M. Rubyan2, J. Sinamo1, A. Hallway1, L. Schoel1, J. Shao1, R. Howard1, S. O’Neill1, A. Ehlers1, D. Telem1 1University Of Michigan, Department Of Surgery, Ann Arbor, MI, USA 2University of Michigan, School Of Public Health, Ann Arbor, MI, USA
Introduction:
Predictive models can guide surgeons in communicating risk to patients but rarely focus on quality of life, arguably the most important outcome following an elective surgery such as hernia repair. Scales that measure patient reported outcomes have been previously validated on small patient panels (N<200) or single-site cohorts, preventing them from being used to develop a predictive model for quality of life that requires population level data. Therefore, predictive models that have been developed are typically focused on short term outcomes such as 30-day complications and do not help patients engaged in shared decision making consider the long-term repercussions of their elective operation.
Methods:
Using the Michigan Surgical Quality Collaborative-Core Optimization Hernia Registry (MSQC-COHR), a representative, random sample of patients from 70 hospitals across the State of Michigan, we identified 1,385 adults (18+) undergoing elective abdominal wall hernia repair (2021-2023). We created a predictive model using a logistic generalized additive model to estimate the probability of reporting feeling or seeing a bulge (based on the previously validated Ventral Hernia Recurrence Inventory) using logistics generalized additive model. We interacted BMI and hernia size by sex and the validated model was verified using k- and concurvity checks, along with clinical generalizability.
Results:
A total of 267 (19.3%) patients reported feeling/seeing a bulge 1-1.5 years post hernia repair. Controlling for patient, hernia specific, and other demographics, the expected risk for healthy patients was 17.1% (no comorbidity, no smoking, BMI < 35). The expected risk for male patients was 17.6% and 21.1% for female patients. The differences in risk profile between male and female patients across combinations of BMI and hernia size was notable (Figure 1). For instance, a reduction from BMI 40 to 30 in a non-smoking male patient with no comorbidities and hernia size of 4 cm was estimated to reduce the risk by 2.4%, from 26.1% to 23.7%. However, for a female patient with the same profile, it was estimated to reduce the risk by 5.4%, from 28.0% to 22.6%. Pre-operative past month smoking (OR=1.7; p<0.05) and mesh use (OR=0.6, p<0.05) were significant risk factors. K- and concurvity checks indicated sufficiency of fit (p>0.05).
Conclusion:
This predictive model for long term outcomes built on a Statewide registry of over 1,300 patients, provides a roadmap for how to develop a clinically generalizable model that can be incorporated into shared-decision making tools to counsel patients about their long term risks of elective surgery. Leveraging population level data that combines long term patient reported outcomes with nuanced hernia characteristics, also illuminates aspects of shared decision making that may not have been previously feasible to consider when counseling patients prior to elective surgery.