Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Anti-obesity drugs reviewed

Systematic review
Results
Weight loss
Adverse events
Comment

How well do anti-obesity drugs work? Interesting question, this, and one asked by a number of readers. Bandolier had spent a few hours beginning to wonder about whether to do a brief review, when happily a really good one arrived [1]. The short answer is that in long term studies of a year or more these drugs achieved weight loss of 3-4% more than doing nothing, or with diets.

Systematic review

The review searched for studies in four databases, reference lists, and reviews, and asked experts and companies for help. Only double blind, randomised trials of at least one year were acceptable. They had to enrol patients with a BMI of at least 30 kg/sq metre, and have a placebo or active control. Full, published studies in any language were accepted.

The primary outcome was average weight loss, both in kg and as percentage change from baseline, and the dichotomous outcomes of the percentage of patients with at least 5% or at least 10% weight loss.

Results

Only 11 studies of orlistat (6,021 participants) and three of sibutramine (929 participants) met the inclusion criteria. All of the orlistat trials had co-interventions that typically included dietary counselling and education, as well as encouragement to exercise, and in nine of the studies participants used a deficit diet as well. In one of the sibutramine studies participants were given dietary advice sheets.

Most studies had a preponderance of women (70-80%) with an average age of about 50 years. Many of the trials enrolled higher risk populations with diabetes and cardiovascular risk factors. The studies had very high withdrawal rates, averaging 33% for orlistat and 43% for sibutramine. Most trials dealt with this by imputing a last observation carried forward analysis.

Weight loss

Orlistat 120 mg three times a day for a year or more produced a consistent reduction in weight above that of placebo (Figure 1) by an average of 2.7 kg or 2.9% of initial weight. Compared with placebo 21% of patients (NNT 5) had at least a 5% weight loss and 12% (NNT 10) achieved a 10% weight loss.



Figure 1: Trials of orlistat versus placebo





Table 1: Summary of outcomes of all trials of orlistat or sibutramine versus placebo lasting at least one year



Outcome
Trials
Patients
Weighted mean difference (active - placebo)
(95% CI)
NNT
95% CI
Orlistat
Weight loss (kg)
11
5473
2.7 (2.3 to 3.1)
Weight loss (%)
10
4941
2.9 (2.3 to 3.4)
5% responders
11
5473
21 (19 to 24)
5 (4 to 5)
10% responders
10
4941
12 (8 to 16)
10 (6 to 13)
Sibutramine
Weight loss (kg)
3
738
4.3 (3.6 to 4.9)
Weight loss (%)
3
738
4.6 (3.8 to 5.4)
5% responders
3
738
34 (28 to 40)
3 (2.5 to 3.6)
10% responders
3
738
15 (4 to 27)
7 (4 to 25)


Sibutramine 15-20 mg daily for a year or more produced a consistent reduction in weight above that of placebo (Figure 2) by an average of 4.3 kg or 4.6% of initial weight. Compared with placebo 34% of patients (NNT 3) had at least a 5% weight loss and 15% (NNT 7) achieved a 10% weight loss.



Figure 2: Trials of sibutramine versus placebo





Adverse events

Gastrointestinal adverse events predominated with orlistat, and included fatty/oily stool, faecal urgency, and oily spotting in 15% to 30% of patients. Faecal incontinence occurred in 7% of those on orlistat compared with 1% on placebo, a number needed to harm of 17.

Sibutramine adverse events included increases in systolic and diastolic blood pressure of 1-3 mmHg, and pulse rate increases of 4-5 beats per minute. Other adverse events such as insomnia, dry mouth, nausea and constipation occurred in 7-20%.

Comment

A minority of patients achieved a 10% or greater weight loss, which is a useful amount. It is also a more useful trial outcome than average weight loss, since the heavy contamination of later data by imputed values from patients who had withdrawn will probably make those results unreal. An interesting discussion for another occasion is the legitimacy of imputing missing value data, both how much is reasonable (a little may be, but a lot not) and how it is done (is last observation carried forward always legitimate?).

The authors of the review comment that there were no predictive factors for responders, and suggested that therapeutic trials made sense because many patients would stop because of adverse events, while the near maximal weight loss was usually seen in the individual studies by six months. The bottom line is that weight reducing drugs are of value in a few, with significant adverse events in many.

Reference:

  1. R Padwal et al. Long-term pharmacotherapy for overweight and obesity: a systematic review and meta-analysis of randomized controlled trials. International Journal of Obesity 2003 27: 1437-1446.

previous or next story