Team Effectiveness in Patient Health Management: An Overview of Reviews

Background: The desire to improve the quality of health care for an aging population with multiple chronic diseases is fostering a rapid growth in inter-professional team care, supported by health professionals, governments, businesses and public institutions. However, the weight of evidence measuring the impact of team care on patient and health system outcomes has not, heretofore, been clear. To address this deficiency, we evaluated published evidence for the clinical effectiveness of team care within a chronic disease management context in a systematic overview. Methods: A search strategy was built for Medline using medical subject headings and other relevant keywords. After testing for performance, the search strategy was adapted to other databases (Cinhal, Cochrane, Embase, PsychInfo) using their specific descriptors. The searches were limited to reviews published between 1996 and 2011, in English and French languages. The results were analyzed by the number of studies favouring team intervention, based on the direction of effect and statistical significance for all reported outcomes. Results: Sixteen systematic and 7 narrative reviews were included. Diseases most frequently targeted were depression, followed by heart failure, diabetes and mental disorders. Effectiveness outcome measures most commonly used were clinical endpoints, resource utilization (e.g., emergency room visits, hospital admissions), costs, quality of life and medication adherence. Briefly, while improved clinical and resource utilization endpoints were commonly reported as positive outcomes, mixed directional results were often found among costs, medication adherence, mortality and patient satisfaction outcomes. Conclusions: We conclude that, although suggestive of some specific benefits, the overall weight of evidence for team care efficacy remains equivocal. Further studies that examine the causal interactions between multidisciplinary team care and clinical and economic outcomes of disease management are needed to more accurately assess its net program efficacy and population effectiveness.


Introduction
To address the growing burden of managing chronic diseases in aging populations, with increasing demand for services over time and across various care settings, innovative models of care are needed. Integrated disease care delivery represents a creative response to the challenges of chronic illness. It applies multidisciplinary care and use of available community-based social networks and interventional resources to provide evidence-based care for entire populations [1,2].
A primary component of such disease management models is the use of multidisciplinary teams in the delivery of care, with the hypothesis that teams will enhance integration of care and improve provider, patient and managerial satisfaction, as well as improving administrative and clinical processes and patient outcomes [3][4][5].
However, in the context of chronic disease management, there have been, as yet, only limited measurements of the efficacy of team care on patient, or health system, outcomes [9,17].
The goal of this overview was to systematically examine all available evidence in an attempt to more reliably answer the question: "Does organised team care, versus individual, or non-organised or non-integrated care, achieve improved outcomes, or care processes, among patients with chronic illness in a disease management setting?"

Definitions
Studying the effectiveness of team care within a chronic disease management context requires definitions. There is a broad range of definitions for chronic disease management in the literature [18,[19][20][21][22][23][24]. For this overview, an intervention conducted within a chronic disease management context was defined as an intervention that targeted a specific chronic disease that involved a care team in outpatient/primary care settings. Our working definition of teamwork was based on WHO's definition: "Coordinated action carried out by two or more individuals jointly, concurrently or sequentially. It implies common agreed goals, clear awareness of, and respect for, others' roles and functions" [25].
According to the Cochrane Collaboration's definition [26], a systematic review was classified as state of the art if the five following criteria were met: clearly stated objectives, predefined eligibility criteria for studies, explicit and reproducible methodology, assessment of risk of bias, systematic presentation and synthesis. Based on the criteria of Cook et al., we considered a review to be narrative if there were no explicit methods for searching literature or for reporting study results. Such reviews have a broader perspective on a given topic, sources and selec-tion of papers usually not specified, no appraisal of studies selected, synthesis often qualitative [27].

Data Sources
In collaboration with a librarian a search strategy was built for MEDLINE using medical subject headings and other relevant keywords. After testing for its performance, the search strategy was adapted to other databases using their specific requirements. Also, reference lists from all articles in our bibliography were searched for additional citations ( Table 1). Searches were limited to reviews published between 1996 and 2011, and to English and French languages.

Study Selection
A first selection was based on titles and abstracts, was performed by two authors (AG, BB) in duplicate; irrelevant articles (i.e., obviously not a review, absence of chronic disease management context, not addressing the effectiveness of team care) were discarded at this stage. Remaining article abstracts were reviewed by two authors (AG, BB) and retained if all of the following inclusion criteria were met: 1) systematic or narrative review; 2) assessment of the effectiveness of team care, provided by at least two health professionals of different speciality and corresponding to different levels of integration (collaborative, coordinated, shared care, multidisciplinary, interdisciplinary, integrated) [28,29]; 3) chronic disease management context (full disease management context if combination of chronic disease, team, and outpatient or primary care or ambulatory setting); 4) where the impact on either a process or an outcome was measured and reported. All reported outcomes were extracted. Articles that characterized the role of individual team members only were excluded. Full papers were retrieved if one or both authors considered the abstract relevant against the four inclusion criteria. The investigators independently assessed full articles and discrepancies were resolved by consensual assessment.

Data Abstraction
A data abstraction form was developed and tested on several papers to ensure that it was consistent with the study objectives. The elements of the abstraction form included the type of review (systematic, narrative); the definition or model of team care and the list of healthcare professionals involved as described by the authors; the context of chronic disease management; the type of patients covered in the review; the chronic diseases targeted; the type of process and outcomes measures, the determinants of effectiveness of team care (where applicable), the conclusions of the authors of the reviews, and the category of effectiveness based on the conclusions of the authors of the reviews as follow: teamwork effective, teamwork possibly effective, teamwork possibly not effective, teamwork not effective, inconclusive. Data abstraction was performed on all eligible papers by two authors (AG, IPB). Any discrepancies were resolved by consensual assessment, and ultimately by discussion with another member of the research team.

Systematic Reviews
We analyzed the results by reporting for each review the number of studies favouring the intervention based on the direction of effect and statistical significance for all reported outcomes. The quality of the systematic reviews was assessed by two elements: the 5 criteria of the Cochrane Handbook of systematic reviews [26] and the AMSTAR, a measurement tool to assess systematic reviews. AMSTAR is an 11-items tool and the total score was computed as the sum of all items answered "yes". The tool was found to have a good agreement, reliability and validity [30][31][32]. Data were then reported by level of quality: low (0 to 3, medium (4 to 7), and high (8 to 11) ( Table 2).

Narrative Reviews
We conducted a separate synthesis of the findings from narrative reviews for they often lack explicit methods for searching literature or reporting results. In addition, there Total score * (low 0 -3, medium 4 -7, high 8 -11) are no methods available as to how to assess the quality of narrative reviews.
Systematic reviews with a meta-analysis were of better quality than systematic reviews without a meta-analysis, on the AMSTAR criteria ( Table 2). The mean score for systematic reviews with meta-analyses was 7 (range 5 to 9, out of a maximum of 11) for a total of 173 primary articles (mean = 21.6) and 5.5 for non-meta-analysis systematic reviews (range 3 to 8) for a total of 135 primary articles (mean = 22.5). As expected, the quality of the systematic reviews that we classified as state of the art was very high (AMSTAR mean score = 7.2) compared to non-state of the art reviews (mean score = 4.3).

Systematic Reviews
Of the 16 systematic reviews reported in this study, 8 were meta-analyses. Effectiveness outcomes most commonly assessed were clinical endpoints, resource utilization (emergency room visit, hospital admission), costs, quality of life, medication adherence. While clinical and resource utilization were the most common positive outcomes, mixed results were more often found with costs, medication adherence, mortality, patient satisfaction. The overarching results are percent of the studies favouring the intervention (PSFI) and, pooled effect estimates (RR, OR, SMD) for reviews with meta-analysis. Individual results are summarized by level of quality and are reported in Table 4.

High Quality Reviews
In the review of Malone et al. (2007), reporting on mental illnesses, the percent of the studies favouring the intervention (PSFI) was 65% (11/17). The PSFI was 85%

Low Quality Reviews
In the review of Vliet Vlieland & Hazes (1997) on rheumatoid arthritis classified as inconclusive, 71% (5/7) of the studies' favoured the intervention on clinical outcomes and no differences for resource utilization outcomes. Seven narrative reviews were included in the overview. Based on the conclusion of the authors of the reviews, four concluded that teamwork was effective; one review found teamwork possibly effective; and, one was classified as inconclusive. The seventh review was disregarded because no conclusion was provided by the authors (Table 4).

Discussion
The findings of this overview do not provide unequivocal evidence that team care is universally effective in improving patient clinical care and outcomes. However, despite the limited number of reviews, the weight of cumulative results is suggestive of positive clinical and resource utilization impacts in the multidisciplinary management of several prevalent chronic diseases, namely: heart failure, depression and diabetes.
To our knowledge, no previous systematic overview has addressed the specific question of the efficacy/effectiveness of team care in chronic disease management. However, in a review of systematic reviews [55] of integrated care programs, with multidisciplinary patient care teams as a component in some of the included reviews, the investigators did report mixed results for costs and mortality outcomes; and, positive results for hospitalization, quality of life and patient satisfaction outcomes. Many unstructured reviews have also supported the benefits of health teams in terms of organisational, team members' and patients' perceived benefits [4].
Although the determinants of teamwork effectiveness were rarely assessed in the reviews analysed for this paper, the findings were in line with other previously reported results [8,15,56,57]. In particular, professional hierarchies, lack of leadership, poor communications and inter-team relations; and, incomplete knowledge of abili-ties and roles are the most frequently perceived factors to have negative impact on teams' effectiveness [10,11].
Another important aspect of teamwork effectiveness and uptake is the model of remuneration or the provision of incentives. The model of reimbursement has been described in two reviews [43,51] as one of the barriers to collaborative care. Experiences in Australia and UK with new reimbursement models have been linked to team performance [15]. In Canada, health officials are increasingly advocating the benefits of interdisciplinary models and providing incentives to health professionals [58].
Finally, no conclusions could be drawn on the relationship between the level of efficiency of teamwork or the intensity of the interventions and effectiveness of patient outcomes.
Our study has several limitations. Many review papers that have examined the effectiveness of disease management programs for various chronic diseases were excluded because they did not report separate results for a multidisciplinary team care component. Thus, effects of team care could therefore not be assessed specifically. Also, in our overview, limited numbers of reviews precluded disease-specific comparisons. Another potential limitation is that vote counting, based on direction of effect and statistical significance, can disregard sample size (number of studies) and be, therefore misleading [59]. And, we did not include subjective outcome measurements for team effectiveness, which often relate to attitudinal aspects measured by team members, such as perceptions of their own team functioning [15,56]. Lastly, the methodological challenges of conducting an overview of reviews are worth noting. Only one quality assessment tool and two methodological papers are currently available for guidance [60]. Further methods need to be developed for this new form of knowledge synthesis, particularly analytical methods that take into consideration the doubling of individual studies between the reviews, the original conclusions of the component reviews, as well as the quality of evidence.

Conclusion
Our systematic overview add to the growing body of evidence suggesting health care teams can have beneficial impact on clinical and health resource endpoints. However, because of the limitations cited, the findings do not provide unqualified evidence of the effectiveness of team care in improving clinical outcomes in the context of chronic disease management. Further studies that examine the causal relations between multidisciplinary team care and clinical and economic outcomes of disease management programs are needed to more accurately assess its efficacy in terms of public health and cost efficiency.