LAUSR

BACKGROUND Abstracts of systematic reviews (SR) are frequently used to guide clinical decision-making. However, if the abstract is inadequately reported, key information may be missing and it may not accurately summarize the results of the review. OBJECTIVE We aimed to investigate 1) if abstracts are fully reported; 2) if abstract reporting is associated with review/journal characteristics in physical therapy for low back pain (LBP); and 3) if these abstracts are consistent with the corresponding full texts. METHODS We searched the Physiotherapy Evidence Database for SRs in physical therapy for LBP published between 2015 and 2017. Associations between abstract reporting quality and review/journal characteristics were explored with linear regression. Abstract reporting was assessed with the 12 item Preferred Reporting Items for Systematic Reviews and Meta-Analyses for abstracts (PRISMA-A) checklist. Consistency of reporting between abstracts and the full text was evaluated by comparing responses to each item of the PRISMA-A using Kappa coefficients. Methodological quality of the reviews was assessed with A MeaSurement Tool to Assess systematic Reviews (AMSTAR-2). RESULTS We included 66 SRs, 9 Cochrane and 57 non-Cochrane. Review methodological quality ranged from 'high' (8%) to 'critically low' (76%). The mean ± SD of the "total number of PRISMA-A fully reported items" (range 0-12 points for fully reported items) was 4.1 ± 1.9 points for non-Cochrane review abstracts and 9.9 ± 1.1 points for Cochrane abstracts. Factors associated with reporting quality of abstracts were: journal impact factor (ß 0.20; 95% CI: 0.06, 0.35), number of words in abstract (ß 0.01; 95% CI: 0.00, 0.01) and review methodological quality ('critically low' with ß -3.06; 95% CI: -5.30, -0.82; with 'high' as reference variable). There was typically inconsistent reporting between abstract and full text, with most Kappa values lower than 0.60. CONCLUSIONS The abstracts of SRs in physical therapy for LBP were poorly reported and inconsistent with the full text. The reporting quality of abstracts was higher in journals with a higher impact factor, in abstracts with a greater number of words, and when the review was of higher methodological quality.