SEM Model Without Interaction Terms
To examine the relationships between the latent variables and the fall response variable, we tested a model without interaction terms (Supplementary Figure 2). Model fit was adequate but not impressive with a χ2(185)=994.23, p<0.01; CFI=0.61; RMSEA=0.055 (90% CI 0.051, 0.058). Unstandardized and standardized model coefficients for direct effects are presented in Supplementary Table 1. Being female was negatively associated with falling in the community within 30 days post-discharge, while having at least one neurological condition was positively associated with falls. Two latent variables were significantly associated with falling, fall concerns/fall history and use of high risk medications. Support and services was negatively but not significantly associated with falling, indicating there was no direct relationship with receiving support and services and falling. However, both the activities of daily living (ADL) impairment and use of high risk medications latent variables were significantly associated with receiving supports and services. For every one standard deviation increase in ADL impairments, there was a 0.84 increase in receiving support and services. Likewise, a one standard deviation increase in use of high risk meds increased support and services by 0.20.
This work was supported by the Agency for Healthcare Research and Quality [grant number R18HS020224]. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.
Gerontologist. 2018;58(6):1075-1084. © 2018 Oxford University Press