Clinical Inquiries

Evidence summary

No therapies are known to improve long-term lung function in COPD; the goal of disease-moderating therapy is therefore to slow the rate of decline compared with the expected rate. All of the studies reviewed used FEV₁ as an objective measure of whether ICS reduce this rate of decline in lung function.

Two recent meta-analyses evaluating medium- to high-dose ICS effects on FEV₁ decline provided conflicting results. One meta-analysis evaluated 8 controlled clinical trials lasting at least 2 years (n=3715) and found that, when compared with placebo, ICS significantly reduced the rate of FEV₁ decline by 7.7 mL/year (P=.02) and that high-dose ICS had a greater effect of 9.9 mL/year (P=.01).¹ Another meta-analysis of 6 randomized, placebo-controlled trials with a duration of at least 2 years (n=3571) found a nonsignificant trend in favor of ICS, with a difference in FEV₁ decline of 5.31 mL/year (P=.08) between the ICS and placebo groups.²

The differences observed in these 2 meta-analyses may be explained by the authors using slightly different approximations to the standard error, applying slightly different statistical analytical methods, and using different inclusion criteria for trials. However, 5 of the trials in these reviews were the same. Both meta-analyses determined only rate of lung function decline and did not evaluate clinical outcomes.

A trial not included in the previously mentioned meta-analyses evaluated post-bronchodilator FEV₁ decline in 48 patients with early signs and symptoms of COPD for 2 years.³ Subjects were assigned to medium-dose fluticasone propionate or placebo. Early initiation of ICS treatment did not affect the progressive deterioration of lung function as no modifying effect on annual FEV₁ decline was observed, however, the study only had power to detect a 60-mL annual drop in FEV₁.

Meta-analyses and trials evaluating COPD progression have focused on a disease-oriented outcome (the rate of FEV₁ decline). However, patient-oriented outcomes such as exacerbation frequency, hospitalization, health-related quality of life, and mortality might be more important measures of successful therapy. Although such patient-oriented outcomes are not the focus of this review or the included meta-analyses, a few of the small randomized controlled trials included in these meta-analyses suggest that ICS may improve such patient-oriented outcomes. Notably, exacerbation rates significantly decreased by 25% (P=.026), and health status improved (P=.0043) among patients with moderate to severe COPD who were taking fluticasone compared with those taking placebo.⁴ In mild to moderate COPD, patients treated with triamcinolone had fewer respiratory symptoms (P=.005), fewer visits to a physician because of respiratory illness (P=.003), and improved airway reactivity (P=.02).⁵ Some systematic reviews and other randomized trials suggest that ICS have significant benefit on these patient outcomes.⁶

Recommendations from others

Scientists from the National Heart, Lung, and Blood Institute and the World Health Organization provided an update of the Global Initiative for Chronic Obstructive Lung Disease (GOLD) in 2003.⁷ They reported that regular treatment with ICS does not modify the long-term decline of FEV₁ in patients with COPD. However, they recommended treatment with ICS for symptomatic COPD patients with an FEV₁ less than 50% of predicted (stage III: severe COPD and stage IV: very severe COPD) and repeated exacerbations (ie, 3 in the last 3 years). Guidelines from other countries also suggest that ICS do not affect the progression of COPD, but support the use of ICS for patients with severe COPD and repeated exacerbations.^8-10

Evidence summary

No therapies are known to improve long-term lung function in COPD; the goal of disease-moderating therapy is therefore to slow the rate of decline compared with the expected rate. All of the studies reviewed used FEV₁ as an objective measure of whether ICS reduce this rate of decline in lung function.

Two recent meta-analyses evaluating medium- to high-dose ICS effects on FEV₁ decline provided conflicting results. One meta-analysis evaluated 8 controlled clinical trials lasting at least 2 years (n=3715) and found that, when compared with placebo, ICS significantly reduced the rate of FEV₁ decline by 7.7 mL/year (P=.02) and that high-dose ICS had a greater effect of 9.9 mL/year (P=.01).¹ Another meta-analysis of 6 randomized, placebo-controlled trials with a duration of at least 2 years (n=3571) found a nonsignificant trend in favor of ICS, with a difference in FEV₁ decline of 5.31 mL/year (P=.08) between the ICS and placebo groups.²

The differences observed in these 2 meta-analyses may be explained by the authors using slightly different approximations to the standard error, applying slightly different statistical analytical methods, and using different inclusion criteria for trials. However, 5 of the trials in these reviews were the same. Both meta-analyses determined only rate of lung function decline and did not evaluate clinical outcomes.

A trial not included in the previously mentioned meta-analyses evaluated post-bronchodilator FEV₁ decline in 48 patients with early signs and symptoms of COPD for 2 years.³ Subjects were assigned to medium-dose fluticasone propionate or placebo. Early initiation of ICS treatment did not affect the progressive deterioration of lung function as no modifying effect on annual FEV₁ decline was observed, however, the study only had power to detect a 60-mL annual drop in FEV₁.

Meta-analyses and trials evaluating COPD progression have focused on a disease-oriented outcome (the rate of FEV₁ decline). However, patient-oriented outcomes such as exacerbation frequency, hospitalization, health-related quality of life, and mortality might be more important measures of successful therapy. Although such patient-oriented outcomes are not the focus of this review or the included meta-analyses, a few of the small randomized controlled trials included in these meta-analyses suggest that ICS may improve such patient-oriented outcomes. Notably, exacerbation rates significantly decreased by 25% (P=.026), and health status improved (P=.0043) among patients with moderate to severe COPD who were taking fluticasone compared with those taking placebo.⁴ In mild to moderate COPD, patients treated with triamcinolone had fewer respiratory symptoms (P=.005), fewer visits to a physician because of respiratory illness (P=.003), and improved airway reactivity (P=.02).⁵ Some systematic reviews and other randomized trials suggest that ICS have significant benefit on these patient outcomes.⁶

Recommendations from others

Scientists from the National Heart, Lung, and Blood Institute and the World Health Organization provided an update of the Global Initiative for Chronic Obstructive Lung Disease (GOLD) in 2003.⁷ They reported that regular treatment with ICS does not modify the long-term decline of FEV₁ in patients with COPD. However, they recommended treatment with ICS for symptomatic COPD patients with an FEV₁ less than 50% of predicted (stage III: severe COPD and stage IV: very severe COPD) and repeated exacerbations (ie, 3 in the last 3 years). Guidelines from other countries also suggest that ICS do not affect the progression of COPD, but support the use of ICS for patients with severe COPD and repeated exacerbations.^8-10

Evidence summary

No therapies are known to improve long-term lung function in COPD; the goal of disease-moderating therapy is therefore to slow the rate of decline compared with the expected rate. All of the studies reviewed used FEV₁ as an objective measure of whether ICS reduce this rate of decline in lung function.

Two recent meta-analyses evaluating medium- to high-dose ICS effects on FEV₁ decline provided conflicting results. One meta-analysis evaluated 8 controlled clinical trials lasting at least 2 years (n=3715) and found that, when compared with placebo, ICS significantly reduced the rate of FEV₁ decline by 7.7 mL/year (P=.02) and that high-dose ICS had a greater effect of 9.9 mL/year (P=.01).¹ Another meta-analysis of 6 randomized, placebo-controlled trials with a duration of at least 2 years (n=3571) found a nonsignificant trend in favor of ICS, with a difference in FEV₁ decline of 5.31 mL/year (P=.08) between the ICS and placebo groups.²

The differences observed in these 2 meta-analyses may be explained by the authors using slightly different approximations to the standard error, applying slightly different statistical analytical methods, and using different inclusion criteria for trials. However, 5 of the trials in these reviews were the same. Both meta-analyses determined only rate of lung function decline and did not evaluate clinical outcomes.

A trial not included in the previously mentioned meta-analyses evaluated post-bronchodilator FEV₁ decline in 48 patients with early signs and symptoms of COPD for 2 years.³ Subjects were assigned to medium-dose fluticasone propionate or placebo. Early initiation of ICS treatment did not affect the progressive deterioration of lung function as no modifying effect on annual FEV₁ decline was observed, however, the study only had power to detect a 60-mL annual drop in FEV₁.

Meta-analyses and trials evaluating COPD progression have focused on a disease-oriented outcome (the rate of FEV₁ decline). However, patient-oriented outcomes such as exacerbation frequency, hospitalization, health-related quality of life, and mortality might be more important measures of successful therapy. Although such patient-oriented outcomes are not the focus of this review or the included meta-analyses, a few of the small randomized controlled trials included in these meta-analyses suggest that ICS may improve such patient-oriented outcomes. Notably, exacerbation rates significantly decreased by 25% (P=.026), and health status improved (P=.0043) among patients with moderate to severe COPD who were taking fluticasone compared with those taking placebo.⁴ In mild to moderate COPD, patients treated with triamcinolone had fewer respiratory symptoms (P=.005), fewer visits to a physician because of respiratory illness (P=.003), and improved airway reactivity (P=.02).⁵ Some systematic reviews and other randomized trials suggest that ICS have significant benefit on these patient outcomes.⁶

Recommendations from others

Scientists from the National Heart, Lung, and Blood Institute and the World Health Organization provided an update of the Global Initiative for Chronic Obstructive Lung Disease (GOLD) in 2003.⁷ They reported that regular treatment with ICS does not modify the long-term decline of FEV₁ in patients with COPD. However, they recommended treatment with ICS for symptomatic COPD patients with an FEV₁ less than 50% of predicted (stage III: severe COPD and stage IV: very severe COPD) and repeated exacerbations (ie, 3 in the last 3 years). Guidelines from other countries also suggest that ICS do not affect the progression of COPD, but support the use of ICS for patients with severe COPD and repeated exacerbations.^8-10

Evidence summary

Office blood pressure (OBP) has traditionally been used in long-term trials to describe the relationship between blood pressure and cardiovascular morbidity and mortality, as well as to establish the efficacy of antihypertensive drug therapy. A prospective randomized trial demonstrating the relationship between therapy based on SMBP to these same outcomes is in progress.¹

Two large prospective cohort studies of the relationship between SMBP and morbidity and mortality made comparative baseline blood pressure measurements and followed the cohorts without suggestions or attempts to change management. The first was a rural population-based study with 1789 subjects (90% of the population) from Ohasama, Japan.² Mean follow-up was 6.6 years with less than 1% dropout rate. The second large cohort study (SHEAF trial) included patients 60 years old with the diagnosis of hypertension.³ A total of 4939 cases were analyzed. Mean follow-up was 3.2 years with less than 1% dropout rate. Both studies show that each mm Hg increase in SMBP was a better predictor of cardiovascular events than an equivalent increase in OBP (Table 1).

Office blood pressure measurements exhibit large variability (decreased precision) and are subject to multiple biases (decreased accuracy). Self-measured blood pressures at home became common when “white-coat hypertension” was recognized to be clinically significant. It allows for a larger number of measurements for individual patients, resulting in greater precision than OBP.⁴ SMBP correlates better than OBP with surrogate measures of hypertensive control, such as ambulatory blood pressure measurement⁵ and left ventricular mass.⁶ Thus, SMBP might some day become the gold standard for defining hypertension in the clinical setting. Meanwhile, the correlation between OBP and SMBP can be derived via three different mathematical models using data from multiple studies. The accepted cutoff for SMBP defined hypertension is 135/85 mm Hg.⁷

The THOP trial⁸ was a single-blinded, randomized controlled trial of hypertensive treatment based on SMBP vs OBP. Four hundred patients were randomized to SMBP or OBP, with medication adjustments made by a blinded clinician. The trial design called for both treatment groups to be titrated to a diastolic blood pressure of 80 to 89 mm Hg. The follow-up was approximately 1 year. Graphical data indicate that both groups were equally effective in meeting the blood pressure goals outlined in the methods.

Other differences in outcomes were proportional to the known difference in normotensive reference ranges (eg, that OBP tend to run higher than SMBP). Patients in the SMBP group were put on less-intensive drug treatment and incurred slightly lower medical costs. SMBP patients were twice as likely to have their blood pressure medication discontinued, possibly indicating SMBP helped to identify white-coat hypertension.

TABLE 1
Increase in cardiovascular mortality for each 1 mm Hg increase in blood pressure

Recommendations from others

In addition to diagnosing white-coat hypertension, World Health Organization/International Society of Hypertension Guidelines Committee has recommended that home blood pressure measurement is useful in the following circumstances:⁹

• unusual variability of blood pressure over the same or different visits
• office hypertension in subjects with low cardiovascular risk
• symptoms suggesting hypotensive episodes
• hypertension resistant to drug treatment.

Standardization and validation protocols are available from the Association for the Advancement of Medical Instrumentation,¹⁰ European Hypertension Society,¹¹ or the British Hypertension Society (available at www.hyp.ac.uk/bhs/bp_ monitors/automatic.htm). Relatively few of the hundreds of available blood pressure measurement devices available meet these criteria. The most current Association for the Advancement of Medical Instrumentation standards are labeled as ANSI/AAMI-SP10:2002/A1:2003 standards. Table 2 lists some devices that meet the various protocols. Devices in this market change rapidly, so buyers should confirm the device they are evaluating meets current standards.

TABLE 2
Devices that meet standards for home BP measurement

Evidence summary

Office blood pressure (OBP) has traditionally been used in long-term trials to describe the relationship between blood pressure and cardiovascular morbidity and mortality, as well as to establish the efficacy of antihypertensive drug therapy. A prospective randomized trial demonstrating the relationship between therapy based on SMBP to these same outcomes is in progress.¹

Two large prospective cohort studies of the relationship between SMBP and morbidity and mortality made comparative baseline blood pressure measurements and followed the cohorts without suggestions or attempts to change management. The first was a rural population-based study with 1789 subjects (90% of the population) from Ohasama, Japan.² Mean follow-up was 6.6 years with less than 1% dropout rate. The second large cohort study (SHEAF trial) included patients 60 years old with the diagnosis of hypertension.³ A total of 4939 cases were analyzed. Mean follow-up was 3.2 years with less than 1% dropout rate. Both studies show that each mm Hg increase in SMBP was a better predictor of cardiovascular events than an equivalent increase in OBP (Table 1).

Office blood pressure measurements exhibit large variability (decreased precision) and are subject to multiple biases (decreased accuracy). Self-measured blood pressures at home became common when “white-coat hypertension” was recognized to be clinically significant. It allows for a larger number of measurements for individual patients, resulting in greater precision than OBP.⁴ SMBP correlates better than OBP with surrogate measures of hypertensive control, such as ambulatory blood pressure measurement⁵ and left ventricular mass.⁶ Thus, SMBP might some day become the gold standard for defining hypertension in the clinical setting. Meanwhile, the correlation between OBP and SMBP can be derived via three different mathematical models using data from multiple studies. The accepted cutoff for SMBP defined hypertension is 135/85 mm Hg.⁷

The THOP trial⁸ was a single-blinded, randomized controlled trial of hypertensive treatment based on SMBP vs OBP. Four hundred patients were randomized to SMBP or OBP, with medication adjustments made by a blinded clinician. The trial design called for both treatment groups to be titrated to a diastolic blood pressure of 80 to 89 mm Hg. The follow-up was approximately 1 year. Graphical data indicate that both groups were equally effective in meeting the blood pressure goals outlined in the methods.

Other differences in outcomes were proportional to the known difference in normotensive reference ranges (eg, that OBP tend to run higher than SMBP). Patients in the SMBP group were put on less-intensive drug treatment and incurred slightly lower medical costs. SMBP patients were twice as likely to have their blood pressure medication discontinued, possibly indicating SMBP helped to identify white-coat hypertension.

TABLE 1
Increase in cardiovascular mortality for each 1 mm Hg increase in blood pressure

Recommendations from others

In addition to diagnosing white-coat hypertension, World Health Organization/International Society of Hypertension Guidelines Committee has recommended that home blood pressure measurement is useful in the following circumstances:⁹

• unusual variability of blood pressure over the same or different visits
• office hypertension in subjects with low cardiovascular risk
• symptoms suggesting hypotensive episodes
• hypertension resistant to drug treatment.

Standardization and validation protocols are available from the Association for the Advancement of Medical Instrumentation,¹⁰ European Hypertension Society,¹¹ or the British Hypertension Society (available at www.hyp.ac.uk/bhs/bp_ monitors/automatic.htm). Relatively few of the hundreds of available blood pressure measurement devices available meet these criteria. The most current Association for the Advancement of Medical Instrumentation standards are labeled as ANSI/AAMI-SP10:2002/A1:2003 standards. Table 2 lists some devices that meet the various protocols. Devices in this market change rapidly, so buyers should confirm the device they are evaluating meets current standards.

TABLE 2
Devices that meet standards for home BP measurement

Evidence summary

Office blood pressure (OBP) has traditionally been used in long-term trials to describe the relationship between blood pressure and cardiovascular morbidity and mortality, as well as to establish the efficacy of antihypertensive drug therapy. A prospective randomized trial demonstrating the relationship between therapy based on SMBP to these same outcomes is in progress.¹

Two large prospective cohort studies of the relationship between SMBP and morbidity and mortality made comparative baseline blood pressure measurements and followed the cohorts without suggestions or attempts to change management. The first was a rural population-based study with 1789 subjects (90% of the population) from Ohasama, Japan.² Mean follow-up was 6.6 years with less than 1% dropout rate. The second large cohort study (SHEAF trial) included patients 60 years old with the diagnosis of hypertension.³ A total of 4939 cases were analyzed. Mean follow-up was 3.2 years with less than 1% dropout rate. Both studies show that each mm Hg increase in SMBP was a better predictor of cardiovascular events than an equivalent increase in OBP (Table 1).

Office blood pressure measurements exhibit large variability (decreased precision) and are subject to multiple biases (decreased accuracy). Self-measured blood pressures at home became common when “white-coat hypertension” was recognized to be clinically significant. It allows for a larger number of measurements for individual patients, resulting in greater precision than OBP.⁴ SMBP correlates better than OBP with surrogate measures of hypertensive control, such as ambulatory blood pressure measurement⁵ and left ventricular mass.⁶ Thus, SMBP might some day become the gold standard for defining hypertension in the clinical setting. Meanwhile, the correlation between OBP and SMBP can be derived via three different mathematical models using data from multiple studies. The accepted cutoff for SMBP defined hypertension is 135/85 mm Hg.⁷

The THOP trial⁸ was a single-blinded, randomized controlled trial of hypertensive treatment based on SMBP vs OBP. Four hundred patients were randomized to SMBP or OBP, with medication adjustments made by a blinded clinician. The trial design called for both treatment groups to be titrated to a diastolic blood pressure of 80 to 89 mm Hg. The follow-up was approximately 1 year. Graphical data indicate that both groups were equally effective in meeting the blood pressure goals outlined in the methods.

Other differences in outcomes were proportional to the known difference in normotensive reference ranges (eg, that OBP tend to run higher than SMBP). Patients in the SMBP group were put on less-intensive drug treatment and incurred slightly lower medical costs. SMBP patients were twice as likely to have their blood pressure medication discontinued, possibly indicating SMBP helped to identify white-coat hypertension.

TABLE 1
Increase in cardiovascular mortality for each 1 mm Hg increase in blood pressure

Recommendations from others

In addition to diagnosing white-coat hypertension, World Health Organization/International Society of Hypertension Guidelines Committee has recommended that home blood pressure measurement is useful in the following circumstances:⁹

• unusual variability of blood pressure over the same or different visits
• office hypertension in subjects with low cardiovascular risk
• symptoms suggesting hypotensive episodes
• hypertension resistant to drug treatment.

Standardization and validation protocols are available from the Association for the Advancement of Medical Instrumentation,¹⁰ European Hypertension Society,¹¹ or the British Hypertension Society (available at www.hyp.ac.uk/bhs/bp_ monitors/automatic.htm). Relatively few of the hundreds of available blood pressure measurement devices available meet these criteria. The most current Association for the Advancement of Medical Instrumentation standards are labeled as ANSI/AAMI-SP10:2002/A1:2003 standards. Table 2 lists some devices that meet the various protocols. Devices in this market change rapidly, so buyers should confirm the device they are evaluating meets current standards.

TABLE 2
Devices that meet standards for home BP measurement

TABLE
Nugent’s Criteria

Evidence summary

Bacterial vaginosis in early pregnancy is a risk factor for preterm delivery.² The role of BV in preterm labor is not well understood, but it has been consistently associated with preterm labor and delivery. The detection of BV in early pregnancy seems to be a stronger risk factor for preterm delivery than BV in later pregnancy.

Studies evaluating the screening and treatment of BV in women at risk for preterm delivery have demonstrated varying results. Most treatment studies have excluded women who are in the first trimester. A meta-analysis of 7 RCTs reviewed the evidence of screening for BV in pregnancy.³ In this meta-analysis, 5 of the trials specified that women were asymptomatic, and the other 2 did not comment on whether the women were symptomatic or not. In general, there was no benefit to routine screening and treatment of BV.

However, a subgroup of high-risk women seems to benefit from screening and treatment. They defined high-risk women as those have had a preterm delivery, premature rupture of membranes, birth weight <2500 g, or spontaneous abortion. Treating BV in women with a high-risk pregnancy decreased preterm delivery (absolute risk reduction [ARR]=0.22; 90% confidence interval [CI], 0.13–0.31; number needed to treat [NNT]=4.5) regardless of antibiotic choice. However, 2 trials of high-risk women who were empirically treated for BV, but did not have BV, showed an increase in preterm delivery less than 34 weeks (number needed to harm [NNH]=11).

A new study evaluating screening for vaginal infections in pregnancy has demonstrated a reduction in preterm delivery.⁴ In this study, looking at a general obstetrical population in Austria, 4429 asymptomatic pregnant women between 15 and 19.6 weeks gestation had a vaginal screen for bacterial vaginosis, candidiasis and trichomoniasis. The 2048 women in the intervention group were given the results of the screen from their maternity care provider. The 2097 women in the control group and their providers did not receive the results of the vaginal screen. There were 447 women in the intervention group and 441 women in the control group with positive screens. Using the Nugent criteria, women who were diagnosed with BV received a 6-day course of intravaginal clindamycin 2% cream. Those with positive test results for Candida were treated with intravaginal clotrimazole 0.1 g; those with positive results for trichomonas received intravaginal metronidazole 500 mg for 7 days. After treatment, women with a positive test result in the intervention group had a second vaginal smear between 24 and 28 weeks. Persistent BV was treated with oral clindamycin 300 mg twice daily for 7 days. If Candida or trichomonas were noted, women were treated with the intravaginal clotrimazole or metronidazole. A statistically significant reduction was seen in preterm births in the intervention group(3.0% vs 5.3%, 95% CI, 1.2–3.6; P=.0001; number needed to screen=44).

A large study in 2000 that looked at the use of metronidazole in the treatment of asymptomatic women for BV did not demonstrate any reduction in preterm birth.⁵ In this study, 21,965 asymptomatic women between 8 and 22 weeks gestation were screened for BV with Gram stain using Nugent’s criteria. Then, 1953 women with BV were randomized to receive either 1 g of metronidazole orally for 2 days or placebo. Between 24 and 29 weeks, all of the women were then rescreened for BV by Gram stain. Even if the results were negative, women received another course of the metronidazole or placebo. In this study, preterm delivery rates did not improve for either low- or high-risk women. Specifically, a subgroup analysis of 213 women with previous preterm delivery did not show any benefit to treatment with metronidazole.

In 2003, a Cochrane meta-analysis of 5 studies involving 622 women with previous preterm birth showed a decrease in the risk of low birth weight infants born to women receiving antibiotics vs placebo for the treatment of BV (odds ratio [OR]=0.31; 95% CI, 0.13–0.75).⁶ Treatment also decreased the risk of preterm-prelabor rupture of membranes (OR=0.14; 95% CI, 0.05–0.38) compared with placebo. Unfortunately, these studies did not always specify whether women were asymptomatic for BV infection. In many of the trials, symptomatic women were excluded as they were automatically treated with antibiotics.

In 2003, 2 RCTs evaluating the early treatment of asymptomatic BV in low- and high-risk patients showed a decrease in preterm labor. The first RCT included 494 asymptomatic pregnant women who presented for prenatal care between 12 and 22 weeks gestation. If women had BV detected by Gram stain using Nugent’s criteria, they were randomized to receive either 300 mg oral clindamycin twice daily for 5 days or placebo. In the general population, treatment with clindamycin reduced the rate of late miscarriage and spontaneous preterm delivery by 10.4% (95% CI, 5.0–15.8). In women with a previous preterm delivery or a late miscarriage the proportion of preterm delivery or late miscarriage was reduced (16.6% vs 42%).⁷

The second RCT included 409 asymptomatic women between 13 and 20 weeks gestation with BV by Gram stain using Nugent’s criteria. Investigators randomized women to intravaginal clindamycin each night for 3 days. At a second visit, 20 to 24 days after treatment, women were retested for BV and if they were positive, they received a 7-day course of intravaginal clindamycin or placebo based on the previous randomization. In this study, the incidence of preterm birth was reduced from 10% to 4% (relative risk [RR]=0.38; 95% CI, 0.16–0.90; NNT=17). This study only included 21 women with previous preterm delivery and a subgroup analysis was not performed.⁸

Intravaginal clindamycin has been associated with worse pregnancy outcomes for patients who do not have bacterial vaginosis. A randomized trial of the prophylactic intravaginal clindamycin 2% cream to prevent preterm birth in high-risk women showed an increase in spontaneous preterm delivery in women who actually used all of the medication and did not have BV (NNH=12.3; P<.05).⁹

Recommendations from others

The US Preventive Services Task Force concludes that the evidence is insufficient to recommend for or against routinely screening for BV for high-risk pregnant women. Furthermore, they recommend against screening for average risk women.¹⁰

The Centers for Disease Control and Prevention recommends that high-risk pregnant women (eg, those women who have had a previous preterm delivery) with asymptomatic BV may be evaluated for treatment. The recommended treatment regimens are metronidazole 500 mg orally twice a day for 5 days, metronidazole gel intravaginally for 5 days, or clindamycin cream intravaginally for 7 days.¹¹

The Cochrane Pregnancy and Childbirth Group finds no evidence supporting routine screening and treatment for asymptomatic bacterial vaginosis in pregnancy, except possibly for women with a history of preterm birth.⁶

The American College of Obstetrics and Gynecology summarizes no data supports screening for BV to prevent preterm birth. Their bulletin references a subgroup of women with previous preterm birth who did show benefit from treatment for BV, but the authors speculated that reanalysis with the inclusion of the largest trial to date, which did not show a benefit for this subgroup, might nullify these results.¹²

TABLE
Nugent’s Criteria

Evidence summary

Bacterial vaginosis in early pregnancy is a risk factor for preterm delivery.² The role of BV in preterm labor is not well understood, but it has been consistently associated with preterm labor and delivery. The detection of BV in early pregnancy seems to be a stronger risk factor for preterm delivery than BV in later pregnancy.

Studies evaluating the screening and treatment of BV in women at risk for preterm delivery have demonstrated varying results. Most treatment studies have excluded women who are in the first trimester. A meta-analysis of 7 RCTs reviewed the evidence of screening for BV in pregnancy.³ In this meta-analysis, 5 of the trials specified that women were asymptomatic, and the other 2 did not comment on whether the women were symptomatic or not. In general, there was no benefit to routine screening and treatment of BV.

However, a subgroup of high-risk women seems to benefit from screening and treatment. They defined high-risk women as those have had a preterm delivery, premature rupture of membranes, birth weight <2500 g, or spontaneous abortion. Treating BV in women with a high-risk pregnancy decreased preterm delivery (absolute risk reduction [ARR]=0.22; 90% confidence interval [CI], 0.13–0.31; number needed to treat [NNT]=4.5) regardless of antibiotic choice. However, 2 trials of high-risk women who were empirically treated for BV, but did not have BV, showed an increase in preterm delivery less than 34 weeks (number needed to harm [NNH]=11).

A new study evaluating screening for vaginal infections in pregnancy has demonstrated a reduction in preterm delivery.⁴ In this study, looking at a general obstetrical population in Austria, 4429 asymptomatic pregnant women between 15 and 19.6 weeks gestation had a vaginal screen for bacterial vaginosis, candidiasis and trichomoniasis. The 2048 women in the intervention group were given the results of the screen from their maternity care provider. The 2097 women in the control group and their providers did not receive the results of the vaginal screen. There were 447 women in the intervention group and 441 women in the control group with positive screens. Using the Nugent criteria, women who were diagnosed with BV received a 6-day course of intravaginal clindamycin 2% cream. Those with positive test results for Candida were treated with intravaginal clotrimazole 0.1 g; those with positive results for trichomonas received intravaginal metronidazole 500 mg for 7 days. After treatment, women with a positive test result in the intervention group had a second vaginal smear between 24 and 28 weeks. Persistent BV was treated with oral clindamycin 300 mg twice daily for 7 days. If Candida or trichomonas were noted, women were treated with the intravaginal clotrimazole or metronidazole. A statistically significant reduction was seen in preterm births in the intervention group(3.0% vs 5.3%, 95% CI, 1.2–3.6; P=.0001; number needed to screen=44).

A large study in 2000 that looked at the use of metronidazole in the treatment of asymptomatic women for BV did not demonstrate any reduction in preterm birth.⁵ In this study, 21,965 asymptomatic women between 8 and 22 weeks gestation were screened for BV with Gram stain using Nugent’s criteria. Then, 1953 women with BV were randomized to receive either 1 g of metronidazole orally for 2 days or placebo. Between 24 and 29 weeks, all of the women were then rescreened for BV by Gram stain. Even if the results were negative, women received another course of the metronidazole or placebo. In this study, preterm delivery rates did not improve for either low- or high-risk women. Specifically, a subgroup analysis of 213 women with previous preterm delivery did not show any benefit to treatment with metronidazole.

In 2003, a Cochrane meta-analysis of 5 studies involving 622 women with previous preterm birth showed a decrease in the risk of low birth weight infants born to women receiving antibiotics vs placebo for the treatment of BV (odds ratio [OR]=0.31; 95% CI, 0.13–0.75).⁶ Treatment also decreased the risk of preterm-prelabor rupture of membranes (OR=0.14; 95% CI, 0.05–0.38) compared with placebo. Unfortunately, these studies did not always specify whether women were asymptomatic for BV infection. In many of the trials, symptomatic women were excluded as they were automatically treated with antibiotics.

In 2003, 2 RCTs evaluating the early treatment of asymptomatic BV in low- and high-risk patients showed a decrease in preterm labor. The first RCT included 494 asymptomatic pregnant women who presented for prenatal care between 12 and 22 weeks gestation. If women had BV detected by Gram stain using Nugent’s criteria, they were randomized to receive either 300 mg oral clindamycin twice daily for 5 days or placebo. In the general population, treatment with clindamycin reduced the rate of late miscarriage and spontaneous preterm delivery by 10.4% (95% CI, 5.0–15.8). In women with a previous preterm delivery or a late miscarriage the proportion of preterm delivery or late miscarriage was reduced (16.6% vs 42%).⁷

The second RCT included 409 asymptomatic women between 13 and 20 weeks gestation with BV by Gram stain using Nugent’s criteria. Investigators randomized women to intravaginal clindamycin each night for 3 days. At a second visit, 20 to 24 days after treatment, women were retested for BV and if they were positive, they received a 7-day course of intravaginal clindamycin or placebo based on the previous randomization. In this study, the incidence of preterm birth was reduced from 10% to 4% (relative risk [RR]=0.38; 95% CI, 0.16–0.90; NNT=17). This study only included 21 women with previous preterm delivery and a subgroup analysis was not performed.⁸

Intravaginal clindamycin has been associated with worse pregnancy outcomes for patients who do not have bacterial vaginosis. A randomized trial of the prophylactic intravaginal clindamycin 2% cream to prevent preterm birth in high-risk women showed an increase in spontaneous preterm delivery in women who actually used all of the medication and did not have BV (NNH=12.3; P<.05).⁹

Recommendations from others

The US Preventive Services Task Force concludes that the evidence is insufficient to recommend for or against routinely screening for BV for high-risk pregnant women. Furthermore, they recommend against screening for average risk women.¹⁰

The Centers for Disease Control and Prevention recommends that high-risk pregnant women (eg, those women who have had a previous preterm delivery) with asymptomatic BV may be evaluated for treatment. The recommended treatment regimens are metronidazole 500 mg orally twice a day for 5 days, metronidazole gel intravaginally for 5 days, or clindamycin cream intravaginally for 7 days.¹¹

The Cochrane Pregnancy and Childbirth Group finds no evidence supporting routine screening and treatment for asymptomatic bacterial vaginosis in pregnancy, except possibly for women with a history of preterm birth.⁶

The American College of Obstetrics and Gynecology summarizes no data supports screening for BV to prevent preterm birth. Their bulletin references a subgroup of women with previous preterm birth who did show benefit from treatment for BV, but the authors speculated that reanalysis with the inclusion of the largest trial to date, which did not show a benefit for this subgroup, might nullify these results.¹²

TABLE
Nugent’s Criteria

Evidence summary

Bacterial vaginosis in early pregnancy is a risk factor for preterm delivery.² The role of BV in preterm labor is not well understood, but it has been consistently associated with preterm labor and delivery. The detection of BV in early pregnancy seems to be a stronger risk factor for preterm delivery than BV in later pregnancy.

Studies evaluating the screening and treatment of BV in women at risk for preterm delivery have demonstrated varying results. Most treatment studies have excluded women who are in the first trimester. A meta-analysis of 7 RCTs reviewed the evidence of screening for BV in pregnancy.³ In this meta-analysis, 5 of the trials specified that women were asymptomatic, and the other 2 did not comment on whether the women were symptomatic or not. In general, there was no benefit to routine screening and treatment of BV.

However, a subgroup of high-risk women seems to benefit from screening and treatment. They defined high-risk women as those have had a preterm delivery, premature rupture of membranes, birth weight <2500 g, or spontaneous abortion. Treating BV in women with a high-risk pregnancy decreased preterm delivery (absolute risk reduction [ARR]=0.22; 90% confidence interval [CI], 0.13–0.31; number needed to treat [NNT]=4.5) regardless of antibiotic choice. However, 2 trials of high-risk women who were empirically treated for BV, but did not have BV, showed an increase in preterm delivery less than 34 weeks (number needed to harm [NNH]=11).

A new study evaluating screening for vaginal infections in pregnancy has demonstrated a reduction in preterm delivery.⁴ In this study, looking at a general obstetrical population in Austria, 4429 asymptomatic pregnant women between 15 and 19.6 weeks gestation had a vaginal screen for bacterial vaginosis, candidiasis and trichomoniasis. The 2048 women in the intervention group were given the results of the screen from their maternity care provider. The 2097 women in the control group and their providers did not receive the results of the vaginal screen. There were 447 women in the intervention group and 441 women in the control group with positive screens. Using the Nugent criteria, women who were diagnosed with BV received a 6-day course of intravaginal clindamycin 2% cream. Those with positive test results for Candida were treated with intravaginal clotrimazole 0.1 g; those with positive results for trichomonas received intravaginal metronidazole 500 mg for 7 days. After treatment, women with a positive test result in the intervention group had a second vaginal smear between 24 and 28 weeks. Persistent BV was treated with oral clindamycin 300 mg twice daily for 7 days. If Candida or trichomonas were noted, women were treated with the intravaginal clotrimazole or metronidazole. A statistically significant reduction was seen in preterm births in the intervention group(3.0% vs 5.3%, 95% CI, 1.2–3.6; P=.0001; number needed to screen=44).

A large study in 2000 that looked at the use of metronidazole in the treatment of asymptomatic women for BV did not demonstrate any reduction in preterm birth.⁵ In this study, 21,965 asymptomatic women between 8 and 22 weeks gestation were screened for BV with Gram stain using Nugent’s criteria. Then, 1953 women with BV were randomized to receive either 1 g of metronidazole orally for 2 days or placebo. Between 24 and 29 weeks, all of the women were then rescreened for BV by Gram stain. Even if the results were negative, women received another course of the metronidazole or placebo. In this study, preterm delivery rates did not improve for either low- or high-risk women. Specifically, a subgroup analysis of 213 women with previous preterm delivery did not show any benefit to treatment with metronidazole.

In 2003, a Cochrane meta-analysis of 5 studies involving 622 women with previous preterm birth showed a decrease in the risk of low birth weight infants born to women receiving antibiotics vs placebo for the treatment of BV (odds ratio [OR]=0.31; 95% CI, 0.13–0.75).⁶ Treatment also decreased the risk of preterm-prelabor rupture of membranes (OR=0.14; 95% CI, 0.05–0.38) compared with placebo. Unfortunately, these studies did not always specify whether women were asymptomatic for BV infection. In many of the trials, symptomatic women were excluded as they were automatically treated with antibiotics.

In 2003, 2 RCTs evaluating the early treatment of asymptomatic BV in low- and high-risk patients showed a decrease in preterm labor. The first RCT included 494 asymptomatic pregnant women who presented for prenatal care between 12 and 22 weeks gestation. If women had BV detected by Gram stain using Nugent’s criteria, they were randomized to receive either 300 mg oral clindamycin twice daily for 5 days or placebo. In the general population, treatment with clindamycin reduced the rate of late miscarriage and spontaneous preterm delivery by 10.4% (95% CI, 5.0–15.8). In women with a previous preterm delivery or a late miscarriage the proportion of preterm delivery or late miscarriage was reduced (16.6% vs 42%).⁷

The second RCT included 409 asymptomatic women between 13 and 20 weeks gestation with BV by Gram stain using Nugent’s criteria. Investigators randomized women to intravaginal clindamycin each night for 3 days. At a second visit, 20 to 24 days after treatment, women were retested for BV and if they were positive, they received a 7-day course of intravaginal clindamycin or placebo based on the previous randomization. In this study, the incidence of preterm birth was reduced from 10% to 4% (relative risk [RR]=0.38; 95% CI, 0.16–0.90; NNT=17). This study only included 21 women with previous preterm delivery and a subgroup analysis was not performed.⁸

Intravaginal clindamycin has been associated with worse pregnancy outcomes for patients who do not have bacterial vaginosis. A randomized trial of the prophylactic intravaginal clindamycin 2% cream to prevent preterm birth in high-risk women showed an increase in spontaneous preterm delivery in women who actually used all of the medication and did not have BV (NNH=12.3; P<.05).⁹

Recommendations from others

The US Preventive Services Task Force concludes that the evidence is insufficient to recommend for or against routinely screening for BV for high-risk pregnant women. Furthermore, they recommend against screening for average risk women.¹⁰

The Centers for Disease Control and Prevention recommends that high-risk pregnant women (eg, those women who have had a previous preterm delivery) with asymptomatic BV may be evaluated for treatment. The recommended treatment regimens are metronidazole 500 mg orally twice a day for 5 days, metronidazole gel intravaginally for 5 days, or clindamycin cream intravaginally for 7 days.¹¹

The Cochrane Pregnancy and Childbirth Group finds no evidence supporting routine screening and treatment for asymptomatic bacterial vaginosis in pregnancy, except possibly for women with a history of preterm birth.⁶

The American College of Obstetrics and Gynecology summarizes no data supports screening for BV to prevent preterm birth. Their bulletin references a subgroup of women with previous preterm birth who did show benefit from treatment for BV, but the authors speculated that reanalysis with the inclusion of the largest trial to date, which did not show a benefit for this subgroup, might nullify these results.¹²

Evidence summary

Chronic daily headache is a heterogeneous primary headache disorder, often defined as a headache duration of more than 4 hours and a headache frequency of more than 15 per month; it affects less than 5% of the US population. Four headache subtypes included in the chronic daily headache definition are chronic (transformed) migraine, chronic tension-type headache, new daily persistent headache, and hemicrania continua. Each subtype may be associated with medication overuse.⁴

Chronic daily headache is challenging to categorize and difficult to manage, and scientific evidence to guide treatment is scant. Despite this, a few studies do offer some hopeful alternatives to those patients who have had conservative measures fail (Table).

A meta-analysis from 2001 reviewed 38 RCTs of antidepressants as prophylaxis for chronic headache. Nineteen studies investigated TCAs, 18 examined serotonin blockers, and 7 focused on SSRIs. Patients taking antidepressants were twice as likely to report headache improvement (rate ratio [RR]=2.0; 95% confidence interval [CI], 1.6–2.4), with the average amount of improvement considered to be large (standard mean difference=0.94; 95% CI, 0.65–1.2). Serotonin blockers, most of which are not available or commonly used in the US, and TCAs were all effective in decreasing the headache burden, while the results for SSRIs were less clear. Dosages of amitriptyline ranged from 10 to 150 mg daily; most of the studies used 60 to 100 mg daily.¹

Medication withdrawal therapy is a treatment strategy for chronic daily headaches associated with the paradoxical induction of headaches by the frequent, long-term use of immediate relief medications such as aspirin, NSAIDs, acetaminophen, caffeine, codeine, ergotamine, and sumatriptan. A retrospective study tracked 101 men and women who underwent a controlled outpatient withdrawal of their overused medications. Headache diaries kept for 1 to 3 months reflected that 56% of the patients had at least a 50% reduction in headache days after removal of overused drugs. Twenty-two patients who had no success with withdrawal and continued to have headaches were treated with amitriptyline. Subsequently, 10 of these patients experienced a 50% reduction in headache frequency.⁵

A systematic review of the therapeutic approaches to medication-induced headache looked at 18 studies from 1966 to 1998. Although most were uncontrolled small trials, medically monitored withdrawal of all symptomatic headache medications is recommended by the authors. No long-term outcome comparisons between withdrawal strategies are available.²

Other therapies for treating chronic daily headache include the skeletal muscle relaxant tizanidine (Zanaflex), which was studied in an industry-sponsored, double-blind, placebo-controlled trial of 92 patients. The medication was used as prophylaxis, titrating up to a dose of 8 mg 3 times daily. The overall headache index (a measure of headache intensity, frequency, and duration) significantly decreased. The headache index decreased in the tizanidine group from 2.6 to 1.2, and in the placebo group from 2.6 to 2.1 (P =.0025). Decreases in headache frequency and headache intensity were less dramatic but still significant. This trial lasted only 12 weeks, so longer-term outcomes are not available.⁶

Stress management, acupuncture, botulinum toxin, behavioral therapy including relaxation therapy, biofeedback, and even Internet-based self-help have all been studied, but most of these therapies do not have significant evidence-based support.

TABLE
Treatment options for chronic daily headache

Recommendations from others

Our literature search and review of major textbooks found no formally organized guidelines or recommendations on the treatment of chronic daily headache.

Evidence summary

Chronic daily headache is a heterogeneous primary headache disorder, often defined as a headache duration of more than 4 hours and a headache frequency of more than 15 per month; it affects less than 5% of the US population. Four headache subtypes included in the chronic daily headache definition are chronic (transformed) migraine, chronic tension-type headache, new daily persistent headache, and hemicrania continua. Each subtype may be associated with medication overuse.⁴

Chronic daily headache is challenging to categorize and difficult to manage, and scientific evidence to guide treatment is scant. Despite this, a few studies do offer some hopeful alternatives to those patients who have had conservative measures fail (Table).

A meta-analysis from 2001 reviewed 38 RCTs of antidepressants as prophylaxis for chronic headache. Nineteen studies investigated TCAs, 18 examined serotonin blockers, and 7 focused on SSRIs. Patients taking antidepressants were twice as likely to report headache improvement (rate ratio [RR]=2.0; 95% confidence interval [CI], 1.6–2.4), with the average amount of improvement considered to be large (standard mean difference=0.94; 95% CI, 0.65–1.2). Serotonin blockers, most of which are not available or commonly used in the US, and TCAs were all effective in decreasing the headache burden, while the results for SSRIs were less clear. Dosages of amitriptyline ranged from 10 to 150 mg daily; most of the studies used 60 to 100 mg daily.¹

Medication withdrawal therapy is a treatment strategy for chronic daily headaches associated with the paradoxical induction of headaches by the frequent, long-term use of immediate relief medications such as aspirin, NSAIDs, acetaminophen, caffeine, codeine, ergotamine, and sumatriptan. A retrospective study tracked 101 men and women who underwent a controlled outpatient withdrawal of their overused medications. Headache diaries kept for 1 to 3 months reflected that 56% of the patients had at least a 50% reduction in headache days after removal of overused drugs. Twenty-two patients who had no success with withdrawal and continued to have headaches were treated with amitriptyline. Subsequently, 10 of these patients experienced a 50% reduction in headache frequency.⁵

A systematic review of the therapeutic approaches to medication-induced headache looked at 18 studies from 1966 to 1998. Although most were uncontrolled small trials, medically monitored withdrawal of all symptomatic headache medications is recommended by the authors. No long-term outcome comparisons between withdrawal strategies are available.²

Other therapies for treating chronic daily headache include the skeletal muscle relaxant tizanidine (Zanaflex), which was studied in an industry-sponsored, double-blind, placebo-controlled trial of 92 patients. The medication was used as prophylaxis, titrating up to a dose of 8 mg 3 times daily. The overall headache index (a measure of headache intensity, frequency, and duration) significantly decreased. The headache index decreased in the tizanidine group from 2.6 to 1.2, and in the placebo group from 2.6 to 2.1 (P =.0025). Decreases in headache frequency and headache intensity were less dramatic but still significant. This trial lasted only 12 weeks, so longer-term outcomes are not available.⁶

Stress management, acupuncture, botulinum toxin, behavioral therapy including relaxation therapy, biofeedback, and even Internet-based self-help have all been studied, but most of these therapies do not have significant evidence-based support.

TABLE
Treatment options for chronic daily headache

Recommendations from others

Our literature search and review of major textbooks found no formally organized guidelines or recommendations on the treatment of chronic daily headache.

Evidence summary

Chronic daily headache is a heterogeneous primary headache disorder, often defined as a headache duration of more than 4 hours and a headache frequency of more than 15 per month; it affects less than 5% of the US population. Four headache subtypes included in the chronic daily headache definition are chronic (transformed) migraine, chronic tension-type headache, new daily persistent headache, and hemicrania continua. Each subtype may be associated with medication overuse.⁴

Chronic daily headache is challenging to categorize and difficult to manage, and scientific evidence to guide treatment is scant. Despite this, a few studies do offer some hopeful alternatives to those patients who have had conservative measures fail (Table).

A meta-analysis from 2001 reviewed 38 RCTs of antidepressants as prophylaxis for chronic headache. Nineteen studies investigated TCAs, 18 examined serotonin blockers, and 7 focused on SSRIs. Patients taking antidepressants were twice as likely to report headache improvement (rate ratio [RR]=2.0; 95% confidence interval [CI], 1.6–2.4), with the average amount of improvement considered to be large (standard mean difference=0.94; 95% CI, 0.65–1.2). Serotonin blockers, most of which are not available or commonly used in the US, and TCAs were all effective in decreasing the headache burden, while the results for SSRIs were less clear. Dosages of amitriptyline ranged from 10 to 150 mg daily; most of the studies used 60 to 100 mg daily.¹

Medication withdrawal therapy is a treatment strategy for chronic daily headaches associated with the paradoxical induction of headaches by the frequent, long-term use of immediate relief medications such as aspirin, NSAIDs, acetaminophen, caffeine, codeine, ergotamine, and sumatriptan. A retrospective study tracked 101 men and women who underwent a controlled outpatient withdrawal of their overused medications. Headache diaries kept for 1 to 3 months reflected that 56% of the patients had at least a 50% reduction in headache days after removal of overused drugs. Twenty-two patients who had no success with withdrawal and continued to have headaches were treated with amitriptyline. Subsequently, 10 of these patients experienced a 50% reduction in headache frequency.⁵

A systematic review of the therapeutic approaches to medication-induced headache looked at 18 studies from 1966 to 1998. Although most were uncontrolled small trials, medically monitored withdrawal of all symptomatic headache medications is recommended by the authors. No long-term outcome comparisons between withdrawal strategies are available.²

Other therapies for treating chronic daily headache include the skeletal muscle relaxant tizanidine (Zanaflex), which was studied in an industry-sponsored, double-blind, placebo-controlled trial of 92 patients. The medication was used as prophylaxis, titrating up to a dose of 8 mg 3 times daily. The overall headache index (a measure of headache intensity, frequency, and duration) significantly decreased. The headache index decreased in the tizanidine group from 2.6 to 1.2, and in the placebo group from 2.6 to 2.1 (P =.0025). Decreases in headache frequency and headache intensity were less dramatic but still significant. This trial lasted only 12 weeks, so longer-term outcomes are not available.⁶

Stress management, acupuncture, botulinum toxin, behavioral therapy including relaxation therapy, biofeedback, and even Internet-based self-help have all been studied, but most of these therapies do not have significant evidence-based support.

TABLE
Treatment options for chronic daily headache

Recommendations from others

Our literature search and review of major textbooks found no formally organized guidelines or recommendations on the treatment of chronic daily headache.

Evidence summary

Constipation in toddlers is characterized by a delay or difficulty in defecation that is present for 2 or more weeks. It is a common problem, accounting for 3% of all general pediatric visits. In most children, constipation is functional without pathological cause. Treatment for pathologic constipation is not addressed here.

Laxative therapies were compared in 2 studies. An unblinded, randomized, crossover trial enrolled 37 children referred for subspecialty evaluation of functional constipation. It found that PEG 3350 and lactulose were equivalent in improving stool frequency, stool form, and ease of passage.

However, there was a significant difference in total stool transit time in subjects taking PEG compared with those taking lactulose (47.6 hours vs 55.3 hours, respectively; P=.038). In addition, twice as many parents and guardians rated PEG effective as those rating lactulose effective (84% vs 46%); and 73% of parents preferred PEG to lactulose.¹

A randomized trial compared PEG with milk of magnesia in 49 children with functional constipation and encopresis. Follow-up at 1, 3, 6, and 12 months revealed similar effectiveness in increasing bowel movement frequency, decreasing soiling episodes, and decreasing abdominal pain. It also revealed that PEG was more palatable and better-tolerated than milk of magnesia (33% of children refused to take milk of magnesia, whereas none refused PEG). No side effects from PEG were reported.²

Behavioral modification has been studied for constipation-related encopresis. A randomized controlled trial of 87 children with fecal soiling compared the effect of enhanced toilet training (including behavioral therapy) with aggressive medical management that included disimpaction, enemas, and regular laxative therapy. After 12 months, the enhanced toilet training with behavioral therapy was more effective in reducing daily frequency of soiling (78% of the children had significantly decreased average daily frequency of soiling compared with 41% in the aggressive medical management group; P<.0001; absolute risk reduction=0.37; number needed to treat for 1 year=2.7).³

Dietary management centers on a balanced diet with whole grains, fruits, and vegetables. A case-control study evaluated 291 subjects with constipation and compared their diet with 1602 controls. Mean daily fiber intake was lower in the constipation group. Compared with fiber intake of more than 29 g/d, the relative risk was 8.0 for fiber intake of less than 12.4 g/d.⁴

A nonrandomized, controlled trial of acupuncture treatment enrolled 17 children with history of constipation for a minimum of 6 months. Bowel movement frequency improved in both males (1.5 ± 0.1/week to 4.4 ± 0.6/week; P<.01) and females (1.4 ± 0.3/week to 5.6 ± 1.1/week; P<.01) after 10 acupuncture sessions. No other bowel movement parameter was reported.⁵

Recommendations from others

The North American Society for Pediatric Gastroenterology and Nutrition (NASPGN) recommends parental education, initial disimpaction as needed, and maintenance therapy with a balanced diet, behavioral modification, and laxatives for all children aged >1 year. Recommended laxatives include mineral oil, magnesium hydroxide, lactulose, and sorbitol. Behavior modification includes regular toileting, unhurried time on the toilet after meals, diary of stool frequency, and a reward system.⁶ The American Academy of Pediatrics supports the above guidelines.

A multispecialty panel from the University of Michigan used a structured literature review as a basis for a consensus guideline. The resulting protocol was similar to the NASPGN protocol, with the addition of stool softeners as an alternative to laxatives.⁷

Evidence summary

Constipation in toddlers is characterized by a delay or difficulty in defecation that is present for 2 or more weeks. It is a common problem, accounting for 3% of all general pediatric visits. In most children, constipation is functional without pathological cause. Treatment for pathologic constipation is not addressed here.

Laxative therapies were compared in 2 studies. An unblinded, randomized, crossover trial enrolled 37 children referred for subspecialty evaluation of functional constipation. It found that PEG 3350 and lactulose were equivalent in improving stool frequency, stool form, and ease of passage.

However, there was a significant difference in total stool transit time in subjects taking PEG compared with those taking lactulose (47.6 hours vs 55.3 hours, respectively; P=.038). In addition, twice as many parents and guardians rated PEG effective as those rating lactulose effective (84% vs 46%); and 73% of parents preferred PEG to lactulose.¹

A randomized trial compared PEG with milk of magnesia in 49 children with functional constipation and encopresis. Follow-up at 1, 3, 6, and 12 months revealed similar effectiveness in increasing bowel movement frequency, decreasing soiling episodes, and decreasing abdominal pain. It also revealed that PEG was more palatable and better-tolerated than milk of magnesia (33% of children refused to take milk of magnesia, whereas none refused PEG). No side effects from PEG were reported.²

Behavioral modification has been studied for constipation-related encopresis. A randomized controlled trial of 87 children with fecal soiling compared the effect of enhanced toilet training (including behavioral therapy) with aggressive medical management that included disimpaction, enemas, and regular laxative therapy. After 12 months, the enhanced toilet training with behavioral therapy was more effective in reducing daily frequency of soiling (78% of the children had significantly decreased average daily frequency of soiling compared with 41% in the aggressive medical management group; P<.0001; absolute risk reduction=0.37; number needed to treat for 1 year=2.7).³

Dietary management centers on a balanced diet with whole grains, fruits, and vegetables. A case-control study evaluated 291 subjects with constipation and compared their diet with 1602 controls. Mean daily fiber intake was lower in the constipation group. Compared with fiber intake of more than 29 g/d, the relative risk was 8.0 for fiber intake of less than 12.4 g/d.⁴

A nonrandomized, controlled trial of acupuncture treatment enrolled 17 children with history of constipation for a minimum of 6 months. Bowel movement frequency improved in both males (1.5 ± 0.1/week to 4.4 ± 0.6/week; P<.01) and females (1.4 ± 0.3/week to 5.6 ± 1.1/week; P<.01) after 10 acupuncture sessions. No other bowel movement parameter was reported.⁵

Recommendations from others

The North American Society for Pediatric Gastroenterology and Nutrition (NASPGN) recommends parental education, initial disimpaction as needed, and maintenance therapy with a balanced diet, behavioral modification, and laxatives for all children aged >1 year. Recommended laxatives include mineral oil, magnesium hydroxide, lactulose, and sorbitol. Behavior modification includes regular toileting, unhurried time on the toilet after meals, diary of stool frequency, and a reward system.⁶ The American Academy of Pediatrics supports the above guidelines.

A multispecialty panel from the University of Michigan used a structured literature review as a basis for a consensus guideline. The resulting protocol was similar to the NASPGN protocol, with the addition of stool softeners as an alternative to laxatives.⁷

Evidence summary

Constipation in toddlers is characterized by a delay or difficulty in defecation that is present for 2 or more weeks. It is a common problem, accounting for 3% of all general pediatric visits. In most children, constipation is functional without pathological cause. Treatment for pathologic constipation is not addressed here.

Laxative therapies were compared in 2 studies. An unblinded, randomized, crossover trial enrolled 37 children referred for subspecialty evaluation of functional constipation. It found that PEG 3350 and lactulose were equivalent in improving stool frequency, stool form, and ease of passage.

However, there was a significant difference in total stool transit time in subjects taking PEG compared with those taking lactulose (47.6 hours vs 55.3 hours, respectively; P=.038). In addition, twice as many parents and guardians rated PEG effective as those rating lactulose effective (84% vs 46%); and 73% of parents preferred PEG to lactulose.¹

A randomized trial compared PEG with milk of magnesia in 49 children with functional constipation and encopresis. Follow-up at 1, 3, 6, and 12 months revealed similar effectiveness in increasing bowel movement frequency, decreasing soiling episodes, and decreasing abdominal pain. It also revealed that PEG was more palatable and better-tolerated than milk of magnesia (33% of children refused to take milk of magnesia, whereas none refused PEG). No side effects from PEG were reported.²

Behavioral modification has been studied for constipation-related encopresis. A randomized controlled trial of 87 children with fecal soiling compared the effect of enhanced toilet training (including behavioral therapy) with aggressive medical management that included disimpaction, enemas, and regular laxative therapy. After 12 months, the enhanced toilet training with behavioral therapy was more effective in reducing daily frequency of soiling (78% of the children had significantly decreased average daily frequency of soiling compared with 41% in the aggressive medical management group; P<.0001; absolute risk reduction=0.37; number needed to treat for 1 year=2.7).³

Dietary management centers on a balanced diet with whole grains, fruits, and vegetables. A case-control study evaluated 291 subjects with constipation and compared their diet with 1602 controls. Mean daily fiber intake was lower in the constipation group. Compared with fiber intake of more than 29 g/d, the relative risk was 8.0 for fiber intake of less than 12.4 g/d.⁴

A nonrandomized, controlled trial of acupuncture treatment enrolled 17 children with history of constipation for a minimum of 6 months. Bowel movement frequency improved in both males (1.5 ± 0.1/week to 4.4 ± 0.6/week; P<.01) and females (1.4 ± 0.3/week to 5.6 ± 1.1/week; P<.01) after 10 acupuncture sessions. No other bowel movement parameter was reported.⁵

Recommendations from others

The North American Society for Pediatric Gastroenterology and Nutrition (NASPGN) recommends parental education, initial disimpaction as needed, and maintenance therapy with a balanced diet, behavioral modification, and laxatives for all children aged >1 year. Recommended laxatives include mineral oil, magnesium hydroxide, lactulose, and sorbitol. Behavior modification includes regular toileting, unhurried time on the toilet after meals, diary of stool frequency, and a reward system.⁶ The American Academy of Pediatrics supports the above guidelines.

A multispecialty panel from the University of Michigan used a structured literature review as a basis for a consensus guideline. The resulting protocol was similar to the NASPGN protocol, with the addition of stool softeners as an alternative to laxatives.⁷

Evidence summary

Based on the advantage of nocturnal sedation of first-generation antihistamines, clinicians frequently use these agents to treat pruritus. Evidence is lacking to support this intuitive approach. Furthermore, not all pruritus can be lumped into a single category, as distinct treatment recommendations exist for different categories.

The best evidence supporting antihistamines is for the treatment of urticarial pruritus. A Medline-based review found 7 double-blind, placebo-controlled trials that compared the benefit of first- and second-generation antihistamines in 720 patients with chronic idiopathic urticaria. Hydroxyzine was used in 682 patients, while the remainder took clemastine. Second-generation agents included cetirizine, loratadine, or acrivastine. The researchers qualitatively summarized outcomes and concluded that the treatment benefits were equivalent and superior to placebo.¹ The clinical practice of doubling the dose of second-generation agents for initial treatment failures was not recommended, due to absence of supporting data for this approach.

A recent review of therapies for urticarial itch concluded that second-generation antihista mines were preferred.² However, the methodology failed to use a systematic search technique. The conclusion was based upon a single double-blind placebo-controlled study of 188 patients at least 12 years of age. They received cetirizine 10 mg daily, hydroxyzine 25 mg 3 times daily, or placebo. This study found both agents produced significant, and equivalent, pruritus reduction relative to placebo.³

In contrast to urticarial itch, pruritus from atopic dermatitis does not improve with antihistamines. An NHS Centre narrative review on relieving pruritus in atopic dermatitis concluded that there was little objective evidence to support the efficacy of first- or second-generation antihistamines; 803 participants from 16 case series and reports were included. There were no large RCTs. Results were not pooled or tested for heterogeneity, so they should be interpreted cautiously.^4,5

Another systematic review focusing on pediatric patients concluded oral antihistamines are not beneficial for pruritus from atopic dermatitis. A search of Cochrane and PubMed revealed only 2 relevant RCTs involving 177 children. Cetirizine and chlorpheniramine were each compared with placebo, and no statistically significant reduction in symptoms was found.⁶

Vigilance must be exercised when interpreting pruritus literature. Many studies are pharmacodynamic only, omit appropriate statistical information, and measure surrogate outcomes in healthy volunteers, such as wheal and flare suppression to injected histamine. Such disease-oriented evidence has filtered into clinical recommendations.

One recent nonsystematic, narrative review of pruritic dermatoses concluded second-generation antihistamines appear to be more effective.⁷ This conclusion was largely based on a study of 14 young, healthy, “light-skinned” Canadian men. No placebo control was used. Seven received fexofenadine 120 mg; the other 7 took diphenhydramine 50 mg. Primary outcomes were concentrations of drug in skin punch biopsies and plasma samples, plus degree of wheal and flare suppression to histamine. In this study, fexofenadine showed statistically significant disease-oriented results.⁸

Recommendations from others

No evidence-based guidelines or consensus statements were found that address antihistamine preference in the treatment of pruritus. Although use of non-antihistamine agents is beyond the scope of this inquiry, an excellent topical review of pruritus was recently published that comprehensively outlined the Twycross classification system and detailed the evidence for usual and nontraditional treatments.⁹ Since antihistamines do not benefit atopic-related pruritus, other options include emollients, counterirritants such as menthol/camphor or capsaicin, EMLA cream, topical pramoxine, topical corticosteroids, topical doxepin, topical immunomodulators such as pimecrolimus or tacrolimus, topical aspirin, and phototherapy with psoralen ultraviolet A-range (PUVA).

Acknowledgments

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

Evidence summary

Based on the advantage of nocturnal sedation of first-generation antihistamines, clinicians frequently use these agents to treat pruritus. Evidence is lacking to support this intuitive approach. Furthermore, not all pruritus can be lumped into a single category, as distinct treatment recommendations exist for different categories.

The best evidence supporting antihistamines is for the treatment of urticarial pruritus. A Medline-based review found 7 double-blind, placebo-controlled trials that compared the benefit of first- and second-generation antihistamines in 720 patients with chronic idiopathic urticaria. Hydroxyzine was used in 682 patients, while the remainder took clemastine. Second-generation agents included cetirizine, loratadine, or acrivastine. The researchers qualitatively summarized outcomes and concluded that the treatment benefits were equivalent and superior to placebo.¹ The clinical practice of doubling the dose of second-generation agents for initial treatment failures was not recommended, due to absence of supporting data for this approach.

A recent review of therapies for urticarial itch concluded that second-generation antihista mines were preferred.² However, the methodology failed to use a systematic search technique. The conclusion was based upon a single double-blind placebo-controlled study of 188 patients at least 12 years of age. They received cetirizine 10 mg daily, hydroxyzine 25 mg 3 times daily, or placebo. This study found both agents produced significant, and equivalent, pruritus reduction relative to placebo.³

In contrast to urticarial itch, pruritus from atopic dermatitis does not improve with antihistamines. An NHS Centre narrative review on relieving pruritus in atopic dermatitis concluded that there was little objective evidence to support the efficacy of first- or second-generation antihistamines; 803 participants from 16 case series and reports were included. There were no large RCTs. Results were not pooled or tested for heterogeneity, so they should be interpreted cautiously.^4,5

Another systematic review focusing on pediatric patients concluded oral antihistamines are not beneficial for pruritus from atopic dermatitis. A search of Cochrane and PubMed revealed only 2 relevant RCTs involving 177 children. Cetirizine and chlorpheniramine were each compared with placebo, and no statistically significant reduction in symptoms was found.⁶

Vigilance must be exercised when interpreting pruritus literature. Many studies are pharmacodynamic only, omit appropriate statistical information, and measure surrogate outcomes in healthy volunteers, such as wheal and flare suppression to injected histamine. Such disease-oriented evidence has filtered into clinical recommendations.

One recent nonsystematic, narrative review of pruritic dermatoses concluded second-generation antihistamines appear to be more effective.⁷ This conclusion was largely based on a study of 14 young, healthy, “light-skinned” Canadian men. No placebo control was used. Seven received fexofenadine 120 mg; the other 7 took diphenhydramine 50 mg. Primary outcomes were concentrations of drug in skin punch biopsies and plasma samples, plus degree of wheal and flare suppression to histamine. In this study, fexofenadine showed statistically significant disease-oriented results.⁸

Recommendations from others

No evidence-based guidelines or consensus statements were found that address antihistamine preference in the treatment of pruritus. Although use of non-antihistamine agents is beyond the scope of this inquiry, an excellent topical review of pruritus was recently published that comprehensively outlined the Twycross classification system and detailed the evidence for usual and nontraditional treatments.⁹ Since antihistamines do not benefit atopic-related pruritus, other options include emollients, counterirritants such as menthol/camphor or capsaicin, EMLA cream, topical pramoxine, topical corticosteroids, topical doxepin, topical immunomodulators such as pimecrolimus or tacrolimus, topical aspirin, and phototherapy with psoralen ultraviolet A-range (PUVA).

Acknowledgments

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

Evidence summary

Based on the advantage of nocturnal sedation of first-generation antihistamines, clinicians frequently use these agents to treat pruritus. Evidence is lacking to support this intuitive approach. Furthermore, not all pruritus can be lumped into a single category, as distinct treatment recommendations exist for different categories.

The best evidence supporting antihistamines is for the treatment of urticarial pruritus. A Medline-based review found 7 double-blind, placebo-controlled trials that compared the benefit of first- and second-generation antihistamines in 720 patients with chronic idiopathic urticaria. Hydroxyzine was used in 682 patients, while the remainder took clemastine. Second-generation agents included cetirizine, loratadine, or acrivastine. The researchers qualitatively summarized outcomes and concluded that the treatment benefits were equivalent and superior to placebo.¹ The clinical practice of doubling the dose of second-generation agents for initial treatment failures was not recommended, due to absence of supporting data for this approach.

A recent review of therapies for urticarial itch concluded that second-generation antihista mines were preferred.² However, the methodology failed to use a systematic search technique. The conclusion was based upon a single double-blind placebo-controlled study of 188 patients at least 12 years of age. They received cetirizine 10 mg daily, hydroxyzine 25 mg 3 times daily, or placebo. This study found both agents produced significant, and equivalent, pruritus reduction relative to placebo.³

In contrast to urticarial itch, pruritus from atopic dermatitis does not improve with antihistamines. An NHS Centre narrative review on relieving pruritus in atopic dermatitis concluded that there was little objective evidence to support the efficacy of first- or second-generation antihistamines; 803 participants from 16 case series and reports were included. There were no large RCTs. Results were not pooled or tested for heterogeneity, so they should be interpreted cautiously.^4,5

Another systematic review focusing on pediatric patients concluded oral antihistamines are not beneficial for pruritus from atopic dermatitis. A search of Cochrane and PubMed revealed only 2 relevant RCTs involving 177 children. Cetirizine and chlorpheniramine were each compared with placebo, and no statistically significant reduction in symptoms was found.⁶

Vigilance must be exercised when interpreting pruritus literature. Many studies are pharmacodynamic only, omit appropriate statistical information, and measure surrogate outcomes in healthy volunteers, such as wheal and flare suppression to injected histamine. Such disease-oriented evidence has filtered into clinical recommendations.

One recent nonsystematic, narrative review of pruritic dermatoses concluded second-generation antihistamines appear to be more effective.⁷ This conclusion was largely based on a study of 14 young, healthy, “light-skinned” Canadian men. No placebo control was used. Seven received fexofenadine 120 mg; the other 7 took diphenhydramine 50 mg. Primary outcomes were concentrations of drug in skin punch biopsies and plasma samples, plus degree of wheal and flare suppression to histamine. In this study, fexofenadine showed statistically significant disease-oriented results.⁸

Recommendations from others

No evidence-based guidelines or consensus statements were found that address antihistamine preference in the treatment of pruritus. Although use of non-antihistamine agents is beyond the scope of this inquiry, an excellent topical review of pruritus was recently published that comprehensively outlined the Twycross classification system and detailed the evidence for usual and nontraditional treatments.⁹ Since antihistamines do not benefit atopic-related pruritus, other options include emollients, counterirritants such as menthol/camphor or capsaicin, EMLA cream, topical pramoxine, topical corticosteroids, topical doxepin, topical immunomodulators such as pimecrolimus or tacrolimus, topical aspirin, and phototherapy with psoralen ultraviolet A-range (PUVA).

Acknowledgments

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

Evidence summary

The long-term safety of PPIs is not completely known. There are 5 PPIs on the US market. Clinical experience with these medications ranges from 3 to 20 years. All of the identified studies addressing long-term use have follow-up of 10 years or less (Table). Studies of longer duration are warranted. We reviewed the possible adverse effects of these medications.

Gastric carcinoid. PPIs cause predictable and sustained hypergastrinemia in response to acid suppression. In rats, this causes enterochro-maffin-like cell (ECL) hyperplasia and carcinoid tumors, raising a safety concern in humans. In a nonsystematic review of 11 studies of 1800 patients who used PPIs from 6 months to 8 years, there were no neoplastic ECL changes or carcinoid tumors.¹ Three other nonsystematic reviews support these findings.^2-4 In a randomized controlled trial comparing efficacy and safety of rabeprazole with omeprazole for gastro-esophageal disease, 123 (51%) out of 243 patients completed 5 years of the study; no patients had neoplastic ECL changes.⁵

Atrophic gastritis and gastric cancer. Atrophic gastritis with intestinal metaplasia is associated with gastric adenocarcinoma. Because PPIs can theoretically cause atrophic gastritis, there is a concern that this could lead to gastric cancer. The evidence regarding atrophic gastritis is contradictory. A nonsystematic review identified 1 cohort study and 1 randomized controlled trial of patients taking omeprazole from 1 to 4 years, which showed no association between PPI use and atrophic gastritis.¹ The same review reported that another cohort study of patients using omeprazole for 1 year showed an increase in atrophic gastritis. None of the studies reviewed showed an association between omeprazole use and intestinal metaplasia or its progression to gastric adenocarcinoma.¹ Three other nonsystematic reviews support these findings.^2,3,5 The available evidence indicates that PPI use is not clearly associated with atrophic gastritis, or with progression from gastritis to metaplasia or cancer.

Enteric infections. Because hypochlorhydria is associated with bacterial enteric infections, bacterial enteritis is a theoretical risk of long-term PPI use. A large case-control study of 54,461 patients using omeprazole for 1 year showed no association with such infections.⁶

Mineral malabsorption. Dietary calcium, phosphorus, magnesium, zinc, and iron depend on gastric acid for absorption. Two separate non-systematic reviews showed no problems with malabsorption of these micronutrients.^1,3

B12 malabsorption. Two nonsystematic reviews showed a decrease in vitamin B12 absorption among patients on high-dose (up to 80 mg of omeprazole daily), long-term PPI therapy (eg, patients with Zollinger-Ellison syndrome).^1,2 This has not been demonstrated for patients taking more typical doses of omeprazole. The clinical significance of this is unknown; however, the authors of these reviews suggested monitoring B12 levels of patients on long-term, high-dose PPI therapy.

TABLE
Potential proton pump inhibitor safety concerns

Recommendations from others

A Federal Drug Commission report indicates that labeling PPIs for cancer risk is not warranted.⁷ The American College of Gastroenterology and the University of Michigan Health System guidelines for treatment of gastroesophageal disease recommend long-term PPI therapy as an option without any warning against their use.^8,9

Evidence summary

The long-term safety of PPIs is not completely known. There are 5 PPIs on the US market. Clinical experience with these medications ranges from 3 to 20 years. All of the identified studies addressing long-term use have follow-up of 10 years or less (Table). Studies of longer duration are warranted. We reviewed the possible adverse effects of these medications.

Gastric carcinoid. PPIs cause predictable and sustained hypergastrinemia in response to acid suppression. In rats, this causes enterochro-maffin-like cell (ECL) hyperplasia and carcinoid tumors, raising a safety concern in humans. In a nonsystematic review of 11 studies of 1800 patients who used PPIs from 6 months to 8 years, there were no neoplastic ECL changes or carcinoid tumors.¹ Three other nonsystematic reviews support these findings.^2-4 In a randomized controlled trial comparing efficacy and safety of rabeprazole with omeprazole for gastro-esophageal disease, 123 (51%) out of 243 patients completed 5 years of the study; no patients had neoplastic ECL changes.⁵

Atrophic gastritis and gastric cancer. Atrophic gastritis with intestinal metaplasia is associated with gastric adenocarcinoma. Because PPIs can theoretically cause atrophic gastritis, there is a concern that this could lead to gastric cancer. The evidence regarding atrophic gastritis is contradictory. A nonsystematic review identified 1 cohort study and 1 randomized controlled trial of patients taking omeprazole from 1 to 4 years, which showed no association between PPI use and atrophic gastritis.¹ The same review reported that another cohort study of patients using omeprazole for 1 year showed an increase in atrophic gastritis. None of the studies reviewed showed an association between omeprazole use and intestinal metaplasia or its progression to gastric adenocarcinoma.¹ Three other nonsystematic reviews support these findings.^2,3,5 The available evidence indicates that PPI use is not clearly associated with atrophic gastritis, or with progression from gastritis to metaplasia or cancer.

Enteric infections. Because hypochlorhydria is associated with bacterial enteric infections, bacterial enteritis is a theoretical risk of long-term PPI use. A large case-control study of 54,461 patients using omeprazole for 1 year showed no association with such infections.⁶

Mineral malabsorption. Dietary calcium, phosphorus, magnesium, zinc, and iron depend on gastric acid for absorption. Two separate non-systematic reviews showed no problems with malabsorption of these micronutrients.^1,3

B12 malabsorption. Two nonsystematic reviews showed a decrease in vitamin B12 absorption among patients on high-dose (up to 80 mg of omeprazole daily), long-term PPI therapy (eg, patients with Zollinger-Ellison syndrome).^1,2 This has not been demonstrated for patients taking more typical doses of omeprazole. The clinical significance of this is unknown; however, the authors of these reviews suggested monitoring B12 levels of patients on long-term, high-dose PPI therapy.

TABLE
Potential proton pump inhibitor safety concerns

Recommendations from others

A Federal Drug Commission report indicates that labeling PPIs for cancer risk is not warranted.⁷ The American College of Gastroenterology and the University of Michigan Health System guidelines for treatment of gastroesophageal disease recommend long-term PPI therapy as an option without any warning against their use.^8,9

Evidence summary

The long-term safety of PPIs is not completely known. There are 5 PPIs on the US market. Clinical experience with these medications ranges from 3 to 20 years. All of the identified studies addressing long-term use have follow-up of 10 years or less (Table). Studies of longer duration are warranted. We reviewed the possible adverse effects of these medications.

Gastric carcinoid. PPIs cause predictable and sustained hypergastrinemia in response to acid suppression. In rats, this causes enterochro-maffin-like cell (ECL) hyperplasia and carcinoid tumors, raising a safety concern in humans. In a nonsystematic review of 11 studies of 1800 patients who used PPIs from 6 months to 8 years, there were no neoplastic ECL changes or carcinoid tumors.¹ Three other nonsystematic reviews support these findings.^2-4 In a randomized controlled trial comparing efficacy and safety of rabeprazole with omeprazole for gastro-esophageal disease, 123 (51%) out of 243 patients completed 5 years of the study; no patients had neoplastic ECL changes.⁵

Atrophic gastritis and gastric cancer. Atrophic gastritis with intestinal metaplasia is associated with gastric adenocarcinoma. Because PPIs can theoretically cause atrophic gastritis, there is a concern that this could lead to gastric cancer. The evidence regarding atrophic gastritis is contradictory. A nonsystematic review identified 1 cohort study and 1 randomized controlled trial of patients taking omeprazole from 1 to 4 years, which showed no association between PPI use and atrophic gastritis.¹ The same review reported that another cohort study of patients using omeprazole for 1 year showed an increase in atrophic gastritis. None of the studies reviewed showed an association between omeprazole use and intestinal metaplasia or its progression to gastric adenocarcinoma.¹ Three other nonsystematic reviews support these findings.^2,3,5 The available evidence indicates that PPI use is not clearly associated with atrophic gastritis, or with progression from gastritis to metaplasia or cancer.

Enteric infections. Because hypochlorhydria is associated with bacterial enteric infections, bacterial enteritis is a theoretical risk of long-term PPI use. A large case-control study of 54,461 patients using omeprazole for 1 year showed no association with such infections.⁶

Mineral malabsorption. Dietary calcium, phosphorus, magnesium, zinc, and iron depend on gastric acid for absorption. Two separate non-systematic reviews showed no problems with malabsorption of these micronutrients.^1,3

B12 malabsorption. Two nonsystematic reviews showed a decrease in vitamin B12 absorption among patients on high-dose (up to 80 mg of omeprazole daily), long-term PPI therapy (eg, patients with Zollinger-Ellison syndrome).^1,2 This has not been demonstrated for patients taking more typical doses of omeprazole. The clinical significance of this is unknown; however, the authors of these reviews suggested monitoring B12 levels of patients on long-term, high-dose PPI therapy.

TABLE
Potential proton pump inhibitor safety concerns

Recommendations from others

A Federal Drug Commission report indicates that labeling PPIs for cancer risk is not warranted.⁷ The American College of Gastroenterology and the University of Michigan Health System guidelines for treatment of gastroesophageal disease recommend long-term PPI therapy as an option without any warning against their use.^8,9

Evidence summary

In the US, GABHS is the cause of acute pharyn-gitis in 5% to 10% of adults and 15% to 30% of children. It is the only commonly occurring cause of pharyngitis with an indication for antibiotic therapy.¹ The main benefit of antibiotic treatment in adults is earlier symptom relief—1 fewer day of fever and pain if antibiotics are begun within 3 days of onset.

Antibiotic treatment also reduces the incidence of acute rheumatic fever, which complicates 1 case per 100,000 in most of the US and Europe (relative risk reduction [RRR]=0.28).² The risk of acute rheumatic fever is higher in some populaHawaiians (13–45 per 100,000).³ Treatment may also reduce suppurative complications (peritonsil-tions, particularly Native Americans and lar or retropharyngeal abscess), which occur in 1 case out of 1000.^2,4

A systematic review of the diagnosis of GABHS evaluated the accuracy of history and physical exam elements.⁵ Clinical prediction rules based on selected symptoms and signs can identify patients at low risk for GABHS. The 4 Centor criteria (history of fever, anterior cervical adenopathy, tonsillar exudates, absence of cough) are well validated in adult populations ( Table 1 ), while other clinical prediction rules (such as McIssac) are validated in populations with children and adults ( Table 2 ). The number of criteria present determines the likelihood ratio (LR), with which to calculate the posttest probability of GABHS.

The usefulness of clinical prediction rules depends on knowing how prevalent GABHS is among cases of pharyngitis in a particular community. In a typical US adult population, GABHS comprises 5% to 10% of cases. The presence of only 1 Centor criterion would reduce the probability of GABHS pharyngitis to 2% to 3%, while meeting all 4 criteria would raise the probability to 25% to 40%, an intermediate value ( Table 1 ). If the prevalence of GABHS pharyngitis were 50%, as in some Native communities in Alaska, meeting all 4 criteria would predict an 86% probability of pharyngitis due to GABHS. Performing additional testing for patients with intermediate or high probability based on clinical prediction rules reduces the likelihood of unnecessary antibiotic treatment.¹

A systematic review⁶ of RAD testing demonstrates that the newer techniques (optical immunoassay, chemiluminescent DNA probes) have a sensitivity of 80% to 90%, which compares closely with that of throat culture (90%–95%). Both have a specificity greater than 95%, so false-positive test results are uncommon (LR+ =16–19). Treatment based on a positive RAD test would result in few unnecessary antibiotic prescriptions.¹

A retrospective outcome study⁴ reviewed the frequency of suppurative complications of GABHS among 30,036 patients with pharyngitis diagnosed with either RAD testing or throat culture. Patients included adults and children in a primary care setting. Complication rates were identical. A prospective study of 465 suburban outpatients with pharyngitis assessed the accuracy of RAD diagnosis using throat culture as a reference. The RAD accuracy was 93% for pediatric patients and 97% for adults.⁵ In another retrospective review of RAD testing, investigators performed 11,427 RAD tests over 3 years in a private pediatric group. There were 8385 negative tests, among which follow-up cultures detected 200 (2.4%) that were positive for GABHS. In the second half of the study, a newer RAD test produced a false-negative rate of 1.4%.⁷ Because of the possibility of higher false-negative RAD test rates in some settings, unless the physician has ascertained that RAD testing is comparable to throat culture in their own setting, expert opinion recommends confirming a negative RAD test in children or adolescents with a throat culture.¹ Patients at higher risk of GABHS or GABHS complications may also warrant throat culture back up of RAD testing.¹

TABLE 1
Centor clinical prediction rules for diagnosis of GABHS (for adults)

TABLE 2
McIssac clinical prediction rules for diagnosis of GABHS (for adults and children)

Recommendations from others

The Infectious Diseases Society of America recommends that if the physician is unable to exclude the diagnosis of GABHS on epidemiological or clinical grounds, either RAD testing or throat culture should be done. A positive result warrants treatment for patients with signs and symptoms of acute pharyngitis. A negative RAD result for a child or adolescent should be confirmed by throat culture unless the physician has ascertained that the sensitivity of RAD testing and throat culture are comparable in his or her practice setting.¹

The American Academy of Pediatrics also recommends laboratory confirmation of GABHS pharyngitis in children with throat culture or RAD testing. If a patient suspected clinically of GABHS has a negative RAD test, a throat culture should be done. Since some experts believe RAD tests using optical immunoassay are sufficiently sensitive to be used without throat culture backup, physicians who wish to use them should validate them by comparison to throat culture in their practice.⁸

Evidence summary

In the US, GABHS is the cause of acute pharyn-gitis in 5% to 10% of adults and 15% to 30% of children. It is the only commonly occurring cause of pharyngitis with an indication for antibiotic therapy.¹ The main benefit of antibiotic treatment in adults is earlier symptom relief—1 fewer day of fever and pain if antibiotics are begun within 3 days of onset.

Antibiotic treatment also reduces the incidence of acute rheumatic fever, which complicates 1 case per 100,000 in most of the US and Europe (relative risk reduction [RRR]=0.28).² The risk of acute rheumatic fever is higher in some populaHawaiians (13–45 per 100,000).³ Treatment may also reduce suppurative complications (peritonsil-tions, particularly Native Americans and lar or retropharyngeal abscess), which occur in 1 case out of 1000.^2,4

A systematic review of the diagnosis of GABHS evaluated the accuracy of history and physical exam elements.⁵ Clinical prediction rules based on selected symptoms and signs can identify patients at low risk for GABHS. The 4 Centor criteria (history of fever, anterior cervical adenopathy, tonsillar exudates, absence of cough) are well validated in adult populations ( Table 1 ), while other clinical prediction rules (such as McIssac) are validated in populations with children and adults ( Table 2 ). The number of criteria present determines the likelihood ratio (LR), with which to calculate the posttest probability of GABHS.

The usefulness of clinical prediction rules depends on knowing how prevalent GABHS is among cases of pharyngitis in a particular community. In a typical US adult population, GABHS comprises 5% to 10% of cases. The presence of only 1 Centor criterion would reduce the probability of GABHS pharyngitis to 2% to 3%, while meeting all 4 criteria would raise the probability to 25% to 40%, an intermediate value ( Table 1 ). If the prevalence of GABHS pharyngitis were 50%, as in some Native communities in Alaska, meeting all 4 criteria would predict an 86% probability of pharyngitis due to GABHS. Performing additional testing for patients with intermediate or high probability based on clinical prediction rules reduces the likelihood of unnecessary antibiotic treatment.¹

A systematic review⁶ of RAD testing demonstrates that the newer techniques (optical immunoassay, chemiluminescent DNA probes) have a sensitivity of 80% to 90%, which compares closely with that of throat culture (90%–95%). Both have a specificity greater than 95%, so false-positive test results are uncommon (LR+ =16–19). Treatment based on a positive RAD test would result in few unnecessary antibiotic prescriptions.¹

A retrospective outcome study⁴ reviewed the frequency of suppurative complications of GABHS among 30,036 patients with pharyngitis diagnosed with either RAD testing or throat culture. Patients included adults and children in a primary care setting. Complication rates were identical. A prospective study of 465 suburban outpatients with pharyngitis assessed the accuracy of RAD diagnosis using throat culture as a reference. The RAD accuracy was 93% for pediatric patients and 97% for adults.⁵ In another retrospective review of RAD testing, investigators performed 11,427 RAD tests over 3 years in a private pediatric group. There were 8385 negative tests, among which follow-up cultures detected 200 (2.4%) that were positive for GABHS. In the second half of the study, a newer RAD test produced a false-negative rate of 1.4%.⁷ Because of the possibility of higher false-negative RAD test rates in some settings, unless the physician has ascertained that RAD testing is comparable to throat culture in their own setting, expert opinion recommends confirming a negative RAD test in children or adolescents with a throat culture.¹ Patients at higher risk of GABHS or GABHS complications may also warrant throat culture back up of RAD testing.¹

TABLE 1
Centor clinical prediction rules for diagnosis of GABHS (for adults)

TABLE 2
McIssac clinical prediction rules for diagnosis of GABHS (for adults and children)

Recommendations from others

The Infectious Diseases Society of America recommends that if the physician is unable to exclude the diagnosis of GABHS on epidemiological or clinical grounds, either RAD testing or throat culture should be done. A positive result warrants treatment for patients with signs and symptoms of acute pharyngitis. A negative RAD result for a child or adolescent should be confirmed by throat culture unless the physician has ascertained that the sensitivity of RAD testing and throat culture are comparable in his or her practice setting.¹

The American Academy of Pediatrics also recommends laboratory confirmation of GABHS pharyngitis in children with throat culture or RAD testing. If a patient suspected clinically of GABHS has a negative RAD test, a throat culture should be done. Since some experts believe RAD tests using optical immunoassay are sufficiently sensitive to be used without throat culture backup, physicians who wish to use them should validate them by comparison to throat culture in their practice.⁸

Evidence summary

In the US, GABHS is the cause of acute pharyn-gitis in 5% to 10% of adults and 15% to 30% of children. It is the only commonly occurring cause of pharyngitis with an indication for antibiotic therapy.¹ The main benefit of antibiotic treatment in adults is earlier symptom relief—1 fewer day of fever and pain if antibiotics are begun within 3 days of onset.

Antibiotic treatment also reduces the incidence of acute rheumatic fever, which complicates 1 case per 100,000 in most of the US and Europe (relative risk reduction [RRR]=0.28).² The risk of acute rheumatic fever is higher in some populaHawaiians (13–45 per 100,000).³ Treatment may also reduce suppurative complications (peritonsil-tions, particularly Native Americans and lar or retropharyngeal abscess), which occur in 1 case out of 1000.^2,4

A systematic review of the diagnosis of GABHS evaluated the accuracy of history and physical exam elements.⁵ Clinical prediction rules based on selected symptoms and signs can identify patients at low risk for GABHS. The 4 Centor criteria (history of fever, anterior cervical adenopathy, tonsillar exudates, absence of cough) are well validated in adult populations ( Table 1 ), while other clinical prediction rules (such as McIssac) are validated in populations with children and adults ( Table 2 ). The number of criteria present determines the likelihood ratio (LR), with which to calculate the posttest probability of GABHS.

The usefulness of clinical prediction rules depends on knowing how prevalent GABHS is among cases of pharyngitis in a particular community. In a typical US adult population, GABHS comprises 5% to 10% of cases. The presence of only 1 Centor criterion would reduce the probability of GABHS pharyngitis to 2% to 3%, while meeting all 4 criteria would raise the probability to 25% to 40%, an intermediate value ( Table 1 ). If the prevalence of GABHS pharyngitis were 50%, as in some Native communities in Alaska, meeting all 4 criteria would predict an 86% probability of pharyngitis due to GABHS. Performing additional testing for patients with intermediate or high probability based on clinical prediction rules reduces the likelihood of unnecessary antibiotic treatment.¹

A systematic review⁶ of RAD testing demonstrates that the newer techniques (optical immunoassay, chemiluminescent DNA probes) have a sensitivity of 80% to 90%, which compares closely with that of throat culture (90%–95%). Both have a specificity greater than 95%, so false-positive test results are uncommon (LR+ =16–19). Treatment based on a positive RAD test would result in few unnecessary antibiotic prescriptions.¹

A retrospective outcome study⁴ reviewed the frequency of suppurative complications of GABHS among 30,036 patients with pharyngitis diagnosed with either RAD testing or throat culture. Patients included adults and children in a primary care setting. Complication rates were identical. A prospective study of 465 suburban outpatients with pharyngitis assessed the accuracy of RAD diagnosis using throat culture as a reference. The RAD accuracy was 93% for pediatric patients and 97% for adults.⁵ In another retrospective review of RAD testing, investigators performed 11,427 RAD tests over 3 years in a private pediatric group. There were 8385 negative tests, among which follow-up cultures detected 200 (2.4%) that were positive for GABHS. In the second half of the study, a newer RAD test produced a false-negative rate of 1.4%.⁷ Because of the possibility of higher false-negative RAD test rates in some settings, unless the physician has ascertained that RAD testing is comparable to throat culture in their own setting, expert opinion recommends confirming a negative RAD test in children or adolescents with a throat culture.¹ Patients at higher risk of GABHS or GABHS complications may also warrant throat culture back up of RAD testing.¹

TABLE 1
Centor clinical prediction rules for diagnosis of GABHS (for adults)

TABLE 2
McIssac clinical prediction rules for diagnosis of GABHS (for adults and children)

Recommendations from others

The Infectious Diseases Society of America recommends that if the physician is unable to exclude the diagnosis of GABHS on epidemiological or clinical grounds, either RAD testing or throat culture should be done. A positive result warrants treatment for patients with signs and symptoms of acute pharyngitis. A negative RAD result for a child or adolescent should be confirmed by throat culture unless the physician has ascertained that the sensitivity of RAD testing and throat culture are comparable in his or her practice setting.¹

The American Academy of Pediatrics also recommends laboratory confirmation of GABHS pharyngitis in children with throat culture or RAD testing. If a patient suspected clinically of GABHS has a negative RAD test, a throat culture should be done. Since some experts believe RAD tests using optical immunoassay are sufficiently sensitive to be used without throat culture backup, physicians who wish to use them should validate them by comparison to throat culture in their practice.⁸

FIGURE
Nonproliferative and proliferative diabetic retinopathy

Proliferative diabetic retinopathy. Note the network of abnormal preretinal new blood vessels just super-otemporal to the macula and the linear accumulations of blood in the preretinal vitreous below the macula.

Evidence summary

The Liverpool Diabetic Eye Study¹ prospectively evaluated the risk of vision-threatening retinopathy in a cohort that included all patients with diabetes mellitus who were registered with a general practitioner and were not under the care of an ophthalmologist. A subgroup of 4770 patients with type 2 diabetes who did not have sightthreatening retinopathy at baseline underwent at least 1 additional screen. Screening included non-stereoscopic 3-field (45° or 50° field) mydriatic photography. Median follow-up was 3.5 years (range, 1–8.5 years).

The patients were divided into cohorts based on level of demonstrated retinopathy. The mean screening interval for a 95% probability of remaining free of sight-threatening retinopathy was calculated for each grade of baseline retinopathy. Screening patients with no retinopathy every 5 years provided a 95% probability of remaining free of sight-threatening retinopathy. Patients with background retinopathy must be screened annually to achieve the same result, and patients with mild preproliferative retinopathy need to be screened every 4 months ( Table ).

A systematic review² of multiple small English-language studies evaluating screening and monitoring of diabetic retinopathy found consistent results. Screening by direct or indirect ophthalmoscopy alone detected 65% of patients with sight-threatening retinopathy. Screening by mydriatic retinal photography, augmented by ophthal-moscopy when the photographs were inconclusive, detected 88% to 100% of such cases.

An RCT of 1700 patients with diabetes and retinopathy evaluated preservation of vision with photocoagulation.³ Patients were not differentiated by type of diabetes. Each patient had initial and follow-up stereoscopic fundus photography. One eye was selected at random to receive treatment and the other remained untreated to serve as a control. Because of the magnitude of difference in vision between the eyes, the study was halted at 2 years to permit photocoagulation of the untreated eyes. Patients whose eyes had new vessels on or near the disk lost vision (defined as visual acuity less than 5/200) more often in untreated eyes (18.3% cumulative rate at 2 years) compared with treated eyes (6.4%; number needed to treat [NNT]=8.4).

Another RCT⁴ of patients with diabetes showed that photocoagulation maintained vision in diabetic retinopathy if the disease was not too advanced. Ninety-nine patients, also not differentiated by type of diabetes, were each treated in 1 eye chosen at random with a xenon-arc photocoagulator. Patients underwent follow-up treatments to the treated eye by clinical indication. The untreated eyes were observed as controls. Blindness occurred significantly less often in the treated eyes (19% total after 5 to 7 years) than in the control eyes (39%; NNT=5 to prevent 1 blind eye). Patients without proliferative retinopathy at onset experienced the most dramatic slowing of deterioration; photocoagulation was more useful in maintaining than in improving vision.

TABLE
Screening frequency for sight-threatening retinopathy

Recommendations from others

The American Diabetes Association 2003 Clinical Practice Recommendations⁵ state that patients with type 2 diabetes should have an initial dilated and comprehensive eye examination by an ophthalmologist or optometrist shortly after diagnosis of diabetes. An ophthalmologist or optometrist who is knowledgeable and experienced in diagnosing diabetic retinopathy and is aware of its management should repeat subsequent examinations for both type 1 and type 2 diabetic patients annually. Examinations are required more frequently if retinopathy is progressing.

FIGURE
Nonproliferative and proliferative diabetic retinopathy

Proliferative diabetic retinopathy. Note the network of abnormal preretinal new blood vessels just super-otemporal to the macula and the linear accumulations of blood in the preretinal vitreous below the macula.

Evidence summary

The Liverpool Diabetic Eye Study¹ prospectively evaluated the risk of vision-threatening retinopathy in a cohort that included all patients with diabetes mellitus who were registered with a general practitioner and were not under the care of an ophthalmologist. A subgroup of 4770 patients with type 2 diabetes who did not have sightthreatening retinopathy at baseline underwent at least 1 additional screen. Screening included non-stereoscopic 3-field (45° or 50° field) mydriatic photography. Median follow-up was 3.5 years (range, 1–8.5 years).

The patients were divided into cohorts based on level of demonstrated retinopathy. The mean screening interval for a 95% probability of remaining free of sight-threatening retinopathy was calculated for each grade of baseline retinopathy. Screening patients with no retinopathy every 5 years provided a 95% probability of remaining free of sight-threatening retinopathy. Patients with background retinopathy must be screened annually to achieve the same result, and patients with mild preproliferative retinopathy need to be screened every 4 months ( Table ).

A systematic review² of multiple small English-language studies evaluating screening and monitoring of diabetic retinopathy found consistent results. Screening by direct or indirect ophthalmoscopy alone detected 65% of patients with sight-threatening retinopathy. Screening by mydriatic retinal photography, augmented by ophthal-moscopy when the photographs were inconclusive, detected 88% to 100% of such cases.

An RCT of 1700 patients with diabetes and retinopathy evaluated preservation of vision with photocoagulation.³ Patients were not differentiated by type of diabetes. Each patient had initial and follow-up stereoscopic fundus photography. One eye was selected at random to receive treatment and the other remained untreated to serve as a control. Because of the magnitude of difference in vision between the eyes, the study was halted at 2 years to permit photocoagulation of the untreated eyes. Patients whose eyes had new vessels on or near the disk lost vision (defined as visual acuity less than 5/200) more often in untreated eyes (18.3% cumulative rate at 2 years) compared with treated eyes (6.4%; number needed to treat [NNT]=8.4).

Another RCT⁴ of patients with diabetes showed that photocoagulation maintained vision in diabetic retinopathy if the disease was not too advanced. Ninety-nine patients, also not differentiated by type of diabetes, were each treated in 1 eye chosen at random with a xenon-arc photocoagulator. Patients underwent follow-up treatments to the treated eye by clinical indication. The untreated eyes were observed as controls. Blindness occurred significantly less often in the treated eyes (19% total after 5 to 7 years) than in the control eyes (39%; NNT=5 to prevent 1 blind eye). Patients without proliferative retinopathy at onset experienced the most dramatic slowing of deterioration; photocoagulation was more useful in maintaining than in improving vision.

TABLE
Screening frequency for sight-threatening retinopathy

Recommendations from others

The American Diabetes Association 2003 Clinical Practice Recommendations⁵ state that patients with type 2 diabetes should have an initial dilated and comprehensive eye examination by an ophthalmologist or optometrist shortly after diagnosis of diabetes. An ophthalmologist or optometrist who is knowledgeable and experienced in diagnosing diabetic retinopathy and is aware of its management should repeat subsequent examinations for both type 1 and type 2 diabetic patients annually. Examinations are required more frequently if retinopathy is progressing.

FIGURE
Nonproliferative and proliferative diabetic retinopathy

Proliferative diabetic retinopathy. Note the network of abnormal preretinal new blood vessels just super-otemporal to the macula and the linear accumulations of blood in the preretinal vitreous below the macula.

Evidence summary

The Liverpool Diabetic Eye Study¹ prospectively evaluated the risk of vision-threatening retinopathy in a cohort that included all patients with diabetes mellitus who were registered with a general practitioner and were not under the care of an ophthalmologist. A subgroup of 4770 patients with type 2 diabetes who did not have sightthreatening retinopathy at baseline underwent at least 1 additional screen. Screening included non-stereoscopic 3-field (45° or 50° field) mydriatic photography. Median follow-up was 3.5 years (range, 1–8.5 years).

The patients were divided into cohorts based on level of demonstrated retinopathy. The mean screening interval for a 95% probability of remaining free of sight-threatening retinopathy was calculated for each grade of baseline retinopathy. Screening patients with no retinopathy every 5 years provided a 95% probability of remaining free of sight-threatening retinopathy. Patients with background retinopathy must be screened annually to achieve the same result, and patients with mild preproliferative retinopathy need to be screened every 4 months ( Table ).

A systematic review² of multiple small English-language studies evaluating screening and monitoring of diabetic retinopathy found consistent results. Screening by direct or indirect ophthalmoscopy alone detected 65% of patients with sight-threatening retinopathy. Screening by mydriatic retinal photography, augmented by ophthal-moscopy when the photographs were inconclusive, detected 88% to 100% of such cases.

An RCT of 1700 patients with diabetes and retinopathy evaluated preservation of vision with photocoagulation.³ Patients were not differentiated by type of diabetes. Each patient had initial and follow-up stereoscopic fundus photography. One eye was selected at random to receive treatment and the other remained untreated to serve as a control. Because of the magnitude of difference in vision between the eyes, the study was halted at 2 years to permit photocoagulation of the untreated eyes. Patients whose eyes had new vessels on or near the disk lost vision (defined as visual acuity less than 5/200) more often in untreated eyes (18.3% cumulative rate at 2 years) compared with treated eyes (6.4%; number needed to treat [NNT]=8.4).

Another RCT⁴ of patients with diabetes showed that photocoagulation maintained vision in diabetic retinopathy if the disease was not too advanced. Ninety-nine patients, also not differentiated by type of diabetes, were each treated in 1 eye chosen at random with a xenon-arc photocoagulator. Patients underwent follow-up treatments to the treated eye by clinical indication. The untreated eyes were observed as controls. Blindness occurred significantly less often in the treated eyes (19% total after 5 to 7 years) than in the control eyes (39%; NNT=5 to prevent 1 blind eye). Patients without proliferative retinopathy at onset experienced the most dramatic slowing of deterioration; photocoagulation was more useful in maintaining than in improving vision.

TABLE
Screening frequency for sight-threatening retinopathy

Recommendations from others

The American Diabetes Association 2003 Clinical Practice Recommendations⁵ state that patients with type 2 diabetes should have an initial dilated and comprehensive eye examination by an ophthalmologist or optometrist shortly after diagnosis of diabetes. An ophthalmologist or optometrist who is knowledgeable and experienced in diagnosing diabetic retinopathy and is aware of its management should repeat subsequent examinations for both type 1 and type 2 diabetic patients annually. Examinations are required more frequently if retinopathy is progressing.

Evidence summary

No studies of inhaled beta-agonists have been conducted with patients who have an explicit diagnosis of acute cough due to URI. While some clinicians feel a distinction between URI and acute bronchitis should be made, there is significant overlap between these diagnoses in clinical practice, as well as in the available studies.

A systematic review looking at beta-agonists for acute bronchitis included the clinical diagnoses of both acute bronchitis and acute cough because a standard definition of bronchitis is lacking.¹ Only two trials in this review examined inhaled beta-agonists. When results from these trials were combined for the outcome of productive cough at 7 days, inhaled beta-agonists showed no benefit. However, the authors note that details of the individual trials may help to clarify the effect of inhaled beta-agonists.

One trial, a randomized controlled trial of adult patients with acute bronchitis in 2 community-based family practices, compared 23 patients receiving albuterolin a multidose inhaler (MDI) with 23 patients receiving placebo inhaler.² Patients were also randomized to receive erythromycin or placebo tablets. Patients with pneumonia or a history of asthma or chronic obstructive pulmonary disease (COPD) were excluded. At 7 days, 61% of patients in the albuterol group reported cough compared with 91% in the control group (P=.02, NNT=3). No statistically significant difference was seen in productive cough or night cough. Smokers responded to inhaled albuterol similarly to nonsmokers. Erythromycin had no effect on cough and side effects were similar among all groups.

The other trial was a randomized controlled trial of 80 adults with cough due to acute respiratory infection; it compared fenoterol aerosol 4 times daily with placebo.³ Inhaled fenoterol is not available in the US but is similar to albuterol. This study showed no difference in cough at 7 days (relative risk [RR]=0.83; 95% confidence interval [CI], 0.52–1.30). In a subgroup analysis, however, smokers and those wheezing on initial exam had lower overall symptom scores when treated with fenoterol.

Recommendations from others

We were unable to find any guidelines on the use of albuterol via MDI for cough from bronchitis or URIs.

Evidence summary

No studies of inhaled beta-agonists have been conducted with patients who have an explicit diagnosis of acute cough due to URI. While some clinicians feel a distinction between URI and acute bronchitis should be made, there is significant overlap between these diagnoses in clinical practice, as well as in the available studies.

A systematic review looking at beta-agonists for acute bronchitis included the clinical diagnoses of both acute bronchitis and acute cough because a standard definition of bronchitis is lacking.¹ Only two trials in this review examined inhaled beta-agonists. When results from these trials were combined for the outcome of productive cough at 7 days, inhaled beta-agonists showed no benefit. However, the authors note that details of the individual trials may help to clarify the effect of inhaled beta-agonists.

One trial, a randomized controlled trial of adult patients with acute bronchitis in 2 community-based family practices, compared 23 patients receiving albuterolin a multidose inhaler (MDI) with 23 patients receiving placebo inhaler.² Patients were also randomized to receive erythromycin or placebo tablets. Patients with pneumonia or a history of asthma or chronic obstructive pulmonary disease (COPD) were excluded. At 7 days, 61% of patients in the albuterol group reported cough compared with 91% in the control group (P=.02, NNT=3). No statistically significant difference was seen in productive cough or night cough. Smokers responded to inhaled albuterol similarly to nonsmokers. Erythromycin had no effect on cough and side effects were similar among all groups.

The other trial was a randomized controlled trial of 80 adults with cough due to acute respiratory infection; it compared fenoterol aerosol 4 times daily with placebo.³ Inhaled fenoterol is not available in the US but is similar to albuterol. This study showed no difference in cough at 7 days (relative risk [RR]=0.83; 95% confidence interval [CI], 0.52–1.30). In a subgroup analysis, however, smokers and those wheezing on initial exam had lower overall symptom scores when treated with fenoterol.

Recommendations from others

We were unable to find any guidelines on the use of albuterol via MDI for cough from bronchitis or URIs.

Evidence summary

No studies of inhaled beta-agonists have been conducted with patients who have an explicit diagnosis of acute cough due to URI. While some clinicians feel a distinction between URI and acute bronchitis should be made, there is significant overlap between these diagnoses in clinical practice, as well as in the available studies.

A systematic review looking at beta-agonists for acute bronchitis included the clinical diagnoses of both acute bronchitis and acute cough because a standard definition of bronchitis is lacking.¹ Only two trials in this review examined inhaled beta-agonists. When results from these trials were combined for the outcome of productive cough at 7 days, inhaled beta-agonists showed no benefit. However, the authors note that details of the individual trials may help to clarify the effect of inhaled beta-agonists.

One trial, a randomized controlled trial of adult patients with acute bronchitis in 2 community-based family practices, compared 23 patients receiving albuterolin a multidose inhaler (MDI) with 23 patients receiving placebo inhaler.² Patients were also randomized to receive erythromycin or placebo tablets. Patients with pneumonia or a history of asthma or chronic obstructive pulmonary disease (COPD) were excluded. At 7 days, 61% of patients in the albuterol group reported cough compared with 91% in the control group (P=.02, NNT=3). No statistically significant difference was seen in productive cough or night cough. Smokers responded to inhaled albuterol similarly to nonsmokers. Erythromycin had no effect on cough and side effects were similar among all groups.

The other trial was a randomized controlled trial of 80 adults with cough due to acute respiratory infection; it compared fenoterol aerosol 4 times daily with placebo.³ Inhaled fenoterol is not available in the US but is similar to albuterol. This study showed no difference in cough at 7 days (relative risk [RR]=0.83; 95% confidence interval [CI], 0.52–1.30). In a subgroup analysis, however, smokers and those wheezing on initial exam had lower overall symptom scores when treated with fenoterol.

Recommendations from others

We were unable to find any guidelines on the use of albuterol via MDI for cough from bronchitis or URIs.