Clinical Inquiries

Evidence summary

Critically ill patients are at increased risk of bleeding from stress-induced gastroduodenal ulceration. Decades ago, ICUs began using pharmacologic prophylaxis on most patients to prevent gastrointestinal bleeding, which had a mortality rate as high as 80%. Before the advent of prophylaxis, the incidence of upper gastro-intestinal bleeding was 6% to 25%.⁴ Since then, improvements in ICU management have decreased this incidence to 0% to 2.8%.⁵ Recent studies suggest that only ICU patients with certain risk factors benefit from ulcer prophylaxis (TABLE).¹

Our search retrieved 20 randomized controlled trials and 6 systematic reviews with meta-analyses from the Medline database since 1990. It was difficult to find a consensus on the matter of stress ulcer prophylaxis because of inconsistencies in the outcomes measured in these studies. We focused on studies examining clinically important bleeding, but even in these studies definitions and measurements vary. Few studies addressed mortality or length of stay; those that did reported no significant difference in either outcome with prophylaxis.

Medications used to prevent gastrointestinal bleeding have included antacids, sucralfate, H2RAs, and PPIs. Sucralfate and H2RAs have been studied most frequently, and both agents significantly reduce the incidence of clinically important bleeding in high-risk patients. Compared with placebo, the odds ratio for clinically important bleeding was 0.44 with ranitidine (95% confidence interval [CI], 0.22–0.88) and 0.58 with sucralfate (95% CI, 0.34–0.99).⁶ In a population with a clinically important bleeding incidence of 3% to 6%, a range consistent with the most recent studies we reviewed, the number needed to treat to prevent 1 bleeding episode is 30 to 60 for ranitidine and 40 to 79 for sucralfate.

Some studies suggest that pharmacologic prophylaxis may increase the incidence of aspiration pneumonia in ventilator-dependent patients. The largest randomized trial addressing this issue (N=1200) found no significant difference between H2RAs and sucralfate in ventilator-associated pneumonia.³ Improved ICU management, such as frequent suctioning, upright positioning, and use of enteral nutrition may help prevent nosocomial pneumonia due to aspiration.

TABLE
Risk factors for stress ulcers

Recommendation from others

In the American Journal of Health-System Pharmacy, Allen et al⁵ state “the frequency of clinically important bleeding is low … the majority of recently published prospective studies and meta-analyses have been unable to demonstrate a reduction in clinically important bleeding with pharmacologic agents.” A 2001 Agency for Healthcare Research and Quality evidence report⁷ states that the evidence is not conclusive that all intensive care patients benefit from stress ulcer prophylaxis and that clinicians “may consider use of prophylactic agents in very high risk patients.”

Evidence summary

Critically ill patients are at increased risk of bleeding from stress-induced gastroduodenal ulceration. Decades ago, ICUs began using pharmacologic prophylaxis on most patients to prevent gastrointestinal bleeding, which had a mortality rate as high as 80%. Before the advent of prophylaxis, the incidence of upper gastro-intestinal bleeding was 6% to 25%.⁴ Since then, improvements in ICU management have decreased this incidence to 0% to 2.8%.⁵ Recent studies suggest that only ICU patients with certain risk factors benefit from ulcer prophylaxis (TABLE).¹

Our search retrieved 20 randomized controlled trials and 6 systematic reviews with meta-analyses from the Medline database since 1990. It was difficult to find a consensus on the matter of stress ulcer prophylaxis because of inconsistencies in the outcomes measured in these studies. We focused on studies examining clinically important bleeding, but even in these studies definitions and measurements vary. Few studies addressed mortality or length of stay; those that did reported no significant difference in either outcome with prophylaxis.

Medications used to prevent gastrointestinal bleeding have included antacids, sucralfate, H2RAs, and PPIs. Sucralfate and H2RAs have been studied most frequently, and both agents significantly reduce the incidence of clinically important bleeding in high-risk patients. Compared with placebo, the odds ratio for clinically important bleeding was 0.44 with ranitidine (95% confidence interval [CI], 0.22–0.88) and 0.58 with sucralfate (95% CI, 0.34–0.99).⁶ In a population with a clinically important bleeding incidence of 3% to 6%, a range consistent with the most recent studies we reviewed, the number needed to treat to prevent 1 bleeding episode is 30 to 60 for ranitidine and 40 to 79 for sucralfate.

Some studies suggest that pharmacologic prophylaxis may increase the incidence of aspiration pneumonia in ventilator-dependent patients. The largest randomized trial addressing this issue (N=1200) found no significant difference between H2RAs and sucralfate in ventilator-associated pneumonia.³ Improved ICU management, such as frequent suctioning, upright positioning, and use of enteral nutrition may help prevent nosocomial pneumonia due to aspiration.

TABLE
Risk factors for stress ulcers

Recommendation from others

In the American Journal of Health-System Pharmacy, Allen et al⁵ state “the frequency of clinically important bleeding is low … the majority of recently published prospective studies and meta-analyses have been unable to demonstrate a reduction in clinically important bleeding with pharmacologic agents.” A 2001 Agency for Healthcare Research and Quality evidence report⁷ states that the evidence is not conclusive that all intensive care patients benefit from stress ulcer prophylaxis and that clinicians “may consider use of prophylactic agents in very high risk patients.”

Evidence summary

Critically ill patients are at increased risk of bleeding from stress-induced gastroduodenal ulceration. Decades ago, ICUs began using pharmacologic prophylaxis on most patients to prevent gastrointestinal bleeding, which had a mortality rate as high as 80%. Before the advent of prophylaxis, the incidence of upper gastro-intestinal bleeding was 6% to 25%.⁴ Since then, improvements in ICU management have decreased this incidence to 0% to 2.8%.⁵ Recent studies suggest that only ICU patients with certain risk factors benefit from ulcer prophylaxis (TABLE).¹

Our search retrieved 20 randomized controlled trials and 6 systematic reviews with meta-analyses from the Medline database since 1990. It was difficult to find a consensus on the matter of stress ulcer prophylaxis because of inconsistencies in the outcomes measured in these studies. We focused on studies examining clinically important bleeding, but even in these studies definitions and measurements vary. Few studies addressed mortality or length of stay; those that did reported no significant difference in either outcome with prophylaxis.

Medications used to prevent gastrointestinal bleeding have included antacids, sucralfate, H2RAs, and PPIs. Sucralfate and H2RAs have been studied most frequently, and both agents significantly reduce the incidence of clinically important bleeding in high-risk patients. Compared with placebo, the odds ratio for clinically important bleeding was 0.44 with ranitidine (95% confidence interval [CI], 0.22–0.88) and 0.58 with sucralfate (95% CI, 0.34–0.99).⁶ In a population with a clinically important bleeding incidence of 3% to 6%, a range consistent with the most recent studies we reviewed, the number needed to treat to prevent 1 bleeding episode is 30 to 60 for ranitidine and 40 to 79 for sucralfate.

Some studies suggest that pharmacologic prophylaxis may increase the incidence of aspiration pneumonia in ventilator-dependent patients. The largest randomized trial addressing this issue (N=1200) found no significant difference between H2RAs and sucralfate in ventilator-associated pneumonia.³ Improved ICU management, such as frequent suctioning, upright positioning, and use of enteral nutrition may help prevent nosocomial pneumonia due to aspiration.

TABLE
Risk factors for stress ulcers

Recommendation from others

In the American Journal of Health-System Pharmacy, Allen et al⁵ state “the frequency of clinically important bleeding is low … the majority of recently published prospective studies and meta-analyses have been unable to demonstrate a reduction in clinically important bleeding with pharmacologic agents.” A 2001 Agency for Healthcare Research and Quality evidence report⁷ states that the evidence is not conclusive that all intensive care patients benefit from stress ulcer prophylaxis and that clinicians “may consider use of prophylactic agents in very high risk patients.”

Evidence summary

Dietary changes are recommended as first-line treatment for mild to moderate dyslipidemia. We examined evidence on 5 common dietary interventions for adults with dyslipidemia. The average effects on lipid levels are reported in the TABLE.

TABLE
Average effect of dietary interventions on serum lipid levels

Low-fat

A meta-analysis of 37 mostly good-quality controlled trials evaluated the former National Cholesterol Education Program (NCEP) Step I and Step II diets in 11,586 participants.¹ The Step I diet restricted in-take of total fat (≤30% of total calories), saturated fat (≤10% of total calories), and cholesterol (≤300 mg/d). Step II goals were lower for saturated fat (<7%) and cholesterol (<200 mg/d). Mean baseline lipid values (mg/dL) were total cholesterol, 233.57; LDL, 155.10; HDL, 47.95; and triglycerides, 147.91. Both of these low-fat diets significantly reduced total cholesterol, LDL, and triglycerides. The Step II diet also reduced HDL.

Soy

A meta-analysis of 23 good-quality controlled trials with 1381 participants reported that soy protein with naturally occurring isoflavones significantly reduced total cholesterol, LDL, and triglycerides while significantly increasing HDL.² The amount of soy isoflavone consumed varied across studies. One subgroup analysis showed that consumption of >80 mg/d was associated with a better effect on lipids. In subjects with baseline hypercholesterolemia (total cholesterol >240 mg/dL), greater reductions in total cholesterol, and greater increases in HDL were reported, with comparable changes in LDL and triglycerides.

Soluble fiber

A meta-analysis of 67 good-quality RCTs evaluated the effects of soluble dietary fiber in 2990 subjects (mean baseline lipid values [mg/dL]: total cholesterol, 240.9; LDL, 164.4).³ Diets high in soluble fiber (average dose of 9.5 g/d) were associated with a statistically significant decrease in total cholesterol and LDL and no significant change in HDL or triglycerides. Type of fiber (oat, psyllium, or pectin) was not influential after controlling for initial lipid level.

“Portfolio” diet

A fair-quality randomized crossover study with 34 participants found that a “portfolio diet,” which combines the fat intake of the NCEP Step II diet with cholesterol-lowering “functional foods” (including plant sterols, nuts, soluble fibers, and soy protein), markedly reduced total cholesterol and LDL.⁴ Mean baseline lipid values (mg/dL) were: total cholesterol, 261.41; LDL, 174.40; HDL, 47.56; triglycerides, 199.28.

Mediterranean diet

A fair-quality RCT with 88 participants reported reduced LDL among subjects assigned to a Mediterranean-type diet.⁵ Mean baseline lipid values (mg/dL) were total cholesterol, 255.22; LDL, 170.15; HDL, 58.01; triglycerides, 141.71.

Recommendations from others

The NCEP Adult Treatment Panel III and the American Heart Association recommend the Therapeutic Lifestyle Changes diet.^6,7 The first stage of this diet emphasizes reduction in dietary saturated fat and cholesterol at the levels of the former NCEP Step II diet (≤7% of energy as saturated fat and ≤200 mg dietary cholesterol). If the LDL goal is not achieved, the second stage emphasizes the addition of functional foods and soluble fiber.

Evidence summary

Dietary changes are recommended as first-line treatment for mild to moderate dyslipidemia. We examined evidence on 5 common dietary interventions for adults with dyslipidemia. The average effects on lipid levels are reported in the TABLE.

TABLE
Average effect of dietary interventions on serum lipid levels

Low-fat

A meta-analysis of 37 mostly good-quality controlled trials evaluated the former National Cholesterol Education Program (NCEP) Step I and Step II diets in 11,586 participants.¹ The Step I diet restricted in-take of total fat (≤30% of total calories), saturated fat (≤10% of total calories), and cholesterol (≤300 mg/d). Step II goals were lower for saturated fat (<7%) and cholesterol (<200 mg/d). Mean baseline lipid values (mg/dL) were total cholesterol, 233.57; LDL, 155.10; HDL, 47.95; and triglycerides, 147.91. Both of these low-fat diets significantly reduced total cholesterol, LDL, and triglycerides. The Step II diet also reduced HDL.

Soy

A meta-analysis of 23 good-quality controlled trials with 1381 participants reported that soy protein with naturally occurring isoflavones significantly reduced total cholesterol, LDL, and triglycerides while significantly increasing HDL.² The amount of soy isoflavone consumed varied across studies. One subgroup analysis showed that consumption of >80 mg/d was associated with a better effect on lipids. In subjects with baseline hypercholesterolemia (total cholesterol >240 mg/dL), greater reductions in total cholesterol, and greater increases in HDL were reported, with comparable changes in LDL and triglycerides.

Soluble fiber

A meta-analysis of 67 good-quality RCTs evaluated the effects of soluble dietary fiber in 2990 subjects (mean baseline lipid values [mg/dL]: total cholesterol, 240.9; LDL, 164.4).³ Diets high in soluble fiber (average dose of 9.5 g/d) were associated with a statistically significant decrease in total cholesterol and LDL and no significant change in HDL or triglycerides. Type of fiber (oat, psyllium, or pectin) was not influential after controlling for initial lipid level.

“Portfolio” diet

A fair-quality randomized crossover study with 34 participants found that a “portfolio diet,” which combines the fat intake of the NCEP Step II diet with cholesterol-lowering “functional foods” (including plant sterols, nuts, soluble fibers, and soy protein), markedly reduced total cholesterol and LDL.⁴ Mean baseline lipid values (mg/dL) were: total cholesterol, 261.41; LDL, 174.40; HDL, 47.56; triglycerides, 199.28.

Mediterranean diet

A fair-quality RCT with 88 participants reported reduced LDL among subjects assigned to a Mediterranean-type diet.⁵ Mean baseline lipid values (mg/dL) were total cholesterol, 255.22; LDL, 170.15; HDL, 58.01; triglycerides, 141.71.

Recommendations from others

The NCEP Adult Treatment Panel III and the American Heart Association recommend the Therapeutic Lifestyle Changes diet.^6,7 The first stage of this diet emphasizes reduction in dietary saturated fat and cholesterol at the levels of the former NCEP Step II diet (≤7% of energy as saturated fat and ≤200 mg dietary cholesterol). If the LDL goal is not achieved, the second stage emphasizes the addition of functional foods and soluble fiber.

Evidence summary

Dietary changes are recommended as first-line treatment for mild to moderate dyslipidemia. We examined evidence on 5 common dietary interventions for adults with dyslipidemia. The average effects on lipid levels are reported in the TABLE.

TABLE
Average effect of dietary interventions on serum lipid levels

Low-fat

A meta-analysis of 37 mostly good-quality controlled trials evaluated the former National Cholesterol Education Program (NCEP) Step I and Step II diets in 11,586 participants.¹ The Step I diet restricted in-take of total fat (≤30% of total calories), saturated fat (≤10% of total calories), and cholesterol (≤300 mg/d). Step II goals were lower for saturated fat (<7%) and cholesterol (<200 mg/d). Mean baseline lipid values (mg/dL) were total cholesterol, 233.57; LDL, 155.10; HDL, 47.95; and triglycerides, 147.91. Both of these low-fat diets significantly reduced total cholesterol, LDL, and triglycerides. The Step II diet also reduced HDL.

Soy

A meta-analysis of 23 good-quality controlled trials with 1381 participants reported that soy protein with naturally occurring isoflavones significantly reduced total cholesterol, LDL, and triglycerides while significantly increasing HDL.² The amount of soy isoflavone consumed varied across studies. One subgroup analysis showed that consumption of >80 mg/d was associated with a better effect on lipids. In subjects with baseline hypercholesterolemia (total cholesterol >240 mg/dL), greater reductions in total cholesterol, and greater increases in HDL were reported, with comparable changes in LDL and triglycerides.

Soluble fiber

A meta-analysis of 67 good-quality RCTs evaluated the effects of soluble dietary fiber in 2990 subjects (mean baseline lipid values [mg/dL]: total cholesterol, 240.9; LDL, 164.4).³ Diets high in soluble fiber (average dose of 9.5 g/d) were associated with a statistically significant decrease in total cholesterol and LDL and no significant change in HDL or triglycerides. Type of fiber (oat, psyllium, or pectin) was not influential after controlling for initial lipid level.

“Portfolio” diet

A fair-quality randomized crossover study with 34 participants found that a “portfolio diet,” which combines the fat intake of the NCEP Step II diet with cholesterol-lowering “functional foods” (including plant sterols, nuts, soluble fibers, and soy protein), markedly reduced total cholesterol and LDL.⁴ Mean baseline lipid values (mg/dL) were: total cholesterol, 261.41; LDL, 174.40; HDL, 47.56; triglycerides, 199.28.

Mediterranean diet

A fair-quality RCT with 88 participants reported reduced LDL among subjects assigned to a Mediterranean-type diet.⁵ Mean baseline lipid values (mg/dL) were total cholesterol, 255.22; LDL, 170.15; HDL, 58.01; triglycerides, 141.71.

Recommendations from others

The NCEP Adult Treatment Panel III and the American Heart Association recommend the Therapeutic Lifestyle Changes diet.^6,7 The first stage of this diet emphasizes reduction in dietary saturated fat and cholesterol at the levels of the former NCEP Step II diet (≤7% of energy as saturated fat and ≤200 mg dietary cholesterol). If the LDL goal is not achieved, the second stage emphasizes the addition of functional foods and soluble fiber.

Evidence summary

Diabetes mellitus and osteoarthritis commonly overlap in patients, many of whom have looked to the nutritional supplement combination of glucosamine and chondroitin sulfates for pain relief. The effectiveness of these supplements in improving patient-oriented outcomes for osteoarthritis is still being evaluated. However, regardless of their effectiveness they remain popular supplements, representing up to a third of the specialized supplement market in the US.¹

The mechanism by which glucosamine is hypothesized to affect blood glucose involves the roles of glucosamine and the hexosamine biosynthesis pathway in the regulation of glucose transport. Overexpression of enzymes involved in this pathway have led to high levels of glucose and insulin resistance in animals given huge doses of intravenous glucosamine (100–200 times higher than oral therapeutic doses in humans).² Studies have specifically investigated the effects of intravenous glucosamine infusion in healthy humans, and it did not show any effect on insulin sensitivity or plasma glucose.

A Cochrane systematic review of 20 randomized controlled trials (RCTs) including 2570 patients in order to evaluate the effectiveness and toxicity of glucosamine in osteoarthritis found that glucosamine was as safe as placebo in terms of adverse reactions. However, they did not comment specifically on diabetic patients or hyperglycemia per se.³

An RCT published in 2004 tested whether glucose intolerance occurs when healthy adults consume normal, recommended dosages of glucosamine sulfate. Nineteen healthy adults were randomized to receive either 1500 mg of glucosamine sulfate or placebo orally each day for 12 weeks. Three-hour oral glucose tolerance tests were performed using 75 g of dextrose. These occurred before the study, at 6 weeks, and at 12 weeks. There were no significant differences between fasting levels of serum insulin or blood glucose.⁴ Glucosamine sulfate supplementation did not alter serum insulin or plasma glucose during the tests. Limitations to the study include the small number of subjects, the short duration period, and the fact that the tests were performed before the patient’s daily glucosamine dosing.

A recent study examined insulin and glucose levels with and without the simultaneous ingestion of 1500 mg of glucosamine. Sixteen fasting volunteers with osteoarthritis but without known diabetes or glucose intolerance received 7 g of glucose with or without ingestion of 1500 mg glucosamine sulfate. The authors unexpectedly uncovered undiagnosed diabetes or impaired glucose tolerance in 3 subjects. These 3 subjects showed a statistically significant (P=.04) 31% increase in the area under the curve of glucose levels following the test. There was no effect of glucosamine sulfate ingestion on patients with normal baseline glucose testing or on insulin levels. Their results might be important since they are the first to suggest that glucosamine ingestion may affect glucose levels in individuals who have untreated diabetes or glucose intolerance.⁵

One double-blinded RCT evaluated whether oral glucosamine supplementation altered glycosylated hemoglobin (HbA_1c) concentrations for patients with well-controlled diabetes mellitus. Thirty-eight patients were randomized to receive either treatment with glucosamine/chondroitin at the recommended doses or placebo. After 3 months of treatment HbA_1c levels did not change and were not significantly different between groups (P=.2).⁶

Another study addressed whether glucosamine taken at recommended doses for the treatment of osteoarthritis had any detrimental effect on glucose metabolism. Fourteen patients participated and had a baseline 4-hour meal tolerance test and a frequently sampled intravenous glucose tolerance test, before and after 4 weeks of glucosamine sulfate treatment (500 mg orally 3 times daily). After 4 weeks they found no change in fasting plasma glucose, insulin, glucose tolerance, or difference in insulin sensitivity in the group of subjects.⁷ Again, the study was limited by a small subject number and short duration of study.

Recommendations from others

The PDR for Nonprescription Drugs and Dietary Supplements states that “glucosamine is likely safe for patients with diabetes that is well controlled with diet only or with one or two oral antidiabetic agents (HbA_1c less than 6.5%). For patients with higher HbA_1c concentrations or for those requiring insulin, closely monitor blood glucose concentrations.”⁸

The American Pain Society encourages adults with osteoarthritis to take 1500 mg of glucosamine daily as a dietary supplement but does not specifically recommend it as pharmacologic management for pain.⁹ The American College of Rheumatology Subcommittee on Osteoarthritis has no recommendations regarding the use of glucosamine or chondroitin in the treatment of knee osteoarthritis.¹⁰

Evidence summary

Diabetes mellitus and osteoarthritis commonly overlap in patients, many of whom have looked to the nutritional supplement combination of glucosamine and chondroitin sulfates for pain relief. The effectiveness of these supplements in improving patient-oriented outcomes for osteoarthritis is still being evaluated. However, regardless of their effectiveness they remain popular supplements, representing up to a third of the specialized supplement market in the US.¹

The mechanism by which glucosamine is hypothesized to affect blood glucose involves the roles of glucosamine and the hexosamine biosynthesis pathway in the regulation of glucose transport. Overexpression of enzymes involved in this pathway have led to high levels of glucose and insulin resistance in animals given huge doses of intravenous glucosamine (100–200 times higher than oral therapeutic doses in humans).² Studies have specifically investigated the effects of intravenous glucosamine infusion in healthy humans, and it did not show any effect on insulin sensitivity or plasma glucose.

A Cochrane systematic review of 20 randomized controlled trials (RCTs) including 2570 patients in order to evaluate the effectiveness and toxicity of glucosamine in osteoarthritis found that glucosamine was as safe as placebo in terms of adverse reactions. However, they did not comment specifically on diabetic patients or hyperglycemia per se.³

An RCT published in 2004 tested whether glucose intolerance occurs when healthy adults consume normal, recommended dosages of glucosamine sulfate. Nineteen healthy adults were randomized to receive either 1500 mg of glucosamine sulfate or placebo orally each day for 12 weeks. Three-hour oral glucose tolerance tests were performed using 75 g of dextrose. These occurred before the study, at 6 weeks, and at 12 weeks. There were no significant differences between fasting levels of serum insulin or blood glucose.⁴ Glucosamine sulfate supplementation did not alter serum insulin or plasma glucose during the tests. Limitations to the study include the small number of subjects, the short duration period, and the fact that the tests were performed before the patient’s daily glucosamine dosing.

A recent study examined insulin and glucose levels with and without the simultaneous ingestion of 1500 mg of glucosamine. Sixteen fasting volunteers with osteoarthritis but without known diabetes or glucose intolerance received 7 g of glucose with or without ingestion of 1500 mg glucosamine sulfate. The authors unexpectedly uncovered undiagnosed diabetes or impaired glucose tolerance in 3 subjects. These 3 subjects showed a statistically significant (P=.04) 31% increase in the area under the curve of glucose levels following the test. There was no effect of glucosamine sulfate ingestion on patients with normal baseline glucose testing or on insulin levels. Their results might be important since they are the first to suggest that glucosamine ingestion may affect glucose levels in individuals who have untreated diabetes or glucose intolerance.⁵

One double-blinded RCT evaluated whether oral glucosamine supplementation altered glycosylated hemoglobin (HbA_1c) concentrations for patients with well-controlled diabetes mellitus. Thirty-eight patients were randomized to receive either treatment with glucosamine/chondroitin at the recommended doses or placebo. After 3 months of treatment HbA_1c levels did not change and were not significantly different between groups (P=.2).⁶

Another study addressed whether glucosamine taken at recommended doses for the treatment of osteoarthritis had any detrimental effect on glucose metabolism. Fourteen patients participated and had a baseline 4-hour meal tolerance test and a frequently sampled intravenous glucose tolerance test, before and after 4 weeks of glucosamine sulfate treatment (500 mg orally 3 times daily). After 4 weeks they found no change in fasting plasma glucose, insulin, glucose tolerance, or difference in insulin sensitivity in the group of subjects.⁷ Again, the study was limited by a small subject number and short duration of study.

Recommendations from others

The PDR for Nonprescription Drugs and Dietary Supplements states that “glucosamine is likely safe for patients with diabetes that is well controlled with diet only or with one or two oral antidiabetic agents (HbA_1c less than 6.5%). For patients with higher HbA_1c concentrations or for those requiring insulin, closely monitor blood glucose concentrations.”⁸

The American Pain Society encourages adults with osteoarthritis to take 1500 mg of glucosamine daily as a dietary supplement but does not specifically recommend it as pharmacologic management for pain.⁹ The American College of Rheumatology Subcommittee on Osteoarthritis has no recommendations regarding the use of glucosamine or chondroitin in the treatment of knee osteoarthritis.¹⁰

Evidence summary

Diabetes mellitus and osteoarthritis commonly overlap in patients, many of whom have looked to the nutritional supplement combination of glucosamine and chondroitin sulfates for pain relief. The effectiveness of these supplements in improving patient-oriented outcomes for osteoarthritis is still being evaluated. However, regardless of their effectiveness they remain popular supplements, representing up to a third of the specialized supplement market in the US.¹

The mechanism by which glucosamine is hypothesized to affect blood glucose involves the roles of glucosamine and the hexosamine biosynthesis pathway in the regulation of glucose transport. Overexpression of enzymes involved in this pathway have led to high levels of glucose and insulin resistance in animals given huge doses of intravenous glucosamine (100–200 times higher than oral therapeutic doses in humans).² Studies have specifically investigated the effects of intravenous glucosamine infusion in healthy humans, and it did not show any effect on insulin sensitivity or plasma glucose.

A Cochrane systematic review of 20 randomized controlled trials (RCTs) including 2570 patients in order to evaluate the effectiveness and toxicity of glucosamine in osteoarthritis found that glucosamine was as safe as placebo in terms of adverse reactions. However, they did not comment specifically on diabetic patients or hyperglycemia per se.³

An RCT published in 2004 tested whether glucose intolerance occurs when healthy adults consume normal, recommended dosages of glucosamine sulfate. Nineteen healthy adults were randomized to receive either 1500 mg of glucosamine sulfate or placebo orally each day for 12 weeks. Three-hour oral glucose tolerance tests were performed using 75 g of dextrose. These occurred before the study, at 6 weeks, and at 12 weeks. There were no significant differences between fasting levels of serum insulin or blood glucose.⁴ Glucosamine sulfate supplementation did not alter serum insulin or plasma glucose during the tests. Limitations to the study include the small number of subjects, the short duration period, and the fact that the tests were performed before the patient’s daily glucosamine dosing.

A recent study examined insulin and glucose levels with and without the simultaneous ingestion of 1500 mg of glucosamine. Sixteen fasting volunteers with osteoarthritis but without known diabetes or glucose intolerance received 7 g of glucose with or without ingestion of 1500 mg glucosamine sulfate. The authors unexpectedly uncovered undiagnosed diabetes or impaired glucose tolerance in 3 subjects. These 3 subjects showed a statistically significant (P=.04) 31% increase in the area under the curve of glucose levels following the test. There was no effect of glucosamine sulfate ingestion on patients with normal baseline glucose testing or on insulin levels. Their results might be important since they are the first to suggest that glucosamine ingestion may affect glucose levels in individuals who have untreated diabetes or glucose intolerance.⁵

One double-blinded RCT evaluated whether oral glucosamine supplementation altered glycosylated hemoglobin (HbA_1c) concentrations for patients with well-controlled diabetes mellitus. Thirty-eight patients were randomized to receive either treatment with glucosamine/chondroitin at the recommended doses or placebo. After 3 months of treatment HbA_1c levels did not change and were not significantly different between groups (P=.2).⁶

Another study addressed whether glucosamine taken at recommended doses for the treatment of osteoarthritis had any detrimental effect on glucose metabolism. Fourteen patients participated and had a baseline 4-hour meal tolerance test and a frequently sampled intravenous glucose tolerance test, before and after 4 weeks of glucosamine sulfate treatment (500 mg orally 3 times daily). After 4 weeks they found no change in fasting plasma glucose, insulin, glucose tolerance, or difference in insulin sensitivity in the group of subjects.⁷ Again, the study was limited by a small subject number and short duration of study.

Recommendations from others

The PDR for Nonprescription Drugs and Dietary Supplements states that “glucosamine is likely safe for patients with diabetes that is well controlled with diet only or with one or two oral antidiabetic agents (HbA_1c less than 6.5%). For patients with higher HbA_1c concentrations or for those requiring insulin, closely monitor blood glucose concentrations.”⁸

The American Pain Society encourages adults with osteoarthritis to take 1500 mg of glucosamine daily as a dietary supplement but does not specifically recommend it as pharmacologic management for pain.⁹ The American College of Rheumatology Subcommittee on Osteoarthritis has no recommendations regarding the use of glucosamine or chondroitin in the treatment of knee osteoarthritis.¹⁰

Evidence summary

Celiac disease (also called celiac sprue, nontropical sprue, or gluten-sensitive enteropathy) is an autoimmune disorder classified by intestinal inflammation and malabsorption in response to dietary gluten—a storage protein component of wheat gliadins. Celiac disease patients—0.5% to 1.0% of the US population¹—are often sensitive to other closely related grain proteins found in oats, barley, and rye.

Celiac disease’s classic histological finding is an infiltrative small intestine lesion characterized by villous flattening, crypt hyperplasia, and lymphocyte accumulation in the lamina propria.² The American Gastroenterological Association’s (AGA) diagnostic criteria include confirmation of this abnormal mucosa and unequivocal improvement on repeat biopsy following a gluten-free diet.³ However, either clinical improvement or biopsy of dermatitis herpetiformis skin lesions (common occurrences in celiac disease) are often considered adequate for diagnosis without repeat intestinal biopsies.

Its reversible nature makes prompt diagnosis of celiac disease important. Three antibodies commonly appear in celiac disease patients: antibodies to tTG, antiendomysial antibodies, and antigliadin antibodies.^4-6 AGA binds dietary gluten and EMA binds the enzyme tTG, which is found in connective tissue surrounding the smooth muscle cells in the intestinal wall.

The gluten-autoantibody interaction in the small intestinal lumen has IgA as its major component; IgG represents a longer-term immune response. Serum assays of the IgA and IgG forms of AGA and tTG are enzyme-linked immunosorbent assays (ELISA); EMA is measured by indirect immunofluorescence.^5,6

In 2 studies of the diagnostic accuracy of IgA tTG, 95% to 98% of biopsy-proven celiac disease patients had positive tests, while only 5% to 6% of controls were positive.^4,5 In a systematic review, there was no statistically significant difference between IgA EMA and IgA tTG—both had sensitivities greater than 90% and specificities greater than 95%.⁷ IgA AGA did not perform as well (sensitivity 80%–90%, specificity 85%–95%). The TABLE summarizes the diagnostic accuracy of the various tests.

TABLE
Diagnostic accuracy of serologic tests for celiac disease patients with normal IgA levels

Two to 3% of patients with celiac disease have selective IgA deficiency.² These patients often have falsely negative serum IgA assays (for EMA, tTG, and AGA), so IgG is a diagnostic alternative.^8,9 In a cross-sectional study, 100% of 20 untreated celiac disease patients with IgA deficiency had positive IgG tests for tTG, AGA, and EMA despite negative IgA tests for the same antibodies.⁹ Eleven patients with celiac disease and no IgA deficiency all had positive tTG, AGA, and EMA tests, whether testing for the IgA or IgG forms.

Despite the performance of the IgG assays in this study, only IgG AGA has performed well in larger studies. In a systematic review, IgG tTG and IgG EMA had specificities of 95% but sensitivities of only 40%.⁷ IgG AGA has similar sensitivity to the IgA assay—approximately 80%—with a slightly lower specificity of 80%. The discrepancy in the sensitivity of IgG tTG and IgG EMA between studies occurs because of differing antibody levels with variations in dietary gluten.^9,10 Therefore, testing for IgG tTG and IgG EMA should be reserved for patients with selective IgA deficiency.

Another notable limitation of using serologic markers to diagnose celiac disease is poor sensitivity in patients with mild disease.¹¹ Diagnosis in these patients may be particularly challenging. Patients with karyotype abnormalities and those with diabetes are also more likely to have false-negative serologic tests.²

Recommendations from others

The AGA recommends using serologic markers to screen patients with either non-specific symptoms or medical conditions that increase the risk of celiac disease.³ Patients whose clinical profile causes a high index of suspicion and negative IgA serologic markers should be tested for selective IgA deficiency. The AGA recommends relying on small intestinal biopsy for the final diagnosis.

Both the AGA and the North American Pediatric Society for Pediatric Gastroenterology state that tissue transglutaminase and endomysial antibodies are the most useful serologic tests. Antigliadin antibody tests are considered inferior in terms of diagnostic accuracy.

Evidence summary

Celiac disease (also called celiac sprue, nontropical sprue, or gluten-sensitive enteropathy) is an autoimmune disorder classified by intestinal inflammation and malabsorption in response to dietary gluten—a storage protein component of wheat gliadins. Celiac disease patients—0.5% to 1.0% of the US population¹—are often sensitive to other closely related grain proteins found in oats, barley, and rye.

Celiac disease’s classic histological finding is an infiltrative small intestine lesion characterized by villous flattening, crypt hyperplasia, and lymphocyte accumulation in the lamina propria.² The American Gastroenterological Association’s (AGA) diagnostic criteria include confirmation of this abnormal mucosa and unequivocal improvement on repeat biopsy following a gluten-free diet.³ However, either clinical improvement or biopsy of dermatitis herpetiformis skin lesions (common occurrences in celiac disease) are often considered adequate for diagnosis without repeat intestinal biopsies.

Its reversible nature makes prompt diagnosis of celiac disease important. Three antibodies commonly appear in celiac disease patients: antibodies to tTG, antiendomysial antibodies, and antigliadin antibodies.^4-6 AGA binds dietary gluten and EMA binds the enzyme tTG, which is found in connective tissue surrounding the smooth muscle cells in the intestinal wall.

The gluten-autoantibody interaction in the small intestinal lumen has IgA as its major component; IgG represents a longer-term immune response. Serum assays of the IgA and IgG forms of AGA and tTG are enzyme-linked immunosorbent assays (ELISA); EMA is measured by indirect immunofluorescence.^5,6

In 2 studies of the diagnostic accuracy of IgA tTG, 95% to 98% of biopsy-proven celiac disease patients had positive tests, while only 5% to 6% of controls were positive.^4,5 In a systematic review, there was no statistically significant difference between IgA EMA and IgA tTG—both had sensitivities greater than 90% and specificities greater than 95%.⁷ IgA AGA did not perform as well (sensitivity 80%–90%, specificity 85%–95%). The TABLE summarizes the diagnostic accuracy of the various tests.

TABLE
Diagnostic accuracy of serologic tests for celiac disease patients with normal IgA levels

Two to 3% of patients with celiac disease have selective IgA deficiency.² These patients often have falsely negative serum IgA assays (for EMA, tTG, and AGA), so IgG is a diagnostic alternative.^8,9 In a cross-sectional study, 100% of 20 untreated celiac disease patients with IgA deficiency had positive IgG tests for tTG, AGA, and EMA despite negative IgA tests for the same antibodies.⁹ Eleven patients with celiac disease and no IgA deficiency all had positive tTG, AGA, and EMA tests, whether testing for the IgA or IgG forms.

Despite the performance of the IgG assays in this study, only IgG AGA has performed well in larger studies. In a systematic review, IgG tTG and IgG EMA had specificities of 95% but sensitivities of only 40%.⁷ IgG AGA has similar sensitivity to the IgA assay—approximately 80%—with a slightly lower specificity of 80%. The discrepancy in the sensitivity of IgG tTG and IgG EMA between studies occurs because of differing antibody levels with variations in dietary gluten.^9,10 Therefore, testing for IgG tTG and IgG EMA should be reserved for patients with selective IgA deficiency.

Another notable limitation of using serologic markers to diagnose celiac disease is poor sensitivity in patients with mild disease.¹¹ Diagnosis in these patients may be particularly challenging. Patients with karyotype abnormalities and those with diabetes are also more likely to have false-negative serologic tests.²

Recommendations from others

The AGA recommends using serologic markers to screen patients with either non-specific symptoms or medical conditions that increase the risk of celiac disease.³ Patients whose clinical profile causes a high index of suspicion and negative IgA serologic markers should be tested for selective IgA deficiency. The AGA recommends relying on small intestinal biopsy for the final diagnosis.

Both the AGA and the North American Pediatric Society for Pediatric Gastroenterology state that tissue transglutaminase and endomysial antibodies are the most useful serologic tests. Antigliadin antibody tests are considered inferior in terms of diagnostic accuracy.

Evidence summary

Celiac disease (also called celiac sprue, nontropical sprue, or gluten-sensitive enteropathy) is an autoimmune disorder classified by intestinal inflammation and malabsorption in response to dietary gluten—a storage protein component of wheat gliadins. Celiac disease patients—0.5% to 1.0% of the US population¹—are often sensitive to other closely related grain proteins found in oats, barley, and rye.

Celiac disease’s classic histological finding is an infiltrative small intestine lesion characterized by villous flattening, crypt hyperplasia, and lymphocyte accumulation in the lamina propria.² The American Gastroenterological Association’s (AGA) diagnostic criteria include confirmation of this abnormal mucosa and unequivocal improvement on repeat biopsy following a gluten-free diet.³ However, either clinical improvement or biopsy of dermatitis herpetiformis skin lesions (common occurrences in celiac disease) are often considered adequate for diagnosis without repeat intestinal biopsies.

Its reversible nature makes prompt diagnosis of celiac disease important. Three antibodies commonly appear in celiac disease patients: antibodies to tTG, antiendomysial antibodies, and antigliadin antibodies.^4-6 AGA binds dietary gluten and EMA binds the enzyme tTG, which is found in connective tissue surrounding the smooth muscle cells in the intestinal wall.

The gluten-autoantibody interaction in the small intestinal lumen has IgA as its major component; IgG represents a longer-term immune response. Serum assays of the IgA and IgG forms of AGA and tTG are enzyme-linked immunosorbent assays (ELISA); EMA is measured by indirect immunofluorescence.^5,6

In 2 studies of the diagnostic accuracy of IgA tTG, 95% to 98% of biopsy-proven celiac disease patients had positive tests, while only 5% to 6% of controls were positive.^4,5 In a systematic review, there was no statistically significant difference between IgA EMA and IgA tTG—both had sensitivities greater than 90% and specificities greater than 95%.⁷ IgA AGA did not perform as well (sensitivity 80%–90%, specificity 85%–95%). The TABLE summarizes the diagnostic accuracy of the various tests.

TABLE
Diagnostic accuracy of serologic tests for celiac disease patients with normal IgA levels

Two to 3% of patients with celiac disease have selective IgA deficiency.² These patients often have falsely negative serum IgA assays (for EMA, tTG, and AGA), so IgG is a diagnostic alternative.^8,9 In a cross-sectional study, 100% of 20 untreated celiac disease patients with IgA deficiency had positive IgG tests for tTG, AGA, and EMA despite negative IgA tests for the same antibodies.⁹ Eleven patients with celiac disease and no IgA deficiency all had positive tTG, AGA, and EMA tests, whether testing for the IgA or IgG forms.

Despite the performance of the IgG assays in this study, only IgG AGA has performed well in larger studies. In a systematic review, IgG tTG and IgG EMA had specificities of 95% but sensitivities of only 40%.⁷ IgG AGA has similar sensitivity to the IgA assay—approximately 80%—with a slightly lower specificity of 80%. The discrepancy in the sensitivity of IgG tTG and IgG EMA between studies occurs because of differing antibody levels with variations in dietary gluten.^9,10 Therefore, testing for IgG tTG and IgG EMA should be reserved for patients with selective IgA deficiency.

Another notable limitation of using serologic markers to diagnose celiac disease is poor sensitivity in patients with mild disease.¹¹ Diagnosis in these patients may be particularly challenging. Patients with karyotype abnormalities and those with diabetes are also more likely to have false-negative serologic tests.²

Recommendations from others

The AGA recommends using serologic markers to screen patients with either non-specific symptoms or medical conditions that increase the risk of celiac disease.³ Patients whose clinical profile causes a high index of suspicion and negative IgA serologic markers should be tested for selective IgA deficiency. The AGA recommends relying on small intestinal biopsy for the final diagnosis.

Both the AGA and the North American Pediatric Society for Pediatric Gastroenterology state that tissue transglutaminase and endomysial antibodies are the most useful serologic tests. Antigliadin antibody tests are considered inferior in terms of diagnostic accuracy.

Evidence summary

SIDS is defined as the sudden death of an infant aged <1 year of age that remains unexplained after a thorough investigation. The SIDS mortality rate is 0.57 per 1000 infants, with peak incidence among 1- to 5-month-olds.¹ Non-supine sleep position and parental tobacco use are established risk factors for SIDS and therefore are not explicitly addressed in this review. Using the 9 best-designed case-control studies published to date, each of which used multivariate analysis to control for infant sleep position and parental tobacco use (among other confounders), we evaluated co-sleeping, room sharing, sleep surfaces, and bedding accessories as risk factors for SIDS (TABLE).

TABLE
Sleeping arrangements and their relationship to SIDS

A number of factors complicated this review. First, although all studies evaluated infants through 1 year of age, some excluded infants <7 days or <28 days old. Second, studies examined different sleep periods; 2 focused on usual sleeping arrangements,^2,3 5 on sleeping arrangement immediately prior to death,^4-8 and 2 evaluated both usual and last sleep arrangements.^9,10 Third, variations in definitions of each risk factor and differences in the confounders controlled for made comparing studies challenging. Fourth, given the difficulty in studying infant deaths, the best evidence available comes from case-control studies.

Co-sleeping. Overall, 5 of 6 studies demonstrated co-sleeping to be an independent risk factor for SIDS (odds ratio [OR]=2.0–16.5),^2,4-7,9 especially for infants younger than 11 weeks old.^6,8 Four stratified analyses indicate that the risk of co-sleeping is greatest among infants of smokers (OR=4.6–17.7) as compared with infants of nonsmokers (OR=1.0–2.2).^3,7,8,10 Some descriptive studies suggest potential benefits of co-sleeping, such as improved breastfeeding and maternal-infant bonding, but these benefits have not been quantified.¹

Room sharing. Three of 4 studies found that infants sleeping in separate rooms from their caregivers had a 3-fold increased risk of SIDS,^5,6,11 while the fourth study found a 10-fold increased risk.⁸ One study found the risk was present in infants less than 20 weeks, but was inconclusive for those greater than 20 weeks.¹¹

Sleep surface. All 4 studies evaluating sleep surface found a significantly increased risk of SIDS for infants sleeping on sofas or armchairs compared with infants sleeping in beds or cribs. Fifty-five of 772 total cases (7.1%) from the 4 studies slept on a non-bed surface compared with 8 of 1854 controls (0.4%).^4-6,9

Bedding accessories. Two of 3 studies found pillow use unrelated to SIDS.^4,7,9 The larger of 2 studies on duvet use found it to be a risk factor for SIDS (OR=1.82).⁸

Recommendations from others

The American Academy of Pediatrics recommends that infants should sleep supine in the same room, but not the same bed, as their caregivers, while on a firm surface without bedding accessories. They should never sleep on a couch or armchair. Infants may be brought into bed briefly for feeding or comforting. Parents should be encouraged to quit smoking.¹⁰

Evidence summary

SIDS is defined as the sudden death of an infant aged <1 year of age that remains unexplained after a thorough investigation. The SIDS mortality rate is 0.57 per 1000 infants, with peak incidence among 1- to 5-month-olds.¹ Non-supine sleep position and parental tobacco use are established risk factors for SIDS and therefore are not explicitly addressed in this review. Using the 9 best-designed case-control studies published to date, each of which used multivariate analysis to control for infant sleep position and parental tobacco use (among other confounders), we evaluated co-sleeping, room sharing, sleep surfaces, and bedding accessories as risk factors for SIDS (TABLE).

TABLE
Sleeping arrangements and their relationship to SIDS

A number of factors complicated this review. First, although all studies evaluated infants through 1 year of age, some excluded infants <7 days or <28 days old. Second, studies examined different sleep periods; 2 focused on usual sleeping arrangements,^2,3 5 on sleeping arrangement immediately prior to death,^4-8 and 2 evaluated both usual and last sleep arrangements.^9,10 Third, variations in definitions of each risk factor and differences in the confounders controlled for made comparing studies challenging. Fourth, given the difficulty in studying infant deaths, the best evidence available comes from case-control studies.

Co-sleeping. Overall, 5 of 6 studies demonstrated co-sleeping to be an independent risk factor for SIDS (odds ratio [OR]=2.0–16.5),^2,4-7,9 especially for infants younger than 11 weeks old.^6,8 Four stratified analyses indicate that the risk of co-sleeping is greatest among infants of smokers (OR=4.6–17.7) as compared with infants of nonsmokers (OR=1.0–2.2).^3,7,8,10 Some descriptive studies suggest potential benefits of co-sleeping, such as improved breastfeeding and maternal-infant bonding, but these benefits have not been quantified.¹

Room sharing. Three of 4 studies found that infants sleeping in separate rooms from their caregivers had a 3-fold increased risk of SIDS,^5,6,11 while the fourth study found a 10-fold increased risk.⁸ One study found the risk was present in infants less than 20 weeks, but was inconclusive for those greater than 20 weeks.¹¹

Sleep surface. All 4 studies evaluating sleep surface found a significantly increased risk of SIDS for infants sleeping on sofas or armchairs compared with infants sleeping in beds or cribs. Fifty-five of 772 total cases (7.1%) from the 4 studies slept on a non-bed surface compared with 8 of 1854 controls (0.4%).^4-6,9

Bedding accessories. Two of 3 studies found pillow use unrelated to SIDS.^4,7,9 The larger of 2 studies on duvet use found it to be a risk factor for SIDS (OR=1.82).⁸

Recommendations from others

The American Academy of Pediatrics recommends that infants should sleep supine in the same room, but not the same bed, as their caregivers, while on a firm surface without bedding accessories. They should never sleep on a couch or armchair. Infants may be brought into bed briefly for feeding or comforting. Parents should be encouraged to quit smoking.¹⁰

Evidence summary

SIDS is defined as the sudden death of an infant aged <1 year of age that remains unexplained after a thorough investigation. The SIDS mortality rate is 0.57 per 1000 infants, with peak incidence among 1- to 5-month-olds.¹ Non-supine sleep position and parental tobacco use are established risk factors for SIDS and therefore are not explicitly addressed in this review. Using the 9 best-designed case-control studies published to date, each of which used multivariate analysis to control for infant sleep position and parental tobacco use (among other confounders), we evaluated co-sleeping, room sharing, sleep surfaces, and bedding accessories as risk factors for SIDS (TABLE).

TABLE
Sleeping arrangements and their relationship to SIDS

A number of factors complicated this review. First, although all studies evaluated infants through 1 year of age, some excluded infants <7 days or <28 days old. Second, studies examined different sleep periods; 2 focused on usual sleeping arrangements,^2,3 5 on sleeping arrangement immediately prior to death,^4-8 and 2 evaluated both usual and last sleep arrangements.^9,10 Third, variations in definitions of each risk factor and differences in the confounders controlled for made comparing studies challenging. Fourth, given the difficulty in studying infant deaths, the best evidence available comes from case-control studies.

Co-sleeping. Overall, 5 of 6 studies demonstrated co-sleeping to be an independent risk factor for SIDS (odds ratio [OR]=2.0–16.5),^2,4-7,9 especially for infants younger than 11 weeks old.^6,8 Four stratified analyses indicate that the risk of co-sleeping is greatest among infants of smokers (OR=4.6–17.7) as compared with infants of nonsmokers (OR=1.0–2.2).^3,7,8,10 Some descriptive studies suggest potential benefits of co-sleeping, such as improved breastfeeding and maternal-infant bonding, but these benefits have not been quantified.¹

Room sharing. Three of 4 studies found that infants sleeping in separate rooms from their caregivers had a 3-fold increased risk of SIDS,^5,6,11 while the fourth study found a 10-fold increased risk.⁸ One study found the risk was present in infants less than 20 weeks, but was inconclusive for those greater than 20 weeks.¹¹

Sleep surface. All 4 studies evaluating sleep surface found a significantly increased risk of SIDS for infants sleeping on sofas or armchairs compared with infants sleeping in beds or cribs. Fifty-five of 772 total cases (7.1%) from the 4 studies slept on a non-bed surface compared with 8 of 1854 controls (0.4%).^4-6,9

Bedding accessories. Two of 3 studies found pillow use unrelated to SIDS.^4,7,9 The larger of 2 studies on duvet use found it to be a risk factor for SIDS (OR=1.82).⁸

Recommendations from others

The American Academy of Pediatrics recommends that infants should sleep supine in the same room, but not the same bed, as their caregivers, while on a firm surface without bedding accessories. They should never sleep on a couch or armchair. Infants may be brought into bed briefly for feeding or comforting. Parents should be encouraged to quit smoking.¹⁰

Evidence summary

In general, strategies for addressing treatment-resistant depression have not been compared in head-to-head studies. Guidelines at this time are based mainly on expert opinion^2,3 and gradually accumulating data from a few randomized controlled studies or low-quality cohort studies.

While it makes sense, as the question implies, to first optimize the dose and duration of SSRI treatment in treatment-resistant depression, it is not clear which strategy to employ next. Switch, augmentation, and combination strategies may each improve clinical outcomes, but which strategy is best is based on expert opinion at this time.

Optimize. The first step in treatment-resistant depression should be optimizing dose and duration of therapy.⁴ For fluoxetine (Prozac), based on a nonrandomized open trial, patients should receive 8 weeks of treatment before the SSRI course is deemed adequate. Only 23% of patients who have not responded to 8 weeks of fluoxetine respond to a still longer course of fluoxetine.

Switch. The strongest evidence is from the recent STAR*D trial, a randomized study that assigned patients in one arm of the study who had no relief from (or did not tolerate) therapy with citalopram (Celexa) to 1 of 3 drugs—sustained-release bupropion (Wellbutrin SR), sertraline (Zoloft), or extended-release venlafaxine (Effexor XR). The study concluded that approximately 1 in 4 patients have remission after switching to an antidepressant from another drug class.¹ Further switches in antidepressant monotherapy have a low success rate (10%–20%).⁵

Add/combine. Mixed evidence supports combining different antidepressants. There is cohort study evidence that combining citalopram and bupropion is more effective than switching to the alternate antidepressant,⁶ but other cohort studies did not find a significant difference between switching and augmenting. An arm of the STAR*D trial added either sustained-release bupropion or buspirone (Buspar) to the failed citalopram therapy. Thirty percent of patients with depression unresponsive to citalopram had remission when bupropion-SR or buspirone was added.⁷ The STAR*D reports do not compare the 2 strategies of switching or combining drugs directly.

Augment. Evidence from a meta-analysis with aggregate data from 3 studies representing a total of 110 patients showed that augmentation of various antidepressants with lithium leads to improved outcomes (number needed to treat [NNT]=3.7).⁸ A cohort study of augmentation with an atypical antipsychotic agent such as aripiprazole (Abilify) suggest improved outcomes, but similar studies found no benefit.⁹ A small (23-patient) randomized trial of lamotrigine (Lamictal) suggests that it may augment the effect of fluoxetine.¹⁰

Psychotherapy. A systematic review of psychological therapies in treatment-resistant depression found 2 controlled studies (of cognitive therapy and cognitive behavioral therapy) out of 12 total studies meeting their inclusion criteria that demonstrated improved scores on the Hamilton Rating Scale for Depression. Further study of these therapies was recommended.¹¹

ECT. The evidence supporting use of ECT for treatment-resistant depression comes from studies following failure of treatment with tricyclic antidepressants and monoamine oxidase (MAO) inhibitors. Methodological problems in these older studies do not permit an estimate of response rate.¹²

Recommendations by others

The American Psychiatric Association treatment guideline recommends changing antidepressant, adding or changing to psychotherapy, or ECT if no response to 4 to 8 weeks of the initial therapy in depression.¹³ A guideline from the University of Michigan recommends referral to a psychiatrist if patients have treatment refractory depression (defined in their guideline as failure of 2 successive trials of antidepressants).¹⁴ The Institute for Clinical Systems Improvement guideline recommends considering switch, augmentation, or other therapies (including adding or modifying psychotherapy).¹⁵

Evidence summary

In general, strategies for addressing treatment-resistant depression have not been compared in head-to-head studies. Guidelines at this time are based mainly on expert opinion^2,3 and gradually accumulating data from a few randomized controlled studies or low-quality cohort studies.

While it makes sense, as the question implies, to first optimize the dose and duration of SSRI treatment in treatment-resistant depression, it is not clear which strategy to employ next. Switch, augmentation, and combination strategies may each improve clinical outcomes, but which strategy is best is based on expert opinion at this time.

Optimize. The first step in treatment-resistant depression should be optimizing dose and duration of therapy.⁴ For fluoxetine (Prozac), based on a nonrandomized open trial, patients should receive 8 weeks of treatment before the SSRI course is deemed adequate. Only 23% of patients who have not responded to 8 weeks of fluoxetine respond to a still longer course of fluoxetine.

Switch. The strongest evidence is from the recent STAR*D trial, a randomized study that assigned patients in one arm of the study who had no relief from (or did not tolerate) therapy with citalopram (Celexa) to 1 of 3 drugs—sustained-release bupropion (Wellbutrin SR), sertraline (Zoloft), or extended-release venlafaxine (Effexor XR). The study concluded that approximately 1 in 4 patients have remission after switching to an antidepressant from another drug class.¹ Further switches in antidepressant monotherapy have a low success rate (10%–20%).⁵

Add/combine. Mixed evidence supports combining different antidepressants. There is cohort study evidence that combining citalopram and bupropion is more effective than switching to the alternate antidepressant,⁶ but other cohort studies did not find a significant difference between switching and augmenting. An arm of the STAR*D trial added either sustained-release bupropion or buspirone (Buspar) to the failed citalopram therapy. Thirty percent of patients with depression unresponsive to citalopram had remission when bupropion-SR or buspirone was added.⁷ The STAR*D reports do not compare the 2 strategies of switching or combining drugs directly.

Augment. Evidence from a meta-analysis with aggregate data from 3 studies representing a total of 110 patients showed that augmentation of various antidepressants with lithium leads to improved outcomes (number needed to treat [NNT]=3.7).⁸ A cohort study of augmentation with an atypical antipsychotic agent such as aripiprazole (Abilify) suggest improved outcomes, but similar studies found no benefit.⁹ A small (23-patient) randomized trial of lamotrigine (Lamictal) suggests that it may augment the effect of fluoxetine.¹⁰

Psychotherapy. A systematic review of psychological therapies in treatment-resistant depression found 2 controlled studies (of cognitive therapy and cognitive behavioral therapy) out of 12 total studies meeting their inclusion criteria that demonstrated improved scores on the Hamilton Rating Scale for Depression. Further study of these therapies was recommended.¹¹

ECT. The evidence supporting use of ECT for treatment-resistant depression comes from studies following failure of treatment with tricyclic antidepressants and monoamine oxidase (MAO) inhibitors. Methodological problems in these older studies do not permit an estimate of response rate.¹²

Recommendations by others

The American Psychiatric Association treatment guideline recommends changing antidepressant, adding or changing to psychotherapy, or ECT if no response to 4 to 8 weeks of the initial therapy in depression.¹³ A guideline from the University of Michigan recommends referral to a psychiatrist if patients have treatment refractory depression (defined in their guideline as failure of 2 successive trials of antidepressants).¹⁴ The Institute for Clinical Systems Improvement guideline recommends considering switch, augmentation, or other therapies (including adding or modifying psychotherapy).¹⁵

Evidence summary

In general, strategies for addressing treatment-resistant depression have not been compared in head-to-head studies. Guidelines at this time are based mainly on expert opinion^2,3 and gradually accumulating data from a few randomized controlled studies or low-quality cohort studies.

While it makes sense, as the question implies, to first optimize the dose and duration of SSRI treatment in treatment-resistant depression, it is not clear which strategy to employ next. Switch, augmentation, and combination strategies may each improve clinical outcomes, but which strategy is best is based on expert opinion at this time.

Optimize. The first step in treatment-resistant depression should be optimizing dose and duration of therapy.⁴ For fluoxetine (Prozac), based on a nonrandomized open trial, patients should receive 8 weeks of treatment before the SSRI course is deemed adequate. Only 23% of patients who have not responded to 8 weeks of fluoxetine respond to a still longer course of fluoxetine.

Switch. The strongest evidence is from the recent STAR*D trial, a randomized study that assigned patients in one arm of the study who had no relief from (or did not tolerate) therapy with citalopram (Celexa) to 1 of 3 drugs—sustained-release bupropion (Wellbutrin SR), sertraline (Zoloft), or extended-release venlafaxine (Effexor XR). The study concluded that approximately 1 in 4 patients have remission after switching to an antidepressant from another drug class.¹ Further switches in antidepressant monotherapy have a low success rate (10%–20%).⁵

Add/combine. Mixed evidence supports combining different antidepressants. There is cohort study evidence that combining citalopram and bupropion is more effective than switching to the alternate antidepressant,⁶ but other cohort studies did not find a significant difference between switching and augmenting. An arm of the STAR*D trial added either sustained-release bupropion or buspirone (Buspar) to the failed citalopram therapy. Thirty percent of patients with depression unresponsive to citalopram had remission when bupropion-SR or buspirone was added.⁷ The STAR*D reports do not compare the 2 strategies of switching or combining drugs directly.

Augment. Evidence from a meta-analysis with aggregate data from 3 studies representing a total of 110 patients showed that augmentation of various antidepressants with lithium leads to improved outcomes (number needed to treat [NNT]=3.7).⁸ A cohort study of augmentation with an atypical antipsychotic agent such as aripiprazole (Abilify) suggest improved outcomes, but similar studies found no benefit.⁹ A small (23-patient) randomized trial of lamotrigine (Lamictal) suggests that it may augment the effect of fluoxetine.¹⁰

Psychotherapy. A systematic review of psychological therapies in treatment-resistant depression found 2 controlled studies (of cognitive therapy and cognitive behavioral therapy) out of 12 total studies meeting their inclusion criteria that demonstrated improved scores on the Hamilton Rating Scale for Depression. Further study of these therapies was recommended.¹¹

ECT. The evidence supporting use of ECT for treatment-resistant depression comes from studies following failure of treatment with tricyclic antidepressants and monoamine oxidase (MAO) inhibitors. Methodological problems in these older studies do not permit an estimate of response rate.¹²

Recommendations by others

The American Psychiatric Association treatment guideline recommends changing antidepressant, adding or changing to psychotherapy, or ECT if no response to 4 to 8 weeks of the initial therapy in depression.¹³ A guideline from the University of Michigan recommends referral to a psychiatrist if patients have treatment refractory depression (defined in their guideline as failure of 2 successive trials of antidepressants).¹⁴ The Institute for Clinical Systems Improvement guideline recommends considering switch, augmentation, or other therapies (including adding or modifying psychotherapy).¹⁵

Evidence summary

A recent Cochrane review identified 11 randomized controlled trials (RCTs) (with a total of 917 patients) addressing antibiotic therapy for COPD exacerbations characterized by 1 or more of the following: an increase in sputum purulence or volume, dyspnea, wheezing, chest tightness, or fluid retention.¹ Eight trials were conducted on hospital wards, 1 was in a medical intensive care unit, and 2 trials were in the outpatient setting. Antibiotics were given orally in 9 of the 11 studies.

Overall, antibiotics reduced risk of short-term mortality by 77% (relative risk [RR]=0.23; 95% confidence interval [CI],0.10–0.52; number needed to treat [NNT]=8), treatment failure by 53% (RR=0.47; 95% CI, 0.36–0.62; NNT=3), and sputum purulence by 44% (RR=0.56; 95% CI, 0.41–0.77; NNT=8). A subgroup analysis that excluded the outpatient and intensive-care unit studies did not change the result. Another subgroup analysis of the 2 outpatient studies failed to find a significant effect, although the studies had very different designs.

These findings are more robust than those of an earlier, lower-quality meta-analysis of 9 randomized controlled trials (RCTs) with 1101 patients with presumed COPD, which also compared antibiotic therapy with placebo for acute exacerbations.² Specific diagnostic criteria were not stated for the diagnosis of either COPD or an acute exacerbation. No single outcome measure was common to all studies. The authors found a summary beneficial effect size of antibiotic therapy of 0.22 (95% CI, 0.10–0.34), which is generally interpreted as small. One clinical parameter, peak expiratory flow rate (PEFR), was reported in 6 of the studies. Antibiotic therapy resulted in an average 10.75 L/min improvement in PEFR compared with placebo (95% CI, 4.96–16.54 L/min).

Two RCTs addressing antibiotic use in the outpatient setting were identified in the Cochrane review. One double-blind crossover trial performed in Canada compared antibiotic with placebo therapy for 173 outpatients with 362 exacerbations classified according to severity.³ The protocol used oral trimethoprim-sulfamethoxazole, amoxicillin, or doxycycline (according to the attending physician’s preference) or a look-alike placebo. Symptom resolution was seen by 21 days in 68% of antibiotic users vs 55% of those on placebo (P<.01, NNT=8). Ten percent of patients taking antibiotics deteriorated to the point where hospitalization or unblinding of the therapy was necessary, compared with 19% in the placebo group (P<.05, NNT=11).

For patients with all 3 cardinal COPD symptoms (increased dyspnea, sputum production, and sputum purulence) at enrollment, there was resolution at 21 days in 63% with antibiotics vs 43% for placebo (P value not given). Antibiotics did not benefit patients with 1 cardinal symptom (74% success with antibiotics vs 70% on placebo; P value not given).

The Cochrane review also identified a Danish RCT that studied 278 patients presenting to their general practitioners with subjective acute worsening of their COPD. Patients were randomized to 7 days of oral amoxicillin or placebo. There was no difference between the groups in terms of symptom resolution at 1 week (odds ratio=1.03, favoring placebo; 95% CI, 0.75–1.41) or in changes in PEFR (weighted mean difference=–0.89, favoring placebo; 95% CI, –29 to 27 L/min).

Recommendations from others

The Veterans Health Administration recommends antibiotics if a patient with COPD has changes in sputum volume or quality as well as increased dyspnea, cough, or fever; infiltrate on x-ray suggesting pneumonia should be treated as such.⁴

The American College of Chest Physicians recommends that with severe COPD exacerbations, narrow spectrum antibiotics are reasonable first-line agents.⁵ They also note that the superiority of newer, more broad-spectrum antibiotics has not been established.

Evidence summary

A recent Cochrane review identified 11 randomized controlled trials (RCTs) (with a total of 917 patients) addressing antibiotic therapy for COPD exacerbations characterized by 1 or more of the following: an increase in sputum purulence or volume, dyspnea, wheezing, chest tightness, or fluid retention.¹ Eight trials were conducted on hospital wards, 1 was in a medical intensive care unit, and 2 trials were in the outpatient setting. Antibiotics were given orally in 9 of the 11 studies.

Overall, antibiotics reduced risk of short-term mortality by 77% (relative risk [RR]=0.23; 95% confidence interval [CI],0.10–0.52; number needed to treat [NNT]=8), treatment failure by 53% (RR=0.47; 95% CI, 0.36–0.62; NNT=3), and sputum purulence by 44% (RR=0.56; 95% CI, 0.41–0.77; NNT=8). A subgroup analysis that excluded the outpatient and intensive-care unit studies did not change the result. Another subgroup analysis of the 2 outpatient studies failed to find a significant effect, although the studies had very different designs.

These findings are more robust than those of an earlier, lower-quality meta-analysis of 9 randomized controlled trials (RCTs) with 1101 patients with presumed COPD, which also compared antibiotic therapy with placebo for acute exacerbations.² Specific diagnostic criteria were not stated for the diagnosis of either COPD or an acute exacerbation. No single outcome measure was common to all studies. The authors found a summary beneficial effect size of antibiotic therapy of 0.22 (95% CI, 0.10–0.34), which is generally interpreted as small. One clinical parameter, peak expiratory flow rate (PEFR), was reported in 6 of the studies. Antibiotic therapy resulted in an average 10.75 L/min improvement in PEFR compared with placebo (95% CI, 4.96–16.54 L/min).

Two RCTs addressing antibiotic use in the outpatient setting were identified in the Cochrane review. One double-blind crossover trial performed in Canada compared antibiotic with placebo therapy for 173 outpatients with 362 exacerbations classified according to severity.³ The protocol used oral trimethoprim-sulfamethoxazole, amoxicillin, or doxycycline (according to the attending physician’s preference) or a look-alike placebo. Symptom resolution was seen by 21 days in 68% of antibiotic users vs 55% of those on placebo (P<.01, NNT=8). Ten percent of patients taking antibiotics deteriorated to the point where hospitalization or unblinding of the therapy was necessary, compared with 19% in the placebo group (P<.05, NNT=11).

For patients with all 3 cardinal COPD symptoms (increased dyspnea, sputum production, and sputum purulence) at enrollment, there was resolution at 21 days in 63% with antibiotics vs 43% for placebo (P value not given). Antibiotics did not benefit patients with 1 cardinal symptom (74% success with antibiotics vs 70% on placebo; P value not given).

The Cochrane review also identified a Danish RCT that studied 278 patients presenting to their general practitioners with subjective acute worsening of their COPD. Patients were randomized to 7 days of oral amoxicillin or placebo. There was no difference between the groups in terms of symptom resolution at 1 week (odds ratio=1.03, favoring placebo; 95% CI, 0.75–1.41) or in changes in PEFR (weighted mean difference=–0.89, favoring placebo; 95% CI, –29 to 27 L/min).

Recommendations from others

The Veterans Health Administration recommends antibiotics if a patient with COPD has changes in sputum volume or quality as well as increased dyspnea, cough, or fever; infiltrate on x-ray suggesting pneumonia should be treated as such.⁴

The American College of Chest Physicians recommends that with severe COPD exacerbations, narrow spectrum antibiotics are reasonable first-line agents.⁵ They also note that the superiority of newer, more broad-spectrum antibiotics has not been established.

Evidence summary

A recent Cochrane review identified 11 randomized controlled trials (RCTs) (with a total of 917 patients) addressing antibiotic therapy for COPD exacerbations characterized by 1 or more of the following: an increase in sputum purulence or volume, dyspnea, wheezing, chest tightness, or fluid retention.¹ Eight trials were conducted on hospital wards, 1 was in a medical intensive care unit, and 2 trials were in the outpatient setting. Antibiotics were given orally in 9 of the 11 studies.

Overall, antibiotics reduced risk of short-term mortality by 77% (relative risk [RR]=0.23; 95% confidence interval [CI],0.10–0.52; number needed to treat [NNT]=8), treatment failure by 53% (RR=0.47; 95% CI, 0.36–0.62; NNT=3), and sputum purulence by 44% (RR=0.56; 95% CI, 0.41–0.77; NNT=8). A subgroup analysis that excluded the outpatient and intensive-care unit studies did not change the result. Another subgroup analysis of the 2 outpatient studies failed to find a significant effect, although the studies had very different designs.

These findings are more robust than those of an earlier, lower-quality meta-analysis of 9 randomized controlled trials (RCTs) with 1101 patients with presumed COPD, which also compared antibiotic therapy with placebo for acute exacerbations.² Specific diagnostic criteria were not stated for the diagnosis of either COPD or an acute exacerbation. No single outcome measure was common to all studies. The authors found a summary beneficial effect size of antibiotic therapy of 0.22 (95% CI, 0.10–0.34), which is generally interpreted as small. One clinical parameter, peak expiratory flow rate (PEFR), was reported in 6 of the studies. Antibiotic therapy resulted in an average 10.75 L/min improvement in PEFR compared with placebo (95% CI, 4.96–16.54 L/min).

Two RCTs addressing antibiotic use in the outpatient setting were identified in the Cochrane review. One double-blind crossover trial performed in Canada compared antibiotic with placebo therapy for 173 outpatients with 362 exacerbations classified according to severity.³ The protocol used oral trimethoprim-sulfamethoxazole, amoxicillin, or doxycycline (according to the attending physician’s preference) or a look-alike placebo. Symptom resolution was seen by 21 days in 68% of antibiotic users vs 55% of those on placebo (P<.01, NNT=8). Ten percent of patients taking antibiotics deteriorated to the point where hospitalization or unblinding of the therapy was necessary, compared with 19% in the placebo group (P<.05, NNT=11).

For patients with all 3 cardinal COPD symptoms (increased dyspnea, sputum production, and sputum purulence) at enrollment, there was resolution at 21 days in 63% with antibiotics vs 43% for placebo (P value not given). Antibiotics did not benefit patients with 1 cardinal symptom (74% success with antibiotics vs 70% on placebo; P value not given).

The Cochrane review also identified a Danish RCT that studied 278 patients presenting to their general practitioners with subjective acute worsening of their COPD. Patients were randomized to 7 days of oral amoxicillin or placebo. There was no difference between the groups in terms of symptom resolution at 1 week (odds ratio=1.03, favoring placebo; 95% CI, 0.75–1.41) or in changes in PEFR (weighted mean difference=–0.89, favoring placebo; 95% CI, –29 to 27 L/min).

Recommendations from others

The Veterans Health Administration recommends antibiotics if a patient with COPD has changes in sputum volume or quality as well as increased dyspnea, cough, or fever; infiltrate on x-ray suggesting pneumonia should be treated as such.⁴

The American College of Chest Physicians recommends that with severe COPD exacerbations, narrow spectrum antibiotics are reasonable first-line agents.⁵ They also note that the superiority of newer, more broad-spectrum antibiotics has not been established.

Evidence summary

Low HDL is recognized as a risk factor for atherosclerosis. Clinicians find raising HDL a challenge, and patients often inquire about dietary advice that may help raise HDL.

No quality evidence exists that specifically looks at the effect of a dietary intervention on HDL or whether it affects survival. However, several dietary intervention studies in specific populations include HDL as a secondary endpoint in the study. This leaves clinicians to act on physiologic data that may or may not increase the overall health and survival of patients. Dietary interventions that raised HDL include low-carbohydrate diets, the DASH diet, supplementation with soy protein including isoflavones, and multivitamin supplementation.

TABLE
Summary of studies evaluating the effect of various diets on HDL cholesterol

Several overall diet interventions appear to raise HDL, but whether this affects cardiovascular events or mortality is unknown. A systematic review with meta-analysis of 5 randomized controlled trials (RCTs) of low-carbohydrate versus low-fat diets showed a 10% increase in HDL attributed to the low-carbohydrate diet, which translated to an absolute increase of 4.6 mg/dL (95% confidence interval [CI], 1.5–8.1).¹ A subsequent uncontrolled prospective trial was consistent with consistent with this systematic review and showed a 12% increase in HDL.² The DASH diet was studied as an intervention in a RCT of 116 patients with the metabolic syndrome. Men responded with an in crease of 21% and women with an increase of 33%.³

Supplementation with several food additives and nutritional supplements has been tested. A systematic review with meta-analysis of 23 RCTs evaluating effect of various amounts of soy protein with isoflavones on lipid profile found a 3% increase in HDL with an absolute difference 1.5 mg/dL (95% CI, 0.0–2.8).⁴ Supplementation with standard multivitamins in a single small, crossover RCT showed a 31% increase in HDL.⁵

Many other strategies, supplements, and plans have been tested in different populations. Other than the above interventions, no other interventions raise HDL when subjected to meta-analysis or quality randomized trials (TABLE).

Recommendations from others

No specific guidelines on dietary therapy of HDL exist; however, the American Heart Association (AHA) published diet and lifestyle recommendations in 2006.¹⁴ These guidelines recommend a diet low in fat, saturated fat, trans fat, and cholesterol in addition to minimizing sodium, added sugars, and alcohol. The AHA also recommends for consumption of oily fish and the DASH diet.

Acknowledgments

The opinions and assertions contained herein are the private views of the author and not to be construed as official, or as reflecting the views of the US Air Force Medical Service or the US Air Force at large.

Evidence summary

Low HDL is recognized as a risk factor for atherosclerosis. Clinicians find raising HDL a challenge, and patients often inquire about dietary advice that may help raise HDL.

No quality evidence exists that specifically looks at the effect of a dietary intervention on HDL or whether it affects survival. However, several dietary intervention studies in specific populations include HDL as a secondary endpoint in the study. This leaves clinicians to act on physiologic data that may or may not increase the overall health and survival of patients. Dietary interventions that raised HDL include low-carbohydrate diets, the DASH diet, supplementation with soy protein including isoflavones, and multivitamin supplementation.

TABLE
Summary of studies evaluating the effect of various diets on HDL cholesterol

Several overall diet interventions appear to raise HDL, but whether this affects cardiovascular events or mortality is unknown. A systematic review with meta-analysis of 5 randomized controlled trials (RCTs) of low-carbohydrate versus low-fat diets showed a 10% increase in HDL attributed to the low-carbohydrate diet, which translated to an absolute increase of 4.6 mg/dL (95% confidence interval [CI], 1.5–8.1).¹ A subsequent uncontrolled prospective trial was consistent with consistent with this systematic review and showed a 12% increase in HDL.² The DASH diet was studied as an intervention in a RCT of 116 patients with the metabolic syndrome. Men responded with an in crease of 21% and women with an increase of 33%.³

Supplementation with several food additives and nutritional supplements has been tested. A systematic review with meta-analysis of 23 RCTs evaluating effect of various amounts of soy protein with isoflavones on lipid profile found a 3% increase in HDL with an absolute difference 1.5 mg/dL (95% CI, 0.0–2.8).⁴ Supplementation with standard multivitamins in a single small, crossover RCT showed a 31% increase in HDL.⁵

Many other strategies, supplements, and plans have been tested in different populations. Other than the above interventions, no other interventions raise HDL when subjected to meta-analysis or quality randomized trials (TABLE).

Recommendations from others

No specific guidelines on dietary therapy of HDL exist; however, the American Heart Association (AHA) published diet and lifestyle recommendations in 2006.¹⁴ These guidelines recommend a diet low in fat, saturated fat, trans fat, and cholesterol in addition to minimizing sodium, added sugars, and alcohol. The AHA also recommends for consumption of oily fish and the DASH diet.

Acknowledgments

The opinions and assertions contained herein are the private views of the author and not to be construed as official, or as reflecting the views of the US Air Force Medical Service or the US Air Force at large.

Evidence summary

Low HDL is recognized as a risk factor for atherosclerosis. Clinicians find raising HDL a challenge, and patients often inquire about dietary advice that may help raise HDL.

No quality evidence exists that specifically looks at the effect of a dietary intervention on HDL or whether it affects survival. However, several dietary intervention studies in specific populations include HDL as a secondary endpoint in the study. This leaves clinicians to act on physiologic data that may or may not increase the overall health and survival of patients. Dietary interventions that raised HDL include low-carbohydrate diets, the DASH diet, supplementation with soy protein including isoflavones, and multivitamin supplementation.

TABLE
Summary of studies evaluating the effect of various diets on HDL cholesterol

Several overall diet interventions appear to raise HDL, but whether this affects cardiovascular events or mortality is unknown. A systematic review with meta-analysis of 5 randomized controlled trials (RCTs) of low-carbohydrate versus low-fat diets showed a 10% increase in HDL attributed to the low-carbohydrate diet, which translated to an absolute increase of 4.6 mg/dL (95% confidence interval [CI], 1.5–8.1).¹ A subsequent uncontrolled prospective trial was consistent with consistent with this systematic review and showed a 12% increase in HDL.² The DASH diet was studied as an intervention in a RCT of 116 patients with the metabolic syndrome. Men responded with an in crease of 21% and women with an increase of 33%.³

Supplementation with several food additives and nutritional supplements has been tested. A systematic review with meta-analysis of 23 RCTs evaluating effect of various amounts of soy protein with isoflavones on lipid profile found a 3% increase in HDL with an absolute difference 1.5 mg/dL (95% CI, 0.0–2.8).⁴ Supplementation with standard multivitamins in a single small, crossover RCT showed a 31% increase in HDL.⁵

Many other strategies, supplements, and plans have been tested in different populations. Other than the above interventions, no other interventions raise HDL when subjected to meta-analysis or quality randomized trials (TABLE).

Recommendations from others

No specific guidelines on dietary therapy of HDL exist; however, the American Heart Association (AHA) published diet and lifestyle recommendations in 2006.¹⁴ These guidelines recommend a diet low in fat, saturated fat, trans fat, and cholesterol in addition to minimizing sodium, added sugars, and alcohol. The AHA also recommends for consumption of oily fish and the DASH diet.

Acknowledgments

The opinions and assertions contained herein are the private views of the author and not to be construed as official, or as reflecting the views of the US Air Force Medical Service or the US Air Force at large.

Evidence summary

A systematic review evaluated beginning combination therapy (adding insulin to oral medication) compared with switching to insulin alone in patients with type 2 diabetes mellitus with inadequate glycemic control on oral medication.¹ Twenty RCTs studied a total of 1811 patients; glycemic control was the primary outcome measure. Oral medication comprised either sulfonylureas (75%), metformin (4%), or both (21%). Individual studies used different insulin dosing schedules and statistical measures. However, overall, combination therapy provided glucose control comparable with insulin alone. In only 1 small, low-quality study did insulin plus metformin reduce HbA_1c more than other combination therapy regimens or insulin alone. Ten studies reported a trend toward less weight gain with combination therapy that included metformin. Fourteen studies found the same incidence of hypoglycemic episodes in combination therapy and insulin alone.

Three later RCTs of overweight patients with inadequate control on oral agents (HbA_1c >7% on a sulfonylurea, metformin, or both) also compared beginning combination therapy with switching to insulin alone (with 70/30 or NPH insulin twice daily). In one study with 64 patients followed for 12 months, HbA_1c fell by 0.14% less (nonsignificant) in the combination therapy group (bedtime NPH plus sulfonylurea and metformin) than in the insulin alone group (70/30 twice daily).² The combination therapy group gained significantly less weight than the insulin-alone group (1.3 kg vs 4.2 kg; P=.01).

In the second study of 261 patients, the combination therapy group (glimepiride [Amaryl] plus bedtime NPH) had a significantly higher HbA_1c after 9 months than 2 groups using insulin alone (twice daily 70/30, and twice daily NPH insulin) (8.9% vs 8.3% and 8.4%).³ Mean weight gain was similar in all 3 groups but only a minority of patients reached a target HbA_1c of 6.5%. In the final study of only 16 patients, HbA1c after 6 months improved significantly and equally in both groups (baseline: 8.3%, combination therapy final: 6.8%; insulin alone final: 7.0%). However, the combination therapy group gained significantly less weight.⁴

An open-label RCT with 341 patients who were inadequately controlled on metformin compared beginning combination therapy (biphasic insulin aspart 30/70 [Novolog Mix 70/30] and metformin) with switching to insulin alone (biphasic insulin aspart 30/70).⁵ A third group added a second oral medication (sulfonylurea and metformin). After 16 weeks, patients taking combination therapy had a significantly lower HbA_1c than those on insulin alone (treatment difference 0.39%, P=.007).

Overall, combination therapy and 2 oral medications reduced HbA_1c by the same amount, but combination therapy reduced HbA_1c more in a subpopulation of patients with HbA_1c >9.0% at baseline (treatment difference 0.46%, P=.027). The group on insulin alone weighed significantly more (4.6 kg, P<.001) at the end of the trial than the group taking 2 oral medications.

An open-label RCT of 756 patients with inadequate glycemic control (HbA_1c >7.5%, mean 8.6%) on either 1 or 2 oral agents (70% taking both metformin and a sulfonylurea) compared combination therapy using bedtime insulin glargine with combination therapy using morning NPH.⁶ Each group titrated insulin doses to achieve a target fasting glucose ≤100. By 24 weeks, both groups had equivalently reduced HbA_1c (mean HbA_1c=6.96% with glargine, and 6.97% with NPH; P=NS), but fewer patients experienced nocturnal hypoglycemia with glargine than with NPH (33.2% vs 26.7%, P<.05).

Another open-label RCT evaluated 281 patients with at least 3 months of inadequate glycemic control (HbA_1c=7.4%–14.7%) on a sulfonylurea.⁷ Patients were randomized to a) switching to a combination of biphasic insulin aspart 30/70 plus pioglitazone, b) adding pioglitazone to the sulfonylurea, or c) switching to insulin alone (biphasic insulin aspart 30/70). After 18 weeks, insulin plus pioglitazone reduced HbA_1c significantly more than either glyburide plus pioglitazone (P=.005) or insulin alone (P=.005). However, the insulin plus pioglitazone group had the most weight gain (mean 4 kg, similar to other pioglitazone trials). There were no major hypoglycemic events.

Another open-label RCT evaluated 217 patients inadequately controlled (HbA_1c=7.5%–11%) on a 2-drug oral regimen (metformin and a sulfonylurea, each drug dosed at ≥50% of the recommended maximum), randomized to add either insulin glargine or rosiglitazone (Avandia).⁸ Both groups reduced HbA_1c equivalently after 24 weeks (–1.7% for glargine vs –1.5% for rosiglitazone). However, in patients with a baseline HbA_1c >9.5%, adding insulin glargine reduced HbA_1c significantly more than rosiglitazone.

Recommendations from others

A comparative analysis of guidelines on diabetes from 13 different countries (including the US) found general agreement in the recommendation to add a second oral agent to maximum doses of an initial agent in patients with poor glycemic control.⁹ However, no consensus was reached on the value or indications of combination therapy with oral agents and insulin.

The European Diabetes Policy Group recommends adding a second oral agent when the maximum dose of a single agent is reached, and using triple therapy when targets are not reached on maximum tolerated doses of 2 agents. Continued therapy with oral agents is advised when initiating insulin.¹⁰

Evidence summary

A systematic review evaluated beginning combination therapy (adding insulin to oral medication) compared with switching to insulin alone in patients with type 2 diabetes mellitus with inadequate glycemic control on oral medication.¹ Twenty RCTs studied a total of 1811 patients; glycemic control was the primary outcome measure. Oral medication comprised either sulfonylureas (75%), metformin (4%), or both (21%). Individual studies used different insulin dosing schedules and statistical measures. However, overall, combination therapy provided glucose control comparable with insulin alone. In only 1 small, low-quality study did insulin plus metformin reduce HbA_1c more than other combination therapy regimens or insulin alone. Ten studies reported a trend toward less weight gain with combination therapy that included metformin. Fourteen studies found the same incidence of hypoglycemic episodes in combination therapy and insulin alone.

Three later RCTs of overweight patients with inadequate control on oral agents (HbA_1c >7% on a sulfonylurea, metformin, or both) also compared beginning combination therapy with switching to insulin alone (with 70/30 or NPH insulin twice daily). In one study with 64 patients followed for 12 months, HbA_1c fell by 0.14% less (nonsignificant) in the combination therapy group (bedtime NPH plus sulfonylurea and metformin) than in the insulin alone group (70/30 twice daily).² The combination therapy group gained significantly less weight than the insulin-alone group (1.3 kg vs 4.2 kg; P=.01).

In the second study of 261 patients, the combination therapy group (glimepiride [Amaryl] plus bedtime NPH) had a significantly higher HbA_1c after 9 months than 2 groups using insulin alone (twice daily 70/30, and twice daily NPH insulin) (8.9% vs 8.3% and 8.4%).³ Mean weight gain was similar in all 3 groups but only a minority of patients reached a target HbA_1c of 6.5%. In the final study of only 16 patients, HbA1c after 6 months improved significantly and equally in both groups (baseline: 8.3%, combination therapy final: 6.8%; insulin alone final: 7.0%). However, the combination therapy group gained significantly less weight.⁴

An open-label RCT with 341 patients who were inadequately controlled on metformin compared beginning combination therapy (biphasic insulin aspart 30/70 [Novolog Mix 70/30] and metformin) with switching to insulin alone (biphasic insulin aspart 30/70).⁵ A third group added a second oral medication (sulfonylurea and metformin). After 16 weeks, patients taking combination therapy had a significantly lower HbA_1c than those on insulin alone (treatment difference 0.39%, P=.007).

Overall, combination therapy and 2 oral medications reduced HbA_1c by the same amount, but combination therapy reduced HbA_1c more in a subpopulation of patients with HbA_1c >9.0% at baseline (treatment difference 0.46%, P=.027). The group on insulin alone weighed significantly more (4.6 kg, P<.001) at the end of the trial than the group taking 2 oral medications.

An open-label RCT of 756 patients with inadequate glycemic control (HbA_1c >7.5%, mean 8.6%) on either 1 or 2 oral agents (70% taking both metformin and a sulfonylurea) compared combination therapy using bedtime insulin glargine with combination therapy using morning NPH.⁶ Each group titrated insulin doses to achieve a target fasting glucose ≤100. By 24 weeks, both groups had equivalently reduced HbA_1c (mean HbA_1c=6.96% with glargine, and 6.97% with NPH; P=NS), but fewer patients experienced nocturnal hypoglycemia with glargine than with NPH (33.2% vs 26.7%, P<.05).

Another open-label RCT evaluated 281 patients with at least 3 months of inadequate glycemic control (HbA_1c=7.4%–14.7%) on a sulfonylurea.⁷ Patients were randomized to a) switching to a combination of biphasic insulin aspart 30/70 plus pioglitazone, b) adding pioglitazone to the sulfonylurea, or c) switching to insulin alone (biphasic insulin aspart 30/70). After 18 weeks, insulin plus pioglitazone reduced HbA_1c significantly more than either glyburide plus pioglitazone (P=.005) or insulin alone (P=.005). However, the insulin plus pioglitazone group had the most weight gain (mean 4 kg, similar to other pioglitazone trials). There were no major hypoglycemic events.

Another open-label RCT evaluated 217 patients inadequately controlled (HbA_1c=7.5%–11%) on a 2-drug oral regimen (metformin and a sulfonylurea, each drug dosed at ≥50% of the recommended maximum), randomized to add either insulin glargine or rosiglitazone (Avandia).⁸ Both groups reduced HbA_1c equivalently after 24 weeks (–1.7% for glargine vs –1.5% for rosiglitazone). However, in patients with a baseline HbA_1c >9.5%, adding insulin glargine reduced HbA_1c significantly more than rosiglitazone.

Recommendations from others

A comparative analysis of guidelines on diabetes from 13 different countries (including the US) found general agreement in the recommendation to add a second oral agent to maximum doses of an initial agent in patients with poor glycemic control.⁹ However, no consensus was reached on the value or indications of combination therapy with oral agents and insulin.

The European Diabetes Policy Group recommends adding a second oral agent when the maximum dose of a single agent is reached, and using triple therapy when targets are not reached on maximum tolerated doses of 2 agents. Continued therapy with oral agents is advised when initiating insulin.¹⁰

Evidence summary

A systematic review evaluated beginning combination therapy (adding insulin to oral medication) compared with switching to insulin alone in patients with type 2 diabetes mellitus with inadequate glycemic control on oral medication.¹ Twenty RCTs studied a total of 1811 patients; glycemic control was the primary outcome measure. Oral medication comprised either sulfonylureas (75%), metformin (4%), or both (21%). Individual studies used different insulin dosing schedules and statistical measures. However, overall, combination therapy provided glucose control comparable with insulin alone. In only 1 small, low-quality study did insulin plus metformin reduce HbA_1c more than other combination therapy regimens or insulin alone. Ten studies reported a trend toward less weight gain with combination therapy that included metformin. Fourteen studies found the same incidence of hypoglycemic episodes in combination therapy and insulin alone.

Three later RCTs of overweight patients with inadequate control on oral agents (HbA_1c >7% on a sulfonylurea, metformin, or both) also compared beginning combination therapy with switching to insulin alone (with 70/30 or NPH insulin twice daily). In one study with 64 patients followed for 12 months, HbA_1c fell by 0.14% less (nonsignificant) in the combination therapy group (bedtime NPH plus sulfonylurea and metformin) than in the insulin alone group (70/30 twice daily).² The combination therapy group gained significantly less weight than the insulin-alone group (1.3 kg vs 4.2 kg; P=.01).

In the second study of 261 patients, the combination therapy group (glimepiride [Amaryl] plus bedtime NPH) had a significantly higher HbA_1c after 9 months than 2 groups using insulin alone (twice daily 70/30, and twice daily NPH insulin) (8.9% vs 8.3% and 8.4%).³ Mean weight gain was similar in all 3 groups but only a minority of patients reached a target HbA_1c of 6.5%. In the final study of only 16 patients, HbA1c after 6 months improved significantly and equally in both groups (baseline: 8.3%, combination therapy final: 6.8%; insulin alone final: 7.0%). However, the combination therapy group gained significantly less weight.⁴

An open-label RCT with 341 patients who were inadequately controlled on metformin compared beginning combination therapy (biphasic insulin aspart 30/70 [Novolog Mix 70/30] and metformin) with switching to insulin alone (biphasic insulin aspart 30/70).⁵ A third group added a second oral medication (sulfonylurea and metformin). After 16 weeks, patients taking combination therapy had a significantly lower HbA_1c than those on insulin alone (treatment difference 0.39%, P=.007).

Overall, combination therapy and 2 oral medications reduced HbA_1c by the same amount, but combination therapy reduced HbA_1c more in a subpopulation of patients with HbA_1c >9.0% at baseline (treatment difference 0.46%, P=.027). The group on insulin alone weighed significantly more (4.6 kg, P<.001) at the end of the trial than the group taking 2 oral medications.

An open-label RCT of 756 patients with inadequate glycemic control (HbA_1c >7.5%, mean 8.6%) on either 1 or 2 oral agents (70% taking both metformin and a sulfonylurea) compared combination therapy using bedtime insulin glargine with combination therapy using morning NPH.⁶ Each group titrated insulin doses to achieve a target fasting glucose ≤100. By 24 weeks, both groups had equivalently reduced HbA_1c (mean HbA_1c=6.96% with glargine, and 6.97% with NPH; P=NS), but fewer patients experienced nocturnal hypoglycemia with glargine than with NPH (33.2% vs 26.7%, P<.05).

Another open-label RCT evaluated 281 patients with at least 3 months of inadequate glycemic control (HbA_1c=7.4%–14.7%) on a sulfonylurea.⁷ Patients were randomized to a) switching to a combination of biphasic insulin aspart 30/70 plus pioglitazone, b) adding pioglitazone to the sulfonylurea, or c) switching to insulin alone (biphasic insulin aspart 30/70). After 18 weeks, insulin plus pioglitazone reduced HbA_1c significantly more than either glyburide plus pioglitazone (P=.005) or insulin alone (P=.005). However, the insulin plus pioglitazone group had the most weight gain (mean 4 kg, similar to other pioglitazone trials). There were no major hypoglycemic events.

Another open-label RCT evaluated 217 patients inadequately controlled (HbA_1c=7.5%–11%) on a 2-drug oral regimen (metformin and a sulfonylurea, each drug dosed at ≥50% of the recommended maximum), randomized to add either insulin glargine or rosiglitazone (Avandia).⁸ Both groups reduced HbA_1c equivalently after 24 weeks (–1.7% for glargine vs –1.5% for rosiglitazone). However, in patients with a baseline HbA_1c >9.5%, adding insulin glargine reduced HbA_1c significantly more than rosiglitazone.

Recommendations from others

A comparative analysis of guidelines on diabetes from 13 different countries (including the US) found general agreement in the recommendation to add a second oral agent to maximum doses of an initial agent in patients with poor glycemic control.⁹ However, no consensus was reached on the value or indications of combination therapy with oral agents and insulin.

The European Diabetes Policy Group recommends adding a second oral agent when the maximum dose of a single agent is reached, and using triple therapy when targets are not reached on maximum tolerated doses of 2 agents. Continued therapy with oral agents is advised when initiating insulin.¹⁰

Evidence summary

Oral agents are commonly prescribed for patients with diabetes mellitus type 2 when diet and exercise fail. Options for initiating therapy include sulfonylureas, metformin (Glucophage), α-glucosidase inhibitors, thiazolidinediones, and nonsulfonylurea secretagogues (repaglinide [Prandin] and nateglinide [Starlix]).

A systematic review with 31 placebo-controlled randomized trials (total n=12,185 patients) evaluated changes in HbA_1c with monotherapy using 5 different classes of oral agents ( TABLE ).¹ Except for the α-glucosidase inhibitor acarbose (Precose), which was less effective, all agents typically reduced HbA_1c by 1% to 2%. However, in an additional 19 out of 23 randomized head-to-head studies (total n=5396) included in the same systematic review, all classes showed equal efficacy.

Head-to-head studies are difficult to compare since hypoglycemic medications may reach peak effects at different times. An RCT compared glimepiride (Amaryl), pioglitazone (Actos), and metformin over 12 months of use by 114 patients with diabetes.³ There was no difference among the groups in overall HbA_1c reduction. However, glimepiride decreased HbA_1c rapidly over 1 month and reached a nadir at 4 months. Pioglitazone did not reduce HbA_1c until 6 months and reached its nadir at 7 to 9 months. Metformin produced an intermediate response.

A meta-analysis of head to head studies involving α-glucosidase inhibitors included 8 trials comparing acarbose with sulfonylureas. In pooled results, sulfonylureas trended towards greater HbA_1c reduction but did not reach significance (additional HbA_1c decrease 0.4%; 95% confidence interval [CI], 0%–0.8%).⁴

A meta-analysis of head-to-head studies involving metformin showed equal efficacy compared with injected insulin (2 trials, 811 participants), α-glucosidase inhibitors (2 trials, 223 participants), and non-sulfonylurea secretagogues (2 trials, 413 participants).⁵ In 12 trials with 2067 patients, metformin decreased HbA_1c more than sulfonylureas did (standardized mean difference [SMD] –0.14; 95% CI, –0.28 to –0.01). In 3 trials with 246 patients, metformin also produced greater HbA_1c decreases than thiazolidinediones (SMD –0.28; 95% CI, –0.52 to –0.03). In the United Kingdom Prospective Diabetes Study (UKPDS), metformin improved diabetes-related outcomes and all-cause mortality in obese patients (relative risk of mortality=0.73; 95% CI, 0.55–0.97; P=.03; number needed to treat [NNT]=19).⁶

A systematic review with 22 RCTs (total n=7370), ranging in length from 12 weeks to 3 years, compared 2 oral agents with a single oral agent or placebo.¹ Combinations of oral agents produced statistically significant additional improvement in HbA_1c in 21 of 22 studies. The magnitude of this effect across the studies was on the order of a 1% change in HbA_1c, although the data were not subject to a formal meta-analysis.

Inhaled insulin may expand the list of initial therapies for type 2 diabetes. A 12-week manufacturer-sponsored RCT with 134 patients (mean HbA_1c=9.5) compared inhaled insulin with rosiglitazone (Avandia).⁷ More patients using inhaled insulin achieved an HbA_1c <8.0 (82.7% vs 58.2%; P=.0003); however, inhaled insulin produced more adverse effects, including cough and hypoglycemia.

TABLE
Oral medications as monotherapy in type 2 diabetes mellitus^1,2

Recommendations from others

The International Diabetes Federation (IDF) recommends metformin as the initial oral agent unless contraindicated.⁸ A sulfonylurea is an acceptable alternative in patients who are not overweight. The IDF states that insulin should be added when oral agents fail.

The Institute for Clinical Systems Improvement (ICSI) says that the “single best choice drug for oral agent therapy for type 2 diabetes has not been determined” and must be chosen in the context of age, weight, and other comorbidities.⁹ The ICSI suggests metformin as an appropriate first agent for obese patients and recommends sulfonylureas or metformin as monotherapy for others because they are both economical and well tolerated. The American Diabetes Association does not specifically recommend a best initial agent or combination of agents for type 2 diabetes.¹⁰

Evidence summary

Oral agents are commonly prescribed for patients with diabetes mellitus type 2 when diet and exercise fail. Options for initiating therapy include sulfonylureas, metformin (Glucophage), α-glucosidase inhibitors, thiazolidinediones, and nonsulfonylurea secretagogues (repaglinide [Prandin] and nateglinide [Starlix]).

A systematic review with 31 placebo-controlled randomized trials (total n=12,185 patients) evaluated changes in HbA_1c with monotherapy using 5 different classes of oral agents ( TABLE ).¹ Except for the α-glucosidase inhibitor acarbose (Precose), which was less effective, all agents typically reduced HbA_1c by 1% to 2%. However, in an additional 19 out of 23 randomized head-to-head studies (total n=5396) included in the same systematic review, all classes showed equal efficacy.

Head-to-head studies are difficult to compare since hypoglycemic medications may reach peak effects at different times. An RCT compared glimepiride (Amaryl), pioglitazone (Actos), and metformin over 12 months of use by 114 patients with diabetes.³ There was no difference among the groups in overall HbA_1c reduction. However, glimepiride decreased HbA_1c rapidly over 1 month and reached a nadir at 4 months. Pioglitazone did not reduce HbA_1c until 6 months and reached its nadir at 7 to 9 months. Metformin produced an intermediate response.

A meta-analysis of head to head studies involving α-glucosidase inhibitors included 8 trials comparing acarbose with sulfonylureas. In pooled results, sulfonylureas trended towards greater HbA_1c reduction but did not reach significance (additional HbA_1c decrease 0.4%; 95% confidence interval [CI], 0%–0.8%).⁴

A meta-analysis of head-to-head studies involving metformin showed equal efficacy compared with injected insulin (2 trials, 811 participants), α-glucosidase inhibitors (2 trials, 223 participants), and non-sulfonylurea secretagogues (2 trials, 413 participants).⁵ In 12 trials with 2067 patients, metformin decreased HbA_1c more than sulfonylureas did (standardized mean difference [SMD] –0.14; 95% CI, –0.28 to –0.01). In 3 trials with 246 patients, metformin also produced greater HbA_1c decreases than thiazolidinediones (SMD –0.28; 95% CI, –0.52 to –0.03). In the United Kingdom Prospective Diabetes Study (UKPDS), metformin improved diabetes-related outcomes and all-cause mortality in obese patients (relative risk of mortality=0.73; 95% CI, 0.55–0.97; P=.03; number needed to treat [NNT]=19).⁶

A systematic review with 22 RCTs (total n=7370), ranging in length from 12 weeks to 3 years, compared 2 oral agents with a single oral agent or placebo.¹ Combinations of oral agents produced statistically significant additional improvement in HbA_1c in 21 of 22 studies. The magnitude of this effect across the studies was on the order of a 1% change in HbA_1c, although the data were not subject to a formal meta-analysis.

Inhaled insulin may expand the list of initial therapies for type 2 diabetes. A 12-week manufacturer-sponsored RCT with 134 patients (mean HbA_1c=9.5) compared inhaled insulin with rosiglitazone (Avandia).⁷ More patients using inhaled insulin achieved an HbA_1c <8.0 (82.7% vs 58.2%; P=.0003); however, inhaled insulin produced more adverse effects, including cough and hypoglycemia.

TABLE
Oral medications as monotherapy in type 2 diabetes mellitus^1,2

Recommendations from others

The International Diabetes Federation (IDF) recommends metformin as the initial oral agent unless contraindicated.⁸ A sulfonylurea is an acceptable alternative in patients who are not overweight. The IDF states that insulin should be added when oral agents fail.

The Institute for Clinical Systems Improvement (ICSI) says that the “single best choice drug for oral agent therapy for type 2 diabetes has not been determined” and must be chosen in the context of age, weight, and other comorbidities.⁹ The ICSI suggests metformin as an appropriate first agent for obese patients and recommends sulfonylureas or metformin as monotherapy for others because they are both economical and well tolerated. The American Diabetes Association does not specifically recommend a best initial agent or combination of agents for type 2 diabetes.¹⁰

Evidence summary

Oral agents are commonly prescribed for patients with diabetes mellitus type 2 when diet and exercise fail. Options for initiating therapy include sulfonylureas, metformin (Glucophage), α-glucosidase inhibitors, thiazolidinediones, and nonsulfonylurea secretagogues (repaglinide [Prandin] and nateglinide [Starlix]).

A systematic review with 31 placebo-controlled randomized trials (total n=12,185 patients) evaluated changes in HbA_1c with monotherapy using 5 different classes of oral agents ( TABLE ).¹ Except for the α-glucosidase inhibitor acarbose (Precose), which was less effective, all agents typically reduced HbA_1c by 1% to 2%. However, in an additional 19 out of 23 randomized head-to-head studies (total n=5396) included in the same systematic review, all classes showed equal efficacy.

Head-to-head studies are difficult to compare since hypoglycemic medications may reach peak effects at different times. An RCT compared glimepiride (Amaryl), pioglitazone (Actos), and metformin over 12 months of use by 114 patients with diabetes.³ There was no difference among the groups in overall HbA_1c reduction. However, glimepiride decreased HbA_1c rapidly over 1 month and reached a nadir at 4 months. Pioglitazone did not reduce HbA_1c until 6 months and reached its nadir at 7 to 9 months. Metformin produced an intermediate response.

A meta-analysis of head to head studies involving α-glucosidase inhibitors included 8 trials comparing acarbose with sulfonylureas. In pooled results, sulfonylureas trended towards greater HbA_1c reduction but did not reach significance (additional HbA_1c decrease 0.4%; 95% confidence interval [CI], 0%–0.8%).⁴

A meta-analysis of head-to-head studies involving metformin showed equal efficacy compared with injected insulin (2 trials, 811 participants), α-glucosidase inhibitors (2 trials, 223 participants), and non-sulfonylurea secretagogues (2 trials, 413 participants).⁵ In 12 trials with 2067 patients, metformin decreased HbA_1c more than sulfonylureas did (standardized mean difference [SMD] –0.14; 95% CI, –0.28 to –0.01). In 3 trials with 246 patients, metformin also produced greater HbA_1c decreases than thiazolidinediones (SMD –0.28; 95% CI, –0.52 to –0.03). In the United Kingdom Prospective Diabetes Study (UKPDS), metformin improved diabetes-related outcomes and all-cause mortality in obese patients (relative risk of mortality=0.73; 95% CI, 0.55–0.97; P=.03; number needed to treat [NNT]=19).⁶

A systematic review with 22 RCTs (total n=7370), ranging in length from 12 weeks to 3 years, compared 2 oral agents with a single oral agent or placebo.¹ Combinations of oral agents produced statistically significant additional improvement in HbA_1c in 21 of 22 studies. The magnitude of this effect across the studies was on the order of a 1% change in HbA_1c, although the data were not subject to a formal meta-analysis.

Inhaled insulin may expand the list of initial therapies for type 2 diabetes. A 12-week manufacturer-sponsored RCT with 134 patients (mean HbA_1c=9.5) compared inhaled insulin with rosiglitazone (Avandia).⁷ More patients using inhaled insulin achieved an HbA_1c <8.0 (82.7% vs 58.2%; P=.0003); however, inhaled insulin produced more adverse effects, including cough and hypoglycemia.

TABLE
Oral medications as monotherapy in type 2 diabetes mellitus^1,2

Recommendations from others

The International Diabetes Federation (IDF) recommends metformin as the initial oral agent unless contraindicated.⁸ A sulfonylurea is an acceptable alternative in patients who are not overweight. The IDF states that insulin should be added when oral agents fail.

The Institute for Clinical Systems Improvement (ICSI) says that the “single best choice drug for oral agent therapy for type 2 diabetes has not been determined” and must be chosen in the context of age, weight, and other comorbidities.⁹ The ICSI suggests metformin as an appropriate first agent for obese patients and recommends sulfonylureas or metformin as monotherapy for others because they are both economical and well tolerated. The American Diabetes Association does not specifically recommend a best initial agent or combination of agents for type 2 diabetes.¹⁰