Article Text
Statistics from Altmetric.com
Low back pain (LBP) is one of the most common reasons for presentation to emergency departments (EDs) internationally,1 and the majority of people presenting are prescribed an opioid.2 3 Tackling overprescribing of medicines is a health priority.4 We recently completed the SHaPED trial5 that evaluated the implementation of an evidence-based model of care for LBP in four EDs in Sydney, Australia, with 269 clinicians and 4625 patient presentations. The intervention aimed to reduce unnecessary care for LBP and resulted in an absolute reduction in opioid prescribing of 13.4% (from 64.6% to 51.2%) across the three of four original hospitals included in this time series (12.3% in the short-term SHaPED analysis which included four EDs), with no deleterious effect on patient outcomes such as pain, function and satisfaction with care. The follow-up period of SHaPED was short (3 months). In this study, we examined how long the reduction in opioid use was sustained for.
The implementation of the model of care included clinician training, educational materials, non-opioid pain management strategies, fast-track referral to outpatient services, audit and feedback about their departments’ opioid prescribing for LBP. The proportion of patients being administered an opioid within the ED was extracted from the electronic medical records of the three metropolitan Sydney EDs. The fourth site in the original trial did not have electronic records of opioid administration and was not included in this study.
We conducted an interrupted time series analysis using Prais-Winsten generalised least-squares regression with robust SEs accounting for first-order autocorrelation. The Durbin-Watson statistic indicated absence of autocorrelation. Monthly opioid administration rates in the preintervention period (1 August 2017 to 31 July 2018) were compared with those in the postintervention period starting on 1 August 2018. Data were adjusted for time so that our results are expressed as months relative to the start of the intervention, accounting for the staggered start times in the stepped-wedge design. We controlled for preintervention time trends. All statistical testing was two-tailed with an α=0.05 using Stata V.14 (Stata).
We included 9962 records of eligible LBP presentations to the three EDs between 1 August 2017 and 30 September 2021. The interrupted time series model indicates that prior to implementation, the rate of opioid use was approximately constant, at 64.6% of presentations (95% CI 59.4% to 69.9%) each month (gradient 0.04, 95% CI –0.80 to 0.88; p=0.923). Following our intervention, there was an immediate decrease in opioid administration in the ED of –13.4% (95% CI –20.2% to –6.6%; p<0.001). In the 36 months after the intervention, opioid use rates slowly increased (0.2% per month, 95% CI 0.04% to 0.37%; p=0.015) (figure 1 and table 1). The decrease in opioid use persisted for 30 months but was uncertain by 36 months.
Opioid administration rates slowly increased from the initial absolute reduction of 13%, but remained lower than preintervention levels for approximately 30 months. This equates to 1023 LBP ED presentations where opioid use was avoided. Our results suggest that the intervention caused changes in practice that led to a meaningful reduction in opioid use over the long-term. The only other ED trial that has aimed to reduce opioid prescribing in the ED setting was very small (n=109 patients) and found no effect at 6 and 12 months.6
A limitation of this study is the lack of data for one of the four sites, but this site only contributed 13% of presentations included in the original trial.
Ethics statements
Patient consent for publication
Ethics approval
Sydney Local Health District Human Research Ethics Committee (X17-0043). Trial registration number is ACTRN12617001160325.
Acknowledgments
The authors thank the SHaPED trial investigators for their assistance in conceptualising this project, and to Sydney Local Health District’s Performance and Innovation Unit for assisting with data extraction from electronic medical records.
Footnotes
Handling editor Steve Rothrock
Twitter @Caitlin_Jones_, @SweekritiSharma, @gustavocmachado
Contributors CJ: conceptualisation, investigation, project administration, visualisation, writing original draft. CWC-L, CGM and CAS: conceptualisation, investigation, project administration, visualisation, supervision, writing, review, editing. SS and DC: conceptualisation, writing, review, editing. QL and AT: conceptualisation, investigation, formal analysis, validation and writing, review, editing. GM: conceptualisation, investigation, project administration, formal analysis, validation, writing, review, editing.
Funding C-WCL (APP1193939) and CGM (APP1194283) are supported by fellowships from Australia’s National Health and Medical Research Council. This was an investigator-initiated study funded by Sydney Health Partners and the NSW Agency for Clinical Innovation.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Author note Reproducible Research Statement: Statistical code: Available on request from Dr Machado (e-mail, gustavo.machado@sydney.edu.au). Dataset: Access to the anonymised data is on application to the Sydney Local Health District Human Research Ethics Committee.