Types of studies

We will include randomized controlled trials, including cluster‐randomized trials. Cluster‐randomized trials will be required to have at least two intervention and two control sites to be considered eligible for inclusion, to reduce potential bias from site‐specific confounding (EPOC 2017). Cross‐over trials will not be included.

The following types of economic evaluation studies will be considered for inclusion: full economic evaluation studies (i.e. cost‐effectiveness analyses, cost‐utility analyses, cost‐benefit analyses), partial economic evaluations (i.e. cost analyses, cost‐description studies, cost‐outcome descriptions); and randomized trials reporting more limited information, such as estimates of resource use or costs associated with intervention(s) and comparator(s). We will only consider relevant health economics studies conducted alongside effectiveness studies that meet eligibility criteria for the effectiveness component of this review (Aluko 2020). We will consider studies irrespective of their publication date, publication status or language of publication. Where possible, we will translate the studies published in non‐English languages.

Types of participants

Eligible study participants include healthcare professionals delivering and/or co‐ordinating healthcare to older adults living in ACFs, and older people who reside in a care home as their place of permanent abode. We defined older people as those aged 60 years or over, and we will include all participants in studies where the mean age is 60 years or more.

Aged care facilities are called different things in different countries. The terms “care home” “residential aged care facility”, “nursing home”, “aged care”, “residential/subacute/extended aged care settings”, “restorative care”, “rest homes”, “skilled nursing facilities” and “homes for the aged” are used interchangeably. No matter what the facility is called, only facilities that meet all the criteria for ‘care home’ set out in Crocker 2013 and Ward 2008 will be eligible for inclusion. Such facilities: 

• provide communal living facilities for long‐term care (as opposed to hospital where there is an expectation that this care is time limited);

• provide overnight accommodation;

• provide nursing or personal care; and

• provide care for people with illness, disability, or dependence.

Types of interventions

Eligible interventions should focus on either ways of delivering primary or specialist health care (or both) or ways of co‐ordinating the delivery of this care. Eligible models of care delivery must investigate changes to at least one of the following delivery arrangement domains (Cochrane Effective Practice and Organisation of Care (EPOC) Group taxonomy of health system interventions EPOC 2015):

• co‐ordination of the primary and/or specialist health care, or management of the primary and/or specialist care processes (e.g. continuity of care models; protocols for care decisions or decision support; nurse practitioners working collaboratively with GPs; care provision by multidisciplinary teams);

• where the primary and/or specialist health care is provided (e.g. co‐location of primary medical care or specialist care services within ACFs; Hospital in the Nursing Home; in‐reach of specialists or specialised nursing teams for routine or emergency care; telemedicine to assist with provision of primary/specialist care services to residents of ACFs); or

• who provides the primary and/or specialist health care (e.g. provision of primary or specialist care services (or both) to residents of ACFs by nurse practitioners; medical treatment provided by multidisciplinary teams of experts).

We will consider studies irrespective of the medical specialisation of the healthcare professional involved in delivering the various models of health care. We will exclude care provided by allied health professionals (e.g. physiotherapy) or pharmacist‐led interventions, except when they are part of a multidisciplinary team or are providing primary or specialist medical care to residents of ACFs.

Eligible comparators may include usual care or another model of care, as defined by the trialists. A key aspect of this review is that both the experimental group and the comparison group need to receive the same primary or specialist healthcare services, just in a different way (e.g. primary care services provided by a GP who is a staff member of the ACF versus provision of primary care to ACF residents on request by an external GP). A detailed description of usual care is important for a meaningful interpretation of the effects of interventions in this review, because the organisation and delivery of medical care to residents in ACFs is expected to be different within and between countries. We will describe the interventions, including usual care, using the Template for Intervention Description and Replication (TIDieR) checklist (Hoffmann 2014).

We will exclude studies focusing on ways of providing dental care. We will also exclude studies looking at more effective ways of providing general care to ACF residents, such as bathing or feeding. We will exclude studies focused primarily on nursing staffing models for existing staff employed within ACFs (i.e. how existing ACF nursing and personal care attendant staffing is organised to meet resident/patient needs, including the mix and level of skills, and staffing ratios), as this is the focus of a separate Cochrane Review that is currently being updated (Hodgkinson 2011).

Medication review for older people in residential care is a focus of several other Cochrane Reviews (Alldred 2016; LaMantia 2010; Rankin 2018), so this intervention will not be considered in this review unless it is part of a more complex intervention (e.g. general medical in‐reach review) that includes other eligible elements. Studies investigating the introduction of new treatments (i.e. adding services such as cognitive behavioural therapy for dementia; or social prescribing, e.g. visits from school children or music therapy providers), will not be considered in this review. Interventions focused exclusively on education of staff, skill development or quality improvement (e.g. interventions that focus primarily on education, information campaigns, audit and feedback, provider reminders, computerised medical records, financial incentives, guideline implementation or guideline adherence) are also outside the scope of this review.

Types of outcome measures

The outcomes in this review are designed to capture the key health, quality‐of‐care and economic effects of alternative ways of delivering or co‐ordinating healthcare (or both) to older adults living in ACFs. While cognitive and functional status outcomes are relevant outcomes to this patient population, they are not the focus of this review, which aims to investigate the effects of different ways of delivering or co‐ordinating the same primary and/or specialist healthcare. Studies will not be selected based on outcomes reported.

Electronic searches

The review authors will develop the search strategies in consultation with the EPOC Information Specialist. We will search the following databases for primary studies, from inception to the date of search.

• Cochrane Central Register of Controlled Trials (CENTRAL), in the Cochrane Library

• MEDLINE Ovid

• Embase Ovid

• Age Line EBSCO

• CINAHL EBSCO (Cumulative Index to Nursing and Allied Health Literature)

The following databases will be searched to identify eligible economic evaluations.

• NHS Economic Evaluation Database (NHS EED), until 2015

• Cost-Effectiveness Analysis Registry (CEA)

• MEDLINE Ovid from 2015 onwards

Search strategies will comprise of keywords and controlled vocabulary terms. We will not apply any limits on language and we will search all databases from inception to the date of search. We will use a study design filter to identify randomized trials.

Searching other resources

To identify completed but unpublished, ongoing and planned trials, the following registries will be searched.

• World Health Organization International Clinical Trials Registry Platform (WHO ICTRP; www.who.int/ictrp)

• US National Institutes of Health Ongoing Trials Register (ClinicalTrials.gov; www.clinicaltrials.gov)

We will search the Cochrane Database of Systematic Reviews (CDSR) and the Database of Abstracts of Reviews of Effects (DARE) for related systematic reviews. We will screen the included studies of these reviews to identify any additional eligible studies. We will also handsearch reference lists of all included studies to identify additional potentially eligible studies. We will contact authors of included studies to clarify reported published information, or to seek unpublished results/data if needed. See Appendix 1 for a draft MEDLINE search.

Selection of studies

All records retrieved from the search will be imported into Covidence (www.covidence.org) to facilitate deduplication and subsequent independent duplicate screening of titles and abstracts and potentially eligible full text papers. We will also use Covidence to facilitate the assessment of risk of bias in included studies.

Two of three review authors (PP, LN, AL) will independently screen titles and abstracts for potentially eligible studies. We will retrieve the full text of all potentially eligible studies and two of the three review authors will independently screen the full texts to identify studies for inclusion. We will list all studies excluded at this stage, together with reasons for exclusion, in the ‘Characteristics of excluded studies’ table. We will resolve any disagreement through discussion; if required, we will consult a senior author (DOC or RB).

We will collate multiple reports of the same study so that each study rather than each published report is the unit of analysis in the review. We will report basic information on any eligible ongoing studies we identify. We will record the study selection process in sufficient detail to complete a PRISMA flow diagram (Liberati 2009).

Data extraction and management

We will adapt the EPOC standard data collection form and use it to extract study characteristics and outcome data (EPOC 2017). We will pilot the form on at least one study in the review. Two of the three review authors (PP, LN, AL) will independently extract the following study characteristics from the included studies:

• Methods: study design, number of study centres and location, study setting, withdrawals, date of study, duration of follow‐up.

• Participants: number, mean age, age range, gender, severity of condition(s) where relevant, inclusion criteria, exclusion criteria, other relevant ACF resident characteristics.

• Intervention and categorised: classified according to the EPOC taxonomy of health system interventions and described using TIDieR checklist (Hoffmann 2014), including nature of primary or specialist healthcare provided.

• Outcomes: all outcomes planned and reported on, with time points and methods of data collection.

• Notes: funding source for trial, conflicts of interest of trial authors, ethical approval.

We will develop a data extraction form for economic evaluations based on the format and guidelines used to produce structured abstracts of economic evaluations for inclusion in the NHS EED, adapted to the specific requirements of this review. We will resolve differences in extracted data by consensus or by involving a fourth review author (DOC). If important information is missing from the full‐text article, we will contact the authors of the publication to obtain it.

Assessment of risk of bias in included studies

Two of three review authors (PP, LN, AL) will independently assess risk of bias for each included study using the Cochrane 'Risk of bias' tool (Higgins 2011) and additional criteria specified by Cochrane EPOC (EPOC 2017). Any disagreement will be resolved by discussion or by consulting a senior author (DOC or RB). The 'Risk of bias' assessment will include the following domains.

• Random sequence generation

• Allocation concealment

• Blinding of participants and personnel

• Blinding of outcome assessment

• Incomplete outcome data

• Selective outcome reporting

• Other bias (bias due to problems not covered by sources of bias specified above; for cluster‐randomized trials, the following specific issues will be assessed: recruitment bias, baseline imbalance, protection against contamination, incorrect analysis)

We will judge each study to be at high, low or unclear risk of bias for each domain listed above, and we will provide justification for our judgement in the 'Risk of bias' table for each study. We will assess information from study reports, protocols, trial registration documents or correspondence with trialists to support our judgement. Where information on risk of bias is related to correspondence with trialists or unpublished data, we will note this in the 'Risk of bias' table. We will summarise the 'Risk of bias' judgements across different studies for each of the domains listed and include the summary figure generated by Review Manager software (Review Manager 2020).

When considering intervention effects, we will take account of the risk of bias for the studies that contribute to that outcome and incorporate this into our judgements about the certainty of the evidence. A summary assessment of the risk of bias of each study will be done using three key domains: sequence generation and allocation concealment (selection bias), and blinding of outcome assessors (detection bias). Studies will be considered to be at low risk of bias if the three key domains are at low risk of bias; unclear risk of bias if at least one of the domains has unclear risk of bias and none of the domains are at high risk of bias; and high risk of bias if at least one of the key domains is at high risk of bias.

Each economic evaluation will be classified as: (1) a type of full economic evaluation; (2) a type of partial economic evaluation; or (3) a type of effectiveness study (e.g. a randomized trial) reporting more limited information on the resource use or costs associated with an intervention. For types (1) and (2), the economic studies will be classified as a single study design (e.g. an economic evaluation alongside a randomized trial) or a model‐based evaluation, involving the synthesis of evidence derived from multiple studies or data sources.

We will critically appraise health economics studies using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) (Husereau 2013) We will assess whether included studies describe methods, assumptions, data and possible biases in a way that is transparent and is easily accessible to critical readers (Aluko 2020). In assessing the methodological quality of economic evaluations, we will aim to identify the key uncertainties in each study and assess the applicability and relevance of each economic evaluation to different settings.

Measures of treatment effect

Unit of analysis issues

We will check to see that analyses in the eligible studies have been performed at the same level as the allocation to ensure that unit‐of‐analyses errors are avoided. Data from cluster‐randomized trials must be appropriately adjusted for clustering when presenting the data at the individual patient level. If the data from cluster‐randomized trials have not been adjusted correctly, we will re‐analyse the results based on guidance provided in Chapter 23 of the Cochrane Handbookfor Systematic Reviews of Interventions (Higgins 2019). Adjusting for clustering requires dividing the original sample size (and number of events in the case of dichotomous data) by the design effect, which is calculated from the average cluster size and the intra‐cluster correlation coefficient (ICC). Where the ICC is not reported, we will impute the most commonly reported value from studies, where it is reported.

Dealing with missing data

If data are missing, we will contact study investigators in order to verify key study characteristics and obtain missing outcome data where possible (e.g. when a study is identified as abstract only). For all outcomes, we intend to analyse the data on an intention‐to‐treat basis. That is, we will include all participants randomized to each group in the analyses, and analyse data according to initial group allocation irrespective of whether or not participants received, or complied with, the planned intervention. Where intention‐to‐treat analyses are not possible due to missing data, we will conduct available case analysis, that is, we will only include the number of participants on whom the outcome was measured in both the intervention and control groups.

Assessment of heterogeneity

In the event that a meta‐analysis of the study data is feasible, we will use the I² statistic to assess heterogeneity among the trials in each analysis. If we identify substantial  heterogeneity (I² = 50% to 90%) or considerable heterogeneity (I² = 75% to 100%) (Deeks 2020), we will note this in the text and explore this heterogeneity through the pre‐specified subgroup analyses (see Subgroup analysis and investigation of heterogeneity). We will use caution in the interpretation of those meta‐analysis results with high levels of unexplained heterogeneity.

Assessment of reporting biases

If we are unable to contact study authors to obtain missing outcome data or they cannot provide it, and the missing data are thought to introduce serious bias, we will explore the impact of including such studies in the overall assessment of results. If we identify more than 10 studies reporting on the same outcome, we will generate funnel plots using Review Manager software (Review Manager 2020) and visually examine them for asymmetry, to explore possible reporting or publication biases (Higgins 2019; Sterne 2011).

For the economic evaluation, a common reporting bias is the non‐reporting of planned economic evaluations. Wherever possible, we will follow up studies that planned to do an economic evaluation in the study protocol but have not yet reported or published these findings, in order to access these data.

Data synthesis

We will combine study data in meta‐analyses only when it is meaningful to do so, i.e. if the interventions, participants, and the underlying question are similar enough for pooling to make sense. We will carry out the statistical analysis using Review Manager software (Review Manager 2020). We will use a random‐effects model to combine the data, as we anticipate that there is likely to be heterogeneity between studies attributable to the different settings, populations and interventions (for example, different models of care, with different usual‐care protocols, varying skills of the nursing staff, age and disease condition of ACF residents).

For meta‐analysis of continuous outcome data we will use the inverse variance method. To combine dichotomous outcome data we will use the method proposed by Mantel‐Haenszel (Deeks 2020). If cluster‐randomized trials are included in the meta‐analysis we will use the generic inverse variance method in Review Manager to combine the data. If both adjusted and non‐adjusted figures are provided we will carry out a sensitivity analysis using the unadjusted figures, to examine any possible impact on the estimate of treatment effect. If trialists report medians and interquartile ranges, it may be because their data are not normally distributed. If this is the case, we will make a note of this and consider the implications of the skewed data on the study findings. If a study has multiple trial arms, we will extract and analyse data from the relevant arms. If two comparisons (e.g. intervention A versus usual care and intervention B versus usual care) must be entered into the same meta‐analysis, we will halve the control group to avoid double‐counting. If it is not possible to combine study data due to clinical or methodological diversity, we will report a structured tabulation of results across studies and use vote counting based on the direction of effect to summarise the results (Higgins 2019).

There are currently no agreed‐upon methods for pooling combined estimates of cost‐effectiveness, extracted from multiple economic evaluations, using meta‐analysis or other quantitative synthesis methods. However, if estimates of measures of resource use and costs (as well as associated measures of uncertainty) are available from two or more studies in a common metric (with due caution given to the setting where costs were measured), for an intervention and its comparator, these will be pooled using a meta‐analysis.

Cost estimates collected from multiple studies will be adjusted to a common currency (using, for example, online cost converter EPPI‐Centre Cost Converter 2019 (EPPI‐Center Cost Converter 2021)) and price year before these data are pooled. If meta‐analyses of resource use or cost data are conducted, a structured summary will be included in the 'Results' section to comment on the direction and magnitude of results and their precision. If meta‐analyses cannot be conducted, a summary of the results of included economic evaluations will be provided in a table, supplemented by a structured summary description in the 'Results'.

Subgroup analysis and investigation of heterogeneity

Where possible, we plan to conduct subgroup analyses for the following factors.

• Type of model according to relevant EPOC delivery arrangement categories, i.e. where care is provided; who provides care; co‐ordination of care

• Type of health care being provided, i.e. primary, specialist

• Age of the ACF patients (less than 80 years versus 80 years or more): increasing age is often associated with decreasing physical/psychological well‐being so it is possible that different models are more or less effective in very old residents.

• Type of condition being treated: it is possible that different models are more effective for different conditions (e.g. patients with dementia might respond differently to a particular model of care compared to patients with congestive heart failure).

Sensitivity analysis

Where possible, we will perform the following sensitivity analyses to assess the robustness of our conclusions and explore their impact on effect sizes.

• Restricting the analysis to studies with a low risk of bias

• Assessing impact of timing of assessment: short‐term (up to 12 months; if multiple time points are available we will select the closest to six months) and longer term (12 to 24 months; if multiple time points are available we will select the closest to 18 months)

• Re‐analysing cluster RCTs to account for within‐cluster correlation

We will conduct the review according to this published protocol and report any deviations from it in the 'Differences between protocol and review' section of the systematic review.

Summary of findings and assessment of the certainty of the evidence

We will create a 'Summary of findings' table(s) for the main intervention comparison: alternative model of care versus usual care. Heterogeneity is expected across interventions. Where meaningful, we will group the interventions with similar content (e.g. care provided by multidisciplinary teams or care provided via teleconsultations at a distance). If usual care is considered substantially different across trials, comparisons will be further split according to the characteristics of the control intervention.

The following outcomes will be included in the 'Summary of findings' tables, together with the certainty of the evidence for each: emergency department visits; unplanned hospital admissions; adverse effects; adherence to clinical‐guideline‐recommended care; health‐related quality of life; mortality; and total costs of care. If, during the review process, we become aware of an important outcome that we failed to list in our planned 'Summary of findings' table(s), we will include the relevant outcome and make a note of this in the section 'Differences between protocol and review'.

Using GRADEpro GDT software (GRADEpro GDT 2014), two of the three review authors (PP, LN, AL) will independently and in duplicate, assess the certainty of the evidence (high, moderate, low, or very low) using the five GRADE considerations (risk of bias, consistency of effect, imprecision, indirectness, and publication bias) (Guyatt 2008; Higgins 2019). We will be guided by the methods and recommendations described in Chapter 14 of the Cochrane Handbook for Systematic Reviews of interventions (Higgins 2019), and the EPOC worksheets (EPOC 2015). We will provide justification for our decisions using table footnotes and we will insert comments to aid readers' understanding of the findings, where necessary. These decisions will be checked by all authors and any disagreements on certainty ratings will be resolved by discussion.

If a study provides data on an outcome but these data cannot be included in the meta analyses, we will add a comment to the 'Summary of findings' table, noting if the findings support or contradict the summary estimate of effect from the meta analyses. We will use plain language statements to report these findings in the review (Cochrane Norway 2019).