OBJECTIVES: The primary objective of this study was to quantify how many publications retracted because of misconduct involved declared medical writers (i.e., not ghostwriters) or declared pharmaceutical industry support. The secondary objective was to investigate factors associated with misconduct retractions. DESIGN: A systematic, controlled, retrospective, bibliometric study. DATA SOURCE: Retracted publications dataset in the MEDLINE database. DATA SELECTION: PubMed was searched (Limits: English, human, January 1966 - February 2008) to identify publications retracted because of misconduct. Publications retracted because of mistake served as the control group. Standardized definitions and data collection tools were used, and data were analyzed by an independent academic statistician. RESULTS: Of the 463 retracted publications retrieved, 213 (46%) were retracted because of misconduct. Publications retracted because of misconduct rarely involved declared medical writers (3/213; 1.4%) or declared pharmaceutical industry support (8/213; 3.8%); no misconduct retractions involved both declared medical writers and the industry. Retraction because of misconduct, rather than mistake, was significantly associated with: absence of declared medical writers (odds ratio: 0.16; 95% confidence interval: 0.05-0.57); absence of declared industry involvement (0.25; 0.11-0.58); single authorship (2.04; 1.01-4.12); first author having at least one other retraction (2.05; 1.35-3.11); and first author affiliated with a low/middle income country (2.34; 1.18-4.63). The main limitations of this study were restricting the search to English-language and human research articles. CONCLUSIONS: Publications retracted because of misconduct rarely involved declared medical writers or declared pharmaceutical industry support. Increased attention should focus on factors that are associated with misconduct retractions.
OBJECTIVES: The primary objective of this study was to quantify how many publications retracted because of misconduct involved declared medical writers (i.e., not ghostwriters) or declared pharmaceutical industry support. The secondary objective was to investigate factors associated with misconduct retractions. DESIGN: A systematic, controlled, retrospective, bibliometric study. DATA SOURCE: Retracted publications dataset in the MEDLINE database. DATA SELECTION: PubMed was searched (Limits: English, human, January 1966 - February 2008) to identify publications retracted because of misconduct. Publications retracted because of mistake served as the control group. Standardized definitions and data collection tools were used, and data were analyzed by an independent academic statistician. RESULTS: Of the 463 retracted publications retrieved, 213 (46%) were retracted because of misconduct. Publications retracted because of misconduct rarely involved declared medical writers (3/213; 1.4%) or declared pharmaceutical industry support (8/213; 3.8%); no misconduct retractions involved both declared medical writers and the industry. Retraction because of misconduct, rather than mistake, was significantly associated with: absence of declared medical writers (odds ratio: 0.16; 95% confidence interval: 0.05-0.57); absence of declared industry involvement (0.25; 0.11-0.58); single authorship (2.04; 1.01-4.12); first author having at least one other retraction (2.05; 1.35-3.11); and first author affiliated with a low/middle income country (2.34; 1.18-4.63). The main limitations of this study were restricting the search to English-language and human research articles. CONCLUSIONS: Publications retracted because of misconduct rarely involved declared medical writers or declared pharmaceutical industry support. Increased attention should focus on factors that are associated with misconduct retractions.