The interventions under study were reported adequately. The rationale for selection of comparators was clear as they represented relevant treatments in the authors setting (recommended for use by the National Institute of Health and Clinical Excellence or supported by a recent large UK national survey).
The clinical evidence was from a small retrospective cohort study that evaluated changes in HbA1c levels, weight and BMI six months before and after treatment with exenatide. This type of study design is open to bias including potential for inclusion bias and the fact that external factors (such as better standards of care over the six months of the study) could have confounded the results. The 95% confidence interval for change in HbA1c levels for patients taking exenatide with OHAs overlapped with zero but the authors still reported a statistical difference of p<0.0001.
A very limited costing study was undertaken by the authors in which only the costs of medications were included. Given that the authors reported that exenatide was effective in reducing patients’ weight and HbA1c levels, it was possible that exenatide would have resulted in lower health care costs due to improved health in patients. Consequently, it was possible that not all relevant costs were included in the study. The price year was not reported and this may hinder further inflation exercises.
Analysis and results:
Cost and outcome data were derived from a small retrospective study. Differences in outcome between baseline and six months were compared using t-tests. Errors in the significance results presented were likely. Despite numerous limitations of the study, the authors reported no limitations.
The quality of the study methodology was poor. The analysis was based on a before-and-after study in which a small number of patients were identified retrospectively, so the results should be treated with caution.