Efficacy and cost-effectiveness of a community-based model of care for older patients with complex needs: a study protocol for a multicentre randomised controlled trial using a stepped wedge cluster design

The study will be conducted at two sites in Far North Queensland (Australia): Cairns and Kuranda. Cairns is a regional centre on the north-east coast of Australia with a population of 160,285, of whom 11.6% are over the age of 65 years [13]. Kuranda is a rural community adjacent to Cairns in the north-west with a population of 4684, of whom 16.4% are over the age of 65 years [14].

A multicentre, randomised controlled trial using a stepped wedge cluster design with repeated cross-sectional samples will be utilised to measure participants’ acceptability of the model, health outcomes, services utilisation and economic viability of the OPEN ARCH intervention (see Fig. 2).

The stepped wedge design allows all participants to receive the intervention, which is the preferred methodology for studies where the intervention is predicted to do more good than harm [15]. It allows patients to act as their own controls and detect trends and changes associated with time [15, 16].

The OPEN ARCH intervention builds on a successful collaboration between Cairns and Hinterland and Hospital and Health Service clinicians with local GP practices. Individual GPs with an interest in geriatrics will be approached via clinical and professional networks to participate in the study. A minimum of 10 GPs within the Cairns area (urban) and the Cairns Hinterland region (rural) will be approached to participate in the study. The project manager will conduct informed consent for the GP to participate.

All recruited GPs will be randomised as ‘clusters’ at baseline using simple randomisation [17]. Such cluster randomisation allows control for the ‘between GPs variance’. A minimum of two clusters (two GPs) will be commencing the intervention at each step, as per the recommendations of Barker et al. [16], to ensure that the intervention effect estimator maintains the nominal 5% significance level and is reasonably unbiased. All clusters will ultimately receive the intervention for the same length of time.

For the purposes of this study, the following staffing will be recruited: a project manager, who will facilitate GP engagement, patient recruitment and consent, data collection, data management and reporting; a community-embedded specialist in geriatric medicine to provide comprehensive geriatric assessments and management plans at the GPs’ rooms; and two community-based care coordinators (enablement officers), who will facilitate access to healthcare services, social supports and self-management options. These positions are solely community embedded, designed to coordinate hospital-based specialist geriatric and other primary care services, such as social care services, allied health and rehabilitation services.

The OPEN ARCH intervention has two components: fast-track specialist geriatric assessment via a community-based referral to a geriatrician in local GP practices; and an enablement service offering coordination and implementation of health and social care services customised to patient need in the community. The OPEN ARCH intervention will be delivered within primary healthcare. Patients who are deemed eligible by GP screening evaluation and meet inclusion criteria will be invited to provide informed consent.

During the first interview and assessment, the geriatrician will provide an in-depth, standardised comprehensive geriatric assessment (CGA), a multi-dimensional, interdisciplinary process used to quantify an older individual’s medical, psychosocial and functional capabilities, including diagnosis, identification of problems, goal-setting and forming an individualised comprehensive management plan for holistic treatment, rehabilitation, support and follow up [19]. Alternatively, if the participant is unable to attend the assessment at the practice, a geriatrician will offer a home visit. The type and number of referrals to primary allied healthcare professionals and community support services will be determined by individual clinical needs and delivered in the community (GP, allied health practice or community health centre) or at home. Examples of allied health services that will be offered include: occupational therapy, physiotherapy, dietetics and social work. Community social support services such as personal care assistance, meals and domestic assistance will be arranged through My Aged Care and community health providers.

Upon completion of the initial interview and assessment, the geriatrician and enablement officer will meet with GP via face-to-face case conferencing, to complete the CGA with recommendations. The following Medicare Benefits Schedule (MBS) items will be available to GPs to claim: Item Numbers 735, 739 and 743 [20].

Based on the CGA recommendations, the enablement officer will meet with participants to develop individual care plans. The enablement officer, who is an advanced health professional specialising in aged care, will further coordinate the participant’s community care by linking them with existing services. Although a registered clinician, their role is not to provide discipline specific services, but rather to draw on existing public, private and non-government allied health, nursing and social support services available in the community to facilitate care coordination goals. All participants will continue to consult with their GPs for usual medical care.

The baseline (pre-intervention or standard care) level of service provision for older patients includes referrals to the Cairns Hospital Outpatient clinic without multidisciplinary care coordination or referral to teams, such as the Aged Care Assessment Team.

As OPEN ARCH requires active involvement, it would not be possible to blind participants or GPs (clusters), nor enablement officers or the project manager [15]. Participants and GPs will be required to make an informed decision for consent, so they will have full disclosure available prior to the consent. Enablement officers providing clinical care will have full knowledge of the participants’ health status throughout the study period. Similarly, the project manager, who will be required to enter the data, will remain un-blinded throughout the study. However, every attempt will be made to blind outcome assessors in order to reduce information or ascertainment bias [21]. Directly asking outcome assessors which intervention they think was administered will judge the success of blinding.

The primary outcome measures, including Emergency Department (ED) presentations, hospital admissions, in-patient bed days, allied health and support services utilisation, will be analysed as panels between each cross-sectional time point of the stepped wedge design, comparing primary outcomes within and between clusters. Each cross-section will include participants undertaking the intervention as well as those receiving standard care (pre-intervention component). In addition to assessing intervention effects, the stepped wedge design can also model the effects of time to determine whether the time of intervention (intervening early versus intervening late) impacts the effectiveness of the intervention. Secondary outcome measures, including functional status, ISAR and QoL measures, will be compared for the pre versus post intervention periods, for each participant, and thus the participants will act as their own control. Participants’ experience of the OPEN ARCH intervention as measured by the Likert scale responses will be analysed using descriptive statistics.

Within-cluster and between-cluster variance, treatment outcomes for intervention versus pre-intervention, across time, will be examined by applying a generalised linear mixed model (GLMM), as recommended for small cluster designed stepped wedge studies [16]. GLMM models can be adjusted for unequal numbers within clusters and cope with non-parametric data distributions. Within the GLMM, primary and secondary outcomes will be assessed against time and intervention status, represented as fixed effects, while study clusters and clients will be represented in the model as random effects [16]. Secondary outcome analyses will include t tests (for continuous data), chi-square tests (for proportions) and GLMMs where applicable, to control for demographic differences. All statistical tests will be two-sided and performed using a 5% significance level with no adjustment for multiplicity. Data analysis will be performed using STATA version 15 [18].

The economic research question is whether the OPEN ARCH intervention is cost-effective in reducing hospital utilisation and enhancing health-related quality of life when compared to the standard care. The perspective for the evaluation model will be a healthcare provider in the Australian setting [27].

Costs of the intervention will be incurred during the CGA and management. The CGA and management costs include the average time with each participant of different personnel, such as a GP, community-embedded specialist in geriatric medicine and community-based care coordinator. Costs of consecutive ED presentations, hospital admissions and hospital length of stay will be included. The cost of the pre-intervention care will include healthcare services utilisation before the initial GP consultation and CGA. Trial initiation, patient accrual and pre-screening activities as well as research and evaluation costs will be recorded and reported separately [28, 29].

Resources will be measured on a patient-specific basis obtained from the purpose-built data input form collated by the project manager from routinely collected computerised patient activity records and medical clinic records, following informed consent processes. Utilisation of objective data is known to eliminate patient recall bias and reduce the amount of missing data [30]. The cost of the time will be derived from a log of time spent on direct patient care activities. A top-down approach will be used to value resource use for capital and operating costs. Healthcare resource use data will be valued using Australian Independent Hospital Pricing Authority standard costs [31] reported in 2018 AUS$.

QALYs will be generated using a multi-attribute utility instrument, the EuroQoL 5-Dimension (EQ-5D-5L) scale [24]. This standardised, five-item descriptive system and visual analogue scale measures mobility, self-care, pain, usual activities and anxiety, and contributes to a ‘utility’ score. The AQOL-8D [32] instrument will be used to generate alternative QALYs. The AQOL-8D consists of eight dimensions: physical, including independent living, senses and pain; and psycho-social, including mental health, happiness, self-worth, coping and relationships. Utility weights for the AQOL were generated using the time-trade-off technique and Australian population sample. Due to its robust psychometric properties, the AQOL is recommended for use in epidemiologic studies where health-related QoL and utility data are required from older populations [33]. The AQOL has been used in Australian community older persons and Indigenous Australians studies [25].

The time horizon of the economic analysis will incorporate the lifetime of the cohort, through extrapolation of quality-adjusted survival from the end of the study period. A discount rate of 5% per year will be applied to both costs and outcomes [34]. The difference in costs and QALYs, between the intervention and pre-intervention periods, will be combined into the incremental cost-effectiveness ratio (ICER) with valuations of the willingness to pay for a marginal QALY to estimate the net monetary benefits for a decision to adopt the intervention:where Cost_Int(Cost_{Pre − int}) and QALY_Int (QALY_{Pre − int}) are the mean cost and QALYs gained in the intervention (pre-intervention) periods, respectively. The maximum willingness to pay for a QALY will be assumed to be AUS$64,000 for the Australian setting [35]. The Markov state transition process will be applied to model participants’ probability of being readmitted to hospital, staying in the community or dying [36]. Sensitivity analysis will be performed to test variations in key parameters including the study perspective, the choice of the comparator and specific cost impacts.

The human rights and dignity of the participants will be protected in accordance with Australia’s National Health Medical Research Council (NHMRC) guidelines [37]. Participants will not receive any financial inducement to participate. Participants will be provided with contact details of the project team to enable them to address any potential queries or concerns. Aboriginal and Torres Strait Islander participants will be offered the opportunity to have an Aboriginal and Torres Strait Islander Liaison Officer or Health Worker present during their CGA and care plan development. To conform to the Data Protection and Freedom of Information Act, all data will be stored securely and de-identified for the analysis. GPs will have access to the information related to the delivery of the study, including geriatric and enablement officer assessment, planning and progress notes. These documents will be sent securely to the GP to be uploaded to the patient’s medical record. Additionally, as a specialist medical officer, the geriatrician will have access to the GP medical notes into which they will write a summary of geriatric assessment, at the time of client appointment. All aspects of the assessment and patient care planning will be discussed with the GP at the case conference. Participants will be informed of their right to access their own results, and the overall results of the research. No published material will contain patient identifiable information. Any complaints will be systematically recorded and acted upon in accordance with NHMRC guidelines [37].

The study populations are, by definition, at risk of natural functional and cognitive decline that may be marked over the study period. Although the study design applies multiple methods to best capture intervention effects, the benefits in slowing down deterioration may not be successful in some participants due to their frailty. Within the quantitative methods are cross-sectional and case–control (pre–post intervention) analysis, which will assess between and within clusters differences. The pre–post analyses will utilise paired analyses where each participant acts as their own control. These methods plus the inclusion of qualitative surveys should tease out even minor benefits of OPEN ARCH, when presented with accelerated decline. The sample size calculation has allowed attrition, so these factors should allow statistical significant effect size within OPEN ARCH achievable.