Background: Well-conducted systematic reviews (SRs) provide 'gold standard' evidence in health care. A good SR should provide enough information about the interventions, populations and outcomes studied to enable it to be both replicated and updated; this information is also critical for appropriate implementation of review findings. A range of complex, non-pharmacological approaches to managing heart failure have been tested in randomized trials (RCTs); and a number of SRs of these trials have been published, all reporting positive results.

Objectives: Using heart failure as an example, we assessed: the quality of SRs of complex interventions; the extent to which these reviews inform implementation; and the feasibility of 'meta-reviewing' such SRs.

Methods: As part of a literature search to update a Cochrane review of complex interventions in heart failure (Taylor 2005), 13 similar systematic reviews were identified. Quality was appraised using the QUOROM Statement criteria, and detailed information was extracted on inclusion criteria and definitions for interventions and subgroups.

Results: Of the 14 systematic reviews in total, nine included meta-analyses, and five were primarily descriptive. Adherence to the QUOROM Statement criteria varied. Although all the SRs had similar objectives, all included a different set of studies. The content of interventions was often unclear and the interventions were described differently across the reviews: e.g. an RCT of a day hospital was variously classified as: a day hospital, a multidisciplinary heart failure clinic, a heart failure clinic with nursing as a significant component, and as telephone or mailing to the patient's home without home visiting. In addition, intervention subgroups, usual care and patient characteristics were poorly defined.