A predictive tool for ability to remain at home after cancer surgery in older adults

Importance — Shared decision-making with older adults regarding cancer surgery is critical. Prognostication tools that report individualized risk estimates of patient-centered outcomes can facilitate discussions.

Objective — To develop and internally validate a risk prediction model, STAYHOME, to estimate the risk of losing the ability to live at home for older adults after cancer surgery.

Design, setting, and participants — This was a retrospective population-based prognostic study conducted in Ontario, Canada. Included were adults 70 years and older undergoing cancer surgery over the period 2007 to 2019. Data were analyzed from June 2023 to January 2024.

Exposures — Predictor variables selected among information available preoperatively. The predictive model included age, sex, rural residence, previous cancer diagnosis, type of surgery, frailty, preoperative home care use, neoadjuvant therapy, cancer site, and cancer stage.

Main outcomes and measures — Inability to stay at home, defined as admission to nursing home. Fine-Gray models accounting for the competing risk of death were used. Discrimination and calibration were assessed. Bootstrap validation using 1000 samples with replacement was performed.

Results — Among 97 353 patients (median [IQR] age, 76 [73-81] years; 61 370 female [63.0%]), there were 2658 events (2.7%) at 6 months and 3746 events (3.8%) events at 12 months. The mean predicted risk of not staying home was 2.4% at 6 months and 3.4% at 12 months. Areas under the curve were 0.76 and 0.75 for 6-and 12-month predictions, respectively. Deviation from the observed risk of not staying home was 0.33% (95% CI, 0.31%-0.34%) for 6-month predictions and 0.46% (95% CI, 0.44%-0.48%) for 12-month predictions. Calibration was maintained across risk deciles.

Conclusions and relevance — Results of this prognostic study reveal that STAYHOME used information available preoperatively to predict the risk of not remaining home after cancer surgery for older adults. It presented good discrimination and was well calibrated. Individualized risk estimates from STAYHOME may support counseling, shared decision-making, and setting of expectations before surgery.