Comprehensive Geriatric Assessment

Darryl Wieland, PhD, MPH; Victor Hirth, MD, MHA


Cancer Control. 2003;10(6) 

In This Article

Effectiveness of Comprehensive Geriatric Assessment

As this review suggests, the past two decades have seen a proliferation of types of CGA programs, if not a standardized CGA "package." Table 3 classifies these types according to whether they are institutionally or community based (or cross-cutting), with a provision of references to recent or key descriptive or controlled clinical studies for each type. In addition, we have indicated which of these types have received relatively recent systematic review.

Interest in CGA effectiveness dates back to a landmark study of a hospital GEM unit in the early 1980s.[18] This trial attracted widespread attention because the effects of the unit on health outcomes were so marked (eg, a reduction of 1-year mortality by 50% vs the control rate), while patients' community tenure was extended and per capita healthcare costs were reduced. Many trials of programs based on CGA began to appear in the literature, reflected in the multiple service types discussed above and noted in Table 3 .

While this literature documents the effectiveness of CGA in a variety of delivery forms for a range of outcomes, including improved or better maintained functional status, survival, increased community tenure, and other outcomes, such results have not been uniform or as impressive or comprehensive as those in the Ruben-stein trial. As noted elsewhere,[61] there are many reasons for this, including variability in the selection of patients, the complexity of the assessments and interventions, the diversity of clinical objectives in the management of geriatrics patients, and the improvement in medical care over time. Thus, even the best designed and described trials lack a fair measure of external validity; it is difficult to conclude what aspects of the particular CGA program (or usual care) may be responsible for the results seen.

Systematic reviewers have attempted to clarify the effectiveness of CGA and to explore aspects of program design that may be associated with improved outcomes.[55,56,57,58,59,60,62,63] However, the complexity inherent in geriatrics interventions that makes single-site intervention trials irreproducible and multicenter studies problematic also creates problems for interpreting the results of such reviews.[56,64]

In the early 1990s, Stuck and colleagues[62] applied systematic review methods to CGA trials. The 28 trials that were available at that time were grouped into institutional and non-institutional intervention programs. Among institutional programs, hospital GEM units and inpatient geriatric consultation teams were identified as discrete types. The former were all geographic units in "control" of the treatments of their patients, at least during their stays. The latter were nongeographic, most did not have intervention "control" like the GEM units; instead, most of these services simply produced lists of treatment recommendations for the primary care physicians and did not follow patients past discharge. Noninstitutional programs were "typed" as home discharge support services (facilitating home placement of elderly patients from acute hospitals), outpatient GEM clinics, and preventive home visit services.

Among the main effects of CGA interventions in the meta-analysis were significant reductions of mortality for GEM programs to 6 months and for preventive home visit programs to 3 years. The likelihood of living in the community at the end of follow-up was significantly greater for GEM unit care, home discharge support, and home visit services. GEM unit care was also associated with significant improvement in physical function (1 year).[62] Favorable effects were also found for interventions that targeted patient selection, had clinical control over the delivery of interventions, and provided follow-up.

As seen in Table 3 , some of the more recent available reviews cut across the program "types" as given here and in fact overlap with one another in terms of trials included (if not methods of review). In their study of 12 controlled day-hospital trials, Forster et al[55] found that patients attending day hospital had significantly lower odds of functional deterioration and "poor" outcome (death or institutionalization) than controls in the subset of studies comparing day hospital with noncomprehensive, usual care services; there were trends for reduced hospital bed use and long-term institutional placement. In an overlapping review of 45 trials,[56] no particular outcome advantages were found for day-hospital models,ACE units, and related services. However, different patterns of positive findings (eg, improved survival, home discharge and tenure, functioning) were found for hospital discharge support services, GEM units, and stroke (but not gero-orthopedic) units. The more focused review of hospital discharge support[57] found evidence for increased patient/family satisfaction and trends for reduced length of index stay and subsequent readmission. In contrast, the review by Hyde and colleagues[59] of nine hospital discharge support studies found evidence of a significantly greater likelihood of remaining at home at 6 to 12 months for the supported group. A review of 16 hospital-at-home trials[58] suggested no significant health benefits. The recent meta-analysis by Stuck et al[60] of preventive home visit programs found no overall effect on survival, nursing home admission, and function. However, services providing more follow-up visits successfully prevented nursing home placement, and those high-visit programs adding multidimensional assessment were more successful at delaying functional decline.

Consistent with the findings of many of the reviews, a recent randomized controlled trial of both inpatient and outpatient GEM units performed at 11 VA medical centers found no effect on survival but significant improvements in functional status as well as mental health with no associated cost increase.[44] The lack of mortality reduction in newer trials as opposed to those from earlier decades may be due to factors discussed above as well as the improvement in medical care over time that affect both intervention and control groups.

In sum, many controlled and randomized clinical trials of CGA have been conducted since the late 1970s, and their results are often positive. Attempts to move the field forward may have been limited by the diversity of elderly patients selected, the variable organization of these interventions (not to mention that of "usual care"), the inherent complexity of geriatric evaluation and management making "standardization" premature and possibly harmful, the inconsistent measurement and reporting of multiple health and other endpoints, and the difficulty in replicating successful single-site studies. However, properly focused and interpreted meta-analytic reviews lend at least some support to the proposition that CGA can be effective and provide general support to the association with benefit of common organizational elements such as targeting, clinical control, and long-term follow-up.[61]