Quality of Life – Feasibility Study

In order to achieve the first study objective 60 care staff completed the DEMQOL-CH on 400 residents between two sites, using programmed iPads. Comparison of DEMQOL-Proxy results to DEMQOL-CH results is a major validity source. Research assistants conducted cognitive interviews with each of the 20 of the 60 care providers to assess response process validity of the DEMQOL-CH and care staff-perceived feasibility of completing it. Research assistants conducted semi-structured interviews with administrators to assess feasibility from their perspectives. To avoid burden on resident care and care staff, we backfilled participating care staff with casual care staff to facilities during this project and refunded related costs.

We assessed time for care staff to complete the DEMQOL-CH (time automatically recorded for electronic surveys), patterns of missing data by resident cognition, floor/ceiling effects, tendency to the mean, lack of variance and other response patterns. Our analysis followed best practice guidelines for psychometric testing of assessment tools.

All cognitive and semi-structured interviews were audio recorded and transcribed verbatim. Our content analysis helped identify themes reflecting care staff understanding of tool items and instructions and participant perceptions of feasibility to collect DEMQOL-CH data.

For the second objective we assessed test-retest reliability, we randomly selected 80 residents (20%) from our sample of 400 residents. The same care staff member who assessed these residents with DEMQOL-CH re-assessed them within 24–48 hours. This time frame is long enough for care staff to not remember their initial ratings and short enough for resident QoL to not change substantially. To assess inter-rater reliability, we randomly selected another 8 residents (20%) from our sample. One of our 6 recruited and trained care staff other than the one who conducted the first assessment completed the DEMQOL-CH a second time within the same work shift as the first assessment.

We assessed internal consistency reliability (Cronbach’s alpha) using data on all 400 residents. This sample size exceeds the minimum for this measure. We assessed test-retest reliability by intra-class correlation with a 2-way mixed effects model of absolute agreement and inter-rater reliability by a 1-way random effects model.

In order to assess criterion validity (agreement between DEMQOL-CH and interviewer-administered DEMQOL-Proxy), we compared tools item by item with weighted Kappa statistics and sensitivity/ specificity analyses. We also compared overall scores of the 2 tools with paired t-tests. A 4-factor model was identified in previous exploratory factor analysis. In this study, we evaluated this hypothesized structure with confirmatory factor analysis. Our sample size of 400 residents was sufficient for Kappa statistics, and paired t-tests and confirmatory factor analyses. Research suggests that residents’ physical function, cognitive function, pain and depression are linked to QoL. To assess construct validity, we assessed whether QoL (measured by DEMQOL-CH) is associated with each of these clinical outcomes (measured by RAI-HC). We ran multiple regression models with QoL as the outcome variable and each clinical outcome as the predictor variable, controlling for resident characteristics (e.g. age, sex, diagnoses), care unit (e.g. unit type, staffing) and facility (e.g. size, ownership model).

To explore the third study objective we collected data for this objective at our final webinars. For this webinar, we generated a tailored report on our findings for each facility, and a summative report with findings on feasibility, barriers and facilitators of our data collection. We developed the feedback reports in collaboration with our stakeholders. In the first half of the webinar, we presented and discuss the tailored QoL reports and the summative report on recommendations and challenges for wider implementation. The second half of the summit convened focus groups to collect additional data on a) more specific experiences of participants in this project at their sites, b) tailored site reports and c) the summative report.

We audio recorded focus groups, transcribed them verbatim and conducted thematic content analysis. We then prepared a report on the focus groups for a final webinar with all available summit attendees in the last 3 months of the study. In this webinar, we presented and discussed focus group findings, received input and then generated a final set of recommendations.