Referral practices for spinal surgery are poorly predicted by clinical guidelines and opinions of primary care physicians

Abstract

Background: Degenerative disease of the lumbar spine is common. Although surgery can benefit selected patients, variation in surgical referrals reduces overall access to care.

Objectives: To compare the actual referral practices for patients with degenerative disease of the lumbar spine with recommendations for surgical referral based on clinical practice guidelines (CPGs) and family physician (FP) opinions.

Research design: An expert panel of primary and specialist physicians, using a Delphi process, came to a consensus on referral recommendations from CPGs based on a series of clinical vignettes. The vignettes were also presented to practicing FPs in Ontario, Canada, to determine their preferences for (or likelihood of) referral.

Subjects: We assembled a 10-member multispecialty expert panel. Practicing FPs were randomly sampled, stratified by county, and their patients were sampled purposefully by the FP.

Measures: Respondents, both panelists and FPs, were asked to rate the appropriateness of surgical referral for a series of clinical vignettes. Patients reported their clinical symptoms and whether they had been referred to a surgeon. Using random-effects probit regression, predictions were compared with actual referral. Receiver operating characteristic curves were constructed and area under the curve (AUC) was measured.

Results: Consensus of the panel on recommendations for referral was achieved after 2 iterations (Cronbach alpha = 0.96). Based on responses from 107 patients and 61 FPs, we found poor concordance of both predicted FP preferences (AUC 0.57) and CPG recommendations (AUC 0.64) with actual referral.

Conclusions: Referral practices are poorly predicted by CPG recommendations and individual FP opinions, based on clinical factors. Understanding other nonclinical factors may be more important in reducing variation in referrals and improving access.