Development and internal validation of a multivariable prediction model for the quality of life of cervical cancer survivors

Abstract

Purpose: To elaborate a prediction model for quality of life of cervical cancer survivors. Methods: We conducted a prospective cohort study on 229 cervical cancer survivors. The quality of life measures included the Functional Assessment Cancer Therapy-Cervix version 4.0 and the World Health Organization Quality of Life-brief version self-administered questionnaires. We imported the data into the statistical software program R and developed a gamma generalized linear model. Results: Our internally validated predictive model for the Functional Assessment Cancer Therapy-Cervix total score was composed of the following predictors: pain, appetite, vaginal bleeding/discharge/odor, and WHOQOL-BREF social relationships domain. The Harrell's concordance index was 0.75. Conclusions: We developed a solid and internally validated predictive model in cervical cancer survivors based on predictors, such as pain, appetite, vaginal bleeding/odor/discharge, and WHOQOL-BREF social relationships subscale score that contribute significantly to quality of life as targets for potential intervention.