symptoms with minimal adverse effects (SOR: A). Emphasize that these treatments have been shown to improve one’s ability to cope with fibromyalgia symptoms. The best-studied antidepressant for treating fibromyalgia is amitriptyline, usually given at 25 to 50 mg, nightly.

Exercise. Prescribe aerobic exercise, at least twice per week for 20 to 60 minutes, targeting a heart rate of 55% to 90% of the predicted maximum (180 beats per minute-age) (SOR: A). One caveat: aerobic exercise in the literature was usually supervised, so the ideal exercise regimen might be a fibromyalgia-specific program.

Medication. Consider a trial of amitriptyline, 25 to 50 mg every night, for up to 6 weeks (SOR A). A caveat: tricyclic antidepressants may also have significant side effects, which could outweigh moderate benefits. Moreover, treatment effectiveness beyond 2 months has not been proven. Therefore, longitudinal measurement of outcomes should be part of ongoing care.

Follow-up. Studies have not determined which measures are best to follow (see “Assessing treatment efficacy”), but they might include the following (SOR: C):

1. Pain (eg, visual analogue scale, pain drawings)
2. Number of tender points, and tenderness
3. Physical function (eg, cardiorespiratory fitness, self-reported physical function measured by the physical-impairment subscale of the FIQ,⁵⁶ strength)
4. Global well-being or perceived improvement (eg, physician-rated change, FIQ total score)
5. Self-efficacy (eg, Arthritis Self-efficacy Questionnaire)
6. Fatigue and sleep (eg, FIQ fatigsubscale, sleep visual analogue scale)
7. Psychological function (eg, FIQ subscales for depression and anxiety)
8. Ability to work
9. Health care consumption and costs.^47,50

Education or psychological coping strategies may also contribute positively to overall patient and family well-being. Consider education/psychological counseling (SOR: C) and acupuncture (SOR: B).

Acknowledgments

The authors would like to express their appreciation to Cheryl Mongillo, Peggy Lardear, and Brian Pellini for their assistance in preparing the manuscript as well as Dolores Moran and Diane Wolf for their library assistance, and to James Newman, MD, rheumatologist, for his expert suggestions. Funding for this project was provided by a grant from the Delaware Department of Health and Social Services, Division of Public Health.of Health and Social Services, Division of Public Health.

Corresponding author
Anna Quisel, MD, c/o Cheryl Mongillo, Department of Family and Community Medicine, Christiana Care Health Systems, 1401 Foulk Road, Wilmington, DE 19803. E-mail: [email protected].

When a patient complains of pain “all over,” consider fibromyalgia, which typically causes a well-documented pattern of pain and characteristic points of tenderness observable on physical exam. Once alternative diagnoses have been ruled out, offer the patient a 2-pronged therapeutic regimen that has proven successful at moderately relieving symptoms.

First rule out concomitant or mimicking disorders

Consider the differential diagnosis carefully.¹ A person who meets the criteria for fibromyalgia may have yet another cause of chronic pain, such as rheumatoid arthritis, or may instead have a different treatable condition that mimics fibromyalgia.

Drug-induced myopathy. Pain suggestive of fibromyalgia should prompt a review of the patient’s medicines. Drug-induced myopathy may occur in persons taking colchicine, statins, corticosteroids, or antimalarial drugs.

Connective tissue, autoimmune, and rheumatologic disorders. Consider this group of disorders next. In 1 study, one fourth of persons referred to a rheumatology clinic with presumed fibromyalgia instead had a spondyloarthropathy.²

Dermatomyositis and polymyositis may present with muscle pain and tenderness but, unlike fibromyalgia, cause proximal muscle weakness.

Systemic lupus erythematosus, rheumatoid arthritis, and polymyalgia rheumatica can also lead to widespread pain.

Blood tests such as antinuclear antibody (ANA), C-reactive protein, or erythrocyte sedimentation rate (ESR) may prove helpful when a patient has a history of unexplained rashes, fever, weight loss, joint swelling, iritis, hepatitis, nephritis, or inflammatory back pain (onset before age 40, insidious onset, present for more than 3 months, associated with morning stiffness, improvement with exercise).³ In the absence of these signs, ANA, rheumatoid factor, and ESR testing in persons with fatigue and diffuse musculoskeletal pain have low positive predictive value.⁴ The rate of false-positive ANA results may be as high as 8% to 11%, especially at low titers.^5,6

Hypothyroidism. Widespread musculoskeletal pain has also been associated with hypothyroidism (level of evidence [LOE]: 2, case-control design),^7,8 supporting the inclusion of a thyroidstimulating hormone in the work-up of fibromyalgia (strength of recommendation [SOR]: B). More recent research suggests that musculoskeletal pain is more related to thyroid microsomal antibodies than to hypothyroidism,⁹ but there has been no further evaluation of antithyroid antibodies in persons with fibromyalgia.

Diagnosis: mostly by clinical judgment

Persons with fibromyalgia have widespread pain, often worst in the neck and trunk.¹ Additional symptoms include fatigue, morning stiffness, waking unrefreshed, paresthesias, and headache.^1,10-15 (See “The toll of fibromyalgia.”)

Accepted criteria

The diagnosis of fibromyalgia is based on 2 criteria:

1. A patient’s report of widespread pain (right and left sides of the body, above and below the waist, and including the axial skeleton) persisting for at least 3 months

2. The clinician’s identification of at least 11 of 18 potential tender points as specified in the American College of Rheumatology (ACR) 1990 Criteria for the Classification of Fibromyalgia (Figure) (LOE: 3, case-control design, nonindependent reference standard).¹

These criteria do not exclude persons with rheumatic diseases or other chronic pain conditions.^1,37-39

Caveats with the criteria

Despite these well-defined criteria, the diagnosis is not as clear-cut as it may appear. In 1990, the ACR convened a panel of 24 experts to define and standardize the diagnosis of fibromyalgia. The basis for this consensus was a group of 293 patients with fibromyalgia, each of whom had been assessed by one of the expert investigators according to “his or her usual method of diagnosis.” ¹

The investigators determined the unique characteristics of fibromyalgia by comparing the 293 cases to 265 controls who had other chronic pain conditions (eg, low back pain syndromes, neck pain syndromes, regional tendonitis, possible systemic lupus erythematosus, rheumatoid arthritis). The investigators considered a multitude of symptoms and signs including sleep disturbance, morning stiffness, paresthesias, irritable bowel syndrome, fatigue, and anxiety. Their conclusion was that widespread pain and tender points were the most sensitive (88.4%) and specific (81.1%) distinguishing criteria for fibromyalgia.¹

No reference standard. However, these calculations of sensitivity and specificity are less meaningful than in studies where an independent reference standard or gold standard is available. The ACR expert panel derived the criteria in a circular way using a nonindependent reference standard—ie, patients thought to have fibromyalgia compared with control patients thought not to have fibromyalgia. The expert panel essentially set the specificity of the criteria at 100%, since the specificity is based on the rate of false positives.

Furthermore, because there was no objective gold standard for determining who truly had fibromyalgia (and we do not yet have an independent biologic “test” for this condition), this panel could not determine whether additional symptoms or signs that should be considered in the diagnosis of fibromyalgia.

Biases, dubious representation? Unknown elements in this analysis are 1) how closely the reference population used to develop these criteria represents the true population of persons with fibromyalgia, and 2) the biases of the ACR experts. Finally, the positive and negative predictive values of these criteria will depend on the prevalence of fibromyalgia and other similar conditions.

Morbidity not predicted by criteria. In addition, the 1990 ACR criteria assume the number of tender points and degree of pain are directly proportional to overall morbidity; however, a person with fewer than 11 tender points may experience significant morbidity, indicating that the sensitivity of the criteria may be low.^40-42 As suggested by Wolfe in 1997, “the tender point count functions as a sedimentation rate for distress” in persons with chronic pain.⁴² Thus, the authors of the 1990 ACR study stated that ACR criteria should not be applied rigidly in diagnosing and treating fibromyalgia,⁴² leaving a large role for clinical judgment.

Subjective factors. A final difficulty with the diagnosis of fibromyalgia is its dependence on patient report and examiner technique.¹ In the 1990 ACR criteria, tender points were defined as a complaint of pain (or any more dramatic response) when an examiner applied 4 kg of pressure with the pulp of the thumb or first two or three fingers, calibrated with a dolorimeter (a device that can measure the amount and rate of pressure applied over a specified surface area).¹ It has been shown that practitioners require training to apply 4 kg of force with regularity.⁴³

However, applying exactly 4 kg of pressure may not be clinically important. Other studies have shown that finger palpation or dolorimetry identifies tender points with equal accuracy (LOE: 3, case-control design with non-independent reference standard).^44,45

Manual palpation

A controlled study of manual palpation was conducted to standardize the tender point survey described in the Figure. ⁴⁶ This method compares well with the ACR 1990 method, with a sensitivity of 88.6% and a specificity of 71.4%.⁴⁶

To speed up the examination, a particular sequence of palpating survey points was established, with the patient positioned as outlined in the Figure. Using the thumb pad of his/her dominant hand, each examiner applied 4 kg of pressure, at a rate of 1 kg per second, just once at each survey point. Examiners learned to apply the proper amount of pressure by standing a patient on a scale and watching the scale while pressing down perpendicularly on the trapezius survey point.

The examinee was seated throughout the exam, except when lying on the side for palpation of the trochanter and lying supine for palpation of the knee. A tender point was identified when the patient rated the pain resulting from palpation at least 2 out of 10 (0, no pain; 10 worst pain) (LOE: 3, case-control design, nonindependent reference standard).⁴⁶

Until a firm biologic basis for fibromyalgia is discovered and a true gold standard for testing is developed, the diagnosis of fibromyalgia will remain a matter of clinical judgment and convention (SOR: C).

Treatment

A diagnosis of fibromyalgia alone may result in health benefits. In a year-long study published in 1986, Cathey et al reported that among 81 persons diagnosed with fibromyalgia, hospitalization rates decreased in the year following diagnosis (LOE: 2, case-control design).⁴⁷

Treatments for fibromyalgia are numerous, ranging from balneotherapy (bathing) to low-energy laser therapy, and studies of interventions are often poorly designed, based on small numbers of patients, report nonstandardized outcomes, and yield conflicting results.⁴⁸

Two interventions—aerobic exercise and antidepressant therapy—appear to improve fibromyalgia.

Aerobic exercise

Though pain relief is insignificant with aerobic exercise, other positive effects are significant (SOR: A). A 2003 Cochrane review identified 7 high-quality studies of aerobic training, defined as: 1) frequency of 2 days per week; 2) intensity sufficient to achieve 40% to 85% of heart rate reserve, or 55% to 90% of predicted maximum heart rate; 3) duration of sessions 20 to 60 minutes, either continuously or intermittently throughout the day, using any mode of aerobic exercise; and 4) a total exercise period of at least 6 weeks (Table 1).⁴⁹

Improved functioning, tender-point threshold. Study subjects engaged in aerobic dancing, whole-body aerobics, stationary cycling, and walking. Persons who exercised improved in global well-being, physical function, and aerobic fitness (by about 17%), and raised the pain threshold of tender points (by about 35%).⁴⁹ Four of the studies were similar enough to be combined for meta-analysis, showing a statistically robust but modest reduction in tender-point threshold (LOE: 1).

Although it seems likely that pain or fatigue might increase at least initially with exercise, participants in the exercise groups were not deterred; the researchers pointed out that reporting of adverse effects of aerobic exercise appeared incomplete, but there was no significant difference in drop-out rates between the exercise (25.1%) and control groups (12.5%).⁴⁹

In the long-term studies (>6 months), improvements were noted up to 1 year after treatment ended but not after 4.5 years.⁴⁹ This Cochrane review further supports aerobic exercise as bring beneficial for persons with fibromyalgia.^50,51

Additional improvement measures. A similar systematic review concluded that although studies were too heterogeneous to draw final conclusions, preliminary data supported aerobic exercise (LOE: 2, with heterogeneous studies).⁵⁰ In another comprehensive meta-analysis of all treatments for fibromyalgia, heterogeneous treatment studies ranging from exercise to physical therapy were identified as physically-based treatments. The analysis revealed a positive effect on physical status (including tender-point index, grip strength, and physician global rating of pain symptoms), fibromyalgia symptoms (including self-reported fatigue and pain using visual analog scales), and psychological status (including measurements of the Hamilton Depression and Anxiety Scales), with no effect on daily functioning (including outcome measures such as the Fibromyalgia Impact Questionnaire [FIQ]) (LOE: 2, with heterogeneous studies).⁵¹

The authors noted that the magnitude of the positive effects of physically-based treatments on fibromyalgia were comparable with drug treatment judged effective for arthritis.⁵¹

TABLE 1
Aerobic exercise for fibromyalgia: the evidence

Less certain nonpharmacologic therapies

Other nonpharmacologic treatments for fibromyalgia are educational interventions, relaxation therapy, cognitive-behavioral therapy, and acupuncture. These therapies have undergone rigorous analysis, but studies have been too heterogeneous to allow for strong conclusions across the studies.⁵⁰

A recent Cochrane review concluded that although physical training plus education had a positive effect at long-term follow up, evidence is insufficient to recommend multidisciplinary rehabilitation, defined as the care of a physician plus psychological, social, and vocational interventions (SOR: C).⁵²

In contrast, other investigators have concluded that multidisciplinary treatment incorporating physically and psychologically based treatments was more successful than treatment with a single modality.⁵¹ A systematic review of acupuncture identified only 1 high-quality randomized controlled trial (Table 2), which did show some improvement in symptoms (SOR: C).⁵³

TABLE 2
Alternative nonpharmacologic therapies for fibromyalgia: the evidence

Therapy with antidepressants

Of all pharmacologic treatments, antidepressants have undergone the most thorough study. Although the optimal role of medications in fibromyalgia has not been delineated, 3 metaanalyses have reported that antidepressants, most commonly amitriptyline, reduce symptoms during treatment of a few months duration (SOR: A) (Table 3).^54,55

Any antidepressants. Pooled results from 13 studies (8 of tricyclics, 3 of selective serotonin reuptake inhibitors, 2 of s-adenosylmethionine) revealed a moderate effect on pain, sleep, and global well-being, and a mild effect on fatigue and number of trigger points.⁵⁴ The authors calculated that persons with fibromyalgia treated with antidepressants were 4 times more likely to improve than persons treated with placebo (number need to treat [NNT]=4). Adverse effects appeared insignificant but were poorly reported in the individual studies.

Tricyclics only. In another meta-analysis, 9 high-quality studies of tricyclic antidepressants (amitriptyline, dothiepin, clomipramine, maprotiline and cyclobenzaprine—classified by the authors as a tricyclic antidepressant) were analyzed for 7 outcomes (patient self-rating of pain, fatigue, stiffness, and sleep; the patient and physician global assessment of improvement; and tenderness of tender points). Significant responses were observed in 25% to 37% of patients. On meta-analysis, outcome measures improved moderately overall with tricyclic treatment, mostly in sleep and global assessment, least in stiffness and tenderness. Long-term safety (more than 8 weeks) and efficacy of tricyclic therapy for fibromyalgia have not been demonstrated.⁵⁵

Combined trials.A third meta-analysis demonstrated improvement when trials of different antidepressants were combined.⁵¹ By pooling studies of antidepressants (amitriptyline, dothiepin, fluoxetine, citalopram, and S-adenosylmethionine) improvements in physical status, fibromyalgia symptoms, and psychological status were found, with no improvement in daily functioning.⁵¹ Although the effect was smaller than physicallybased treatments, the effect size was still comparable to drug treatment for arthritis.⁵¹

Muscle relaxants (primarily cyclobenzaprine) and nonsteroidal anti-inflammatories have been studied, with no evidence of a positive effect.⁵¹ Thus, the best evidence currently supports the use of aerobic exercise and antidepressants, particularly tricyclics, for the treatment of fibromyalgia.

TABLE 3
Antidepressant therapy for fibromyalgia: the evidence

Instructions to patients, management follow-up

Persons with fibromyalgia should know that although specific symptoms, particularly pain, may be not be dramatically reduced with treatment, aerobic exercise and tricyclic antidepressants alleviate some symptoms with minimal adverse effects (SOR: A). Emphasize that these treatments have been shown to improve one’s ability to cope with fibromyalgia symptoms. The best-studied antidepressant for treating fibromyalgia is amitriptyline, usually given at 25 to 50 mg, nightly.

Exercise. Prescribe aerobic exercise, at least twice per week for 20 to 60 minutes, targeting a heart rate of 55% to 90% of the predicted maximum (180 beats per minute-age) (SOR: A). One caveat: aerobic exercise in the literature was usually supervised, so the ideal exercise regimen might be a fibromyalgia-specific program.

Medication. Consider a trial of amitriptyline, 25 to 50 mg every night, for up to 6 weeks (SOR A). A caveat: tricyclic antidepressants may also have significant side effects, which could outweigh moderate benefits. Moreover, treatment effectiveness beyond 2 months has not been proven. Therefore, longitudinal measurement of outcomes should be part of ongoing care.

Follow-up. Studies have not determined which measures are best to follow (see “Assessing treatment efficacy”), but they might include the following (SOR: C):

1. Pain (eg, visual analogue scale, pain drawings)
2. Number of tender points, and tenderness
3. Physical function (eg, cardiorespiratory fitness, self-reported physical function measured by the physical-impairment subscale of the FIQ,⁵⁶ strength)
4. Global well-being or perceived improvement (eg, physician-rated change, FIQ total score)
5. Self-efficacy (eg, Arthritis Self-efficacy Questionnaire)
6. Fatigue and sleep (eg, FIQ fatigsubscale, sleep visual analogue scale)
7. Psychological function (eg, FIQ subscales for depression and anxiety)
8. Ability to work
9. Health care consumption and costs.^47,50

Education or psychological coping strategies may also contribute positively to overall patient and family well-being. Consider education/psychological counseling (SOR: C) and acupuncture (SOR: B).

Acknowledgments

The authors would like to express their appreciation to Cheryl Mongillo, Peggy Lardear, and Brian Pellini for their assistance in preparing the manuscript as well as Dolores Moran and Diane Wolf for their library assistance, and to James Newman, MD, rheumatologist, for his expert suggestions. Funding for this project was provided by a grant from the Delaware Department of Health and Social Services, Division of Public Health.of Health and Social Services, Division of Public Health.

Corresponding author
Anna Quisel, MD, c/o Cheryl Mongillo, Department of Family and Community Medicine, Christiana Care Health Systems, 1401 Foulk Road, Wilmington, DE 19803. E-mail: [email protected].

Despite the introduction of new anticonvulsant drugs, phenytoin is still a first-line medication for common types of epileptic seizures, particularly those caused by focal brain lesions.^1-4 Available in parenteral and oral form, phenytoin (or its pro-drug, fosphenytoin) is widely used. An estimated 873,000 prescriptions for phenytoin were issued during office visits in 2001.⁵

Phenytoin carries a special risk of dose-related toxicity, due to its saturation (zero-order) pharmacokinetics: serum levels often rise much more than would ordinarily be expected after initiating or increasing a maintenance dose. This predicts a vulnerability to toxicity, but does not predict exactly when this will occur in the individual.

The risk of toxicity can be minimized, however, by applying practical dosing and monitoring strategies based on the understanding of phenytoin pharmacokinetics, and by educating patients appropriately.

Patients at risk: the scope of the problem

Extrapolating from the more than 5000 hospitals in the US⁶ to our experience in an urban community hospital, we estimate there may be as many as 25,000 cases of phenytoin intoxication presenting annually to emergency departments or resulting in hospitalization in the United States. In 1 study, a tertiary hospital recorded phenytoin intoxication from all causes at a rate of 1 inpatient admission per month over a 10-year period.⁷ Another study at a major hospital found 143 instances of phenytoin levels >25 μg/mL in 1 year; 86% of 120 studied cases were toxic, representing 33% of all adverse drug reactions reported.⁸ Thus, evidence points to a substantial problem with patient safety nationwide.

Adverse drug events like phenytoin intoxication increase morbidity, causing such injuries as falls due to ataxia and resulting in expenses of office or emergency department visits and hospitalization. While no prescription strategy, system of monitoring, or “safety net” is likely to eliminate phenytoin mishaps, an informed and active approach to therapeutic management can minimize instances of intoxication.

Action points and safety tips in phenytoin therapy

Safe therapy with phenytoin depends on observing particular courses of action at 4 stages:

1. Loading
2. Institution of a maintenance regimen
3. Adjustment of the regimen
4. Monitoring, follow-up, and patient education.

Loading

Loading is indicated when the risk of seizures is so great that adequate serum levels of the drug must be reached rapidly. Such situations would include status epilepticus; repeated new seizures (excluding most withdrawal seizures, for example); breakthrough seizures with a low anticonvulsant level; and a first seizure with a high likelihood of repeating, as with a demonstrated focal brain lesion. Depending on the degree of urgency, loading can be accomplished with intravenous phenytoin (at an infusion rate of no more than 50 mg/min), with intravenous or intramuscular fosphenytoin, or with oral phenytoin.

To load initially, or to add a supplemental load to increase an insufficient phenytoin level, the following formula based on a distribution constant for phenytoin indicates the amount of drug needed to raise the level by a specified amount.⁹

Use the loading formula. The peak serum level of phenytoin after intravenous loading is a function of the drug’s distribution in the body and is independent of the pharmacokinetics of elimination. Subsequent metabolism, which may be affected by other drugs or impairments (eg, liver disease), will affect elimination of the loaded drug but not ordinarily the calculated loading dose. Overloading phenytoin has been documented as a cause of early toxicity.⁷ According to the formula above, a 60-kg patient with no detectable starting level and an (arbitrary) target serum level of 15 μg/mL should need only 675 mg of phenytoin, and not the 1000 mg often administered.

This loading formula is also applicable to supplementary (“booster”) loading to reach a higher serum level quickly, either because the initial loading dose did not achieve the intended level or because that level was inadequate to control seizures. In this context, simply increasing the existing or planned daily maintenance dose raises the level too slowly. In addition, the increased maintenance dosage may be inappropriate if the cause of the low level is noncompliance. A higher maintenance dose, under conditions of complete or improved compliance, probably will lead to toxicity.

Measure serum levels. A sound preventive approach¹⁰ is to measure the peak serum level shortly after loading^11,12—one-half hour to 1 hour or more after giving intravenous phenytoin, 2 hours or more after intravenous fosphenytoin, 4 hours or more after intramuscular fosphenytoin, and 16 to 24 hours after accelerated oral loading. While measuring a post-load serum level is not established as a standard of care, the rationale is that a relatively high serum level forewarns of increased risk of early intoxication because of a high starting point for maintenance therapy, and a low level indicates greater vulnerability to seizures (Table 1).

TABLE 1
Preventing phenytoin intoxication at loading (independent of pharmacokinetics of elimination)

Initial maintenance dosing

A useful maintenance dose formula yields the dose ordinarily needed to achieve a specified serum level or maintain it after loading:⁹

For a target maintenance level of 15 μg/mL in a 60-kg adult, the dose would be 5.71 mg/kg/d x 60 kg = 343 mg/d (which can guide selection of a practical, starting dosage regimen, such as 300 or 350 mg/d, or 5–6 mg/kg/d, as is often recommended).¹³ This formula is more accurate than guessing at 5 mg/kg vs 6 mg/kg; increasingly, such formulas will be incorporated into computerized dosing protocols, thus putting the advised dose only a click or 2 away.

Even calculated dosages should be subject to modification by individual patient factors, including age, reliability, health status (such as liver function), and potential medication interactions. Computerized protocols will help, but the uncertainty of individual responses¹⁴ simply means that close symptomatic or serum monitoring must be implemented, while not overreacting to isolated variations (Table 2).¹⁵

Important caveat. Phenytoin is typically 90% protein-bound in the serum. Active free phenytoin may be higher than expected if serum albumin is decreased or if bound phenytoin is displaced by other drugs (eg, valproate). Thus, toxicity may be present despite non-elevated total levels of phenytoin, and successive increases in the dosage may cause or exacerbate toxicity. Consider obtaining a free phenytoin level, commonly available by specific requisition, if clinical toxicity is suspected despite total levels that do not suggest toxicity. Consultation or careful review of all potential metabolic interactions helps to ensure proper management in such cases.

Adjusting dosage and the maintenance