CRD summary: This review concluded that patients with osteoporosis treated with teriparatide have a reduced risk of developing back pain up to 30 months after treatment. The manufacturer of teriparatide funded the review, sponsored the included studies and employed three of the authors. Important systematic review methods that help ensure the findings are reliable were not mentioned in the report.

[The included studies compared teriparatide (20 or 40 microg/day) with placebo, alendronate (10 mg/day) or hormone replacement therapy.]

CRD commentary: The review had a clear objective although the inclusion criteria appeared to be largely determined by the studies already known to the authors (see Other Publications of Related Interest). The review was funded by the manufacturer of teriparatide, also the sponsor of the included studies, and three of the authors were employees of the company. The search to identify relevant studies was not extensive and there was no information about restrictions on language, publication status or the screening procedure.

There was insufficient information on study quality to make an independent assessment of the potential for bias in the included studies. The data handling and analysis methods appeared appropriate (and were strengthened by the authors having access to the individual patient data) but some aspects were not reported, such as weighting in the meta-analysis (presumably by trial variance). The meta-analysis graphs suggested a possible unit of analysis error (possibly artificially inflating the power) if the control group participants in the trials with more than one treatment arm (different doses) were included more than once, but it was impossible to be certain. How familiar the authors were with the data (which they already had on file) before planning the analysis was not known. The authors' conclusion with regard to the non-placebo comparators was not fully supported (in terms of statistical significance) by their subgroup analyses. Overall, key aspects of systematic review methods were absent from the report and the authors' conclusions might not be entirely impartial.