Article type
Year
Abstract
Background: The use of rhBMP-2 in spinal fusion increased 25-fold from 2002 to 2006, with much of that increase due to off-label use. In the midst of concerns over the safety of rhBMP-2 and questions regarding the accurate reporting of adverse events in early industry-sponsored trials, we were afforded access to individual patient data (IPD), study protocols, and manufacturer’s reports which we used to assess benefits, harms, and reporting bias.
Objectives: To assess reporting bias in publications of industry-sponsored trials of rhBMP-2 in spinal fusion.
Methods: We searched MEDLINE, EMBASE, the Cochrane Central Register of Controlled Trials and other databases through August 2012 for published studies of rhBMP-2. For each industry-sponsored trial we identified a primary publication and associated companion publications. Using a previously published protocol, we identified and classified examples of reporting bias by comparing journal publications with corresponding study protocols, internal manufacturer reports, and IPD provided by the manufacturer.
Results: Nine of 16 included studies were published inmedical journals as individual trials; two were partly described in articles that analyzed multiple studies together; five were unpublished. When we compared published articles to manufacturer reports submitted to the United States Food and Drug Administration, study protocols, and IPD, we identified selective reporting, reporting of inappropriately pooled data, and underreporting of outcome measures. While published studies of rhBMP-2 indicated no increase in adverse events due to rhBMP-2 along with the same or better fusion rates compared with bone graft, our analysis of individual patient data indicated no increase in fusion rates with possible increased cancer risk.
Conclusions: Compared with individual patient data and internal manufacturer reports, the journal publications exaggerated the benefits and minimized the potential harms of rhBMP-2 for both on-label and off label use.
Objectives: To assess reporting bias in publications of industry-sponsored trials of rhBMP-2 in spinal fusion.
Methods: We searched MEDLINE, EMBASE, the Cochrane Central Register of Controlled Trials and other databases through August 2012 for published studies of rhBMP-2. For each industry-sponsored trial we identified a primary publication and associated companion publications. Using a previously published protocol, we identified and classified examples of reporting bias by comparing journal publications with corresponding study protocols, internal manufacturer reports, and IPD provided by the manufacturer.
Results: Nine of 16 included studies were published inmedical journals as individual trials; two were partly described in articles that analyzed multiple studies together; five were unpublished. When we compared published articles to manufacturer reports submitted to the United States Food and Drug Administration, study protocols, and IPD, we identified selective reporting, reporting of inappropriately pooled data, and underreporting of outcome measures. While published studies of rhBMP-2 indicated no increase in adverse events due to rhBMP-2 along with the same or better fusion rates compared with bone graft, our analysis of individual patient data indicated no increase in fusion rates with possible increased cancer risk.
Conclusions: Compared with individual patient data and internal manufacturer reports, the journal publications exaggerated the benefits and minimized the potential harms of rhBMP-2 for both on-label and off label use.