Exploring spillover effects following surgical de-implementation: An observational evaluation of primary care, referrals, and time to surgical intervention following reductions in the use of tonsillectomy and Dupuytren’s contracture

Background: Reducing the use of low-value surgery is important to maintain effective, safe and financially sustainable health systems. However, following de-implementation there are inevitably wider impacts for health systems beyond the targeted reduction in surgery, which have hitherto not been explored. Here we describe the spillover effects following the reduction in the use of two surgical procedures; tonsillectomy and Dupuytren’s contracture release (DCR) identified in 2019 by the Evidence Based Interventions (EBI) programme a de-implementation initiative in England.

Methods: This longitudinal observational study used linked primary (Clinical Practice Research Datalink) and secondary (hospital episode statistics) care electronic health records from 1st April 2016 to 29th February 2020 to identify care for patients presenting with symptoms of tonsillitis or Dupuytren’s contracture (DC). Outcome measures include GP visits; related prescriptions; outpatient appointments, referral to secondary care and surgery. Differences were explored between cohorts of patients seeking care for tonsillitis or DC before and after EBI guidelines. Using a two-part Generalised Linear Model (GLM), we explored changes in likelihood of surgery and the time-to-surgery (if it occurred) before and after EBI guidelines.

Results: Repeat GP visits for tonsillitis increased by 540 per 10,000 patients-per-year post-EBI, but remained stable for DC. Antibiotic prescriptions for tonsillitis fell, and outpatient appointments remained stable for both conditions.
The likelihood of surgery reduced for both conditions post-EBI with an odds ratio of 0.75 {95%CI 0.71, 0.79} and 0.88 {95%CI 0.81, 0.95}), with a shorter time-to-tonsillectomy of -5.2 days (95%CI {-9.5 days to -1 day}). Reduction in time-to-DCR was less evident (-2.2 days CI {95%-7.1 to + 3}) but should be considered in the context of nationally increasing elective waiting time trends.

Conclusions: Reductions in both surgical procedures were associated with small, but potentially important, changes to primary care utilisation and specialist treatment. Policy makers should identify potential spillovers from de-implementation, design policies to mitigate negative effects, and monitor these wider impacts alongside the direct influence on the targeted procedure rates.