A.L. Kappelman1,2,3, A.G. Antunez1,4, F. Jacobson-Davies1, R.J. Kazemi5, C. Richburg6, C. Pesavento7, A. Vastardis3, E. Kim3, D. Nanua3, H. Pediyakkal3, S.N. Smith8,9, J. Henderson9,10, V. Gavrila9, A. Cuttitta9, H. Nathan1,11, L.A. Dossett1,9,11 1University Of Michigan, Center For Healthcare Outcomes And Policy, Ann Arbor, MI, USA 2University Of Michigan, Department Of Epidemiology, Ann Arbor, MI, USA 3University Of Michigan, Medical School, Ann Arbor, MI, USA 4Brigham And Women’s Hospital, Center For Surgery And Public Health, Boston, MA, USA 5University Of California – Davis, Department Of Surgery, Sacramento, CA, USA 6Yale University School Of Medicine, Department Of Surgery, New Haven, CT, USA 7Vanderbilt University Medical Center, Department Of Surgery, Nashville, TN, USA 8University Of Michigan, Department Of Health Management And Policy, Ann Arbor, MI, USA 9University Of Michigan, Michigan Program On Value Enhancement, Ann Arbor, MI, USA 10University Of Michigan, Institute For Healthcare Policy And Innovation, Ann Arbor, MI, USA 11University Of Michigan, Department Of Surgery, Ann Arbor, MI, USA
Introduction: De-implementation of routine preoperative testing before low-risk surgery is a priority. Multicomponent de-implementation strategies (the intervention), including decision support, stakeholder engagement, and provider education, may aid in reducing this unnecessary practice. We sought to evaluate effectiveness of such an intervention in a large, quaternary referral hospital.
Methods: Following implementation of our intervention with general surgeons and physician assistants at our hospital, we used administrative claims data from the Michigan Value Collaborative (MVC) to conduct a pre-post analysis of its effect on preoperative testing rates (proportion of operations for which testing was conducted) across three low-risk general surgery ambulatory procedures (inguinal hernia repair, lumpectomy, laparoscopic cholecystectomy). Using difference-in-difference methods, we compared the intervention group (cases at a large referral hospital) to a control group (cases at other comparable hospitals in the state captured in the MVC) before and after the intervention. We employed a general linear model with a log link function and hospital-level clustered standard errors and adjusted for patient sex, Charlson-Dayo Score, operation type, and month. The target surgeries and related preoperative tests were defined using current procedural terminology (CPT) codes. The study period lasted from June 2022-July 2023 (four months pre, six months intervention, four months post). Analyses were conducted using Stata 18.
Results: At the intervention hospital, 54 cases occurred pre, 92 during, and 51 post-intervention. Control hospital case numbers were 1031, 1700, and 1043, respectively. At the intervention hospital, the mean rate of preoperative testing decreased across study periods (41.1%, 35.4%, 32.7%), while at control hospitals the mean rate of preoperative testing remained stable (about 38% in all three periods). This change over time at the intervention site compared to the control sites was statistically significant: when accounting for existing trends, the intervention was associated with a significant reduction in the probability of preoperative testing (aOR=0.792, p<0.001).
Conclusion: We provide evidence for an effective multicomponent strategy to reduce unnecessary preoperative testing before low-risk procedures applied in a large population of general surgery patients. Our use of quasi-experimental methods minimizes the risk that the observed effect is due to temporal trends.