Protocol for a randomised controlled trial of an outreach support program for family carers of older people discharged from hospital

Trial design

This is a mixed methods study addressing objectives and testing hypotheses in a single
blind Randomised Controlled Trial (RCT) (concealed allocation), adhering to CONSORT guidelines for transparent reporting
16], during which Qualitative Evaluation and FECH program Process Evaluation will also be undertaken. The study will also be supported by an expert reference
group. The RCT will compare outcomes between the control condition (usual care) and
the experimental condition (usual care plus family carer inclusion on the new FECH
program). Approval has been received from the Sir Charles Gairdner Group Human Research
Ethics Committee (HREC) (2014–133), the Department of Health WA HREC (2014/78), and
from the Curtin University HREC (HR14/2015).

Study setting and participants

The MAU is a 36-bed unit providing intensive assessment and treatment for patients
with acute medical conditions, most of whom are over 70 years of age. MAUs are widely
implemented in the UK, Australia and New Zealand in response to the need to increase
patient through-put 17] and sustained health system pressures suggest that their use will continue to expand.
Care is provided by physician-led multidisciplinary teams for up to 72 h, after which
patients are discharged or transferred to inpatient units for ongoing management.
A key component of the MAU model is that ongoing management for discharged patients
is provided through prompt general practitioner follow-up 9].

Included participants will comprise dyads of older people (aged 70+) discharged home
from the MAU during the recruitment period and their adult (aged 18+), English speaking,
family carers (one per patient). The definition of a family carer is that used in
a recent Australian study investigating the care of frail older people: “a family
member or friend who provides unpaid personal care, support and assistance” 18]. Dyads already recruited into the study will be excluded from further recruitment
upon any subsequent readmissions to the MAU.

Experimental condition (inclusion in the FECH program)

Carer participants in the intervention group will receive standard care plus the FECH
program implemented by the FECH nurse who will be appropriately skilled and trained.
Currently there is no nurse involved in providing the FECH or any similar program
to patients or carers.

Contacts will be made by telephone, after discharge (Fig. 1). The FECH Nurse will make initial contact with the family carer within one week
of the discharge (Contact 1). This contact will include the FECH nurse introducing
him/herself, explaining the intervention, and scheduling ‘Contact 2’. Contact 2 is
likely to comprise a series of brief telephone interviews, if this is the arrangement
that suits the family carer best, within the subsequent few days to (a) determine
and respond to the extent to which the family carer understands the copy of the discharge
letter to the general practitioner that has been provided to them; (b) administer
the CSNAT 14] to the family carer, resulting in the carer’s self-identified and highest prioritised
support needs; and (c) initiate responses to the three prioritised needs, helping
ensure family carers’ linkage and engagement with appropriate existing resources.
A few days after Contact 2 is finalised (within 14 days of the discharge), the FECH
nurse will check to determine if access to support has been achieved as planned, advising
as appropriate (Contact 3). This contact will include asking if the services have
been sourced (eg, an appointment made) and/or if the service has been accessed. These
three contact points will be coordinated with the three data collection time points
(T1, T2, and T3) also shown in Fig. 1.

Fig. 1. Data collection and intervention points (experimental condition, family participants)

Control condition (usual care)

Standard discharge ‘care’ includes the provision of a letter from the MAU’s physician
to the patient’s general practitioner, with a copy provided to the patient. Medications
are provided/organised by the MAU pharmacist. The MAU information booklet provides
information about post-discharge support options and is intended to be taken home.
Carers regarded as ‘at particular risk’ by the social work team receive social work
assessment and links to services. Services put in place for patients may include a
variety of care packages or programs. Information packs from Carers Australia are
made available in the hospital rooms. Any one or more of these options considered
by treating ward staff to be relevant to the patient and their carer will be provided
as usual care.

Carer outcome and data collection

Primary outcome variable

The primary outcome is preparedness for caregiving, to be measured with the Preparedness
for Caregiving Scale (PCS) from the Family Care Inventory 19] (T1- T3). This measure is an 8-item scale, with five response options (0?=?not at
all prepared, 4?=?very well prepared) and is designed for use with carers of older
people receiving homecare or experiencing care transitions. Construct validity has
been established in older people, and the measure has been shown to be reliable (Cronbach’s
alpha coefficients: 0.88-0.93) 19]. Testing in patients with a life threatening illness confirmed satisfactory internal
consistency reliability as well as unidimensionality 20].

Potentially confounding variables

Data on potentially confounding variables will be collected to help inform interpretation
of results. Carer rated patient Symptom Assessment Scale scores 21] and carer rated scores evaluating dependence of the patient in ten activities (Barthel
Activities of Daily Living Index) 22] will be collected (T1-T3). Robust psychometric properties have been documented for
these tools. Carer resilience will be measured at T1 using the 10-item Connor-Davidson
Resilience Scale 23].

Secondary outcome variables

Based upon recent findings using the CSNAT tool 24], caregiver strain will also be measured at each time point using the 25-item Family
Appraisal of Caregiving Questionnaire 25]. Carers’ ratings of their own health will also be collected (T1-T3) using the SF12
Version 2, a tool exhibiting robust psychometric properties with older people 26]; this tool has been used in large scale surveys such as those with veterans in the
United States 27].

Data collection

A Research Officer (RO1) will develop a schedule for collecting data from all participating
family carers (both groups) by telephone, at the time of recruitment to the study.
At baseline assessment (T1), RO1 will also collect demographic details of the patient
and the carer from the carer, including the carer’s familial relationship with the
patient, age and gender (carer and patient), highest education level, occupation,
usual place of residence, country of birth, years in Australia, support provided for
the patient, contact with the patient, length of time caring for the patient, education
received to help in the caring role, location of care, whether currently living with
patient (and if so, whether in patient’s or carer’s home), any care provided by others,
and known current medical conditions (patient and carer). Brief, robust measures to
minimise questionnaire burden will then be administered (each contact is estimated
to last approximately 30 min). Questionnaire administration across the three assessments
(T1, T2, and T3) is summarised in Table 1.

Table 1. Planned questionnaire administration at each time point

Patient outcomes and data collection

Quantitative patient outcome data will address Western Australian health care system
utilisation and will be obtained through linked administrative health data provided
via the WA Data Linkage System (WADLS) 28]. These Data Linkage Unit (DLU) data will capture reason for service and mode of transport
plus: date of service, symptom/presenting problem, and triage code (ED presentation);
admission and separation dates, primary diagnosis code, co-diagnosis code, and diagnosis
related group (hospitalisation data). In addition, to account for time at risk, and
to capture deaths resulting from acute events during the study follow-up period, date
and cause of death will be collected.

The data retrieved from data linkage will cover the time period from the date of the
index separation of the patient whose carer is the first to be recruited to the study
until 3 months after the index separation of the patient whose carer is the last to
be recruited. Since recruitment is planned to last approximately 6 months, this time
period will cover approximately 9 months.

Sample size

The primary outcome is the total score on the Preparedness for Caregiving Scale 19]. An improvement of two points in the total score is regarded as clinically relevant,
given that this would mean a change such as progressing to ‘very well prepared’ from
‘well prepared’ in 25 % of items. To detect a change of this magnitude with 80 % power
(assuming that the standard deviation of the change in mean score is approximately
0.5, as in previous work) 19], 29] a sample size of 63 per group will be required. Based upon previous studies by the
research team, attrition of approximately 30 % is expected so a sample of 180 carers
in total will be recruited.

Recruitment and randomisation

Figure 2 shows the proposed study flow. There were more than eight MAU discharges of patients
aged 70+ each day in 2013, approximately half returning home (estimated 120 per month).
Therefore, recruitment of 180 participants is considered achievable within the 6 month
recruitment period (30 per month).

Fig. 2. Modified consort diagram to illustrate trial study flow and participant numbers

Although the primary study outcome relates to carers of patients discharged from the
MAU, other outcomes relate to the former patients themselves so carer/patient dyads
will be recruited. Whenever a patient aged 70 years or older is admitted to the MAU
during the recruitment period, family carers – and, when appropriate, patients – will
be provided with information about the project and asked to provide written consent
to participate in the study by RO1. However, many patients are likely to be too unwell
to be approached with information about the study during their hospitalisation and
are likely to experience ongoing poor health that may also fluctuate 9]. Others will lack (cognitive) capacity to consider consenting. In these instances,
data about patients’ use of health care services will be sought via a waiver of consent.
An opt out form will be provided for such patients in case there is a later opportunity
– if and when the patient’s health has improved – for them to withhold consent for
inclusion of their data in the study, should this be their wish.

To facilitate the assignment to group (control or intervention) of participants, a
list of treatment allocations will be prepared before the study commences; it will
contain a study number (whole number, starting from one), and a code to indicate the
group (intervention or control). The list of codes will be obtained using a sequence
of computer-generated random numbers, and organised so that recruitment to the two
study arms occurs at an approximately equal rate (using a permuted random block strategy).
Allocation of participants to their treatment group will occur as follows: as each
new dyad is recruited to the study, the next study number will be allocated to them
(next in sequence on the list), and the treatment allocation for that study number
will be read from the list.

Qualitative feedback from patients and family carers

An ‘exit’ telephone interview will be conducted with an estimated 30 carers in the
intervention group and, when possible, the older people receiving care from them (estimated
20). These participants will be purposively sampled on the basis of questionnaire
data to cover a wide variety of carer profiles (eg, carer sex, age [older versus younger
carers], duration of caregiving [new carers versus those who are more experienced],
relationship to patient [husbands, wives, sons, daughters, others], and location of
care). The sample will be extended if issues need further exploration and until data
saturation is reached.

Interviews will occur within two weeks following administration of the final measures,
to: investigate how the FECH model affected the participants, assess the support provided,
identify factors influencing feasibility/usefulness, and assess how the FECH model
could be improved. Written informed consent for participation will be obtained separately
for this step and will involve mailing the information sheet and consent forms (for
the carer and for the patient); then following up with a phone call during which questions
can be answered, before requesting the return of the completed consent form in a prepaid
study envelope. It is anticipated that the interviews will take approximately 30 min.
A second telephone call will be offered if the interview takes longer than 30 min
and the carer wishes to provide further relevant information.

Qualitative feedback from the hospital staff

A focus group interview will be convened with the MAU staff as soon as carer recruitment
for the trial has been completed. The staff working on the unit during the trial will
be invited to take part and will be asked to comment on any impact from the FECH program
on the Unit and to suggest: program refinements that would minimise negative outcomes
and maximise those that are positive, how the program could be integrated into practice,
and how its sustainability could be addressed. The staff members will be invited to
take part by mail, and will also be sent an information sheet and consent form.

Process evaluation

Evaluation of FECH program processes will involve the FECH nurse documenting, during
the trial: (a) adherence to, or deviation from, planned FECH processes; (b) information
provided to carers and the resources to which they were referred; (c) the extent to
which carers engaged with resources; (d) the contextual factors that were barriers
to, or facilitators of, resource and service access and engagement; and (e) costs
issues – primarily time taken to implement processes. This documentation will be achieved
using an electronic database developed at the same time as the process manual and
completed by the FECH nurse as the process is implemented.

Blinding

Group assignment will be concealed from RO1. When RO1 conducts the T1-T3 data collection
telephone questionnaire administration, he or she will commence the conversation with
a request to the participant (the carer) not to mention any phone calls from other
study personnel. This reminder will not always be sufficient to prevent disclosure,
and instances when it occurs will be documented to inform study reports. In a previous
study, quantification of any ‘unblinding’ involved asking the hospital staff to identify
to which group they thought patient participants had been assigned 30]. In this study, a similar question will be asked of the research staff collecting
outcome data. The person conducting the qualitative interviews will be a Research
Officer employed for this purpose alone (RO2), so that blinding to group allocation
will be maintained for RO1. Investigator blinding will also be maintained by allocating
responsibility for this (qualitative) component of the study to one of the research
team, an experienced qualitative researcher who will not be involved in quantitative
data collection or analyses.

Statistical analysis

Data analyses will be performed using the IBM SPSS statistical software package 31]. A p-value of 0.05 will be taken to indicate a statistically significant result
in all tests.

Baseline data

Standard descriptive statistics (means, standard deviations, medians for continuous
variables and frequencies and percentages for variables measured on a categorical
scale) will be used to summarise the characteristics of participants in the study
(primarily demographic data, but also including some data relating to the care recipient
to describe the caregiving situation). Chi-square tests and t-tests will be used to
compare the treatment groups (intervention versus control) on the basis of these variables.
It is anticipated that there will be no significant differences between groups in
terms of demographic characteristic or potentially confounding variables, which will
confirm that the randomisation process has allocated participants evenly to each arm
of the study.

Carer outcome

Outcome analysis will be conducted using an Intention to Treat approach. The change
in the total Preparedness for Caregiving Scale 19] score between T1 and T3 will be calculated for each carer. An initial t-test will compare changes between treatment groups. If changes in scores are not
normally distributed, a non-parametric method will be used instead (Wilcoxon 2-sample
test). In the (unlikely) event that groups differ on the basis of baseline variables,
a regression model will be used instead of the t-test, so that these potentially confounding variables can be taken into account.
A random effects regression model will be used to examine changes in the Preparedness
for Caregiving Scale score over all three data collection periods (instead of just
baseline/end of study). This model will be used so that correlations between measurements
made on the same participants can be taken into account. Results of the model will
help show whether changes in scores happen soon after recruitment, or later. The model
can be extended to adjust for other potentially confounding variables.

Patient outcomes

Patient outcomes will be compared between groups using t-tests or regression models.
For example, the total length of hospital stay will be calculated and compared between
groups using a t-test (after log-transformation of the data if indicated). Number of re-admissions
to hospital or presentations to ED will be compared using the same method. A regression
model will be used to take into account potentially confounding variables (if relevant).
The association between change in preparedness and change in health care utilisation
will be evaluated using carer outcomes linked to the administrative data.

Economic costs and associated analysis

Since a randomised controlled trial is being undertaken to determine the efficacy
of the new support program for older patients’ carers, the economic evaluation reflects
a service substitution model without cost sharing or transfer. Costs of the program
will be evaluated using prospective data collection for each patient/carer dyad (cases
and controls) and will include the costs associated with the intervention and outcomes
using a Western Australian health system perspective.

Program costs

Program costs in addition to those for usual care will include those documented during
the process evaluation, in particular: (1) training of the FECH nurse to implement
the program; (2) the FECH nurse salary; (3) costs associated with providing discharge
information to carers (costed on a case by case basis) – FECH nurse time and web access
(to identify service providers), any costs for communication that is part of the intervention,
and stationery used. Costs associated with research components/activities will be
excluded since they would not be incurred if the program was taken up. Outcome costs
associated with utilisation of health services during follow up for patients (cases
and controls) will include: (1) cost of ED visits, (2) cost of additional in-patient
hospitalisation, and (3) cost of ambulance use.

In-patient costs

In-patient costs will be calculated using Diagnostic Related Group (DRG) based costings
from the appropriate cost report available from the Australian Government Department
of Health. Since Urgency Related Group (URG) is not available directly from the ED
data, cost of ED attendances will be based on derivation of the closest urgency related/disposition
group using data obtained from the Standard Emergency Record Information. The derivation
algorithm follows as closely as possible the URG Grouper application developed by
the Independent Hospital Pricing Authority (IHPA) which uses Episode End Status, Type
of Visit, Triage, Sex, and Diagnosis Code. Costing will then be undertaken using the
price weight of the URG and National Efficient Price provided in the closest IHPA
National Efficient Price Determination report to the date of the episode 32].

Patient ambulance utilisation

Patient ambulance utilisation will be costed at $AUD898 per service. The use of ambulance
services will be obtained via the hospital mortality data system data (source of referral
transport) and ED (arrival type-transport mode) data sets.

Analysis

Costs and outcomes associated with delivering the intervention will be compared using
cost-consequence analysis, a variant of cost-effectiveness analysis in which the components
of incremental costs and outcomes are computed and listed without aggregating these
results into an overall ratio. Cost-consequence analysis provides a more comprehensive
presentation of information than other types of economic evaluation and is appropriate
for complex interventions that generate outcomes that cannot meaningfully be expressed
using a single metric such as those in this study. Consequences (outcome measures
as described above under patient and carer data) and net costs (cost of intervention
minus cost savings produced by the intervention) will be tabulated to allow analysis
of incremental cost per net change for each outcome.

A decision tree analysis using TreeAge Pro 2015 33] will evaluate cost-consequence separately for each outcome. Decision trees are widely
used to illustrate the conceptual model of a cost effectiveness analysis. The tree
begins with a decision node depicting treatment options (intervention versus usual
care) for study participants. Each option becomes a main branch off the box, which
further divides into smaller branches at a ‘chance node’ as certain pre-defined events
(outcomes) occur. Decision trees illustrate both the probability of each outcome and
the costs associated with the resultant outcome. The likelihood of each consequence
is expressed as a probability of occurrence and cost calculated from trial data. Thus
it will be possible to calculate expected cost and expected outcome of each option.
For a given option, the expected cost is the sum of costs of each consequence weighted
by the probability of that consequence.

As with all research, this study will incorporate assumptions, uncertainties, and
variability in the data. To evaluate the importance of the assumptions and uncertainties,
comprehensive sensitivity analyses (deterministic and probabilistic) will be performed.
Sensitivity analyses strengthen economic evaluations by indicating the stability of
the reported outcome and identifying which variables have the most influence on it.
This step will provide an estimation of likely generalisability of the cost-consequence
estimate outside the trial. Examples of uncertainties and variables are in Table 2. Where average costs are used, best and worst case scenarios will be investigated
for potential impact on cost-consequence. Effectiveness variations (eg, patient risk
profiles) will be considered but most model variations will be performed on cost variables.

Table 2. Variables to be included in the sensitivity analysis

Qualitative analysis

All interviews will be audio-recorded and transcribed verbatim. Analysis will be aided
by the use of the NVivo software program 34]. Responses will be coded on a question-by-question basis, codes will be collapsed
into themes across questions, and quotes from participants will illustrate each theme.
This qualitative analysis will be undertaken independently by the investigator responsible
for overseeing qualitative evaluations and the interviewer to ensure consideration
of the non-verbal context. These analysts will compare themes and sub-themes until
agreement is reached.

Analysis for process evaluation

Findings from the process evaluation will be summarised to support recommendations.
In addition, a feasible ‘caseload’ for a full-time FECH nurse will be determined.