Limitations and sources of bias in clinical knee cartilage research

摘要

The purpose of this study was to systematically review the limitations and biases inherent to surgical trials on the management of knee chondral defects. A literature search of PubMed/Medline, CINAHL (Cumulative Index to Nursing and Allied Health Literature), EMBASE, and the Cochrane Central Register of Controlled Trials was conducted in September 2010 and updated in August 2011 to identify all English-language, Level I evidence, prospective, randomized controlled trials published from 1996 to present. The keyword search included the following: "autologous chondrocyte," "cartilage graft," "cartilage repair," "chondroplasty," "microfracture, " "mosaicplasty," and/or "osteochondral." Nonoperative studies, nonhuman studies, ex vivo studies, non-knee studies, and/or studies with follow-up of less than 1 year were excluded. A systematic review was performed on all included studies, and limitations and/or biases were identified and quantitated. Of 15,311 citations, 33 abstracts were reviewed and 11 prospective, randomized controlled trials were included. We identified 9 major limitations (subject age, subject prior surgery, subject duration of symptoms, lesion location, lesion size, lesion number, procedure selection, procedure standardization, and limited histologic analysis) and 7 common biases (selection, performance, transfer, nonresponder, detection, publication, and study design). Level I therapeutic studies investigating the surgical management of human knee cartilage defects have substantial identified biases and limitations. This review has limitations because other classifications of bias or limitation exist. Optimal management of cartilage defects is controversial, and future rigorous research methods could minimize common biases through strict study design and patient selection criteria, larger patient enrollment, more extended follow-up, and standardization of clinical treatment pathways. Level I, systematic review of Level I studies.