Todd McDiarmid, MD

Evidence summary

rGH has been available since 1985. The Food and Drug Administration has approved it for such conditions as growth hormone deficiency, chronic renal insufficiency, Turner syndrome, small size for gestational age, and Prader-Willi syndrome.¹ The use of rGH to treat idiopathic short stature introduces many clinical, economic, and ethical questions. We have attempted to discern the clinical effectiveness of treatment by focusing on RCTs of rGH therapy while leaving the other substantive issues unexplored.

Final height is arguably the most important outcome measure for the effects of rGH and may be represented as actual height or as a standard deviation score (SDS)—actual height minus mean height for age divided by standard deviation of height for age.² This measure standardizes height comparisons for different age groups and is comparable to the percentile values on growth charts.

Growth hormone increases height in girls and boys

A 2003 Cochrane systematic review identified 9 RCTs that evaluated treatment with rGH in children with idiopathic short stature. Only 1 used near final height as its main outcome. Inclusion criteria for this RCT comprised prepubertal girls in the bottom third percentile for height without a known cause.

Of the 40 subjects, only 18 provided consent for randomization. Seven of the 10 girls randomized to the treatment group and 6 of the 8 randomized to the control group completed the study to final height measurement. The average age of the treated girls at the start of therapy was 8.07 years; the average duration of treatment was 6.2 years. All participants reached stage 4 breast development, menarche, and a growth velocity of <2 cm per year in the year preceding final height measurement. Mean final height in the treatment group was 155.3 cm compared to 147.8 cm in the control group—a 7.5-cm difference (95% confidence interval [CI], 3.14-11.86 cm).^2,3

A double-blind, placebo-controlled RCT published after the Cochrane review assessed final height in a peripubertal, predominantly male population with non-growth-hormone-deficient short stature.⁴ Inclusion criteria comprised a height SDS <–2.50, but 6 participants with a height SDS between –2.25 and –2.5 were included because of a change in the criteria.⁵ Sixty-eight children were initially randomized. Of the 37 randomized to treatment, 22 were available for final height measurement. The placebo group had a higher dropout rate—only 11 of 31 patients were available for final height measurement. In an attempt to reduce the dropout rate, the final height criterion for discontinuation of injections was changed from <0.5 to <1.5 cm per year. The mean age of the treatment group was 12.5 years at initiation of treatment; average duration of treatment was 4.6 years.

Intent-to-treat analysis of patients who received at least 6 months of treatment with final height assessment revealed a positive treatment effect on height (SDS) of 0.51. This is the equivalent of a 3.7-cm difference in final height for the treatment group compared with the placebo group (P<.02; 95% CI, 0.10-0.92 SDS).⁵

Recommendations

The FDA has approved rGH for use in children with height SDS ≤–2.25—equivalent to the lowest 1.2% of children. The Lawson-Wilkins Pediatric Endocrinology Society Drug and Therapeutics Committee states that rGH therapy should be considered only after accurate diagnosis, careful monitoring of growth velocity, and estimation of final height by a pediatric endocrinologist.^6,7

Evidence summary

rGH has been available since 1985. The Food and Drug Administration has approved it for such conditions as growth hormone deficiency, chronic renal insufficiency, Turner syndrome, small size for gestational age, and Prader-Willi syndrome.¹ The use of rGH to treat idiopathic short stature introduces many clinical, economic, and ethical questions. We have attempted to discern the clinical effectiveness of treatment by focusing on RCTs of rGH therapy while leaving the other substantive issues unexplored.

Final height is arguably the most important outcome measure for the effects of rGH and may be represented as actual height or as a standard deviation score (SDS)—actual height minus mean height for age divided by standard deviation of height for age.² This measure standardizes height comparisons for different age groups and is comparable to the percentile values on growth charts.

Growth hormone increases height in girls and boys

A 2003 Cochrane systematic review identified 9 RCTs that evaluated treatment with rGH in children with idiopathic short stature. Only 1 used near final height as its main outcome. Inclusion criteria for this RCT comprised prepubertal girls in the bottom third percentile for height without a known cause.

Of the 40 subjects, only 18 provided consent for randomization. Seven of the 10 girls randomized to the treatment group and 6 of the 8 randomized to the control group completed the study to final height measurement. The average age of the treated girls at the start of therapy was 8.07 years; the average duration of treatment was 6.2 years. All participants reached stage 4 breast development, menarche, and a growth velocity of <2 cm per year in the year preceding final height measurement. Mean final height in the treatment group was 155.3 cm compared to 147.8 cm in the control group—a 7.5-cm difference (95% confidence interval [CI], 3.14-11.86 cm).^2,3

A double-blind, placebo-controlled RCT published after the Cochrane review assessed final height in a peripubertal, predominantly male population with non-growth-hormone-deficient short stature.⁴ Inclusion criteria comprised a height SDS <–2.50, but 6 participants with a height SDS between –2.25 and –2.5 were included because of a change in the criteria.⁵ Sixty-eight children were initially randomized. Of the 37 randomized to treatment, 22 were available for final height measurement. The placebo group had a higher dropout rate—only 11 of 31 patients were available for final height measurement. In an attempt to reduce the dropout rate, the final height criterion for discontinuation of injections was changed from <0.5 to <1.5 cm per year. The mean age of the treatment group was 12.5 years at initiation of treatment; average duration of treatment was 4.6 years.

Intent-to-treat analysis of patients who received at least 6 months of treatment with final height assessment revealed a positive treatment effect on height (SDS) of 0.51. This is the equivalent of a 3.7-cm difference in final height for the treatment group compared with the placebo group (P<.02; 95% CI, 0.10-0.92 SDS).⁵

Recommendations

The FDA has approved rGH for use in children with height SDS ≤–2.25—equivalent to the lowest 1.2% of children. The Lawson-Wilkins Pediatric Endocrinology Society Drug and Therapeutics Committee states that rGH therapy should be considered only after accurate diagnosis, careful monitoring of growth velocity, and estimation of final height by a pediatric endocrinologist.^6,7

Evidence summary

rGH has been available since 1985. The Food and Drug Administration has approved it for such conditions as growth hormone deficiency, chronic renal insufficiency, Turner syndrome, small size for gestational age, and Prader-Willi syndrome.¹ The use of rGH to treat idiopathic short stature introduces many clinical, economic, and ethical questions. We have attempted to discern the clinical effectiveness of treatment by focusing on RCTs of rGH therapy while leaving the other substantive issues unexplored.

Final height is arguably the most important outcome measure for the effects of rGH and may be represented as actual height or as a standard deviation score (SDS)—actual height minus mean height for age divided by standard deviation of height for age.² This measure standardizes height comparisons for different age groups and is comparable to the percentile values on growth charts.

Growth hormone increases height in girls and boys

A 2003 Cochrane systematic review identified 9 RCTs that evaluated treatment with rGH in children with idiopathic short stature. Only 1 used near final height as its main outcome. Inclusion criteria for this RCT comprised prepubertal girls in the bottom third percentile for height without a known cause.

Of the 40 subjects, only 18 provided consent for randomization. Seven of the 10 girls randomized to the treatment group and 6 of the 8 randomized to the control group completed the study to final height measurement. The average age of the treated girls at the start of therapy was 8.07 years; the average duration of treatment was 6.2 years. All participants reached stage 4 breast development, menarche, and a growth velocity of <2 cm per year in the year preceding final height measurement. Mean final height in the treatment group was 155.3 cm compared to 147.8 cm in the control group—a 7.5-cm difference (95% confidence interval [CI], 3.14-11.86 cm).^2,3

A double-blind, placebo-controlled RCT published after the Cochrane review assessed final height in a peripubertal, predominantly male population with non-growth-hormone-deficient short stature.⁴ Inclusion criteria comprised a height SDS <–2.50, but 6 participants with a height SDS between –2.25 and –2.5 were included because of a change in the criteria.⁵ Sixty-eight children were initially randomized. Of the 37 randomized to treatment, 22 were available for final height measurement. The placebo group had a higher dropout rate—only 11 of 31 patients were available for final height measurement. In an attempt to reduce the dropout rate, the final height criterion for discontinuation of injections was changed from <0.5 to <1.5 cm per year. The mean age of the treatment group was 12.5 years at initiation of treatment; average duration of treatment was 4.6 years.

Intent-to-treat analysis of patients who received at least 6 months of treatment with final height assessment revealed a positive treatment effect on height (SDS) of 0.51. This is the equivalent of a 3.7-cm difference in final height for the treatment group compared with the placebo group (P<.02; 95% CI, 0.10-0.92 SDS).⁵

Recommendations

The FDA has approved rGH for use in children with height SDS ≤–2.25—equivalent to the lowest 1.2% of children. The Lawson-Wilkins Pediatric Endocrinology Society Drug and Therapeutics Committee states that rGH therapy should be considered only after accurate diagnosis, careful monitoring of growth velocity, and estimation of final height by a pediatric endocrinologist.^6,7

Evidence summary

Ten percent of preeclampsia occurs in pregnancies at less than 34 weeks of gestation. Traditionally, physicians often recommended bed rest to preterm, preeclamptic patients in the belief that it would improve neonatal outcomes.

RCTs find no difference between bed rest and ad lib movement

A single-center RCT investigated bed rest treatment for 105 patients with preeclampsia and gestational ages between 26 to 38 weeks. Patients were assigned to either strict bed rest with bathroom privileges in the hospital until delivery, or to bed rest with the ability to move freely around the hospital. Outcome assessors were not blinded to patient treatment allocation. There was no statistical difference between the 2 groups in perinatal or neonatal mortality, or in the neonatal morbidities of preterm births, endotracheal intubations, or NICU admissions.¹

Similarly, a small, unblinded RCT of 40 preeclamptic patients treated in the hospital with strict bed confinement or without restrictions found no significant difference in fetal or perinatal mortality.² No power calculations were reported for detecting differences in neonatal outcome rates in either of these studies.

Studies in nonproteinuric hypertension were no different

In addition to the studies in patients with preeclampsia, 2 RCTs measured neonatal outcomes with bed rest compared with normal activity in pregnancies complicated by nonproteinuric hypertension. These studies also found that bed rest did not improve neonatal outcomes.

The first trial was a multicenter RCT involving 218 patients between 28 to 38 weeks gestation with nonproteinuric hypertension (blood pressure >140/90 mm Hg). The patients were randomized to 2 groups: bed rest in the hospital but allowed to move around the ward, and normal activity at home without restrictions. The outcomes were measured by masked assessors. There were no statistical differences in perinatal or neonatal mortality, or in the neonatal morbidities of preterm birth, newborns small for their gestational age, or NICU admissions between the 2 groups.³

A second RCT of 135 nonproteinuric but hypertensive pregnant patients with diastolic blood pressures between 90 and 109 mm Hg also demonstrated no difference between patients treated with bed rest and sedation or normal activity in fetal or neonatal outcomes.⁴

Recommendations from others

An American College of Obstetrics and Gynecology practice bulletin on diagnosis and management of preeclampsia and eclampsia does not mention bed rest.⁵ The Canadian Hypertension Society Consensus Conference in 1997 stated that a “policy of hospital admission and strict bed rest is not advised for gestational hypertension with or without proteinuria.”⁶

Evidence summary

Ten percent of preeclampsia occurs in pregnancies at less than 34 weeks of gestation. Traditionally, physicians often recommended bed rest to preterm, preeclamptic patients in the belief that it would improve neonatal outcomes.

RCTs find no difference between bed rest and ad lib movement

A single-center RCT investigated bed rest treatment for 105 patients with preeclampsia and gestational ages between 26 to 38 weeks. Patients were assigned to either strict bed rest with bathroom privileges in the hospital until delivery, or to bed rest with the ability to move freely around the hospital. Outcome assessors were not blinded to patient treatment allocation. There was no statistical difference between the 2 groups in perinatal or neonatal mortality, or in the neonatal morbidities of preterm births, endotracheal intubations, or NICU admissions.¹

Similarly, a small, unblinded RCT of 40 preeclamptic patients treated in the hospital with strict bed confinement or without restrictions found no significant difference in fetal or perinatal mortality.² No power calculations were reported for detecting differences in neonatal outcome rates in either of these studies.

Studies in nonproteinuric hypertension were no different

In addition to the studies in patients with preeclampsia, 2 RCTs measured neonatal outcomes with bed rest compared with normal activity in pregnancies complicated by nonproteinuric hypertension. These studies also found that bed rest did not improve neonatal outcomes.

The first trial was a multicenter RCT involving 218 patients between 28 to 38 weeks gestation with nonproteinuric hypertension (blood pressure >140/90 mm Hg). The patients were randomized to 2 groups: bed rest in the hospital but allowed to move around the ward, and normal activity at home without restrictions. The outcomes were measured by masked assessors. There were no statistical differences in perinatal or neonatal mortality, or in the neonatal morbidities of preterm birth, newborns small for their gestational age, or NICU admissions between the 2 groups.³

A second RCT of 135 nonproteinuric but hypertensive pregnant patients with diastolic blood pressures between 90 and 109 mm Hg also demonstrated no difference between patients treated with bed rest and sedation or normal activity in fetal or neonatal outcomes.⁴

Recommendations from others

An American College of Obstetrics and Gynecology practice bulletin on diagnosis and management of preeclampsia and eclampsia does not mention bed rest.⁵ The Canadian Hypertension Society Consensus Conference in 1997 stated that a “policy of hospital admission and strict bed rest is not advised for gestational hypertension with or without proteinuria.”⁶

Evidence summary

Ten percent of preeclampsia occurs in pregnancies at less than 34 weeks of gestation. Traditionally, physicians often recommended bed rest to preterm, preeclamptic patients in the belief that it would improve neonatal outcomes.

RCTs find no difference between bed rest and ad lib movement

A single-center RCT investigated bed rest treatment for 105 patients with preeclampsia and gestational ages between 26 to 38 weeks. Patients were assigned to either strict bed rest with bathroom privileges in the hospital until delivery, or to bed rest with the ability to move freely around the hospital. Outcome assessors were not blinded to patient treatment allocation. There was no statistical difference between the 2 groups in perinatal or neonatal mortality, or in the neonatal morbidities of preterm births, endotracheal intubations, or NICU admissions.¹

Similarly, a small, unblinded RCT of 40 preeclamptic patients treated in the hospital with strict bed confinement or without restrictions found no significant difference in fetal or perinatal mortality.² No power calculations were reported for detecting differences in neonatal outcome rates in either of these studies.

Studies in nonproteinuric hypertension were no different

In addition to the studies in patients with preeclampsia, 2 RCTs measured neonatal outcomes with bed rest compared with normal activity in pregnancies complicated by nonproteinuric hypertension. These studies also found that bed rest did not improve neonatal outcomes.

The first trial was a multicenter RCT involving 218 patients between 28 to 38 weeks gestation with nonproteinuric hypertension (blood pressure >140/90 mm Hg). The patients were randomized to 2 groups: bed rest in the hospital but allowed to move around the ward, and normal activity at home without restrictions. The outcomes were measured by masked assessors. There were no statistical differences in perinatal or neonatal mortality, or in the neonatal morbidities of preterm birth, newborns small for their gestational age, or NICU admissions between the 2 groups.³

A second RCT of 135 nonproteinuric but hypertensive pregnant patients with diastolic blood pressures between 90 and 109 mm Hg also demonstrated no difference between patients treated with bed rest and sedation or normal activity in fetal or neonatal outcomes.⁴

Recommendations from others

An American College of Obstetrics and Gynecology practice bulletin on diagnosis and management of preeclampsia and eclampsia does not mention bed rest.⁵ The Canadian Hypertension Society Consensus Conference in 1997 stated that a “policy of hospital admission and strict bed rest is not advised for gestational hypertension with or without proteinuria.”⁶

Evidence summary

Fractures within 1.5 cm of the fifth metatarsal tuberosity, without extension distal to the fourth-fifth intermetatarsal articulation, occurring with less than 2-week symptom prodrome and without a history of previous fracture, are defined as “acute Jones’ fractures” (FIGURE). In a recent RCT by Mologne et al,¹ 37 active-duty military personnel with acute Jones’ fractures were randomized to either 8 weeks of no weight-bearing in a short leg casting, followed by a walking cast or hard-soled shoe until clinical union; or to early outpatient intramedullary screw fixation followed by no weight-bearing for 2 weeks, then weight-bearing as tolerated in a hard-soled shoe until clinical union. Screw fixation significantly reduced both time to clinical union and time to return to sports—by nearly 50% when compared with non-weightbearing short leg casting. Furthermore, at 26 weeks the casting group saw a significant 44% failure rate compared with only 5% in the surgical group (number needed to treat [NNT]=2.6). Six patients in the surgical group had mild discomfort from the screw head, and 3 needed the screw to be removed. Generalization of the results was limited by the mostly male military population.

The rates and times of union with short leg casting vary over a wide range in the research literature. The casting group in the RCT above had union rates of 56% and median time to union of 14.5 weeks (lower and upper quartile range, 10.5–18.5).¹ A prospective registry of 68 consecutive acute Jones’ fractures in primarily young military service members showed a 72% union rate with non-weightbearing short leg casting with average time to union of 21.2 weeks.² A heterogeneous group of 5 retrospective follow-up studies of short leg casting reported wide ranges in union rates of 72% to 100%, and in time to healing of 7 weeks to 21 months.^3-7 These studies varied in average age, sex, and athletic ability of their samples as well as type of immobilization and weight-bearing status during treatment.

Tuberosity avulsion fractures are proximal fifth metatarsal styloid fractures resulting from a forceful pull of the lateral band of the long plantar ligament or the peroneus brevis tendon during ankle inversion. A 12-week RCT in 89 consecutive patients presenting to an emergency department with fifth metatarsal tuberosity avulsion fractures compared a nonrigid, soft Jones’ dressing consisting of alternating layers of cast padding and elastic bandages with a rigid short leg casting.⁸ The Jones’ dressing had a significant 28% reduction in time to return to pre-injury levels of activity. Other outcomes—time in treatment modality, time to radiographic healing, and functional foot score—were not different between intervention groups. Validity was limited by the 32% lost to follow-up rate.

FIGURE
Acute fracture of the fifth metatarsal

Recommendations from others

We were unable to locate any consensus statements or clinical guidelines regarding the treatment of Jones’ fractures.

DeLee and Drez’s Orthopaedic Sports Medicine recommends immobilization in a cast or below-the-knee boot with strict non-weightbearing for at least 6 weeks for acute Jones’ fractures.⁹ It recommends surgical treatment, followed by 6 weeks of cast immobilization, then progression to weight bearing based on radiographic findings, for nonoperative treatment failures or with desire to return high-performance athletes to activity.

In Fracture Management for Primary Care, the authors recommend posterior splinting and non-weightbearing with crutches for acute Jones’ fractures, followed by non-weightbearing short leg casting application at 3 to 5 days from injury.¹⁰ After a minimum of 6 to 8 weeks of casting, they recommend options of 4 additional weeks of casting or internal fixation for clinical or radiographic nonunion.

For tuberosity avulsion fractures, the authors recommend use of a firm-soled shoe for 4 to 8 weeks. For patients with discomfort at an initial 4- to 7-day follow-up, they give an option of using a walking short leg casting for 2 weeks, with follow-up every 2 to 4 weeks until clinical healing.

Evidence summary

Fractures within 1.5 cm of the fifth metatarsal tuberosity, without extension distal to the fourth-fifth intermetatarsal articulation, occurring with less than 2-week symptom prodrome and without a history of previous fracture, are defined as “acute Jones’ fractures” (FIGURE). In a recent RCT by Mologne et al,¹ 37 active-duty military personnel with acute Jones’ fractures were randomized to either 8 weeks of no weight-bearing in a short leg casting, followed by a walking cast or hard-soled shoe until clinical union; or to early outpatient intramedullary screw fixation followed by no weight-bearing for 2 weeks, then weight-bearing as tolerated in a hard-soled shoe until clinical union. Screw fixation significantly reduced both time to clinical union and time to return to sports—by nearly 50% when compared with non-weightbearing short leg casting. Furthermore, at 26 weeks the casting group saw a significant 44% failure rate compared with only 5% in the surgical group (number needed to treat [NNT]=2.6). Six patients in the surgical group had mild discomfort from the screw head, and 3 needed the screw to be removed. Generalization of the results was limited by the mostly male military population.

The rates and times of union with short leg casting vary over a wide range in the research literature. The casting group in the RCT above had union rates of 56% and median time to union of 14.5 weeks (lower and upper quartile range, 10.5–18.5).¹ A prospective registry of 68 consecutive acute Jones’ fractures in primarily young military service members showed a 72% union rate with non-weightbearing short leg casting with average time to union of 21.2 weeks.² A heterogeneous group of 5 retrospective follow-up studies of short leg casting reported wide ranges in union rates of 72% to 100%, and in time to healing of 7 weeks to 21 months.^3-7 These studies varied in average age, sex, and athletic ability of their samples as well as type of immobilization and weight-bearing status during treatment.

Tuberosity avulsion fractures are proximal fifth metatarsal styloid fractures resulting from a forceful pull of the lateral band of the long plantar ligament or the peroneus brevis tendon during ankle inversion. A 12-week RCT in 89 consecutive patients presenting to an emergency department with fifth metatarsal tuberosity avulsion fractures compared a nonrigid, soft Jones’ dressing consisting of alternating layers of cast padding and elastic bandages with a rigid short leg casting.⁸ The Jones’ dressing had a significant 28% reduction in time to return to pre-injury levels of activity. Other outcomes—time in treatment modality, time to radiographic healing, and functional foot score—were not different between intervention groups. Validity was limited by the 32% lost to follow-up rate.

FIGURE
Acute fracture of the fifth metatarsal

Recommendations from others

We were unable to locate any consensus statements or clinical guidelines regarding the treatment of Jones’ fractures.

DeLee and Drez’s Orthopaedic Sports Medicine recommends immobilization in a cast or below-the-knee boot with strict non-weightbearing for at least 6 weeks for acute Jones’ fractures.⁹ It recommends surgical treatment, followed by 6 weeks of cast immobilization, then progression to weight bearing based on radiographic findings, for nonoperative treatment failures or with desire to return high-performance athletes to activity.

In Fracture Management for Primary Care, the authors recommend posterior splinting and non-weightbearing with crutches for acute Jones’ fractures, followed by non-weightbearing short leg casting application at 3 to 5 days from injury.¹⁰ After a minimum of 6 to 8 weeks of casting, they recommend options of 4 additional weeks of casting or internal fixation for clinical or radiographic nonunion.

For tuberosity avulsion fractures, the authors recommend use of a firm-soled shoe for 4 to 8 weeks. For patients with discomfort at an initial 4- to 7-day follow-up, they give an option of using a walking short leg casting for 2 weeks, with follow-up every 2 to 4 weeks until clinical healing.

Evidence summary

Fractures within 1.5 cm of the fifth metatarsal tuberosity, without extension distal to the fourth-fifth intermetatarsal articulation, occurring with less than 2-week symptom prodrome and without a history of previous fracture, are defined as “acute Jones’ fractures” (FIGURE). In a recent RCT by Mologne et al,¹ 37 active-duty military personnel with acute Jones’ fractures were randomized to either 8 weeks of no weight-bearing in a short leg casting, followed by a walking cast or hard-soled shoe until clinical union; or to early outpatient intramedullary screw fixation followed by no weight-bearing for 2 weeks, then weight-bearing as tolerated in a hard-soled shoe until clinical union. Screw fixation significantly reduced both time to clinical union and time to return to sports—by nearly 50% when compared with non-weightbearing short leg casting. Furthermore, at 26 weeks the casting group saw a significant 44% failure rate compared with only 5% in the surgical group (number needed to treat [NNT]=2.6). Six patients in the surgical group had mild discomfort from the screw head, and 3 needed the screw to be removed. Generalization of the results was limited by the mostly male military population.

The rates and times of union with short leg casting vary over a wide range in the research literature. The casting group in the RCT above had union rates of 56% and median time to union of 14.5 weeks (lower and upper quartile range, 10.5–18.5).¹ A prospective registry of 68 consecutive acute Jones’ fractures in primarily young military service members showed a 72% union rate with non-weightbearing short leg casting with average time to union of 21.2 weeks.² A heterogeneous group of 5 retrospective follow-up studies of short leg casting reported wide ranges in union rates of 72% to 100%, and in time to healing of 7 weeks to 21 months.^3-7 These studies varied in average age, sex, and athletic ability of their samples as well as type of immobilization and weight-bearing status during treatment.

Tuberosity avulsion fractures are proximal fifth metatarsal styloid fractures resulting from a forceful pull of the lateral band of the long plantar ligament or the peroneus brevis tendon during ankle inversion. A 12-week RCT in 89 consecutive patients presenting to an emergency department with fifth metatarsal tuberosity avulsion fractures compared a nonrigid, soft Jones’ dressing consisting of alternating layers of cast padding and elastic bandages with a rigid short leg casting.⁸ The Jones’ dressing had a significant 28% reduction in time to return to pre-injury levels of activity. Other outcomes—time in treatment modality, time to radiographic healing, and functional foot score—were not different between intervention groups. Validity was limited by the 32% lost to follow-up rate.

FIGURE
Acute fracture of the fifth metatarsal

Recommendations from others

We were unable to locate any consensus statements or clinical guidelines regarding the treatment of Jones’ fractures.

DeLee and Drez’s Orthopaedic Sports Medicine recommends immobilization in a cast or below-the-knee boot with strict non-weightbearing for at least 6 weeks for acute Jones’ fractures.⁹ It recommends surgical treatment, followed by 6 weeks of cast immobilization, then progression to weight bearing based on radiographic findings, for nonoperative treatment failures or with desire to return high-performance athletes to activity.

In Fracture Management for Primary Care, the authors recommend posterior splinting and non-weightbearing with crutches for acute Jones’ fractures, followed by non-weightbearing short leg casting application at 3 to 5 days from injury.¹⁰ After a minimum of 6 to 8 weeks of casting, they recommend options of 4 additional weeks of casting or internal fixation for clinical or radiographic nonunion.

For tuberosity avulsion fractures, the authors recommend use of a firm-soled shoe for 4 to 8 weeks. For patients with discomfort at an initial 4- to 7-day follow-up, they give an option of using a walking short leg casting for 2 weeks, with follow-up every 2 to 4 weeks until clinical healing.

Evidence summary

Tramadol is a novel, central-acting synthetic opioid with weak mu-opioid activity, and is approved for treatment of moderate to moderately severe pain in adults. Anecdotally, some clinicians have assumed this popular analgesic’s nonscheduled status under the Controlled Substance Act (CSA) means tramadol has no substance abuse potential. (The term “abuse” herein denotes substance abuse or dependence.)

Evidence of tramadol abuse in the US comes primarily from federally operated programs collecting adverse drug event (ADE) data. The MedWatch program of the Food and Drug Administration (FDA) provides a central depository for receiving and compiling postmarketing voluntary case reports. While passive reporting systems can significantly underestimate serious ADE numbers, these reports are often the first evidence of an ADE after a new drug’s release into the market.¹ MedWatch has received 766 case reports of abuse associated with tramadol, as well as 482 cases of withdrawal associated with tramadol from the drug’s initial US marketing in 1995 through September 2004.^2,3

The Drug Abuse Warning Network (DAWN) is a federally operated, national surveillance system that monitors trends in drug-related emergency department visits. Over the period from 1995 to 2002, DAWN reported drug-related emergency department visits mentioning tramadol in more than 12,000 cases. Tramadol case numbers significantly increased 165% during this time. For perspective, during the same period, DAWN found nalbuphine (Nubain, also not CSA scheduled) in 118 cases, propoxyphene drug combinations (CSA Class IV) in more than 45,000 cases, codeine drug combinations (CSA Classes III & V) in about 50,000 cases, and hydrocodone drug combinations (CSA Class III) in around 128,000 cases.⁴

Using data from observational postmarketing studies, investigators have extrapolated a tramadol abuse rate for the general tramadolexposed population.^5,6 Ortho-McNeil, Ultram’s manufacturer, funded a surveillance program that compiled tramadol abuse and withdrawal case reports from 2 sources: (1) periodic surveys for tramadol abuse case reports from a group of 255 substance abuse experts studying and caring for addiction communities, and (2) voluntary ADE case reports from health care professionals and consumers received by Ortho-McNeil. Over 3 years of surveillance, the program received 454 case reports classified as tramadol abuse. Over 5 years of surveillance, 422 cases of substance withdrawal, with primarily opioid withdrawal symptoms, were reported. There are significant threats to the validity and generalizability of the investigators’ estimated abuse rate of 1 to 3 cases per 100,000 tramadol-exposed patients. The abuse cases were collected in nonrepresentative samples of the tramadol-exposed population. Tramadol exposure is likely suppressed in addiction communities with access to preferred, more potent or euphoriant opioids than tramadol. Voluntary case reports of tramadol abuse significantly underestimate the actual number of abuse cases in the tramadol-exposed population. In addition, the low survey return rate (49%) further decreases the accuracy of any estimation of tramadol abuse rates.

Prospective studies among patients with known abuse, or at high risk of abuse, reported a tramadol abuse rate, as well as subjective experiences of tramadol withdrawal. A 3-year post-marketing cohort study measured tramadol’s nonmedical misuse rates using urine drug testing for tramadol among 1601 participants in 4 US state monitoring programs for impaired healthcare professionals.⁷ Tramadol exposure occurred in 140 (8.7%) participants. Thirty-nine (28%) were classified as extensive experimentation or abuse of tramadol. Overall, the rate of extensive experimentation or abuse was 18 cases per thousand personyears. The Hawthorne effect, where awareness of being monitored alters a subject’s behavior, may threaten these measured frequency rates’ generalizability. Another prospective study assessed the subjective tramadol withdrawal experience in 219 patients with a diagnosis of “Tramadol misuse” who were attending 6 drug detoxification centers in China.⁸ Validated drug dependence symptom scales found that while the degree of physical dependence reported was uniformly mild, the majority of patients reported the psychic dependence symptom of tramadol craving.

The FDA’s Drug Abuse Advisory Committee performed a formal review of the tramadol abuse evidence in 1998, including the data from OrthoMcNeil’s surveillance studies and federal case reporting/surveillance programs. The FDA did not recommend changing tramadol’s unscheduled status.⁹ The FDA’s considered decision to not schedule tramadol as a controlled substance implies its abuse risk to the general population is low. in comparison to its novel analgesic benefit.

Recommendations from others

Ortho-McNeil’s revised 2001 product package insert for Ultram states, “Tramadol may induce psychic and physical dependence of the morphine type (mu-opioid). Dependence and abuse, including drug-seeking behavior and taking illicit actions to obtain the drug are not limited to those patients with prior history of opioid dependence.” (italics in original, emphasizing 2001 addition). The risk for patients with a history of substance abuse has been observed to be higher.¹⁰

Evidence summary

Tramadol is a novel, central-acting synthetic opioid with weak mu-opioid activity, and is approved for treatment of moderate to moderately severe pain in adults. Anecdotally, some clinicians have assumed this popular analgesic’s nonscheduled status under the Controlled Substance Act (CSA) means tramadol has no substance abuse potential. (The term “abuse” herein denotes substance abuse or dependence.)

Evidence of tramadol abuse in the US comes primarily from federally operated programs collecting adverse drug event (ADE) data. The MedWatch program of the Food and Drug Administration (FDA) provides a central depository for receiving and compiling postmarketing voluntary case reports. While passive reporting systems can significantly underestimate serious ADE numbers, these reports are often the first evidence of an ADE after a new drug’s release into the market.¹ MedWatch has received 766 case reports of abuse associated with tramadol, as well as 482 cases of withdrawal associated with tramadol from the drug’s initial US marketing in 1995 through September 2004.^2,3

The Drug Abuse Warning Network (DAWN) is a federally operated, national surveillance system that monitors trends in drug-related emergency department visits. Over the period from 1995 to 2002, DAWN reported drug-related emergency department visits mentioning tramadol in more than 12,000 cases. Tramadol case numbers significantly increased 165% during this time. For perspective, during the same period, DAWN found nalbuphine (Nubain, also not CSA scheduled) in 118 cases, propoxyphene drug combinations (CSA Class IV) in more than 45,000 cases, codeine drug combinations (CSA Classes III & V) in about 50,000 cases, and hydrocodone drug combinations (CSA Class III) in around 128,000 cases.⁴

Using data from observational postmarketing studies, investigators have extrapolated a tramadol abuse rate for the general tramadolexposed population.^5,6 Ortho-McNeil, Ultram’s manufacturer, funded a surveillance program that compiled tramadol abuse and withdrawal case reports from 2 sources: (1) periodic surveys for tramadol abuse case reports from a group of 255 substance abuse experts studying and caring for addiction communities, and (2) voluntary ADE case reports from health care professionals and consumers received by Ortho-McNeil. Over 3 years of surveillance, the program received 454 case reports classified as tramadol abuse. Over 5 years of surveillance, 422 cases of substance withdrawal, with primarily opioid withdrawal symptoms, were reported. There are significant threats to the validity and generalizability of the investigators’ estimated abuse rate of 1 to 3 cases per 100,000 tramadol-exposed patients. The abuse cases were collected in nonrepresentative samples of the tramadol-exposed population. Tramadol exposure is likely suppressed in addiction communities with access to preferred, more potent or euphoriant opioids than tramadol. Voluntary case reports of tramadol abuse significantly underestimate the actual number of abuse cases in the tramadol-exposed population. In addition, the low survey return rate (49%) further decreases the accuracy of any estimation of tramadol abuse rates.

Prospective studies among patients with known abuse, or at high risk of abuse, reported a tramadol abuse rate, as well as subjective experiences of tramadol withdrawal. A 3-year post-marketing cohort study measured tramadol’s nonmedical misuse rates using urine drug testing for tramadol among 1601 participants in 4 US state monitoring programs for impaired healthcare professionals.⁷ Tramadol exposure occurred in 140 (8.7%) participants. Thirty-nine (28%) were classified as extensive experimentation or abuse of tramadol. Overall, the rate of extensive experimentation or abuse was 18 cases per thousand personyears. The Hawthorne effect, where awareness of being monitored alters a subject’s behavior, may threaten these measured frequency rates’ generalizability. Another prospective study assessed the subjective tramadol withdrawal experience in 219 patients with a diagnosis of “Tramadol misuse” who were attending 6 drug detoxification centers in China.⁸ Validated drug dependence symptom scales found that while the degree of physical dependence reported was uniformly mild, the majority of patients reported the psychic dependence symptom of tramadol craving.

The FDA’s Drug Abuse Advisory Committee performed a formal review of the tramadol abuse evidence in 1998, including the data from OrthoMcNeil’s surveillance studies and federal case reporting/surveillance programs. The FDA did not recommend changing tramadol’s unscheduled status.⁹ The FDA’s considered decision to not schedule tramadol as a controlled substance implies its abuse risk to the general population is low. in comparison to its novel analgesic benefit.

Recommendations from others

Ortho-McNeil’s revised 2001 product package insert for Ultram states, “Tramadol may induce psychic and physical dependence of the morphine type (mu-opioid). Dependence and abuse, including drug-seeking behavior and taking illicit actions to obtain the drug are not limited to those patients with prior history of opioid dependence.” (italics in original, emphasizing 2001 addition). The risk for patients with a history of substance abuse has been observed to be higher.¹⁰

Evidence summary

Tramadol is a novel, central-acting synthetic opioid with weak mu-opioid activity, and is approved for treatment of moderate to moderately severe pain in adults. Anecdotally, some clinicians have assumed this popular analgesic’s nonscheduled status under the Controlled Substance Act (CSA) means tramadol has no substance abuse potential. (The term “abuse” herein denotes substance abuse or dependence.)

Evidence of tramadol abuse in the US comes primarily from federally operated programs collecting adverse drug event (ADE) data. The MedWatch program of the Food and Drug Administration (FDA) provides a central depository for receiving and compiling postmarketing voluntary case reports. While passive reporting systems can significantly underestimate serious ADE numbers, these reports are often the first evidence of an ADE after a new drug’s release into the market.¹ MedWatch has received 766 case reports of abuse associated with tramadol, as well as 482 cases of withdrawal associated with tramadol from the drug’s initial US marketing in 1995 through September 2004.^2,3

The Drug Abuse Warning Network (DAWN) is a federally operated, national surveillance system that monitors trends in drug-related emergency department visits. Over the period from 1995 to 2002, DAWN reported drug-related emergency department visits mentioning tramadol in more than 12,000 cases. Tramadol case numbers significantly increased 165% during this time. For perspective, during the same period, DAWN found nalbuphine (Nubain, also not CSA scheduled) in 118 cases, propoxyphene drug combinations (CSA Class IV) in more than 45,000 cases, codeine drug combinations (CSA Classes III & V) in about 50,000 cases, and hydrocodone drug combinations (CSA Class III) in around 128,000 cases.⁴

Using data from observational postmarketing studies, investigators have extrapolated a tramadol abuse rate for the general tramadolexposed population.^5,6 Ortho-McNeil, Ultram’s manufacturer, funded a surveillance program that compiled tramadol abuse and withdrawal case reports from 2 sources: (1) periodic surveys for tramadol abuse case reports from a group of 255 substance abuse experts studying and caring for addiction communities, and (2) voluntary ADE case reports from health care professionals and consumers received by Ortho-McNeil. Over 3 years of surveillance, the program received 454 case reports classified as tramadol abuse. Over 5 years of surveillance, 422 cases of substance withdrawal, with primarily opioid withdrawal symptoms, were reported. There are significant threats to the validity and generalizability of the investigators’ estimated abuse rate of 1 to 3 cases per 100,000 tramadol-exposed patients. The abuse cases were collected in nonrepresentative samples of the tramadol-exposed population. Tramadol exposure is likely suppressed in addiction communities with access to preferred, more potent or euphoriant opioids than tramadol. Voluntary case reports of tramadol abuse significantly underestimate the actual number of abuse cases in the tramadol-exposed population. In addition, the low survey return rate (49%) further decreases the accuracy of any estimation of tramadol abuse rates.

Prospective studies among patients with known abuse, or at high risk of abuse, reported a tramadol abuse rate, as well as subjective experiences of tramadol withdrawal. A 3-year post-marketing cohort study measured tramadol’s nonmedical misuse rates using urine drug testing for tramadol among 1601 participants in 4 US state monitoring programs for impaired healthcare professionals.⁷ Tramadol exposure occurred in 140 (8.7%) participants. Thirty-nine (28%) were classified as extensive experimentation or abuse of tramadol. Overall, the rate of extensive experimentation or abuse was 18 cases per thousand personyears. The Hawthorne effect, where awareness of being monitored alters a subject’s behavior, may threaten these measured frequency rates’ generalizability. Another prospective study assessed the subjective tramadol withdrawal experience in 219 patients with a diagnosis of “Tramadol misuse” who were attending 6 drug detoxification centers in China.⁸ Validated drug dependence symptom scales found that while the degree of physical dependence reported was uniformly mild, the majority of patients reported the psychic dependence symptom of tramadol craving.

The FDA’s Drug Abuse Advisory Committee performed a formal review of the tramadol abuse evidence in 1998, including the data from OrthoMcNeil’s surveillance studies and federal case reporting/surveillance programs. The FDA did not recommend changing tramadol’s unscheduled status.⁹ The FDA’s considered decision to not schedule tramadol as a controlled substance implies its abuse risk to the general population is low. in comparison to its novel analgesic benefit.

Recommendations from others

Ortho-McNeil’s revised 2001 product package insert for Ultram states, “Tramadol may induce psychic and physical dependence of the morphine type (mu-opioid). Dependence and abuse, including drug-seeking behavior and taking illicit actions to obtain the drug are not limited to those patients with prior history of opioid dependence.” (italics in original, emphasizing 2001 addition). The risk for patients with a history of substance abuse has been observed to be higher.¹⁰

Evidence summary

Transient ischemic attack (Figure) is a temporary, focal brain or retinal deficit caused by vascular disease that clears completely in less than 24 hours.¹ A large prospective cohort study recently estimated the risk of stroke after a TIA or minor stroke to be 8% to 12% at 7 days and 11% to 15% at 1 month.²

In a large retrospective cohort study, 5% of TIA patients returned to the emergency department with a stroke within the first 2 days after TIA.³ Another 6% returned with a stroke within 90 days. Five independent risk factors were identified: age >60 years, diabetes mellitus, duration of TIA longer than 10 minutes, signs or symptoms of weakness, and speech impairment. Thirty-four percent of patients with all 5 risk factors, and none of the patients without any risk factors, had a stroke within 90 days. Of note, 13% of the TIA patients had an arrhythmia, congestive heart failure, unstable angina, myocardial infarction, stroke, or recurrent TIA within 4 days of initial presenting with a TIA. Twenty-five percent of the patients experienced 1 of these cardiovascular events during the 3 months of follow-up.

In a retrospective case review of TIA and stroke patients, the hospital admissions of 4 of 21 TIA patients were retrospectively categorized as medically justified.⁴ Admission was categorized as medically justified if the patient had 1 or more of the following criteria: another diagnosis that warranted admission, inadequate home situation, altered mental status, an adverse event during hospitalization including worsening of the deficit, and if the patient underwent some hospital-based treatment that could not be provided on an out-patient basis. Ease and rapidity of evaluation was not considered medically justifiable and outcome improvement (stroke prevention) was not studied.

Two retrospective chart reviews of TIA found considerable practice variability in the evaluation of TIA patient. In 1 study of TIA patients presenting to an emergency department, 81% had a computed tomography scan, 75% had electrocardiogram, and 74% had a complete blood count.⁵ Carotid Doppler imaging was performed in the emergency department in 16%, and 26% were referred for outpatient Doppler studies. One percent had an ECG in the emergency department, and 16% were given ECGs as outpatients. Seventy-five percent of patients were discharged home. Those hospitalized had a median length of stay of 1 day. In the second study, 31% of the TIA patients had no diagnostic studies performed during the first month after presenting to their primary care physician.⁶

FIGURE
Expeditious evaluation of TIA is imperative

Recommendations from others

The American Heart Association (AHA) recommends that physicians use a stepwise approach to TIA evaluation as outlined in the Table. The AHA also recommends that the diagnostic evaluation of patients seen within 7 days of a TIA should be completed within 1 week or less. The AHA leaves the decision whether to hospitalize a patient up to the physician based on a patient’s circumstances. The goals of diagnostic testing are to identify or exclude causes of TIA requiring specific therapy, to assess modifiable risk factors, and to determine prognosis.⁷

The National Stroke Association recommends that patients with known high-grade stenosis in a vascular territory appropriate to the symptoms, and patients with recurrent symptoms, undergo urgent evaluation. Evaluation includes imaging and ruling out other causes of TIA. Patients should be admitted to the hospital if imaging is not immediately available. If indicated, carotid endarterectomy should be performed without delay.⁸

TABLE
Stepwise diagnostic evaluation for patients with transient ischemic attack

Evidence summary

Transient ischemic attack (Figure) is a temporary, focal brain or retinal deficit caused by vascular disease that clears completely in less than 24 hours.¹ A large prospective cohort study recently estimated the risk of stroke after a TIA or minor stroke to be 8% to 12% at 7 days and 11% to 15% at 1 month.²

In a large retrospective cohort study, 5% of TIA patients returned to the emergency department with a stroke within the first 2 days after TIA.³ Another 6% returned with a stroke within 90 days. Five independent risk factors were identified: age >60 years, diabetes mellitus, duration of TIA longer than 10 minutes, signs or symptoms of weakness, and speech impairment. Thirty-four percent of patients with all 5 risk factors, and none of the patients without any risk factors, had a stroke within 90 days. Of note, 13% of the TIA patients had an arrhythmia, congestive heart failure, unstable angina, myocardial infarction, stroke, or recurrent TIA within 4 days of initial presenting with a TIA. Twenty-five percent of the patients experienced 1 of these cardiovascular events during the 3 months of follow-up.

In a retrospective case review of TIA and stroke patients, the hospital admissions of 4 of 21 TIA patients were retrospectively categorized as medically justified.⁴ Admission was categorized as medically justified if the patient had 1 or more of the following criteria: another diagnosis that warranted admission, inadequate home situation, altered mental status, an adverse event during hospitalization including worsening of the deficit, and if the patient underwent some hospital-based treatment that could not be provided on an out-patient basis. Ease and rapidity of evaluation was not considered medically justifiable and outcome improvement (stroke prevention) was not studied.

Two retrospective chart reviews of TIA found considerable practice variability in the evaluation of TIA patient. In 1 study of TIA patients presenting to an emergency department, 81% had a computed tomography scan, 75% had electrocardiogram, and 74% had a complete blood count.⁵ Carotid Doppler imaging was performed in the emergency department in 16%, and 26% were referred for outpatient Doppler studies. One percent had an ECG in the emergency department, and 16% were given ECGs as outpatients. Seventy-five percent of patients were discharged home. Those hospitalized had a median length of stay of 1 day. In the second study, 31% of the TIA patients had no diagnostic studies performed during the first month after presenting to their primary care physician.⁶

FIGURE
Expeditious evaluation of TIA is imperative

Recommendations from others

The American Heart Association (AHA) recommends that physicians use a stepwise approach to TIA evaluation as outlined in the Table. The AHA also recommends that the diagnostic evaluation of patients seen within 7 days of a TIA should be completed within 1 week or less. The AHA leaves the decision whether to hospitalize a patient up to the physician based on a patient’s circumstances. The goals of diagnostic testing are to identify or exclude causes of TIA requiring specific therapy, to assess modifiable risk factors, and to determine prognosis.⁷

The National Stroke Association recommends that patients with known high-grade stenosis in a vascular territory appropriate to the symptoms, and patients with recurrent symptoms, undergo urgent evaluation. Evaluation includes imaging and ruling out other causes of TIA. Patients should be admitted to the hospital if imaging is not immediately available. If indicated, carotid endarterectomy should be performed without delay.⁸

TABLE
Stepwise diagnostic evaluation for patients with transient ischemic attack

Evidence summary

Transient ischemic attack (Figure) is a temporary, focal brain or retinal deficit caused by vascular disease that clears completely in less than 24 hours.¹ A large prospective cohort study recently estimated the risk of stroke after a TIA or minor stroke to be 8% to 12% at 7 days and 11% to 15% at 1 month.²

In a large retrospective cohort study, 5% of TIA patients returned to the emergency department with a stroke within the first 2 days after TIA.³ Another 6% returned with a stroke within 90 days. Five independent risk factors were identified: age >60 years, diabetes mellitus, duration of TIA longer than 10 minutes, signs or symptoms of weakness, and speech impairment. Thirty-four percent of patients with all 5 risk factors, and none of the patients without any risk factors, had a stroke within 90 days. Of note, 13% of the TIA patients had an arrhythmia, congestive heart failure, unstable angina, myocardial infarction, stroke, or recurrent TIA within 4 days of initial presenting with a TIA. Twenty-five percent of the patients experienced 1 of these cardiovascular events during the 3 months of follow-up.

In a retrospective case review of TIA and stroke patients, the hospital admissions of 4 of 21 TIA patients were retrospectively categorized as medically justified.⁴ Admission was categorized as medically justified if the patient had 1 or more of the following criteria: another diagnosis that warranted admission, inadequate home situation, altered mental status, an adverse event during hospitalization including worsening of the deficit, and if the patient underwent some hospital-based treatment that could not be provided on an out-patient basis. Ease and rapidity of evaluation was not considered medically justifiable and outcome improvement (stroke prevention) was not studied.

Two retrospective chart reviews of TIA found considerable practice variability in the evaluation of TIA patient. In 1 study of TIA patients presenting to an emergency department, 81% had a computed tomography scan, 75% had electrocardiogram, and 74% had a complete blood count.⁵ Carotid Doppler imaging was performed in the emergency department in 16%, and 26% were referred for outpatient Doppler studies. One percent had an ECG in the emergency department, and 16% were given ECGs as outpatients. Seventy-five percent of patients were discharged home. Those hospitalized had a median length of stay of 1 day. In the second study, 31% of the TIA patients had no diagnostic studies performed during the first month after presenting to their primary care physician.⁶

FIGURE
Expeditious evaluation of TIA is imperative

Recommendations from others

The American Heart Association (AHA) recommends that physicians use a stepwise approach to TIA evaluation as outlined in the Table. The AHA also recommends that the diagnostic evaluation of patients seen within 7 days of a TIA should be completed within 1 week or less. The AHA leaves the decision whether to hospitalize a patient up to the physician based on a patient’s circumstances. The goals of diagnostic testing are to identify or exclude causes of TIA requiring specific therapy, to assess modifiable risk factors, and to determine prognosis.⁷

The National Stroke Association recommends that patients with known high-grade stenosis in a vascular territory appropriate to the symptoms, and patients with recurrent symptoms, undergo urgent evaluation. Evaluation includes imaging and ruling out other causes of TIA. Patients should be admitted to the hospital if imaging is not immediately available. If indicated, carotid endarterectomy should be performed without delay.⁸

TABLE
Stepwise diagnostic evaluation for patients with transient ischemic attack

Evidence summary

Three systematic reviews addressing the efficacy of systemic corticosteroids in managing acute exacerbations of COPD found consistent, good-quality evidence supporting short courses of systemic steroids. The improvement in outcomes included decreases in airflow obstruction, treatment failure, and length of hospital stay.^1-3

The optimal initial doses of systemic corticosteroids to achieve these benefits are uncertain. Variable study designs limit combining study results into a dose-response curve, and there are no comparative trials of high- vs low-dose regimens. A panel consensus judgment from a collaboration of the National Heart, Lung, and Blood Institute and the World Health Organization recommended 30–40 mg of oral prednisone daily for 10 to 14 days.⁴

A Cochrane systematic review analyzed 7 randomized, placebo-controlled trials of systemic steroids for acute exacerbations of COPD.¹ While most of the studies reporting symptom outcomes used disparate methods of measurement, 2 small studies^5,6 reported changes in quality of life using validated visual analogue scales. This allowed their results to be combined into a summary estimate of the effect of corticosteroids compared with placebo. Combining the visual analogue scales using a standardized mean difference showed a significant improvement of this summary quality of life measure in the steroid-treated group.

Other small randomized controlled trials of systemic steroids¹ demonstrated trends towards improvement in symptom outcomes. A Taiwanese study randomized 138 patients presenting to an emergency department to treatment with 100 mg intravenous hydrocortisone or placebo within 15 minutes of arrival.¹ Using a 6-point scale, patients gave self-assessments of the severity of their attack on arrival and at 6 hours. Compared with placebo, the steroid group showed a 6-hour improvement of uncertain significance.

Similarly, a British trial of 30 mg prednisone vs placebo in 56 inpatients with acute exacerbations of COPD measured a daily composite symptom score based on 7 pulmonary and functional symptoms.⁸ There was a nonsignificant trend towards greater improvement in the steroid-treated group.

Finally, a multicenter, 3-armed, placebo-controlled, double-blinded, parallel design study enrolled 199 COPD inpatients, who were randomized to oral prednisone, inhaled budesonide, or placebo treatment groups.⁹ Dyspnea was assessed using a validated, modified Borg scale every 12 hours for 72 hours. The reduction in the modified Borg scale rating was of comparable magnitude in the 3 groups, but again there was a nonsignificant greater reduction in the systemic steroid group compared with both the placebo and inhaled budesonide groups. Power calculations were not provided, so it is unclear whether sample size in this study was sufficient to detect important differences in outcomes.

Three randomized controlled trials prospectively measured adverse events rates of systemic steroids in acute exacerbations of COPD.^9-11 Hyperglycemia or glycosuria was more common in the steroid-treated groups. The SCCOPE study, the largest of the 3 trials, found hyperglycemia requiring treatment occurred in a greater proportion of the steroid-treated group than placebo (15% vs 4%; P=.002; number needed to harm=9).

Recommendations from others

A recent review provides a concise summary of practice guidelines for the management of acute exacerbations of COPD from widely recognized professional societies.¹² Systemic steroids are endorsed in the evidence-based systematic review guidelines from the American College of Chest Physicians–American Society of Internal Medicine, along with the National Heart, Lung, and Blood Institute with the World Health Organization cosponsored Global Initiative for Chronic Obstructive Lung Disease (GOLD), and the consensus guidelines of the American Thoracic Society.¹³

Evidence summary

Three systematic reviews addressing the efficacy of systemic corticosteroids in managing acute exacerbations of COPD found consistent, good-quality evidence supporting short courses of systemic steroids. The improvement in outcomes included decreases in airflow obstruction, treatment failure, and length of hospital stay.^1-3

The optimal initial doses of systemic corticosteroids to achieve these benefits are uncertain. Variable study designs limit combining study results into a dose-response curve, and there are no comparative trials of high- vs low-dose regimens. A panel consensus judgment from a collaboration of the National Heart, Lung, and Blood Institute and the World Health Organization recommended 30–40 mg of oral prednisone daily for 10 to 14 days.⁴

A Cochrane systematic review analyzed 7 randomized, placebo-controlled trials of systemic steroids for acute exacerbations of COPD.¹ While most of the studies reporting symptom outcomes used disparate methods of measurement, 2 small studies^5,6 reported changes in quality of life using validated visual analogue scales. This allowed their results to be combined into a summary estimate of the effect of corticosteroids compared with placebo. Combining the visual analogue scales using a standardized mean difference showed a significant improvement of this summary quality of life measure in the steroid-treated group.

Other small randomized controlled trials of systemic steroids¹ demonstrated trends towards improvement in symptom outcomes. A Taiwanese study randomized 138 patients presenting to an emergency department to treatment with 100 mg intravenous hydrocortisone or placebo within 15 minutes of arrival.¹ Using a 6-point scale, patients gave self-assessments of the severity of their attack on arrival and at 6 hours. Compared with placebo, the steroid group showed a 6-hour improvement of uncertain significance.

Similarly, a British trial of 30 mg prednisone vs placebo in 56 inpatients with acute exacerbations of COPD measured a daily composite symptom score based on 7 pulmonary and functional symptoms.⁸ There was a nonsignificant trend towards greater improvement in the steroid-treated group.

Finally, a multicenter, 3-armed, placebo-controlled, double-blinded, parallel design study enrolled 199 COPD inpatients, who were randomized to oral prednisone, inhaled budesonide, or placebo treatment groups.⁹ Dyspnea was assessed using a validated, modified Borg scale every 12 hours for 72 hours. The reduction in the modified Borg scale rating was of comparable magnitude in the 3 groups, but again there was a nonsignificant greater reduction in the systemic steroid group compared with both the placebo and inhaled budesonide groups. Power calculations were not provided, so it is unclear whether sample size in this study was sufficient to detect important differences in outcomes.

Three randomized controlled trials prospectively measured adverse events rates of systemic steroids in acute exacerbations of COPD.^9-11 Hyperglycemia or glycosuria was more common in the steroid-treated groups. The SCCOPE study, the largest of the 3 trials, found hyperglycemia requiring treatment occurred in a greater proportion of the steroid-treated group than placebo (15% vs 4%; P=.002; number needed to harm=9).

Recommendations from others

A recent review provides a concise summary of practice guidelines for the management of acute exacerbations of COPD from widely recognized professional societies.¹² Systemic steroids are endorsed in the evidence-based systematic review guidelines from the American College of Chest Physicians–American Society of Internal Medicine, along with the National Heart, Lung, and Blood Institute with the World Health Organization cosponsored Global Initiative for Chronic Obstructive Lung Disease (GOLD), and the consensus guidelines of the American Thoracic Society.¹³

Evidence summary

Three systematic reviews addressing the efficacy of systemic corticosteroids in managing acute exacerbations of COPD found consistent, good-quality evidence supporting short courses of systemic steroids. The improvement in outcomes included decreases in airflow obstruction, treatment failure, and length of hospital stay.^1-3

The optimal initial doses of systemic corticosteroids to achieve these benefits are uncertain. Variable study designs limit combining study results into a dose-response curve, and there are no comparative trials of high- vs low-dose regimens. A panel consensus judgment from a collaboration of the National Heart, Lung, and Blood Institute and the World Health Organization recommended 30–40 mg of oral prednisone daily for 10 to 14 days.⁴

A Cochrane systematic review analyzed 7 randomized, placebo-controlled trials of systemic steroids for acute exacerbations of COPD.¹ While most of the studies reporting symptom outcomes used disparate methods of measurement, 2 small studies^5,6 reported changes in quality of life using validated visual analogue scales. This allowed their results to be combined into a summary estimate of the effect of corticosteroids compared with placebo. Combining the visual analogue scales using a standardized mean difference showed a significant improvement of this summary quality of life measure in the steroid-treated group.

Other small randomized controlled trials of systemic steroids¹ demonstrated trends towards improvement in symptom outcomes. A Taiwanese study randomized 138 patients presenting to an emergency department to treatment with 100 mg intravenous hydrocortisone or placebo within 15 minutes of arrival.¹ Using a 6-point scale, patients gave self-assessments of the severity of their attack on arrival and at 6 hours. Compared with placebo, the steroid group showed a 6-hour improvement of uncertain significance.

Similarly, a British trial of 30 mg prednisone vs placebo in 56 inpatients with acute exacerbations of COPD measured a daily composite symptom score based on 7 pulmonary and functional symptoms.⁸ There was a nonsignificant trend towards greater improvement in the steroid-treated group.

Finally, a multicenter, 3-armed, placebo-controlled, double-blinded, parallel design study enrolled 199 COPD inpatients, who were randomized to oral prednisone, inhaled budesonide, or placebo treatment groups.⁹ Dyspnea was assessed using a validated, modified Borg scale every 12 hours for 72 hours. The reduction in the modified Borg scale rating was of comparable magnitude in the 3 groups, but again there was a nonsignificant greater reduction in the systemic steroid group compared with both the placebo and inhaled budesonide groups. Power calculations were not provided, so it is unclear whether sample size in this study was sufficient to detect important differences in outcomes.

Three randomized controlled trials prospectively measured adverse events rates of systemic steroids in acute exacerbations of COPD.^9-11 Hyperglycemia or glycosuria was more common in the steroid-treated groups. The SCCOPE study, the largest of the 3 trials, found hyperglycemia requiring treatment occurred in a greater proportion of the steroid-treated group than placebo (15% vs 4%; P=.002; number needed to harm=9).

Recommendations from others

A recent review provides a concise summary of practice guidelines for the management of acute exacerbations of COPD from widely recognized professional societies.¹² Systemic steroids are endorsed in the evidence-based systematic review guidelines from the American College of Chest Physicians–American Society of Internal Medicine, along with the National Heart, Lung, and Blood Institute with the World Health Organization cosponsored Global Initiative for Chronic Obstructive Lung Disease (GOLD), and the consensus guidelines of the American Thoracic Society.¹³

Evidence summary

Amitryptyline was moderately efficacious in 3 small nonblinded trials.^1,2 The largest and best-designed prospective cohort trial studied 192 children. Of the 146 patients available for the first follow-up visit, 84% noted subjective improvement of symptoms. Headache frequency decreased from 17.1 ± 10.1 to 9.2 ± 10.0 days/month (P<.001).¹

Propranolol, although widely used in children, has conflicting evidence regarding effectiveness. One small randomized controlled trial showed reduced headache frequency in children when compared with placebo.³ However, these results were not duplicated in a larger randomized controlled trial using slightly smaller doses.⁴

A comparative randomized controlled trial with multiple crossovers involving 33 children found that a self-hypnosis placebo decreased mean headache frequency from 13.3 per 3-month interval to 5.8 (P=.045), but found propranolol no different than placebo.⁵ Propranolol was also studied in a 3-armed randomized controlled trial in comparison with flunarizine—a drug likely to be efficacious—and placebo. Both drugs were equally efficacious and superior to placebo according to reviews; however, these results were not published in English and could not be critiqued by this author.²

In 2 small retrospective case studies, valproic acid demonstrated >50% improvement in symptoms in 65%⁶ and 78%⁷ of subjects. A single uncontrolled interventional trial of valproic acid in 10 children showed a significant trend of improvement in frequency (mean of 6 attacks/month to 0.8 attacks/month) and duration (mean 5.5 hours per attack to 1.1 hour).⁸

Two similar vasodilatory calcium channel blockers, flunarizine and nimodipine, have the best evidence as migraine prophylactics in children. Flunarizine was found to be effective in multiple well-designed randomized controlled trials and case series, as well as in multiple comparative trials with other agents.²

In a double-blinded, placebo-controlled randomized controlled trial of 48 children, flunarizine decreased mean headache frequency (3.0 attacks/3 months vs 6.5 [P<.001]).⁹ A repeat randomized controlled trial in 70 children had similar outcomes.¹⁰

Nimodipine, in a single randomized controlled trial with crossover design in 37 children decreased headache frequency from a mean of ~2.7 attacks/month to ~1.9 vs. no change for placebo (P<.05).¹¹ A small, prospective, nonblinded comparative trial found that nimodipine and flunarizine have similar efficacy and are superior to placebo.¹²

Cyproheptadine is widely used in children but is not as effective as amitriptyline and propranolol.² In adults it is not considered a first-line agent due to lack of evidence of efficacy.¹³ Nonsteroidal anti-inflammatory drugs have insufficient data to recommend them as prophylactic medications in children.²

RECOMMENDATIONS FROM OTHERS

Nelson Textbook of Pediatrics recommends propranolol as a first-line agent for prevention.¹⁴

A recent review article15 recommends cyproheptadine as an initial agent in children <10 years of age. This article also has a patient handout discussing nonpharmacologic prophylactics such as regular sleep, exercise, stress reduction, and avoiding certain foods.

UpToDate recommends propranolol, cyproheptadine, valproate, and amitriptyline as prophylactic options based on patient parameters such as age, sex, and comorbid conditions.¹⁶

Evidence summary

Amitryptyline was moderately efficacious in 3 small nonblinded trials.^1,2 The largest and best-designed prospective cohort trial studied 192 children. Of the 146 patients available for the first follow-up visit, 84% noted subjective improvement of symptoms. Headache frequency decreased from 17.1 ± 10.1 to 9.2 ± 10.0 days/month (P<.001).¹

Propranolol, although widely used in children, has conflicting evidence regarding effectiveness. One small randomized controlled trial showed reduced headache frequency in children when compared with placebo.³ However, these results were not duplicated in a larger randomized controlled trial using slightly smaller doses.⁴

A comparative randomized controlled trial with multiple crossovers involving 33 children found that a self-hypnosis placebo decreased mean headache frequency from 13.3 per 3-month interval to 5.8 (P=.045), but found propranolol no different than placebo.⁵ Propranolol was also studied in a 3-armed randomized controlled trial in comparison with flunarizine—a drug likely to be efficacious—and placebo. Both drugs were equally efficacious and superior to placebo according to reviews; however, these results were not published in English and could not be critiqued by this author.²

In 2 small retrospective case studies, valproic acid demonstrated >50% improvement in symptoms in 65%⁶ and 78%⁷ of subjects. A single uncontrolled interventional trial of valproic acid in 10 children showed a significant trend of improvement in frequency (mean of 6 attacks/month to 0.8 attacks/month) and duration (mean 5.5 hours per attack to 1.1 hour).⁸

Two similar vasodilatory calcium channel blockers, flunarizine and nimodipine, have the best evidence as migraine prophylactics in children. Flunarizine was found to be effective in multiple well-designed randomized controlled trials and case series, as well as in multiple comparative trials with other agents.²

In a double-blinded, placebo-controlled randomized controlled trial of 48 children, flunarizine decreased mean headache frequency (3.0 attacks/3 months vs 6.5 [P<.001]).⁹ A repeat randomized controlled trial in 70 children had similar outcomes.¹⁰

Nimodipine, in a single randomized controlled trial with crossover design in 37 children decreased headache frequency from a mean of ~2.7 attacks/month to ~1.9 vs. no change for placebo (P<.05).¹¹ A small, prospective, nonblinded comparative trial found that nimodipine and flunarizine have similar efficacy and are superior to placebo.¹²

Cyproheptadine is widely used in children but is not as effective as amitriptyline and propranolol.² In adults it is not considered a first-line agent due to lack of evidence of efficacy.¹³ Nonsteroidal anti-inflammatory drugs have insufficient data to recommend them as prophylactic medications in children.²

RECOMMENDATIONS FROM OTHERS

Nelson Textbook of Pediatrics recommends propranolol as a first-line agent for prevention.¹⁴

A recent review article15 recommends cyproheptadine as an initial agent in children <10 years of age. This article also has a patient handout discussing nonpharmacologic prophylactics such as regular sleep, exercise, stress reduction, and avoiding certain foods.

UpToDate recommends propranolol, cyproheptadine, valproate, and amitriptyline as prophylactic options based on patient parameters such as age, sex, and comorbid conditions.¹⁶

Evidence summary

Amitryptyline was moderately efficacious in 3 small nonblinded trials.^1,2 The largest and best-designed prospective cohort trial studied 192 children. Of the 146 patients available for the first follow-up visit, 84% noted subjective improvement of symptoms. Headache frequency decreased from 17.1 ± 10.1 to 9.2 ± 10.0 days/month (P<.001).¹

Propranolol, although widely used in children, has conflicting evidence regarding effectiveness. One small randomized controlled trial showed reduced headache frequency in children when compared with placebo.³ However, these results were not duplicated in a larger randomized controlled trial using slightly smaller doses.⁴

A comparative randomized controlled trial with multiple crossovers involving 33 children found that a self-hypnosis placebo decreased mean headache frequency from 13.3 per 3-month interval to 5.8 (P=.045), but found propranolol no different than placebo.⁵ Propranolol was also studied in a 3-armed randomized controlled trial in comparison with flunarizine—a drug likely to be efficacious—and placebo. Both drugs were equally efficacious and superior to placebo according to reviews; however, these results were not published in English and could not be critiqued by this author.²

In 2 small retrospective case studies, valproic acid demonstrated >50% improvement in symptoms in 65%⁶ and 78%⁷ of subjects. A single uncontrolled interventional trial of valproic acid in 10 children showed a significant trend of improvement in frequency (mean of 6 attacks/month to 0.8 attacks/month) and duration (mean 5.5 hours per attack to 1.1 hour).⁸

Two similar vasodilatory calcium channel blockers, flunarizine and nimodipine, have the best evidence as migraine prophylactics in children. Flunarizine was found to be effective in multiple well-designed randomized controlled trials and case series, as well as in multiple comparative trials with other agents.²

In a double-blinded, placebo-controlled randomized controlled trial of 48 children, flunarizine decreased mean headache frequency (3.0 attacks/3 months vs 6.5 [P<.001]).⁹ A repeat randomized controlled trial in 70 children had similar outcomes.¹⁰

Nimodipine, in a single randomized controlled trial with crossover design in 37 children decreased headache frequency from a mean of ~2.7 attacks/month to ~1.9 vs. no change for placebo (P<.05).¹¹ A small, prospective, nonblinded comparative trial found that nimodipine and flunarizine have similar efficacy and are superior to placebo.¹²

Cyproheptadine is widely used in children but is not as effective as amitriptyline and propranolol.² In adults it is not considered a first-line agent due to lack of evidence of efficacy.¹³ Nonsteroidal anti-inflammatory drugs have insufficient data to recommend them as prophylactic medications in children.²

RECOMMENDATIONS FROM OTHERS

Nelson Textbook of Pediatrics recommends propranolol as a first-line agent for prevention.¹⁴

A recent review article15 recommends cyproheptadine as an initial agent in children <10 years of age. This article also has a patient handout discussing nonpharmacologic prophylactics such as regular sleep, exercise, stress reduction, and avoiding certain foods.

UpToDate recommends propranolol, cyproheptadine, valproate, and amitriptyline as prophylactic options based on patient parameters such as age, sex, and comorbid conditions.¹⁶

Evidence summary

A systematic review of glucosamine sulfate use for osteoarthritis, based on early research (1956–1991), found that it has anti-inflammatory properties and rebuilds damaged cartilage.¹ These studies evaluated chondrocytes grown in culture and animal models.^1,2 Chondroitin sulfate also stimulates chondrocyte biosynthesis in both animal and in vitro studies. There is insufficient evidence to demonstrate glucosamine sulfate or chondroitin sulfate stimulates chrondrocyte growth in humans with osteoarthritis.^2,3

Joint space narrowing on radiographs suggests progression of osteoarthritis. This narrowing is thought to imply cartilage destruction or loss due to osteoarthritis. A double-blinded randomized controlled trial studied the effect of glucosamine sulfate on tibial-femoral compartment joint space narrowing in 212 patients older than 50 with mild to moderate knee osteoarthritis.⁴ Patients took either 1500 mg/day of glucosamine sulfate or placebo over 3 years. Knee radiographs in a standing anterior-posterior view, using visual and digital analysis, were used to assess joint space narrowing.⁵ The average mean joint space loss was 0.31 mm in the placebo group and 0.07 mm in the treatment group (P<.05; 95% confidence interval, 0.13–0.48).

The clinical relevance of knee joint space narrowing is undetermined. Radiographic evaluation of a weight-bearing joint space may not be an accurate or reproducible technique. A study of 15 patients with mild to moderate knee osteoarthritis used standing and semi-flexed radiographic views after an analgesic and nonsteroid anti-inflammatory drug washout period, and 1 to 12 weeks after resumption of analgesic therapy (mean 6.0 weeks).⁶ Knee pain significantly decreased radiographic joint space in the standing anteriorposterior position, but not in the semiflexed position. Using the standing anterior-posterior method may confound accurate interpretation of joint space narrowing and changes in articular cartilage since glucosamine may have an anti-inflammatory effect.⁶

One double-blinded randomized controlled trial, comparing chondroitin sulfate with placebo, evaluated joint space in patients with symptomatic hand osteoarthritis.⁷ One hundred sixty-five Caucasian patients, aged 40 to 70 years, were randomized to receive either a 50-mg intramuscular injection of chondroitin polysulfate, twice weekly, for 8 weeks, every 4 months, versus placebo, or 400 mg of oral chondroitin sulfate, 3 times a day, versus placebo.

Osteoarthritis progression in the metacarpalphalangeal and interphalangeal joints was assessed with radiographs over 3 years. Evaluators used the Anatomic Lesion Progression Scale to assess the development of osteophytes and joint space narrowing, with or without subchondral bone changes, to determine osteoarthritis progression. This scale makes it very difficult to determine whether improvements are clinically significant.

Chondroitin sulfate and polysulfate did not prevent osteoarthritis from occurring in previously normal joints. In joints already affected, intramuscular chondroitin polysulfate significantly reduced progression of distal interphalangeal, proximal interpharangeal, and metacarpophalangeal joint space narrowing (P<.013), using the progression scale. Oral chondroitin sulfate did not prevent progression.⁷

Recommendations from others

The American College of Rheumatology stated in 2000 that recommending glucosamine sulfate or chondroitin sulfate for osteoarthritis might be premature due to the methodology, lack of standardization, and insufficient information on study designs. More research was recommended.⁸

These products are sold as supplements in the United States. Their purity is often questionable and thus may affect study results. When studying glucosamine, the National Institutes of Health was forced to manufacture the drug itself due to lack of a reliable amount present in commercial products.⁹

Evidence summary

A systematic review of glucosamine sulfate use for osteoarthritis, based on early research (1956–1991), found that it has anti-inflammatory properties and rebuilds damaged cartilage.¹ These studies evaluated chondrocytes grown in culture and animal models.^1,2 Chondroitin sulfate also stimulates chondrocyte biosynthesis in both animal and in vitro studies. There is insufficient evidence to demonstrate glucosamine sulfate or chondroitin sulfate stimulates chrondrocyte growth in humans with osteoarthritis.^2,3

Joint space narrowing on radiographs suggests progression of osteoarthritis. This narrowing is thought to imply cartilage destruction or loss due to osteoarthritis. A double-blinded randomized controlled trial studied the effect of glucosamine sulfate on tibial-femoral compartment joint space narrowing in 212 patients older than 50 with mild to moderate knee osteoarthritis.⁴ Patients took either 1500 mg/day of glucosamine sulfate or placebo over 3 years. Knee radiographs in a standing anterior-posterior view, using visual and digital analysis, were used to assess joint space narrowing.⁵ The average mean joint space loss was 0.31 mm in the placebo group and 0.07 mm in the treatment group (P<.05; 95% confidence interval, 0.13–0.48).

The clinical relevance of knee joint space narrowing is undetermined. Radiographic evaluation of a weight-bearing joint space may not be an accurate or reproducible technique. A study of 15 patients with mild to moderate knee osteoarthritis used standing and semi-flexed radiographic views after an analgesic and nonsteroid anti-inflammatory drug washout period, and 1 to 12 weeks after resumption of analgesic therapy (mean 6.0 weeks).⁶ Knee pain significantly decreased radiographic joint space in the standing anteriorposterior position, but not in the semiflexed position. Using the standing anterior-posterior method may confound accurate interpretation of joint space narrowing and changes in articular cartilage since glucosamine may have an anti-inflammatory effect.⁶

One double-blinded randomized controlled trial, comparing chondroitin sulfate with placebo, evaluated joint space in patients with symptomatic hand osteoarthritis.⁷ One hundred sixty-five Caucasian patients, aged 40 to 70 years, were randomized to receive either a 50-mg intramuscular injection of chondroitin polysulfate, twice weekly, for 8 weeks, every 4 months, versus placebo, or 400 mg of oral chondroitin sulfate, 3 times a day, versus placebo.

Osteoarthritis progression in the metacarpalphalangeal and interphalangeal joints was assessed with radiographs over 3 years. Evaluators used the Anatomic Lesion Progression Scale to assess the development of osteophytes and joint space narrowing, with or without subchondral bone changes, to determine osteoarthritis progression. This scale makes it very difficult to determine whether improvements are clinically significant.

Chondroitin sulfate and polysulfate did not prevent osteoarthritis from occurring in previously normal joints. In joints already affected, intramuscular chondroitin polysulfate significantly reduced progression of distal interphalangeal, proximal interpharangeal, and metacarpophalangeal joint space narrowing (P<.013), using the progression scale. Oral chondroitin sulfate did not prevent progression.⁷

Recommendations from others

The American College of Rheumatology stated in 2000 that recommending glucosamine sulfate or chondroitin sulfate for osteoarthritis might be premature due to the methodology, lack of standardization, and insufficient information on study designs. More research was recommended.⁸

These products are sold as supplements in the United States. Their purity is often questionable and thus may affect study results. When studying glucosamine, the National Institutes of Health was forced to manufacture the drug itself due to lack of a reliable amount present in commercial products.⁹

Evidence summary

A systematic review of glucosamine sulfate use for osteoarthritis, based on early research (1956–1991), found that it has anti-inflammatory properties and rebuilds damaged cartilage.¹ These studies evaluated chondrocytes grown in culture and animal models.^1,2 Chondroitin sulfate also stimulates chondrocyte biosynthesis in both animal and in vitro studies. There is insufficient evidence to demonstrate glucosamine sulfate or chondroitin sulfate stimulates chrondrocyte growth in humans with osteoarthritis.^2,3

Joint space narrowing on radiographs suggests progression of osteoarthritis. This narrowing is thought to imply cartilage destruction or loss due to osteoarthritis. A double-blinded randomized controlled trial studied the effect of glucosamine sulfate on tibial-femoral compartment joint space narrowing in 212 patients older than 50 with mild to moderate knee osteoarthritis.⁴ Patients took either 1500 mg/day of glucosamine sulfate or placebo over 3 years. Knee radiographs in a standing anterior-posterior view, using visual and digital analysis, were used to assess joint space narrowing.⁵ The average mean joint space loss was 0.31 mm in the placebo group and 0.07 mm in the treatment group (P<.05; 95% confidence interval, 0.13–0.48).

The clinical relevance of knee joint space narrowing is undetermined. Radiographic evaluation of a weight-bearing joint space may not be an accurate or reproducible technique. A study of 15 patients with mild to moderate knee osteoarthritis used standing and semi-flexed radiographic views after an analgesic and nonsteroid anti-inflammatory drug washout period, and 1 to 12 weeks after resumption of analgesic therapy (mean 6.0 weeks).⁶ Knee pain significantly decreased radiographic joint space in the standing anteriorposterior position, but not in the semiflexed position. Using the standing anterior-posterior method may confound accurate interpretation of joint space narrowing and changes in articular cartilage since glucosamine may have an anti-inflammatory effect.⁶

One double-blinded randomized controlled trial, comparing chondroitin sulfate with placebo, evaluated joint space in patients with symptomatic hand osteoarthritis.⁷ One hundred sixty-five Caucasian patients, aged 40 to 70 years, were randomized to receive either a 50-mg intramuscular injection of chondroitin polysulfate, twice weekly, for 8 weeks, every 4 months, versus placebo, or 400 mg of oral chondroitin sulfate, 3 times a day, versus placebo.

Osteoarthritis progression in the metacarpalphalangeal and interphalangeal joints was assessed with radiographs over 3 years. Evaluators used the Anatomic Lesion Progression Scale to assess the development of osteophytes and joint space narrowing, with or without subchondral bone changes, to determine osteoarthritis progression. This scale makes it very difficult to determine whether improvements are clinically significant.

Chondroitin sulfate and polysulfate did not prevent osteoarthritis from occurring in previously normal joints. In joints already affected, intramuscular chondroitin polysulfate significantly reduced progression of distal interphalangeal, proximal interpharangeal, and metacarpophalangeal joint space narrowing (P<.013), using the progression scale. Oral chondroitin sulfate did not prevent progression.⁷

Recommendations from others

The American College of Rheumatology stated in 2000 that recommending glucosamine sulfate or chondroitin sulfate for osteoarthritis might be premature due to the methodology, lack of standardization, and insufficient information on study designs. More research was recommended.⁸

These products are sold as supplements in the United States. Their purity is often questionable and thus may affect study results. When studying glucosamine, the National Institutes of Health was forced to manufacture the drug itself due to lack of a reliable amount present in commercial products.⁹

EVIDENCE-BASED ANSWER

Antioxidant supplements of vitamins E and C do not reduce cardiovascular death in people with coronary artery disease. Vitamin E supplementation, in a variety of doses, does not decrease the incidence of cardiovascular or all-cause mortality (grade of recommendation: A, 4 high quality randomized controlled trials [RCTs]). There is no evidence that vitamin C decreases mortality in patients at risk for coronary disease (grade of recommendation: A, meta-analysis of 3 small RCTs). Combination antioxidant regimens (Vitamins E, C, and betacarotene) seem safe, but do not decrease mortality or incidence of major coronary and vascular events (grade recommendation: A, 1 high-quality RCT).

Evidence summary

Four large, well-designed RCTs with a combined enrollment of nearly 25,000 individuals with known coronary artery disease (CAD) or high risk for CAD receiving vitamin E (50–800 IU/d) collectively demonstrated no change in all-cause mortality or incidence of total cardiovascular events.¹ Three of these studies were double-blind, placebo-controlled and the fourth was an open-label design with central randomization and 4 treatment arms.^2-5 Two of the studies did suggest that vitamin E may reduce the incidence of non-fatal myocardial infarctions. One study of 2002 persons receiving 400–800 IU/d showed a statistically significant reduction of non-fatal coronary events (relative risk [RR], 0.62)² In a subgroup analysis of another, 1862 men with history of MI also had reduced risk of non-fatal MI (RR, 0.23).³ However, in both of these groups, the increase in coronary death was not significant.¹

Three small RCTs enrolling a total of 1034 geriatric patients, with follow-up of less than 2 years, evaluated vitamin C (50–200 mg/d) versus placebo and showed no mortality benefit.¹ Meta-analysis of these studies showed a non-significant increase in the relative risk of death (RR, 1.08).⁶

A randomized, placebo-controlled study of simvastatin 40 mg and antioxidants (vitamin E 600 mg, vitamin C 250 mg, beta-carotene 20 mg) enrolled 20,536 adults aged 40 to 80 years with known CAD or high risk for CAD. No significant difference was found in all-cause mortality (RR, 1.04), major coronary events (RR, 1.02), any stroke (RR, 0.99), or any major vascular event (RR, 1.00).⁷ The investigators found no evidence of an adverse affect of the antioxidants on the substantial outcome benefits demonstrated with 40 mg daily of simvastatin. This finding eases some concern from a smaller prior study, which had suggested a negative interaction between simvastatin plus niacin and antioxidant supplementation (composed of vitamins E and C, beta-carotene, and selenium).⁸

Recommendations from others

A 2002 systematic review of antioxidant vitamins (carotene, tocopherol, and ascorbic acid) in primary and secondary prevention of cardiovascular disease concluded simply that “antioxidant vitamins as food supplements cannot be recommended in the primary or secondary prevention against cardiovascular disease.”⁹

The American Heart Association guidelines do not advocate antioxidant vitamin supplements, rather a well-balanced diet “with emphasis on anti-oxidant rich fruits and vegetables and whole grains.”¹⁰

EVIDENCE-BASED ANSWER

Antioxidant supplements of vitamins E and C do not reduce cardiovascular death in people with coronary artery disease. Vitamin E supplementation, in a variety of doses, does not decrease the incidence of cardiovascular or all-cause mortality (grade of recommendation: A, 4 high quality randomized controlled trials [RCTs]). There is no evidence that vitamin C decreases mortality in patients at risk for coronary disease (grade of recommendation: A, meta-analysis of 3 small RCTs). Combination antioxidant regimens (Vitamins E, C, and betacarotene) seem safe, but do not decrease mortality or incidence of major coronary and vascular events (grade recommendation: A, 1 high-quality RCT).

Evidence summary

Four large, well-designed RCTs with a combined enrollment of nearly 25,000 individuals with known coronary artery disease (CAD) or high risk for CAD receiving vitamin E (50–800 IU/d) collectively demonstrated no change in all-cause mortality or incidence of total cardiovascular events.¹ Three of these studies were double-blind, placebo-controlled and the fourth was an open-label design with central randomization and 4 treatment arms.^2-5 Two of the studies did suggest that vitamin E may reduce the incidence of non-fatal myocardial infarctions. One study of 2002 persons receiving 400–800 IU/d showed a statistically significant reduction of non-fatal coronary events (relative risk [RR], 0.62)² In a subgroup analysis of another, 1862 men with history of MI also had reduced risk of non-fatal MI (RR, 0.23).³ However, in both of these groups, the increase in coronary death was not significant.¹

Three small RCTs enrolling a total of 1034 geriatric patients, with follow-up of less than 2 years, evaluated vitamin C (50–200 mg/d) versus placebo and showed no mortality benefit.¹ Meta-analysis of these studies showed a non-significant increase in the relative risk of death (RR, 1.08).⁶

A randomized, placebo-controlled study of simvastatin 40 mg and antioxidants (vitamin E 600 mg, vitamin C 250 mg, beta-carotene 20 mg) enrolled 20,536 adults aged 40 to 80 years with known CAD or high risk for CAD. No significant difference was found in all-cause mortality (RR, 1.04), major coronary events (RR, 1.02), any stroke (RR, 0.99), or any major vascular event (RR, 1.00).⁷ The investigators found no evidence of an adverse affect of the antioxidants on the substantial outcome benefits demonstrated with 40 mg daily of simvastatin. This finding eases some concern from a smaller prior study, which had suggested a negative interaction between simvastatin plus niacin and antioxidant supplementation (composed of vitamins E and C, beta-carotene, and selenium).⁸

Recommendations from others

A 2002 systematic review of antioxidant vitamins (carotene, tocopherol, and ascorbic acid) in primary and secondary prevention of cardiovascular disease concluded simply that “antioxidant vitamins as food supplements cannot be recommended in the primary or secondary prevention against cardiovascular disease.”⁹

The American Heart Association guidelines do not advocate antioxidant vitamin supplements, rather a well-balanced diet “with emphasis on anti-oxidant rich fruits and vegetables and whole grains.”¹⁰

EVIDENCE-BASED ANSWER

Antioxidant supplements of vitamins E and C do not reduce cardiovascular death in people with coronary artery disease. Vitamin E supplementation, in a variety of doses, does not decrease the incidence of cardiovascular or all-cause mortality (grade of recommendation: A, 4 high quality randomized controlled trials [RCTs]). There is no evidence that vitamin C decreases mortality in patients at risk for coronary disease (grade of recommendation: A, meta-analysis of 3 small RCTs). Combination antioxidant regimens (Vitamins E, C, and betacarotene) seem safe, but do not decrease mortality or incidence of major coronary and vascular events (grade recommendation: A, 1 high-quality RCT).

Evidence summary

Four large, well-designed RCTs with a combined enrollment of nearly 25,000 individuals with known coronary artery disease (CAD) or high risk for CAD receiving vitamin E (50–800 IU/d) collectively demonstrated no change in all-cause mortality or incidence of total cardiovascular events.¹ Three of these studies were double-blind, placebo-controlled and the fourth was an open-label design with central randomization and 4 treatment arms.^2-5 Two of the studies did suggest that vitamin E may reduce the incidence of non-fatal myocardial infarctions. One study of 2002 persons receiving 400–800 IU/d showed a statistically significant reduction of non-fatal coronary events (relative risk [RR], 0.62)² In a subgroup analysis of another, 1862 men with history of MI also had reduced risk of non-fatal MI (RR, 0.23).³ However, in both of these groups, the increase in coronary death was not significant.¹

Three small RCTs enrolling a total of 1034 geriatric patients, with follow-up of less than 2 years, evaluated vitamin C (50–200 mg/d) versus placebo and showed no mortality benefit.¹ Meta-analysis of these studies showed a non-significant increase in the relative risk of death (RR, 1.08).⁶

A randomized, placebo-controlled study of simvastatin 40 mg and antioxidants (vitamin E 600 mg, vitamin C 250 mg, beta-carotene 20 mg) enrolled 20,536 adults aged 40 to 80 years with known CAD or high risk for CAD. No significant difference was found in all-cause mortality (RR, 1.04), major coronary events (RR, 1.02), any stroke (RR, 0.99), or any major vascular event (RR, 1.00).⁷ The investigators found no evidence of an adverse affect of the antioxidants on the substantial outcome benefits demonstrated with 40 mg daily of simvastatin. This finding eases some concern from a smaller prior study, which had suggested a negative interaction between simvastatin plus niacin and antioxidant supplementation (composed of vitamins E and C, beta-carotene, and selenium).⁸

Recommendations from others

A 2002 systematic review of antioxidant vitamins (carotene, tocopherol, and ascorbic acid) in primary and secondary prevention of cardiovascular disease concluded simply that “antioxidant vitamins as food supplements cannot be recommended in the primary or secondary prevention against cardiovascular disease.”⁹

The American Heart Association guidelines do not advocate antioxidant vitamin supplements, rather a well-balanced diet “with emphasis on anti-oxidant rich fruits and vegetables and whole grains.”¹⁰

Evidence summary

A randomized, double-blinded, placebo-controlled study in 1496 subjects examined side-effect rates of 3 iron salt formulations using equal dosages of elemental iron (Table).¹ Gastrointestinal (GI) side-effect rates were not significantly different. The side-effect rate in the ferrous sulfate group (23%) was significantly different from that of the placebo group (14%); thus, for every 11 patients treated with ferrous sulfate, 1 patient would have GI side effects attributable to the iron salt (number needed to harm [NNH] = 11).

Two formulations—controlled-release iron preparations and polysaccharide–iron complexes—decrease the amount of iron presented to the proximal GI tract. Three large randomized trials assessed tolerability of controlled-release iron preparations compared with ferrous sulfate.^2–4 The only double-blinded study found a lower rate of nausea and epigastric pain in the controlled-release iron formulation among 1376 blood donors receiving 200 mg/day elemental iron (3.3% vs 6.4%, P < .05, NNH = ~32).² A nonblinded randomized trial of 543 non-anemic adult patients taking 50 mg/day elemental iron also found a lower rate of stomach-related side effects in the controlled-release group (12.2% vs 27.2%, P < .001, NNH = ~7).³ However, none of the 3 studies showed a difference in the discontinuation rates between the 2 iron formulations. Comparative constipation rates among the trials were conflicting.

Two small, nonblinded, randomized trials of polysaccharide–iron complexes reported conflicting results. A study of 159 subjects found fewer subjects discontinuing the polysaccharide–iron complex taken with meals than ferrous sulfate taken on an empty stomach.⁵ A study of 60 subjects taking both preparations on an empty stomach found no difference in side-effect rates.⁶ Two small, randomized, blinded studies found no difference in rates of GI side effects between carbonyl iron and ferrous sulfate.^7,8

TABLE
Representative average wholesale prices* for various iron supplement formulations

Recommendations from others

Wintrobe’s Clinical Hematology⁹ and Williams Hematology¹⁰ recommend ferrous sulfate as the standard oral iron preparation, and assert that claims of improved tolerability of one oral iron preparation over another have not been substantiated.

Clinical Commentary by Andrea Gordon, MD, at http://www.fpin.org.

Evidence summary

A randomized, double-blinded, placebo-controlled study in 1496 subjects examined side-effect rates of 3 iron salt formulations using equal dosages of elemental iron (Table).¹ Gastrointestinal (GI) side-effect rates were not significantly different. The side-effect rate in the ferrous sulfate group (23%) was significantly different from that of the placebo group (14%); thus, for every 11 patients treated with ferrous sulfate, 1 patient would have GI side effects attributable to the iron salt (number needed to harm [NNH] = 11).

Two formulations—controlled-release iron preparations and polysaccharide–iron complexes—decrease the amount of iron presented to the proximal GI tract. Three large randomized trials assessed tolerability of controlled-release iron preparations compared with ferrous sulfate.^2–4 The only double-blinded study found a lower rate of nausea and epigastric pain in the controlled-release iron formulation among 1376 blood donors receiving 200 mg/day elemental iron (3.3% vs 6.4%, P < .05, NNH = ~32).² A nonblinded randomized trial of 543 non-anemic adult patients taking 50 mg/day elemental iron also found a lower rate of stomach-related side effects in the controlled-release group (12.2% vs 27.2%, P < .001, NNH = ~7).³ However, none of the 3 studies showed a difference in the discontinuation rates between the 2 iron formulations. Comparative constipation rates among the trials were conflicting.

Two small, nonblinded, randomized trials of polysaccharide–iron complexes reported conflicting results. A study of 159 subjects found fewer subjects discontinuing the polysaccharide–iron complex taken with meals than ferrous sulfate taken on an empty stomach.⁵ A study of 60 subjects taking both preparations on an empty stomach found no difference in side-effect rates.⁶ Two small, randomized, blinded studies found no difference in rates of GI side effects between carbonyl iron and ferrous sulfate.^7,8

TABLE
Representative average wholesale prices* for various iron supplement formulations

Recommendations from others

Wintrobe’s Clinical Hematology⁹ and Williams Hematology¹⁰ recommend ferrous sulfate as the standard oral iron preparation, and assert that claims of improved tolerability of one oral iron preparation over another have not been substantiated.

Clinical Commentary by Andrea Gordon, MD, at http://www.fpin.org.

Evidence summary

A randomized, double-blinded, placebo-controlled study in 1496 subjects examined side-effect rates of 3 iron salt formulations using equal dosages of elemental iron (Table).¹ Gastrointestinal (GI) side-effect rates were not significantly different. The side-effect rate in the ferrous sulfate group (23%) was significantly different from that of the placebo group (14%); thus, for every 11 patients treated with ferrous sulfate, 1 patient would have GI side effects attributable to the iron salt (number needed to harm [NNH] = 11).

Two formulations—controlled-release iron preparations and polysaccharide–iron complexes—decrease the amount of iron presented to the proximal GI tract. Three large randomized trials assessed tolerability of controlled-release iron preparations compared with ferrous sulfate.^2–4 The only double-blinded study found a lower rate of nausea and epigastric pain in the controlled-release iron formulation among 1376 blood donors receiving 200 mg/day elemental iron (3.3% vs 6.4%, P < .05, NNH = ~32).² A nonblinded randomized trial of 543 non-anemic adult patients taking 50 mg/day elemental iron also found a lower rate of stomach-related side effects in the controlled-release group (12.2% vs 27.2%, P < .001, NNH = ~7).³ However, none of the 3 studies showed a difference in the discontinuation rates between the 2 iron formulations. Comparative constipation rates among the trials were conflicting.

Two small, nonblinded, randomized trials of polysaccharide–iron complexes reported conflicting results. A study of 159 subjects found fewer subjects discontinuing the polysaccharide–iron complex taken with meals than ferrous sulfate taken on an empty stomach.⁵ A study of 60 subjects taking both preparations on an empty stomach found no difference in side-effect rates.⁶ Two small, randomized, blinded studies found no difference in rates of GI side effects between carbonyl iron and ferrous sulfate.^7,8

TABLE
Representative average wholesale prices* for various iron supplement formulations

Recommendations from others

Wintrobe’s Clinical Hematology⁹ and Williams Hematology¹⁰ recommend ferrous sulfate as the standard oral iron preparation, and assert that claims of improved tolerability of one oral iron preparation over another have not been substantiated.

Clinical Commentary by Andrea Gordon, MD, at http://www.fpin.org.