The Problem
Prescription of opioids following surgery is central to pain management. However, misuse of opioids is an immense problem in the United States, and, unfortunately, during the COVID-19 pandemic, this problem only grew.[1,2] The over-prescription of pain medication contributes to the epidemic of abuse and many physicians, including gynecologists, often overprescribe postoperatively.[3,4] Despite this, there is little evidence-based guidance on post-operative opioid prescription that allows for minimizing excess postoperative prescribing while precisely prescribing the right amount to meet each patient’s unique needs. Physicians and trainees need an evidence-based tool that can help to provide patient-personalized guidance on appropriate post-operative opioid prescriptions. In two prior prospective studies enrolling 382 subjects, we had recently developed and validated a predictive nomogram post-operative opioid prescribing for women undergoing surgery in the Gynecologic Oncology division for either benign or malignant indications. This application was incorporated into a Shiny app (an open-sourced R package allowing web applications to be built using R, the statistical programming language). However, the use of this app required manual data entry during every use and the studies did not incorporate methods of applying this innovation into clinical practice.
Our Solution
We developed a Tableau mobile application that uses the validated Gynecologic Oncology Postoperative Opioid Use Predictive (GO-POP) algorithm to generate an estimated number of opioid pain pills a patient will need following surgery.[5] The application automatically identified current post-operative Gynecologic Oncology division patients. It immediately retrieved the GO-POP algorithm’s required information from the electronic health record (EHR). Automated nomogram inputs included operative time, pregabalin administration, patient age, patient education attainment, and smoking history, as well as their pre-operative anxiety regarding surgery and anticipated need for post-operative pain medication.
The application presented these inputs on the left while allowing the care provider using it to update them on the right if they saw fit. Furthermore, this application’s integration with the EHR allowed the algorithm to calculate and auto-populate values into a phone-sized Tableau dashboard, easing use for the prescribing provider. The application showed the user a single predicted number of pills that a specific patient was predicted to require, as well as the percentage of patients with similar demographics who needed more than 5, 10, and 15 pills
Impact
During the five-month study period, GO-POP algorithm use increased from 33% in November to 64% in March (p = .087). Compared to a historical cohort, the implementation of GO-POP resulted in a 33% decrease in the median number of post-operative opioids prescribed from 15 to 10 (p < .001). Thankfully, overall, GO-POP use did not change the number of refill prescriptions, provider emergency or urgent care visits, or readmissions for pain during the follow-up time period. The use of GO-POP also did not change how patients rated their average pain over the prior week (scale of 1-10) at the time of the first follow-up (median 4 vs 4, p = .834). Given this, we believe the development and implementation of the GO-POP application was successful in providing an evidence-based tool to providers that allowed them to minimize their opioid prescribing without sacrificing the quality of pain management. Integration of GO-POP allowed for evidence-based opioid prescribing that maximized pain control while reducing the number of excess opioids entering our community.
ACADEMIC OUTPUT
Posters presented at the North Carolina Obstetrical and Gynecological Society (NCOGS) 2022 Annual Meeting. Kiawah, SC, 2022; Machine Learning Conference for Healthcare, Durham, NC, 2022.
References
1. Drug Overdose Deaths in the U.S. Top 100,000 Annually. https://www.cdc.gov/nchs/pressroom/nchs_press_releases/2021/20211117.htm. Accessed May 23, 2022.
2. Products – Vital Statistics Rapid Release – Provisional Drug Overdose Data. https://www.cdc.gov/nchs/nvss/vsrr/drugoverdose-data.htm. Accessed May 23, 2022.
3. As-Sanie S, Till SR, Mowers EL, et al. Opioid prescribing patterns, patient use, and postoperative pain after hysterectomy for benign indications. Obstet Gynecol. 2017;130(6):1261-1268. doi:10.1097/AOG.0000000000002344
4. Lamvu G, Feranec J, Blanton E, Perioperative pain management: an update for obstetrician-gynecologists. Am J Obstet Gynecol. 2018;218(2):193-199. doi:10.1016/j.ajog.2017.06.021
5. Davidson B, Jelovsek JE, Rodriguez I, et al. Development and validation of the Gynecologic Oncology Predictor Of Postoperative opioid use (GO-POP) model. Gynecol Oncol. 2021;162:S55. doi:10.1016/S0090-8258(21)00748-4
