Work-Related Chronic Low Back Pain

Return-to-Work Outcomes After Referral to Interventional Pain and Spine Clinics

Rathin N. Vora, MD, MPH; Bruce A. Barron, MD, MS; Anthony Almudevar, PHD; Mark J. Utell, MD


Spine. 2012;37(20):E1282-E1289. 

In This Article

Materials and Methods

We conducted a chart review of all consecutive patients diagnosed with work-related chronic LBP who were treated in 1 of 2 occupational medicine clinics (Strong Occupational and Environmental Medicine [OEM] and Finger Lakes Occupational Health Services [FLOHS]), which are part of University of Rochester Medical Center (URMC), between January 1, 2001, and December 31, 2007. A more detailed description of the 2 clinics and referral sources has been previously published.[18] The general mix of patients treated in the 2 clinics was representative of work-related chronic low back injuries in the greater Rochester region. This study was approved by the Research Subjects Review Board of URMC.

Study Design

All patients met the International Association of Study of Pain criteria defined as any type of pain lasting greater than 3 months. Inclusion criteria for this study included the following: (1) all consecutive work-related chronic LBP cases that were treated in either OEM or FLOHS between January 1, 2001, and December 31, 2007, and (2) duration of LBP for 6 months or more. Exclusion criteria included (1) back surgery prior to the initial office visit to OEM or FLOHS and (2) pre-existing LBP prior to the work-related injury.

Patient referrals to PTC/SC were based on clinical judgment, with perceived pain and functional benefit as determined by the treating provider. Because of low turnover rates of providers in the 2 clinics, referral patterns were similar over the course of the study period. In general, all patients whose medical condition did not improve significantly after at least 8 weeks of conservative care were considered for referral. Although the scope of services provided at PTC/SC can be variable, all of the referral clinics in this study specialized in interventional procedures for the lower back. Comprehensive pain management centers had the capability to provide additional modalities including psychological evaluations and/or treatment with pain medications. The purpose of these referrals was primarily to improve pain and return-to-work status in patients and not pain medication management; therefore, both types of referrals were combined. OEM and FLOHS provided overall medical management and continued to follow these patients irrespective of their referral status.

Data Collection

All medical record reviews and data collection were completed by 1 of the 2 clinicians involved in this study (B.A.B or R.V.). The jobs were graded on a 5-point scale from sedentary (1) to very heavy (5). The pain and return-to-work/functional levels were determined at baseline (at the time of first visit to OEM or FLOHS) for both the referral and nonreferral groups. The pain and functional levels of the referral group were recorded at 6 months and 1 year after PTC/SC evaluation, whereas these levels were recorded at 6 months and 1 year after initial evaluation at OEM/FLOHS in the nonreferral group. OEM and FLOHS pain measurement methods were combined to form a pain scale score of 0 through 5. The functional level was measured on the basis of whether the subject returned to the original work (functional level = 1), modified work (functional level = 2), or did not return to work (functional level = 3).

Outcomes Measured

The primary outcomes measured in this group of patients were work status change and change in pain score between the referred and the nonreferred group. As a secondary objective, the investigators sought to identify factors such as time since injury (TSI) that would be predictive of positive or negative functional outcomes.

Statistical Analysis

We estimated that to obtain a power of 80% for a 2-sided significance at 0.05, the sample size needed would be 202, with 101 in each group. All data were entered into an Excel spreadsheet and were de-identified prior to submission for statistical analysis.

Two sample comparisons were performed using the Wilcoxon rank-sum test for independent samples and the Wilcoxon signed-rank test for paired samples. Differences in proportions were tested with a χ2 test. Correlations were tested using Kendall τ test for concordance. Logistic regression was used to model a dichotomous functional level outcome against TSI.

To model TSI (TSI = first office visit minus injury date in days), functional level end points were reduced to function improvement (FI = 1 if functional level decreased from baseline to 6 mo). Thirteen patients with functional level = 1 (working at their original job) at baseline were removed for this analysis. We anticipated FI to be highly dependent on TSI; therefore, the relationship between referral status, FI, and TSI interactions was examined.

Next, we built a predictive model for FI on the basis of clinic referral status and TSI. Because of the large difference in TSI between sites, the model was fit separately for OEM and FLOHS as well as for the pooled sample. The model used logistic regression with FI as an independent variable and TSI (log transformed) and clinic referral status as dependent variables (continuous and dichotomous, respectively). The fitted prediction curves were superimposed on nonparametric (kernel smoother) estimates of the curves to assess goodness of fit.