Abstract
This systematic review examined the effectiveness of nurse care coordinator (CC) roles on patient-reported and health service outcomes. Multiple electronic databases (Medline, CINAHL, and EMBASE) were searched and the Cochrane Risk of Bias Assessment Tool was applied by two independent reviewers. The Grades of Recommendation, Assessment, Development, and Evaluation system was used to assess the quality of evidence. A total of 45 articles (reporting on 36 studies) were included. The majority of studies (n = 28, 78%) were conducted in the United States and published after 2009 (n = 24, 67%). Thirteen studies (36%) used a randomized controlled trial design. A total of 17 studies evaluated patient-reported outcomes and 29 studies reported health service outcomes. The individual components of nurse CC roles that were evaluated ranged considerably. The impact of nurse care coordination on patient-reported and health service outcomes was inconsistent. There was an indication from higher quality studies that nurse care coordination roles were more likely to result in improved patient and health service outcomes where they involved frequent, in-person interactions, had ongoing follow-up with monitoring of disease status, and involved transition care and the application of behavior change principles.
The fragmented process in which health care is typically delivered is failing people who have complex needs, potentially resulting needlessly in increased morbidity and distress as well as increased overall costs to the system. Improving the coordination of health care is one solution to this issue. Care coordination refers to a “person-centred, assessment based, interdisciplinary approach to integrating health care and social support services in a cost-effective manner in which an individual’s needs and preferences are assessed, a comprehensive care plan is developed and services are managed and monitored by an evidence-based process which typically involves a designated lead care coordinator” (National Coalition on Care Coordination, 2011, p. 1). Internationally, an increasing number of health services are employing nurses into dedicated care coordinator (CC) positions as a strategy to improve coordination of care (Nutt & Hungerford, 2010). This systematic review examined the literature with regard to CC roles performed by nurses to assist future development and implementation. The review was structured according to the following specific aims:
To describe the various components of the CC role within the health system; and To examine the impact of the CC role and its various components on patient outcomes, and service delivery, from the perspective of patients, clinicians, and those working in the role;
Method
Criteria for Including Studies in the Review
Types of studies
Studies that used randomized controlled trial, nonrandomized controlled trial, quasi-experimental, and pre–posttest designs were included.
Types of clinical settings
Studies that included participants who received health care in hospital and/or community settings were included.
Types of participants
Studies that included participants of any age (adults and/or children) receiving treatment for a health condition (acute or chronic) were included. Obstetric care was not considered as a treatment in this review, so any studies that addressed pregnancy care were not included.
Types of interventions
Studies that included a group of participants who were allocated a nurse who provided them with a dedicated care coordination service were included. Studies that included a group of participants who were allocated a dedicated CC to manage cancer were excluded because this was the subject of a recent comprehensive systematic review (Langbecker, Hunt, & Yates, 2014). Although it is acknowledged that coordination functions are an important part of the role for all health professionals, this review is focused solely on literature relating to the effectiveness of dedicated CC roles. This review is not focused on examining the impact of other roles that have significant coordination functions, such as community mental health nurses or practice nurses. It is also not focusing on studies that examined replacing care usually delivered by medical practitioners by nurses or nurse practitioners that have significant coordination functions.
Types of comparisons
Studies with a comparison group of participants who were not allocated a CC (historical or prospective) were included. Studies that did not have a comparison group (i.e., cohort studies reporting outcomes of participants who were allocated a CC) were also included.
Types of outcomes
The types of patient-reported outcomes considered for inclusion in this review were quality of life, symptom severity (physical and psychological), concerns and problems, self-efficacy, knowledge (disease/self-management), continuity of care, and satisfaction with care (patient and family). The types of health service outcomes considered for inclusion were resource use (hospitalization, length of stay, emergency department [ED] use, outpatient/clinic use, home visit use, hospice use, doctor visits, allied health service use, community service use, and other health service use), receiving appropriate care, treatment time/delay, treatment adherence, follow-up adherence, survival, throughput, costs, coordinator satisfaction, staff-reported understanding of the CC role, and staff-reported effectiveness of CC role.
Search Methods
Our review of the literature was limited to articles published in English from January 1, 1990, to November 2015. We searched Medline, CINAHL, and Embase. Reference lists of all included articles and articles excluded following full-text screening (including reviews) were reviewed to identify additional literature not identified through database and web-based searches. Search terms are listed in Supplementary Material.
Screening and Selection
Title and abstract screening was conducted by two reviewers independently. After the removal of articles which did not meet the initial title/abstract screening, the full text of remaining articles was obtained and screened. Some included longitudinal studies involved more than one published article where results from the same outcomes were reported at different follow-up time points. Only the report with the longest follow-up was included in this review. Some care coordination programs had also been tested using multiple research designs (e.g., a pilot study followed up by a fully powered randomized controlled trial). In these circumstances, only outcomes from better designed studies were considered in this review.
Data Abstraction and Critical Appraisal
Data were extracted from each article, including country, year of publication, study design, sample characteristics, key features of the CC role, outcome measures, and key findings. Randomized controlled trial, nonrandomized controlled trial, quasi-experimental, and pre–posttest designs were evaluated for risk of bias using domains of the Cochrane Risk of Bias Tool. Quality of evidence for outcomes was not downgraded due to performance bias because personnel could not be blinded for this form of intervention. An additional domain to assess risk of bias due to confounding was assessed for the nonrandomized studies included in the review. Nonrandomized studies included in this review were considered at high risk of bias due to confounding if the study did not (i) restrict participant selection so that all groups had the same value for the confounder, (ii) demonstrate balance between groups for the confounder, (iii) match on the confounder, or (iv) adjust for the confounder in statistical analyses to quantify the effect size. The quality of evidence from nonrandomized studies was assigned low quality or downgraded to very low quality if concerns regarding risk of bias or imprecision were detected. We used the principles of the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) system (Guyatt et al., 2008) to assess the quality of the body of evidence associated with the outcomes.
Data Synthesis
Results are presented in two sections: (1) a description of the CC roles used. Information about the individual components of CC roles were extracted and summarized in a table included in Supplementary Material and (2) an analysis of the impact of CCs on outcomes. For consistency, results refer to “CC” and “noncare coordinator (non-CC)” patients, regardless of the terminology used in the original study. It was not possible to conduct meta-analyses due to clinical heterogeneity and inconsistent reporting of outcomes. A narrative synthesis of the results from the included studies was performed.
Results
Search Results
Results of the literature search are presented in Figure 1. A total of 45 articles, which reported outcomes from 36 studies, were included in this review.
Search results and study selection.
Study Characteristics
The majority of studies (n = 28, 78%) were conducted in the United States and published after 2009 (n = 24, 67%). Thirteen studies (36%) used a randomized controlled trial design. One study reported results from the Medicare Care Coordination Demonstration project, which involved 18,309 participants randomized to receive care from a dedicated CC or control group in 15 separate sites (Peikes, Chen, Schore, & Brown, 2009). Although results from the sites were reported separately in this study, the research design used in each site was the same apart from slightly different implementation of the CC role. For this reason, risk of bias assessment was applied to the whole Medicare Coordinated Care Demonstration (MCCD) program evaluation. It should also be noted that an additional study where results from one individual MCCD site (health quality partners) was reported separately was included in this review (Coburn, Marcantonio, Lazansky, Keller, & Davis, 2012).
There was a high or unclear risk of selection bias for six randomized clinical trials (RCTs; 46%) and detection bias for 26 studies (72%). We found no clear evidence of selective reporting or other biases in the studies. Several nonrandomized studies included in this review were considered at high risk of bias due to confounding (n = 10, 43%).
A description of the study design, populations, outcomes reported, and results of risk of bias assessments for each included study are presented in Table 1. More detailed summaries of the features of the CC roles, populations they were targeted toward, and results of outcomes included in this review are provided in Supplementary Material Table S1.
Characteristics of Included Studies.
Note. RSG= randomization sequence generation; AC = allocation concealment; BO = blinded outcome assessment; IO = incomplete outcome reporting; SR = selective outcome reporting; C = confounding; O = other sources of bias; U = unclear risk of bias; H = high risk of bias; L = low risk of bias; NA = not applicable for the study design; COPD = chronic obstructive pulmonary disease; CHF = congestive heart failure; US = United States; NZ = New Zealand; RCT= randomized clinical trial; CCHT = Coordinated Care/Home Telehealth
Individual components of CC roles
CC roles targeted toward populations defined as “high risk” were common among the included studies (n = 18, 50%; Aiken et al., 2006; Aldeen, Courtney, Lindquist, Dresden, & Gravenor, 2014; Boult et al., 2013; Ciccone et al., 2010; Coburn et al., 2012; Coleman, Parry, Chalmers, & Min, 2006; Dajczman et al., 2013; Gravelle et al., 2007; Hajewski & Shirey, 2014; Hawkins et al., 2015; Kind et al., 2012; Koehler et al., 2009; Marek et al., 2013; Parsons et al., 2012; Peikes et al., 2009; Plant et al., 2015; Sidorov, Fisher, Girolami, & Wolke, 2002; Sidorov, Shull, et al., 2002). Patients with diabetes were targeted in five studies (14%; Barnett et al., 2006; Gabbay et al., 2013; Kogut, Johnson, Higgins, & Quilliam, 2012; Price, 1996; Sidorov, Shull, et al., 2002). People with dementia (Eloniemi-Sulkava, Notkola, & Hentinen, 2001; Engelhardt et al., 2008), with a terminal illness (Aiken et al., 2006; Sweeney, Halpert, & Waranoff, 2007), and children with special care needs (Berry et al., 2011; Taylor et al., 2013) were the target populations in two studies each. Other studies focused on people recovering from a stroke (Fens et al., 2015, 2014), patients with chronic obstructive pulmonary disease (COPD; Dajczman et al., 2013), people with bipolar disorder (Bauer et al., 2006), disabled with functional impairments (Jingping, Goehring, & Mancuso, 2015), and people either residing in (Boyd et al., 2014) or at high risk of requiring care in a residential aged care facility (Parsons et al., 2012).
A variety of different components were included in each CC role. It was noted that advanced practice registered nurses were employed as CC in seven studies (19%; Barnett et al., 2006; Gravelle et al., 2007; Koehler et al., 2009; Kruse et al., 2010; Marek, Adams, Stetzer, Popejoy, & Rantz, 2010; Marek et al., 2013; Parsons et al., 2012). Other studies simply noted that the CCs were registered nurses or nurses without stating their experience or qualifications for performing this role.
A frequent component of the CC roles was “transition care,” which was applied in some form in 11 studies (31%; Boult et al., 2013; Coburn et al., 2012; Coleman et al., 2006; Dajczman et al., 2013; Hawkins et al., 2015; Kind et al., 2012; Koehler et al., 2009; Marek et al., 2013; Peikes et al., 2009; Plant et al., 2015; Taylor et al., 2013; White, Carney, Flynn, Marino, & Fields, 2014). Inclusion of an action plan for situations of clinical deterioration was a component of transition care for three of these CC roles (Coleman et al., 2006; Kind et al., 2012; Koehler et al., 2009). Three studies (9%), which all evaluated care coordination programs based in the primary care setting, used a “Reach in” model of transition care that involved the CC working with in-patient teams when participants were hospitalized (Hawkins et al., 2015; Marek et al., 2013; White et al., 2014).
Development of a care plan was also a component of 11 studies (31%; Aldeen et al., 2014; Berry et al., 2011; Boult et al., 2013; Ciccone et al., 2010; Coburn et al., 2012; Engelhardt et al., 2008; Gravelle et al., 2007; Hawkins et al., 2015; Peikes et al., 2009; Sweeney et al., 2007; Taylor et al., 2013). Arranging consultations with health-care or community service providers was a feature of the CC role used in 12 studies (33%; Aldeen et al., 2014; Boult et al., 2013; Ciccone et al., 2010; Coburn et al., 2012; Eloniemi-Sulkava et al., 2001; Gabbay et al., 2013; Gravelle et al., 2007; Hajewski & Shirey, 2014; Kruse et al., 2010; Parsons et al., 2012; Plant et al., 2015; Sidorov, Fisher, et al., 2002). The CC delivered services for patients for a defined time period in eight studies (22%; Fens et al., 2014; Gabbay et al., 2013; Kind et al., 2012; Koehler et al., 2009; Plant et al., 2015; Sidorov, Fisher, et al., 2002; Sweeney et al., 2007; White et al., 2014).
Education about disease and self-management was a component of the CC role evaluated in 11 studies (31%; Aiken et al., 2006; Barnett et al., 2006; Boult et al., 2013; Dajczman et al., 2013; Engelhardt et al., 2008; Koehler et al., 2009; Kruse et al., 2010; Peikes et al., 2009; Price, 1996; Sidorov, Fisher, et al., 2002). Five studies (14%; Ciccone et al., 2010; Coburn et al., 2012; Gabbay et al., 2013; Gravelle et al., 2007; Kogut et al., 2012) used behavior change or health coaching principles to promote the uptake of recommended health maintenance practices. Interventions targeted at medication management or adherence were included in nine (25%) of the CC roles (Coburn et al., 2012; Coleman et al., 2006; Engelhardt et al., 2008; Gabbay et al., 2013; Hawkins et al., 2015; Kogut et al., 2012; Kruse et al., 2010; Marek et al., 2013; Price, 1996).
Contact with participants was made using in-home visits in eight (22%) of the CC roles included in this review (Coburn et al., 2012; Eloniemi-Sulkava et al., 2001; Engelhardt et al., 2008; Fens et al., 2014; Hawkins et al., 2015; Kind et al., 2012; Marek et al., 2013; Sweeney et al., 2007). Participants visited the CC in a clinic in four studies (11%; Bauer et al., 2006; Coburn et al., 2012; Gabbay et al., 2013; Kruse et al., 2010). Telephone follow-up was common, being used in 13 studies (36%; Barnett et al., 2006; Bauer et al., 2006; Coburn et al., 2012; Coleman et al., 2006; Eloniemi-Sulkava et al., 2001; Gabbay et al., 2013; Hawkins et al., 2015; Kind et al., 2012; Koehler et al., 2009; Kruse et al., 2010; Peikes et al., 2009; Sweeney et al., 2007; White et al., 2014). Health status of patients was monitored by the CC in eight (22%) studies (Barnett et al., 2006; Boult et al., 2013; Gabbay et al., 2013; Gravelle et al., 2007; Hawkins et al., 2015; Marek et al., 2013; Price, 1996; Sweeney et al., 2007), including one study that utilized daily telemonitoring (Barnett et al., 2006). It was noted in five studies (14%) that the CC was the designated key contact person for patients or that they were available to be contacted (Dajczman et al., 2013; Eloniemi-Sulkava et al., 2001; Gabbay et al., 2013; Kruse et al., 2010; Taylor et al., 2013).
Physician collaboration was a core feature of the CC role employed in six studies (17%; Bauer et al., 2006; Boyd et al., 2014; Coburn et al., 2012; Dajczman et al., 2013; Hajewski & Shirey, 2014; Parsons et al., 2012), whereas the CC provided education to patients about how to communicate with health-care team in three further studies (9%; Engelhardt et al., 2008; Peikes et al., 2009; Taylor et al., 2013). One study reported that the CC would communicate with members of the health-care team on the patient’s behalf if it was required (Sweeney et al., 2007).
Impact of the CC Role on Patient-Reported Outcomes
Quality of life
Twelve studies reported about the effect of care coordination on health-related quality of life (Supplementary Material Table S2). Moderate- and high-quality evidence from four RCTs, in addition to the report of outcomes from the 15 MCCD sites, indicated that receiving care from a CC did not have an impact on health-related quality of life for elderly patients or those with chronic disease (Boult et al., 2013; Gabbay et al., 2013; Parsons et al., 2012; Peikes et al., 2009; Plant et al., 2015). In contrast, high-quality evidence from a trial that enrolled elderly patients who had impaired ability to manage medications identified that intervention participants had better quality of life than the control group (Marek et al., 2013). It should be noted that CC in this trial involved more frequent contact with participants compared with the other trials (CC provided biweekly in-home visits for the duration of follow-up). Bauer et al. (2006) reported that the participants with bipolar disorder who received care from the CC had better scores in the mental component summary score of the SF-36 (Medical Outcomes Short Form-36 item health survey) but not the physical component score. Although greater quality of life was reported by the CC group in a trial that enrolled patients with a terminal illness (Aiken et al., 2006), the evidence rating was downgraded to low quality due to attrition bias (only 45% of the 190 participants included in analysis).
Implementation Recommendations.
Psychological morbidity
Six studies reported outcomes that were grouped under the term psychological morbidity (Supplementary Material Table S3). High-quality evidence from a trial of a care coordination role that involved biweekly home visits from an advanced practice nurse for elderly patients with impaired ability to manage medications reported that depression scores were significantly better than the control group (Marek et al., 2013). Moderate-quality evidence from a randomized controlled trial of nurse case management of diabetes that involved care coordination found that this intervention reduced depression symptoms at the 2-year follow-up time point (Gabbay et al., 2013). In contrast, no effect of CC on depression was observed in any of the 15 programs included in the MCCD (Peikes et al., 2009). Only 3 of the 15 MCCD programs produced a moderate improvement in emotional distress (Peikes et al., 2009). However, it should be noted that a validated tool to measure severity of depressive symptoms or emotional distress was not used in this study and we downgraded the quality of evidence to moderate quality for this reason. The other included studies that reported on psychological morbidity were nonrandomized comparisons and the evidence was rated as low or very low quality.
Physical symptoms
Four studies reported on the effect of CC roles on physical symptoms (Supplementary Material Table S4). Two of the 12 programs from the MCCD identified moderate improvement in sleep (defined as a rating of being bothered by poor sleep most or all of the time) and pain (defined as pain interfering with usual activities; Peikes et al., 2009). However, it should be noted that a validated tool to measure these symptoms was not used in this study and we downgraded the quality of evidence to moderate quality for this reason.
Evidence regarding the effect of a CC on physical symptoms from another RCT was also downgraded to moderate due to risk of attrition bias and imprecision (Aiken et al., 2006). The effect of a CC on physical symptoms was inconsistent. Participants randomized to a CC who had a diagnosis of COPD reported lower symptom distress than the control group while participants randomized to CC who had a diagnosis of congestive heart failure reported higher symptom distress. The authors noted that the criteria used for study entry resulted in one third of the sample dying or being transferred to a hospice within the first 3 months (Aiken et al., 2006).
Satisfaction with care and CC
High-quality evidence from two RCTs was identified regarding the effect of CC on satisfaction with care (Supplementary Material Table S5). In a trial of CC for bipolar disorder, participants who were randomized to the CC reported greater satisfaction with care (measured using the patient satisfaction index) than the control group (p < .001; Bauer et al., 2006). Likewise, elderly patients at high risk of readmission who were randomized to a CC also reported greater satisfaction with care (measured with the patient assessment of chronic illness care; Boult et al., 2013). In contrast, no difference in Diabetes Treatment Satisfaction Questionnaire scores were observed in a randomized trial at 2-year follow-up (Gabbay et al., 2013). There was also low-quality evidence identified that nearly all patients (>97%) recovering from a stroke who received care from a CC were satisfied both with their care in general and care provided by the CC in particular (Fens et al., 2014).
Only low-quality or very low-quality evidence from either single group or nonrandomized comparisons was identified for the outcomes of family-reported and staff-reported satisfaction with care and CCs (Supplementary Material Tables S6–S9). All studies reported positive satisfaction with care and CCs, including the two that enrolled children (Berry et al., 2011; Engelhardt et al., 2008; Fens et al., 2015, 2014; Taylor et al., 2013).
Impact of the CC Role on Health Service Outcomes
Hospitalizations
The effect of a CC on hospitalizations was reported by 18 studies (Supplementary Material Table S10; Boyd et al., 2014; Hajewski & Shirey, 2014). High-quality evidence was identified from the MCCD program that CCs did not impact hospitalizations (Peikes et al., 2009). A statistically significant reduction in average annualized rate of hospitalizations was observed in the CC group for only 1 of the 15 programs included in the evaluation (Peikes et al., 2009). One additional program, Health Quality Partners, resulted in 11% (not statistically significant) reductions in hospitalizations for the CC group (Peikes et al., 2009). The differences between CC and usual care in this program were concentrated in the “high-risk” group, suggesting that targeting this subpopulation for CC intervention may be most appropriate and requires further investigations (Peikes et al., 2009).
Further high-quality evidence from a trial conducted within an ED in Australia also identified that rates of hospitalizations were not different between groups over 2 years follow-up (Plant et al., 2015). Evidence from a matched pair cluster RCT conducted with elderly patients identified to be at high risk of readmission was downgraded to moderate quality due to imprecision (Boult et al., 2013). Reductions in the rate of overall hospital admissions (6% less in CC group) and 30-day readmissions (13% less in CC group) were identified but the difference between groups was not statistically significant (Boult et al., 2013).
In contrast, the majority (n = 9, 69%) of the low-quality evidence derived from studies that utilized nonrandomized comparison group and single group pre–postdesigns reported that significant reductions in hospitalizations were associated with CCs (Barnett et al., 2006; Boyd et al., 2014; Dajczman et al., 2013; Engelhardt et al., 2008; Hawkins et al., 2015; Jingping et al., 2015; Sweeney et al., 2007).
Length of stay
There were 12 studies that reported on the effect of a CC on length of stay included in this review (Supplementary Material Table S11). Moderate-quality evidence from the two RCTs indicated that there was no difference in hospital length of stay (Bauer et al., 2006; Plant et al., 2015). Low-quality and very low-quality evidence from several nonrandomized studies was inconsistent (Aldeen et al., 2014; Barnett et al., 2006; Dajczman et al., 2013; Engelhardt et al., 2008; Hajewski & Shirey, 2014; Kind et al., 2012; Koehler et al., 2009; Kruse et al., 2010; Sidorov, Shull, et al., 2002; Sweeney et al., 2007).
ED use
A total of 11 studies reported on the effect of a CC on ED utilization (Supplementary Material Table S12). There were no statistically significant differences identified between CC and non-CC groups in the three RCTs that reported on this outcome (Aiken et al., 2006; Boult et al., 2013; Plant et al., 2015). However, the quality of evidence was downgraded to moderate quality due to imprecision (Boult et al., 2013; Plant et al., 2015) or attrition bias (Aiken et al., 2006).
The low-quality and very low-quality evidence was less consistent. The nonrandomized studies with small sample sizes tended to produce more positive results for CC on ED use (Dajczman et al., 2013; Kruse et al., 2010; Sweeney et al., 2007). By contrast, a propensity score matched study of the implementation of the Evercare care coordination program into nine National Health Service primary care trusts in the United Kingdom identified that CC did not reduce the likelihood of an admission to the ED (Gravelle et al., 2007). Another large nonrandomized study which was focused on patients with chronic disease failed to detect a statistically different difference between CC and non-CC groups in the mean number of ED visits over 2 years follow-up (Sidorov, Fisher, et al., 2002).
Other health service use
Two nonrandomized studies reported on outpatient/clinic use (Supplementary Material Table S13; Barnett et al., 2006; Kruse et al., 2010). No statistically significant differences between groups were identified. One RCT reported a 29% reduction in home health-care use for the group of participants who were randomized to CC (Supplementary Material Table S14; Boult et al., 2013). Five studies reported on the effect of CC on use of primary care services (Supplementary Material Table S15 and S16). One used an RCT design (Boult et al., 2013). No difference between groups was identified downgraded to moderate quality due to imprecision (Boult et al., 2013). The low-quality evidence from three studies that used nonrandomized comparison groups was inconsistent.
Receipt of appropriate care and treatment adherence
Two studies reported on the effect of CCs on receipt of appropriate care for diabetes (Supplementary Material Table S16). In one study, CC patients were similarly likely to have all five recommended processes of care performed (40.1% vs. 38.9%; p = .543) as propensity score matched controls (graded as low-quality evidence; Kogut et al., 2012). Screening for diabetic complications was more likely to be carried out for participants randomized to receive nurse case management, which involved care coordination (Gabbay et al., 2013). One RCT reported that the average percentage of correct doses per month was 98.8% in the CC group who received a medication dispenser and 97.4% in the CC group who received a pill organizer (Supplementary Material Table S17; Marek et al., 2013).
Survival
A total of 13 studies reported on the effect of CC on survival (Supplementary Material Table S18). High-quality evidence from a large RCT identified that CC improved survival over a 6-year follow-up period (adjusted Hazard Ratio [HR] .73; 95% confidence interval [CI] [.55, .98]; Coburn et al., 2012). This study reported results from the Health Quality Partners program, which was one of the MCCD sites. It involved care coordination by a nurse with frequent telephone follow-up. The evidence from the five other RCTs that reported on survival was downgraded to moderate quality due to imprecision (Bauer et al., 2006; Boult et al., 2013; Eloniemi-Sulkava et al., 2001; Parsons et al., 2012; Plant et al., 2015).
Large-scale observational studies, which used propensity score matching to control confounding, did not identify differences in survival between CC and non-CC groups. For example, the implementation of the Evercare program into 62 primary care practices in the United Kingdom was found to have no effect on survival (mortality increased 5.5%; 95% CI [−3.5%, 14.5%]; Gravelle et al., 2007). The Evercare program involved development of individualized care plans followed by monitoring by a nurse CC for people aged over 65 with a previous admission to an ED (Gravelle et al., 2007). There was no difference in the likelihood of 12-month all-cause mortality between patients with a disability who received in-person education and self-management support using behavior management principles from a nurse CC and non-CC groups (odds ratio .82; p = .39; Jingping et al., 2015). In contrast, diabetic veterans who received a home telehealth care coordination intervention survived longer than a propensity score matched control group (adjusted HR .7; 95% CI [.5, .9]; Chumbler et al., 2009).
Health service costs
There were 14 studies included in this review that reported on the effect of CC on health-care costs (Supplementary Material Table S19). Most of the studies compared total health-care costs between CC and non-CC groups. A cost–utility analysis was undertaken for one program (Barnett et al., 2007). There was considerable uncertainty regarding the cost-effectiveness of the program under evaluation (CC involving home telehealth care using disease management protocols for diabetics; Barnett et al., 2007). The mean Incremental Cost-effectiveness Ratio (ICER) was more than US $60,000 and the program was considered cost-effective for only one third of the participants (Barnett et al., 2007).
A noteworthy finding is that none of the programs in the MCCD evaluation generated savings in health-care costs (calculated from Medicare claims; Peikes et al., 2009). In contrast, high-quality evidence from two other RCTs included in this review did identify that CC interventions with frequent follow-up and transition care reduced health-care costs (Coleman et al., 2006; Marek et al., 2014). The Care Transitions program, which involved in-person contact with inpatients and a 28-day postdischarge follow-up program involving care coordination, resulted in lower health-care costs at 90 and 180 days posthospitalization (Coleman et al., 2006). A care coordination intervention targeted at geriatric patients with an impaired ability to manage medications, which involved a program of biweekly in-person contact during which interventions to improve medication adherence and care coordination were provided, resulted in $296 savings per month (Marek et al., 2014).
Discussion
A key finding of this systematic review was that the individual components of nurse CC roles ranged considerably. Not surprisingly, the impact of nurse care coordination on patient-reported and health service outcomes was also inconsistent. There was an indication from higher quality studies that nurse care coordination roles did result in improved patient and health service outcomes where they involved frequent, in-person interactions, had ongoing follow-up with monitoring of disease status, and involved transition care and the application of behavior change principles (Coburn et al., 2012; Gabbay et al., 2013; Marek et al., 2013). Our finding that nurse CC roles were more likely to have positive effects on some patient-reported and health service outcomes when they were characterized by more frequent contact over time and based on evidence-based principles of behavior change is consistent with previous systematic reviews of other similar CC roles (Manderson, McMurray, Piraino, & Stolee, 2012; Sutherland & Hayter, 2009). It was reported that nurse case management resulted in positive results on health outcomes such as objective clinical measurements, quality of life, patient satisfaction, adherence to treatment, self-care, and service use (Sutherland & Hayter, 2009). Manderson, McMurray, Piraino, and Stolee (2012) reported that some evidence was identified in support of a “navigator” role that involved a variety of interventions including care plans and treatment goals. There is an important methodological difference between the previous and our present review that should be considered though. We assessed the risk of bias for each included study and incorporated the GRADE system for appraising the quality of evidence, as is recommended for systematic reviews (Moher, Liberati, Tetzlaff, & Altman, 2009). A tool to appraise the quality of the evidence was not used in the previous systematic reviews of similar CC roles (Manderson et al., 2012; Sutherland & Hayter, 2009). As such, it is unclear whether or not the conclusions drawn in the previous systematic reviews about the effects of care coordination may have been different had the quality of research design in the primary studies been assessed.
It should be noted that we limited our review to studies that included a CC role performed by a nurse. In a recently published review of CC roles (defined as navigators), which did not have professional qualification as a nurse as an inclusion criteria, it was identified that nurses were most frequently employed as the CCs (Manderson et al., 2012). Laypersons or allied health professionals, such as physiotherapists or social workers, were also commonly employed to fulfill the CC roles (Manderson et al., 2012). In that review, results of individual studies were not compared to determine whether or not the professional qualifications of the CC had an impact on the effectiveness of the model of care. Reassuringly, in a different systematic review, it was identified that successful programs assigned a nurse to be the case manager (Naylor, Aiken, Kurtzman, Olds, & Hirschman, 2011). Therefore, our focus on nurse care coordination was justified.
Our finding that nurse CC roles that included a transitional care component frequently improved a range of patient and health service outcomes is also consistent with results from studies that specifically evaluated transitional care interventions performed by nurses. A systematic review identified that transition care was consistently associated with reduced readmissions in the short term (typically 30 days) and, subsequently, a reduction in overall health-care costs (Naylor et al., 2011). Due to the nature of the research designs of studies included in our review, though, it is unclear whether, or to what extent, the transitional care component of the nurse CC intervention exerted an effect on outcomes. A randomized controlled trial that directly compares a transitional care intervention in isolation with a care coordination intervention that involves transitional care as one component of care would be required for such clarity.
A high-quality study of a nurse CC role produced positive long-term health outcomes (Coburn et al., 2012). Previous randomized controlled trials that did not involve care coordination have identified that disease self-management programs guided by behavior change principles also improved patient outcomes (Rubak, Sandbæk, Lauritzen, & Christensen, 2005). Therefore, it is possible that that the use of behavior change principles to guide intervention components focused on disease self-management may have been a mediating factor for the positive impact observed in the CC trial included in our review (Coburn et al., 2012).
Although a key feature of nurse care coordination that produced positive impacts on patient-reported and health service outcomes was targeted at specific populations deemed to be at “high risk,” the actual patient selection strategies employed across this subgroup of studies still differed considerably. As a result of this variation, we were unable to determine whether a specific “high-risk” patient selection strategy was the most effective in recruiting patients who would benefit most from care coordination. It is notable that most studies did not utilize a validated risk-stratification model for patient selection. Instead, participants were selected as “high risk” based on arbitrarily set cutoffs for age (typically 65); the presence of specific chronic conditions, such as heart failure or diabetes; and recent unplanned hospital admissions (or a combination of these). It may be more appropriate for future evaluations of nurse CC programs to devise a patient selection strategy based upon a risk stratification model developed using local data in order to target the role toward patients who are most likely to derive the greatest benefit. However, the impact of conducting patient selection using information derived from a risk stratification model on the operational requirements of participant recruitment for a nurse CC program is unknown. It is possible that extra resources are required to implement this particular patient selection strategy, which could potentially offset the (already limited) potential for overall cost savings associated with such programs. For these reasons, we recommend that further research be conducted to explore the issues and challenges associated with particular patient selection strategies.
An alternative patient selection strategy is to identify patients who will benefit from the main target of the particular nurse CC role being employed. Only one study included in our review utilized such a strategy (Marek et al., 2013). A randomized controlled trial of a nurse CC focused on promoting medication adherence appropriately enrolled only elderly patients who had confirmed difficulty managing medications, based on assessment using a validated instrument (Marek et al., 2013).
There are a number of possible explanations for the lack of/inconsistent evidence regarding the impact of nurse CC roles across a broad range of outcomes. One potential reason why some of the nurse CC roles failed to improve health service outcomes, such as hospitalizations and health-care costs, could have been their strategy used for patient selection. Some studies relied on clinician referrals for inclusion in the nurse care coordination program. There is a possibility that provider referral identifies patients who are challenging to manage, but not necessarily those at high risk for future health-care utilization and costs. In this regard, it should be noted that the general purpose of care coordination is to assist patients to effectively engage with the complex range of health services required to manage complex disease states. However, an actual measure of “care complexity” was not used to guide patient selection for any of the included studies in this review. Further research into whether measuring care complexity and using this information to inform patient selection would be a feasible and (more) effective strategy for selection of only patients with confirmed complex care needs into nurse CC programs would be worthwhile.
Methodological limitations of the available studies also need to be considered. Some studies produced only low-quality or very low-quality evidence, meaning that further research is likely to have an impact on estimates of the impact of the nurse CC role on patient and health service outcomes. For this reason, conducting further research of better quality (i.e., randomized trials adhering to Consolidated Standards of Reporting Trials (CONSORT) recommendations for allocation concealment and blinding of outcomes assessment with larger sample sizes to improve effect estimate precision) would increase confidence in our estimates of the potential clinical benefits for some of the nurse CC roles evaluated.
There was also a considerable degree of inconsistency across the included studies in terms of the outcomes assessed. As the role focused specifically on coordination of care, outcomes relating to health service use and patient experience of care may be a more appropriate target than other patient reported outcomes that are subject to a broader range of other influences that are beyond the scope of the coordinator role.
Many studies provided insufficient detail about the intervention (e.g., the expertise and qualifications of the CC). As such, it was difficult to determine how frequently specific elements of CC roles were implemented for participants across different studies. Inadequate descriptions of intervention components are a major problem because without this vital information, other researchers will not be able to replicate as well as build upon research findings and clinicians will be unable to implement the effective intervention that was evaluated. For this reason, we recommend that future research into evaluations of nurse CC roles consider utilizing existing tools to improve the reporting of intervention descriptions, such as the Template for intervention description and replication (TiDIER) checklist (Hoffmann et al., 2014).
Selection bias may have impacted the results of the individual studies included in this review. Patients who enrolled into research studies focused on improving care coordination were likely to have been more motivated to engage in related activities, while the more vulnerable patients who were less motivated to participate in research may not have been as likely to be included. Regarding the generalizability of the evidence identified in this review, it should be noted that the evidence concerning nurse CCs identified in this review was almost exclusively derived from an adult population. Only two studies included in the review enrolled children (Berry et al., 2011; Taylor et al., 2013). The outcome reported in these studies was family-reported satisfaction with the CC. Further research into the effectiveness of nurse care coordination on patient and health service outcomes is therefore required.
It should be noted that our review was focused on the effect of nurse CCs on patient-reported and health service outcomes, regardless of the terms that were used in the individual studies to define the role. The lack of standardized definitions in this field meant that several terms were used to describe the roles performed by nurses included in this review, such as nurse CC, nurse case manager, and nurse navigator. Several studies that used the term “case management” or “case manager” that we reviewed in full text were excluded because care coordination was not a component of the role.
Implications for Practice
The findings of this review confirm the potential that nurse CC roles improve patient and health service outcomes. Suggested actions to support implementation of our findings into a health service’s nurse care coordination functions are provided in Table 2. To achieve optimal outcomes for the health system, the findings suggest that CC roles should have some key features, including ongoing contact with the patient over time. However, given the variability in patient needs, the available studies do not provide specific guidance on a particular “dose” or frequency of contact. Presumably, such decisions need to be based on a careful assessment of patient need. Furthermore, the findings of this review also suggest that CCs need to be capable of implementing evidence-based principles of behavior change. Such capabilities reflect a more advanced level of nursing practice and indicate that nurse CCs need specific education in such techniques.
Supplemental Material
Supplementary_file - The Effectiveness of the Nurse Care Coordinator Role on Patient-Reported and Health Service Outcomes: A Systematic Review
Supplementary_file for The Effectiveness of the Nurse Care Coordinator Role on Patient-Reported and Health Service Outcomes: A Systematic Review by Aaron Conway, Chris O’Donnell and Patsy Yates in Evaluation & the Health Professions
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article: This review was conducted with funding received from the Office of the Chief Nursing & Midwifery Officer of Queensland.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
