This systematic review of interventions to improve outcomes of patients with CHF through changing care delivery processes in clinical settings is consistent with literature reporting that interventions that attempt to improve patient outcomes through impacting the organizations in which care is delivered have mixed results [2
]. In this review, the majority (72%) of interventions were not effective in significantly improving outcomes. Our analysis of these interventions through the lens of a CAS perspective again demonstrates that interventions consistent with a CAS perspective are more likely to be effective in improving outcomes. We interpret this as providing further evidence that the clinical settings are CASs. For interventions to be effective in improving patient outcomes, they must take this into account.
The difference in the individual CAS characteristics associated with intervention effectiveness for patients with type 2 diabetes and CHF brings a new perspective to the consideration of clinical settings as CASs. Specifically, we believe that when implementing interventions to improve outcomes of patients with chronic disease, not only must the characteristics of the organization be considered, but so must the characteristics of the disease and its treatment. Interventions must be appropriately matched to the level of complexity of not only the organization, but also of the disease, as disease and treatment characteristics may influence what interventions are more likely to be effective.
The interplay of differences between chronic diseases within the context of CAS clinical systems will affect which approaches are more or less likely to be effective for patients with a specific disease. The level of uncertainty inherent in diseases and their treatments may be an important contributor to these differences. To illustrate this point, Table outlines how potential differences between type 2 diabetes and CHF in terms of uncertainty may influence the CAS characteristics that were associated with intervention effectiveness for each disease. For example, the myriad combinations of lifestyle and medication approaches to managing type 2 diabetes may be more complex and nuanced than those for CHF, and the symptoms of worsening glycemic control may be more subtle and insidious than worsening volume status, leading to greater uncertainty in the management of diabetes relative to CHF [72
]. This greater uncertainty may influence the effectiveness of CAS characteristics in interventions, or the effectiveness of combinations of characteristics. Because one way that individuals can navigate uncertainty is through relationships, interconnections may be particularly important with increasing uncertainty.
Potential differences between type 2 diabetes and CHF with regards to uncertainty, and how they might influence CAS characteristic effectiveness
However, for both CHF and type 2 diabetes, patients have a chronic disease that is changing over time, and the recognition of the dynamic nature of the evolution of disease in interventions is important. Also in both cases, the fact that the clinical settings in which patients receive care are CASs is an important contextual consideration, as no two are exactly alike.
The implication of these findings for implementation research whose goal is to change organizations to improve care of patients with chronic disease is that we must shift our focus in intervention design. While considerations such as cost, ease of implementation, and level of disruption to the clinical setting are important, the levels of complexity of the organization and the disease are even more important. Intervention design for chronic disease requires a greater level of nuance, individualization, flexibility, and assessment over time. Specific implications of this insight include the need to pay attention to or explicitly change the relationships between individuals as a strategy to improve outcomes, the importance of allowing 'local' input or control into the intervention design, and the need to provide feedback regarding the impact of the intervention and the possibility to change the intervention based on this feedback.
This study has several limitations. The first is the relatively small number of studies of organizational interventions. However, despite these small numbers, the associations found are significant ones. A more overarching limitation is the difficulty in applying the lens of a CASs perspective to traditional intervention design, with specific regard to assigning scores regarding CAS characteristics retrospectively. Our method of independent review of interventions and results using groups of separate reviewers was intended to offset this methodological limitation, and our kappa scores suggest that reviewers did have a consistent ability to make these retrospective assessments.
Other limitations include the possibility of publication bias, which may have led either to negative studies not being published, or to the interventions being described in less detail, making assessment of CAS characteristics more difficult. Negative studies are well represented in the distribution of outcomes in the included studies, and the inter-rater consistency suggests that sufficient information was available to make an assessment. All raters are from the same institution, and the possibility of bias in terms of the CAS characteristics on which we focused is possible, but made less likely by our use of characteristics consistently recognized as key in the CAS literature.
Finally, this analysis is limited to studies of patients with type 2 diabetes and CHF. The applicability of our findings to other chronic diseases, or to acute disease processes, has yet to be demonstrated. However, we believe that the strength of our results across a combined 76 examples of interventions (46 CHF and 32 diabetes) for two distinct chronic diseases is at least suggestive of the range of application of the CAS framework in clinical settings.