Development and validation of a predictive index of elder self-neglect risk among a Chinese population

Bei Wang, Ying Xiao Hua, Xin Qi Dong

Publish Year : 2020

Objective: To develop a predictive index that estimates the individual risk of incident self-neglect onset among the US Chinese older adults.

Methods: The study used two waves of longitudinal data from 2713 participants of the Population Study of Chinese Elderly (PINE). Data were collected during 2011–2015 in Chicago, Illinois, with approximately 2-year follow-up intervals. The main outcomes are incident self-neglect cases. Variables in 14 potential predictive domains were considered, which are (1) sociodemographic/socioeconomic, (2) neighborhood/community, (3) immigration and acculturation, (4) adverse events, (5) culture, (6) general wellbeing, (7) health behavior, (8) medical health, (9) health care, (10) physical function, (11) cognitive function, (12) social wellbeing, (13) violence, and (14) psychological wellbeing. Stepwise selection in multivariable logistical regression models and bootstrapping were used to develop and validate the predictive index.

Results: The 2-year self-neglect incidence rate was 237 (8.7%). A 19-item predictive model (with a c-statistic of 0.74) was developed. After correcting for overfitting by validating in 100 bootstrapping samples, the model demonstrated moderate predictive accuracy by a c-statistic of 0.68. A point-based risk index was developed and has an area under the receiver operating characteristic curve of 0.73.

Discussion: The study developed an efficient index with a moderate-to-good predictive ability of self-neglect. With further external validation, modification, and impact studies, the index could be a culturally relevant tool for practitioners to quantify the risk of self-neglect among the US Chinese older population.