A prediction model for low functional status after colorectal cancer surgery: A retrospective cohort study using administrative data

Introduction

Colorectal cancer (CRC) accounts for 10% of all cancer incidences globally and often affects older populations and people with poor nutrition. As such, we developed a prediction model for low functional status following CRC surgery using inpatient data collected during routine practice, and investigated the relationship between low functional status and outcomes.

Methods

Data from 690 patients who underwent CRC surgery were analysed, and the Barthel Index was used to evaluate functional status [activity of daily living (ADL)]. A low-ADL status was defined as a lower score at discharge than at admission, and unchanged complete dependence from admission to discharge. The model input data included 10 basic characteristics, eight comorbidities, and four laboratory parameters. The final model was developed using stepwise logistic regression.

Results

The low-ADL predictive model was successfully developed using nine variables: age, ADL dependence, nursing home residency, ambulance use, disturbance of consciousness on admission, diabetes, cerebrovascular disease, low creatinine, and low protein (c-statistics = 0.857). Only 6.5% of high-ADL patients were unable to return home following discharge; in contrast, 53.5% of low-ADL patients were unable to return home. Low-ADL patients also had significantly longer post-operative hospital stays and higher medical costs than high-ADL patients.

Conclusion

Low-ADL patients had decreased rates of discharge to homes, experienced longer hospital stays, and incurred higher medical costs than high-ADL patients. Pre-operative prediction of low ADL status is important, and essential for taking efficient preventive measures.