Time‐limited home‐care reablement services for maintaining and improving the functional independence of older adults
Abstract
Background
Reablement, also known as restorative care, is one possible approach to home‐care services for older adults at risk of functional decline. Unlike traditional home‐care services, reablement is frequently time‐limited (usually six to 12 weeks) and aims to maximise independence by offering an intensive multidisciplinary, person‐centred and goal‐directed intervention.
Objectives
To assess the effects of time‐limited home‐care reablement services (up to 12 weeks) for maintaining and improving the functional independence of older adults (aged 65 years or more) when compared to usual home‐care or wait‐list control group.
Search methods
We searched the following databases with no language restrictions during April to June 2015: the Cochrane Central Register of Controlled Trials (CENTRAL); MEDLINE (OvidSP); Embase (OvidSP); PsycINFO (OvidSP); ERIC; Sociological Abstracts; ProQuest Dissertations and Theses; CINAHL (EBSCOhost); SIGLE (OpenGrey); AgeLine and Social Care Online. We also searched the reference lists of relevant studies and reviews as well as contacting authors in the field.
Selection criteria
We included randomised controlled trials (RCTs), cluster randomised or quasi‐randomised trials of time‐limited reablement services for older adults (aged 65 years or more) delivered in their home; and incorporated a usual home‐care or wait‐list control group.
Data collection and analysis
Two authors independently assessed studies for inclusion, extracted data, assessed the risk of bias of individual studies and considered quality of the evidence using GRADE. We contacted study authors for additional information where needed.
Main results
Two studies, comparing reablement with usual home‐care services with 811 participants, met our eligibility criteria for inclusion; we also identified three potentially eligible studies, but findings were not yet available. One included study was conducted in Western Australia with 750 participants (mean age 82.29 years). The second study was conducted in Norway (61 participants; mean age 79 years).
We are very uncertain as to the effects of reablement compared with usual care as the evidence was of very low quality for all of the outcomes reported. The main findings were as follows.
Functional status: very low quality evidence suggested that reablement may be slightly more effective than usual care in improving function at nine to 12 months (lower scores reflect greater independence; standardised mean difference (SMD) ‐0.30; 95% confidence interval (CI) ‐0.53 to ‐0.06; 2 studies with 249 participants).
Adverse events: reablement may make little or no difference to mortality at 12 months' follow‐up (RR 0.97; 95% CI 0.74 to 1.29; 2 studies with 811 participants) or rates of unplanned hospital admission at 24 months (RR 0.94; 95% CI 0.85 to 1.03; 1 study with 750 participants).
The very low quality evidence also means we are uncertain whether reablement may influence quality of life (SMD ‐0.23; 95% CI ‐0.48 to 0.02; 2 trials with 249 participants) or living arrangements (RR 0.92, 95% CI 0.62 to 1.34; 1 study with 750 participants) at time points up to 12 months. People receiving reablement may be slightly less likely to have been approved for a higher level of personal care than people receiving usual care over the 24 months' follow‐up (RR 0.87; 95% CI 0.77 to 0.98; 1 trial, 750 participants). Similarly, although there may be a small reduction in total aggregated home and healthcare costs over the 24‐month follow‐up (reablement: AUD 19,888; usual care: AUD 22,757; 1 trial with 750 participants), we are uncertain about the size and importance of these effects as the results were based on very low quality evidence.
Neither study reported user satisfaction with the service.
Authors' conclusions
There is considerable uncertainty regarding the effects of reablement as the evidence was of very low quality according to our GRADE ratings. Therefore, the effectiveness of reablement services cannot be supported or refuted until more robust evidence becomes available. There is an urgent need for high quality trials across different health and social care systems due to the increasingly high profile of reablement services in policy and practice in several countries.
PICOs
Plain language summary
Time‐limited home‐care reablement services (up to 12 weeks) for supporting older adults to live independently
Review question
We aimed to assess the effectiveness of time‐limited reablement for older people (aged 65 years or more) in helping them to maintain or improve their independence. We included two studies in the review.
Background
Services that help older people to remain living in their own home have obvious appeal for service‐users, family members, care‐providers and policy makers alike, especially if those services help to reduce pressure on hospitals or the need for long‐term care, or both. Reablement (or restorative care) is one potentially useful service that helps an older person to continue living at home. The service is typically provided by a team of health/social care professionals and care‐workers who work with an older person to restore their independence. The service is time‐limited (usually six to 12 weeks) and normally involves multiple visits to a person's home. It sets out to achieve goals set by the older person, and help them to regain ability to complete everyday tasks and activities.
Study characteristics
The evidence is current to April 2015. The review included two studies, one each from Australia (750 participants) and Norway (61 participants). In both studies, half of the participants received a reablement‐based home‐care package and half usual home‐care provision.
Key results
The very low quality evidence for all of the results means that we are uncertain about the effects of reablement when compared with usual care.
Reablement may help some older adults to improve their abilities to engage in everyday activities (functional status) to a small degree, but may make little or no difference to death rates or admissions to hospital. The findings mean we are also uncertain whether reablement affects quality of life or living arrangements. Reablement may lead to a small decrease in numbers of people needing higher levels of personal care, and may decrease care costs to a small degree, but neither study reported satisfaction of those using the reablement service.
Quality of the evidence
While there may be some small positive effects of reablement, the evidence was very low quality, meaning that we are very uncertain about how large or important these effects may be. There is a need for more studies to be conducted in a range of countries and situations before the effectiveness and safety of reablement can be determined with certainty.
Authors' conclusions
Summary of findings
| Reablement compared with usual home‐care for maintaining independence | ||||||
| Patient or population: adults aged ≥ 65 years Settings: clients' own home Intervention: reablement services Comparison: usual home‐care service | ||||||
| Outcomes | Illustrative comparative risks (95% CI) | Effect estimate (95% CI) | No of participants (studies) | Quality of the evidence (GRADE) | Comments | |
| Assumed risk* | Corresponding risk | |||||
| Usual care | reablement | |||||
| Functional status** ADL and IADL (Lewin 2013) COPM (Tuntland 2015) Lower scores indicate greater independence Follow‐up: 9 to 12 months | ‐ | The mean score in the intervention group was 0.3 SD lower (0.53 to 0.06 lower) | SMD ‐0.30 (‐0.53 to ‐0.06) | 249 (2 studies) | ⊕ooo very low a,b | ‐ |
| Mortality*** Follow‐up: 9 to 12 months | 198 per 1000 | 6 fewer per 1000 (51 fewer to 57 more) | RR 0.97 (0.74 to 1.29) | 811 (2 studies) | ⊕ooo very lowa,c | ‐ |
| Unplanned hospital admission** Follow‐up: 24 months | 707 per 1000 | 42 fewer per 1000 (106 fewer to 21 more) | RR 0.94 (0.85 to 1.03) | 750 (1 study) | ⊕ooo very lowa,c,d | ‐ |
| Quality of life** AAQ (Lewin 2013) COOP/Wonka (overall health; Tuntland 2015) Lower scores indicate improvement Follow‐up: 9 to 12 months | ‐ | The mean score in the intervention group was 0.2 SD lower (0.48 lower to 0.02 higher) | SMD ‐0.23 (‐0.48 to 0.02) | 249 (2 studies) | ⊕ooo very lowa,b,c | ‐ |
| Level of emerging personal care needs (approved for higher level of care)*** Follow‐up: 24 months | 643 per 1000 | 84 fewer per 1000 (148 fewer to 13 fewer) | RR 0.87 (0.77 to 0.98) | 750 (1 study) | ⊕ooo very lowa,d | ‐ |
| Living arrangements (transferred to residential care) Follow‐up: 12 months | 128 per 1000 | 10 fewer per 1000 (49 fewer to 45 more) | RR 0.92 (0.62 to 1.34) | 750 (1 study) | ⊕ooo very lowa,c,d | ‐ |
| Cost effectiveness*** Total aggregated costs for home‐ and healthcare (emergency department and unplanned hospital admissions) Follow‐up: 24 months | The mean costs were AUD 2869 lower for the reablement group compared with usual care (AUD 19,888 with intervention versus AUD 22,757 with control) | ‐ | 750 (1 study) | ⊕ooo very lowa,d | ‐ | |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). AAQ: Assessment of Quality of Life; ADL: activities of daily living; CI: confidence interval; COPM: Canadian Occupational Therapy Performance; IADL: instrumental activities of daily living; RR: risk ratio; SMD: standardised mean difference; SD: standard deviation. | ||||||
| GRADE Working Group grades of evidence | ||||||
| * median control group risk calculated by GRADEpro. ** unpublished data. *** Data from Lewin 2014. a Downgraded twice (‐2) for risk of bias concerns (Lewin 2013 was at high risk of bias on all domains). b Downgraded once (‐1) for imprecision (fewer than 400 participants). c Downgraded once for imprecision (‐1) as confidence interval was likely to include minimally important difference and no difference). d Downgraded once for imprecision (‐1) as data were from single study. | ||||||
Background
Description of the condition
As the population ages and people live longer, the proportion of dependent older people is likely to increase (Brodsky 2003; Wittenberg 2004). As a result, the cost of long‐term care for people aged over 65 years living in OECD (Organisation for Economic Co‐operation and Development) countries is expected to double or even triple by 2050 (Martins 2006). Therefore, many high‐income countries have actively promoted a shift from residential to home‐based care as a potentially more effective and financially sustainable approach to meeting the health and social care needs of older adults (Rostgaard 2011). Importantly, most older people prefer to 'age in place' (Wiles 2012), and therefore, to remain in their own homes for as long as possible, provided they have appropriate levels of support to meet their changing needs (Cutchin 2009).
Government policies in various high‐income countries reflect the need to reconfigure health and social service provision in order to meet the current and future requirements of an ageing population. In England, for example, the Department of Health has articulated a vision for the integration of health and social care services, with a greater focus on individualised preventative services to delay the need for more costly forms of care (Xie 2012). Similar key policy goals of early intervention, person‐centred care and restoration of function have been identified and developed in Australia (Cartwright 2009), Sweden (Löfqvist 2012), and New Zealand (King 2012), mainly with a view to reducing pressure on health and social care systems. However, despite these changes, little is known about the effectiveness and cost‐effectiveness of models of care provision across different geographical and socioeconomic contexts.
Description of the intervention
In recent years, there has been increasing interest in high‐income countries in reablement (also known as restorative care in Australia and the USA) ‐ an innovative approach to improving home‐care services for older adults in need of care and support or at risk of functional decline (Francis 2011). There is a lack of clarity regarding the boundaries between reablement and other related interventions in health and social care (including intermediate care, occupational therapy and traditional domiciliary care) (Wood 2012). While reablement shares features with other interventions, it is distinguished by a re‐orientation of home care away from treating disease and creating dependency to maximising independence; it achieves this by offering intensive (i.e. multiple visits), time‐limited (typically six to 12 weeks' duration), multidisciplinary, person‐centred and goal‐directed home‐care services (Ryburn 2009). It is important to note that reablement is not designed to resolve specific healthcare issues (e.g. Crotty 2010), but may help an older person to regain confidence and functional abilities after recovering from an illness or a period of hospitalisation. Therefore, a reablement programme typically includes a range of targeted components designed to optimise functioning in the performance of activities of daily living (ADL). These may include exercise and training to support behavioural change, education about self management and healthy ageing, environmental adjustments, provision of equipment and use of local resources (Kent 2000; Lewin 2010; Tinetti 2002). So, for example, rather than providing a meals‐on‐wheels service, a reablement approach would enable an older person to develop the confidence and skills to prepare lunch through task analysis/redesign, the use of assistive technology and physical exercises (Glendinning 2010).
Thus, reablement differs from usual home care/domiciliary care, which tends to focus on doing things for older people rather than enabling/reabling them to do things for themselves. Indeed, traditional models of home care have been shown to increase dependency, with an associated loss of function (Parsons 2013). Furthermore, the assumption underpinning usual home‐care services is that they will continue indefinitely (Montgomery 2008), whereas reablement is specifically time‐limited and aims to reduce the need for home care into the future (King 2012; Ryburn 2009). Reablement, therefore, is particularly valued for its potential to decrease demand on home‐care services and to reduce the attendant costs of ongoing care (Jones 2009). Nevertheless, this form of care provision may have considerable resource implications in terms of retraining staff and effecting organisational change (Francis 2011).
The reablement approach has become increasingly popular and has been implemented widely in the UK (Department of Health 2010), as well as in a number of other countries (e.g. New Zealand (King 2012; Parsons 2013), Australia (Ryburn 2009), USA (Tinetti 2002)). The provision of reablement reflects a wider change agenda that promotes person‐centred care through individually tailored services that permit greater choice and control for consumers (Xie 2012). Additionally, the growth in this type of approach is in line with the increasing demands of people as they age; older consumers are becoming increasingly likely to demand greater choice, more personalised services and better quality home‐care support in the future (Rostgaard 2011).
How the intervention might work
The reablement approach emphasises the active participation of an older person in working towards agreed goals that are designed to maximise independence and confidence. For example, these goals might include regaining confidence in self care management and improving mobility. The content of the intervention may encompass graduated practice in completing tasks, environmental adjustments and adaptive equipment, or enabling an older person to build up social networks (Ryburn 2009). Improved outcomes across similar domains, including self care, mobility and quality of life (QoL), have been reported (Kent 2000; Tinetti 2002). Furthermore, the ability to function effectively in the home may reduce the need for unscheduled hospital admission, while postponing or preventing admission to residential care (Tinetti 2002). A reduction in the care hours required following the intervention is frequently used as a measure of success (Kent 2000; Lewin 2010), although this may not always be a desired or possible outcome for some older people, particularly people who are socially isolated or in failing health (Francis 2011). Arguably therefore, a decrease in hours of care with regard to older people with high dependency needs may not be an appropriate outcome measure. Importantly, additional outcomes that are valued by older people themselves as indicators of effective services should be measured (Clark 2001).
There may be different routes and thresholds for entry into a reablement‐based service. For instance, some hospital discharge support schemes select only older people who are most likely to benefit from the approach (i.e. people with relatively low levels of ongoing need), whereas a reablement service that accepts referrals directly from the community may adopt a more flexible approach and screen out only those people who are terminally ill or who have advanced dementia (Glendinning 2010). Nevertheless, it seems likely that outcomes will vary depending on the route of entry and also on the functional abilities of the older person on entry to the service. For example, people with a high level of need may not benefit as much as people with lower support requirements (Francis 2011). Indeed, reablement represents only one end of the continuum of care and may not be suitable for people with chronic or relatively intractable problems such as dementia that may require a different type of longer‐term service model (CSED 2007).
Why it is important to do this review
There has been strong international interest in developing effective and cost‐effective interventions to support older people living in their own homes and, in turn, to reduce the demand on acute hospital services and residential care provision. Arguably, a lack of, or poorly developed, rehabilitation services has contributed to increasing pressure on acute hospital beds, delayed discharge, more frequent re‐admissions to hospital, and increased use of costly residential and nursing home care (Audit Commission 2000). One approach to freeing up hospital beds is to support early discharge by providing acute care at home. For example, one Cochrane review of 'hospital at home' services found that older people with a mix of conditions were less likely to need residential care at follow‐up after receiving these services, although only a small proportion of older people were deemed to be eligible or were willing to take part (Shepperd 2011).
There is currently limited evidence as to which setting or model(s) of care may be most effective for the rehabilitation and maintenance of the independence of older adults (Huss 2008; Ward 2009). This appears to be due, in large part, to the challenges involved in comparing different, often multi‐component interventions across a range of settings. For example, Beswick 2008 (89 participants) and Huss 2008 (21 participants) reviewed a range of heterogeneous studies such as community‐based nursing care following discharge from hospital, falls prevention, group education and annual health assessments. These reviews concluded that, while multidimensional home‐based programmes had the potential to reduce the burden of disability among older adults, it was not possible, on the basis of the available evidence, to identify which one of the various models/types of care provision was the most effective. There is a need to undertake a more focused systematic review in order to assess the comparative effectiveness and disentangle the effects of each type of intervention and their potentially active ingredients or components.
While a number of previous Cochrane and non‐Cochrane reviews have examined a range of rehabilitation and home‐visiting programmes, as yet there has not been a systematic review that has focused specifically on the effects of reablement‐based interventions. Important questions about the effectiveness and cost‐effectiveness of these types of interventions remain unanswered. For example, does reablement reduce health service utilisation (such as hospital re‐admissions)? Do specific subgroups benefit more than others (e.g. younger populations), and people with lower levels of need? Is there evidence to support personalisation of the service? We undertook this review to try to address these important gaps in our knowledge provided that a sufficient number of randomised controlled trials (RCTs) were eligible for inclusion.
Objectives
To assess the effects of time‐limited home‐care reablement services (up to 12 weeks) for maintaining and improving the functional independence of older adults (aged 65 years or more) when compared to usual home‐care or wait‐list control group.
Methods
Criteria for considering studies for this review
Types of studies
Randomised controlled trials (RCT), cluster randomised trials and quasi‐RCTs of reablement compared to usual care (i.e. home‐care support, which included unpaid informal care) or wait list.
We deemed the inclusion of cluster randomised and quasi‐RCTs necessary to consider trials where individual random assignment may have been impractical due to the nature of the intervention (e.g. where only the reablement intervention was available in one geographical area or there may have been restrictions in terms of the availability of care staff to deliver either reablement or usual care).
We also planned to include studies that examined the costs or cost‐effectiveness of the intervention versus usual care, which had been conducted alongside, or subsequent to, trials that met the eligibility criteria (Shemilt 2011).
Types of participants
Older adults aged 65 years or more living in their own home who required assistance to perform tasks of daily living and to participate in normal activities due to poor physical or mental health. We excluded trials involving older adults living outside their own homes (e.g. in nursing homes). We anticipated that we could encounter trials with mixed populations because reablement is offered to younger people in some settings. We included trials with 80% or more older adults (aged 65 years or more) in the overall sample, and contacted study authors to determine the age profile in situations where younger people had been recruited into the trial.
Types of interventions
Reablement interventions compared with groups receiving usual home‐care services or with a wait list control group. Studies were required to meet the following criteria:
-
participants must have had an identified need for formal care and support or be at risk of functional decline (Francis 2011);
-
the intervention must have been time‐limited (up to 12 weeks) and intensive (e.g. multiple home visits) (Ryburn 2009);
-
the intervention must have been delivered in the older person’s own home, and provided by an interdisciplinary team (Glendinning 2010);
-
the intervention must have been focused on maximising independence; and
-
the intervention must have been person‐centred and goal‐directed (Parsons 2013).
We excluded trials that focused on the provision of acute care (e.g. nursing care in the home), or those describing interventions outside of existing home‐care services.
The control group was in receipt of, or awaiting, usual home‐care services, which may have been defined as ongoing assistance with completion of household activities or personal care (or both) by an outside agency (i.e. paid support) or informal (unpaid) care (or both), with or without professional input (e.g. nurses, occupational therapists). The control group could also have included people waiting for the intervention (wait list).
Types of outcome measures
We recognised the possibility that specific outcomes may have been measured using different tools across trials. Where we found studies with more than one relevant outcome per outcome category we selected the primary outcome identified by the publication authors. If no primary outcome was identified, we planned to select the one specified in the sample size calculations; if the sample size calculation was not stated, we would have ranked the effect estimates and selected the median effect estimate.
We only included studies that assessed functional outcomes (e.g. ADL).
Primary outcomes
-
Functional status including measures of the skills and abilities to complete ADL.
-
Adverse events including mortality, hospital (re)admission.
Secondary outcomes
-
Quality of life (QoL). We evaluated studies that assessed health‐related quality of life (HRQoL) or social care‐related quality of life (SCRQoL) (or both) using validated uni‐ or multi‐dimensional questionnaires. Examples of generic HRQoL questionnaires include the 36‐item Short Form (SF‐36) and EuroQol five dimensions questionnaire (EQ‐5D); SCRQoL measures include ASCOT (Adult Social Care Outcome Toolkit; Netten 2011).
-
User satisfaction.
-
Service outcomes, including level of ongoing home‐care service (e.g. care hours) or use of external health services (e.g. visits to emergency department).
-
Living arrangements (i.e. in own home or other setting).
-
Cost‐effectiveness (as measured by comparing the costs of the intervention versus usual care; and health service utilisation). Full economic evaluations of reablement interventions may be relatively rare (e.g. Pilkington 2011), and as such, we also included cost analyses, provided that these were conducted alongside, or subsequent to, trials that otherwise met the eligibility criteria.
Timing of outcome assessment
All outcomes measured at baseline and on discharge from the reablement service (typically six to 12 weeks). We analysed follow‐up at nine and 12 months (and longer) as available.
Main outcomes for 'Summary of findings' table
We prepared summary of findings Table for the main comparison based on the methods described in theCochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011), for the following main outcomes:
-
functional status;
-
adverse events (mortality; unplanned hospital admissions);
-
QoL;
-
service outcomes (level of ongoing/emergent personal care needs;
-
living arrangements (i.e. living in own home or elsewhere); and
-
cost‐effectiveness.
Search methods for identification of studies
We identified studies through: key word and text word searches of relevant electronic databases and government and non‐government agencies; searches of grey literature including conference papers, unpublished theses and reference lists of other rehabilitation reviews; and personal communications with experts in the field.
Electronic searches
We searched the following electronic databases with no language restrictions:
-
Cochrane Central Register of Controlled Trials (CENTRAL) (2015 Issue 5) (Appendix 1);
-
MEDLINE (OvidSP) (1945 to 21 April 2015) (Appendix 2);
-
Embase (OvidSP) (1988 to 4 April 2015) (Appendix 3);
-
PsycINFO (OvidSP) (1967 to 14 May 2015) (Appendix 4);
-
ProQuest (ERIC; Dissertations and Theses; Sociological Abstracts; earliest to 14 May 2015) (Appendix 5);
-
CINAHL complete (EBSCOhost) (1982 to 25 May 2015) (Appendix 6);
-
SIGLE (System for Information on Grey Literature in Europe) (1980 to 17 June 2015) (Appendix 7);
-
AgeLine (EBSCOhost) (1978 to 27 May 2014) (Appendix 6); and
-
Social Care Online (earliest to 20 May 2015) (Appendix 8).
Searches were up to date as of April 2015 (AgeLine was only available to us in May 2014); detailed search strategies are presented in the Appendices.
Searching other resources
We contacted key experts in the field and first authors of included studies for advice as to other relevant published, unpublished and ongoing studies (e.g. conference papers, unpublished dissertations, working papers or government reports) that might be eligible for inclusion. We searched reference lists of included studies and relevant reviews to identify further relevant studies.
We also searched online trial registers (Clinical Trials Register; ClinicalTrials.gov; WHO International Clinical Trials Registry) for ongoing and recently completed studies. Appendix 9 shows the search terms for these registers.
Data collection and analysis
Selection of studies
The lead author (AC) screened all titles and abstracts identified from searches to determine which met the inclusion criteria. We retrieved the full text of any papers identified as potentially relevant. Two authors (AC, MF) independently screened full‐text articles for inclusion or exclusion, with discrepancies resolved by discussion and by consulting a third author (SMcG), where necessary. We collated duplicate publications and considered these by individual study. All potentially relevant papers excluded from the review at this stage are listed as excluded studies, with reasons provided in the Characteristics of excluded studies table. We presented the available information about three ongoing studies in the Characteristics of ongoing studies table. The screening and selection process is outlined in a PRISMA flow chart (Figure 1; Liberati 2009).
Study flow diagram.
Data extraction and management
Two authors (AC, MF) independently extracted data from the included studies. We resolved any discrepancies by discussion until we reached consensus, or through consultation with a third author (SMcG), where necessary. We used an adapted data extraction form based on the Cochrane Consumers and Communication Group Data Extraction Template (available at: cccrg.cochrane.org author‐resources). We piloted the adapted form using three trials (two of which we subsequently excluded) before finalising the design. Data extracted included: aim of intervention, study design, sample size and attrition, description of the comparison group, all outcomes and funding sources. See Characteristics of included studies for full details. One author (AC) entered extracted data into Review Manager 5 including outcome data and results (RevMan 2014), and one author (MF) independently checked for accuracy against the data extraction sheets.
Assessment of risk of bias in included studies
Two authors (AC, MF) independently assessed and reported on the methodological risk of bias of the included studies in accordance with the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), and the Cochrane Consumers and Communication Group guidelines (Ryan 2011). Both recommended the explicit reporting of the following elements for RCTs: random sequence generation, allocation sequence concealment, blinding (participants and personnel), blinding (outcome assessment), completeness of outcome data (including data on attrition) and selective outcome reporting. We also considered other sources of bias: contamination, baseline comparability of groups and fidelity/delivery of interventions including any co‐interventions. We considered blinding separately for different outcomes. We judged each item as being at high, low or unclear risk of bias as outlined in the criteria provided by Higgins 2011, and we provided a quote from the study report as well as a justification for our judgement for each domain in 'Risk of bias' tables. If we had included any cluster‐RCTS we would have assessed and reported the risk of bias associated with selective recruitment of cluster participants and potential contamination between intervention and control groups. We would have assessed any quasi‐RCTs as high risk of bias on the sequence generation item of the 'Risk of bias' tool.
Two authors (AC, MF) independently assessed the risk of bias of the included studies, with any disagreements resolved by discussion to reach consensus. We contacted the study authors for additional information and for clarification of the study methods. We incorporated the results of the risk of bias assessment into the review by means of a standard table, and systematic narrative description/commentary about each of the elements, thereby providing an overall assessment of the risk of bias of the included studies.
With regard to the cost‐effectiveness analysis, we used the Drummond checklist to appraise the methodological quality of the included costs study critically (Shemilt 2011).
Measures of treatment effect
Dichotomous data
For dichotomous outcomes (e.g. living at home versus other location), we analysed data based on the number of events and the number of people assessed in the intervention and usual care group. We used these to calculate the risk ratio (RR) and 95% confidence intervals (CI) using a random‐effects model to analyse such data where pooled.
Continuous data
We planned to analyse continuous data (e.g. ADL) based on the mean and standard deviation (SD), and number of people assessed for both the intervention and usual care groups to calculate mean difference (MD) and 95% CI. However, the studies reported all continuous data using different scales for the same outcome (e.g. QoL). Therefore, we estimated SMDs and 95% CI and used the inverse‐variance method to analyse data in Review Manager 5 (RevMan 2014).
In cases where the mean and SD were not available in the published report, we obtained data from the study authors. If these had not been available, we planned to calculate effect sizes (e.g. from t tests, F tests or exact P values).
Time‐to‐event data
If we had encountered time‐to‐event (e.g. transfer to nursing home) data, we would have extracted the log of the hazard ratio (log(HR)) and its standard error from trial reports. If these were not available, we would have estimated the log(HR) using published methods (Parmar 1998; Tierney 2007). We would have pooled HRs using the generic inverse‐variance method of Review Manager 5 (Deeks 2011).
Economic data
We had planned to tabulate the characteristics of any health economic studies included by subgroups (i.e. full economic evaluations, partial economic evaluations and analyses reporting more limited information). We would only present a pooled estimate if there was evidence of little variation in resource or cost use between studies. As we were unable to pool data, due to the lack of eligible data, we presented a narrative summary for the single study regarding the design and analytical viewpoints adopted, the primary outcome measures used for the evaluation and resource‐use cost data (Drummond 1996; Table 1).
| Study design | Issue addressed | Explanation |
| 1. The research question is stated | Yes | Compared health and aged care service use and costs of a restorative home care service versus a conventional service |
| 2. The economic importance of the research question is stated | Yes | Investigated whether an effective intervention was also cost‐effective |
| 3. The viewpoint(s) of the analysis are clearly stated and justified | Yes | Multi‐agency perspective including providers of health and social (aged) care |
| 4. The rationale for choosing alternative programmes or interventions compared is stated | Yes | Intervention and usual care services funded by the Government and provided by a not‐for‐profit organisation |
| 5. The alternatives being compared are clearly described | Yes | Time‐limited individualised restorative service delivered in the home compared to usual personal care service |
| 6. The form of economic evaluation used is stated | No | Not explicitly stated, information provided on the use and cost of home‐care and healthcare services |
| 7. The choice of form of economic evaluation is justified in relation to the questions asked | Yes | A partial evaluation that provided costs data on service utilisation in both the intervention and control conditions. An ICER was not conducted |
| Data collection | ||
| 8. The source(s) of effectiveness estimates are stated | Yes | RCT (Lewin 2013): 750 participants (aged ≥ 65 years) eligible for government‐funded home care and allocated to a time‐limited (12 weeks maximum) reablement intervention or to usual home‐care services |
| 9. Details of the design and results of effectiveness study are given (if based on a single study) | Yes | See Lewin 2013 |
| 10. Details of the methods of synthesis of estimates are given (if based on a synthesis of a number of effectiveness studies) | N/A | ‐ |
| 11. The primary outcome measure(s) for the economic evaluation are clearly stated | Yes | Comparison of intervention and control groups on:
|
| 12. Methods to value benefits are stated | N/A | Not stated |
| 13. Details of the participants from whom valuations were obtained were given | Yes | ‐ |
| 14. Productivity changes (if included) are reported separately | N/A | Productivity costs not included |
| 15. The relevance of productivity changes to the study question is discussed | N/A | Productivity costs not discussed |
| 16. Quantities of resource use are reported separately | No | The mean hours of aged‐care services provided, but only mean cost per client for first year and total 2‐year period reported rather than cost per hour Number of emergency department visits provided, but only mean total costs per client for 2‐year period rather than cost per visit reported Number of hospital visits and mean length of stay (episodic and cumulative) provided, but only mean total cost per client for all hospital admissions reported |
| 17. Methods for the estimation of quantities and unit costs are described | Yes | Quantities data were obtained via the Western Australian Data Linkage System: the Emergency Department Data Collection; Hospital Morbidity Data System; the Mortality Register; the HACC database and the Aged Care Assessment Program database Unit costs: home care costs provided by the Western Australian DoH; emergency department costs provided by the National Hospital Cost Data Collection Cost Report (2007‐8); inpatient data were provided by the Public Sector Estimated Round 12 (2007‐8) AR‐DRG 5.1 Cost Report for Western Australia (DoHA 2008) |
| 18. Currency and price data are recorded | Australian dollars | |
| 19. Details of currency of price adjustments for inflation or currency conversion are given | No | ‐ |
| 20. Details of any model used are given | Yes | General linear model used to analyse aggregated health and social care costs over time (2‐year period) and adjusted for living arrangements, carer status, gender and dependency level |
| 21. The choice of model used and the key parameters on which it is based are justified | Yes | Aggregated health and social care costs presented |
| Analysis and interpretation of results | ||
| 22. Time horizon of costs and benefits is stated | Yes | Data collected for 3‐year period commencing 1 year prior to the date when the participant assigned to either intervention or control groups |
| 23. The discount rate(s) is stated | No | Not discussed |
| 24. The choice of discount rate(s) is justified | N/A | ‐ |
| 25. An explanation is given if costs and benefits are not discounted | No | None provided |
| 26. Details of statistical tests and confidence intervals are given for stochastic data | N/A | ‐ |
| 27. The approach to sensitivity analysis is given | N/A | Not conducted |
| 28. The choice of variables for sensitivity analysis is justified | N/A | Not conducted |
| 29. The ranges over which the variables are varied are justified | N/A | Sensitivity analyses not conducted |
| 30. Relevant alternatives are compared | No | ‐ |
| 31. Incremental analysis is reported | No | ‐ |
| 32. Major outcomes are presented in a disaggregated as well as aggregated form | Yes, in some cases only | Home‐care services, presentations to emergency departments and unplanned admissions were disaggregated. However, costs of the intervention itself were not disaggregated. The study authors reported that they were unable to include costs of residential/hospice care as the data they used related to approval for the service and there was no certainty that this translated into an actual admission |
| 33. The answer to the study question is given | Yes | Clients who received intervention were less costly to the aged and healthcare services over time than those who received standard home care, p. 334 |
| 34. Conclusions follow from the data reported | Yes | Statistical significance achieved more often in as‐treated analysis, suggesting that the success of the intervention depended heavily on compliance with the HIP [intervention] protocol. p. 334 |
| 35. Conclusions are accompanied by the appropriate caveats | Yes | Limitations and possible effects on the findings noted on p. 335. Specifically, authors were unable to match the dates of home‐care referral exactly with the financial year date of assessment or utilisation. Therefore, there may have been some over‐ or under‐estimation of the number of hours of service(s) the client used in each year. This measurement bias was likely to have affected the intervention and control groups equally |
AR‐DRG: Australian Refined Diagnosis Related Group; DoH: Department of Health; DoHA: Department of Health and Ageing; HACC: home and community care; HIP: Home Independence Program; ICER: incremental cost‐effectiveness ratio; RCT: randomised controlled trial.
Unit of analysis issues
We presented the relevant outcomes assessed at the end of the intervention (three months) and at follow‐up (nine, 12 and 24 months) separately.
If we had identified cluster RCTs for inclusion, we planned to check for unit‐of‐analysis errors. If we had found such errors and sufficient information was available, we planned to re‐analyse the data using the appropriate unit of analysis by taking account of the intracluster coefficient (ICC). We planned to contact study authors to obtain ICC estimates if these had not been clearly available from the trial reports, or to impute them using estimates from external sources (i.e. from a study of a similar population). If ICCs from other sources were used, we planned to undertake sensitivity analyses to investigate the effect of variation in the ICC. If it had not been possible to obtain sufficient information to re‐analyse the data, we planned to report the effect estimate and annotate unit‐of‐analysis error.
Dealing with missing data
We contacted the study authors to obtain data not included in the original article (i.e. means and SD of outcomes). We reported on the levels of loss to follow‐up and assessed this as a source of potential bias. We used intention‐to‐treat (ITT) data when available for our analyses. The ITT analysis in one study did not include all participants as randomised, and we contacted the study authors to determine if and how values for the missing data were imputed. The authors reported that they excluded cases if there was missing data and did not use any methods to impute these values (Lewin 2013).
We planned to conduct sensitivity analysis excluding studies with 20% or more of data missing for one of the primary outcomes to assess potential bias, but this was not possible due to the small number of included studies.
Assessment of heterogeneity
Where studies were sufficiently similar (e.g. based on considerations of population, intervention duration and intensity) to allow for pooling of data, we assessed the degree of heterogeneity by the visual inspection of the forest plots and by examining the Chi² test for heterogeneity. We quantified heterogeneity using the I² statistic (Higgins 2011). We considered an I² value of 50% or more to represent substantial heterogeneity, and interpreted it in view of size and direction of effects and the strength of the heterogeneity based on the P value from the Chi² test. If there was evidence of heterogeneity, we planned to discuss any possible reasons, and if there had been sufficient trials, we would have conducted subgroup analyses accordingly; the issue of sample size and power in each study would be considered in the interpretation and reporting of results.
If there had been substantial clinical, methodological or statistical heterogeneity across the included studies we would not have pooled results, but instead used a narrative approach to data synthesis. In addition, we would have explored possible clinical or methodological reasons for any variation to examine differences in intervention effects. Since the review included only two studies it was not possible to perform this analysis.
Assessment of reporting biases
We included insufficient studies to test for possible publication bias.
We had planned to assess reporting bias qualitatively based on the characteristics of the included studies (e.g. if only small studies that indicate positive findings are identified for inclusion), and if information that we had obtained from contacting experts and authors of studies suggested that there were relevant unpublished studies. If we had identified sufficient studies (at least 10) for inclusion, we planned to construct funnel plots to investigate any relationship between effect size and standard error. Such a relationship could be due to publication or related biases, or due to systematic differences between small and large studies. Where there was such a relationship, the methodological diversity of the studies was to be further examined as a possible explanation (Egger 1997). Findings were to be incorporated into 'Risk of bias' tables in the domain 'Other sources of bias'.
Data synthesis
We decided whether to meta‐analyse data based on if the included trials were sufficiently similar in terms of participants, interventions, comparisons and outcome measures to ensure meaningful conclusions from statistically pooled results. Due to the variability in the interventions and participants, we used a random‐effects model for meta‐analysis, with data analysis conducted in Review Manager 5 (RevMan 2014).
Lewin 2013 employed two different self report measures of functional status, ADL and IADL (instrumental activities of daily living), reported by the study authors as secondary outcomes. Therefore, to derive a summary effect estimate for this outcome from Lewin 2013, we calculated a mean effect size and standard error across both measures. We also calculated a mean effect size and standard error from the means, SDs and numbers of participants reported in Tuntland 2015 for their primary outcome (function measured using the Canadian Occupational Therapy Performance; COPM). Both summary estimates could then be pooled within the generic inverse‐variance analytic method, using the effect measure of SMD.
We calculated and reported the appropriate effect estimate (RR and 95% CI) where data were based on a single study.
summary of findings Table for the main comparison shows the findings for those outcomes most important to decision makers.
Subgroup analysis and investigation of heterogeneity
We were unable to conduct subgroup analyses due to an insufficient number of included studies. We had planned analyses based on the following subgroup parameters that emerged from the literature.
-
Context of recruitment to intervention. We anticipated that participants who had been recently discharged from hospital may have higher level of need or be at greater risk of re‐admission (or both) than participants recruited from the community, and thus some differences in outcome may emerge (Francis 2011).
-
Mean age of participants. There was some indication that younger participants (aged under 75 years) may gain greater benefit from reablement (Glendinning 2010). Therefore, we had planned to examine two groups, people aged 65 to 75 years and people aged over 75 years, to explore this effect.
-
Living circumstances (i.e. alone or with others). Isolated older people may experience the service differently from people with a higher level of support (Francis 2011).
-
Duration of intervention. Defined as standard (six weeks); long (seven to 12 weeks) as some trials may offer an extended period of reablement to meet individual needs (Jones 2009).
Sensitivity analysis
We were unable to conduct the planned sensitivity analysis due to a small number of included studies. We had planned to evaluate the robustness of any pooled effect sizes across various components of methodological quality to examine the robustness of the various effect estimates. We would have analysed the effects of excluding trials that were judged at high risk of bias across one or more of the domains of sequence generation, allocation concealment, attrition (rates larger than 20%) and outcome reporting (greater than 20% of data missing) for meta‐analysis of the primary outcomes. We planned to include the data if the exclusion of trials at high risk of bias did not substantially alter the direction of effect or the precision of the effect estimates. We would also have undertaken a sensitivity analysis to assess the effects of including trials where we used imputed values (e.g. ICC values from external sources for cluster‐RCT trials).
'Summary of findings' table
We presented results for each of the major outcomes as outlined in Types of outcome measures (Schünemann 2011). We provided sources and rationales for the assumed risk cited in the table. two authors (AC, MF) independently assessed the quality of the evidence using the GRADE criteria for each of the following parameters: risk of bias, inconsistency, imprecision, indirectness and publication bias using the GRADEprofiler (GRADEpro) software (Schünemann 2011). We downgraded a rating of high quality evidence by one level for serious concerns, and by two levels for very serious concerns. The findings for the outcomes where meta‐analysis was not possible are presented in summary of findings Table for the main comparison using a narrative format (Chan 2011).
Consumer participation
Consumer participation and content expertise were considered important because reablement reflects a partnership between the older person and the service providers. Therefore, feedback on the protocol was received from a consumer referee (a Cochrane volunteer), while feedback from referees with content expertise was sought at both protocol and review stages.
Results
Description of studies
Results of the search
Searches of electronic databases carried out between April and June 2015 yielded 11,507 abstracts. Handsearching of the reference lists within included studies and previous reviews yielded 809 references. After removal of 3463 duplicates, we screened titles and abstracts and identified 33 as being potentially relevant. Following assessment of the full text of these papers, and in some cases contact with study authors, we identified two studies (one with an associated cost evaluation) that met our eligibility criteria (Lewin 2013; Tuntland 2015). We identified three potentially eligible studies but these were ongoing or in the planning stages and findings not yet available (Grimmer 2013; Langeland 2015; Whitehead 2014); see Characteristics of ongoing studies table for further details. See Figure 1 for the PRISMA study flow diagram. We excluded the remaining 28 studies; see Characteristics of excluded studies table.
Included studies
The Characteristics of included studies table describes the main features of the two included RCTs. One study was conducted in Perth, Western Australia (Lewin 2013), and the other in a rural municipality of Norway (Tuntland 2015).
Participants
Lewin 2013 included 750 older adults (aged 65 years or more), eligible for home‐care services defined as needing assistance with one or more activity of living. Recruitment took place between June 2005 and August 2007. The mean age of the participants was 82.28 years (SD 7.45), and the sample was predominantly female (67.33%; 505/750). At baseline, over half of the intervention (57.8%) and usual care (67.7%) groups had a carer available, and more of the intervention (51.2%) group lived alone relative to the usual care group (42.4%).
Tuntland 2015 recruited 61 people between May 2012 and February 2014. The participants reported activity limitations and had been referred to home‐based services available for people aged 18 years and over. We contacted the study authors to determine the age profile of participants; 8.2% (5/61) of the full sample were aged 64 years or younger. The trial therefore met our inclusion criterion because more than 80% of included participants were aged 65 years or over. The mean age of participants was 79 years (SD 10.1), and most were female (67.2%; 41/61). The study authors did not report information related to living situation and carer availability.
Reablement and usual care
The interventions were similar in the two studies and in both cases there was an emphasis on encouraging participants to achieve individualised goals and to perform daily activities themselves rather than letting others do it for them. In addition, the intervention included exercises to improve mobility, adaptations to tasks and equipment, and strategies to promote social connectedness. Both interventions involved interdisciplinary teams including occupational therapists and physiotherapists, who conducted the initial assessments and developed the rehabilitation plan tailored to the aims and needs of each participant. Tuntland 2015 reported that the reablement service lasted 10 weeks on average, with a mean number of seven home visits per person per week lasting on average 2.1 hours (based on a 12‐week period). The reablement group received more home visits from therapists than the usual care group reflecting the enhanced emphasis on rehabilitation (Tuntland 2015). The participants in Lewin 2013 continued with reablement until they achieved their goals or for up to 12 weeks, whichever occurred first; additional information regarding the mean number of visits or duration of these visits was not provided. Neither study provided data on the percentage of participants who received the full three months of sessions, nor how many achieved their goals earlier.
There were no changes to the usual home‐care services for the control groups in either study; Lewin 2013 described "standard" home care as typically involving three visits a week to help with personal care (bathing/showering) and house cleaning. Tuntland 2015 described usual care as the "compensating help they applied for" and for most participants this consisted of personal or practical assistance, meals on wheels or assistive technology; the participants received, on average, six visits per week lasting 1.7 hours. The limited information provided suggested that usual home care appears to be broadly comparable across the two studies. However, participants in the Norwegian study appeared to have received twice as many visits per week relative to Lewin 2013. Additionally, the participants in the usual care group in Tuntland 2015 accessed a significantly higher amount of co‐interventions in terms of outpatient physiotherapy during the first three months compared to the reablement group. Lewin 2013 did not report on co‐interventions, so we were unable to make any direct comparisons between the studies on this issue.
Outcomes
The two studies used different tools to measure functional status, the primary outcome for our review. Lewin 2013 used standardised measures of function (ADLs and IADLs; lower scores indicated greater functional independence) for a subgroup of participants recruited by a research assistant from the full sample (150 in the intervention group and 150 in the usual care group), at three and 12 months. Tuntland 2015 employed the COPM to measure self perceived activity performance and satisfaction with performance on individualised activity goals at baseline, and at three and nine months. The COPM was used to enable participants to identify and prioritise problems with their self care or other activities (or both) at baseline; each participant rated their five most important activities on a 10‐point scale (higher scores indicated better function; scores were transformed by inserting a minus sign in the analyses to be consistent with data from Lewin 2013 ); the COPM informed the individualised components and targets of the intervention as well as providing the outcome measure. Therefore, for the outcome of functional status a decrease represented an improvement.
Tuntland 2015 also measured separate domains of HRQoL (COOP/Wonka; Weel 1993; rated 1 to 5, lower scores indicating better QoL) whereas Lewin 2013 used a single measure (Assessment of Quality of Life Scale; Hawthorne 1999; rated 0 to 1, where higher scores indicated better QoL; scores were transformed by inserting a minus sign to be consistent with Tuntland 2015). Therefore, a decrease represented an improvement in QoL.
The study authors provided us with additional data (means and SDs) on request that had not been reported in the published studies.
Both studies reported on mortality rates at nine to 12 months' follow‐up. A paper linked to the Australian study, Lewin 2014, provided additional data related to mortality and service use at the one‐ and two‐year follow‐up. Additionally, Lewin 2014 examined the costs for the intervention and usual care groups for home and healthcare service utilisation at one‐ and two‐year follow‐up, and calculated mean costs per participant across three outcomes:
-
use of aged care services;
-
visits to emergency departments and
-
unplanned hospital admissions
The studies did not report the costs associated with the implementation of the intervention itself when compared to usual care, or the costs associated with residential or hospice care.
Excluded studies
We excluded 28 studies; see Characteristics of excluded studies table. Reasons for exclusion included design issues, that is not randomised at either individual or cluster level (Glendinning 2010; Heebøll 2012; Kent 2000; Le Mesurier 1999; Lewin 2010; McLeod 2009; Newbronner 2007; Tinetti 2002; Tinetti 2012; Winkel 2015); intervention did not meet criteria for duration or intensity (or both) (e.g. over six months or a limited number of visits to the home (or both); Gitlin 2006a; Gitlin 2006b; King 2012; Parsons 2012; Parsons 2013; Sheffield 2013; Szanton 2011); intervention was not exclusively delivered at home (Crotty 2008; Cunliffe 2004; Nikolaus 1999; Senior 2014); intervention focused on medical rather than social care (Avlund 2002; Friedman 2014; Li 2013; Martin 1994; Melis 2008); study did not measure the primary outcome (Crawford Shearer 2010); or the control group received an alternative intervention (i.e. not usual care; Gill 2002). We also excluded economic studies that included costs data based on ineligible studies.
Risk of bias in included studies
We conducted a risk of bias assessment in line with the guidelines provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We sought further information from study authors to inform our assessment when necessary. See Characteristics of included studies table and Figure 2 for a summary. We used the Drummond checklist (Table 1) to assess any risk of bias of the costs paper (Lewin 2014); this information is presented in Other potential sources of bias, Economic evaluation).
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Overall, we rated Lewin 2013 at high risk of bias on all domains. Tuntland 2015 was deemed to be largely adequate, with issues around performance and detection bias and other sources of bias (use of co‐interventions and treatment fidelity) only.
Allocation
Lewin 2013 used a computerised system to randomise participants; however, the system was open to manipulation, and a 'handful' of operators had, in some cases, purposefully influenced the randomisation, based on the assumption that some participants (e.g. those living alone) might benefit more from allocation to either the intervention of usual care group; we judged this at high risk of bias. The recruitment of participants to subgroups in Lewin 2013 occurred after the initial group assignment; but there was insufficient information for us to be able to judge how these participants were selected from the full sample.
Tuntland 2015 reported appropriate randomisation and allocation concealment procedures, and so was rated at low risk of bias.
Blinding
It was not possible to blind participants or care providers (performance bias) to the allocated condition in either study due to the nature of the intervention, therefore both studies were rated as high risk of bias on this item.
Service outcomes
In Lewin 2013, the provider of home‐care services who collected data on service outcomes (e.g. ongoing personal care needs, hospital admissions) routinely were likely to be blind to group assignment. They used official records maintained by the Western Australian Data Linkage System to collect data relating to home and community care services as well as healthcare utilisation (Lewin 2014).
Functional status and quality of life
The research assistants involved in collecting outcome data from the subgroups in Lewin 2013 were initially blinded, but as the older people would often talk about the type of care they were receiving during the assessments, it was possible for research assistants to deduce which group participants had been allocated. Similarly, participants in Tuntland 2015 were urged not to reveal their group allocation to researchers, but the study authors reported this was only partly successful. Therefore, we rated both studies to be at a high risk of bias on this domain.
Incomplete outcome data
See the Characteristics of included studies table for details on incomplete outcome data.
Before the start of the Lewin 2013 trial, it had been agreed that any client randomised to receive the intervention who was not fully participating in the programme after two weeks would be reassigned to usual care; 20 participants (5.3%; 20/375) were transferred to the usual care group for this reason. A further 45 participants (6.0%; 45/750) were excluded from the as‐treated analysis because they did not receive sufficient levels of service (defined as three visits for the intervention group or three hours of personal care for the usual care group). Figure 1 in Lewin 2013 (page 74) suggested that all 45 participants were originally randomised to receive the intervention.
Dealing with missing data
Missing data due to mortality and participants' illness are commonly reported in research conducted with older populations. Imputing values for the deceased is inappropriate (Little 2002), but a number of different imputation approaches for ITT analysis have been proposed to address other sources of attrition by, for example, imputing missing values of the second follow‐up using information from both the baseline and first follow‐up (Ning 2013). There was no evidence from Lewin 2013 to suggest they used any specific statistical method to deal with missing data in their ITT analysis for personal care at three and 12 months or for the data collected for the functional status and QoL outcomes (provided by study authors), in which case this may be more correctly considered an 'available case analysis' (Higgins 2011). In our judgement, the missing outcome data were not managed appropriately for the purposes of an ITT analysis, so we rated this at high risk of bias.
Lewin 2013 did include all 750 participants in the descriptive data for mortality and service outcomes and Lewin 2014 included all 750 participants when reporting on the 12‐month and the full 24‐month period, and thus can be considered ITT analyses. Tuntland 2015 also included all participants in their ITT analyses.
The incomplete outcome data for the two studies are summarised below.
Mortality and service outcomes
We contacted the study authors to seek clarification on some minor differences in mortality rates reported in Lewin 2013 versus the costs paper Lewin 2014. The study authors indicated that the mortality data in the earlier paper were sourced from a database (maintained by the service provider) that was prone to occasional delays in updating, and that may therefore, have been less accurate due to some degree of under‐reporting. By contrast, the data used in the costs paper may be considered to be more reliable because these figures were obtained using a Mortality Register via the Western Australia Data Linkage System (Lewin 2014). This may explain the higher death rates reported in the costs study that was completed some time after the RCT.
The analysis for ongoing personal care included only 78.9% (592/750) of participants at three months and 63.1% (473/750) of participants at one year (Lewin 2013). People were excluded from the analysis if they: had died, moved either into residential care or out of the area, were terminally ill, declined follow‐up or had missing data for any included variable. The Characteristics of included studies table provides details on attrition. The incomplete data for service outcomes appear to be relatively comparable across the two groups, apart from 30 from the intervention group who declined follow‐up at three months (intervention: 8.0% (30/375); usual care: 2.4% (9/375)).
The ITT analysis conducted by Lewin 2014 included all participants, with the exception of people who had died in the first year and were removed from the second year analysis (intervention: 19.7% (74/375); usual care: 20.5% (77/375)). The ITT analyses for the full 24‐month follow‐up period included all 750 participants (Lewin 2014).
Functional and quality of life outcomes
Attrition was low for both groups in Tuntland 2015; we considered that missing data in the intervention group were unlikely to be due to adverse effects of the intervention, and thus rated the study low risk of bias for this domain.
Functional and QoL outcomes were examined for a subgroup of 40% (300/750) of the originally randomised sample (Lewin 2013). Furthermore, the analysis included only two‐thirds (66.7%) of the intervention subgroup (100/150) and a similar proportion (65.3% 98/150) of the usual care group. The proportions of missing data for functional (ADL and IADL) and QoL outcomes were similar for the intervention and usual care subgroups and at both time points (three and 12 months), with attrition rates ranging from 3.3% to 18.7%. These were related to participants who had declined follow‐up, whose health had deteriorated or who were no longer contactable. As these outcomes were assessed for a subgroup only, we rated this domain at high risk of bias.
See the Characteristics of included studies table for details on attrition,
Selective reporting
All outcomes proposed in the Tuntland 2014 protocol were reported and there was no evidence of selective reporting; we rated this domain at low risk of bias.
There was some indication of selective reporting in Lewin 2013; the outcomes for the subgroup (relating to both functional status and QoL) were not fully reported. However, we did receive unpublished data from the study authors. Some analyses reported in the study appeared to have been conducted on a post‐hoc basis (e.g. individual items of the IADL and ADL rather than full scales) thereby contributing to a high risk of reporting bias.
Other potential sources of bias
We noted several methodological issues that may have affected the magnitude of effect estimates. Lewin 2013 indicated that some participants may have already improved in self care abilities before the baseline assessments were conducted. Contamination of the group receiving usual care may have occurred in both studies. Tuntland 2015 did not monitor therapist and participant adherence to the intervention protocol, so it was not clear whether the reablement intervention led to the changes; the benefits achieved in the usual care group may have been due to the extra co‐interventions they received. We rated both studies to be at high risk of bias on this domain.
Economic evaluation
The reliability of any economic evaluation depends, at least in part, on its use of reliable clinical data (Shemilt 2011). As indicated, we judged the associated trial to be at high risk of bias in terms of randomisation and other domains (Lewin 2013), and this should therefore, be borne in mind when considering the costs reported in Lewin 2014. According to the Drummond checklist guidelines (Table 1), Lewin 2014 conducted a partial evaluation and used a regression‐based analysis to compare the costs associated with services used by the intervention and usual care groups over a two‐year period. The total costs per person were calculated for home‐care, visits to emergency departments and unplanned inpatient admissions for the first and second year, and for the 24 months combined. The cost of the 12‐week intervention when compared to 12 weeks of usual care were not reported separately; neither were the costs of residential or hospice care.
There were some additional limitations to the economic evaluation as indicated by the Drummond guidelines (Table 1). Specifically, an incremental cost‐effectiveness ratio linking the benefits to costs for the intervention and control conditions, was not calculated. In addition, the authors did not take into account discount rates to control for inflation and there was no mention of productivity costs. They did not conduct sensitivity analyses. The authors acknowledged they were unable to exactly match the date of home‐care referral with the financial year of assessment or with actual service use, so it is possible this may have led to some over‐ or under‐estimation of the hours of service(s) that participants had received in each year under investigation; however, this measurement bias was likely to have affected the intervention and usual care groups equally.
Effects of interventions
We were able to pool data for functional status, mortality and QoL (at three‐month and nine‐ to 12‐month follow‐ups; Lewin 2013; Tuntland 2015). The remaining outcomes: hospital admission, emergency department presentation, level of personal care and living arrangements, came from a single study (Lewin 2013); and where indicated, from the associated costs paper (Lewin 2014). According to our GRADE assessment, the evidence was very low quality for all outcomes. summary of findings Table for the main comparison presents a summary of the main results.
We reported ITT analyses as described by the study authors in line with recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), although as indicated, we did not consider all analyses conducted by Lewin 2013 to meet the criteria for ITT.
Reablement versus usual care
Primary outcomes
Function
Self report measures
We pooled the self report function measures from Lewin 2013 (ADLs and IADLs) and Tuntland 2015 (COPM) using the generic inverse‐variance method. There was very low quality evidence that time‐limited reablement may slightly improve functional status at three months (SMD ‐0.40; 95% CI ‐0.81 to ‐0.00; Analysis 1.1; 2 studies; 252 participants), and at the nine‐ to 12‐month follow‐up period (SMD ‐0.30; 95% CI ‐0.53 to ‐0.06; Analysis 1.2; 2 studies; 249 participants).
Mortality
There was very low quality evidence from the pooled data that reablement may lead to little or no difference in mortality at nine to 12 months (RR 0.97; 95% CI 0.74 to 1.29; Analysis 2.1; 2 studies; 811 participants). Only Lewin 2014 pooled mortality rates at 24 months of follow‐up; there is very low quality evidence that reablement may lead to little or no difference in mortality rates (RR 0.84; 95% CI 0.69 to 1.03; Analysis 2.2; 1 study; 750 participants).
Unplanned hospital admissions
Very low quality evidence from a single study with 750 participants suggested that the intervention may make little or no difference to unplanned hospital admissions at 12 months (RR 0.94; 95% CI 0.83 to 1.07; Analysis 3.1) or 24 months (RR 0.94; 95% CI 0.85 to 1.03; Analysis 3.2) (Lewin 2013; Lewin 2014).
Secondary outcomes
Quality of life
We combined QoL scores from Lewin 2013 (Assessment of Quality of Life Scale; three and 12 months) and the overall health rating from Tuntland 2015 (COOP/Wonka; three and nine months) so that lower scores indicated an improvement in QoL. The very low quality findings indicated that reablement may make little or no difference to QoL at three months (SMD ‐0.18; 95% CI ‐0.43 to 0.07; Analysis 4.1; 2 studies; 252 participants) or at the nine‐ to 12‐month follow‐up period (SMD ‐0.23; 95% CI ‐0.48 to 0.02; Analysis 4.2; 2 studies; 249 participants).
Service use ‐ level of personal care
The three‐ and 12‐month data are from Lewin 2013; there was very low quality evidence that the intervention may reduce the need for personal care at three months (RR 0.43; 95% CI 0.36 to 0.52; Analysis 5.1; 750 participants). At 12‐month follow‐up, Lewin 2014 reported on both people receiving ongoing care as well as new (emergent) clients. We combined these scores and found there was very low quality evidence that the reablement group may be less likely to need personal care services at 12 months (RR 0.45; 95% CI 0.36 to 0.56; Analysis 5.2; 750 participants). Lewin 2014 also reported on people assessed and approved for a higher level of care (residential care or equivalent home care) over the 24‐month follow‐up period. There was very low quality evidence that the reablement group may have been slightly less likely to have been approved for a higher level of care than the usual care group (RR 0.87; 95% CI 0.77 to 0.98; Analysis 5.3; 750 participants).
Service use ‐ visits to emergency departments
Only Lewin 2014 reported on presentations to emergency departments. The very low quality results suggest that the intervention may make little or no difference to the rates of emergency department visits at 12 months (RR 0.90; 95% CI 0.79 to 1.04; Analysis 6.1; 750 participants) or at 24‐month follow‐up (RR 0.93; 95% CI 0.84 to 1.03; Analysis 6.2; 750 participants).
Living arrangements
Lewin 2013 reported the number of people who were in residential care at three and 12 months. There was very low quality evidence that reablement may make little or no difference to the rates of transfer to a residential setting (RR 0.76; 95% CI 0.40 to 1.44; Analysis 7.1, three‐month data; RR 0.92 95% CI 0.62 to 1.34; Analysis 7.2, 12‐month data; 750 participants).
User satisfaction
Neither of the included trials reported user satisfaction.
Costs of services
The intervention group used fewer hours of personal care and other home‐care services and this translated into lower costs per client relative to people in the usual care group for the total two‐year period (Lewin 2014; intervention AUD 5833; usual care AUD 8374). The costs related to hospital admissions over the two‐year period were only slightly lower for the intervention group (intervention AUD 13,369; usual care AUD 13,675). The mean total aggregated costs for home‐care and healthcare (emergency department and unplanned hospital admissions) in the intervention group over 24 months was AUD 19,888 compared to AUD 22,757 for the usual care group (RR 0.89; 95% CI 0.78 to 1.02; P = 0.08 as reported in Lewin 2014).
Discussion
Summary of main results
The findings of this review were based on two studies (summary of findings Table for the main comparison), and the evidence was uniformly very low quality for all outcomes. From the results of meta‐analyses of the two included studies, reablement may slightly improve functional status but may have little or no effect on QoL of older adults, or mortality rates at nine to 12 months. Other outcomes were measured by one study (Lewin 2013), and an associated costs paper (Lewin 2014). The very low quality evidence suggested there is uncertainty regarding the effects of reablement on living arrangements, unplanned hospital admissions or visits to an emergency department at both the 12‐month follow‐up and for the overall 24‐month period, or for mortality at 24 months (Lewin 2014).
There was very low quality evidence from one study to indicate that the reablement intervention may reduce need for either ongoing home‐care, or a new episode of personal care at 12‐month follow‐up (Lewin 2013), and may slightly reduce the likelihood of being assessed as needing a higher level of care (i.e. residential care or equivalent home care) at 24 months (Lewin 2014). Neither study measured user satisfaction, which is possibly an important factor in ensuring uptake and adherence related to such interventions. The accompanying cost data, again of very low quality, indicated that health and social‐care services cost per client were lowered to a small degree in the intervention group when compared to people receiving usual care at 24 months' follow‐up, suggesting that reduced need for personal care may translate into lower costs per client in the intervention group. However, these results were based on very low quality evidence and so we are uncertain about the size and importance of these effects.
Overall completeness and applicability of evidence
Despite a thorough search of the existing evidence for the effectiveness of reablement type services, we identified only two trials that were eligible for inclusion in this review. Our review focused on reablement services that were time‐limited (12 weeks or less) as this model has been widely adopted in practice (Mishra 2016; e.g. UK: Glendinning 2010, Denmark: Winkel 2015). We excluded trials lasting longer than 12 weeks; in one study with an extended time frame (Parsons 2012), a reablement‐type intervention continued until the clients were admitted to residential care or died. We concluded that interventions that continued indefinitely would not be directly comparable to time‐limited services for our outcomes of interest, and especially in terms of achieving functional independence and a reduced need for personal care services in the home. We also focused our search on those studies that included a measure of functional outcomes. These two factors resulted in excluding a number of studies, and we may consider broadening our inclusion criteria to enable a comparison of time‐limited versus extended reablement services in an update of this review. We identified no trials that compared reablement with a wait‐list control, a finding that may reflect ethical concerns related to withholding a service to a vulnerable group. Nevertheless, as usual home care itself could have some beneficial effects on the outcomes of interest, a wait‐list comparison may have provided an opportunity to determine the absolute effect of reablement.
While acknowledging these limitations, Lewin 2013 and Tuntland 2015 did yield some interesting information on reablement services developed and delivered in two different countries and health/social care settings. The characteristics of the participants were broadly similar to those in the studies that we excluded (e.g. not time‐limited or non‐randomised) in terms of age, level of dependency and proportion living alone, thereby suggesting they were generally representative of older adults requiring home care. The components and ethos of the two services are comparable to each other and to those offered elsewhere (e.g. Glendinning 2010), although there may be limits in the extent to which the findings can be generalised to other settings with differing levels of integration between health and social care systems. There are also likely to be differences in the content and delivery of usual care provided in different jurisdictions; for example, there were some differences in the frequency of visits for the usual care group between the included studies. Indeed, home‐care can be provided and resourced in a number of different ways (e.g. public or private sector funds/publicly or privately funded organisations; Rostgaard 2011), all of which may add to the complexity associated with delivering and evaluating this intervention elsewhere.
Many of our objectives could not be addressed in this review due to the lack of trials that met our inclusion criteria. Unfortunately, it was not possible to undertake analyses to determine whether reablement is more suitable for particular groups of older people, such as people living alone, or younger versus older cohorts. Future studies are required to address these gaps in our knowledge. Encouragingly, one large multicentre trial is currently underway in Norway using the same design as Tuntland 2015, and data should be available by 2017 (Langeland 2015).
Quality of the evidence
We are very uncertain of the effectiveness of reablement because the evidence was very low quality for all outcomes according to our GRADE assessment (summary of findings Table for the main comparison). We downgraded for risk of bias and imprecision (primarily as some of the outcomes were from a single study only) and as a result, any interpretation of the findings and effects should be treated with considerable caution. To their credit, the authors of both trials highlighted methodological concerns that are probably not uncommon in RCTs of this type. The randomisation process was compromised in Lewin 2013, in that some personnel involved had manipulated the procedure in the well‐intentioned belief that particular clients might benefit more from assignment to either the intervention or usual care group. Neither study was able to blind participants/personnel adequately, something that is difficult to achieve in this type of intervention. Furthermore, blinding of outcome assessors was not completely successful in either trial. There were also some potential issues related to treatment fidelity and adherence. Incomplete data collection resulting from participant drop‐out was also evident in Lewin 2013; indeed, reaching, recruiting and retaining older participants whose health may deteriorate over the course of the study presents a methodological challenge for studies of this type. Similarly, issues such as contamination, baseline differences or differences in what control groups receive, and fidelity of such complex interventions are all challenging when undertaking research in this area.
Potential biases in the review process
It is possible that we may have missed some studies when searching, including those not published in English. Following advice from Cochrane Consumers and Communication, we excluded all non‐randomised trials. Arguably, these would have provided a more extensive analysis of the effectiveness of reablement, but it seems unlikely that such research designs would be rated as higher quality evidence than the identified RCTs, and therefore would not necessarily have added to the findings with any certainty. Our criteria did limit the number of trials for inclusion although we do not consider this to have introduced bias to the review.
Agreements and disagreements with other studies or reviews
A number of previous reviews were unable to identify which of the various models/types of care provision may be most effective for the rehabilitation and maintenance of the independence of older adults. One (non‐Cochrane) review examined interventions designed to reduce dependency in personal care in adults aged 18 years and older (Whitehead 2015). The review considered a range of single and multi‐component interventions and non‐randomised designs. Whitehead 2015 included 13 studies, five of which (including Lewin 2013) were classified as reablement/restorative services. We excluded four of these studies from the current review because they did not meet our inclusion criteria (i.e. non‐randomised designs; not time‐limited). Whitehead 2015 focused on changes in ADL scores as a measure of dependency and concluded that there was some limited evidence to support the effectiveness of the interventions. Consistent with our findings, these authors indicated that such interventions could reduce the use and costs associated with ongoing care services. However, the use of broad selection criteria led to considerable heterogeneity in the content of the interventions; this was something we were keen to avoid by using more focused inclusion criteria.
Another non‐Cochrane review aimed to determine whether home‐care reablement interventions reduced the need for support and assistance from social care home‐care services (Legg 2016). Consistent with our review, Legg 2016 focused on time‐limited services, but used six weeks as their cut‐off; Legg 2016 found no studies that met their inclusion criteria and interestingly did not list Lewin 2013 as one of their excluded studies. Legg 2016 concluded that there was no evidence to suggest reablement is effective at increasing independence or reducing the use of personal care services.
A further non‐Cochrane review included 10 studies, six of which were non‐randomised designs, and suggested that reablement had a positive impact on HRQoL and service utilisation; however, the inclusion of non‐randomised studies may have introduced some bias and uncertainty into their findings, and this may have inflated the effect estimates in this case (Tessier 2016). In addition, while a time‐limit to the service was part of the inclusion criteria, Tessier 2016 included some trials that could continue beyond the six to 12 weeks, which we excluded on that basis from our review.
Another non‐Cochrane review is currently underway that will examine the impact of a range of programmes for older adults in receipt of home‐care services (Sims‐Gould 2015). The review will consider services that aim to provide recipients with the capacity and ability to self manage some aspects of their care, principally ADLs; it will include trials lasting a few weeks to a few months, while excluding long‐term programmes (those lasting more than six months).
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Comparison 1 Reablement versus usual care: functional status, Outcome 1 Functional status at 3 months.
Comparison 1 Reablement versus usual care: functional status, Outcome 2 Functional status at 9 to 12 months.
Comparison 2 Reablement versus usual care: mortality, Outcome 1 Mortality at 9 to 12 months.
Comparison 2 Reablement versus usual care: mortality, Outcome 2 Mortality at 24 months.
Comparison 3 Reablement versus usual care: unplanned hospital admissions, Outcome 1 Hospital admissions at 12 months.
Comparison 3 Reablement versus usual care: unplanned hospital admissions, Outcome 2 Hospital admissions at 24 months.
Comparison 4 Reablement versus usual care: health‐related quality of life, Outcome 1 Quality of life at 3 months.
Comparison 4 Reablement versus usual care: health‐related quality of life, Outcome 2 Quality of life at 9 to 12 months.
Comparison 5 Reablement versus usual care: level of personal care, Outcome 1 Personal care at 3 months.
Comparison 5 Reablement versus usual care: level of personal care, Outcome 2 Personal care at 12 months.
Comparison 5 Reablement versus usual care: level of personal care, Outcome 3 Approved for higher level of personal care at 24 months.
Comparison 6 Reablement versus usual care: emergency department (ED) presentations, Outcome 1 ED presentation at 12 months.
Comparison 6 Reablement versus usual care: emergency department (ED) presentations, Outcome 2 ED presentation at 24 months.
Comparison 7 Reablement versus usual care: living arrangements, Outcome 1 Residential care at 3 months.
Comparison 7 Reablement versus usual care: living arrangements, Outcome 2 Residential care at 12 months.
| Reablement compared with usual home‐care for maintaining independence | ||||||
| Patient or population: adults aged ≥ 65 years Settings: clients' own home Intervention: reablement services Comparison: usual home‐care service | ||||||
| Outcomes | Illustrative comparative risks (95% CI) | Effect estimate (95% CI) | No of participants (studies) | Quality of the evidence (GRADE) | Comments | |
| Assumed risk* | Corresponding risk | |||||
| Usual care | reablement | |||||
| Functional status** ADL and IADL (Lewin 2013) COPM (Tuntland 2015) Lower scores indicate greater independence Follow‐up: 9 to 12 months | ‐ | The mean score in the intervention group was 0.3 SD lower (0.53 to 0.06 lower) | SMD ‐0.30 (‐0.53 to ‐0.06) | 249 (2 studies) | ⊕ooo very low a,b | ‐ |
| Mortality*** Follow‐up: 9 to 12 months | 198 per 1000 | 6 fewer per 1000 (51 fewer to 57 more) | RR 0.97 (0.74 to 1.29) | 811 (2 studies) | ⊕ooo very lowa,c | ‐ |
| Unplanned hospital admission** Follow‐up: 24 months | 707 per 1000 | 42 fewer per 1000 (106 fewer to 21 more) | RR 0.94 (0.85 to 1.03) | 750 (1 study) | ⊕ooo very lowa,c,d | ‐ |
| Quality of life** AAQ (Lewin 2013) COOP/Wonka (overall health; Tuntland 2015) Lower scores indicate improvement Follow‐up: 9 to 12 months | ‐ | The mean score in the intervention group was 0.2 SD lower (0.48 lower to 0.02 higher) | SMD ‐0.23 (‐0.48 to 0.02) | 249 (2 studies) | ⊕ooo very lowa,b,c | ‐ |
| Level of emerging personal care needs (approved for higher level of care)*** Follow‐up: 24 months | 643 per 1000 | 84 fewer per 1000 (148 fewer to 13 fewer) | RR 0.87 (0.77 to 0.98) | 750 (1 study) | ⊕ooo very lowa,d | ‐ |
| Living arrangements (transferred to residential care) Follow‐up: 12 months | 128 per 1000 | 10 fewer per 1000 (49 fewer to 45 more) | RR 0.92 (0.62 to 1.34) | 750 (1 study) | ⊕ooo very lowa,c,d | ‐ |
| Cost effectiveness*** Total aggregated costs for home‐ and healthcare (emergency department and unplanned hospital admissions) Follow‐up: 24 months | The mean costs were AUD 2869 lower for the reablement group compared with usual care (AUD 19,888 with intervention versus AUD 22,757 with control) | ‐ | 750 (1 study) | ⊕ooo very lowa,d | ‐ | |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). AAQ: Assessment of Quality of Life; ADL: activities of daily living; CI: confidence interval; COPM: Canadian Occupational Therapy Performance; IADL: instrumental activities of daily living; RR: risk ratio; SMD: standardised mean difference; SD: standard deviation. | ||||||
| GRADE Working Group grades of evidence | ||||||
| * median control group risk calculated by GRADEpro. ** unpublished data. *** Data from Lewin 2014. a Downgraded twice (‐2) for risk of bias concerns (Lewin 2013 was at high risk of bias on all domains). b Downgraded once (‐1) for imprecision (fewer than 400 participants). c Downgraded once for imprecision (‐1) as confidence interval was likely to include minimally important difference and no difference). d Downgraded once for imprecision (‐1) as data were from single study. | ||||||
| Study design | Issue addressed | Explanation |
| 1. The research question is stated | Yes | Compared health and aged care service use and costs of a restorative home care service versus a conventional service |
| 2. The economic importance of the research question is stated | Yes | Investigated whether an effective intervention was also cost‐effective |
| 3. The viewpoint(s) of the analysis are clearly stated and justified | Yes | Multi‐agency perspective including providers of health and social (aged) care |
| 4. The rationale for choosing alternative programmes or interventions compared is stated | Yes | Intervention and usual care services funded by the Government and provided by a not‐for‐profit organisation |
| 5. The alternatives being compared are clearly described | Yes | Time‐limited individualised restorative service delivered in the home compared to usual personal care service |
| 6. The form of economic evaluation used is stated | No | Not explicitly stated, information provided on the use and cost of home‐care and healthcare services |
| 7. The choice of form of economic evaluation is justified in relation to the questions asked | Yes | A partial evaluation that provided costs data on service utilisation in both the intervention and control conditions. An ICER was not conducted |
| Data collection | ||
| 8. The source(s) of effectiveness estimates are stated | Yes | RCT (Lewin 2013): 750 participants (aged ≥ 65 years) eligible for government‐funded home care and allocated to a time‐limited (12 weeks maximum) reablement intervention or to usual home‐care services |
| 9. Details of the design and results of effectiveness study are given (if based on a single study) | Yes | See Lewin 2013 |
| 10. Details of the methods of synthesis of estimates are given (if based on a synthesis of a number of effectiveness studies) | N/A | ‐ |
| 11. The primary outcome measure(s) for the economic evaluation are clearly stated | Yes | Comparison of intervention and control groups on:
|
| 12. Methods to value benefits are stated | N/A | Not stated |
| 13. Details of the participants from whom valuations were obtained were given | Yes | ‐ |
| 14. Productivity changes (if included) are reported separately | N/A | Productivity costs not included |
| 15. The relevance of productivity changes to the study question is discussed | N/A | Productivity costs not discussed |
| 16. Quantities of resource use are reported separately | No | The mean hours of aged‐care services provided, but only mean cost per client for first year and total 2‐year period reported rather than cost per hour Number of emergency department visits provided, but only mean total costs per client for 2‐year period rather than cost per visit reported Number of hospital visits and mean length of stay (episodic and cumulative) provided, but only mean total cost per client for all hospital admissions reported |
| 17. Methods for the estimation of quantities and unit costs are described | Yes | Quantities data were obtained via the Western Australian Data Linkage System: the Emergency Department Data Collection; Hospital Morbidity Data System; the Mortality Register; the HACC database and the Aged Care Assessment Program database Unit costs: home care costs provided by the Western Australian DoH; emergency department costs provided by the National Hospital Cost Data Collection Cost Report (2007‐8); inpatient data were provided by the Public Sector Estimated Round 12 (2007‐8) AR‐DRG 5.1 Cost Report for Western Australia (DoHA 2008) |
| 18. Currency and price data are recorded | Australian dollars | |
| 19. Details of currency of price adjustments for inflation or currency conversion are given | No | ‐ |
| 20. Details of any model used are given | Yes | General linear model used to analyse aggregated health and social care costs over time (2‐year period) and adjusted for living arrangements, carer status, gender and dependency level |
| 21. The choice of model used and the key parameters on which it is based are justified | Yes | Aggregated health and social care costs presented |
| Analysis and interpretation of results | ||
| 22. Time horizon of costs and benefits is stated | Yes | Data collected for 3‐year period commencing 1 year prior to the date when the participant assigned to either intervention or control groups |
| 23. The discount rate(s) is stated | No | Not discussed |
| 24. The choice of discount rate(s) is justified | N/A | ‐ |
| 25. An explanation is given if costs and benefits are not discounted | No | None provided |
| 26. Details of statistical tests and confidence intervals are given for stochastic data | N/A | ‐ |
| 27. The approach to sensitivity analysis is given | N/A | Not conducted |
| 28. The choice of variables for sensitivity analysis is justified | N/A | Not conducted |
| 29. The ranges over which the variables are varied are justified | N/A | Sensitivity analyses not conducted |
| 30. Relevant alternatives are compared | No | ‐ |
| 31. Incremental analysis is reported | No | ‐ |
| 32. Major outcomes are presented in a disaggregated as well as aggregated form | Yes, in some cases only | Home‐care services, presentations to emergency departments and unplanned admissions were disaggregated. However, costs of the intervention itself were not disaggregated. The study authors reported that they were unable to include costs of residential/hospice care as the data they used related to approval for the service and there was no certainty that this translated into an actual admission |
| 33. The answer to the study question is given | Yes | Clients who received intervention were less costly to the aged and healthcare services over time than those who received standard home care, p. 334 |
| 34. Conclusions follow from the data reported | Yes | Statistical significance achieved more often in as‐treated analysis, suggesting that the success of the intervention depended heavily on compliance with the HIP [intervention] protocol. p. 334 |
| 35. Conclusions are accompanied by the appropriate caveats | Yes | Limitations and possible effects on the findings noted on p. 335. Specifically, authors were unable to match the dates of home‐care referral exactly with the financial year date of assessment or utilisation. Therefore, there may have been some over‐ or under‐estimation of the number of hours of service(s) the client used in each year. This measurement bias was likely to have affected the intervention and control groups equally |
| AR‐DRG: Australian Refined Diagnosis Related Group; DoH: Department of Health; DoHA: Department of Health and Ageing; HACC: home and community care; HIP: Home Independence Program; ICER: incremental cost‐effectiveness ratio; RCT: randomised controlled trial. | ||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Functional status at 3 months Show forest plot | 2 | 252 | Std. Mean Difference (Random, 95% CI) | ‐0.40 [‐0.81, 0.00] |
| 2 Functional status at 9 to 12 months Show forest plot | 2 | 249 | Std. Mean Difference (Random, 95% CI) | ‐0.30 [‐0.53, ‐0.06] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Mortality at 9 to 12 months Show forest plot | 2 | 811 | Risk Ratio (M‐H, Random, 95% CI) | 0.97 [0.74, 1.29] |
| 2 Mortality at 24 months Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Hospital admissions at 12 months Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| 2 Hospital admissions at 24 months Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Quality of life at 3 months Show forest plot | 2 | 252 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.18 [‐0.43, 0.07] |
| 2 Quality of life at 9 to 12 months Show forest plot | 2 | 249 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.23 [‐0.48, 0.02] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Personal care at 3 months Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| 2 Personal care at 12 months Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| 3 Approved for higher level of personal care at 24 months Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 ED presentation at 12 months Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| 2 ED presentation at 24 months Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Residential care at 3 months Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
| 2 Residential care at 12 months Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Totals not selected | |
