THE COST-EFFECTIVENESS OF THREE REFERRAL STRATEGIES FOR ACUTE LOW BACK PAIN IN THE PRIMARY CARE SETTING

Purpose:   Low back pain (LBP) is a significant public health issue and the fifth most common reason people seek care from primary care physicians.  A single universally effective intervention for LBP does not exist.  As a result the care delivered is often uncoordinated and inefficient.   One approach to LBP management directs patient referrals using an established Treatment Based Classification (TBC) algorithm which matches patients with interventions.

Method:   We constructed a Markov model to estimate the incremental cost-effectiveness of three strategies for acute LBP management in the primary care setting: no referral, mixed referral, and TBC matched referral.  Patients in the no referral arm were managed solely within the primary care setting.  Patients in the mixed referral arm were either managed within primary care or with a combination of primary care and specialty care, chiropractic care, or physical therapy.  Finally, patients in the matched referral arm were managed with primary and specialty care or with a combination of primary care and one of four matched interventions based on TBC algorithms.  The four interventions were specific exercise (McKenzie), spinal stabilization exercises, thrust manipulation or traction.  Each cohort was followed for two years.  Probabilities and utilities were estimated from published sources, professional costs were from Medicare reimbursement and pharmacologic costs were obtained from Walgreens’ posted retail rates.  Costs and utilities were discounted 3% annually. 

Result:   The TBC matched referral strategy dominated both the no referral and mixed referral strategies.  Results are sensitive to several parameters.  When the probability of recovery with pharmacologic treatment alone is ≥ 0.65 (base case = 0.41) or the probability of being matched to the stabilization exercises is ≤ 0.50 (base case = 0.75) the TBC matched and mixed referral strategies were dominated by the no referral strategy.  Mixed referral dominates the other strategies when the probability of recovery is ≤ 0.175 (base case = 0.35) for patients that are inappropriately matched to specific exercise and/or stabilization exercise interventions. 

Conclusion:   The TBC matched referral strategy for acute LBP is clinically and economically reasonable.  Our analysis identifies key parameters that influence cost-savings which could occur with TBC algorithm use and should be verified in future clinical trials.