Description
- Abstract:
- Background & Aims Food insecurity, which is associated with increased dementia risk and cognitive decline, affects millions of older people in the US. Meals on Wheels (MOW) programs aim to decrease food insecurity and increase independence among this population. Given increased demand and decreased resources, MOW programs have introduced frozen shipped meals as a cheaper alternative to traditional daily delivered meals. Some studies have shown that people receiving shipped meals had higher feelings of loneliness and fall risk than those receiving daily delivered meals. The aim of this study was to evaluate the causal effect of these delivery models on nursing home and inpatient usage among older people in a randomized controlled pilot trial. Specifically, we were interested in the time to inpatient hospitalization, time to nursing home visit, and proportion of time spent in either. Because the data is from a pilot trial with small sample size, we implement additional statistical methods to increase power and precision in estimating the causal effects. Methods We use a combination of record linkage, historical data, and Bayesian imputation to increase statistical power and assess causal effects. To assess causal effects, we use the potential outcomes framework. We used deterministic and probabilistic record linkage to connect MOW participants to their Medicare records to obtain comorbidity and outcome data. The current pilot RCT data was augmented with historical data using power prior and data-adaptive weighting methods. We fit a Weibull model for survival times, and multiply impute survival times and estimate hazard ratios from a Cox PH model. For the proportion outcome, we fit a Beta regression and multiply impute the missing potential outcomes to estimate the average treatment effect. Results Subjects receiving daily delivered meals had lower risk of nursing home or inpatient placement, but these findings were not statistically significant. Daily delivered meals did significantly decrease the proportion of time spent in a nursing home or as an inpatient. As the amount of historical borrowing increased, the power increased and interval estimates became narrower. Conclusions The nursing home and inpatient hazard was reduced in the daily delivered meal group, however the study lacked sufficient power to detect significant differences. Daily delivered meals did significantly reduce the proportion of time spent in a hospital or nursing home. The statistical methods we applied did increase power and precision in our analysis.
- Notes:
- Thesis (Sc. M.)--Brown University, 2025
Citation
Lindner-Liaw, Maia,
"Increasing Power in a Meals on Wheels Pragmatic Pilot Trial: A Record Linkage, Historical Data, and Bayesian Imputation Approach"
(2025).
Biostatistics Theses and Dissertations.
Brown Digital Repository. Brown University Library.
https://repository.library.brown.edu/studio/item/bdr:pebabdr5/
Relations
Collection:
-
Biostatistics Theses and Dissertations
Theses and Dissertations for the Biostatistics department....