Risk Model for Predicting Gaps in Surgical Oncology Care Among Patients With Stage I-III Rectal Cancer From KwaZulu-Natal, South Africa.

PURPOSE

Gaps in surgical oncology care (GISOC), including delayed or nonreceipt of surgery, are detrimental to cancer control. This research sought to develop a risk model for predicting GISOC in South African rectal cancer (RC) patients with localized disease.

METHODS

This retrospective cohort study analyzed data from an existing colorectal cancer patient registry. GISOC was defined as surgery received >62 days after diagnosis with stage I-III RC or nonreceipt of surgery for stage I-III RC. Patient demographics, comorbidity, disease staging, and neoadjuvant therapy receipt were included as covariates in the analysis. A supervised logistic regression machine learning algorithm was used to train and test an appropriate risk model, which was translated into a nomogram. Receiver operating characteristic curve analyses and AUC assessments were used to establish the nomogram's performance.

RESULTS

The analysis included 490 patients (training data set = 245, testing data set = 245). Overall, there were 242 patients who experienced GISOC (49.4%), of whom 33 (13.6%) did not receive surgery and 209 (86.4%) had a delay in receiving surgery. The trained risk model consisted of patient race (Indian, odds ratio [OR] = 0.24; White, OR = 0.23; Black), comorbidity (OR = 2.29 no comorbidity), and neoadjuvant therapy receipt (OR = 18.40 nonreceipt). AUCs for the risk model were >0.800.

CONCLUSION

An accurate, setting-specific risk model and nomogram was developed for predicting GISOC in patients with RC. The nomogram can be implemented without the use of technology to identify patients at high risk for GISOC, who can then be targeted with risk-reduction interventions. The impact of the nomogram on existing surgical unit workflows requires further investigation.