Limitations of the Evidence Base

Many important questions about case management have not been answered by the body of evidence that is available. For example, there is a surprising lack of evidence about the effectiveness of case management for facilitating the delivery of multidrug treatment regimens to patients with HIV infection. We found few studies of this population that used outcome measures that met the criteria defined for this review. Thus, we concluded that the evidence for this outcome is insufficient. Another important unanswered question pertains to the comparison of different delivery models for CM and role definitions for case managers. The multiplicity of roles and variability of day-to-day activities means that evaluations of CM can never fully specify the content of the intervention. Furthermore, few organizations have the potential scope (in terms of patient base and clinical resources) to conduct evaluations that directly compare different CM approaches. Thus, nearly all evaluations have compared a customized CM program with a “usual care” model in which patients receive no CM services. Synthesizing the evidence about CM requires indirect comparisons among different types of clinical programs. Because the published studies have not compared case managers with differing qualifications, there is no evidence about the efficacy of specialized training programs or case manager certification.

Despite these extensive methodological challenges, the evidence base about CM is still very useful. This review included 70 randomized trials that have been conducted in a variety of patient populations, and a smaller number of good quality observational studies also have been reported. The total number of participants in these studies approaches 100,000. The majority of these studies have given good descriptions of the patient populations, making it possible to organize the evidence by population groupings (as was done in this report). In some cases, there has been enough similarity in patient populations that indirect comparisons of different types of programs can be made with moderate confidence.

Most of the individual clinical trials of CM have had modest sample sizes (less than 500 participants per intervention arm). This size limitation has been a barrier to the analysis of patient subgroups, and many of the trials have not reported results by subgroup. Consequently, analyses of subgroup results are mostly based on indirect comparisons. In fact, the available evidence permits all conclusions about subgroup comparisons to have only a low strength of evidence. Furthermore, for some of the outcomes of interest (particularly resource utilization outcomes in several population groups), the conclusions generally had only a low strength of evidence.

The broad scope of the review and the high heterogeneity of included studies, particularly heterogeneity in the nature of the interventions and the outcomes evaluated, constrained our ability to assess applicability in great detail. The bodies of evidence for each of the Key Questions had good general applicability for the patient populations as generally defined by each disease/condition. However, because of heterogeneity in the inclusion criteria among studies of CM for particular diseases/conditions and limited descriptions of subgroups, we were not able to assess applicability for the many possible specific patient subgroups of potential interest within the disease/condition-based population groups. The unique characteristics and circumstances of so many of the diverse CM interventions and the variety of particular outcomes that they evaluated made even a general assessment of applicability related to these domains of populations, interventions, comparators, outcomes, timing, and setting (PICOTS) impractical.

Another important limitation of our review is that we examined only studies that met our definition of CM, and in most cases the CM program was not compared with other types of care management interventions. Some of the outcomes achieved by CM may have been achievable using less intensive, more focused interventions. CM typically involves nurses or other health professionals performing multiple functions to meet patients' needs. Our review did not address whether the outcomes achieved by successful CM interventions could have been achieved with more narrowly tailored interventions, targeting the specific deficits in care most likely to cause poor outcomes. However, the published research does not provide a model for how such targeted interventions would be designed. One approach that has been widely deployed is disease management programs, which generally use telephone-based interactions with patients to address specific treatments and self-care measures for individual chronic diseases. While assessment of disease management programs is beyond the scope of this review, recent evaluations suggest that these narrow disease-focused interventions are often ineffective.¹³ Overall, we were able to draw conclusions only with a low strength of evidence for the relationships between characteristics of the CM intervention and any clinical outcomes (Key Question 3).

Case managers vary in their experience and training, and there is a very limited evidence base about the expertise of case managers in any of the CM programs that have been studied.

While most studies included registered nurses as the case managers, there are no good quality studies that have compared the outcomes achieved by registered nurses with case managers from other disciplines. Many of the programs that have been studied provided CM via a team (such as a nurse and a social worker), and the distinctive roles of the team members were not well described. Because of the lack of studies providing comparisons of differing skill sets, it is not possible to answer important questions about the necessary qualifications and training of case managers.