Prediction of clinical outcomes in individuals with chronic low back pain: a protocol for a systematic review with meta-analysis

We will systematically search Ovid MEDLINE (PubMed), Scopus and Web of Science databases for longitudinal studies, published until June 2017, including adults with CLBP (defined as persistent pain with ≥ 3 months duration), which studied the association between multivariable models and/or predictors with at least one of the selected clinical outcomes after ≥ 3 months of follow-up. Articles’ screening and selection will be conducted by two reviewers, blindly and independently. Disagreements will be resolved by a third reviewer. Models’ discriminative ability will be assessed using C-statistic. The link between multivariable models and predictors with the clinical outcome will be analysed through association measures. Qualitative and quantitative synthesis of the available evidence will be performed. Meta-analysis will be conducted to aggregate each type of measure. In the absence or in the presence of only slight to moderate of heterogeneity, we will use the fixed or random effects model, respectively. In case of moderate to severe heterogeneity, an attempt to explain variability in detail will be made through subgroups and sensitivity analyses. Subgroup analysis will be conducted according to clinical outcome, follow-up duration (≤ 6 months versus > 6 months) and type of context (pain management clinics versus other therapeutic settings).