Increased cardiovascular risk in rheumatoid arthritis: mechanisms and implications

Selection bias in BMJ "state of the art" reviews

The title "state of the art" for BMJ published clinical reviews is potentially misleading as these reports are at a high risk of selection bias in relation to the published literature.

The recent review on 'Increased cardiovascular risk in rheumatoid arthritis: mechanisms and implications' (16 June 2018) was restricted to searching only PubMed and identified 2,756 articles.[1] Optimal searching to identify relevant articles for any review needs to include more than one bibliographic database and should also include 'free-text' searching alongside 'controlled vocabulary'. Because only 'medical subject heading' (MeSH) terms were used, only the Medline database has been searched rather than other components of PubMed.

We carried out the same search on Embase [Emtree: “rheumatoid arthritis” and “cardiovascular disease” or “cardiovascular system”] covering the same dates and employed the ‘exclude Medline journals' limit filter. This identified 1,890 results unique to Embase. Without going through all the articles one-by-one it is not possible to identify which 'translational and basic science' studies may have been overlooked by the "state of the art" review. But further searching based on epidemiological study-design identified 40 hits that included two systematic reviews, two randomised clinical trials and one cohort study that were relevant to the review topic.[2-6] The "state of the art" MesH-based search strategy identified none of these five studies, despite two of them being available in PubMed [3, 5].

We are concerned that this reflects a fundamental problem with these clinical reviews. The "state of the art" series was launched in 2014 and the first clinical review included searching Medline, PubMed, Embase and Cumulative Index to Nursing and Allied Health Literature (CINAHL). This high standard has now slipped. Of the 14 BMJ "state of the art" reviews published over the past 12-months, 10 (71%) limited their searching to a single database - almost exclusively PubMed. Only 4 reviews (29%) searched more than one database - all searching three or more databases.

In order to reduce the potential bias identified we suggest that the commissioning and peer reviewing of BMJ “state of the art” clinical reviews should include consideration of the adequacy of the bibliographic database searching.

Emma Whiteley
Medical Student undertaking intercalated MPH
Aberdeen University Medical School
emma.whiteley.13@aberdeen.ac.uk

Paul Manson
Clinical Librarian
NHS Grampian
p.manson@nhs.net

Mike Crilly
Clinical Epidemiologist
Aberdeen University Medical School
mike.crilly@abdn.a.cuk

Competing interests: No competing interests.

References:

[1] England B, Thiele G, Anderson D, Mikuls T. Increased cardiovascular risk in rheumatoid arthritis: mechanisms and implications. BMJ 2018;361:k1036.

[2] Duan H, Heng L, Xiao L, et al. Association of rheumatoid arthritis and the prevalence of metabolic syndrome: An update meta-analysis. Int J Clin Exp Med 2016; 9(9):17334-17344.

[3] Missala I, Kassner U, Steinhagen-Thiessen E. A Systematic Literature Review of the Association of Lipoprotein(a) and Autoimmune Diseases and Atherosclerosis. Int J Rheumatol. 2012;2012:480784.

[4] Komolafe AO, Ally MT, Van Tonder JJ, Greef OB. The anti-inflammatory properties of simvastatin can benefit statin-naive rheumatoid arthritis patients with associated risks for cardiovascular disease. S Afr Fam Pract 2015:57(1):28-30.

[5] Abdollahzad H, Alipour B, Aghdashi M, Jafarabadi M. Coenzyme Q10 supplementation in patients with rheumatoid arthritis: Are there any effects on cardiovascular risk factors? Eur J Integr Med 2015:7(5): 534-539.

[6] Joseph RM, Movahedi M, Dixon WG, Symmons DP. Risks of smoking and benefits of smoking cessation on hospitalisations for cardiovascular events and respiratory infection in patients with rheumatoid arthritis: a retrospective cohort study using the Clinical Practice Research Datalink. RMD Open. 2017;3(2):e000506.

Table: Databases searched in BMJ ‘Sate of the Art’ Clinical Reviews: July 2017 to June 2018
See also for formatting: https://www.researchgate.net/publication/326129377_Selection_bias_in_BMJ...

BMJ "state of the art" review topic area: Databases searched
Autism spectrum disorder: Medline, Embase, CINAHL, PsychINFO, ERIC
Pre-exposure prophylaxis for HIV: PubMed, Embase, CINAHL
Step down chronic asthma drugs: Medline, Embase, Cochrane Central & DSR
Diagnosis and management of neck pain: Medline, Embase, Cochrane DSR, Google Scholar
Perioperative tobacco use treatments: Medline
Cardiovascular risk in rheumatoid arthritis: PubMed
Asthma-COPD overlap syndrome: PubMed
Community acquired pneumonia in adults: PubMed
Pneumonia in children: PubMed
Eosinophilic esophagitis: PubMed
Gut microbiome in systemic inflammatory disease: PubMed
Pulmonary arterial hypertension: PubMed
Robotics in colorectal surgery: PubMed
Secondary peritonitis: PubMed

Cumulative Index to Nursing and Allied Health Literature (CINAHL); Database of Systematic Reviews (DSR); Education Resources Information Centre (ERIC).