Predictive biomarkers (or moderators of treatment) are features, or more likely feature clusters, that discriminate individuals who are more likely to experience a favourable or unfavourable effect from a specific treatment. Utilization of validated predictive biomarkers for chronic low back pain (CLBP) treatments is a plausible strategy to guide patients more rapidly to effective treatments thereby reducing wastage of finite healthcare funds on treatments that are ineffective (or potentially harmful). Yet, few predictive biomarkers have been successfully validated in clinical studies. This paper summarizes work by the Phenotype/Precision Spine Focus Group of the International Society for the Study of the Lumbar Spine that addressed: (1) relevant definitions for terminology
(2) advantages and disadvantages of different research approaches for the specification of predictive biomarkers
(3) methods for assessment of clinical validity
(4) approaches for their implementation
(5) barriers to predictive biomarker identification
and (6) a prioritised list of recommendations for the development and refinement of predictive biomarkers for CLBP. Key recommendations include the harmonisation of data collection, data sharing, integration of theoretical models, development of new treatments, and health economic analyses to inform cost-benefit of assessments and the application of matched treatments. The complexity of CLBP demands large datasets to derive meaningful progress. This will require coordinated and substantive collaboration involving multiple disciplines and across the research spectrum from the basic sciences to clinical applications.