Variation in pain NRS screening scores by patient and facility characteristics in a national system of care

Variation in patient reported pain has been observed between healthcare facilities within the same system, yet little is known about the sources of that variation. We examined pain screening results from 126 Veterans Health Administration (VHA) facilities, a national integrated healthcare system. We used administrative data to construct a sample of Veterans with musculoskeletal disorders (MSD) diagnosed between 2001 and 2011, retaining the 0-10 pain numeric rating scale (NRS) recorded on the diagnosis date. We used hierarchical linear models to assess NRS score variation by: patient demographic, military, and clinical characteristics, including depression, BMI, and substance use disorders, and year of MSD diagnosis; facility characteristics, including complexity score based on volume, case-mix, teaching, research, and ICU capability, as well as the proportion of MSD patients prescribed opioids at each facility. The sample comprised 2,817,264 Veterans, with mean age 58: 6.6% were women, 15.9% African American, 9.7% served in OEF/OIF, and 18.7% had depression. Facilities ranged from 4,121 – 68,208 patients in the analysis, and 48% have a high complexity score. Patient characteristics varied by facility (min-max); 2.8%-15.4% women, 0.2%-65.8% African American, 51-65 mean age, 3.2%-16.9% OEF/OIF, and 11.3%-24.0% depression. The overall mean NRS was 3.2, ranging from 1.8 to 4.1 between facilities. Patients at high complexity facilities had significantly higher NRS scores. The intra-class correlation, a measure of similarity of patients within facility, was 0.02, which justifies the need for hierarchical models. Adding patient characteristics accounted for 34% of the variation, facility characteristics an additional 37%, and year of MSD diagnosis <1%. Substantial differences in pain exist by facility and were partially explained by the patient and facility characteristics measured. An unmeasured source of variation may include fidelity to screening procedures. Awareness of these differences can help facilities implement pain treatment policies and procedures.