This is a randomised controlled trial of a complex intervention of a package of supports and services provided by District Health Boards (DHBs) interdisciplinary team outreach to RAC facilities.
The study is taking place in the greater Auckland area of New Zealand. Auckland is the largest metropolitan area of New Zealand with a population of 1.4 million people. Within this region there are 175 aged care facilities, from which the participating facilities were selected on the basis of higher than ‘expected’ hospital admission rates. A total sample size of 1400 resident-years was originally anticipated to provide 80% power (5% significance) to detect a 25% reduction in rate of ASH hospitalisations in the intervention group compared to the control group when an event rate of 60 events per 100
years was expected. However, observed event rate in another cohort after commencement of the trial showed a lower event rate, so power estimates were recalculated. With 18 facilities for each group each with 14
months of follow-up, an average of 38 beds per facility and bed occupancy of 90%, we expect a total of 1500 resident years of follow-up. Revised power is estimated at 53%, considering:
• Inflated sample size as the design effect of 2.0 will allow for moderate intracluster correlation for hospitalisation rates of 0.025
• Rate of 35 ASH admissions per 100 resident years in control facilities vs. 26 ASH admissions in intervention facilities, assuming a Poisson distribution where the mean equals the variance. The control event rate was estimated from re-analysis of the RACIP study (Boyd et al., 2008) and the OPAL study cohort (results not yet published).
However, we anticipate an improvement in power because;
• the facilities were chosen (from statistical modelling) for their higher event rates (e.g. an event rate of 0.40 would provide power of 0.67)
• short-stay residents (under-represented in OPAL-based rates) have a higher event rate
• adjustment for covariates in our analysis will reduce confidence intervals around effect size
Eligible facilities were all those certified by the DHBs as providing long-term care in the region. Details were obtained of bed numbers and care level provided. Prior to facility recruitment, we examined structural aspects of RAC and evaluated potential associations with avoidable hospitalisations, known as ambulatory sensitive hospitalisations (ASH), taking into account individuals’ demographics and health status. Routine reporting of ASH events in New Zealand ceases at age 75, partly because established classifications of ASH diagnoses are perceived to be ‘less relevant’ for older people, particularly those in RAC. We thus developed an ASH classification based on diagnoses that were relevant for the RAC population. This classification method was used to identify facilities with high hospitalisation rates using multivariate modelling techniques that will be reported separately. Facilities were selected from those identified as at moderate or greater risk of resident hospitalisation based on factors identified during modelling.
Initially facilities were contacted by phone by the project manager and invited to participate in the study. This was followed by a visit to confirm written informed consent and to obtain base-line facility data after which the facility was advised whether allocated to intervention or control. In all, a total of 50 facilities were contacted and invited to participate. Twelve facilities declined to take part, citing work load and the similarity to another study recently undertaken as the main reasons. Two facilities withdrew shortly after randomisation because they did not desire to continue with the research project. These facilities were replaced with two other facilities as the intervention had not begun. One control facility has withdrawn half way through the study as it had changed ownership and the new owners did not want to continue participation in the research. Overall there has been an very positive response to the study as most of the facilities are keen to receive the perceived benefits of the intervention (Figure
Process for recruitment and randomisation of facilities.
Thirty-six facilities from three Auckland DHBs were randomised following consent. Facility randomisation was conducted by random number allocation, stratified by DHB and mix of care types (rest home only, or a mix of rest home, dementia and hospital beds). Facility identification was blinded to the main investigators wherever possible (some of the researchers were involved in clinical aspects of the intervention so were required to know only those facilities to which they provided clinical input). The control group were offered the same interventions following the completion of the intervention stage of the study. Control facilities were all blinded to all investigators.
Full ethics approval was given by the Northern Y Ethics Committee in January 2011 (NTY 10/11/090). Consent was obtained for facility participation rather than individual resident participation, as no identifying data is being collected and none of the interventions will be outside the usual clinical practice provided by facility staff. Only National Health Index (NHI) numbers (a national healthcare identification number for each person in New Zealand) were collected by the researchers. All individual residents’ information was kept anonymous to the researchers.
The intervention builds on aged care programmes already in place in the three Auckland region DHBs. Each DHB has gerontology clinical nurse specialists (GNSs) providing outreach to facilities, but the models of care vary between the DHBs and there is no consistent method to Identify high-risk facilities. The trial intervention delivers outreach tailored to facility needs. The multi-disciplinary team is considered a critical element
]. These teams were established at each of the three Auckland DHBs, building on existing RAC outreach services. The team comprises the facility General Medical Practitioner and Nurse Manager, a DHB geriatrician, a DHB Gerontology Nurse Specialist and a community pharmacist that services the facility or DHB clinical pharmacist.
The interventions in this study include:
• Initial baseline facility assessment to identify areas of need and facility care plan
• Benchmarking monitoring resident quality indicators that are linked to the quality of care provided (falls, nutrition, use of restraints, weight loss, UTIs, residents on nine or more medications).
• Three one-hour multi-disciplinary team (MDT) meetings to be held monthly for the first three months at each intervention facility, including medication reviews by the geriatrician in conjunction with the GP, pharmacist and nurse manager. At most, six residents were considered at each meeting with new admissions, those recently hospitalised, and those residents on nine or more medications given priority.
• Gerontology education and clinical coaching for RAC nurses and caregivers including advanced care planning for end-of-life care, nutrition and hydration, early identification of illness, falls prevention, end-stage dementia care, communication with families and practical aspects concerned with resident care.
The intervention continued for nine months with the intensity of the intervention decreasing over time to foster facility independence prior to the conclusion of active involvement, including months 6 and 8 where facilities did not receive any input by the GNSs. The GNSs began the intervention with one new facility per month in order to allow sufficient time for the organisation and delivery of the intervention.
Facilities provided monthly lists of all residents showing only residents’ unique identifiers, care type, and admission and discharge dates, to facilitate tracking residents and for sub-group analyses. The data for all acute hospitalisations and deaths will be retrieved from the Ministry of Health’s routinely collected public hospital admissions data on presentation of the unique identifier.
The primary outcome measurements comprise:
1. Rate of ASH admissions classified using a pre-determined set of ICD codes.
2. Number of acute admission hospital days.
3. Number of deaths from any cause, including death in acute hospital or elsewhere obtained from a national mortality database.
Numbers and rates for these three outcomes will be calculated as the proportion of residents hospitalised (1), number of hospital days per occupied bed per year (2) and deaths per person-year (3). Sub-group analyses will be conducted, by age group and gender, and by facility type.
This data will be collected from the date of randomisation to 14
months after the first visit by the GNS in the intervention arm, and for the paired facility in the control arm. Hospitalisation records will be sought five months after the final facility completes its intervention, in order to allow for any delays in registering admissions. Paper reports from each participating facility will enable checking against national records.
Analysis of primary endpoints will compare rate of ASH admissions between treatment groups followed by other main endpoints. Simple unadjusted rates, relative risks and 95% confidence intervals will be obtained initially, with subsequent multiple regression analysis adjusting for paired randomisation and other variables. Negative binomial regression will be used. Baseline rates will likely be included as a covariate in regression models as pre-intervention hospitalisation rates will be highly predictive of post-intervention rates.
Sub-group analyses to check for effects that differ from the overall treatment effect will be performed for the following:
• facilities classed as charitable, religious or welfare facilities vs. for “for profit” facilities
• for facilities that have only rest-home beds vs. those with both rest-home and hospital beds
• for long-stay vs. short-term (e.g. respite or palliative care) residents
• for facilities providing after-hours primary care cover through either their usual GP or a contracted after-hours primary care provider, vs. those without such cover (effectively using ambulance and emergency department for after-hours primary care)
All main analyses will be formally analysed on an "intention to treat" basis. Tests of significance will be two-tailed. Analysis of secondary outcomes will use standard statistical procedures applicable to categorical, continuous, or failure-time data as appropriate.