Practical methodological strategies for individual patient data systematic reviews

BackgroundDiabetes-related leg amputations are associated with substantial morbidity and mortality and are usually preceded by foot ulceration. Annual assessment procedures are recommended to identify people with diabetes who are at risk of foot ulceration. Some evidence suggests the use of specific diagnostic tests, symptoms, signs, and elements from the patients' history, but the role of other contributory factors is unclear. Clinical guidelines for foot screening are largely based on consensus and on the findings from individual studies rather than on any systematic integration of all available data. Systematic reviews to integrate evidence of predictive factors exist, but are compromised because both adjusted and unadjusted estimates are present in the primary studies. Because adjusted meta-analyses of aggregate data can be challenging, the best way to standardise the analytical approach is to use individual patient data (IPD). There are many challenges associated with this type of systematic review including its time-consuming and costly nature. We will share the key methodological strategies that underpin our IPD systematic review of prognostic factors for foot ulceration in diabetes. We have three aims: (1) to systematically review individual patient data from cohort studies in a meta-analysis to estimate the predictive value of clinical characteristics and diagnostic tests for diabetic foot ulceration; (2) to develop a prognostic model of the risk factors for diabetic foot ulceration on the basis of data obtained worldwide; and (3) to test the robustness of the model in different demographic profiles (eg, age, duration of diabetes, control of diabetes [insulin, diet, or oral medication], and type of diabetes [type I or II]).MethodsWe searched Medline and Embase databases from inception until 2012 (appendix). We contacted the corresponding authors of eligible primary studies and invited them to contribute the data to an IPD analysis. The review protocol is available from PROSPERO.FindingsWe identified 14 eligible cohort studies of more than 20 000 patients worldwide, and all corresponding authors have given signed agreements to share their data. These datasets share common variables from which meta-analyses of predictive factors for foot ulceration in diabetes are feasible. Several key methodological strategies positively contributed to our review. Previous conduct and publication of an aggregate systematic review allowed us to develop search strategies and communicate with authors about their primary studies and resultant data; the absence of industry involvement meant that authors were in possession of their dataset; the continual involvement of the authors in the specialty of diabetes foot research sustained their interest in the review question; and a detailed protocol including items for an assessment of quality and ethical considerations helped to establish a contract between the collaborators. Importantly, the project has received funding to support research activity and international collaborations.InterpretationPractical methodological strategies have positively contributed to the conduct of this IPD systematic review, which is central to the development of a global evidence-based strategy for the risk assessment of the diabetic foot. IPD systematic reviewers should carefully consider practical effects in the preparatory stages of their work.FundingThe National Institute of Health Research Health Technology Assessment Clinical Evaluation and Trials Board (project: 10/57/08).