Volume 56 | Number 6 | December 2021

Abstract List

Dominic Hodgkin


To examine the relationships between nonpharmacological treatment (NPT) utilization and opioid prescriptions and doses and whether these relationships vary according to the type of NPT service received.

Data Source

Secondary data from the US Military Health System, nationwide.

Study Design

Patterns of NPT utilization and opioid prescriptions were analyzed over the 23 months after initial pain treatment (index visit). Regression models were used to examine the relationship between opioid prescription use in a given month and NPT service utilization in the three preceding months, using person fixed effects to control for time‐invariant patient characteristics, as well as time fixed effects. Analyses were stratified by whether the patient filled an opioid prescription in the first 30 days post index visit.

Data Extraction Methods

Administrative data on health care utilization were extracted from the US Military Health System Data Repository for Army service members who returned from deployments in Afghanistan and Iraq that ended in fiscal years 2008 to 2014 and had at least one outpatient visit with a primary diagnosis of musculoskeletal pain in the subsequent year.

Principal Findings

Utilization of any NPT service in the past 3 months was positively associated with filling an opioid prescription in the given month, regardless of whether the patient was initially prescribed opioids (percentage point difference [PP] =2.87,  < 0.01) or not (PP = 0.83,  < 0.01). However, for those not initially prescribed opioids, use of any NPT service in the past 3 months was negatively associated with mean daily opioid dose in the given month (morphine milligram equivalent dose = −0.4017,  < 0.01). For those initially prescribed opioids, NPT was not associated with opioid dose.


NPT only reduced the prescription opioid daily dose for some patients, whereas the probability of receiving an opioid prescription was positively associated with NPT. Future research should assess whether recent system‐level policies and program changes influence referral and opioid prescribing patterns.