• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

LeBlanc E, O'Connor E, Whitlock EP, et al. Screening for and Management of Obesity and Overweight in Adults [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2011 Oct. (Evidence Syntheses, No. 89.)

Cover of Screening for and Management of Obesity and Overweight in Adults

Screening for and Management of Obesity and Overweight in Adults [Internet].

Show details

Appendix FDetails on Pharmacology Harms

Key Questions 4 and 4a. What Are the Adverse Effects of Primary Care–Relevant Interventions in Obese or Overweight Adults? Are There Differences in Adverse Effects Between Patient Subgroups?

In addition to evaluating all 61 studies from KQs 2 and 3 for harms, we abstracted an additional 27 weight loss studies for harms data (see methods for inclusion and quality criteria for additional studies).

Orlistat

General characteristics of studies. We included a total of 23 studies on the harms of orlistat (120 mg tid) (Table 16). Seventeen were RCTs from KQs 2 and 3,180-184,187,189-191,193,194,197-202 five were additional published RCTs,126,127,129,130,132 and one was an event monitoring study from the United Kingdom.133 The event monitoring study relied on doctors' retrospective reports of adverse events and had low response rates. We chose to include the study because we wanted to capture rare adverse events that might not be picked up in relatively small RCTs. Of the RCTs, eight recruited unselected populations129,182,184,189,190,193,199,200 and 14 recruited participants with at least one clinical or subclinical cardiovascular risk factor.126,127,130,132,180,181,183,187,191,194,197,198,201,202

Seven of the 22 trials (32 percent) were conducted in the United States.126,127,182,189-191,197 All trials included both men and women (overall weighted average percent of female participants, 66 percent). The overall weighted average age of the entire group was 46.9 years (range, 41 to 59 years). Only nine of 23 trials reported ethnicity of the participants, and in these trials the weighted average percent of nonwhite participants was 15.6 percent (range, 0 to 28 percent). The median trial duration was 52 weeks (range, 24 to 208 weeks), but five provided data beyond 52 weeks.

Withdrawals due to adverse effects. More participants who were randomized to orlistat were withdrawn from the study due to adverse effects compared with those who were randomized to placebo. Twenty-two trials included data on withdrawals due to harms and were combined by meta-analysis.126,127,129,130,132,180-184,187,189-191,193,194,197-202 Particpants taking orlistat were 1.6 times more likely to withdraw from the study due to adverse effects (RR, 1.63 [95% CI, 1.28-2.09]; I2=51.1%; k=22; n=11,920) (Figure 14). In absolute terms, the weighted mean withdrawal rates in the orlistat and placebo groups were 8 (range, 2 to 15 percent) and 4 percent (range, 2 to 14 percent), respectively. Many studies did not list specific adverse effects that led to withdrawal. In three of the four studies that listed reasons for withdrawal, gastrointestinal-related symptoms were the main cause of withdrawal.126,129,133 The fourth study reported that syncope, bradycardia, vomiting, and vomiting/trauma led to withdrawal.295

Total number reporting adverse effects. More participants reported adverse effects in the orlistat group compared with the placebo group. Data on the total proportion of participants with adverse effects from eight of 22 orlistat trials were combined by meta-analysis. Participants given orlistat were 1.1 times more likely to have an adverse effect than participants in the placebo group (RR, 1.10 [95% CI, 1.03-2.17]; I2=70.8%; k=8; n=11,920) (Figure 13). In absolute terms, the weighted mean rate of adverse effects was 78 percent (range, 32 to 95 percent) in the orlistat group and 70 percent (range, 26 to 93 percent) in the placebo group. Gastrointestinal events were the leading etiology of excess adverse effects.126,129,130,200

Number with serious adverse effects. Serious adverse effects were those labeled by the authors as “serious” or “severe” adverse effects. A similar number of participants reported serious adverse effects in the orlistat group compared with the placebo groups. Data on serious adverse effects from 12 of 22 studies were combined in a meta-analysis. Those taking orlistat were not more likely to suffer serious adverse effects compared with those in the placebo group (RR, 1.21 [95% CI, 0.88-21.68]; I2=62.3%; k=12; N=7724) (Figure 15). In absolute terms, the weighted mean average serious rate of adverse effects was 10 percent (range, 0 to 15 percent) in the orlistat group and 9 percent (range, 0 to 18 percent) in the placebo group. Three trials reported an elevated risk of serious adverse effects in the orlistat group compared with the control group (RR, 3.11-6.15).129,199,209 The rate of serious adverse effects in these three orlistat trials ranged from less than 1 percent199 to 10 percent.194 The serious adverse effects in these studies included fecal incontinence, diverticulitis, and abdominal pain.

Number with gastrointestinal-related adverse effects. Orlistat was associated with more gastrointestinal-related adverse effects than the placebo group. Data on gastrointestinal adverse effects from 18 studies were combined in a meta-analysis. Participants given orlistat had a 1.4 greater risk of suffering from a gastrointestinal-related adverse effect than those given placebo (RR, 1.42 [95% CI, 1.33-1.52]; I2=81.5%; k=18; N=10,401) (Figure 16). In absolute terms, the weighted mean average rate of gastrointestinal side effects in the orlistat group was 83 percent (range, 63 to 95 percent) and 59 percent (range, 39 to 82 percent) in the placebo group. Gastrointestinal side effects included loose stools, increased defecation, uncontrolled oily discharge/oily evacuation, oily spotting, fatty/oily stool, fecal urgency, discolored feces, flatus with discharge, fecal incontinence, and abdominal pain. Most gastrointestinal adverse effects were mild to moderate in intensity, occurred early in treatment, and resolved spontaneously. In an orlistat event monitoring study from the United Kingdom, gastrointestinal symptoms were the main adverse effect that general practitioners reported as the cause of patients stopping orlistat treatment.133

Hypoglycemia. Data were limited and contradictory regarding whether orlistat led to hypoglycemia in drug-treated patients with type 2 diabetes. Two studies found an increased incidence of hypoglycemia in participants treated with orlistat compared with placebo (16.9 vs. 9.7 percent; 10 percent in intervention group vs. 4 percent in control group),127,197 although the difference was not statistically significant in one study.127 A third study found no difference in the number of hypoglycemic episodes between treatment arms.187

Bone density. Data were insufficient to determine whether orlistat had detrimental effects on bone density. In a small subsample (N=30) of participants from a larger study,200 bone density did not differ between orlistat and placebo groups.216

Vitamins. Orlistat treatment appeared to be associated with a decrease in some fat-soluble vitamin levels compared with placebo. Data were strongest for vitamins E and beta carotene, but there were also several reports for vitamin D. Evidence was sparser and/or conflicting for vitamins A and K. Five trials examined the effects of orlistat on changes in vitamin E levels, and all found that orlistat resulted in a greater decrease in vitamin E compared with placebo.129,190,191, 199,202 All four trials examining beta-carotene129,190,191,199 and all three examining vitamin D129,199, 202 noted a greater decrease in vitamin levels in the orlistat group compared with placebo. One trial noted a decrease in vitamins A and K in the orlistat group compared with placebo;202 however, another study did not find that 120 mg of orlistat resulted in a lower vitamin A level compared with placebo.129

Two trials compared the number of participants in the orlistat and placebo groups with low vitamin levels at multiple measurement time points throughout the trial.200,202 More orlistat participants compared with placebo participants experienced at least two low vitamin E levels (3.2 to 4.6 percent in orlistat groups vs. 0.5 to 0.9 percent in placebo groups).200,202 Neither trial found that more orlistat participants had two or more low vitamin A levels.200,202 Data on orlistat's effects on the development of low vitamin D and beta carotene levels were mixed.200,202

More orlistat participants than placebo participants required vitamin supplementation during the study.129,182,191,199,200 In the one study that listed the type of vitamin supplementation required, vitamins D and beta carotene, but not vitamin E, were required more in the orlistat group compared with placebo.191

Liver injury. Data to evaluate orlistat's effects on the liver were insufficient. No trial reported specifically screening for liver disease. No trial recorded liver injury as an adverse effect. In an orlistat event monitoring study in the United Kingdom, no cases of serious hepatic adverse reactions were reported.133,134 There were reports of elevated liver tests with two cases felt to be causally related to orlistat treatment.133

Dosage effect. In terms of dosing, all 22 trials prescribed orlistat 120 mg tid.126,127,129,130,132,180-184, 187,189-191,193,194,197-202 Four trials included more than just a 120 mg tid dosage group (30 to 240 mg tid).129,189,190,199 Although none of the studies presented statistical comparisons between dosing groups, their data do not suggest that dosage was associated with different adverse effect rates, although the results were somewhat mixed. Three of the four trials reported similar adverse effect rates with increasing dose. For example, in one study,129 withdrawal rates due to adverse effect were 6, 5, 2, and 3 percent in the 30, 60, 120, and 240 mg tid treatment groups, respectively. In another study, severe gastrointestinal event rates were 6.6 percent in the 60 mg tid group and 10.3 percent in the 120 mg tid group; however, withdrawals for gastrointestinal events were 5 percent in the 60 mg tid group and 3.7 percent in the 120 mg tid group.199 In contrast, in the fourth trial, a weight maintenance trial, overweight and obese unselected/low risk participants who took 30, 60, and 120 mg tid of orlistat had 5.4, 7.0, and 11.7 percent, respectively, withdrawal from adverse effects (however, no statistical testing was reported to determine if these were statistically different).190

Subgroup analysis. Withdrawals from adverse effects were more likely in trials of unselected participants taking orlistat than in participants with cardiovascular risk factors, regardless of age. In eight studies of unselected populations,129,182,184,189,190,193,199,200 those who were randomized to orlistat were 2.2 times more likely to withdraw due to adverse effects than those taking placebo (RR, 2.18 [95% CI, 1.57-3.01]; I2=21.2%; k=8; N=4029). In contrast, in the 12 studies of participants with type 2 diabetes, hypertension, or dyslipidemia,126,127,130,132,180,181,183,187,191,194,197, 198,201,202 the orlistat group had no greater risk of withdrawing due to adverse effects (RR, 1.34 [95% CI, 0.93-1.94]; I2=50.5%; k=12; N=4277). Similarly, in the four trials of participants with a mean age of at least 55 years who had type 2 diabetes, hypertension, or dyslipidemia,127,180,187, 191 the orlistat group did not withdraw more than the placebo group (RR, 0.8 [95% CI, 0.43-1.49]; I2=46.9%; k=4; N=1475).

Similarly, serious adverse effects from orlistat may also be less likely in those with cardiovascular risk factors than unselected participants. In eight studies of participants with type 2 diabetes, hypertension, or dyslipidemia,126,132,181,183,194,198,201,202 serious adverse effects were not increased in the orlistat group compared with placebo (RR, 1.08 [95% CI, 0.59-1.97]; I2=63.3%; k=6; N=1992). In the four studies of unselected populations,129,193,199,200 however, there was a nonsignificant increase in the risk of serious adverse effects in those who were randomized to orlistat (RR, 2.01 [95% CI, 0.91-4.47]; I2=65.9%; k=4; N=2118). This elevated risk ratio was primarily the result of two studies.129,199 The serious adverse effects in these two studies included fecal incontinence, diverticulitis, and abdominal pain.

Metformin

General characteristics of studies. We included a total of four studies on the harms of metformin (850 mg twice daily) (Table 16). Three trials were RCTs from KQs 2 and 3142,185,186 and one was an additional published RCT.131 Recruitment criteria included impaired fasting glucose or impaired glucose tolerance,142 high waist-to-hip ratios,185 or PCOS.131,186 Only one trial was conducted in the United States.142 The overall weighted average percent of female participants in all trials was 83.6 percent (range, 66 to 100 percent; two trials included only women). The overall weighted average age of participants was 39.8 years (range, 27 to 50 years), and 45.3 percent of the participants in the largest trial of metformin were nonwhite.142 The other trials did not describe ethnicity. The average trial duration was 84 weeks (range, 26 to 208 weeks).

Withdrawals due to adverse effects. More participants who were randomized to metformin withdrew from the study due to adverse effects compared with those who were randomized to placebo. Two of the four trials included data on withdrawals due to harms and were combined by meta-analysis. Participants taking metformin were almost four times more likely to withdraw from the study due to adverse events (RR, 3.92 [95% CI, 1.23-12.57]; k=2; I2=0%; N=4118) (Figure 14). In absolute terms, the weighted mean average rate of withdrawal due to adverse effects was 5 percent (range, 0 to 7 percent) in the metformin group and 1 percent (range, 0 to 1 percent) in the placebo group. Studies did not list what adverse effects led to withdrawal. The largest trial and only study rated as good quality, DPP, did not list withdrawals due to adverse effects.142

Total number with adverse effects. More participants experienced adverse effects in the metformin group compared with the placebo group. Two of the four metformin trials listed the total proportion of participants with adverse effects and were combined by meta-analysis. Participants given metformin were almost five times more likely to suffer an adverse effect compared with those in the placebo group (RR, 4.83 [95% CI, 0.84-27.63]; I2=87.6%; k=2; N=517) (Figure 13). In absolute terms, the weighted mean average rate of adverse effects was 46 percent (range, 4 to 100 percent) in the metformin group and 16 percent (range, 6 to 17 percent) in the placebo group. Excess adverse effects were mostly due to gastrointestinal events in these two trials. DPP was not combined in the meta-analysis because it did not record the total number of adverse effects; it only reported gastrointestinal and musculoskeletal adverse effects.206

Number with serious adverse effects. No studies reported the number of participants with serious adverse effects in the two treatment groups.

Number with gastrointestinal-related adverse effects. Gastrointestinal adverse effects were more likely to occur in participants who were randomized to metformin compared with placebo. One small study of 40 women with PCOS found that two women had transient abdominal gastrointestinal events (abdominal swelling, mild diarrhea, and flatulence) during the first two weeks of treatment (RR, 5.0 [95% CI, 0.26-98.00]; N=40).185 In DPP, participants taking metformin had an increased risk of gastrointestinal symptoms (diarrhea, flatulence, nausea, vomiting) compared with the placebo group (77.8 vs. 30.7 events/100 person-years; p<0.05).206 This pattern was consistent across age groups.210 In another study of 457 people with high waist-to-hip ratio, diarrhea and nausea/vomiting were more common in the metformin group compared with placebo (diarrhea: 45/457 [9.8 percent] vs. 10/457 [2.2 percent]; nausea/vomiting: 14/457 [3.1 percent] vs. 6/457 [1.3 percent]).185 However, the incidence of abdominal pain and cramps was not different between treatment groups.185

Bone density. There were no data about the effects of metformin on bone density.

Hypoglycemia. No metformin study reported rates of hypoglycemia in treatment groups.

Dosage effects. We were unable to examine the relationship between metformin dose and adverse effects, as all four studies prescribed the same dose of metformin (850 mg twice daily).

Subgroup analysis. In DPP, the relative increase in gastrointestinal adverse events in the metformin group did not appear to differ by age.210 No other subgroup analyses were reported in DPP or could be done with meta-analysis.

Comparison of the Two Drugs

One study randomized 150 obese women to 6 months of one of three weight loss drugs: sibutramine, orlistat, and metformin.136 Abdominal discomfort occurred in the orlistat (22/50) and metformin groups (14/50). Both metformin and orlistat resulted in decreases in blood pressure and heart rate.136

Heterogeneity of Medication Studies (Meta-Regression Analysis)

We performed meta-regression to examine whether certain study characteristics influenced the association between the medication and the proportion of participants withdrawing due to adverse effects, reporting any adverse effects, reporting any serious adverse effects, and reporting gastrointestinal-related adverse effects in all cases controlling for risk status of participants and medication type. We examined multiple trial factors, including how many participants returned for followup, whether the study was conducted in the United States, and duration of the study. None of these trial factors influenced the harms effect size of the medications. Sex and age did not predict effect size for any adverse effect associated with medications. We were unable to examine ethnicity because of the paucity of reporting (nine studies) and low percentage of nonwhite participants in all of the medication studies combined (13 percent).

We had limited ability to detect differences in harms between medications, since we did not include trials that did not have placebo comparison groups. Only one trial of medication harms included head-to-head comparisons of orlistat and metformin in 150 obese women (50 in each medication group and 50 in a sibutramine group) after 6 months of treatment.136 The trial reported only two participants withdrawing from the orlistat group due to side effects, none of which were reported as serious, and there were no differences in blood pressure or heart rate. The type of medication did not influence withdrawal due to adverse effects, total adverse effects, or serious adverse effects in any of the meta-regression models.

PubReader format: click here to try

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (11M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...