NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Hickam DH, Weiss JW, Guise JM, et al. Outpatient Case Management for Adults With Medical Illness and Complex Care Needs [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2013 Jan. (Comparative Effectiveness Reviews, No. 99.)

Cover of Outpatient Case Management for Adults With Medical Illness and Complex Care Needs

Outpatient Case Management for Adults With Medical Illness and Complex Care Needs [Internet].

Show details

Results

Search Results

A summary of the search results is presented in Figure 3. We obtained 5,645 citations from the initial and updated database searches, by reviewing reference lists of published studies and systematic reviews, and by reviewing registries of ongoing studies.

This figure is a flow chart that summarizes the search and selection of articles related to Outpatient Case Management for Adults with Medical Illness and Complex Care Needs. Citations were identified through bibliographical databases (MEDLINE, CINAHL, CCRCT, CDSR, DARE) and other sources (experts, reference lists, ongoing studies registries). There were 5,645 abstracts of potentially relevant articles reviewed in the original search. Excluded abstracts numbered 4,444, and 1,201 full-text articles were reviewed for relevance. Articles excluded on full-text review numbered 1,051. Studies were excluded at the full-text level for the following reasons: not case management, 349; outcomes not included in this review, 143; no original data, 345; wrong population (e.g., mental illness only or not adults), 90; wrong setting, 86; background, 31; unable to locate, 7. Of the 1,202 articles reviewed at the full-text level, a total of 153 articles (109 studies) met inclusion criteria. The number of articles and studies included for various conditions were: 18 studies reported in 28 articles on chronic disease in older adults; 14 studies reported in 17 articles on frail elderly; 15 studies reported in 26 articles on dementia; 12 studies reported in 12 articles on congestive heart failure; 12 studies reported in 24 articles on diabetes mellitus; 6 studies reported in 8 articles on cancer; 15 studies reported in 17 articles on serious chronic infections; and 14 studies reported in 18 articles on other medical problems.

Figure 3

Study flow diagram.

After a review of the citation titles and abstracts, 1,201 were selected as possibly relevant by at least one of the two reviewers. Full articles were retrieved for all of these. After review of the full articles, a total of 152 articles were selected as relevant by two reviewers and included in the review. One additional study was identified through a registry of VA Cooperative Studies but was published after the date of the update search.28 This study was added, bringing the total to 153 articles. Due to multiple publications for some studies, this represented 109 total studies of case management (CM). After examination of the gray literature search results, no additional studies were included. Appendix D contains a list of included and excluded articles.

Overall Effectiveness of Case Management

Our review examined studies of CM that provide longitudinal services to patients, generally for a duration of at least 6 months and often extending for longer than 1 year. The individual studies were diverse but generally fell into two categories. The first category is evaluations of innovative programs targeted at specific patient groups. These studies often featured a close relationship between program developers and the evaluation teams. While some such studies included small sample sizes and short durations of follow-up, others included hundreds of participants and sometimes lengthy follow-up. 29-31 The second major category was a group of formal demonstration projects, most commonly funded by the U.S. Centers for Medicare & Medicaid Services. These studies tended to be large, including geographically diverse clinical sites, and they commonly had evaluations performed by research teams that had no history of working with the clinical programs.32-37 While the first category of studies examined novel programs that may be uniquely suited to their patient populations, the second category examined clinical programs that were likely to be typical of how such programs are implemented and disseminated in community settings. In general, these two types of studies are complementary, and we attempted to account for program diversity in estimating the strength of evidence for the Key Questions.

The overarching finding of our review is that, when CM was deployed in a variety of community settings, its impacts tended to be limited to narrowly specified outcomes such as patterns of the care received and certain measures of the status of the underlying disease. However, it had minimal impact on more general patient-centered outcomes, quality of care, and resource utilization among patients with chronic medical illness. On balance, CM interventions tested in randomized trials were more often unsuccessful than successful in improving prespecified outcomes. The most notable example of the limited impact of CM was the Medicare Coordinated Care Demonstration project (MCCD), in which over 18,000 patients, predominantly elderly persons with multiple chronic illnesses, were enrolled in a prospective randomized trial conducted in 15 separate CM programs across the United States.32 Twelve of these 15 programs met our criteria for inclusion in this report. In assessing multiple outcomes—including health outcomes, quality of care, hospitalizations, and overall expenditures—there were only sporadic and isolated successes. Only three of the programs, one of which was small and could not be sustained, showed potential return on investment.

Although this summative conclusion of minimal impact reflects the balance of findings from our review, it was not a consistent finding across all studies. Some studies enrolled general populations with chronic illness, while others targeted patients with clinical or sociodemographic characteristics that put them at risk for inadequate care, poor outcomes, or high resource utilization (e.g., patients with high utilization of services or with limited social support). Other studies tested CM for the management of specific clinical conditions (e.g., diabetes, dementia). The goals of CM varied across different clinical conditions, patient populations, and settings. For instance, CM intended to delay nursing home placement for community-dwelling patients with dementia was very different—in content, implementation, and intensity—from CM intended to improve physiologic and metabolic measures (such as glycemic control) among outpatients with diabetes. We therefore synthesized data for specific patient groups (typically defined by clinical condition), in which the goals of CM interventions were relatively similar. We then sought common themes that cut across groups. In this section we present the findings of our crosscutting synthesis.

Key Question 1a. In adults with chronic medical illness and complex care needs, is case management effective in improving patient-centered outcomes?

Mortality

While reducing mortality was rarely the principal outcome examined in the studies, it was often measured and reported. Patients who were provided CM did not experience lower mortality in general populations of patients with chronic illness, in the frail elderly, those with AIDS, or in patients with congestive heart failure.

Quality of Life and Functional Status

CM interventions produced mixed results in terms of improving patient QOL and functional status. In general, CM was sometimes successful in improving aspects of functioning and QOL that were directly targeted by the interventions. For instance, CM was successful in improving caregiver stress among persons caring for patients with dementia and congestive heart failure (CHF)-related QOL among patients with CHF. The measures used to evaluate QOL and functional status varied across studies, and overall, the improvements in QOL and functional status achieved by CM were either small or of unclear clinical significance. CM was less successful in improving overall QOL and functioning, as indicated by global measures not specific to a particular condition.

Patient Satisfaction With Care

CM interventions were generally associated with improved patient (and caregiver) satisfaction, although satisfaction with CM varied across interventions. Studies measuring patient satisfaction typically reported overall satisfaction with care, rather than satisfaction in specific domains. Some studies found that CM improves patient perceptions of coordination among health care providers.

Ability to Remain at Home

One measure of the clinical significance of improvements in functioning for elderly patients with chronic conditions is the ability to remain at home and avoid nursing home placement. This outcome was often the primary objective of CM programs for patients with dementia. In most studies of the frail elderly and of patients with dementia, CM was not effective in maintaining patients' ability to live at home. Evidence from one study suggests that a high-intensity CM intervention sustained over a period of several years can produce a substantial delay in nursing home placement for patients with dementia.

Disease-Specific Health Outcomes

The effect of CM on disease-specific outcomes was inconsistent. In some studies, CM had a positive impact on specific symptoms, including pain and fatigue in patients with cancer and depressive symptoms among caregivers of patients with dementia. Some studies also found that CM had a positive impact on glycohemoglobin levels for adults with diabetes. However, CM has not been found to have a significant benefit for improving lipid levels or body weight in this population.

Patient Satisfaction With Care

CM interventions were generally associated with improved patient (and caregiver) satisfaction, although satisfaction with CM varied across interventions. Studies measuring patient satisfaction typically reported overall satisfaction with care, rather than satisfaction in specific domains. Satisfaction was most substantially improved in the domain of coordination among health care providers.

Key Question 1b. In adults with chronic medical illness and complex care needs, is case management effective in improving quality of care?

Disease-Specific Process Measures and Receipt of Recommended Services

CM was effective in increasing the receipt of recommended health care services when it was an explicit objective of the CM intervention. For instance, CM interventions designed to improve cancer therapy for patients with breast and lung cancer were successful in increasing the receipt of radiation treatment, as recommended in clinical guidelines. In a study of low-income adults who already were enrolled in primary care, CM was found to improve measures of cardiac risk. The effect of CM on guideline-recommended care in general, however, was less consistent. Studies showed only sporadic effects on elements of quality of care, such as receipt of appropriate medications for patients with CHF or diabetes, or receipt of appropriate preventive services for elderly patients.

Patient Self-Management

CM was effective in improving patients' self-management behaviors, including dietary and medication adherence, for specific conditions such as CHF or tuberculosis, when patient education and self-management support were included within CM interventions.

Missed Appointments

Few studies measured the frequency of missed appointments as an outcome of CM interventions.

Key Question 1c. In adults with chronic medical illness and complex care needs, is case management effective in improving resource utilization?

Hospitalization Rates

Although hospitalization rates were often included as an outcome, trials of CM generally did not demonstrate reductions in these rates.

Emergency Department Use

CM had a variable effect on ED use. Several studies found reduced ED use in patients receiving CM, but other studies found no effect.

Clinic Visits

Few studies measured the frequency of clinic visits as an outcome of CM interventions. Those that did found varying results. CM sometimes was associated with increased rates of physician visits and sometimes with decreased rates.

Overall Expenditures

Most studies examining the impact of CM on the overall cost of care showed no significant difference between groups of patients receiving CM and control groups. Although the cost of CM programs often was modest relative to overall costs among patients with high utilization, the effect of CM on reducing utilization is minimal.

Key Question 2. Does the effectiveness of case management differ according to patient characteristics?

Medical Conditions

Individual studies had inconsistent findings on whether CM interventions are more successful for patients with high disease burden. While it is possible that there is a mid-range of disease burden for which CM is most effective, the evidence base does not permit defining how to identify such patients.

Age

Most studies of CM included mainly elderly patients, making it difficult to determine impact of age on CM effectiveness.

Socioeconomic Status

Studies did not routinely report the effect of CM according to socioeconomic indicators among enrolled patients. Some studies explicitly targeted low-income populations. There was no apparent pattern to suggest an influence of patient socioeconomic status on the effectiveness of CM.

Social Support

Few studies explicitly evaluated patients' level of social support. However, studies that targeted patients with limited social support did not tend to find better results.

Formally Assessed Health Risk

Some studies explicitly targeted patients considered to be at high risk of poor outcomes. The methods used to evaluate risk, however, varied substantially across studies. The studies have not defined a specific level of risk for which CM is most effective for improving outcomes.

Key Question 3. Does the effectiveness of case management differ according to intervention characteristics?

Setting

Characteristics of the setting in which CM was implemented (e.g., integrated health system, home health agency, outpatient clinic) did not clearly influence the effectiveness of CM.

Case Manager Experience, Training, Skills

Studies did not consistently provide details about the experience, training, or skills of case managers. In most studies the case managers were registered nurses, and some had specialized training in caring for patients with the conditions targeted by the CM intervention (e.g., diabetes, cancer). There was some evidence that pre-intervention training of nurses in providing CM for the targeted conditions, the use of protocols or scripts to guide clinical management, and collaboration between a case manager and a physician (or multidisciplinary team) specializing in the targeted clinical condition, resulted in more successful interventions.

Case Management Intensity, Duration, Integration With Other Care Providers

Studies across multiple patient groups suggested that more intense CM interventions, as indicated by greater contact time, longer duration, and face-to-face (as opposed to only telephone) visits, produced better outcomes, including functional outcomes and lower hospitalization rates. The most successful interventions generally had more contacts between case managers and patients and were more integrated with the hospitals and physicians where patients received care.

Case manager Functions

Case managers typically performed multiple functions. These included but were not limited to assessment and planning, patient education, care coordination, and clinical monitoring. In general, emphasis on specific functions varied according to patients' conditions and the primary objectives of specific CM interventions. For example, interventions among patients with cancer typically focused on coordination and navigation, while interventions for patients with diabetes and CHF focused more on patient education (for self-management) and clinical monitoring. Most studies did not carefully measure the amount of effort case managers devoted to different functions, making it difficult to discern the degree to which emphasis on different case manager functions impacted CM effectiveness.

Effectiveness of Case Management in Defined Patient Populations

Population: Older Adults With One or More Chronic Diseases

Contemporary models of CM use clinical approaches that are applicable to a variety of diseases and conditions. Clinical programs that meet the needs of a broad patient population potentially are more sustainable, and the largest clinical trials of CM have been studies of programs that take a generalist approach. The primary goal of many of these studies has been to determine whether CM can reduce health care expenditures by preventing acute hospitalizations and reducing use of other expensive services. At the same time, CM programs for the elderly frequently have been dominated by approaches that attempt to define subpopulations at particular risk. The basic premise is that a healthy, highly functional older adult is less likely to need CM than one of the same age who has a greater burden of illness. Selection of older adults for inclusion in CM, therefore, has taken a wide variety of approaches. These include purely administrative assessments such as previous utilization, especially hospitalization, certain chronic illnesses, or prior costs of care. Evaluations of such CM programs are included in this section. Subsequent sections of this report will review the evidence about programs that select participants on the basis of either targeted assessments of patient-reported functional and health status (the frail elderly) or on the basis of specific clinical diagnoses such as dementia or congestive heart failure. However, it is important to note that the studies of general populations of older adults with various chronic illnesses (analyzed in this section) include populations of patients that have characteristics and medical problems that are very similar to the populations included in the studies of particular diseases (such as congestive heart failure, diabetes, or cancer) that are described in the subsequent sections of this report.

Description of Studies

We identified seven randomized trials of CM programs that delivered services to broad populations of older adults (see Appendix I, Evidence Table 1). Four were rated good quality,29, 32-34, 38 and three were rated fair35, 39-41 (see Appendix G). Six trials were conducted in the United States29, 32, 34, 35, 38, 39 and one in the Netherlands.40, 41 Two trials were published between 1994 and 1997.34, 39 Five trials were published between 2003 and 2011.29, 32, 35, 38, 40, 41 In addition to the seven randomized trials, we identified four studies of CM for community-dwelling Medicare populations that used nonexperimental designs, one good quality,42 and three fair quality studies.43-45 These four studies examined groups of patients who received CM services in existing programs and used matching techniques to construct comparison groups. We also identified nine other observational studies that used either historical controls, a nonequivalent comparison group, or did not have a comparison group;46-55 all but two46, 49 of these studies were poor quality (see Appendix I, Evidence Table 2, and Appendix G).

Other closely related clinical approaches have been developed for older adults with chronic diseases. These were not included in this review, based on our definition of CM. First, team-based geriatric practices, including the Program for All-Inclusive Care of the Elderly (PACE) and the Home-based Primary Care (HBPC) program of the Department of Veterans Affairs, were excluded because they tended to have provider-led interventions, and the role of the case manager was less clear in most of them.56, 57 Rather, these approaches tended to involve team-based discussion and coordination that was either the source of primary care or essentially replaced primary care. Similarly, the Geriatric Resources for Assessment and Care of Elders project (GRACE)58 also was excluded. This model used home-based care by a team consisting of a nurse practitioner and social worker to provide guidance and assistance to older adults.

The largest randomized trial of CM was conducted between 2002 and 2005 in multiple sites in the United States.32, 33, 59, 60 Known as the Medicare Coordinated Care Demonstration (MCCD), the study was funded by the Centers for Medicare & Medicaid Services (CMS). The 15 clinical sites had submitted proposals to CMS to participate in the project. The evaluation was managed by a separate organization that collected all outcome data. Participants were enrolled and randomized through 2005. Because each clinical program was managed separately, this study was in fact a set of single-site clinical trials, each using identical methodology. The study reports listed outcome data separately for each site. Of the 15 sites, one was a hospice program, one was conducted in a long-term care facility, and one did not provide care coordination. Because these did not meet our definitions for study setting or intervention characteristics, we dropped these three sites from our analyses, leaving 12 sites used for this report. The total sample size for these 12 sites was 16,301. There was a significant variation in size across these 12 sites, ranging from 211 to 2,657 participants per site. For all seven of the clinical trials in this category, the total number of participants is 31,935.

The populations in five of the six U.S. trials were Medicare beneficiaries living independently who were judged to be at high risk of medical complications and the attendant utilization of health care services.29, 32, 34, 35, 38 Eligibility criteria for all but one of the programs included in the MCCD trial32 included one or more targeted chronic conditions; seven of the 12 programs also required a recent hospitalization—either within a year prior to enrollment (six programs) or within the prior 60 days (one program). The average monthly Medicare expenditures at baseline for the study sample overall was nearly three times that of beneficiaries nationwide; baseline expenditures for study participants in six programs averaged more than $2000 per month, but less than $600 per month in three programs. In the study reported by Boult et al.29 participants were identified as being at high risk of heavy health services use during the upcoming year by using a claims-based predictive model. Study participants (n=904) had four chronic diseases on average, over 40 percent rated their health as fair/poor, and 25-30 percent had diminished functional status by activities of daily living (ADL) or instrumental activities of daily living (IADL) measures. The study reported by Newcomer et al.35 had enrollment criteria of either being age 80 years or older or being 65 or older with at least one qualifying chronic condition; over 70 percent of the sample population (n=3079) was 80 years or older, which is a notable difference compared with the percentage of this age group in the other study samples. The study reported by Martin et al.38 also had a notable difference in the study sample; enrollment was open to all members of a health maintenance organization (HMO) who resided within the study catchment area and were at least 65 years of age (n= 8504). During the study period, a total of 1,640 participants in the intervention group (38.5 percent) were evaluated for CM based on an electronic algorithm or a low score on a general health measurement. The study reported by Schore et al.34 enrolled patients with one of a set of qualifying diagnoses who had been hospitalized over the prior year. The most frequent diagnoses were congestive heart failure and chronic obstructive pulmonary disease (COPD). The study reported by Fitzgerald et al.39 enrolled patients being discharged from an acute hospitalization at a VA medical center. The mean age of participants was 64 years, and comorbidities included COPD, heart disease, diabetes, and heart disease. The study conducted in the Netherlands (n=208)40 enrolled participants being discharged from an acute hospitalization with a case complexity score indicating the need for case management. The mean age of participants was 64 years and comorbidities included circulatory, respiratory, and gastrointestinal disorders.

CM interventions in these studies focused on patient self-management education, health status monitoring, and coordination of health care (see Table 2). Case managers in all of the studies were nurses. Across studies, the vast majority of contacts with patients were via the telephone. In-person contacts generally were reserved for initial assessments, although in four programs included in the MCCD trial,32 participants were contacted in person nearly once a month. The length of CM intervention was 6 months in one trial,40 12 months in one,35 and 20 months in one.29 In the MCCD trial,32 programs varied widely in participant's average length of exposure to a CM intervention, with a range of 18 to 38 months. One study38 did not report exposure time for the participants who received CM during the study period. In one study, CM was managed via teams having caseloads of 800-1000 study participants on each of four teams. A small fraction of the cases (50-70 participants per team) received more intense CM.38

Table 2. Characteristics of case management interventions for older adults with one or more chronic diseases (randomized trials).

Table 2

Characteristics of case management interventions for older adults with one or more chronic diseases (randomized trials).

The comparator in these trials was “usual care,” meaning the standard services provided in each study setting but without the CM intervention.

All of these trials examined both utilization and health status outcomes (see Table 3). Patient-centered outcomes included mortality, measures of mental and physical health, QOL, and patient satisfaction. Quality of care outcomes also were examined in two trials32, 62 and included measures of self-management support, service arrangement, and general and disease-specific preventative care. Resource utilization measures included hospitalizations, skilled nursing facility admissions, ED utilization, outpatient visits, home care, and overall costs.

Table 3. Characteristics and outcomes of studies of case management for older adults with one or more chronic diseases (randomized trials).

Table 3

Characteristics and outcomes of studies of case management for older adults with one or more chronic diseases (randomized trials).

The timing of the CM interventions in two of the studies29, 32 was similar in that participants were identified as already being high utilizers of health care services. CM was initiated to improve patient health and reduce the need for ED, hospitalization, and acute care services. In two studies, CM was initiated upon hospital discharge after an acute event;39, 40 in one study,35 CM was initiated proactively among a population with increased risk of high service utilization due to advanced age or chronic conditions; and in one study,38 the CM intervention was offered to a subset of disease management program participants at a point when their health care needs were deemed to have become complex.

The settings of the CM programs varied. The MCCD trial32 included three hospital-based programs, five commercial disease management or care coordination programs, two programs operated in academic medical center, a program in an integrated health care system, and a program in a retirement community. The majority of these programs serviced large metropolitan areas but four serviced rural areas. The study reported by Schore examined three CM programs.34 One program was integrated with the family medicine and geriatrics departments of a teaching hospital, and the other two were based in free-standing community organizations. In the other four studies conducted in the United States,29, 35, 38, 39 the CM programs were health plan based or health system based, while the study conducted in the Netherlands was hospital based.40

Key Points Related to Older Adults With One or More Chronic Diseases

  • CM programs that serve patients with multiple chronic diseases do not reduce overall mortality (strength of evidence: high). (See Appendix H. Strength of Evidence.)
  • CM programs that serve patients with one or more chronic diseases do not result in clinically important improvements in functional status (strength of evidence: high).
  • CM programs that serve patients with one or more chronic diseases increase patients' perceptions that their care is better coordinated and of higher quality (strength of evidence: high).
  • CM programs that serve patients with one or more chronic diseases do not reduce Medicare expenditures (strength of evidence: high).
  • CM programs that serve patients with one or more chronic diseases do not reduce overall rates of hospitalization (strength of evidence: moderate).
  • CM is more effective for reducing hospitalization rates among patients with greater disease burden (strength of evidence: low).
  • CM is more effective for preventing hospitalizations when case managers have greater personal contact with patients and physicians (strength of evidence: low).

Detailed Analysis: Effectiveness of Case Management by Outcome

Patient-Centered Outcomes
Mortality

Five clinical trials and four observational studies examined mortality among patients who received CM. In the MCCD trial,32 3-year mortality rates ranged from 10 to 40 percent for the 11 programs for which mortality was reported. Mortality rates in the groups receiving CM were slightly lower in six of these programs and higher in the other five.

Overall mortality rates in three other trials were low. In Martin's trial,38 19-month mortality was 4 percent in the CM group and 5 percent in the control group. Newcomer et al.35 reported 12-month mortality of 3 percent in both the CM and control groups. Fitzgerald et al.39 reported 12-month mortality of 10.5 percent for each group. In the trial conducted in the Netherlands, 6-month mortality rates were similar (7-8 percent) in both the CM and control groups.40

One observational study reported a mortality benefit with CM.42 In this study of U.S. Medicare beneficiaries the CM group included patients who were referred to and completed intake into a CM program linked to primary care clinics. The comparison group included patients followed in similar clinics that did not have CM programs. Patients in the comparison group were selected by matching for age and diagnosis. Two-year mortality rates were 13 percent in the CM group and 17 percent in the control group. This difference was marginally significant (p=0.07).

Another U.S. observational study examined mortality over 5 years of followup and found no effect of CM on this outcome.43 Two family medicine clinics were compared, with only one offering a CM program. Study participants in both clinics were individuals who had three or more clinic visits in the prior year. Average age of the participants was 76 years, and the CM was provided by a nurse practitioner based in the experimental clinic. Five-year mortality was 27 percent in both groups. Two European observational studies found similar mortality rates between CM groups and comparison groups of similar age.44, 52

Because of the minimal changes in mortality rates across multiple clinical settings, we concluded that CM programs that serve broad populations of patients with chronic diseases do not affect mortality rates. This has a high strength of evidence due to the large cumulative sample size of these studies (including the MCCD trial).

Functional Status

Evidence about functional outcomes was reported in three clinical trials. The MCCD trial32 conducted surveys of random samples of participants 10 months after entry into the study. One site did not participate in the survey due to dropping out of the study and another site did not participate because of program focus (it enrolled only patients receiving active cancer treatment). For the remaining 10 sites, response rates were reported to be about 95 percent. Sample sizes were at least 350 participants in each of the CM and control groups for each site. The MCCD collected self-reports for ADLs and IADLs. In none of the programs was there consistent improvement in ADLs or IADLs with CM. Martin38 also used a survey measure to assess patient functioning at 18 months. The only significant change was a slightly lower rate of deterioration of social functioning in the CM group. Newcomer35 also found no difference between CM and control groups in measures of physical and mental functioning at 12 months.

Psychological Measures

The MCCD trial32 also examined psychological measures in the 10-month participant survey. Three of the 10 programs found significantly better scores on a measure of stress in the CM groups. However, CM was not associated with better scores on a depression screen in any of the programs. Another trial examined measures of caregiver depression and burden but found no difference in these measures between the CM and control groups. This finding did not change when evaluating subgroups of caregivers who had higher and lower levels of time commitments to caregiving.61

Patient Satisfaction

Two trials assessed participant satisfaction. The MCCD survey included four items on satisfaction with explanations received from providers. There were no consistent trends in these measures for any of the ten programs when comparing the CM and control groups.32 The study reported by Schore also found minimal effects on patient satisfaction in three CM programs that served a Medicare population.34

Quality of Care Outcomes

The MCCD survey included two types of quality measures: perception of care coordination and self-care behaviors. In eight of the ten programs in the MCCD trial, participants in the CM group gave higher ratings of the impression that clinicians kept in touch with each other, and this difference was statistically significant in six.32 The MCCD survey also included several measures of health behavior associated with chronic illness care. No more than one program showed an effect of CM on each of four measures of diet and exercise. None of the programs showed differences between the CM and control groups for self-reported medication adherence. In addition, none of the programs showed an effect of CM on a question about planning for physician visits. Similar findings were found in an earlier trial, with no effect of CM on medication adherence or on self-monitoring of blood pressure.33 In another trial, both patients and caregivers were asked to rate care coordination. Both patients and caregivers in the CM group gave significantly higher ratings.61, 62 None of the other trials included measures of care coordination or self-care.

Using Medicare claims data the MCCD trial also measured receipt of preventive services. No consistent effects of CM on vaccination rates or rates of colon cancer screening were found.32 Two of 11 programs had higher mammography rates in the CM group.32 for patients with diabetes, effects of CM on quality measures were mixed. One of 11 programs had higher rates of eye examinations and microalbumin measurements with CM. Two other programs had higher rates of glycosylated hemoglobin testing with CM. In three out of 11 programs, CM was associated with higher rates of lipid testing among patients with diabetes and/or coronary disease.32 Another trial conducted in a Medicare population found no effect of CM on rates of influenza vaccination or smoking cessation.33

An observational study having a pre-post design examined changes in physiological measures with 3 months of CM.49 Blood pressure, glucose, and cholesterol levels decreased moderately, compared with the pre-CM values. However, there was no non-CM comparison group in this study.

Resource Utilization Outcomes

All seven of the randomized trials and eight observational studies included utilization outcomes. The most common utilization measure was hospitalization rates. In the MCCD study, one of the 12 programs found a significantly lower hospitalization rate in the CM group.32 This program had a per capita yearly hospitalization rate of 0.98 in the control group and 0.82 in the CM group. A second program that had a high hospitalization rate in the control group (per capita rate of 2.1) had a marginally significant (p=0.07) reduction in the rate to 1.6 with CM.

There were similar findings in the other clinical trials. Newcomer35 found no difference between the CM and control groups in the following measures: overall hospital admissions, readmissions, or nursing home admissions. Boult29 also found no significant difference between CM and control groups in the rates of hospital admissions, 30-day readmissions, and nursing home admissions. Boult did find an interaction between insurance coverage and CM effects on nursing home admissions. There was a greater reduction of nursing home admissions with CM for members of a staff-model health maintenance organization than for patients covered by fee for service plans.29 Hospital admissions and total inpatient days also were not different between CM and control groups in the trials reported by Martin,38 Fitzgerald,39 and Schore.33 Martin's study found that nursing home admission rates were low in both groups (less than 4 percent per year), but total nursing home days was modestly lower in the CM group.38 In the Netherlands trial, hospital admission rates were similar over 6 months in the CM and control groups.40

The observational studies had differing findings on hospitalizations. In a good quality study, 2-year hospitalization rates were not significantly reduced (32 percent in CM group; 35 percent in control group).42 Three other studies43-45 also found no difference between CM and control groups in hospitalization rates or total inpatient days. A poor quality Australian observational study compared acute hospitalization rates for patients currently receiving CM with rates during the 12 preceding months. The rates were 28 percent lower, while the rates did not change in a comparison group.46 Two poor quality observational studies reported a significant reduction in hospital admissions with CM over a 6-month period48, 50

Three clinical trials29, 35, 38 and two observational studies 45, 52 examined nursing home utilization in this population. Overall, the findings were inconsistent. A good-quality clinical trial38 and the two observational studies found that CM was associated with lower rates of nursing home utilization. However, the utilization was very low in the clinical trial (fewer than one nursing home day per person per year). The fair quality observational study found average one-year nursing home use to be 8.4 days in a CM group and 12.6 days in a comparison group.45 A poor quality European observational study found 1-year rates of nursing home placement to be 7 percent in the CM group and 13 percent in the comparison group.52 Another good-quality clinical trial 29 found no significant effect of CM on nursing home admission rates, although a patient subgroup enrolled in a health maintenance organization had lower nursing home use with CM. Finally, a fair quality trial35 found the nursing home placement rate to be significantly higher among patients who received CM. Because of the inconsistency of these findings, we concluded that there is insufficient evidence to draw a conclusion about the impact of CM on nursing home use among elderly patients with one or more chronic diseases.

Two trials and three observational studies examined ED visits. Both the Boult and Newcomer trials found no difference in ED visits between the CM and control groups.29, 35 However, an observational study found significantly lower rates of visits to both EDs and urgent care clinics in the CM group,43 and another observational study found a 54 percent reduction in ED visits in a CM group.53 An Australian observational study also found lower ED visit rates in a CM group, compared with the 12-month period prior to enrolling in CM.46

In this population, there are not consistent findings on the effect of CM on the utilization of a variety of outpatient services. One trial39 and one observational study45 reported modest increases in primary care visits for the CM group. Another trial29 and an observational study43 both found no effects of CM on rates of primary care or specialty clinic visits in the United States. The U.S. trial also found that the CM group had significantly lower use of home health services.29 An observational study of European programs found no difference in utilization of home nursing, caregiver services, physical therapy, and occupational therapy between a CM group and a comparison group.52

In this population, CM had minimal effects on the overall costs of care. In the MCCD trial, none of the 12 programs had significantly lower overall Medicare expenditures in the CM group.32 Total costs also were not significantly different between CM and control groups in another U.S. trial.38 An additional U.S. trial measured only the costs of inpatient hospitalizations. It found no difference between the CM and control groups.35 A fair quality observational study in the United States found higher overall costs in a group receiving CM compared with a similar group that did not receive CM.45 A regression analysis that controlled for costs in the previous year estimated a cost savings. However, the comparability of the control group was not well described in this study.

Effectiveness of Case Management by Patient Characteristics

Although the studies of CM in this population group included large numbers of participants, there were few analyses of patient subgroups. One of the programs included in the MCCD study conducted a risk stratification of its participants at the time of enrollment. For the 30 percent of participants having the highest severity, hospitalization rates were 29 percent lower with CM, and total expenditures were 20 percent lower. This higher risk group was defined as patients having average Medicare monthly expenditures of between $900 and $1200 per month.32

One observational study compared mortality and hospitalization rates among people with diabetes with the entire population of participants.42 In the subgroup of patients with diabetes, 2-year mortality rates were similar to those in the entire patient sample. However, those who received CM had significantly lower mortality (18 vs. 13 percent at 2 years).

The hospitalization rate also was significantly lower with CM (30 percent in CM group; 39 percent in control group).42 No other studies have examined subgroups of people with diabetes for these outcomes.

Effectiveness of Case Management by Intervention Characteristics

In the MCCD study, the two CM programs that had the greatest reductions in reducing hospitalization rates were compared with the remaining programs by a variety of programmatic characteristics. Several differences were found. First, the two successful programs averaged one in-person contact between the patient and case manager per month, compared with a median of 0.3 such contacts in the other programs. Second, participants in these two successful programs were more likely to report that they had received instructions on how to take their medications. he successful programs also tended to be closely linked to providers. The case managers frequently traveled to primary care sites for direct communication with physicians and also had close contacts with hospitals to provide followup of patients after acute hospitalizations. Another feature of the successful CM programs was the continuity of the relationship between the case manager and medical providers, defined as a single case manager assigned to each physician's patients.32

Indirect comparisons can be made between the MCCD study and other trials by intervention characteristics. The large trial reported by Martin38 featured high caseloads by the case managers and consequently little face-to-face patient contact. This trial showed few benefits of CM. The Newcomer trial35 also had relatively high caseloads (about 250 per case manager), and this study found minimal benefits of CM. These findings suggest that CM effectiveness may be related to face-to-face time with patients. However, Schore et al.34 found that a case management program that had more face to face contact with clients resulted in no difference in outcomes when compared with two other programs that used primarily telephone contact with clients (with similar case manager caseloads across the three programs).

Population: The Frail Elderly

As people with multiple chronic illnesses age, the cumulative result is a declining ability to live independently. CM programs potentially can help the frail elderly to avoid or reduce functional loss, improve QOL, and maintain independence. For people who are frail, these programs also have the potential to forestall hospitalizations, ED visits, and skilled nursing facility use. The reduction of utilization of these services potentially can be accomplished through coordinating care for complex illnesses, preventing adverse events (such as urinary tract infections, pressure ulcers, falls, and the like), and preventing disease exacerbations. The approach to CM is often broad and holistic so as to meet the needs of individual patients, rather than an emphasis on single disease indicators.

The CM programs included in the frail elderly category share many characteristics with programs that targeted the population of patients with multiple chronic diseases. Case managers in the frail elderly programs also needed to coordinate care for multiple chronic diseases. We used two criteria to differentiate between the two types of programs. These included:

  • Patients in the frail elderly programs tended to have a higher prevalence of functional deficits.
  • The frail elderly programs placed more emphasis on maintaining functioning and delaying nursing home placement, while the chronic disease programs tended to emphasize care of specific diseases.

The two types of programs nearly always had mean patient ages greater than 70 years, but more of the frail elderly programs had mean ages greater than 80 years.

Description of Studies

We found eight randomized trials of CM programs for the frail elderly (see Appendix I, Evidence Table 3). Four were rated good quality,30, 64-66 three were rated fair,67-69 and one was rated poor70 (see Appendix G). The trials were conducted in the United States,30, 66, 68 Canada,67 Italy,64 Sweden,65 and Hong Kong.69, 70 The studies were published between 1998 and 2010. Sample sizes ranged from 92 to 792 participants (total N=2,417). We also identified six observational studies of CM for the frail elderly (see Appendix I, Evidence Table 4). Three were rated as having fair methodological quality,71-73 and three were rated poor.74-76 Four of the studies were conducted outside the United States.71, 72, 74, 76 All the studies defined cases on the basis of older age and presence of functional deficits.

The populations in the clinical trials were all elderly with some marker of frailty. All used an assessment of functional status in screening patients for eligibility, primarily through assessment of ADLs or IADLs. Mean patient age ranged from 74 to 85 years, with the mean in four studies being 80 or older.64-67 Three trials included a recent hospital admission or ED visit among the eligibility criteria.67, 69, 70

CM interventions in these studies focused on health care and community resource coordination (see Table 4). The clinical functions most often assessed were propensity to fall or functional status. Case managers were most commonly nurses, although some studies utilized other type of health care worker with geriatric expertise (e.g., physician assistant, social worker, allied health worker). Average caseloads varied widely among studies, ranging from 10 to 100. Interventions almost uniformly involved home visits in addition to telephone followup; the frequency of contacts varied among the studies. The case manager in one study initiated contact during a clinic visit and subsequent contact was via telephone only.30 The duration of study interventions ranged from 3 to 24 months (see Table 4). In general, reporting of case manager activity and location was poor; few studies identified how much the case manager interacted with the patient.

Table 4. Characteristics of case management interventions for the frail elderly (randomized trials).

Table 4

Characteristics of case management interventions for the frail elderly (randomized trials).

Comparators for CM were dependent on setting. In each study the comparator was usual care but without the CM component. The hospital-based studies69, 70 used usual hospital discharge services as comparators, the health care plan-based study66 used usual plan care, one study30 used usual primary clinic care, and four64, 65, 67, 68 used the package of home care and community services available to all study participants.

Targeted outcomes in the trials included patient-centered outcomes and resource utilization (Table 5). Patient-centered outcomes included mortality, measures of mental and physical health and functional status, satisfaction with health care, QOL, and measures of caregiver burden. Resource utilization measures included ED utilization, hospitalizations, nursing home admissions, outpatient visits, community service use, and overall costs. One study30 measured a quality outcome: the recognition and evaluation of common geriatric clinical problems.

Table 5. Characteristics and outcomes of studies of case management for the frail elderly (randomized trials).

Table 5

Characteristics and outcomes of studies of case management for the frail elderly (randomized trials).

The timing of the CM interventions varied and depended on how the study populations were identified. In essence, the interventions were initiated either in the course of the slow process of becoming frail or following a high risk clinical event. In three studies, CM interventions were initiated for participants with a recent history of hospitalization or ED use;67, 69, 70 three were initiated for participants enrolled or enrolling in a home-care assistance program;64, 65, 68 and two were initiated in populations already followed in primary care practices.30, 66

Settings for the trials varied; one was health plan-based,66 two were hospital-based,69, 70 one was conducted within the Veterans Affairs Medical Center (VAMC) health care system30 and four were community-based, one in the United States68 and three within national health care systems.64, 65, 67

Key Points Related to the Frail Elderly

  • CM does not affect mortality in frail elders (strength of evidence: low). (See Appendix H. Strength of Evidence.)
  • CM does not decrease acute hospitalizations in the frail elderly (strength of evidence: low).
  • CM does not decrease nursing home admissions in the frail elderly (strength of evidence: low).

Detailed Analysis: Effectiveness of Case Management by Outcome

Patient-Centered Outcomes
Mortality

Two of the good-quality trials measured mortality, and both found no reduction in the intervention group at 1 year64 or 3 years of followup.30 A fair quality trial69 reported 12-month mortality of 4 percent in the intervention group and 9 percent in the control group, but this study had a total sample size of only 92, so there was low confidence in this difference. Another fair quality trial68 reported no difference in mortality. The other four trials did not report mortality. A fair quality observational study72 also reported no difference in one-year mortality between the CM and comparison groups.

Functional Outcomes

There was marked heterogeneity in the studies of the frail elderly for the effects of CM on functional status. The study reported by Rubenstein and colleagues30 was rated as having good methodological quality, had the largest sample size, and had the longest followup (3 years). This study found that measures of functional status did not change significantly over time in either the CM or the control group. However, another good-quality trial64 found significantly better improvement in ADLs in the CM group. A fair quality trial66 found no change in ADL or IADL scores in the CM group over 2 years but worsening of these scores in the control group. Four other trials67-70 also found no difference between CM and control groups in ADL or IADL scores over 6-12 months. A poor quality observational study76 found improvement in functional status with CM, but a fair quality observational study72 did not find improvement with CM in their frail elderly group. A good quality trial that had a small sample size (23 participants per study arm) found no effect of CM on self-rated health,65 and a fair quality observational study had a similar finding.71

Quality of Care Outcomes

One good-quality trial30 had measures of the process of care as a primary outcome. This evaluation focused on five geriatric conditions that were assessed by medical record review. Documentation of all five problems was substantially higher for the CM patients. Clinical evaluation of the problems also was higher in the CM group. None of the other studies evaluated such outcomes.

Resource Utilization Outcomes

A primary rationale for CM for the frail elderly is to avoid unnecessary hospitalizations or ED visits. All seven of the eight clinical trials examined one or more utilization measures. In the good-quality trial by Rubenstein,30 about one-third of participants in both groups were hospitalized in each of 3 years of followup, with no difference in rates between the CM and control groups. In the other good-quality trial conducted in the United States66 hospitalization rates averaged 37 percent per year, without a significant difference between the CM and control groups. A fair quality trial and a fair quality observational study also found no difference in rates of hospitalization between the CM and control groups at 668, 73 and 18 months.68 Significant differences were not found in hospitalization rates between CM and control groups in either the trial conducted in Canada,67 or in a Canadian observational study,76 or in a trial in Hong Kong.70 However, trials conducted in Italy64 and a second Hong Kong trial69 found reductions in hospitalization rates with CM. Two poor quality observational studies found opposite effects of CM on hospitalization rates, with a small study in the United States reporting reduced hospitalizations 75 and a larger study in the United Kingdom finding no significant effect.74

Three trials looked at changes in ED visits. Marshall66 found no effect of CM on ED visits in the United States, while Gagnon67 found that CM was associated with higher rates of ED visits in Canada. In the Italian trial,64 the CM group had significantly fewer ED visits. A fair quality observational study in the United States found that ED visit rates were similar in CM and comparison groups.73 One trial64 also examined nursing home admissions and found no difference between the CM and control groups over 12 months.

CM has variable effects on use of outpatient services. The good-quality U.S. trial30 found that outpatient referrals to a variety of specialty services were significantly higher in the CM group than in the control group. However, the other U.S. trial66 found no significant difference in the numbers of outpatient visits between the CM and control groups. A trial conducted in Hong Kong70 found only small changes in outpatient visits with CM.

Three of the trials evaluated costs of care. A fair quality trial in the United States used total Medicare payments as the measure of cost and found no significant difference between the CM and control groups over 18 months.68 A good-quality trial in the United States estimated the total costs of care using approximations.66 The estimated costs were higher in the CM group than in the control group in both years of the study. A good-quality Italian trial64 also used an approximation method to estimate costs and found total costs to be significantly lower in the CM group, primarily due to the lower hospitalization rate. One poor quality observational study found a decrease in health care expenditures in the CM group,71 and two poor quality observational studies found a reduction in costs due to decreased hospital admissions.74, 75 Because of the inconsistency of these findings, we concluded that the evidence is insufficient to draw a conclusion about the effect of CM on overall costs for the frail elderly population.

Effectiveness of Case Management by Patient Characteristics

The modest sample size of the trials of CM for the frail elderly generally precludes subgroup analysis within this patient category. No studies examined age as a variable, and there generally were not good measures of comorbidity burden. There is no particular patient subgroup that appears to achieve greater success with CM.

Effectiveness of Case Management by Intervention Characteristics

The studies of frail elders generally included little information about the intensity of CM delivered, although all used relatively low caseloads (fewer than 100 patients) for the case managers. The greatest variation in outcomes was in measures of functional status, but none of the studies identified unique program characteristics that were linked to better functional outcomes.

Population: Patients With Dementia

Dementia is a disabling chronic disease for which the prevalence steadily increases with advancing age. It is estimated that about 14 percent of people in the United States who are older than 70 currently have dementia79 People with dementia have decreasing functional abilities over time, requiring the assistance of caregivers for their daily needs. Providing such assistance in institutional settings (such as nursing homes) is expensive and often is associated with isolation and medical complications. Avoiding or delaying placement in nursing homes has been widely regarded as a desirable clinical goal. There have been many major initiatives to examine possibly beneficial interventions. CM is one approach that has been studied.

Description of Studies

We identified 13 randomized trials of CM programs for patients with dementia (see Appendix I, Evidence Table 5); seven were rated good quality,31, 80-85 two were rated fair quality,86, 87 and four were rated poor quality36, 88-91 (see Appendix G). The trials were conducted in the United States,31, 36, 80, 84, 85, 89, 91 the United Kingdom,85 Hong Kong,86, 87 Canada,88 Finland,81, 82 the Netherlands,83 and Australia.85 They were published between 1990 and 2011. Sample sizes ranged from 78 to 8,138 participants (total N = 10,160). However, the majority of these studies were relatively small with 10 of the 12 trials having fewer than 100 participants in their CM intervention arms.80-83, 85-89, 91

The populations in all 13 studies were patients with dementia still living at home. The majority of patients lived with a caregiver. Each study enrolled a primary caregiver along with the patient (a study dyad) or involved the caregiver in the CM intervention. Mean patient age ranged from 68 to 83, with eight studies having a patient population averaging 78 years or older. In three studies that required the primary caregiver be a spouse, the mean age range of the spouse caregiver was 71 to 74 years.31, 82, 85 In studies that included caregivers other than spouses (most commonly a patient's child), the mean age of caregivers ranged from 44 to 66. Patient eligibility for five of the studies included a diagnosis of Alzheimer's Disease,31, 80, 85, 86, 88 the other eight a diagnosis of dementia (unspecified type).36, 81-84, 87, 89, 91 One study also included patients with a diagnosis code for memory loss.91

CM interventions in these studies focused on both patient and caregiver, with the majority emphasizing caregiver support (see Table 6). Intervention components aimed at caregivers included education on problem solving, communication, and coping skills provided through workshops, support groups, and individual counseling sessions. Those CM programs with control over budgeted services had the ability to provide caregivers additional services, such as respite and homemaking. Intervention components aimed at the patient included social and recreational activities, behavioral interventions, pharmacotherapy, and monitoring. Case managers in these studies were generally registered nurses or social workers. Caseloads (reported in only six studies) ranged from 25 to 100, most commonly 50 to 75. Case managers generally had face-to-face contact with patients and/or caregivers, in addition to telephone followup. The time horizon of most studies was 12-24 months, although one study31 followed the participants for more than 5 years.

Table 6. Characteristics of case management interventions for patients with dementia (randomized trials).

Table 6

Characteristics of case management interventions for patients with dementia (randomized trials).

The comparator group in 12 of the trials received “usual care”, which was defined as customary care through a primary care clinic, or more often through a community agency, without an assigned case manager. One study89, 90 was a head-to-head comparison: CM by an individual nurse case manager compared with CM by a team that included a nurse and a social worker. The team-based model in this study entailed more direct in-person interactions with clients, while the individual model was based on telephone interactions.

Targeted outcomes in these studies included patient and/or caregiver health, patient/caregiver satisfaction, quality of care, and resource utilization (see Table 7). Patient health outcomes included measures of dementia-related behavioral problems, cognition and function, QOL, and most often (8 of 13 studies) the ability to remain in the home. Caregiver health outcomes included measures of burden, depression, and QOL. Quality of care was measured by receipt of care consistent with clinical guidelines and measures of medication management (cholinesterase inhibitors, antidepressants, and other protocol driven treatments). Resource utilization measures included ED utilization, hospitalizations, nurse and physician visits, use of community services, and overall costs. Note that nursing home placement was classified as a patient health outcome due to its strong relationship to QOL.

Table 7. Characteristics and outcomes of studies of case management for patients with dementia (randomized trials).

Table 7

Characteristics and outcomes of studies of case management for patients with dementia (randomized trials).

The timing of a CM intervention can be considered in terms of where the patients are in the course of their disease process. Dementia is nearly always a progressive disorder, with decline in mental function and functional status over time. There is no clinical consensus on when in the course of the illness an intervention like CM would be most effective. As mentioned previously, all the patients in these 13 studies were still living at home. The majority had dementia of mild or moderate severity (for example, mean scores on the Folstein Mini Mental Status Scale of 15-20). Two studies specifically targeted patients with early dementia.83, 88

The setting for CM programs varied. Two were aligned with primary care clinics,80, 84 but more commonly they were situated within community agencies31, 85, 86, 89, 91 or national health care entities.81, 82, 88

In addition to the 13 randomized trials described above, we identified two observational studies, one rated fair quality92 and one rated poor quality.93 (See Appendix I, Evidence Table 6.) One study was conducted in the United Kingdom.92 Mean age of the participants was 80 and 70 percent were women, the majority of whom lived alone. The other trial was conducted in the United States.93 Participants ranged in age from 43 to 95 years and almost 75 percent lived with a caregiver. The comparison group in each of these studies included individuals followed in a similar community program that did not offer CM.

Key Points Related to Patients With Dementia

  • Patients with dementia who receive services from CM programs do not have lower mortality rates (strength of evidence: high). (See Appendix H. Strength of Evidence.)
  • CM programs that serve patients with dementia and have a duration of no longer than 2 years do not confer clinically important delays in time to nursing home placement (strength of evidence: moderate).
  • CM programs that serve patients with dementia reduce depression and strain among caregivers (strength of evidence: moderate).
  • CM programs that serve patients with dementia do not reduce problematic behavioral symptoms (strength of evidence: moderate).
  • CM programs that focus on clinical guideline measures for dementia increase adherence to those measures (strength of evidence: low).
  • CM does not change total health care expenditures for patients with dementia (strength of evidence: moderate).
  • CM programs that serve patients with dementia who have in-home spouse caregivers and continue services for longer than 2 years are more effective for delaying nursing home placement than programs providing services for 2 years or less (strength of evidence: low).

Detailed Analysis: Effectiveness of Case Management by Outcome

Patient-Centered Outcomes
Mortality

Ten clinical trials31, 37, 80-84, 86, 88, 89 and two observational studies92, 93 reported mortality rates. The time frames ranged from 1 to 3 years in all but one study, which followed patients for more than 10 years.31 Deaths often were not recorded after nursing home placement, which could bias the reported rates. The death rates varied considerably in the control groups, ranging from 3 percent at 18 months88 to 35 percent at 2 years.92 Across this group of studies, there was no trend toward significantly different mortality rates in the groups that received CM.

Patient's Ability to Remain at Home

A total of eight randomized trials and one observational study examined the patient's ability to remain at home. Two clinical trials had sample sizes of more than 100 participants per group.31, 36 Mittelman and colleagues31 conducted a long-term trial of CM for caregivers of patients with dementia at a single clinical site (New York City). The study had good methodological quality. It began in 1987, and participant accrual extended over 10 years.

Caregivers were required to be the spouse, the primary caregiver, and living with the person with Alzheimer's disease. The case managers were family counselors, who interacted primarily with the caregiver, and followed a protocol focused on strategies for coping with stressful situations in the caregiving role. The CM activities extended over the entire duration of followup (as long as 10 years). There are no other studies of CM in this clinical domain that continued the intervention longer than 2 years.

Over the initial 6 years of followup in the New York trial, nursing home placement was about 12 percent a year in the control group and about 9 percent a year in the intervention group. By 11 years, about 80 percent of the control group patients and 70 percent of the intervention group patients had either died or moved to a nursing home. The authors estimated that the intervention delayed nursing home placement by an average of about 18 months.

The Medicare Alzheimer's Disease Demonstration and Evaluation (MADDE) was a uniquely large clinical trial.36, 37, 99, 100 The MADDE study enrolled more than 8000 volunteers into a prospective randomized trial of CM between 1989 and 1991. The project was conducted in eight states, and the models of CM varied considerably across sites. While the programs included components designed to reduce caregiver stress, the fidelity of the intervention across the multiple sites is unknown. The overall rating of methodological quality is poor. The overall rate of nursing home placement in the MADDE study was 43.5 percent at 3 years of followup. There was no significant difference in this rate between the intervention and control groups. Subgroup analyses examined the case manager's caseload and relationship of the caregiver to the patient (spouse vs. nonspouse). There were no significant effects of receipt of CM on nursing home placement rates in these subgroups.37

Six smaller clinical trials examined nursing home placement rates as an outcome measure. Four of these were judged to have good methodological quality. All continued the CM for 2 years or less. Eloniemi-Sulkava and colleagues reported a randomized trial of CM in Finland, with a total of 100 participants enrolled between 1993 and 1995.81 Thirty-one percent of patients had moved to nursing homes at the end of 2 years. While the overall rate did not differ between the experimental and control groups, a Cox regression analysis found that patients in the intervention group moved to nursing homes significantly later (p=0.04) than patients in the control group. These results suggest a mild benefit of CM in maintaining patients at home that is not sustained over time. Eloniemi-Sulkava then reported on a second clinical trial, also in Finland, with the participants being recruited in 2004.82 This trial included a total of 125 participants and had very similar results to the earlier Finnish trial. The overall rate of nursing home placement was 26 percent at 2 years, with no significant difference in the overall rate between intervention and control groups.

Two other good-quality studies failed to find an effect of CM on rates of nursing home placement. Mittelman85 repeated the model of CM that previously had been found to delay nursing home placement when continued long-term.31 This replication trial was conducted in the United Kingdom, Australia, and the United States. However, the number of participants was small (between 52 and 54 participants in each country), and the duration of CM was only up to 2 years. The mean time to nursing home placement was 4.1 years in the intervention group and 4.3 years in the control group. Overall nursing home placement rates were lower in the United States than in the other two countries. Callahan80 also conducted a trial of CM for patients with dementia. The intervention lasted 12 months and emphasized caregiver skills for coping with bothersome patient symptoms. The nursing home placement rate was 5 percent at 18 months, with no difference between the intervention and control groups. A fair quality trial in Hong Kong found decreased rates and duration of institutionalization for the intervention group at 12 months.86

One fair quality observational study92 and two poor quality trials88-90 did not provide evidence that would change the conclusions reached from the studies described above. Chu88 reported a small (total of 74 participants) clinical trial that had poor methodological quality. At 18 months of followup, the nursing home placement rate was 28 percent in the control arm. There was no significant difference in placement rate between the intervention and control groups. The authors estimated that CM delayed nursing home placement by an average of 53 days among patients with more severe dementia. In an observational study92 conducted in the United Kingdom 43 patients in a CM program were compared with 43 matched controls who did not receive CM. At 2 years, 31 percent of all patients had died. Twenty-one percent of the CM patients had been placed in a nursing home compared with 33 percent of the patients in the comparison program. Finally, a poor quality clinical trial of two types of CM in the State of New York included a subgroup analysis of the participants who had dementia.90 Nursing home utilization was nearly identical among patients receiving CM by individual case managers when compared with patients who received a more intensive model of CM by a team that included a nurse and a social worker.

Patient and Caregiver Health Outcomes

One of the major challenges in caring for patients with dementia is management of problematic behavioral symptoms. Studies of CM have used a variety of methods to measure such symptoms, using two different but related approaches. The first approach is to use a questionnaire such as the neuropsychiatric inventory (NPI) that measures the caregiver's report of the severity of symptoms. The second approach is to assess measures of strain or burden experienced by the caregiver due to performing the caregiving role. Thus, these are measures of the frequency/severity of patient behaviors and the caregiver's stress in dealing with these behaviors. Since the same caregiver usually completes both types of measure, the measurements are not independent. Also, because a variety of different instruments have been used, we will report the trends in such measurements for each study rather than separating out each type of assessment.

Of the 13 randomized trials of CM for dementia, 11 included measures of the caregivers' perceptions of the patients' behaviors. Five of these trials had good methodological quality. The trial reported by Mittelman had both the longest duration of CM and the longest followup period.95 This study found no difference over time between the CM and control groups in the frequency of problematic behaviors. However, caregiver stress associated with the behaviors was significantly lower in the CM group and this effect persisted over a 4-year period. The caregivers in the CM group also reported lower scores on a depression scale, but this difference did not persist beyond 3 years.

Mittelman and colleagues also performed a second randomized trial to replicate the original study.85 This trial continued CM for only 2 years. While caregiver burden scores were lower in the CM group, this was not statistically significant. Caregiver depression scores were significantly lower in the CM group during the followup period. While depression scores increased over time in the control group, they decreased over time in the CM group.

Other clinical trials rated as either good or fair quality have had shorter followup periods, but their results generally are consistent with those found by Mittelman.85 Callahan80 included caregiver assessments of the NPI, a rating of patient depression, and a measure of caregiver stress at 6, 12, and 18 months, although the CM program ended at 12 months. The NPI scores were better in the CM arm at both 12 and 18 months. Measures of caregiver stress also were better in the CM arm at 12 and 18 months. Vickrey84 assessed caregiver confidence and QOL after 18 months of CM. Confidence increased modestly in the CM group, but measures of QOL and caregiver strain did not change. Jansen83 conducted a trial of 12 months of CM. This trial found no differences between the CM and control groups in measures of caregiver depression, QOL, or caregiver burden. Chien86 conducted a fair quality trial in which patient NPI scores and a caregiver burden measure did not change over 12 months in the control group. In the CM group, both measures significantly improved at 12 months. Lam87 assessed caregiver outcomes after a 4-month CM intervention. There was no change in measures of caregiver stress and QOL for the CM and control groups at 4 and 12 months. Psychological health scores were unchanged at 4 months but increased for CM caregivers at 12 months.

Of three other clinical trials rated as poor quality, one91 found mild effect of CM on improvement of patient symptoms at 12 months and one88 found no effect on symptoms. Two of these trials88, 99 found no effect of CM on caregiver burden or depression. An observational study92 found decreased caregiver burden in the group receiving CM. Another observational study93 found a positive effect of CM on caregiver stress, well-being, and endurance.

Quality of Care Outcomes

The clinical trials of CM for patients with dementia generally have provided only limited data about the effects of the programs on processes of care. However, Vickrey and colleagues84 reported a good-quality randomized trial that had adherence to dementia care guidelines as its primary outcome. The study had 23 prespecified dementia guidelines that were included in the clinical protocol for CM. These fell into four clinical domains: assessment, treatment, education/support, and safety. At 18 months, the care was judged to be adherent to a mean of 33 percent of the guidelines in the control group and 64 percent of the guidelines in the CM group. No other studies of CM have examined its effect on guideline adherence.

Resource Utilization Outcomes

Multiple studies have examined the effect of CM on the use of outpatient and inpatient care. While various individual utilization measures have been studied, there is a sufficient body of evidence to draw a conclusion about the effects of CM only for measures of overall costs. Three randomized trials and one observational study evaluated the effect of CM on costs of care for patients with dementia. These studies evaluated costs over 1-2 years of followup. Duru102 examined costs in a good-quality trial evaluating health care, caregiving, and out-of-pocket costs over 18 months. The monthly cost for CM was modest (mean $118). Total costs (from either a societal or payer perspective) were slightly higher in the control group, but this was not statistically significant. Another good-quality trial also found slightly higher total costs in the control group, but the difference was not statistically significant.82 The MADDE trial was a large trial that included an incentive to use home-care services by the CM group. It found that CM had little effect on Medicare expenditures.36 In an observational study92 total costs were higher in the CM group, primarily due to higher utilization of clinic visits and acute care hospitalizations. Overall there is a moderate strength of evidence indicating that CM has little effect on the overall cost of care in this population.

There is insufficient evidence to draw conclusions about CM effects on specific types of utilization in this population. In the good-quality trial reported by Callahan,80 the frequency of primary care clinic visits was higher in the CM group but acute care hospitalization rates did not differ between groups. A lower quality trial91 found that the CM group had a higher rate of physician visits but lower rates of ED visits and hospitalizations. In an observational study, both psychiatric and medical hospitalizations were higher in the CM group.92

For in-home services, one good-quality trial84 and a fair quality trial87 found that patients in the CM groups had higher utilization of respite and outside caregiver services. Jansen's good-quality study83 found no differences in utilization of in-home services between the CM and control groups. A lower-quality trial88 also found that the CM and control groups did not differ in the use of in-home services. Another low-quality trial100 found increased use of community services among patients receiving CM, but this trial included a financial benefit for these services (in the CM but not the control group), so it is a biased evaluation of this effect. Overall, there is only a small body of evidence about the effect of CM on use of in-home services among patients with dementia.

Effectiveness of Case Management by Patient Characteristics

As previously described, most of these studies had sample sizes of less than 100 participants in each study arm, which provided little power for subgroup analyses. In two clinical trials, the participants were stratified by severity of dementia. Using time to nursing home placement as the outcome, the differences between intervention and control groups was greatest among those with the greatest severity of dementia, suggesting that these individuals were more likely to benefit from CM.82, 88 Another trial91 performed regression analyses to see if patient characteristics were associated with utilization outcomes, but these results found no clear trends.

Effectiveness of Case Management by Intervention Characteristics

The only head-to-head trial comparing two different approaches with CM was an older randomized trial of individual compared with team-based CM.90 This trial tested the hypothesis that a team-based approach that provided more frequent patient contact and more home visits would lead to superior outcomes. It had poor methodological quality and had negative findings. For indirect comparisons, the major evidence comes from Mittelman's good quality trial conducted in New York City.31 As described above, this program provided long-term CM (up to 10 years) and specialized in providing services to live-in spouse caregivers. All other CM programs that have been studied served a variety of spouse and nonspouse caregivers and continued services no longer than 2 years. The positive findings in the Mittelman study suggest that long-term specialized CM programs for this clinical problem may have superior success in reducing caregiver depression and stress and in delaying nursing home placement.

Population: Patients With Congestive Heart Failure

Congestive heart failure (CHF) is an illness associated with substantial morbidity and mortality in the elderly and is characterized by frequent exacerbations that make it the leading cause of hospitalizations among Medicare beneficiaries.103, 104 Nurse-led interventions, including CM, are commonly used to improve CHF management, with the goals of improving patients' QOL, maintaining clinical stability, and preventing CHF exacerbations and hospitalizations. CM functions used to achieve these goals typically include educating patients to enhance their self-management knowledge and skills; coordinating and facilitating access to multiple clinical providers and services; monitoring clinical parameters; and sometimes adjusting medication regimens and doses.

Description of Studies

We found 11 clinical trials of CM for patients with CHF (see Appendix I, Evidence Table 7); five were rated as good quality,105-109 four fair,110-113 and two poor114, 115 (see Appendix G). Sample sizes of the included trials ranged from 58 to 1049 patients (total N for all studies = 3,804). The studies were published between 1993 and 2010. We also identified one, small, poor quality observational study from our search.116 (See Appendix I, Evidence Table 8.)

The populations in the 11 trials varied, ranging from members of a large health maintenance organization who were at low risk for hospitalization105 to patients with predominantly severe CHF, living in a low-income, urban neighborhood.109 Other studies fell within this spectrum, with patients who had moderate (New York Heart Association class II and III) heart failure106, 107, 110-113, 115 and were at increased risk for hospitalization.107, 110, 111, 115 Three studies included only patients with systolic heart failure, typically indicated by a left ventricular ejection fraction of less than 35 to 45 percent,108-110 while others included patients with both systolic and diastolic dysfunction.105-107, 111-113 The mean age in most studies ranged from 60 to 80. A feature of most of these studies that differs from the studies of CM in all other clinical categories in this report is that in all of the studies except two,108, 109 the patients were enrolled during an acute hospitalization.

CM interventions varied in nature and intensity (see Table 8). The focus of the interventions ranged from predominantly clinical management—including self-management education, monitoring of clinical parameters, and adjustment of medications—to a more comprehensive CM approach that included a strong element of service coordination and social support. All interventions employed telephone contacts, six included home visitation,106-108, 111, 114, 115 and four held face-to-face clinic visits.106, 107, 109, 114 Case managers were nurses in all interventions; some received supervision from physicians or more senior nursing staff. Most interventions employed protocols or algorithms to guide clinical management of CHF. Most studies evaluated the effect of CM as an isolated intervention, but some included CM as a component of a multidisciplinary team approach to discharge planning and disease management.107, 111, 114, 115

Table 8. Characteristics of case management interventions for patients with congestive heart failure (randomized trials).

Table 8

Characteristics of case management interventions for patients with congestive heart failure (randomized trials).

The comparator group in most studies received usual care without CM. What constituted usual care in most studies was a CHF-specific discharge plan for patients enrolled during a hospitalization with outpatient primary care followup that was not standardized. In a study from the Netherlands, outpatient followup care was provided by cardiologists.106 That study also included a third arm, in addition to CM and usual care, in which patients received nurse-led CHF management that focused on clinical management alone (without coordinating functions) and was less intensive than the CM intervention.106

The primary outcome in all studies enrolling inpatients was hospital admission,105-107, 110-113, 115 with some studies targeting the composite outcome of admission or death.106, 107, 111 for two studies enrolling outpatients, the primary outcomes were all-cause hospitalizations109 and health-related QOL.108 Other outcomes included patient satisfaction,108, 110, 112 patient adherence to self-care plans,107, 108, 110 receipt of guideline-recommended CHF medications,105, 107, 110 and the overall cost of care.110-113 All studies examined multiple outcomes (see Table 9).

Table 9. Characteristics and outcomes of studies of case management for patients with congestive heart failure (randomized trials).

Table 9

Characteristics and outcomes of studies of case management for patients with congestive heart failure (randomized trials).

The timing and setting of CM interventions was in most of the studies related to the principal objective of preventing readmissions among patients hospitalized for CHF. Case managers typically engaged with patients prior to hospital discharge and followed them for 3 to 18 months, depending on the duration of CM stipulated in different study protocols.105-107, 110-115 Two studies enrolled outpatients from community medical practices.108, 109

In one good-quality study, the authors reported a conflict of interest, indicating that the Division of Cardiology in which they worked had stock in and was entitled to royalties from the disease management company whose CM intervention they implemented and evaluated.107 The company also provided funding for the study, which demonstrated multiple benefits of CM over usual care.

Key Points Related to Patients With Congestive Heart Failure

  • CM does not reduce mortality among adults with CHF (strength of evidence: low). (See Appendix H. Strength of Evidence.)
  • CM improves CHF-related QOL (strength of evidence: low).
  • CM increases patient satisfaction (strength of evidence: moderate).
  • CM increases patients' adherence to self-management behaviors recommended for patients with CHF (strength of evidence: moderate).
  • CM is more effective in improving outcomes among CHF patients when case managers are part of a multidisciplinary team of health care providers (strength of evidence: low).

Detailed Analysis: Effectiveness of Case Management by Outcome

Patient-Centered Outcomes
Mortality

None of the included studies was explicitly designed to examine the impact of CM on mortality, although three trials included mortality as part of a composite primary outcome measure, usually coupled with rehospitalization.106, 107, 111 Three other trials reported mortality rates without explicitly defining it as an outcome.105, 112, 114 No study found a statistically significant improvement in either all-cause or CHF-related mortality, but all but one study114 reported lower mortality rates in the CM group compared with controls (RR 0.74 to 0.88). The small number of studies, coupled with heterogeneity of the patient populations, CM interventions, and duration of followup, precluded pooling of data to derive a meaningful estimate of potential mortality reduction with CM. The consistency of relative risk across five studies, however, raises the possibility that CM may provide a survival benefit over usual care for patients with CHF. However, because none of the studies found a statistically significant mortality improvement, the overall impact on improvement appears to be low.

Quality of Life

Six studies examined the effect of CM on QOL, using a variety of CHF-specific instruments, including the Minnesota Living with Heart Failure Questionnaire,107, 109 the Kansas City Cardiomyopathy Questionnaire,108 and the Chronic Heart Failure Questionnaire.111 Four of these studies also used global measures of functional status that are not specific to CHF: the Medical Outcomes Study SF-36108, 109, 114 and the EuroQOL EQ-5D™.113 Among these six studies, three found significant improvements in CHF-related QOL among patients receiving CM,107, 109, 111 one of which also found improvements in overall functional status.109 In the other three studies, QOL scores were similar in the CM and control groups, with minimal evidence of trends towards better QOL in either the CM or control groups. In the study showing improvements in overall functional status with CM, the improvement occurred in both physical and emotional domains of functioning.109 Notably, in the one study that followed patients beyond the end of the intervention period, functional status declined in the CM group at a rate similar to that in the control group,109 suggesting that the benefits of CM may not be durable unless the intervention is continued. Because of the heterogeneity of findings across the studies, the strength of evidence for the effect of CM on QOL was rated as low.

Patient Satisfaction

Three studies reported the impact of CM on patient satisfaction with care.108, 110, 112 Two used general measures of patient satisfaction designed or adapted specifically for their studies and found modest but statistically significant improvements in satisfaction in the CM groups compared with controls.110, 112 The third study used the Patient Assessment of Chronic Illness Care (PACIC) instrument and found significant improvements in patient ratings with CM.108 Because of the consistency of positive findings across three studies, we judged the strength of evidence to be moderate that CM improves satisfaction among patients with CHF.

Quality of Care Outcomes

Four studies evaluated the impact of CM on indicators of quality of care for CHF. Three examined the use of appropriate pharmacotherapy (e.g., angiotensin-converting enzyme inhibitors or angiotensin-receptor blockers and beta-blockers for patients with systolic heart failure).105, 107, 110 One study showed improvements in the use of recommended medications with CM,107 while the other two did not.105, 110 Three studies examined adherence to self-care recommendations (e.g., low-sodium diet, monitoring weight).107, 108, 110 All three found that patients' adherence to self-management recommendations improved with CM.107, 108, 110 Because of the consistency of positive findings across these three studies, we judged the strength of evidence to be moderate that CM improves adherence to self care behaviors for CHF.

Resource Utilization Outcomes

Nine studies reported the impact of CM on all-cause hospitalization rates.105-107, 109-113, 115 Results were mixed, with four studies showing lower hospitalization rates with CM107, 109, 111, 112, 115 and five showing no difference between CM and controls.105, 106, 110, 113-115 In the five studies showing no difference, the relative rates of hospitalization in CM compared with control groups ranged from 1.02 to 1.12. In the four studies reporting significantly lower hospitalization rates with CM,107, 109, 111, 112 the relative rates ranged from 0.56 to 0.79, and absolute differences ranged from 19 fewer hospitalizations per 100 patients over a 12-month observation period109 to 30 fewer hospitalization per 100 patients over a 3-month period.111 Reductions in all-cause hospitalization rates were driven primarily by lower rates of hospitalization for CHF. Five studies examined the total number of hospital days during the study period, with one reporting fewer hospital days per patient in the CM compared with control group (3.9 vs. 6.2 days over a 3-month period),111 and four reporting no difference.107, 110, 112, 113

We examined whether study quality was associated with the effects of CM interventions on inpatient utilization. Of the four studies that were rated as having the highest methodological quality,105-107, 109 two107, 109 found lower hospitalization rates and two105, 106 found no decrease in hospitalization rates with CM. The study with the largest sample size (conducted in the Netherlands) found no reduction in hospital admissions.106 We concluded that there is heterogeneity of results for this outcome. While CM may reduce hospitalization rates for patients with CHF, there presently is insufficient evidence to draw a conclusion about this effect.

Six studies estimated the impact of CM on the overall or hospital-related cost of care.107, 110-114 Total costs were dominated by the cost of inpatient care, with estimated costs for CM interventions being comparatively small. Accordingly, three studies demonstrating reductions in hospitalization rates with CM also found reductions in cost,111, 112 though in one study this difference was not significant.107 The two studies that did not find improvements in hospitalization rates also did not find a positive or negative impact of CM on the cost of care.110, 113

We examined study characteristics, patient populations, and intervention components across studies to try to determine which elements might explain the mixed results for the impact of CM on various outcomes. Broadly speaking, three studies can be classified as “negative,” demonstrating no differences between CM and control groups across outcome measures.105, 106, 113 Four studies can be considered “positive,” demonstrating improvements in QOL, hospitalization rates, and/or cost of care.107, 109, 111, 112 Finally, two studies can be considered “intermediate,” showing some improvements in patient-reported measures of satisfaction and self-care but not in health outcomes or hospitalization rates.108, 110 We omitted two small poor quality studies from this analysis.114, 115

Effectiveness of Case Management by Patient Characteristics

Case management is considered to be most appropriate for patients at high risk for poor outcomes. Three studies selected hospitalized patients who had features considered to put them at high risk for readmission.107, 110, 111 Two of these studies showed lower hospitalization rates with CM,107, 111 while the third demonstrated improvements in patient-reported outcomes.110 Four other hospital-based studies enrolled either low-risk105 or unselected patients106, 112, 113 with CHF. Results were negative in three of these studies105, 106, 113 and positive in one.112

In the three studies enrolling high-risk patients, high risk was defined as having one or more risk factors for readmission. The specific risk factors varied across the three studies. In two of them, a prior history of CHF was considered a risk factor in and of itself.110, 111 All three studies included recent hospitalizations as a risk factor, though the specific criteria varied (e.g., four hospitalizations for any reason over the prior 5 years compared with one CHF hospitalization during the prior year). Two studies showing a positive impact of CM on readmission used clinical parameters (e.g., uncontrolled blood pressure) to select high-risk patients,107, 111 while the other study, which found an impact on self-care but not on readmission, used mainly social and behavioral factors (e.g., knowledge deficits, potential for lack of adherence, living alone).110

Baseline hospitalization rates (those observed in control groups) can also be considered a proxy measure of risk within the sampled populations across studies. We calculated control-group hospitalization rates for each study reporting them and adjusted rates for the duration of followup. Three studies demonstrating a reduction in hospitalization rates with CM107, 111, 112 had relatively high baseline rates of hospitalization (1.74 to 2.69 per person per year), while three negative studies105, 106, 113 had relatively low rates (0.74 to 0.99 per person per year). A study with an intermediate baseline hospitalization rate (1.47 per person per year) showed no reduction in hospitalizations with CM but improvements in self-care and patient satisfaction.110 These findings suggested a pattern of higher success with CM in populations at higher risk of hospitalization. One study did not fit this pattern. In that study,109 CM successfully reduced hospitalization rates in a population with a relatively low baseline rate (0.89 per person per year). This study also differed from others in that patients were not hospitalized at the time of recruitment.109

There was no clearly discernible pattern in study outcomes based on whether the study sample included only patients with systolic heart failure or patients with either systolic or diastolic dysfunction. Likewise, functional status, as measured by New York Heart Association class did not appear to be associated with the impact of CM. The studies with the most class III and IV patients included one positive109 and one negative study.113

Few studies specifically targeted vulnerable patient populations. One included predominantly ethnic minority patients with relatively poor functional status in a low-income, urban neighborhood (Harlem, New York City) and found improvements in both QOL and hospitalization.109 Another study, however, enrolled Spanish-speaking patients in Southern California, also with poor functional status, and found no improvement with CM.113 The authors of this latter study used a CM intervention that was essentially identical to one they used in an earlier study, in which they enrolled predominantly English-speaking patients and found significant reductions in hospitalization rates and cost with CM.112 Linguistic and cultural factors may explain the difference in success in these two interventions. However, the populations in these two studies also differed in other ways, including more class IV heart failure patients in the unsuccessful study.

In two studies which found CM not to be superior to usual care for any outcome measure, the authors reported that the baseline quality of CHF care may have been sufficiently high such that there was minimal room for the CM intervention to improve quality of care and thereby result in better outcomes.105, 106 These studies were conducted in a large HMO (Kaiser Permanente) with a strong quality improvement focus105 and in a group of cardiology practices.106 While it is possible that these settings may have resulted in control groups that received higher quality care than in other studies, we did not observe higher rates of appropriate pharmacotherapy in the control groups of those two studies, compared with other studies that reported superior outcomes with CM.

Overall, there were no consistent trends when examining patient subgroups in this set of studies. We concluded that patient characteristics do not appear to mediate the effects of CM for patients with CHF.

Effectiveness of Case Management by Intervention Characteristics

No studies included head-to-head comparisons of different models of CM. One study compared an intensive CM program with a more basic and less intensive disease management intervention.106 Neither the CM nor the less intensive intervention was superior to the control group on any outcomes.

There were few discernible patterns in terms of intervention characteristics that predict successful CM interventions (Table 8). Interventions that were longer in duration did not produce more positive results, nor did the use of home visits, as opposed to telephone care alone or the amount of contact time. Only one study reported CM caseloads.110 In most studies, CM functions were heavily weighted towards clinical activities, as opposed to coordinating functions; the specific CM functions employed did not track with intervention success, though few studies described CM functions with enough specificity to allow clear delineations in the nature and intensity of those functions. The ability of nurses to adjust medications was present in only two studies, one with negative and one with positive results.105, 107

The degree to which the care delivered by case managers was integrated with patients' usual care providers (usually primary care physicians or cardiologists) was not well described in most studies. Interventions that appeared to include higher levels of integration with usual care providers did not clearly produce better results than others. One study, however, reported significantly lower hospitalization rates among patients whose usual care providers were in the local vicinity where the case manager worked and with whom he or she had closer contact.110 No such improvement in hospitalization rates was observed among patients with nonlocal providers.

The presence of physician supervision of case managers was not clearly associated with better outcomes. Two studies, however, that embedded case managers within teams that included other health professionals (e.g., cardiologist, social worker, dietitian) demonstrated better outcomes across multiple domains in the intervention compared with control group.107, 111 Preintervention training for nurse CMs and care protocols to guide clinical management were not more prevalent in successful compared with unsuccessful CM interventions.

Population: Patients With Diabetes Mellitus

Diabetes mellitus (DM) is a significant health problem, currently estimated to affect 26 million Americans and approximately 27 percent of adults over age 65.117 The prevalence of diabetes continues to rise, as do the associated increased risks of cardiovascular disease, end stage renal disease, neuropathy, and retinopathy. Considerable health care resources have been devoted to seeking mechanisms to optimize care as a strategy to diminish the morbidity and mortality associated with this chronic health condition. Diabetes is especially complex in that its management requires avid and persistent participation from both providers and patients. Desired patient behaviors often are complex, with permanent alterations of habits (diet and exercise) and complex medication regimens. The ability of case managers to work with patients to improve education and individual goal setting may positively impact patients' understanding of their diabetes and their self-directed care activities. Case manager involvement may also aid providers via improved tracking and implementation of systems to monitor glucose control and to obtain routine tests that screen for disease complications. Although the overall functional status of many patients with diabetes is relatively good, the rationale for CM is that they need assistance and training to improve both self-management skills and the overall coordination of their health care.

Description of Studies

We identified 12 studies in total, nine clinical trials and three observational studies, of adults with diabetes (see Appendix I, Evidence Table 9 and 10). Of the clinical trials, two were rated good,118, 119 six were rated fair,120-134 and one was rated poor135 (see Appendix G). The study by Brown and colleagues was listed as poor quality due to unevenness of characteristics between groups at baseline as well as no reporting of withdrawals, attrition, or crossover between groups. Due to the poor quality of this study, its data will not be further included in this discussion and it is not incorporated in the tables within this chapter. All eight included trials were conducted in the United States and published between 2002 and 2009. Sample sizes of included trials ranged from 147 to 1,665 participants (total N = 3,776); notably, the majority of these studies were relatively small, with five of the eight having fewer than 400 participants.118, 120, 123, 125-127, 136 Of the three observational studies, two were rated fair137, 138 and one was rated good.139, 140 All three observational studies utilized a retrospective cohort design, were conducted in the United States, included a total patient population of greater than 1,000 (range 1,076 to 5,925), and were published between 2005 and 2009. Study duration ranged from 12 months to 5 years. Only one study examined outcomes at 5 years,130 however, and 10 of the 11 included studies limited their followup or retrospective analysis to 2 years or less.

The populations examined by the 11 included studies varied significantly. The mean age of participants ranged from 48 to 71. There was notable heterogeneity in racial/ethnic backgrounds, as some trials limited their patient populations to African Americans,121, 122 American Indians/Alaskan Natives,137, 138 or Latinos.120 Five trials examined only individuals with type II diabetes.118, 120, 121, 123, 124, 126, 127, 136 The trials also exhibited different levels of complexity defined by differences in disease severity (measured as mean hemoglobin A1c [HgA1c] and duration of DM).118, 120, 122, 125-127, 136 Population complexity also varied between studies due to different degrees of socioeconomic disparity in that five of the eight trials included populations of lower socioeconomic status120-124, 126-130 and four studies documented a low educational level in the majority of their included population.120, 121, 123, 124, 126 One of the eight trials documented that a large percentage utilized medical assistance programs or were uninsured.122

The intensity of the CM intervention was similar in seven of the 11 studies, in that face-to-face interaction was the primary mode of CM delivery supplemented with telephone contact118, 120, 122, 125-127, 136, 138, 139 (see Table 10). One study, rated good quality, included two face-to-face visits but relied primarily on telephone contact for most of the CM intervention.119 Unfortunately, very few trials reported adherence to the CM intervention or the number of case manager interactions/visits achieved, making true intervention intensity difficult to assess. Case managers for the included studies were primarily nurses, although some were registered dietitians118, 126, 127, 136 or social workers.139 The comparator group for each study was defined as usual care, which uniformly referred to care by a patient's primary care or usual care provider. Targeted outcomes in the included studies included patient health, patient satisfaction, quality of care, and resource utilization (see Tables 11 and 12). Patient health outcomes included hard endpoints (i.e., mortality and QOL) as well as a number of intermediate measures such as HgA1c, cholesterol management, blood pressure control, and weight/body mass index (BMI) among others. Quality of care measures included eye and foot examinations, medication adherence, and glucose self-monitoring. Resource utilization measures included ED utilization, hospitalizations, primary care utilization, and a cost analysis. The timing of the CM intervention was mentioned by only one trial, which limited their population to individuals with a new diagnosis of type II diabetes (within 6 months of enrollment).120 The settings for the CM intervention was conducted in an outpatient clinic setting in eight of the eleven studies,118, 120, 121, 123, 124, 126, 127, 136, 138, 139 while the setting in the remaining three studies was unclear.

Table 10. Characteristics of case management interventions for patients with diabetes (randomized trials).

Table 10

Characteristics of case management interventions for patients with diabetes (randomized trials).

Table 11. Characteristics and outcomes of studies of case management for patients with diabetes (randomized trials).

Table 11

Characteristics and outcomes of studies of case management for patients with diabetes (randomized trials).

Table 12. Intermediate health outcomes among trials of case management for diabetes mellitus.

Table 12

Intermediate health outcomes among trials of case management for diabetes mellitus.

Key Points Related to Patients With Diabetes

  • CM does not reduce mortality among adults with diabetes (strength of evidence: low). (See Appendix H. Strength of Evidence.)
  • CM improves glucose control among adults with diabetes (strength of evidence: low).
  • CM does not improve lipid management or weight/BMI in patients with diabetes (strength of evidence: moderate).
  • CM does not reduce hospitalization rates among adults with diabetes (strength of evidence: low).

Detailed Analysis: Effectiveness of Case Management by Outcome

Patient-Centered Outcomes
Mortality

One study, a clinical trial rated as fair quality, examined 5-year mortality in adults with diabetes who were exposed to a CM intervention.130 This study did not find a mortality benefit from this CM intervention after 5 years (hazard ratio for mortality 1.01, 95% confidence interval [CI] 0.82 to 1.24).

Quality of Life

Two clinical trials examined differences in QOL among adults exposed to CM.118, 120 These trials utilized similar CM intervention techniques but found discordant results (one positive study in favor of CM and one negative study).118, 120 The positive study observed differences in seven of nine examined categories in a standard scale commonly used to assess QOL.118 There is insufficient evidence to draw a conclusion about the effect of CM on QOL among people with diabetes.

Changes in Hemoglobin A1c

All of the eight included clinical trials examined differences in glucose control over time, measured by HgA1c, and the majority identified no difference in this intermediate outcome with CM intervention (see Table 11).118, 119, 122, 123, 125, 126, 129, 130, 136 The intervention setting and duration were similar in six of these trials (see Table 11). One good-quality trial119 found statistically significant improvement in HgA1c to less than 8 percent in the CM group compared with usual care. Two fair quality trials126, 129, 130 found statistically significant declines in HgA1c in the CM groups compared with usual care. One trial, also rated as fair quality, provided information on within-group change in HgA1c over time and identified a possible benefit of CM for HgA1c improvement by this metric. Four trials, three fair quality and one good quality, found no significant difference between CM and usual care groups.118, 122, 123, 125 Three observational studies also examined changes in HgA1c between CM and control groups. Two of these three studies, one rated good quality and one rated fair quality, found improvement in HgA1c among individuals exposed to CM while the third study found no significant difference between groups.137-139 Taken together, this evidence suggests that CM intervention improves glucose control in patients with diabetes, but there is marked heterogeneity of results for this outcome. The strength of evidence for the conclusion of a positive effect on glucose control is low.

Additional Intermediate Health Outcomes

Seven clinical trials and one observational study examined a cholesterol-related outcome—change in total cholesterol, triglyceride levels, low-density lipoprotein (LDL) cholesterol, or high-density lipoprotein (HDL) cholesterol118, 119, 122, 123, 125, 126, 129, 130, 139 (See Table 12). Of these, the vast majority identified no benefit of CM for improving measures of cholesterol control. Limited improvement was identified in two studies122, 129, 130 with regards to two specific measures (LDL and triglyceride levels). The most commonly measured outcome was LDL level, and only one trial129, 130 showed a benefit of CM for improving this outcome. Because of the consistently negative findings in the other studies, we concluded that there is moderate evidence that CM does not improve lipid measures, when compared with usual care.

Changes in blood pressure were predominantly examined as changes in systolic blood pressure (SBP) and diastolic blood pressure (DBP) (see table 12). The one exception is the trial by Ishani and colleagues, which examined both components of blood pressure together with goal to reach less than 130/80 mmHg; this trial found a significant improvement in blood pressure in CM compared with usual care (p=0.047). One fair quality study also identified benefit of CM for hypertension control.128-130 Five trials (all fair quality) examined changes in SBP and DBP and the majority (four of the five) identified no benefit or excess harm of CM for SBP management.122, 123, 125, 126 In total, three of the seven studies which examined blood pressure control identified a benefit of CM including one good-quality study, compared with four studies (all fair quality) which did not. Because of the inconsistency of these results, the evidence is insufficient to discern whether CM leads to improved control of hypertension in people with diabetes.

Four trials examined changes in BMI and none of these identified a benefit of CM120, 122, 123), 126 (see Table 12). Two trials, one good quality118 and one fair quality,126 examined change in weight and describe discordant results. In total, five trials found no benefit in BMI/weight adjustment with CM intervention, while one did find a benefit.

Patient Satisfaction

Krein et al. assessed “general satisfaction” of individuals who received CM compared with usual care and found significantly greater satisfaction among patients in the CM group (p=0.04).125

Quality of Care Outcomes

Quality of care was examined via process measure outcomes. Process measure outcomes include those tests or examinations that are recommended to help curb disease severity or to follow/manage other disease manifestations. In the case of diabetes, this ranged from screening examinations for diabetes-related illness (i.e., neuropathy, nephropathy, or retinopathy) to prescription of appropriate medication regimens such as aspirin and angiotensin converting enzyme inhibitor use. The majority of the available quality of care outcomes were examined by a single study. Two studies (one trial and one observational study), both rated as fair, examined the frequency of patient receipt of recommended dilated eye examinations among adults with diabetes but identified discordant results.125, 138 While the trial was negative (no significant improvement in CM group),125 the observational study found improved regularity of eye examinations in the CM intervention group.138 Similarly, these same studies examined patterns of medication use and again the trial was negative and the observational study was positive (observed significantly more aspirin use in the CM arm).

Resource Utilization Outcomes

Resource utilization outcomes can include analysis of trends or frequency with which the examined population utilized the health care system. In the case of diabetes, CM might be expected to improve hospitalization rates, both via influence on glucose control (e.g., hypoglycemia, hyperglycemia) and via improvement in diabetes-related complications such as cardiovascular disease and peripheral neuropathy. CM could similarly influence ED visits, primary care provider appointments, and overall costs of health care. Four studies examined resource utilization outcomes.118, 120, 121, 125

Two trials,120, 123 both rated as fair quality, examined differences in rates of ED utilization and had opposing findings. One identified fewer ED visits in the CM intervention group, whereas the other found no significant difference in ED visits between groups. These two studies varied significantly by CM intervention strategy and patient populations (see Table 10).120, 123

Two trials123, 125 examined rates of hospitalizations. Despite some design differences between studies (see Table 11), results of these two studies for this outcome were both negative (no significant benefit of CM in decreasing rates of hospitalization). This conclusion was rated as having low strength of evidence, due to the small number of studies.

Effectiveness of Case Management by Patient Characteristics

Several studies examined specific patient subgroups of people with diabetes. By far the most common subpopulation examined was that of patients with type II diabetes (examined by 5 of 10 studies).118, 120-124, 126, 136 All five of these studies (four rated as fair and one rated as poor quality) examined HgA1c and BMI as outcomes. The results of these five studies, however, did not differ from the three studies that did not examine this patient subgroup.

Two trials limited their patient populations to urban, inner-city patient populations,120, 122 and two trials examined CM among African-American adults with diabetes.122-124 All three of these studies were rated as fair quality, and all examined BMI as an outcome. CM was not associated with improved BMI in any of these studies. The two clinical trials of African-American adults with diabetes122-124 also did not find an effect of CM on other physiologic outcomes (e.g., HgA1c, SBP, and HDL cholesterol). Two observational studies limited their populations to American Indians/Alaskan Natives with diabetes.137, 138 These studies both examined change in HgA1c but had discordant results. Wilson et al.138 observed significant improvement in HgA1c among individuals who received the CM intervention, while Curtis et al. did not. Two other patient populations (Latinos and adults living in under-served areas) were each examined by only one study.120, 128-130 Overall, there was minimal evidence suggesting that CM is more effective for improving outcomes for diabetes in any subpopulation.

Effectiveness of Case Management by Intervention Characteristics

Only one observational study evaluated different CM strategies head-to-head.137 This study included three intervention arms which differed in intensity: usual care by primary care provider alone (least intensive), primary care provider and nurse case manager combined intervention (intermediate intensity), and primary care provider and nurse case manager intervention which allowed case managers to alter medications (most intensive). Interpretation of results from this study was challenging because of the distribution of patients among the study arms. Although the total population for this study was large (n=2300), the vast majority (98 percent) of participants were in either the least intensive (usual care) or intermediate intensity CM arm, with only 60 patients included in the high intensity CM group. Because of the small sample size in the high intensity arm, the precision of the results is low. This study identified no benefit of CM for HgA1c between arms but did observe a statistically significant increase in hypoglycemic events in the most intensive arm (p=0.035). However, this is based upon a single hypoglycemic event in the most intensive arm.

CM strategies employed by the included studies overall were quite variable (see Table 10). There were no consistent similarities in CM strategies among trials with positive results. Only one trial128-130 reported results which consistently showed a benefit of CM in diabetes (to improve HgA1c, LDL, and blood pressure). That trial, rated as fair quality, was the only trial to utilize solely telephone interactions between case manager and participant. One other trial found a significant improvement in HgA1c with CM utilized a strategy of both face-to-face and telephone interactions by care managers. Of note, this trial only identified a positive result for HgA1c improvement but did not find that CM improved other health outcomes.126

Population: Patients With Cancer

The goals of CM for patients with cancer are generally to support and navigate patients through intensive and complex treatment regimens (e.g., surgery, chemotherapy, and radiation); to manage symptoms (e.g., pain, dyspnea, nausea, and fatigue) associated with cancer or its treatment; to maintain patients' physical, mental, and emotional well-being and independence in the context of serious illness and often debilitating treatment; and to help patients, families, and caregivers plan for and cope with the psychosocial and emotional burden imposed by the diagnosis, treatment, and prognosis of cancer. Many CM activities in the setting of cancer care overlap with other interventions such as hospice and palliative care services. What differentiates CM is that comprehensive care coordination is usually the primary focus, while hospice and palliative care interventions tend to focus primarily on symptom management.141

Description of Studies

We found six clinical trials of CM for patients with cancer (see Appendix I, Evidence Table 11). Of these, four were rated fair quality18, 142-144 and two poor quality145, 146 (see Appendix G). Sample sizes of the included trials ranged from 203 to 335 patients (total N for all studies = 1,406). The earliest included study was published in 1989 and the most recent in 2006. No observational studies were identified for this category of patients.

The populations in which CM interventions were tested varied substantially across studies. Two studies evaluated CM for patients with breast cancer,143, 146 two for patients with lung cancer,144, 145 and two for patients with a variety of cancer types.142 Patients ranged in age from 21 to 85 years old, although the mean ages reported by the majority of the studies ranged from 55 to 72 years old. One study recruited only patients meeting criteria for being homebound.145 None of the other five studies explicitly targeted patients with functional limitations or specified complex care needs beyond the vulnerability and complexity inherent in undergoing treatment for and coping with cancer. Some studies were conducted in patient populations with high levels of comorbidity144 or low socioeconomic status.142

CM interventions across the six studies shared some common elements but varied in both content and implementation (see Table 13). In all six studies, case managers performed a variety of functions, including developing management plans; addressing the psychosocial and emotional needs of patients and their families or caregivers; educating them about cancer and its treatment; assessing, monitoring, and treating symptoms; and coordinating care and making referrals. CM functions were deployed mainly through home visits, face-to-face encounters in a clinic setting, and telephone calls. The duration of CM interventions, as implemented in each trial, ranged from 3 months to 2 years. The intensity of CM also varied, from multifaceted and comprehensive CM that included home visits, telephone calls, and accompanying the patient to doctor visits143 to lighter interventions involving primarily telephone calls to evaluate and coordinate simple care needs.18 In most cases, case managers were nurses with specialized training in cancer care. Although protocols and care scripts were used in some interventions, case managers had the flexibility to individualize care according to specific patient needs in all studies.

Table 13. Characteristics of case management interventions for patients with cancer (randomized trials).

Table 13

Characteristics of case management interventions for patients with cancer (randomized trials).

The comparator group in most CM trials was described as usual, standard, or conventional medical care. In most studies, the nature of usual care was not explicitly described. One study of patients undergoing lung cancer treatment in the United Kingdom described usual care as outpatient visits in the post-treatment period and then at 2-3 month intervals.144 Another study of patients with lung cancer compared CM both with a “standard” home care intervention carried out by a multidisciplinary team (without a case manager) and with usual outpatient care.145

Targeted outcomes in CM trials (see Table 14) included health outcomes such as QOL, functional status, cancer-related symptoms, and survival;18, 143-146 patient and caregiver satisfaction with care;142-144 receipt of specific treatments and services considered to represent high-quality cancer care;142-144 utilization of resources not considered to represent high-quality care (e.g., hospitalization, ED visits);18, 144, 145 and the overall cost of care.142, 144, 146

Table 14. Characteristics and outcomes of studies of case management for patients with cancer (randomized trials).

Table 14

Characteristics and outcomes of studies of case management for patients with cancer (randomized trials).

The timing of CM interventions varied across studies. In four studies, case managers primarily supported and coordinated the care of patients undergoing treatment for cancer.18, 143, 144, 146 CM began before initial treatment in two studies143, 146 and after treatment in two studies.18, 144 The other two studies included trials that enrolled patients at different stages in the course of their illness and focused more generally on addressing patients' care needs related to cancer.142, 145

The settings for CM interventions included managed care organizations,142, 146 VAMCs,142 community hospitals and clinics,18, 143 home care organizations,142, 145 and cancer care centers.144 Five studies were conducted in the United States and one in the United Kingdom.144

Key Points Related to Patients With Cancer

  • CM is effective in improving selected cancer-related symptoms and functioning (physical, psychosocial, and emotional) but not overall QOL or survival (strength of evidence: low). (See Appendix H. Strength of Evidence.)
  • CM improves patient satisfaction with care (strength of evidence: moderate).
  • CM is effective in increasing the receipt of appropriate (i.e., guideline-recommended) cancer treatment (strength of evidence: moderate).
  • CM does not affect overall health care utilization or cost among cancer patients (strength of evidence: low).
  • Greater intervention intensity and duration, integration of CM with patients' usual care providers, and greater structuring of interventions through preintervention training and care protocols enhance the effectiveness of CM (strength of evidence: low).

Detailed Analysis: Effectiveness of Case Management by Outcome

Patient-Centered Outcomes
Quality of Life/Health Outcomes

Overall QOL and survival were generally not improved by CM in any of the studies that examined those outcomes.18, 144, 145 CM was effective, however, in improving outcomes that were directly targeted by the intervention. For instance, using an intervention intended to help women recover after breast cancer surgery, Goodwin et al. found that CM was effective in restoring normal ipsilateral arm function compared with usual care (93 vs. 84 percent).143

Similarly, two studies in which case managers provided symptom management and psychosocial support for patients with lung cancer demonstrated improvements in symptoms and psychosocial or emotional functioning.144, 145 However, in one of these studies, significant improvements were found in only three of 36 prespecified outcome measures,144 raising the possibility that the improvements resulted by chance rather than as a result of CM. Another study found no differences in symptoms or functional outcomes with CM.18 In one study, patients receiving CM had declining perceived health status over the course of the study, while control patients' perceived health status steadily improved,145 even in the presence of greater symptom distress and worse functioning. This seemingly contradictory finding may have indicated, as suggested by the authors, that education and monitoring by case managers instilled more realistic evaluations of health status among homebound patients with lung cancer.145 Due to the inconsistent findings and changes that were sometimes of small magnitude, the strength of evidence for the effect of CM on these outcomes was rated as low.

Patient Satisfaction

Of four studies that analyzed various aspects of patient experience with the care they received,18, 142-144 three found CM to be superior to usual care. Two studies found that CM increased patients' (and caregivers') satisfaction with care.142, 144 Another CM intervention improved breast cancer patients' sense of having a choice in their treatment.143 The fourth study showed no difference in perceived unmet needs18 among patients receiving CM compared with controls. A study examining the effect of CM on patients who had died found that CM increased the proportion of cancer patients dying at home rather than in an institution.144 Whether or not home deaths reflected patients' and families' preferences was not reported in this study. Due to the consistency of findings across three of these studies, the effect of CM on patient satisfaction was rated as having moderate strength.

Quality of Care Outcomes

Three studies examined the effect of CM on the use of health care services considered to represent high-quality care.142-144 All three found that CM improved the use of recommended services. An intervention specifically targeting the use of advanced directives succeeded in increasing the number of completed advanced directives.142 Other studies demonstrated increased use of use of breast-conserving surgery (with lymph node dissection and radiation treatment) for women with early-stage breast cancer143 and the early use of radiation as adjunctive therapy for lung cancer.144 The strength of evidence for this outcome was rated as moderate because of the consistent findings across the studies.

Resource Utilization Outcomes

Five studies examined the impact of CM on resource utilization (including hospitalizations, ED visits, medical visits, and testing) and overall cost of care18, 142, 144-146 and found no reduction in overall cost of care. One study found that CM reduced the number of radiographic studies patients underwent but did not affect referrals, hospitalization rates, or the overall cost of care.144 Other studies similarly demonstrated no difference between CM and controls in utilization of services.18, 142, 145, 146 In general, the estimated cost of the CM interventions was small. Thus, the cost of implementing CM had a minimal impact on the overall cost of care, which was driven mainly by the cost of hospitalizations. We rated the overall strength of evidence for these outcomes as low. Although there were consistently negative results across all the studies, the sample sizes were not large and the studies may not have been sufficiently powered for these outcomes.

Effectiveness of Case Management by Patient Characteristics

CM is a high-intensity intervention that is most often deployed for patients with complex care needs. While cancer and its treatment may in and of themselves create complex care needs, it is possible that the utility of CM is greatest among high-risk or vulnerable patient subgroups. In our review, only one study explicitly targeted a high-risk group (homebound patients with lung cancer). This study did not show a stronger effect of CM than other studies.145 Three studies evaluated whether measures of vulnerability or level of care needs predicted the success of CM within their study samples. In one study, patients were stratified into three groups based on a statistical model of predicted unmet needs.18 This study found no differences in any outcomes for any subgroups. Two other studies, however, both using CM for women undergoing treatment for breast cancer, found that CM was primarily effective in women with lower levels of social support, as indicated by being unmarried or living alone.143, 146 CM was most effective in this population of women in terms of ensuring use of appropriate services143 and improving QOL.146 Overall, these subgroup analyses were limited and had inconsistent results. Thus, there was insufficient evidence to draw conclusions about subgroups of patients with cancer.

Effectiveness of Case Management by Intervention Characteristics

No studies included head-to-head comparisons of different models of CM. One study compared a specialized home care CM program for cancer patients with a standard home care program delivered by a multidisciplinary team.145 In that study of homebound lung cancer patients, both home care programs produced similar outcomes in terms of symptoms, functional status, and hospitalization rates. However, the study was poor quality and did not clearly specify the differences in activities and functions performed by the different home care models.

The CM interventions described in the included studies varied widely in their implementation. We analyzed this variation in an attempt to discern the features of successful compared with unsuccessful CM interventions. Heterogeneity in the outcome measures used across studies precluded a quantitative analytic approach (e.g., meta-regression). Our findings therefore derived from a qualitative synthesis of the six included studies.

Two studies reported on interventions that demonstrated significant improvements in multiple outcomes, including health outcomes, patient experience, and quality and utilization of care.143, 144 Another intervention was successful in achieving more focused improvements in targeted outcomes, including patient satisfaction and advanced directive completion.142 There were several features that, while not unique to these successful interventions, in the aggregate appeared to distinguish them from others (Table 13). Specifically, the interventions reported by Goodwin et al. and Moore et al. represented more intensive forms of CM, in that they included more contacts and were sustained over a longer period of time than most others. They also explicitly included integration between the case managers and the patients' usual care providers. Finally, those interventions, as well as the one reported by Engelhardt et al., appeared to be more structured, as indicated by explicit descriptions of pre-intervention training for case managers and the use of care protocols to guide CM activities.

Aside from CM intensity, integration with primary care, and structure, we found no discernible pattern indicating that other aspects of CM—including modes of contact or principal CM functions—influenced effectiveness. It should be noted, however, that the specificity with which authors described the core functions performed by case managers was variable, which limited our ability to evaluate whether specific core functions influenced CM effectiveness. Likewise, no studies provided information on the average caseload of case managers at any given time, and only one explicitly reported the degree to which case managers received supervision from a physician. Most case managers were nurses, and most had flexibility for individualizing care plans, limiting our ability to comment on the value of these intervention components. Overall, because of the limited number of studies, the strength of evidence for the influence of intervention characteristics on patient outcomes (programs with higher intensity and better integration being associated with better quality/satisfaction outcomes) was rated as low.

Population: Patients With Serious Chronic Infections

HIV and mycobacterium tuberculosis (TB) are serious infectious agents that, when inadequately treated, can be fatal. Both require treatment with multiple drugs and for long durations of time. For HIV, treatment must be continued indefinitely. Both are infectious, and treatment can reduce the chance of transmitting the infection to others. Thus, effective treatment of these infections is a clinical and public health priority. Treatment requires engagement by the infected person and adherence to regimens that are complex and can be associated with unpleasant side effects. Problems such as low health literacy, unstable living situations, and substance abuse can be important barriers to carrying out treatment plans. For both of these infections, a variety of public health programs have been tried to improve medication adherence and thereby to enhance clinical outcomes.

Description of Studies

Studies of Case Management for People With HIV

We identified five randomized trials and six observational studies of CM programs for people infected with HIV (see Appendix I, Evidence Tables 12 and 13). Of the five randomized trials, two were rated fair quality149, 150 and three were rated poor151-153 (see Appendix G). Four of the trials were conducted in the United States and one in Canada.151 The earliest included study was published in 1992 and the most recent in 2007. Sample sizes ranged from 57 to 250 participants (total N = 736).

Of the six observational studies, one was rated good quality,154 two were rated fair,155, 156 and three were rated poor.157-159 All six observational studies were conducted in the United States, and the majority included a relatively small number of participants (sample sizes of 51, 78, 132, 280, 588, and 988). One study was published in 1991.155 The other five studies were published between 2000 and 2009.

All 11 of these studies targeted low income populations except for one observational study restricted to women.157 The majority of participants in each study were male (54 to 93 percent). In three of the studies the majority (70 percent or more) of participants were Caucasian.151, 153, 155 In six studies, 49 to 90 percent of participants were African American or Latino. One study evaluated CM services specific to American Indians, Alaska Natives, and Native Hawaiians.159 The average age of study participants was 35–45 years. One study targeted homeless and marginally housed individuals154 and three studies specifically included current intravenous drug use or other substance abuse as study eligibility criteria.149, 152, 157

Studies of Case Management for People With Tuberculosis

We identified two randomized trials (see Appendix I, Evidence Table 12), both of which were rated fair quality (see Appendix G). One was conducted in the United States and published in 2006160, 161 and one was conducted in Taiwan and published in 2007.162 Sample sizes, respectively, were 520 and 114 (three study arms). We also identified two observational studies (see Appendix I, Evidence Table 13). One was rated good quality163 and one was rated poor.164 Both studies utilized a retrospective cohort design and were of similar sample size (n=343 and n=369). One was conducted in the United States and published in 2002163 and one was conducted in Taiwan and published in 2006.164 One of the trials restricted enrollment to individuals with latent tuberculosis infection.160 The other three studies examined programs serving patients with active TB infection. The majority of participants in both U.S. studies were nonwhite and male; in one of the U.S. studies, more than 30 percent were substance abusers and more than 40 percent had concurrent infection with HIV; eligibility for the other U.S. study included spending the previous night in a homeless shelter. Participants in the Taiwan studies were mostly male with a mean age range of 53 to 68; socioeconomic status was not reported.

Approach to Case Management for Chronic Infections

CM interventions in all of the studies focused on linking individuals to needed services, including medical, mental health, social, and drug treatment services (see Table 15). The programs generally included counseling and education components. The TB programs tended to have a greater emphasis on the coordination and monitoring of medications. In one of the HIV programs,153 the participants were housebound patients with AIDS, and the case managers had caseloads of only 12 or less. Mode of case manager/client contact (reported in three studies) was either strictly face-to-face or in combination with telephone contact. The disciplines of the case managers were usually nurses or counselors. The length of the interventions was 6 months in all of the TB studies and 6 to 12 months in the HIV studies.

Table 15. Characteristics of case management interventions for patients with HIV/AIDS or tuberculosis (randomized trials).

Table 15

Characteristics of case management interventions for patients with HIV/AIDS or tuberculosis (randomized trials).

While one of the trials was a head-to-head comparison of less intensive to more intensive CM for TB treatment,162 the rest of the studies used a usual care comparison group. The control groups generally had access to all the same services as the intervention groups (community-based services, or usual clinic or in-home care), but acquired them through self-direction or without the assistance or involvement of a designated case manager.

The patient-centered outcomes included in these studies (see Table 16) often were measures of response to antibiotic treatment. All of the TB studies used measures of successful suppression of the infection. Two of the HIV studies150, 154 included viral load or CD4 count as outcome measures. Other patient-centered outcomes included measures of mental health, QOL, and risk behaviors. Quality of care outcomes included and medication adherence rates and receipt of community services. Resource utilization measures included outpatient and ED utilization, hospitalization rates, and overall program costs.

Table 16. Characteristics and outcomes of studies of case management for patients with HIV/AIDS or tuberculosis (randomized trials).

Table 16

Characteristics and outcomes of studies of case management for patients with HIV/AIDS or tuberculosis (randomized trials).

The settings for these CM programs included HIV/AIDS service organizations,151, 152, 159 public health clinics,150, 163 public hospitals,149, 156, 162, 164 and homeless shelters.160 All of these studies were conducted in large metropolitan areas.

Key Points Related to Patients With Serious Chronic Infections

  • CM does not improve survival among patients with HIV infection (strength of evidence: low). (See Appendix H. Strength of Evidence.)
  • Short-term CM management programs that emphasize medication adherence improve rates of successful treatment for TB in vulnerable populations (strength of evidence: moderate).
  • Evidence is insufficient to determine whether CM improves antiviral treatment of HIV infection.

Detailed Analysis: Effectiveness of Case Management by Outcome

Patient-Centered Outcomes

Two clinical trials of HIV patients included survival as a primary outcome. A fair quality trial149 reported 16 percent mortality at 18 months and a poor quality clinical trial of patients with AIDS reported 50 percent mortality at 6 months.153 Neither study found a significant difference in mortality between the CM and control groups.

Some clinical trials in HIV populations also measured psychological distress149, 151 and quality of well-being.153 Changes in these measures showed little difference between the CM and control groups. One fair quality observational study156 found that CM counseling on mental health issues had a positive relationship with client QOL, and one poor quality observational study found improvement in self-reported QOL after CM program enrollment.159 Due to the overall small changes found in these studies, this evidence was judged as insufficient to conclude whether CM affects measures of QOL in these populations.

Quality of Care Outcomes

The studies of populations with TB had CM programs in which the case manager emphasized adherence to drug treatment regimens, and these programs generally found higher rates of successful treatment with CM. The study with the best methodological quality was a good quality interrupted-time-series evaluation.163 Using a measure of achieving adequate treatment, a successful outcome was achieved for 69 percent of patients during the time period in which conventional directly observed therapy (DOT) was used. This rate increased from 81 percent to 86 percent in successive time periods in which CM was added to DOT. These rates stayed consistent over four successive 6-month time periods, suggesting that this finding was not due to a time trend unrelated to the use of CM. Higher rates of treatment completion with CM were also observed in two fair quality clinical trials of patients with TB.160, 162 A poor quality observational study compared a population of TB patients receiving CM with a population in a different health system. Treatment success was 87 percent in the CM group and 73 percent in the comparison group.164 Due to the consistently positive findings in these studies, we concluded that the overall strength of evidence is moderate that CM programs emphasizing medication adherence improve rates of successful treatment of TB.

In a fair quality clinical trial, HIV patients were randomized to CM, directly observed antiretroviral administration, or usual care.150 Viral load dropped in all three groups, without significant differences among the programs. The CM used in this trial was of moderately high intensity (weekly contacts by case manager) and was not found to bring significant improvement in self-reported medication adherence.150 In a good quality cohort study, the quantity of CM was used as a predictor variable in a multivariate analysis.154 The quantity of CM had a moderate association with rise in the CD4 count but was not associated with drops in viral load. Due to the small number of studies, the evidence is insufficient to conclude whether CM has an effect on the quality of treatment for HIV.

Other quality measures have included (for HIV patients) behaviors associated with viral transmission. CM has not been demonstrated to improve viral transmission behaviors.149, 152

Resource Utilization Outcomes

Because the studies in this clinical category often include vulnerable and underserved populations, the CM programs focus on facilitating and increasing provider visits. However, CM generally had little effect on the rate of clinic visits. While CM was associated with increased clinic visits in a poor quality observational study of HIV patients,157 the visit rates were not significantly changed in a fair quality clinical trial150, 165 and a good quality observational study.154 In a fair quality clinical trial of an HIV population, hospitalization rates were lower in the CM group than in the usual care group, but ED visits were not significantly different.165 However, in a good quality observational study, receiving CM was not associated with either ED or inpatient utilization.154 The study finding a reduction of hospitalizations also found lower overall health care costs in the CM group.165 A poor quality clinical trial151 also found lower (but not significant) overall costs in the CM group. Due to the small number of studies and inconsistent findings, the evidence about the effect of CM on measures of utilization was judged to be insufficient in these populations.

Effectiveness of Case Management by Patient Characteristics

Because all of the studies in this clinical category had relatively small sample sizes, there were few subgroup analyses. The influence of CM on patient outcomes applies only to the limited populations that were studied. As previously stated, this group of studies included mostly underserved and impoverished populations.

Effectiveness of Case Management by Intervention Characteristics

One fair quality clinical trial for TB patients had a head-to-head comparison of two levels of intensity of CM (weekly home visits vs. monthly home visits).162 The sample size was small (32 participants per study arm). The measure of treatment success was significantly higher in the group that received weekly visits. A good quality observational study of HIV patients154 measured intensity of CM by frequency of contact with case managers. However, the highest-frequency category could still be less often than monthly. The intensity of CM was evaluated for a large number of possible outcomes. The only outcome that showed a positive association with CM intensity was improvement in CD4 count. Due to the small number of studies and inconsistent results, the evidence was judged to be insufficient for drawing conclusions about variation by intervention characteristics.

Population: Patients With Other Medical Problems

CM can be adapted to a wide variety of community settings and clinical problems. While the clinical categories described earlier in this report captured most of the studies of CM, there were 15 additional studies that do not fall into those categories, nine trials28, 166-177 and six observational studies.178-183 These additional studies related to three care coordination themes. The first is coordinating services for low income individuals who often have serious problems with access to clinical services. The second theme is patient education and coordination of services following hospital discharge for acutely disabling medical conditions (stroke and renal failure requiring dialysis).The third theme is case management that focuses on self-care for patients with obstructive lung disease. In general, these studies had findings that were consistent with the results described earlier in this report.

Description of Studies

Of the nine randomized trials of CM programs for clinical populations different from those already described in this report (see Appendix I, Evidence Table 14), six were good quality,28, 166, 167, 168, 169, 170, 172, 173, 177 one was fair quality,174 and two were rated poor175, 176 (see Appendix G). Six were conducted in the United States,28, 166-170, 175, 176 two were conducted in Canada,172, 177 and the remaining trial was conducted in Hong Kong.174 These trials were published between 2002 and 2012. Of the observational studies (See Appendix I, Evidence Table 15) two were rated as having fair quality methods,180, 182 and the other four were rated as poor quality.178, 179, 181, 183 Six of the studies (two trials166, 169 and four observational studies178-181) examined low income populations, although the nature of the CM programs was quite variable across these studies. Four clinical trials examined patients undergoing home peritoneal dialysis174 or patients undergoing rehabilitation after a stroke.175-177 Three clinical trials28, 170, 172 and one observational study183 examined patients with chronic obstructive pulmonary disease (COPD), while one observational study examined adults with bronchial asthma.182

A good quality clinical trial evaluated a CM program for patients followed in primary care clinics operated by a county health department in California.166, 167 Patients were eligible for the study if they had DM, coronary artery disease, peripheral vascular disease, cerebrovascular disease, hypertension, or elevated cholesterol and/or triglyceride levels. Of the 419 participants, mean age was 56 years, 65 percent were female, 63 percent were Hispanic, and 38 percent were employed. Sixty-three percent had type 2 diabetes. In the intervention group, CM was performed by a team consisting of a registered nurse and a dietician. The case managers used protocols that focused on lifestyle modifications and the intervention lasted 15 months. The control group received the usual model of primary care provided in the four participating clinics. The primary outcome was a measure of risk factors for atherosclerosis (the Framingham risk score).

The second clinical trial that focused on a low income population examined homeless patients who were recruited at the time of an acute hospitalization in the United States.169 The participants had a wide variety of chronic medical conditions. The intervention included CM for up to 18 months, and the intervention group patients also were provided placement in stable housing. The control group received no specific services following hospital discharge, but there were other CM services available in the community. The outcomes were counts of hospitalizations and ED visits. This study design makes it difficult to discern the unique effects of CM, in that there was an important cointervention (placement in permanent housing) that was not available to the control group.

The fair quality observational study examined a group of patients followed in a California safety net clinic who had high rates of emergency department use or hospital stays over a one-year period.180 A group who were assigned to case management services was compared with a group who did not receive such services, but assignment to groups was not randomized or otherwise controlled. The non-CM group had a significantly lower comorbidity score than the CM group. The three poor quality observational studies were all conducted in the United States and evaluated CM programs for low-income people.178, 179, 181 The first evaluated 492 uninsured adults, 70 percent of whom were female (mean age 35 years).179 CM was provided by a team consisting of a registered nurse and social worker. There was no comparison group and the primary outcome was ED visit rates (measured 6 months prior to starting CM and 6 months after completing CM). Mean duration of CM was 179 days. The second observational study evaluated 159 patients who received services from a CM program designed for low-income patients with epilepsy.178 Mean age was 41 years, and 58 percent were male. Two-thirds were uninsured, 59 percent were unemployed, and none had yearly incomes greater than $5,000. Self-reported estimates of seizure control and ED visits were assessed by a questionnaire administered after completing the CM program. A third study using a pre-post design enrolled 53 patients who had used the ED five times or more in 12 months.181 Study subjects were assigned to a social worker case manager who was responsible for providing and coordinating all needed services. Hospital service utilization and cost and psychosocial outcomes, including homelessness and access to care, were measured at 12 months.

A fair quality clinical trial conducted in Hong Kong evaluated a 6-week CM program for patients who perform home peritoneal dialysis.174 The 85 study participants were recruited during an acute hospitalization. The outcome data were derived from patient questionnaires administered at 6 and 12 weeks after hospital discharge.

One good quality and two poor-quality clinical trials evaluated CM programs for patients who had recently undergone acute rehabilitation following a stroke. The good-quality trial was performed in Canada among 190 people hospitalized for an acute stroke. They were randomized at hospital discharge to a short-term (6-week) CM program or to a comparison group that received only instructions on how to make an appointment with a primary care provider.177 The case managers were nurses with geriatrics experience. The primary outcomes were physical functioning and healthcare utilization. Another randomized trial evaluated utilization outcomes of 28 stroke patients who were being discharged from an inpatient rehabilitation service in New York.175 The case managers were social workers, and the program focused on ameliorating barriers to ongoing rehabilitation. Control group patients received usual care without the services of the social workers. The outcomes were measures of utilization over 3 months. The third trial also enrolled patients (N=96) who were being discharged from an inpatient stroke unit.176 The case managers were advanced practice nurses. The CM focused on coordination between neurology consultants and the primary care physicians. Patients in the control group did not receive these coordination services. The outcome measures included functional status, QOL, and measures of stroke-related quality of medical care (all measured at 3 months after hospital discharge).

Three good quality clinical trials examined programs for patients with COPD. All three programs emphasized training patients in self management, including self-administration of medications (steroids and/or antibiotics) for acute exacerbations.28, 170, 172 One of the trials was conducted in Canada.172 The other two trials were conducted at Veteran Affairs (VA) medical centers in the United States.28, 170 Study participants in each trial had severe COPD, with mean values of the forced expiratory volume in one second (FEV1) ranging from 1.0 to 1.2 liters across the trials. Outcomes measured in all the trials included mortality, number of hospital admissions or time to first hospital admission, number of acute exacerbations, and other health status and quality of life measures.

All three of these trials emphasized self-management of symptoms. In the Canadian trial,172 eight hour-long educational sessions were conducted weekly in the intervention participant's home for the first two months. The case manager (either a nurse or respiratory therapist) made follow-up phone calls weekly during the 8-week educational period and then monthly for the remainder of the 1 year study. In one of the VA studies,170 intervention group patients received a single 1- to 1.5-hour educational session and monthly calls from a case manager. In the other VA study,28 the educational program consisted of 4 weekly 90-minute sessions and followup phone calls from the case manager once per month for 3 months and then every 3 months thereafter.

A poor quality observational study conducted in New Zealand evaluated 16 patients with severe COPD (mean FEV1 0.64 liters) who were enrolled in a program in which a registered nurse provided weekly telephone calls and monthly in-home visits.183 Hospitalization rates were compared with a control group of 16 patients followed at a different hospital.

A fair quality observational study evaluated a case management program for adults with a clinical diagnosis of bronchial asthma.182 Nurses conducted case management by telephone. Rates of unscheduled outpatient visits and hospitalizations over 24 months were compared with a baseline period.

Key Points Related to Other Populations

  • Evidence is insufficient to assess the effect of CM programs on mortality among patients with severe COPD. (See Appendix H. Strength of Evidence.)
  • CM programs that serve populations that have COPD or are homeless reduce ED visits (strength of evidence: low).

Detailed Analysis: Effectiveness of Case Management by Outcome

Patient-Centered Outcomes

In general the studies included in this category had short durations of followup. They also included diverse populations and used a variety of outcome measures. Of the good quality clinical trials, three (all in populations of patients with COPD) examined mortality rates. In two studies, cumulative mortality was measured at one year after enrollment.170, 172 In both the mortality rate was less than 12 percent and was slightly lower in the CM groups. The third trial was stopped early because of a finding of a higher mortality rate in the CM arm of the trial.28 With a mean follow-up of 250 days, mortality was 13 percent in the CM group and 5 percent in the control group. When cause of death was assessed, mortality attributable to COPD was also higher in the CM group. Due to the heterogeneity in mortality rates across these three trials, the overall evidence is insufficient to conclude whether CM affects mortality in the population of patients with severe COPD. These trials also included measures of symptom status and QOL, but there generally were only small changes in these measures.

One trial conducted in a low-income population measured a variety of cardiac risk factors.166, 167 In this trial the mean Framingham risk score was one point lower in the intervention group at 15 months. The major contributor to the difference between groups was better achievement of blood pressure goals in the intervention group. Because there no other similar studies, there is insufficient evidence from this single study to conclude whether CM is effective for this clinical goal.

Two of the three clinical trials of patients with recent strokes measured patient-centered outcomes. 176, 177 The good quality trial measured physical functioning and found no difference between the CM and control groups at 6 months of follow-up.177 The second study, which had poor methodological quality, had a small sample size and used multiple outcome measures, suggesting that some changes may have been due to chance. The study found small improvements in QOL in the CM group but no differences in functional status or blood pressure control.176

Studies in two other clinical settings also found improvements in patient-centered outcomes with CM. In the trial of CM for patients undergoing home peritoneal dialysis, patients in the CM group had small improvements in several measures of functioning and satisfaction compared with patients in the control group.174 The observational study of patients with seizures found a reduction in self-reported seizure rates.178 However, there was no comparison group in this study and it is possible that part of this change was due to regression to the mean. Due to the small number of studies, the evidence was judged insufficient to draw conclusions about the influence of CM on physical functioning or seizure rates.

Resource Utilization Outcomes

Many of the studies in this category reported on utilization of health care services, with ED visits being the most commonly measured type of utilization. Three of the good quality trials had ED visits as a primary outcome.169, 170, 172 Compared with the usual care group, homeless CM patients had, on average, about one fewer ED visit per year, but this group also received housing assistance in addition to CM. In a trial of patients with COPD, the group receiving CM had half as many ED visits over one year (0.21 visits/year in CM group vs. 0.42 visits/year in the comparison group).170 In another COPD trial ED visit rates were higher, but visits attributable to exacerbations of pulmonary symptoms were significantly lower in the CM group.172 A good quality trial of CM following stroke found no difference between CM and control groups in ED visit rates during the 6 weeks that patients received CM. However, in the following 6 months, 16percent of patients who had received CM and 23percent of control-group patients made ED visits.177 Two fair quality observational studies compared ED utilization rates for patients who received CM.180, 182 CM was associated with reductions in rates of ED use when compared with a group of patients matched by race and age180 and when compared with usage by the same patients in a historical time period.182 Four other studies that were rated as poor quality175, 178, 179, 181 also found lower ED visit rates in patient groups who received CM. Due to the consistency of findings across studies, it was concluded that there is a low strength of evidence that CM leads to fewer ED visits in these populations.

Five good quality trials of CM examined hospitalization rates as a utilization outcome. Three were studies of patients with COPD,28, 170, 172 one was a study of stroke patients, and one was a study of CM for homeless people.169 All three of the COPD trials found lower hospitalization rates in the groups receiving CM, with this result being statistically significant in two.170, 172 In the trial of CM conducted among homeless people, the hospitalization rates did not differ significantly between the CM and control groups, but patients in the CM group had about three fewer hospital days per year. However, this difference in length of hospital stays may be due to the housing assistance provided as a cointervention to the CM group.169 The trial of stroke patients did not find a significant effect of CM on hospitalization rates.177 A poor quality observational study of patients with COPD also found that a group receiving CM had shorter lengths of stay but no difference from a comparison group in the hospitalization rates.183 Due to the inconsistency of findings for hospitalization rates, the evidence was rated as insufficient for this outcome.

Effectiveness of Case Management by Patient Characteristics

Although four of the studies in this category166, 169, 178, 179 addressed CM for low income individuals, the populations were quite diverse, ranging from homeless people to patients who were followed regularly in safety net clinics. The outcome measures in these studies were diverse, and the only outcome that was measured in multiple studies was ED visits. This measure improved in all the studies, so the utilization outcome did not appear to be influenced by any particular patient characteristics. The other outcomes in these studies are different enough that it is not possible to draw conclusions based on patient subgroups. The studies of CM for COPD did not perform comparisons by sub-groups with differing severity of lung disease, although most participants in these trials had severe disease.28, 170, 172

Effectiveness of Case Management by Intervention Characteristics

The studies in this category tended to examine CM programs that were tailored to the patient populations (i.e., cardiac risk factor reduction, management of home dialysis, management of respiratory symptoms, or coordination of care for the uninsured) and the outcomes were specific to each type of program. The main difference that can be examined is length of CM. In the studies of CM for COPD or for low income people,28, 166, 169, 170, 172, 178, 179 the CM was continued for 6 to 18 months. In the four other studies (of home dialysis174 and stroke175, 176, 177) the CM lasted 3 months or less. Nevertheless, there were no clear trends in outcomes based on CM duration within these ranges.

PubReader format: click here to try

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (3.6M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...