Community care navigation intervention for people who are at‐risk of unplanned hospital presentations
Abstract
Objectives
This is a protocol for a Cochrane Review (intervention). The objectives are as follows:
Main objective
To assess the effects of care navigation, delivered in the community, on hospital presentations and PROMS in those at‐risk of unplanned hospital presentation.
Secondary objective
To assess whether the effects of care navigation on the proposed outcomes differ according to the type of clinician delivering the intervention and the population groups receiving the intervention.
Background
People with complex multi‐morbidities and chronic diseases utilise hospital services frequently, especially when their health care needs are unmet during hospitalisation and postdischarge. Increasingly short lengths of hospital stay and rapid discharge often mean that assessment of needs and circumstance, such as medical complexity, home care accessibility and level of care/support in the community, are not addressed holistically (Reed 2015). Lack of hospital liaison, postdischarge support and poor communication between hospital and community care providers may result in unplanned hospital presentations and suboptimal patient‐reported outcomes (PROMS) (Tierney 1993). Consequently, there are numerous services worldwide targeting chronic disease management, people with complex conditions, or both.
Care navigation, a common model applying care co‐ordination strategies, is one approach to manage chronic diseases through individualised case management pathways by a trained professional not involved in the person's direct care. Current evidence suggests that care navigation may reduce unplanned hospital presentations and improve PROMS (Leppin 2014; Prieto‐Centurion 2019).
Description of the condition
Global increases in elderly, people with complex conditions and multi‐morbidities mean that greater numbers of people may be at risk of unplanned hospital presentations, frequent use of public health services and poor health outcomes. An unplanned hospital presentation is defined as an unplanned hospital admission or Emergency Department (ED) presentation that is not related to planned routine care such as chemotherapy, dialysis, outpatient appointments, etc. (ACSQHC 2019). In Australia, it is estimated that 10% of hospital readmissions are unplanned, with a large percentage (56%) of these attributed to the management of chronic disease (Page 2007). A study in Western Australia reported the top 5% of hospital users accounted for 38% of inpatient costs and 26% of inpatient admissions. This top 5% mainly comprised people with chronic medical conditions such as congestive heart failure, cancer and end‐stage renal disease (Calver 2006). In the USA, 30% of all adult non‐obstetric admissions to hospital were for people at high‐risk of readmission. The definition of 'high‐risk' was based on the presence of social complexity, for example, homelessness or frailty; or medical complexity, such as having cardiac and respiratory chronic conditions or sepsis (AHRQ 2016). An Australian study reported additional factors associated with unplanned readmissions, which included old age, multi‐morbidity, presence of chronic disease, previous emergency presentations and hospital admissions or longer length of stay during prior hospital admissions (Considine 2019).
Multi‐morbidity is defined as the co‐existence of multiple chronic diseases, or "health problems that require ongoing management over a period of years or decades" (WHO 2002). Multi‐morbidity is known to affect a person's quality of life or is associated with functional decline, poorer patient health outcomes (AIHW 2012; Salisbury 2013; Sasseville 2019; Warner 2017), and increased health care utilisation, including emergency admissions. Prominent chronic diseases are cardiovascular diseases, cancer, chronic obstructive pulmonary disease and type 2 diabetes (WHO 2002). These conditions alone have been reported to result in an average 13‐fold longer hospital stay and a five times greater risk of readmission (Longman 2012; Rosted 2016). A recent study showed that having a higher number of chronic diseases or body systems involved was strongly associated with a higher number of potentially avoidable readmissions (Aubert 2019). Admission rates for chronic conditions where hospital admission can generally be avoided with appropriate care are highest in areas of greater socioeconomic disadvantage (Page 2007). Although unplanned hospital presentations are multifactorial, at‐risk groups are likely to have difficulty in accessing health services within the context of declining health. There is evidence to suggest that care navigation may improve the person's outcome and access to services that may reduce unplanned hospital presentations in at‐risk people (Balaban 2015).
Description of the intervention
Care navigation is the provision of services by a third party who co‐ordinates the care of the individual between multiple healthcare and service providers, but is not directly involved in delivering clinical care. Care navigation is defined as “a partnership between a patient or caregiver and a navigator that seeks to proactively guide patients through the healthcare continuum to facilitate timely access to care and foster self‐management and autonomy through education and emotional support (AMA 2012; Luke 2018; Psooy 2004). The goal is to ensure that people receive the right care at the right time, in the right place with the right outcome. Care navigation was initially used to support people with cancer to access hospital and community services (Freeman 2011; Gilbert 2011). It has since been adopted in other groups of people with chronic and complex conditions, for example, people with HIV (Steward 2018), older people with chronic diseases (Manderson 2012; McBrien 2018), and homeless people (Bishop 2009). Care navigation can take place in a single discipline or in a multidisciplinary team.
Care navigation is a type of care that applies a care co‐ordination strategy, with care co‐ordination encompassing a broad and diverse range of care types. It is commonly used to manage chronic diseases, with a view to reducing unplanned hospital presentations (Ouwens 2005), and improve PROMS. Other common types of care co‐ordination include: (i) discharge support, which focuses on transition from hospital to a postdischarge setting and may include discharge planning, patient education and shared medical information (Gonçalves‐Bradley 2016); and (ii) shared care, which involves clinical management shared by two clinicians, whereby two parties equally contribute to the planned delivery of care (Smith 2017).
Care navigation is not the same as hospital discharge care planning. Hospital discharge care planning is when the individual, carer, family and any staff involved make the necessary arrangements to ensure there is a smooth transition from hospital to home. Discharge planning is the development of a personalised plan for each person who is leaving hospital, with the aim of containing costs and improving patient outcomes (Gonçalves‐Bradley 2016; Jack 2009). By contrast, care navigation is a continuum of care with support facilitated by a trained person and not necessarily initiated during a hospital admission. Care navigation can also be integrated with other community services to prevent unplanned hospital presentations. Therefore, this review will not cover the following care types: hospital discharge planning, discharge support and shared care.
Care navigation can be delivered in the hospital or in the community. In this review, we will specifically focus on care navigation delivered predominantly in the community and provided by a care navigator. A care navigator may be an individual with or without clinical expertise, for example, a lay person or a trained professional, who supports individuals to navigate through health services. Some studies have described the role of care navigator as a case manager when the person who delivers the role is trained with clinical expertise. Therefore we will include studies if they used the term ‘case manager’ to describe the person providing the care navigation intervention, as long as the description for case manager meets our prespecified definition of care navigator. In this review, a care navigator is neither a lay person nor a family member or a carer; the care navigator is a trained clinician who has formal professional/paraprofessional tertiary education and is involved in paid care. This allows the care navigator to help identify the individual’s clinical needs, with an understanding of their clinical conditions and hospital service utilisation. The care navigator also has in‐depth knowledge of healthcare systems, e.g. local service availability, service eligibility criteria, and which services are likely to meet the needs of clients. Care navigator roles have been defined and mapped to ensure validity and professional competency standards (Carter 2018; Willis 2013). The trained clinician can be of any discipline in healthcare, and frequently includes hospital or community nurses, social workers, occupational therapists, physiotherapists, dieticians or pharmacists. The care navigator co‐ordinates the care of the individual between multiple healthcare and service providers but is not directly involved in delivering clinical care. Their role is to help the person navigate their local healthcare and support networks, and advocate for the person’s health care, as well as social and support needs.
How the intervention might work
People with complex needs and multi‐morbidities require long‐term holistic care to manage their progressive chronic conditions. Lack of understanding of their chronic disease management and physiological changes to their health also mean that their healthcare needs change over time. In general, people with complex conditions and multi‐morbidities rely on their primary care providers to manage their care needs. The range of community services available to support people with complex multi‐morbidities varies between countries and healthcare systems. The services are often highly fragmented and come from a wide range of sources, both formal and informal, including health and social care services, family, friends and neighbours. Inability to meet care needs in the community, especially in the presence of progressive chronic diseases, may result in unplanned hospital presentations and worsening of health outcomes. When the person is hospitalised, the acute medical care teams mainly focus on treating the immediate health concerns. The treatment provided may often be only a partial solution to the person's needs and may not necessarily respond to their preferences. Lack of communication and inconsistent process of care are common during transition to home in this group (Mansukhani 2015). Some people may present with other healthcare and psychosocial needs that require different follow‐up and support. Consequently, it is often very difficult for people with complex conditions to co‐ordinate health service access and advocate for their own needs.
As a result, follow‐up care may become a series of standard short‐term care services, without providing person‐centred care. This management pattern can become cyclic as people re‐present to hospitals repeatedly in response to existing services not meeting their needs. Having a trained professional to proactively guide the person through the healthcare continuum and advocate for their needs, assist in organising medical appointments specific to their needs and foster self‐management, education and support for the person and their carer may help break this cycle and reduce unplanned hospital presentations in this group. Care navigation used in various population groups has reported indications of optimising continuity of care and assisting access to services specific to individual healthcare needs, thereby improving PROMS, patient satisfaction and healthcare processes. As such, care navigation has the potential to reduce unplanned hospital presentations and improve PROMS through enhanced patient management (Prieto‐Centurion 2019).
Why it is important to do this review
This lack of a clear and unified care navigation definition has contributed to inconsistency in the evidence to support the effectiveness of care co‐ordination as an intervention. Previous Cochrane Reviews have predominantly focused on other types of care co‐ordination. Results from these reviews report that ‘shared care’ interventions showed little improvement on clinical outcomes, mixed results for PROMS and little or no difference for hospital admission rates, process of care, participation rate and participant health behaviours (Smith 2017). By contrast, 'discharge support’ interventions showed a small reduction in hospital service utilisations, such as, hospital admission and length of stay; improved participant satisfaction and lower short‐term health service costs (Gonçalves‐Bradley 2016). Smith 2016 also performed a Cochrane Review examining interventions to improve outcomes in people with multi‐morbidities. Although the interventions were not specifically restricted to care co‐ordination, most (approximately two thirds) involved some form of this care, most commonly case management. Some improvements were demonstrated for PROMS as well as mental health and functional outcomes.
Non‐Cochrane systematic reviews have predominantly included general care co‐ordination, involving a variety of intervention types with outcomes focused on reducing unplanned hospital presentations. Hansen 2011 examined discharge support, whereas Soril 2015 and Leppin 2014 examined the overview concept of care co‐ordination, i.e. a mixture of discharge support, shared care and care navigation. Results from these reviews showed varying results. Hansen 2011 suggested that no single type of intervention implemented alone was regularly associated with reduced risk of 30‐day rehospitalisation. Soril 2015 found that case management interventions yielded moderate cost savings, but resulted in variable reductions in ED use. Leppin 2014 concluded that case management and supportive interventions resulted in reduced hospital admissions.
Although previous reviews have reported some benefits of care co‐ordination in reducing unplanned hospital presentations and improving patient outcomes, the effectiveness of specific care types and the most beneficial components of complex interventions in terms of these outcomes are not clear. In particular, there are no systematic reviews that specifically address the effectiveness of care navigation as a subset of care co‐ordination. Further evidence is needed specific to care navigation to guide clinical and policy decision making and assess the effects of care navigation interventions on unplanned hospital presentations, PROMS (quality of life, well‐being, measures of disability or functional status), patients‐reported experience measures, hospital service utilisations and associated adverse events.
Objectives
Main objective
To assess the effects of care navigation, delivered in the community, on hospital presentations and PROMS in those at‐risk of unplanned hospital presentation.
Secondary objective
To assess whether the effects of care navigation on the proposed outcomes differ according to the type of clinician delivering the intervention and the population groups receiving the intervention.
Methods
Criteria for considering studies for this review
Types of studies
We will include eligible randomised trials and cluster‐randomised trials RCTs. The rationale for including cluster‐randomised trials is that it can be difficult to implement individually‐randomised trials to assess the effect of care navigation on unplanned hospital presentations and PROMS. Additionally, this is a relatively new research area with few randomised trials. Cluster‐randomised trials focus on implementation studies, that is, requiring a change of practice in the study setting, and with the model of care delivered in a group setting. Eligible cluster‐randomised trials must involve at least two control and two intervention sites. We will include full‐text studies, conference abstracts, and unpublished data. We will include studies irrespective of their publication status and language of publication.
Types of participants
Participants can be people with any condition, residing in the community (including but not limited to: home, aged care facilities or supported care), who may be at risk of unplanned hospital presentations. There are no restrictions on the types of study populations that can receive the intervention. However, if feasible we will perform subgroup analyses on individuals with single conditions or multiple comorbidities if the care navigation is delivered differently due to the clinical nature of the groups, such as those with mental health concerns.
Types of interventions
Care navigation is a complex intervention involving multiple components, designed for individual needs. Commonly, these components include all or some of the following.
-
Development of a community‐based healthcare plan
-
Discharge planning and ensuring care plans are sent to general practitioners (GPs)
-
Follow‐up phone calls to the person
-
Organising follow‐up appointments with GPs, specialists and outpatient care
-
Organising referrals for appropriate medical and mental health assessments
-
Facilitating timely access to care
-
Fostering self‐management, education and support for the person and their carer
-
Proactively guiding the person through the healthcare continuum
-
Communicating with health and community care providers
-
Helping the person get community care and support services
-
Assisting with appointment, planning and transport
-
Planning for future health service needs
-
Reducing barriers to care, including geographic, cultural, socioeconomic, and organisational barriers (Esparza 2013)
Care navigation may involve a number of strategies. We will classify care navigation interventions as 'simple' if they used only one of these approaches, and as 'multi‐faceted' if they incorporated more than one feature of care navigation.
Comparison
We will include trials comparing care navigation with usual care. Usual care is defined as the routine care received by people for prevention or treatment of diseases (Harlapur 2013).
Types of outcome measures
We will report the following outcomes.
Primary outcomes
-
Unplanned hospital presentation rate, defined as:
-
unplanned hospital admission(s); or
-
emergency department presentation(s) that occur within one month (28 or 30 days), three months (90 days) or 12 months of delivery of the intervention.
-
-
PROMS: self‐reported quality of life (well‐being, health status or functional status) reported using quantitative scores measured at one month (28 or 30 days), three months (90 days) or 12 months of delivery of the intervention.
Secondary outcomes
-
Treatment satisfaction: for example, but not limited to, acceptability of the service to recipients and providers
-
Utilisation of community services: for example, but not limited to, hospital outpatient, GP, ambulatory, community and ambulance services
-
Quality of care (process measures): participation and default rates, participant adherence to treatment/care plans and non‐attendance of appointments
-
Adverse events
Search methods for identification of studies
Electronic searches
The Cochrane Effective Practice and Organisation of Care (EPOC) Information Specialist developed the search strategies in consultation with the review authors (Appendix 1).
We will search the following databases for primary studies and related systematic reviews, without language and date restrictions, from inception to the date of search.
-
Cochrane Database of Systematic Reviews (CDSR)
-
Epistemonikos
-
Cochrane Central Register of Controlled Trials (CENTRAL; current issue), in the Cochrane Library
-
MEDLINE In‐Process and other non‐indexed citations, and MEDLINE OvidSP (1946 to present)
-
Embase Ovid (1974 to present)
-
CINAHL EBSCO (Cumulative Index to Nursing and Allied Health Literature; 1982 to present)
Searching other resources
We will also search the following databases.
-
WHO ICTRP (World Health Organization International Clinical Trials Registry Platform; www.who.int/ictrp; to present)
-
US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov; to present)
We will review reference lists of all included studies and relevant systematic reviews for additional potentially eligible primary studies. We will contact authors of included studies/reviews to clarify reported published information and to seek unpublished results/data. We will contact researchers with expertise relevant to the review topic/EPOC interventions. We will provide appendices for all strategies used, including a list of sources screened and relevant reviews/primary studies reviewed.
Data collection and analysis
Selection of studies
We will download all titles and abstracts retrieved by electronic searching to a reference management database and remove duplicates. Two review authors (RP and BS) will independently screen titles and abstracts for inclusion. We will retrieve the full‐text study reports/publications and two review authors (RP and BS) will independently screen the full text to identify studies for inclusion. We will identify and record reasons for exclusion of the ineligible studies. We will resolve any disagreement through discussion or, if required, we will consult a third review author (CW or NA).
Data extraction and management
We will use the EPOC standard data collection form and adapt it for study characteristics and outcome data (EPOC 2017a); we will pilot the form on at least one study in the review.
Two review authors (RP and BS) will independently extract the following study characteristics from the included studies and enter the data into Review Manager 5.4.1 (Review Manager 2020). The data to be collected follows the reporting guidelines for systematic reviews as stated below, which should adequately describe the included studies, support the construction of tables and figures, facilitate the risk of bias assessment, and enable syntheses and meta‐analyses.
-
Information about data extraction from reports
-
Name of data extractor, date of data extraction, and identification features of each report from which data are being extracted
-
-
Eligibility criteria
-
Confirm eligibility of the study for the review
-
Reason for exclusion
-
-
Study methods
-
Study design
-
Parallel, factorial, cross‐over, cluster aspects of design for randomised trials
-
Single or multicentre study; if multicentre, number of study/recruiting centres and locations
-
-
Recruitment and sampling procedures used (including at the level of individual participants and clusters/sites if relevant)
-
Enrolment start and end dates; length of participant follow‐up
-
Details of random sequence generation, allocation sequence concealment, and masking for randomised trials, and methods used to prevent and control for confounding and selection biases
-
Methods used to prevent and address missing data
-
Statistical analysis
-
Unit of analysis (e.g. individual participant, episode of care, clinic/hospital attendance, readmission rate)
-
Statistical methods used if computed effect estimates are extracted from reports, including any covariates included in the statistical model
-
-
Likelihood of reporting and other biases
-
Source(s) of funding or other material support for the study
-
Authors’ financial relationship and other potential conflicts of interest
-
-
Participants
-
Setting
-
Region(s) and country/countries from which study participants were recruited
-
Study eligibility criteria, including diagnostic criteria
-
Characteristics of participants at the beginning (or baseline) of the study (e.g. age, mean age, age range, gender, comorbidity, socioeconomic status)
-
-
Intervention
-
Description of the intervention(s) and comparison intervention(s), ideally with sufficient detail for replication
-
Components, intervention protocols, length of intervention
-
Factors relevant to implementation (e.g. staff qualifications, training and equipment requirements)
-
Integrity of interventions (i.e. the degree to which specified procedures or components of the intervention were implemented as planned)
-
Description of co‐interventions (single vs multiple component of intervention)
-
Definition of ‘control’ groups (e.g. no intervention or components of usual care)
-
Components involved in delivering intervention (ambulance call‐outs)
-
Fidelity assessment
-
-
-
Outcomes
-
For each prespecified outcome domain (e.g. unplanned hospital presentations and PROMS) in the systematic review
-
Whether there is evidence that the outcome domain was assessed (especially important if the outcome was assessed but the results not presented)
-
Measurement tool or instrument (including definition of clinical outcomes or endpoints); for a scale, name of the scale (e.g. the Charlson comorbidity index), upper and lower limits, and whether a high or low score is favourable, definitions of any thresholds if appropriate
-
Specific metric (e.g. quality of life scores from baseline to a postintervention time point)
-
Method of aggregation (e.g. mean and standard deviation of unplanned hospital presentation rates in each group, or proportion of people who represented to hospital)
-
Timing of outcome measurements (e.g. assessments at one month (28 or 30 days), three months (90 days) or 12 months)
-
Adverse outcomes need special attention, depending on whether they are collected systematically or non‐systematically (e.g. by voluntary report).
-
-
-
Results
-
For each group, and for each outcome at each time point: the number of participants randomly assigned and included in the analysis; and the number of participants who withdrew, were lost to follow‐up or were excluded (with reasons for each).
-
Summary data for each group (e.g. 2×2 table for dichotomous data; means and standard deviations for continuous data)
-
Between‐group estimates that quantify the effect of the intervention on the outcome, and their precision (e.g. risk ratio, odds ratio, mean difference)
-
If subgroup analysis is planned, the same information will be extracted for each participant subgroup.
-
-
Miscellaneous
-
Key conclusions of the study authors
-
Reference to other relevant studies
-
Correspondence required
-
Miscellaneous comments from the study authors or by the review authors
-
Two review authors (RP and BS) will independently extract outcome data from the included studies, evaluate study quality, and judge the certainty of the evidence using the GRADE approach. We will conduct a meta‐analysis of the results where possible and carry out a narrative synthesis for the remainder of the results, including any qualitative results reported in the secondary outcomes. We will present the results in a summary of findings table and tabular format to show effect sizes across all outcome types. We will note in the characteristics of included studies table if publications reported outcome data in an unusable way. We will resolve disagreements by consensus or by involving a third review author (CW or NA).
If data are missing, we will contact study authors, when possible, to obtain the missing information.
Assessment of risk of bias in included studies
Two review authors (RP and BS) will independently assess risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions Version 5, Section 8 (Higgins 2011). We will resolve any disagreements by discussion or by involving a third review author (CW or NA). We will assess the risk of bias according to the following domains.
-
Random sequence generation (selection bias)
-
Allocation concealment (selection bias)
-
Blinding or knowledge of the allocated interventions by participants and personnel during the study (performance bias)
-
Blinding or knowledge of the allocated interventions by outcome assessors (detection bias)
-
Selective outcome reporting (reporting bias)
-
Amount, nature or handling of incomplete outcome data (attrition bias)
-
Problems not covered elsewhere (other bias for the cluster‐randomised trials, as listed below)
We will use the Cochrane RoB tool to assess risk of bias, which reflects current understanding of how the causes of bias can influence study results, and the most appropriate ways to assess this risk.
We will judge each potential source of bias as 'high', 'low', or 'unclear' and provide a quote from the study report together with a justification for our judgement in the risk of bias table, as suggested in Chapter 8.5 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We will summarise the risk of bias judgements across different studies for each of the domains listed. We will consider studies as:
-
'high' risk of bias where the trial has serious bias that decreases the certainty of the conclusions;
-
'low' risk of bias where it seems unlikely for bias to seriously alter the results of individual trials;
-
'unclear' risk of bias where the trial is unclear, provides insufficient details, or we judge it to have some bias that could plausibly raise doubts about the conclusions.
We will assign an overall risk of bias assessment for individual studies using the approach suggested in Chapter 8.7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Where information on risk of bias relates to unpublished data or correspondence with a trialist, we will note this in the risk of bias table. We will not exclude studies on the grounds of their risk of bias, but will clearly report the risk of bias when presenting the results of the studies.
When considering treatment effects, we will take into account the risk of bias for the studies that contribute to that outcome.
As we will be including cluster‐randomised trials, we will also consider the following additional biases, as proposed in chapters 9 and 16 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
-
Identification or recruitment bias
-
Baseline imbalance
-
Incorrect outcomes analysis
-
Compatibility with individual randomised trial
We will perform and measure approximately correct analyses for cluster‐randomised trials.
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.
Measures of treatment effect
We will estimate the effect of the intervention using the mean difference or standardised mean difference for continuous data, together with the 95% confidence interval (Higgins 2019). We will ensure that an increase in scores for continuous outcomes can be interpreted in the same way for each outcome, explain the direction to the reader, and report where we reversed the directions, if this was necessary.
Unit of analysis issues
The participant will be the unit of analysis for individually‐randomised trials. If we include cluster‐randomised trials, the cluster will be the unit of analysis. We will adjust the unit of analysis for data clustering. If the study does not report the number of clusters, we will attempt to contact the authors. For studies that involved multiple intervention types, we will combine the intervention groups to create a single pair‐wise comparison to avoid unit of analysis errors.
Dealing with missing data
If there is evidence of missing data we will contact 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). If we consider the data to be missing at random, we will analyse the available information. If we do not consider the data to be missing at random, we will make explicit assumptions of any methods we use to cope with the missing data, for example that we assumed the missing values indicate a poor outcome for dichotomous outcomes. We will contact trial authors to seek clarification regarding descriptions of interventions (e.g. setting, mode of delivery, format, duration etc.), trial conduct (e.g. method of random sequence generation, method of allocating participants to intervention groups), and availability of unpublished data for outcomes that were measured. We will try to compute missing summary data from other reported statistics. Whenever it is not possible to obtain data, we will report the level of missingness and consider how that might impact the certainty of the evidence.
Assessment of heterogeneity
We will consider included studies in terms of clinical and statistical heterogeneity. We will explore clinical heterogeneity by examining potentially influential factors, e.g. participants and intervention components, and will report these in the characteristics of included studies table. We will explore statistical heterogeneity when undertaking meta‐analysis. We will use the I2 statistic to measure heterogeneity among the trials in each analysis. This examines the percentage of total variation across the studies due to heterogeneity rather than to chance. The level of heterogeneity is defined as suggested in the Cochrane Handbook for Systematic Reviews of Interventions Chapter 10.10 (Higgins 2019).
-
0% to 40%: might not be important;
-
30% to 60%: may represent moderate heterogeneity;
-
50% to 90%: may represent substantial heterogeneity;
-
75% to 100%: considerable heterogeneity.
If we identify considerable heterogeneity i.e. I2 > 75%, we will explore it by prespecified subgroup analysis.
Assessment of reporting biases
We will assess reporting bias by comparing outcomes listed in the methods section versus those reported in the results section and, when possible, we will check outcomes in published protocols.
We will attempt to contact study authors, asking them to provide missing outcome data. Where this is not possible, and we think the missing data could introduce serious bias, we will explore the impact of including such studies in the overall assessment of results.
Data synthesis
Data analysis will be based on the list of outcome measures defined in this protocol. We will conduct meta‐analysis using RevMan 5.4.1 when there are two or more separated studies reporting sufficient data for the outcome. We will use standardised mean differences in meta‐analyses when different scales are used to report the same outcome. Because of the nature of the intervention, we expect to see clinical heterogeneity among studies. If pooling is possible, we will undertake meta‐analyses using random‐effects models, and will use forest plots to present outcomes. Where quantitative synthesis is not possible, i.e. if analyses indicate substantial heterogeneity (I2 > 50%) or if studies have too much clinical or methodological diversity, we will perform a non‐quantitative synthesis. We will use a narrative synthesis to describe the overall effect of the quantitative results related to the included studies, as suggested in Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2019). These acceptable synthesis methods include summarising effect estimates, combining P values or vote counting based on direction of effect. We will describe the limitations of the chosen methods and will word our conclusions with appropriate caution to make the synthesis process more transparent and reproducible, and ensure that use of methods and interpretation is appropriate. The review will not include a secondary analysis of qualitative data reported in eligible studies. However, we will provide a descriptive narrative of relevant qualitative results within the context of the review.
When meta‐analysis is not possible due to limited evidence for a prespecified comparison, we plan to combine one or more of the population groups, intervention or outcomes at a broader level. If the reported outcome or effect is incomplete, we will calculate the effect estimates, summary effect estimates and measures of precision e.g. standard deviations, from the available statistics. If the effect measures are different across studies, or we identify bias in the evidence, we will transform standardised mean differences into hazard ratios or odds ratios, as outlined in the Cochrane Handbook for Systematic Reviews of Interventions Section 12 (Higgins 2019).
Subgroup analysis and investigation of heterogeneity
We plan to perform the following subgroup analyses if we find substantive variations in the included studies and there is sufficient information available to perform a subgroup analysis.
-
Care navigation as a single intervention type versus care navigation as part of combination or multiple interventions: care navigation that includes care navigation alone versus models that do not just consist of care navigation. This is important because different levels and formulations of care navigation, i.e. provision of care and carer support, are likely to have a degree‐of‐service response depending on the level and combination of components of care navigation (Dhalla 2014; Koehler 2009; Plant 2015), which we would like to test. We will examine the degree‐of‐service related impact by level of care navigation (i.e. model of care or number of staff involved) to ascertain how strong the effect of care navigation was in each trial.
-
Number of contacts impact on effectiveness of outcome: lower number of contacts (i.e. one follow‐up) versus multiple visits (i.e. more than one) at different time points. This is important because our initial review of the literature showed that studies varied in the number of contacts provided to participants, with multiple contacts likely to have better outcomes, and we would like to test this. We will record the actual number of contacts for each trial in the characteristics of included studies table.
-
Individuals with single versus multiple comorbidities. This is important because care navigation intervention will require different skillsets/understanding for those with one chronic condition compared to those with multiple comorbidities (e.g. chronic obstructive pulmonary disease and chronic heart failure). In addition, we will consider a subgroup analysis for specific health conditions (e.g. mental health or cancer) as care navigation intervention delivered in these groups may vary. We will record the actual number of morbidities for each trial in the characteristics of included studies table.
-
Care navigation intervention to support any patient group conducted in high income countries versus those conducted in middle‐ to low‐income countries. This is important because the availability, nature, and scope of health and social care services, support and integration will vary between countries.
We will perform tests for interaction for subgroup analyses, and use meta‐regression techniques to test for subgroup interactions providing that sufficient studies are available (i.e. five or more).
Sensitivity analysis
We will perform sensitivity analyses if sufficient data are available to assess the robustness of effect estimates for our primary outcomes. Sensitivity analysis will involve restricting the analysis to:
-
published studies;
-
studies with low risk of bias, for example, selection bias (excluding trials with inadequate or unclear allocation concealment), detection bias (excluding trials with unclear or inadequate blinding of outcome assessors) and attrition bias (excluding trials with incomplete outcome data);
-
imputing missing data; and
-
by 10‐year publication group to account for likely changes over time.
Summary of findings and assessment of the certainty of the evidence
We will assess the evidence for care navigation using GRADE criteria. GRADE assessments of certainty are determined through consideration of five domains: risk of bias, inconsistency, indirectness, imprecision and publication bias. Two review authors (RP and BS) will independently assess the certainty of the evidence, and we will present our judgements in a summary of findings table that includes the following outcomes: unplanned hospital presentation rate, self‐reported quality of life, treatment satisfaction, utilisation of community services, quality of care, and adverse events, as outlined in the section 'Types of outcome measures'. These outcomes are important to guideline, policy or other stakeholders. We will use the methods and recommendations described in the Cochrane Handbook for Systematic Reviews of Interventions Section 8 and Chapter 12 (Higgins 2019), and use GRADEpro software.
The GRADE approach specifies four levels of the certainty for a body of evidence for a given outcome: high, moderate, low and very low. We will upgrade and downgrade the certainty of the evidence with justifications, if we have concerns about study limitations, risk of bias (lack of blinding), consistency of effect (changes in population demographic), imprecision and indirectness (number of studies), and publication bias. We will use the EPOC worksheets to guide this process (EPOC 2017b). If a new outcome is identified during the review process, we will include the relevant outcome and explain the reasons for this is the section 'Differences between protocol and review'.
