Aims and objectives
The aims and objectives of this work, undertaken in addition to those stated in the research proposal, were to:
evaluate outcome tools to measure PA in the CH population to inform the outcome assessment in WS5 (the feasibility cluster randomised trial)
evaluate outcome tools to assess staff job satisfaction and the CH environment
review methods of data collection at the level of the residents and CHs.
Methods
The service improvement work of WS4 also provided the opportunity to refine resident recruitment procedures and to clarify our proposed outcome assessment tools from WS2. Earlier WSs had highlighted the importance of PA on not only physical health/function (decelerating decline), but also more general well-being (to improve or maintain the quality of life of residents). It is also possible that depression may influence the take-up of PA and may be affected by PA. Therefore, we undertook further pilot assessment of outcome measures to assess these domains with CH residents (details can be found in the WS4 protocol; see Report Supplementary Material 6). We carried out additional feasibility work in relation to the testing of these outcome measures, which involved the ‘questerview’ method of voice-recording the outcome measure questionnaires completed with residents. This was to capture additional insights into the appropriateness of the various measures from a resident perspective. Through capturing the responses (by voice-recording residents’ answers and their conversations with researchers in relation to the questions posed), we hoped to gain an understanding of how residents interpret questions and the meanings attached to their responses.
The intention was not to collect complete data sets on all outcomes, but rather to determine what was feasible among this very challenging client group and, ultimately, to inform data collection for WS5. With this in mind, we did not use all of the outcome measures with all residents, choosing instead to test each of the outcome measures with a few residents and then, if a measure was deemed not to be fit for purpose (e.g. questions too complex), we did not pursue the use of that measure further.
For some key outcomes (e.g. measurement of PA, physical function, and health-related quality of life), we hoped to gain some preliminary insight into whether or not the tools might be sensitive to change, including the tools measuring PA, among CH residents.
All residents in the participating CHs were screened for eligibility. For a full description of recruitment procedures, see the study protocol (see Report Supplementary Material 6).
Resident recruitment
A total of 147 residents across the four CHs were eligible to participate in the research; of these, 136 were approached by researchers. Mental capacity was confirmed in 60 residents, and 75 letters were sent out to PCs. As a result of this process, 11 residents were assented to participate in the study by PCs, and 19 residents were assented by NCs. In total, 61 residents [mean 15 (range 8–31) residents per CH] were recruited for data collection. This is in line with our estimate of feasibly, being able to recruit 8–12 residents per home, as outlined in the protocol for the WS.
Evaluation of outcome measures
Questionnaire outcome measures
Working with residents and staff, we assessed the feasibility of a range of resident- and proxy-completed outcome measures.
Resident-completed outcome measures: the measures that were tested related to grip strength, physical function (EMS), mood (GDS), perceived health [EQ-5D-5L and ICEpop CAPability measure for Older people (ICECAP-O)] and quality of life [World Health Organization Quality of Life-BREF, Control, Autonomy, Self-Realization and Pleasure (CASP-19) and Ageing Well Profile]. We also tested the DEMQOL and DEMQOL Proxy, which were not included in the original grant application, as research within our unit has suggested that these are useful measures to use with frail older adults.
Proxy-completed outcome measures
The BI and PAM-RC were completed by researchers in collaboration with the CH staff for all (n = 61) participating residents at baseline. The mean BI score was 10.6 (SD 5.4) and the median score was 10 (IQR 6–14; range 1–20). More than half of the residents [n = 35 (57%)] had a score of ≤ 11, indicating dependence in ADL. The mean total PAM-RC score was 10.5 (SD 6.0), and the median score was 10 (IQR 5–15; range 1–21). The mobility and balance domains of the PAM-RC may be considered together as questions related to ‘abilities’, and thus comprise the first subsection of the scale. The mean ability subscore was 6.0 (SD 3.0), and the median score was 6 (IQR 3–8; range 1–10). Equally, walking frequency, wandering and outdoor mobility may be considered together as ‘activity’ questions, and make up the second subsection of the scale. The mean activity subscore was 4.5 (SD 3.4) and the median score was 4 (IQR 2–7; range 2–11).
At follow-up, one resident had died and one resident had moved out of the CH; therefore, 59 residents were available for data collection. As was the case at baseline, the BI and PAM-RC were to be completed by researchers in collaboration with the CH staff. However, one of the CHs (CH9) withdrew from data collection, and the data from an additional participant were marked as missing. Thus, the BI and PAM-RC were completed for 47 residents. The mean BI score was 10.0 (SD 5.6), and the median score was 9 (IQR 5.5–14.5; range 0–19). A considerable proportion [n = 29 (62%)] of these residents had a score of ≤ 11, indicating dependence in ADL.
The mean total PAM-RC score was 11.4 (SD 6.2), and the median score was 12 (IQR 5–17; range 1–20). The mean ability subscore was 6.7 (SD 3.2) and the median score was 7 (IQR 5–10; range 0–10). The mean activities subscore was 4.7 (SD 3.3) and the median score was 4 (IQR 1–7.5; range 0–10).
In accordance with the findings from WS2, the completion rates for the BI and PAM-RC were high, thus providing additional support for the feasibility of these measures.
Accelerometer
Feasibility of using accelerometers
Across the four homes, 50 (82%) of the 61 recruited residents agreed to wear a hip accelerometer at baseline. An accelerometer was not administered to the other participants for the following reasons: resident was ill during data collection (n = 1), consultee deemed it inappropriate (n = 3) and resident wore a wrist accelerometer (n = 7). There were no differences in the personal characteristics of those residents to whom a hip-worn accelerometer was administered (n = 50) and those who did not receive one (n = 11) (p > 0.05).
Of those 50 residents who wore a hip accelerometer, nine did not meet the criteria for valid wear time (i.e. ≥ 8 hours on ≥ 4 days of the week) and a further three accelerometers were lost. Thus, accelerometer data from 38 (62% of recruited residents) residents were valid. There were no differences in the personal characteristics and mobility scores of those residents who met the criteria for valid accelerometer data (n = 38) and those who did not (n = 11) (p > 0.05).
Those residents (n = 38) who provided valid wear time for analysis at baseline were approached and asked to wear the accelerometer again. Of these, 27 residents agreed to wear the accelerometer. An accelerometer was not administered to the other 11 residents for the following reasons: resident deceased (n = 1), resident moved out of home (n = 1), resident declined to wear (n = 6) and health had deteriorated (n = 3). Twenty-one of the 27 residents who agreed to wear an accelerometer met the criteria for valid wear. As was the case at baseline, there were no differences in the personal characteristics and mobility scores of those residents who met the criteria for valid accelerometer data (n = 21) and those who did not (n = 6) (p > 0.05).
Refinement of data collection procedures
Based on learning from this WS, a more structured approach to the accelerometer data collection procedures was proposed for use in the feasibility trial (WS5), in an attempt to ensure the quality of accelerometer data. Key aspects of the data collection procedures are detailed here.
First, whenever possible, accelerometers were to be administered to all participants by a researcher (with support from CH staff, when appropriate) between 09:00 and 11:00 in an attempt to ensure sufficient wear time (i.e. ≥ 8 hours per day). Furthermore, in an attempt to ensure that the accelerometer was worn on a weekend day, we proposed that accelerometers should be administered on either a Thursday or a Friday, given that compliance in the current WS tended to decline over the measurement period following initial enthusiasm. The need for more support from the research team than was originally proposed was acknowledged in this WS, and a member from the research team visited or telephoned each of the CHs periodically over the measurement period. However, these visits were predominantly done on an ad hoc basis, as the frequency was based on perceived need. It was felt that a more systematic approach to the provision of this support would be needed for a larger trial. Thus, we proposed that a prompting schedule was implemented in the feasibility trial. Specifically, a researcher would make a visit or telephone call to the CH two or three times over the measurement period. Flexibility would be afforded over when these prompts were made, with the exception that all homes would receive a prompt on day 2 of the measurement period.
In addition, we suggested that, when appropriate, the following techniques (each of which was trialled in the current WS) be employed, in an effort to improve compliance:
provide the research lead with a list of residents wearing the accelerometers and request that this is displayed prominently in the CH
provide CH staff with an ‘activity monitor how-to guide’ (which includes contact details for the research team).
Exploration of content and methods of collecting routine and anonymous data in care homes
The research assistant discussed with the CH managers in the four selected homes the content and method of collecting routine data at the level of the resident (e.g. age, sex, length of residence) and CH (e.g. number of beds, staff levels), clarifying what information is recorded, how it is reported and the frequency/time scale of reporting. This informed data collection in the feasibility study, including the economic evaluation.
Members of the research team also discussed with the manager the feasibility of collecting anonymised data at the level of the home and resident. Staff successfully completed the BI and PAM-RC for each of the residents in the home on an anonymised basis.
Evaluate tools to assess staff job satisfaction and the care home environment
A range of tools (see Report Supplementary Material 6) was considered. After review of the literature and discussion with the care managers and care staff, the P-CAT was chosen as an appropriate tool for staff completion.
