Background:Comparative effectiveness studies using observational data are increasingly used. Despite their high potential for bias, there are no detailed recommendations on how these studies should best be analysed and reported in rheumatology.Objectives:To conduct a systematic literature review of comparative effectiveness research in rheumatology to inform the EULAR task force developing points to consider when analysing and reporting comparative effectiveness research with observational data.Methods:All original articles comparing drug effectiveness in longitudinal observational studies of ≥100 patients published in key rheumatology journals (Scientific Citation Index > 2) between 1.01.2008 and 25.03.2019 available in Ovid MEDLINE® were included. Titles and abstracts were screened by two reviewers for the first 1000 abstracts and independently checked to ensure sufficient agreement has been reached. The main information extracted included the types of outcomes used to assess effectiveness, and the types of analyses performed, focusing particularly on confounding and attrition.Results:9969 abstracts were screened, with 218 articles proceeding to full-text extraction (Figure 1), representing a number of rheumatic and musculoskeletal diseases. Agreement between the two reviewers for the first 1000 abstracts was 92.7% with a kappa of 0.6. The majority of the studies used several outcomes to evaluate effectiveness (Figure 2A). Most of the studies did not explain how they addressed missing data on the covariates (70%) (Figure 2B). When addressed (30%), 44% used complete case analysis and 10% last observation carried forward (LOCF). 25% of studies did not adjust for confounding factors and there was no clear correlation between the number of factors used to adjust and the number of participants in the studies. An important number of studies selected covariates using bivariate screening and/or stepwise selection. 86% of the studies did not acknowledge attrition (Figure 2C). When trying to correct for attrition (14%), 38% used non-responder (NR) imputation, 24% used LUNDEX1, a form of NR imputation, and 21% LOCF.Conclusion:Most of studies used multiple outcomes. However, the vast majority did not acknowledge missing data and attrition, and a quarter did not adjust for any confounding factors. Moreover, when attempting to account for attrition, several studies used methods which potentially increase bias (LOCF, complete case analysis, bivariate screening…). This systematic review confirms the need for the development of recommendations for the assessment and reporting of comparative drug effectiveness in observational data in rheumatology.References:[1]Kristensen et al. A&R. 2006 Feb;54(2):600-6.Acknowledgments:Support of the Standing Committee on Epidemiology and Health Services ResearchDisclosure of Interests:Kim Lauper: None declared, Joanna KEDRA: None declared, Maarten de Wit Grant/research support from: Dr. de Wit reports personal fees from Ely Lilly, 2019, personal fees from Celgene, 2019, personal fees from Pfizer, 2019, personal fees from Janssen-Cilag, 2017, outside the submitted work., Consultant of: Dr. de Wit reports personal fees from Ely Lilly, 2019, personal fees from Celgene, 2019, personal fees from Pfizer, 2019, personal fees from Janssen-Cilag, 2017, outside the submitted work., Speakers bureau: Dr. de Wit reports personal fees from Ely Lilly, 2019, personal fees from Celgene, 2019, personal fees from Pfizer, 2019, personal fees from Janssen-Cilag, 2017, outside the submitted work., Bruno Fautrel Grant/research support from: AbbVie, Lilly, MSD, Pfizer, Consultant of: AbbVie, Biogen, BMS, Boehringer Ingelheim, Celgene, Lilly, Janssen, Medac MSD France, Nordic Pharma, Novartis, Pfizer, Roche, Sanofi Aventis, SOBI and UCB, Thomas Frisell: None declared, Kimme Hyrich Grant/research support from: Pfizer, UCB, BMS, Speakers bureau: Abbvie, Florenzo Iannone Consultant of: Speaker and consulting fees from AbbVie, Eli Lilly, Novartis, Pfizer, Roche, Sanofi, UCB, MSD, Speakers bureau: Speaker and consulting fees from AbbVie, Eli Lilly, Novartis, Pfizer, Roche, Sanofi, UCB, MSD, Pedro M Machado Consultant of: PMM: Abbvie, Celgene, Janssen, Lilly, MSD, Novartis, Pfizer, Roche and UCB, Speakers bureau: PMM: Abbvie, BMS, Lilly, MSD, Novartis, Pfizer, Roche and UCB, Lykke Midtbøll Ørnbjerg Grant/research support from: Novartis, Ziga Rotar Consultant of: Speaker and consulting fees from Abbvie, Amgen, Biogen, Eli Lilly, Medis, MSD, Novartis, Pfizer, Roche, Sanofi., Speakers bureau: Speaker and consulting fees from Abbvie, Amgen, Biogen, Eli Lilly, Medis, MSD, Novartis, Pfizer, Roche, Sanofi., Maria Jose Santos Speakers bureau: Novartis and Pfizer, Tanja Stamm Grant/research support from: AbbVie, Roche, Consultant of: AbbVie, Sanofi Genzyme, Speakers bureau: AbbVie, Roche, Sanofi, Simon Stones Consultant of: I have been a paid consultant for Envision Pharma Group and Parexel. This does not relate to this abstract., Speakers bureau: I have been a paid speaker for Actelion and Janssen. These do not relate to this abstract., Anja Strangfeld Speakers bureau: AbbVie, BMS, Pfizer, Roche, Sanofi-Aventis, Robert B.M. Landewé Consultant of: AbbVie; AstraZeneca; Bristol-Myers Squibb; Eli Lilly & Co.; Galapagos NV; Novartis; Pfizer; UCB Pharma, Axel Finckh Grant/research support from: Pfizer: Unrestricted research grant, Eli-Lilly: Unrestricted research grant, Consultant of: Sanofi, AB2BIO, Abbvie, Pfizer, MSD, Speakers bureau: Sanofi, Pfizer, Roche, Thermo Fisher Scientific, Sytske Anne Bergstra: None declared, Delphine Courvoisier: None declared
Analysing and reporting of observational data: a systematic review informing the EULAR points to consider when analysing and reporting comparative effectiveness research with observational data in rheumatology
