A systematic review of the quality of cardiovascular surgery studies that extracted data from the MAUDE database.

Objective: To investigate opportunities and limitations of using the Manufacturer and User Facility Device Experience (MAUDE) database for cardiovascular surgery research, we analyzed the quality of studies having ever used MAUDE, in the field of cardiovascular surgery.

Methods: We systematically searched the Cochrane Library, PubMed, EMBASE, and Google Scholar for randomized and nonrandomized studies, from inception to July 2019. Two authors evaluated the quality of the retrieved observational studies, according to the National Institutes of Health quality assessment tool for either case series or cross-sectional studies. These tools quantify the quality of case series and cohorts/cross-sectional studies, respectively, with nine and 14 queries.

Results: Fifty-eight studies were included in the final qualitative review. Of 58 identified studies, 32 were case series, 8 were abstracts of case series, and 13 were reviews or case discussion with an included series from MAUDE. Also, five articles were cross-sectional studies. Of the 32 formal case series, 26 (81%) were found to have poor quality. The most common reasons for a poor quality designation included a lack of consecutive participants, undetermined comparability of participants, and undetermined follow-up adequacy. Only one out of five cross-sectional studies had fair quality; four others were evaluated as poor quality studies.

Conclusions: Cardiovascular surgery studies using the MAUDE database, whether case series or cross-sectional design, are mostly of poor quality. Their low quality is partly caused by poor study design, but mainly by intrinsic limitations to the MAUDE database: cases recruited are not consecutive; patient characteristics are not detailed enough to allow a meaningful comparison of patient characteristics between different patient entries; outcome measures are unclear; there is a limited follow-up; and time-to-event data are lacking. We conclude that the quality of cardiovascular surgery publications that rely on data from MAUDE could be improved if investigators were to extract all relevant data points from MAUDE entries, then apply standard quality assessment tools in compiling and reporting the data. MAUDE might be improved if it used medical case report standards during the process of reporting and indexing adverse events. To calculate the incidence rate of any adverse event, all event-free cases, as well as all adverse events in patients using a device, are required. Neither of these two variables is available in the MAUDE at the time of writing.