Development of memorial Sloan Kettering (MSK) surgery stress score for older adults with cancer.

Abstract

e24014 Background: The stress from oncologic surgeries for older adults may lead to poor surgical outcomes. In this study, we aim to develop a MSK Surgery Stress Score to measure the complexity of surgery for these patients. The score could then be used for future analyses using datasets where patients undergoing a variety of procedures. Methods: This is a retrospective analysis of adults aged 75 and older who underwent elective cancer surgery in our institution with hospital length of stay of at least one day in 2015 to 2020. Based on surgery stress score developed by Haga et al., we explored various models including at least one or some of these variables as linear or non-linear terms: operation time, weight (kg), blood loss (cc), type of surgery, body mass index, and incision score. The primary outcome of interest evaluated was defined as the composite outcome of death within 90-day of surgery, readmission or emergency room visit or major complications (grade 3-5) within 30 days of surgery. We additionally explored models for a secondary outcome, where minor complications (grade 1-2) were included in the definition, and lastly, we looked at major complications alone as a tertiary outcome. For each of the model and outcomes, we used a logistic regression. We then used the logit transformation of the predicted probability to represent the proposed surgery stress score. Using this score, we evaluated the area under the curve (AUC) for each outcome. Results: In total, 1573 patients were included in the study. The median age was 80 (quartiles 77, 83) and just under half (49%) were male. The median (quartile) operation time was 181 minutes (115, 259), weight was 71 kg (62, 84), blood loss was 100 mL (50, 300), and just over (51%) of patient underwent an abdominopelvic procedure. The rate of 90-day mortality was 3.8%, while the rate of 30-day major complication, readmission and emergency room visit was 7.4%, 10% and 13%, respectively Furthermore, 21%, and 35% experienced primary, and secondary outcomes, respectively. Overall, 18 predictive models for each of the outcomes were developed and assessed. AUC for our different models ranged from 0.59 to 0.73 for the different definitions of our outcomes. Among the various models, the one defined using whether patients underwent an abdominopelvic procedure, incision score, operation time, weight, and blood loss (the latter two both included as non-linear terms) appeared to the front runner. Conclusions: We explored potential models to be used as the MSK Surgery Stress Score. Currently the model is being optimized by additional work. Following optimization of the model, future studies should validate this score in other cohorts of older surgical patients with cancer.