Note from the National Guideline Clearinghouse (NGC): The National Institute for Health and Clinical Excellence (NICE) commissioned an independent academic centre to perform a systematic literature review on the technology considered in this appraisal and prepare an Evidence Review Group (ERG) report. The ERG report for this technology appraisal was prepared by the Centre for Health Economics, University of York and Centre for Reviews and Dissemination, University of York (see the "Availability of Companion Documents" field).
Clinical Effectiveness
Description of Manufacturers Search Strategy and Comment on whether the Search Strategy Was Appropriate
The submission reports a search of most of the required databases for records of reviews and randomized controlled trials relating to effects of rimonabant, sibutramine and orlistat. NICE requires a search of the Cochrane Library, but the submission reports only a search of the Cochrane Database of Systematic Reviews. This may mean that the CENTRAL Register of Clinical Trials, Database of Abstracts of Reviews of Effectiveness (DARE) and the Health Technology Assessment (HTA) database were not searched. The submission reports that an additional relevant database, Biosis, was searched.
A MEDLINE search strategy only is reported. The database searches were reported to have been run on Datastarweb, but the search syntax (truncation symbols, etc.) reported is not correct for Datastarweb. The ERG was unable to rerun the strategy, as presented, in the PubMed, Datastarweb or Ovid interfaces to MEDLINE. The ERG was also unable to verify how the strategy was adapted for databases other than MEDLINE. However, the structure of the search strategy as reported is suitable for capturing the topic in MEDLINE.
The words used in the strategies for identifying evidence on the effects of rimonabant, sibutramine and orlistat are adequate to capture the topic. One search term reported is not a MeSH term (HYPERLIPIDAEMIA). The relevant MeSH terms for the topics in this section of the strategy should be DISLIPIDEMIAS/, HYPERLIPIDEMIAS/ and HYPERCHOLESTEROLEMIA/.
The submission records that reference lists of retrieved papers were reviewed to identify additional articles. This is accepted practice.
The manufacturer identified data from unpublished trials presented at conferences from its own files only: "unpublished data held on file by Sanofi-Aventis". Data from SERENADE and REBA trials are included in the submission. Searches of other external resources for trial information in the form of presentations, abstracts and posters were not reported to have been undertaken. In response to a request for clarification, the manufacturer stated that they did not search for data from ongoing (soon to report) studies.
Statement of the Inclusion/Exclusion Criteria Used in the Study Selection and Comment on whether They Were Appropriate
The manufacturer identified three base-case populations:
- Overweight or obese patients with treated type 2 diabetes
- Overweight or obese patients with dyslipidaemia (defined as triglycerides >1.7 mmol/L or total plasma cholesterol >5.0 mmol/L or low-density lipoprotein cholesterol [LDL-C] >3.0 mmol/L or high-density lipoprotein cholesterol [HDL-C] <1.03 mmol/L for men, and triglycerides >1.7 mmol/L or total plasma cholesterol >5.0 mmol/L or LDL-C >3.0 mmol/L or HDL-C <1.29 mmol/L for females) not treated with a statin, and without type 2 diabetes
- Overweight and/or obese patients with or without comorbidities, without diabetes.
These groups seem to reflect the Rimonabant in Obesity (RIO) trials rather than subgroups of importance in clinical practice. A notable omission is a subgroup of patients with hypertension.
For the review of orlistat and sibutramine the inclusion criteria were:
- Studies of 1 year duration (or data available for 1 year).
- Diet and exercise administered to placebo and treatment arms.
- Data for intention to treat (ITT) population available (if this was not stated, it was assumed that data presented in the studies were for the ITT population and they were not excluded).
- Orlistat dose of 120 mg three times a day (tid) or 120 mg with each meal.
- Sibutramine dose of 10 or 15 mg/day.
- Data relating to trial run-in periods (if applicable) were excluded from the analysis.
The inclusion criteria for the doses of orlistat and sibutramine appear clinically appropriate. All the trials evaluating orlistat included the dose of 120 mg three times daily as specified in the inclusion criteria; several trials also evaluated 30 mg or 60 mg three times daily. Two of the included sibutramine trials did not appear to meet the inclusion criteria; these two trials evaluated 20 mg of sibutramine once daily. Data for orlistat and sibutramine were only sought for 1 year; although this is appropriate for sibutramine given its licence, orlistat can be prescribed for longer, and two year data may have been available for comparison with the longer-term outcomes reported in the RIO trials.
The submission states that studies were screened by a single reviewer at both the title/abstract stage and the full paper stage, with a second reviewer screening only approximately 10% of identified studies. This could lead to missed studies and selection bias, particularly when considering the orlistat and sibutramine trials as it seems that some of the reasons for exclusion could be deemed subjective and judgements may vary between reviewers. In addition, no description is provided of methods for resolving disagreements where dual screening was undertaken.
Cost-Effectiveness
Existing Cost-Effectiveness Evidence
As part of the manufacturer's submission, a systematic search was undertaken with the aim of identifying all studies evaluating the cost-effectiveness of rimonabant, orlistat and sibutramine for the treatment of obesity. No studies of the cost-effectiveness of rimonabant were identified by the manufacturer as part of this search. However, one study which appears to have been published after the search was undertaken, was reported by the manufacturer. The search strategy was critically appraised by an experienced information scientist within the ERG.
Although the manufacturer undertook a search of most of the required databases for studies of the cost-effectiveness, a search of the Cochrane Library was not conducted. However, searches of additional relevant databases were undertaken: including National Health Service Economic Evaluation Database (NHS EED), Health Economic Evaluations Database (HEED) and Biosis.
A MEDLINE search strategy only is reported. The database searches were reported to have been run on Datastarweb. However, the search syntax reported is not correct for Datastarweb. The ERG was therefore unable to rerun the strategy as presented in the PubMed, Datastarweb or Ovid interfaces to MEDLINE. In addition, the ERG was unable to verify how the strategy was adapted for databases other than MEDLINE. However, the structure of the search strategy as reported was considered suitable for capturing rimonabant cost-effectiveness studies in MEDLINE. In addition, the words used in the strategy are adequate to capture the topic. Sensitivity might have been enhanced by the use of additional quality of life terms, such as "quality-adjusted", "qalys", etc.
Although unable to re-run the searches as reported in the submission, the ERG translated the strategy generously (assuming broad searches of all fields for terms that were not subject headings) and ran it in MEDLINE (1950 to 28 September 2007) on Datastarweb. The translated search yielded 10 records. The strategies used with the other databases were not reported so it was not possible to replicate or translate them. No additional studies relating to the cost-effectiveness were identified using the translated search.
The study mentioned in the manufacturer's submission evaluated the cost-effectiveness of rimonabant compared to diet and exercise from a UK NHS perspective. This study is based on the same model used as part of the manufacturer's own submission and hence is not considered in any more detail by the ERG.