Predictors of multidimensional outcome after spinal surgery

The variables identified as predictors of surgical outcome often differ depending on the specific outcome variable chosen to designate "success''. A short set of multidimensional core outcome measures was recently developed, in which each of the following domains was addressed with a single question and then combined in an index: pain, function, symptom-specific well-being, general well-being ( quality of life), disability ( work and social). The present study examined the factors that predicted surgical outcome as measured using the multidimensional core measures. 163 spinal surgery patients ( mixed indications) completed questionnaires before and 6 months after surgery enquiring about demographics, medical/clinical history, fear-avoidance beliefs (FABQ), depression (Zung self-rated depression), and the core measures domains. Multiple regression analyses were used to identify predictor variables for each core domain and for the multidimensional combined core-set index. The combination of baseline symptoms, medical variables ( pain duration, previous spine operations, number of levels treated, operative procedure) and psychosocial factors ( FABQ and Zung depression) explained 34% of the variance in the core measures index ( P < 0.001). With regard to the individual domain items, the medical variables were better in predicting the items "pain'' and "symptom-specific well-being'' (R-2 = 6 - 7%) than in predicting "function'', "general well-being'' or "disability'' ( each R-2 < 4%). The inverse pattern was shown for the psychosocial predictors, which accounted for in each case approximately 20% variance in "function'', `` general well-being'' and "disability'' but only 12 - 14% variance in "pain'' and `` symptom-specific well-being''. Further to previous studies establishing the sensitivity to change of the core-set, we have shown that a large proportion of the variance in its scores after surgery could be predicted by "well-known'' medical and psychosocial predictor variables. This substantiates the recommendation for its further use in registry systems, quality management projects, and clinical trials.