Survey of the reporting characteristics of systematic reviews in rehabilitation

Article type
Authors
Gianola S1, Gasparini M2, Agostini M3, Castellini G4, Corbetta D5, Gozzer P6, Li L7, Sirtori V5, Taricco M8, Tetzlaff J9, Turolla A3, Moher D10, Moja L11
1Clinical Epidemiology Unit, I.R.C.C.S. Orthopedic Institute Galeazzi, Milan, Italy
2Department of rehabilitation, Asl Biella, Italy
3Laboratory of Kinematics and Robotics. I.R.C.C.S. Fondazione Ospedale San Camillo, Venezia, Italy
4COF Lanzo Hospital, Lanzo d’Intelvi, Como, Italy
5Unit of Functional Recovery, Fondazione Centro San Raffaele del Monte Tabor, Milan, Italy
6APSS Tn, Villa Igea, Trento, Italy
7Department of Physical Therapy, University of British Columbia, Vancouver, British Columbia, Canada; Arthritis Research Centre of Canada, Vancouver, British Columbia, Canada
8Rehabilitation Unit, University Hospital St. Orsola-Malpighi Polyclinic, Bologna, Italy
9Department of Epidemiology & Community Medicine, Faculty of Medicine, University of Ottawa, Ottawa, Ontario, Canada
10Clinical Epidemiology Program, Ottawa Hospital Research Institute, Ottawa, Canada
11Department of Biomedical Sciences for Health, University of Milan, Italy; Clinical Epidemiology Unit, I.R.C.C.S. Orthopedic Institute Galeazzi, Milan, Italy
Abstract
Background: Systematic reviews (SRs) have become increasingly important for informing clinical practice; however little is known about the reporting characteristics and the quality of the SRs relevant to practice of rehabilitation health professionals.

Objectives: To examine the reporting quality of a representative cross-sectional sample of published SRs on rehabilitation. We specifically focused on the descriptive, reporting and bias-related characteristics.

Methods: We searched Medline for aggregative and configurative SRs indexed in 2011 focused on rehabilitation as restoring of functional limitations, written in English. Two reviewers independently screened and selected the SRs and extracted data using a 38-item data collection form derived from PRISMA. The data were analyzed descriptively.

Results: We sampled 88 SRs published in 59 journals, with most journals publishing only one Sr. Eight were Cochrane Reviews (10%). Nearly half [41/88 (47%)] of SRs were focused on musculoskeletal and connective tissue diseases followed from neurological diseases. Over two-thirds of SRs assessed the quality of primary studies (74/88 [84%]). One-third of the studies included a meta-analysis. All Cochrane Reviews included only randomized controlled trials (RCT) whereas 45% of non-Cochrane Reviews used also other designs. Half of the Cochrane Reviews reported a statistically favorable result for the primary outcome, whereas only 11% of non-Cochrane Reviews did so. Both non-Cochrane (83%) reviews and Cochrane Reviews (100%) assessed dimensions of risk of bias (e.g. allocation concealment).

Conclusions: Our sample of SRs in the rehabilitation field shows heterogeneous characteristics and a moderate quality of reporting. Poor control of potential source of bias might be improved if more widely agreed upon evidence-based reporting guidelines will be actively endorsed and adhered to by authors and journals.