Amy C. Denham, MD, MPH

Evidence summary

The prevalence of elevated blood lead levels varies widely among different demographic groups and geographic regions, and it has decreased dramatically in the last several decades. Racial and ethnic minorities and children of families with low incomes, who live in the Northeast or Midwest, or who live in older houses continue to be at increased risk.¹ Children with blood lead levels 10 μg/dL have been shown to have poorer cognitive and behavioral functioning.²

No studies have demonstrated that screening for lead poisoning improves outcomes. To justify screening, one must therefore extrapolate from indirect evidence, demonstrating that screening tests are accurate and that treatment of children detected by screening is effective. Capillary blood samples are comparable with venous samples for detecting elevated blood lead levels. The sensitivity of capillary samples ranges from 86% to 96% compared with venous samples.³

In low-prevalence areas, questionnaires may inform screening decisions. A questionnaire inquiring about age of housing, presence of peeling paint, ongoing renovations, siblings or playmates with elevated blood lead levels, adults in the home with occupational exposures to lead, and proximity to industrial sources of lead has a sensitivity for detecting blood lead levels 10 μg/dL ranging from 32% to 87%. Sensitivity varies depending on the population and geographic location in which the questionnaire is tested. Accuracy is improved by tailoring the questionnaire based on locally important risk factors.⁴

Proposed treatments for elevated blood lead levels include chelation therapy, education about hygiene and nutrition, household dust control measures, and soil lead abatement. No good-quality trials have demonstrated that lowering slightly to moderately elevated blood lead levels (10–55 μg/dL) improves patient-oriented outcomes such as cognitive and behavioral functioning. Although 1 observational study of chelation therapy linked lowering blood lead levels with improved cognitive function,³ a randomized controlled trial showed that chelation had no effect on cognitive or behavioral outcomes.⁵

All other trials evaluating treatment for lead poisoning looked at the intermediate outcome of blood lead levels. A systematic review of randomized controlled trials showed that home dust control interventions reduced the proportion of children with elevated blood lead levels (15 μg/dL) from 14% to 6%.⁶ A randomized controlled trial of high-efficiency particulate air (HEPA) filtration vacuuming showed no effect.⁷ More intensive interventions such as soil lead abatement and paint remediation have not proven effective in good-quality randomized controlled trials.

Increasing dietary calcium and iron and decreasing dietary fat are also commonly recommended for children with elevated blood lead levels, based on animal models and cross-sectional studies. The only randomized controlled trial that investigated calcium supplementation showed no effect on blood lead levels.⁸ Our search revealed no good-quality studies on the effect of iron or fat intake on lead poisoning.

In summary, because the prevalence of lead poisoning varies between communities and continues to change, standard recommendations are not possible. Clinicians must rely on local epidemiologic data to make screening decisions. Although questionnaires are accurate in predicting elevated blood lead levels in some settings, no specific set of questions can be recommended for all populations.

No treatment options for those with mild to moderate elevations in blood lead levels have been shown to improve clinically important outcomes, although some interventions may decrease blood lead levels.

Recommendations from others

The Centers for Disease Control and Prevention (CDC) recommends

• that individual states develop screening plans based on local data
• universal screening at 12 and 24 months of age:

Otherwise screening should be targeted based on a questionnaire on age of housing, recent or ongoing remodeling, and having a sibling or playmate diagnosed with lead poisoning, in addition to questions on locally important risk factors.⁹

The American Academy of Pediatrics endorses the CDC recommendations.² The US Preventive Services Task Force, the American Academy of Family Physicians, and the American College of Preventive Medicine all recommend screening for lead poisoning at 12 months of age in children with demographic or geographic risk factors.^3,10,11

Evidence summary

The prevalence of elevated blood lead levels varies widely among different demographic groups and geographic regions, and it has decreased dramatically in the last several decades. Racial and ethnic minorities and children of families with low incomes, who live in the Northeast or Midwest, or who live in older houses continue to be at increased risk.¹ Children with blood lead levels 10 μg/dL have been shown to have poorer cognitive and behavioral functioning.²

No studies have demonstrated that screening for lead poisoning improves outcomes. To justify screening, one must therefore extrapolate from indirect evidence, demonstrating that screening tests are accurate and that treatment of children detected by screening is effective. Capillary blood samples are comparable with venous samples for detecting elevated blood lead levels. The sensitivity of capillary samples ranges from 86% to 96% compared with venous samples.³

In low-prevalence areas, questionnaires may inform screening decisions. A questionnaire inquiring about age of housing, presence of peeling paint, ongoing renovations, siblings or playmates with elevated blood lead levels, adults in the home with occupational exposures to lead, and proximity to industrial sources of lead has a sensitivity for detecting blood lead levels 10 μg/dL ranging from 32% to 87%. Sensitivity varies depending on the population and geographic location in which the questionnaire is tested. Accuracy is improved by tailoring the questionnaire based on locally important risk factors.⁴

Proposed treatments for elevated blood lead levels include chelation therapy, education about hygiene and nutrition, household dust control measures, and soil lead abatement. No good-quality trials have demonstrated that lowering slightly to moderately elevated blood lead levels (10–55 μg/dL) improves patient-oriented outcomes such as cognitive and behavioral functioning. Although 1 observational study of chelation therapy linked lowering blood lead levels with improved cognitive function,³ a randomized controlled trial showed that chelation had no effect on cognitive or behavioral outcomes.⁵

All other trials evaluating treatment for lead poisoning looked at the intermediate outcome of blood lead levels. A systematic review of randomized controlled trials showed that home dust control interventions reduced the proportion of children with elevated blood lead levels (15 μg/dL) from 14% to 6%.⁶ A randomized controlled trial of high-efficiency particulate air (HEPA) filtration vacuuming showed no effect.⁷ More intensive interventions such as soil lead abatement and paint remediation have not proven effective in good-quality randomized controlled trials.

Increasing dietary calcium and iron and decreasing dietary fat are also commonly recommended for children with elevated blood lead levels, based on animal models and cross-sectional studies. The only randomized controlled trial that investigated calcium supplementation showed no effect on blood lead levels.⁸ Our search revealed no good-quality studies on the effect of iron or fat intake on lead poisoning.

In summary, because the prevalence of lead poisoning varies between communities and continues to change, standard recommendations are not possible. Clinicians must rely on local epidemiologic data to make screening decisions. Although questionnaires are accurate in predicting elevated blood lead levels in some settings, no specific set of questions can be recommended for all populations.

No treatment options for those with mild to moderate elevations in blood lead levels have been shown to improve clinically important outcomes, although some interventions may decrease blood lead levels.

Recommendations from others

The Centers for Disease Control and Prevention (CDC) recommends

• that individual states develop screening plans based on local data
• universal screening at 12 and 24 months of age:

Otherwise screening should be targeted based on a questionnaire on age of housing, recent or ongoing remodeling, and having a sibling or playmate diagnosed with lead poisoning, in addition to questions on locally important risk factors.⁹

The American Academy of Pediatrics endorses the CDC recommendations.² The US Preventive Services Task Force, the American Academy of Family Physicians, and the American College of Preventive Medicine all recommend screening for lead poisoning at 12 months of age in children with demographic or geographic risk factors.^3,10,11

Evidence summary

The prevalence of elevated blood lead levels varies widely among different demographic groups and geographic regions, and it has decreased dramatically in the last several decades. Racial and ethnic minorities and children of families with low incomes, who live in the Northeast or Midwest, or who live in older houses continue to be at increased risk.¹ Children with blood lead levels 10 μg/dL have been shown to have poorer cognitive and behavioral functioning.²

No studies have demonstrated that screening for lead poisoning improves outcomes. To justify screening, one must therefore extrapolate from indirect evidence, demonstrating that screening tests are accurate and that treatment of children detected by screening is effective. Capillary blood samples are comparable with venous samples for detecting elevated blood lead levels. The sensitivity of capillary samples ranges from 86% to 96% compared with venous samples.³

In low-prevalence areas, questionnaires may inform screening decisions. A questionnaire inquiring about age of housing, presence of peeling paint, ongoing renovations, siblings or playmates with elevated blood lead levels, adults in the home with occupational exposures to lead, and proximity to industrial sources of lead has a sensitivity for detecting blood lead levels 10 μg/dL ranging from 32% to 87%. Sensitivity varies depending on the population and geographic location in which the questionnaire is tested. Accuracy is improved by tailoring the questionnaire based on locally important risk factors.⁴

Proposed treatments for elevated blood lead levels include chelation therapy, education about hygiene and nutrition, household dust control measures, and soil lead abatement. No good-quality trials have demonstrated that lowering slightly to moderately elevated blood lead levels (10–55 μg/dL) improves patient-oriented outcomes such as cognitive and behavioral functioning. Although 1 observational study of chelation therapy linked lowering blood lead levels with improved cognitive function,³ a randomized controlled trial showed that chelation had no effect on cognitive or behavioral outcomes.⁵

All other trials evaluating treatment for lead poisoning looked at the intermediate outcome of blood lead levels. A systematic review of randomized controlled trials showed that home dust control interventions reduced the proportion of children with elevated blood lead levels (15 μg/dL) from 14% to 6%.⁶ A randomized controlled trial of high-efficiency particulate air (HEPA) filtration vacuuming showed no effect.⁷ More intensive interventions such as soil lead abatement and paint remediation have not proven effective in good-quality randomized controlled trials.

Increasing dietary calcium and iron and decreasing dietary fat are also commonly recommended for children with elevated blood lead levels, based on animal models and cross-sectional studies. The only randomized controlled trial that investigated calcium supplementation showed no effect on blood lead levels.⁸ Our search revealed no good-quality studies on the effect of iron or fat intake on lead poisoning.

In summary, because the prevalence of lead poisoning varies between communities and continues to change, standard recommendations are not possible. Clinicians must rely on local epidemiologic data to make screening decisions. Although questionnaires are accurate in predicting elevated blood lead levels in some settings, no specific set of questions can be recommended for all populations.

No treatment options for those with mild to moderate elevations in blood lead levels have been shown to improve clinically important outcomes, although some interventions may decrease blood lead levels.

Recommendations from others

The Centers for Disease Control and Prevention (CDC) recommends

• that individual states develop screening plans based on local data
• universal screening at 12 and 24 months of age:

Otherwise screening should be targeted based on a questionnaire on age of housing, recent or ongoing remodeling, and having a sibling or playmate diagnosed with lead poisoning, in addition to questions on locally important risk factors.⁹

The American Academy of Pediatrics endorses the CDC recommendations.² The US Preventive Services Task Force, the American Academy of Family Physicians, and the American College of Preventive Medicine all recommend screening for lead poisoning at 12 months of age in children with demographic or geographic risk factors.^3,10,11

BACKGROUND: Episiotomy was initially used based on theoretical benefit, with little evidence supporting claims that it prevented severe perineal lacerations or pelvic floor dysfunction. As principles of evidence-based medicine have begun to influence obstetrical practice, the utility of routine episiotomy has been called into question. Several observational studies have suggested that episiotomy increases the risk of third- and fourth-degree lacerations. A recent Cochrane review of 6 randomized controlled clinical trials comparing routine versus restricted use of episiotomy showed that episiotomy was associated with more second-degree perineal trauma, without significant differences in dyspareunia, severe perineal trauma, or severe pain. Although all but one of the trials included in the review used mediolateral episiotomy, the one randomized trial conducted in North America (which used midline episiotomy) showed similar results. Despite these data, episiotomy remains a common practice performed in more than 40% of deliveries in the United States.

POPULATION STUDIED: The authors of this study enrolled 80 pregnant women at term who had not had previous vaginal deliveries. The 62 who went on to have vaginal deliveries were included in the analysis. The participants’ mean age was 26.3 years. The majority (92%) had prenatal care, and most (88%) had epidural analgesia during labor. Approximately one fourth of the women (28%) had forceps or vacuum-assisted delivery. A few had malpresentations, with 6% in the occiput posterior position.

STUDY DESIGN AND VALIDITY: This small observational study looked at a range of variables hypothesized to be related to perineal laceration length, including maternal demographics, size of genital hiatus and perineal body, fetal size and presentation, duration of second stage of labor, level of experience of birth attendant, operative vaginal delivery, and episiotomy. After delivery, one of the study authors measured perineal laceration length, and for 10 patients 3 additional observers measured laceration length to assess inter-rater reliability. Observers were blinded to one another’s measurements but not to the other variables included in the analysis. The authors used logistic regression and Mann-Whitney U test to determine which variables were associated with laceration length.

OUTCOMES MEASURED: Perineal laceration length was the primary outcome measured in this study. The authors also assessed laceration severity. The study did not include variables relevant to quality of life, such as healing complications, severity of pain, duration of symptoms, dyspareunia, or incontinence.

RESULTS: Of the 62 patients in the final analysis, 76% had a perineal laceration, with a median length of 4 cm. Five patients (8%) had a third-degree laceration, and one patient (2%) had a fourth-degree laceration. Approximately half (44%) had an episiotomy. The mean laceration length was 3 cm longer for patients who had an episiotomy (4.9 cm vs 1.9 cm; P < 001). Patients who had a forceps- or vacuum-assisted delivery had a longer average length of laceration, but this association was not independent of episiotomy. When assisted deliveries were excluded from the analysis, the association between episiotomy and laceration length remained significant.

BACKGROUND: Episiotomy was initially used based on theoretical benefit, with little evidence supporting claims that it prevented severe perineal lacerations or pelvic floor dysfunction. As principles of evidence-based medicine have begun to influence obstetrical practice, the utility of routine episiotomy has been called into question. Several observational studies have suggested that episiotomy increases the risk of third- and fourth-degree lacerations. A recent Cochrane review of 6 randomized controlled clinical trials comparing routine versus restricted use of episiotomy showed that episiotomy was associated with more second-degree perineal trauma, without significant differences in dyspareunia, severe perineal trauma, or severe pain. Although all but one of the trials included in the review used mediolateral episiotomy, the one randomized trial conducted in North America (which used midline episiotomy) showed similar results. Despite these data, episiotomy remains a common practice performed in more than 40% of deliveries in the United States.

POPULATION STUDIED: The authors of this study enrolled 80 pregnant women at term who had not had previous vaginal deliveries. The 62 who went on to have vaginal deliveries were included in the analysis. The participants’ mean age was 26.3 years. The majority (92%) had prenatal care, and most (88%) had epidural analgesia during labor. Approximately one fourth of the women (28%) had forceps or vacuum-assisted delivery. A few had malpresentations, with 6% in the occiput posterior position.

STUDY DESIGN AND VALIDITY: This small observational study looked at a range of variables hypothesized to be related to perineal laceration length, including maternal demographics, size of genital hiatus and perineal body, fetal size and presentation, duration of second stage of labor, level of experience of birth attendant, operative vaginal delivery, and episiotomy. After delivery, one of the study authors measured perineal laceration length, and for 10 patients 3 additional observers measured laceration length to assess inter-rater reliability. Observers were blinded to one another’s measurements but not to the other variables included in the analysis. The authors used logistic regression and Mann-Whitney U test to determine which variables were associated with laceration length.

OUTCOMES MEASURED: Perineal laceration length was the primary outcome measured in this study. The authors also assessed laceration severity. The study did not include variables relevant to quality of life, such as healing complications, severity of pain, duration of symptoms, dyspareunia, or incontinence.

RESULTS: Of the 62 patients in the final analysis, 76% had a perineal laceration, with a median length of 4 cm. Five patients (8%) had a third-degree laceration, and one patient (2%) had a fourth-degree laceration. Approximately half (44%) had an episiotomy. The mean laceration length was 3 cm longer for patients who had an episiotomy (4.9 cm vs 1.9 cm; P < 001). Patients who had a forceps- or vacuum-assisted delivery had a longer average length of laceration, but this association was not independent of episiotomy. When assisted deliveries were excluded from the analysis, the association between episiotomy and laceration length remained significant.

BACKGROUND: Episiotomy was initially used based on theoretical benefit, with little evidence supporting claims that it prevented severe perineal lacerations or pelvic floor dysfunction. As principles of evidence-based medicine have begun to influence obstetrical practice, the utility of routine episiotomy has been called into question. Several observational studies have suggested that episiotomy increases the risk of third- and fourth-degree lacerations. A recent Cochrane review of 6 randomized controlled clinical trials comparing routine versus restricted use of episiotomy showed that episiotomy was associated with more second-degree perineal trauma, without significant differences in dyspareunia, severe perineal trauma, or severe pain. Although all but one of the trials included in the review used mediolateral episiotomy, the one randomized trial conducted in North America (which used midline episiotomy) showed similar results. Despite these data, episiotomy remains a common practice performed in more than 40% of deliveries in the United States.

POPULATION STUDIED: The authors of this study enrolled 80 pregnant women at term who had not had previous vaginal deliveries. The 62 who went on to have vaginal deliveries were included in the analysis. The participants’ mean age was 26.3 years. The majority (92%) had prenatal care, and most (88%) had epidural analgesia during labor. Approximately one fourth of the women (28%) had forceps or vacuum-assisted delivery. A few had malpresentations, with 6% in the occiput posterior position.

STUDY DESIGN AND VALIDITY: This small observational study looked at a range of variables hypothesized to be related to perineal laceration length, including maternal demographics, size of genital hiatus and perineal body, fetal size and presentation, duration of second stage of labor, level of experience of birth attendant, operative vaginal delivery, and episiotomy. After delivery, one of the study authors measured perineal laceration length, and for 10 patients 3 additional observers measured laceration length to assess inter-rater reliability. Observers were blinded to one another’s measurements but not to the other variables included in the analysis. The authors used logistic regression and Mann-Whitney U test to determine which variables were associated with laceration length.

OUTCOMES MEASURED: Perineal laceration length was the primary outcome measured in this study. The authors also assessed laceration severity. The study did not include variables relevant to quality of life, such as healing complications, severity of pain, duration of symptoms, dyspareunia, or incontinence.

RESULTS: Of the 62 patients in the final analysis, 76% had a perineal laceration, with a median length of 4 cm. Five patients (8%) had a third-degree laceration, and one patient (2%) had a fourth-degree laceration. Approximately half (44%) had an episiotomy. The mean laceration length was 3 cm longer for patients who had an episiotomy (4.9 cm vs 1.9 cm; P < 001). Patients who had a forceps- or vacuum-assisted delivery had a longer average length of laceration, but this association was not independent of episiotomy. When assisted deliveries were excluded from the analysis, the association between episiotomy and laceration length remained significant.

BACKGROUND: Osteoarthritis is a major problem in primary care, but its optimal management is unclear. Nonsteroidal anti-inflammatory drugs (NSAIDs), analgesics, and intra-articular corticosteroids have significant toxicity and do not alter disease progression; thus, many patients have turned to glucosamine and chondroitin.

POPULATION STUDIED: The authors of this meta-analysis reviewed 15 randomized double-blind placebo-controlled trials of glucosamine and chondroitin (both together and separately) for patients with osteoarthritis of the knee or hip. Trials were included if they were longer than 4 weeks in duration and used an accepted osteoarthritis outcome measure. A total of 1710 patients were enrolled; no information was provided on age, sex, severity of osteoarthritis, or other treatments, making generalizability to a typical family practice difficult.

STUDY DESIGN AND VALIDITY: The authors’ search included MEDLINE, the Cochrane Controlled Trials Registry, rheumatology meeting abstracts, citation lists from review articles, and consultation with experts. Two trained reviewers independently examined outcomes, quality, and financial sponsorship of the articles. Disagreements were resolved by discussion. A pooled effect size was calculated by dividing the difference in mean outcomes between treatment group and control group by the standard deviation of the outcome value in the control group. The resulting effect measure combines pain outcomes with disability outcomes. Publication bias was addressed by funnel plots and regression of effects on the inverse of study variance.

OUTCOMES MEASURED: The primary outcome measured was improvement in symptoms at 4 weeks, measured by either a pain scale or a disability index. The article did not address other outcomes important for osteoarthritis, such as quality of life, cost, side effects, or radiographic progression of osteoarthritis.

RESULTS: Interrater reliability of the reviewers was excellent. The quality of the available studies was relatively poor, with only 2 employing intention-to-treat analysis. None of the studies reported independent funding from a governmental or not-for-profit organization, and there was evidence of a publication bias toward positive trials. Glucosamine showed an effect size of 0.44 (95% confidence interval [CI], 0.24-0.64), and chondroitin had an effect size of 0.96 (95% CI, 0.63-1.3). An effect size of 0 indicates equivalency with placebo, with less than 0.2 indicating a small effect, 0.2 to 0.8 a moderate effect, and greater than 0.8 a large effect. The studies were heterogenous; one trial with chondroitin seemed to be significantly different from the others. Excluding that trial weakened but did not eliminate the pooled effect of chondroitin. Shorter duration of treatment was associated with a lower effect size. Poorer-quality studies showed a larger effect than better-quality studies.

BACKGROUND: Osteoarthritis is a major problem in primary care, but its optimal management is unclear. Nonsteroidal anti-inflammatory drugs (NSAIDs), analgesics, and intra-articular corticosteroids have significant toxicity and do not alter disease progression; thus, many patients have turned to glucosamine and chondroitin.

POPULATION STUDIED: The authors of this meta-analysis reviewed 15 randomized double-blind placebo-controlled trials of glucosamine and chondroitin (both together and separately) for patients with osteoarthritis of the knee or hip. Trials were included if they were longer than 4 weeks in duration and used an accepted osteoarthritis outcome measure. A total of 1710 patients were enrolled; no information was provided on age, sex, severity of osteoarthritis, or other treatments, making generalizability to a typical family practice difficult.

STUDY DESIGN AND VALIDITY: The authors’ search included MEDLINE, the Cochrane Controlled Trials Registry, rheumatology meeting abstracts, citation lists from review articles, and consultation with experts. Two trained reviewers independently examined outcomes, quality, and financial sponsorship of the articles. Disagreements were resolved by discussion. A pooled effect size was calculated by dividing the difference in mean outcomes between treatment group and control group by the standard deviation of the outcome value in the control group. The resulting effect measure combines pain outcomes with disability outcomes. Publication bias was addressed by funnel plots and regression of effects on the inverse of study variance.

OUTCOMES MEASURED: The primary outcome measured was improvement in symptoms at 4 weeks, measured by either a pain scale or a disability index. The article did not address other outcomes important for osteoarthritis, such as quality of life, cost, side effects, or radiographic progression of osteoarthritis.

RESULTS: Interrater reliability of the reviewers was excellent. The quality of the available studies was relatively poor, with only 2 employing intention-to-treat analysis. None of the studies reported independent funding from a governmental or not-for-profit organization, and there was evidence of a publication bias toward positive trials. Glucosamine showed an effect size of 0.44 (95% confidence interval [CI], 0.24-0.64), and chondroitin had an effect size of 0.96 (95% CI, 0.63-1.3). An effect size of 0 indicates equivalency with placebo, with less than 0.2 indicating a small effect, 0.2 to 0.8 a moderate effect, and greater than 0.8 a large effect. The studies were heterogenous; one trial with chondroitin seemed to be significantly different from the others. Excluding that trial weakened but did not eliminate the pooled effect of chondroitin. Shorter duration of treatment was associated with a lower effect size. Poorer-quality studies showed a larger effect than better-quality studies.

BACKGROUND: Osteoarthritis is a major problem in primary care, but its optimal management is unclear. Nonsteroidal anti-inflammatory drugs (NSAIDs), analgesics, and intra-articular corticosteroids have significant toxicity and do not alter disease progression; thus, many patients have turned to glucosamine and chondroitin.

POPULATION STUDIED: The authors of this meta-analysis reviewed 15 randomized double-blind placebo-controlled trials of glucosamine and chondroitin (both together and separately) for patients with osteoarthritis of the knee or hip. Trials were included if they were longer than 4 weeks in duration and used an accepted osteoarthritis outcome measure. A total of 1710 patients were enrolled; no information was provided on age, sex, severity of osteoarthritis, or other treatments, making generalizability to a typical family practice difficult.

STUDY DESIGN AND VALIDITY: The authors’ search included MEDLINE, the Cochrane Controlled Trials Registry, rheumatology meeting abstracts, citation lists from review articles, and consultation with experts. Two trained reviewers independently examined outcomes, quality, and financial sponsorship of the articles. Disagreements were resolved by discussion. A pooled effect size was calculated by dividing the difference in mean outcomes between treatment group and control group by the standard deviation of the outcome value in the control group. The resulting effect measure combines pain outcomes with disability outcomes. Publication bias was addressed by funnel plots and regression of effects on the inverse of study variance.

OUTCOMES MEASURED: The primary outcome measured was improvement in symptoms at 4 weeks, measured by either a pain scale or a disability index. The article did not address other outcomes important for osteoarthritis, such as quality of life, cost, side effects, or radiographic progression of osteoarthritis.

RESULTS: Interrater reliability of the reviewers was excellent. The quality of the available studies was relatively poor, with only 2 employing intention-to-treat analysis. None of the studies reported independent funding from a governmental or not-for-profit organization, and there was evidence of a publication bias toward positive trials. Glucosamine showed an effect size of 0.44 (95% confidence interval [CI], 0.24-0.64), and chondroitin had an effect size of 0.96 (95% CI, 0.63-1.3). An effect size of 0 indicates equivalency with placebo, with less than 0.2 indicating a small effect, 0.2 to 0.8 a moderate effect, and greater than 0.8 a large effect. The studies were heterogenous; one trial with chondroitin seemed to be significantly different from the others. Excluding that trial weakened but did not eliminate the pooled effect of chondroitin. Shorter duration of treatment was associated with a lower effect size. Poorer-quality studies showed a larger effect than better-quality studies.