| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
COMMUNITY-BASED SERVICES |
a Center for the Study of Aging and Human Development, Duke University
Correspondence: Kirsten Corazzini-Gomez, Duke University School of Nursing, Box 3322 DUMC, Durham, NC 27710. E-mail: gomez006{at}mc.duke.edu.
Decision Editor: Laurence G. Branch, PhD
 |
Abstract |
|---|
 TOP Abstract Conceptual ModelMethodResultsDiscussionReferences |
|---|
Key Words: Case management decision making • Client care preferences
Book Reviews
Practice Concepts
The Forum
Case managers in the home- and community-based care arena are charged with applying professional judgment to assess clients through the performance of a set of core tasks, including targeting and identifying clients, determining client service need, tailoring service plans to client need, and monitoring clients across time with pertinent follow-up activities (Davies 1992
). Case managers in the publicly funded service arena are often in the position of serving as gatekeepers for a diverse array of community care services. Consequently, case managers have the potential to substantially affect the lives of community-residing frail elders (Kane 1995). Thus, accurate measurement of how case management judgment shapes the assessment process is vital to understanding whether or not scarce home care resources are allocated in an efficient and effective manner. Although considerable previous research has identified characteristics of either the elderly client or the case manager that affect care plan decisions, such research has largely failed to consider the dynamic process of the case manager and elderly client interaction on care plan decisions.
More recent research has begun to address this deficit by considering the client as an active participant in the care-planning process (Degenholtz, Kane, Kane, and Finch 1999; Degenholtz, Kane, and Kivnick 1997; Diwan 1999; Kane 1995). Diwan, Berger, and Manns 1997 argued that the ultimate care plan outcome represents a negotiated outcome between the case manager and the client. Degenholtz and associates 1999 and Kane 2000 articulated the need to consider client preferences in care plan development, such as the level of independence desired or the level of personal risk an elder is willing to accept. Furthermore, Degenholtz and associates 1997 have piloted a means of training case managers to include client care preferences into the care-planning process in a more standardized way.
Qualitative research findings explore the relevance of aspects of the interpersonal dynamic between case manager and elderly client, such as the client's personality or the initial impressions of the client held by the case manager. Although these additional characteristics are not desirable to include in the assessment process in a standardized way, they appear to shape final care plan decisions by the case manager (Miles 1993). Clemens and Hayes 1997, for example, discussed how the initial face-to-face assessment conducted by case managers often leads to a summary impression that is used to predict client behavior. Importantly, this impression is quite resistant to change or to the addition of further information and is not entirely based on actual assessment data. Similarly, Hagan-Hennessy 1987 also presented evidence of case managers who use a dominant label to drive care plan decision making (e.g., "squeaky wheel" clients; p. 149). Corazzini 2000 showed that case managers use nonstandardized cues of client need to support or refute client responses to standardized assessment protocols. For example, case managers may use their perceptions of the monetary value of the client's home furnishings to support or refute client statements of financial status.
By more comprehensively modeling the "black box" of the assessment process as a dynamic between elderly individual and case manager, we can consider the effects of the elder's agency on the care-planning process. Importantly, we can determine if significant biases arise in this dynamic in a systematic, quantitatively measurable manner to ultimately systematically affect home care resource allocation and the implementation of home care policy initiatives. This study asks the following research question: "Are there nonstandardized cues of elders' need arising from the interpersonal dynamic that are integrated into care plan decisions in a quantitatively measurable, systematic manner?" Nonstandardized cues are defined as client characteristics that are salient during the interpersonal dynamic between case manager and elderly client but that are not readily linked to actual care needs. These cues may not be linked to care needs because of a paucity of widely used standardized assessment tools (e.g., client preferences or values) or because such cues capture elements of client personality or demeanor.
|
Conceptual Model |
|---|
|
TOPAbstractConceptual ModelMethodResultsDiscussionReferences |
|---|
|
; Challis 1993; Degenholtz 1997; Hagan-Hennessy 1993), pathology (Degenholtz 1997; Hagan-Hennessy 1993), and cognitive impairment (Degenholtz 1997; Ginther, Webber, Fox, and Miller 1993; Hagan-Hennessy 1987). Increased total ADL/IADL deficits, increased severity of pathology, and increased cognitive impairments have all significantly predicted increased case manager–determined care plan.
Considering the second domain, client resources, strong levels of informal support have been related to lower levels of formal care (Hagan-Hennessy 1987, Hagan-Hennessy 1993), and home care clients with unstable informal caregiving networks have been more likely to receive increased care plan levels (Ginther et al. 1993). Decreased income has been related to increased likelihood of receiving publicly funded home care services (Caro and Leventhal Stern 1995). Therefore, Research Hypotheses 1 and 2 acknowledge the direct effects of standardized client cues on care plan as follows:
Within the third domain, client descriptors, quantitative research findings have primarily included sociodemographic descriptors. Examples include age (Degenholtz 1997; Yee 1990) and ethnicity (Chumbler, Dobbs-Kepper, Beverly, and Beck 2000; Yee 1990). Increased age has been related to less case management time and lower care plan cost (Yee 1990) as well as to an increased probability of case management recommendation of institutionalization (Degenholtz 1997). Ethnic minority status has significantly predicted increased care plan cost (Yee 1990) and decreased likelihood of receiving respite services (Chumbler et al. 2000).
Regarding nonsociodemographic descriptors of central importance to the research question, only Degenholtz and colleagues 1999 have presented quantitative findings relating client receptivity to services to care plan decision. Specifically, willingness to have family help, to have paid home care, or to relocate to a nursing home all significantly predicted whether case managers recommended nursing home placement (Degenholtz et al. 1999). Therefore, the third research hypothesis draws on Degenholtz and colleagues 1999 and previously cited qualitative research to test the main research question,
Case manager background characteristics previously identified in the literature include sociodemographic measures of age (Yee 1990), education (Capitman, Haskins, and Bernstein 1986; Yee 1990), and ethnicity (Chumbler et al. 2000); work history measures of length of employment (Yee 1990); and variants in core task arrangement, such as specialization (Capitman et al. 1986; Degenholtz et al. 1999) and caseload size (Degenholtz 1997; Yee 1990). Increased age, for example, has been related to increased diversity in care plan service mix (Yee 1990). Yee related increased education level to increased overall care plan cost. Chumbler et al. demonstrated a relationship between Caucasian ethnic origin and assessing fewer ADL deficits in clients relative to African American ethnic origin. Specialization in the performance of core case management tasks has been related to such aspects as increased diversity in care plan service mix (Yee 1990) and to increased likelihood of recommending institutionalization (Degenholtz et al. 1999). The dynamic model of the assessment process proposed in this study integrates the interactive effects of case manager background characteristics with nonstandardized elder characteristics central to the research question. Therefore, the fourth research hypothesis is as follows:
|
Method |
|---|
|
TOPAbstractConceptual ModelMethodResultsDiscussionReferences |
|---|
factorial vignette survey methodology. Surveys were administered in accordance with Dillman 1978, Dillman 2000 total design method for administering mail surveys.
Sample
The sampling frame included the total census of case managers employed by 26 of the 27 regional nonprofit agencies administering the MHCP (N = 507). A previous study of the MHCP (Yee 1990) suggested that case managers are typically young, White women, for whom this position represents the first job in human services received on graduation from a 4-year college. The MHCP provides a range of home care services for low-income elders in Massachusetts who have impairments in ADLs and/or IADLs, using funds obtained directly from state revenues. Case managers are responsible for between 80 to just over 100 clients each and are expected to meet their clients' needs within specific budgetary guidelines, usually about $200 per client per month. A typical service package includes 3 hr of homemaker services per week, plus home-delivered meals (Kent 1995).
Measures
Home Care Clients
The central component of the survey instrument included factorial survey vignettes (Rossi and Nock 1982) describing potential home care clients for whom the respondent was asked to assess care plan eligibility levels. Fig. 2 presents a sample vignette client. Factorial survey vignettes are a unique method of constructing vignettes. Typical vignette surveys are characterized by the researcher's providing the identical set of vignettes to each respondent. By contrast, factorial survey vignettes are constructed by randomly sampling from the entire matrix of possible prespecified components of the vignettes. For the home care clients, 20 client characteristics (or dimensions) were specified to capture physical and psychological functioning, resources, and descriptors (see Table 1 ). Within each dimension, several levels were specified. For every home care client vignette, one level was randomly selected for each dimension. The number of home care client vignettes, V, from sampling 20 dimensions, D, is expressed by the following equation, V = D1 * D2 * ... D20, where Di (i = 1 ... 20) is the number of levels for each characteristic. Each of the 507 sampled case managers rated six vignettes that were randomly, independently sampled from the entire matrix of specified client characteristics (N = 3,042 sampled vignette clients). This method has several distinct advantages, including the ability to control for the multicollinearity problem typical in real-life vignettes, where combinations of dimensions may commonly occur, and the ability to include a wide range of dimensions and levels (Rossi and Nock 1982).
|
|
Client resources were defined to include informal care and formal care currently received, living arrangements, marital and parental status, housing type and neighborhood, and insurance status. Informal care was operationalized with a single vignette dimension describing currently received informal care. Formal care received was captured by a dimension stating the current status and history of formal care. Marital and parental status were each operationalized with a single dimension describing both the presence and the capacity of a spouse or an adult child to assist with informal care. The housing type vignette dimension described the type of housing within which the elder currently resides. The neighborhood vignette dimension stated the socioeconomic status of the neighborhood. Last, the dimension capturing insurance status described whether the elder was enrolled in Medicaid or did not have any health insurance.
Client descriptors were defined as client presentation and demeanor. Presentation was operationalized by vignette dimensions describing the client's age, ethnicity, immigration status (both the length of residency in the United States and the ability to speak English), and sex. Finally, client demeanor was captured by two vignette dimensions. The first dimension described whether the client was receptive to help from the home care program and the second described the elder's attitude toward the case manager.
Care Plan Decision
After reading each vignette client, the case manager was asked to determine care plan eligibility level as measured on a 5-point scale (M = 2.07, SD = 1.01) of care plan eligibility level in relation to the average home care plan, ranging from 0 (not at all eligible for services) to 4 (eligible for much more than the average home care plan). Previous evidence suggests the persistence of a typical care plan prescribed by case manager to clients, irrespective of actual case manager–ascribed flexibility in care plan development and variability in extant services (Kane 1995). Kent 1995 presented evidence of a similar, typical care plan package assigned to clients by case managers in the MHCP across agencies.
Case Manager Characteristics
Additional close-ended survey questions captured information on case manager background characteristics. Questions were organized within three areas: core task variants, training and education, and demographics. Core task variants were defined to include whether the case manager was a specialist (e.g., only conducts intake assessments) and whether the case manager had a caseload size greater than the median caseload size. Training and education were defined to include whether the case manager received at least a 4-year college degree, whether the case manager was licensed as a social worker, and the number of years the case manager was employed by the home care agency. The area of demographics was defined to include the age, ethnicity, and sex of the case manager. An additional item tapped into overall satisfaction with work.
Analyses
The dependent variable was treated as having an underlying continuum with an interval level of measurement. Thus, the distribution of the dependent variable was examined for normality. Measures of age, parental status, cleanliness of house, hygiene, nutritional status, mobility, and informal care were recoded into single measures owing to the underlying continuums integrated in these variables during vignette dimension development. Case manager characteristics were examined for adequate discriminatory ability, and items with greater than a 90/10 split were dropped from the analyses (DeVellis 1991).
Ordinary least squares regression was conducted with a robust regression equation adjustment to test the effects of client and case manager characteristics on care plan decision. Although the argument has been made that the random selection of vignette characteristics allows for independent counting of each rated vignette as a separate case (Rossi and Nock 1982), a more conservative approach was taken by using a Huber correction (STATA 1999) to recalculate more accurate standard errors. Data were clustered by case manager, and an iterative weighting process adjusted for the response distribution attributable to nonindependence of observations. Five models were estimated hierarchically by entering blocks of predictors into the model in accordance with the four sets of predictors in the conceptual model and interaction terms. First, elderly client physical and psychological functioning were regressed on care plan to test Hypothesis 1. Second, physical and psychological functioning and client resources were regressed on care plan to test Hypothesis 2. Third, client descriptors were added to the model, testing Hypothesis 3. Fourth, case manager characteristics were added. Last, interaction terms were included to test Hypothesis 4. F tests tested if the inclusion of each new set of predictors significantly decreased total unexplained variance in care plan. Model robustness was assessed by examination of plots of the residuals. Importantly, model results were compared with an ordinal logistic regression model to refute concerns regarding the treatment of the measure of care plan eligibility level as having an underlying continuum.
|
Results |
|---|
|
TOPAbstractConceptual ModelMethodResultsDiscussionReferences |
|---|
), 89% of the case managers completing the survey were female, 91% were White, and the mean age was 39 years (SD = 13.1). Almost all (89%) had at least a 4-year college degree, 23% were licensed as social workers, and the mean number of years of employment with the MHCP was less than 5 years. Although no data were collected on nonrespondents to assess response bias, respondents from agencies with lower levels of completion rates were more likely to have worked for fewer years for the agency (p < .05), to have had lower caseload sizes (p < .05), and to be younger (p < .10).
Model Estimation
All five models of the robust regression analysis of client and case manager characteristics on care plan are presented in Table 2 . All five models were significant (p < .001), and the final model explained 18.5% of the variance in care plan eligibility determinations. Each successive block of predictors significantly improved (p < .05) on our ability to explain variance in the care plan eligibility level assigned to vignette clients. Examination of the variance inflation factors of the coefficients did not indicate problems with multicollinearity. In addition, the final model sample size allowed for more than twice the recommended minimum number of cases per predictor of 15 (Stevens 1996), for adequate power to test the research hypothesis. Comparison of sets of predictors in the final robust regression model with the ordinal logistic regression model indicates comparable significance and direction of influence, supporting the ordinary least squares treatment of the dependent variable as having an underlying continuum.
|
Contrasted with dimensions capturing physical and psychological functioning, fewer than half of the client resource measures were significant when entered in Model 2 and remained insignificant even when client descriptors, case manager characteristics, and interaction terms were entered in Models 3–5 providing mixed support of Hypothesis 2. Current levels of informal care did not significantly predict care plan eligibility level, with the exception of marginal significance (p < .10) when client descriptors were entered into the model. Current levels of formal care, however, were consistently important across models in the case management decision-making process. When presented with information about a client currently receiving home health care services (distinct from the state-funded home care services allocated by the respondents), case managers assessed lower care plan eligibility levels, on average. When presented with information about clients having had home health care services recently terminated, case managers assessed more generous care plan eligibility levels. This effect persisted regardless of whether the reason for termination was noncompliance or having been classified as no longer having a skilled need. The effect of termination due to no longer having a skilled need increased in magnitude and significance when controlling for case manager characteristics and interaction effects of social work licensure, demeanor, and attitude.
Other client resource dimensions significant in the second model included marital status and housing status. Relative to clients who were married with a spouse in good health, clients who had never married, who had a spouse in poor health, or who were widowed were assessed more generous care plans. The standardized coefficients highlight a case manager–perceived progression of need from least to greatest. These effects persisted when descriptors, case manager characteristics, and interaction terms were entered into the model. For housing status, only living in an apartment predicted increased care plan eligibility relative to living in a single-family home in the second model. However, this relationship was explained by the relationships between client descriptors, case manager characteristics, interaction terms, and care plan level and was therefore no longer significant in the final, full model. Parental status, socioeconomic status of the neighborhood, and insurance status did not significantly predict care plan eligibility level.
Client descriptors, when entered as a block in Model 3, showed no significant, direct effects on care plan eligibility level, with one very important exception. One measure of client demeanor, receptivity to services, had an effect that remained in the full model, providing limited support for Hypothesis 3. Clients who denied needing any home care assistance and who refused services were assessed care plan eligibility levels significantly lower on average than clients who were appreciative and passive service recipients.
Model Results: Case Manager Characteristics and Interactive Effects of Case Manager Characteristics With Client Characteristics
Of the case manager characteristics entered into the model in Model 4, sex and education level significantly predicted care plan eligibility level. Male case managers allocated more generous care plans, on average. Having at least a 4-year college degree predicted an increased care plan eligibility level relative to having less than a 4-year college degree. In the final model, when interaction terms were entered, however, the significance of these effects decreased to only p < .10. Interestingly, social work licensure was not significant when entered in Model 4, but it did have a strong, significant effect when interaction terms with social work licensure were entered in the final model. When controlling for the differential effects of demeanor and attitude by social work licensure status, case managers licensed as social workers assessed more generous care plans.
Two of the four interaction terms between social work licensure and elderly client attitude and demeanor were significant when entered into the final model, in support of Hypothesis 4. Both interactions indicated the significant effect of social work licensure on the case management care plan eligibility judgment process. The first interaction term of social work licensure with whether a client denied needing help indicated that case managers who are social workers assess less generous care plans to elders denying needing help relative to case managers who are not social workers. The second interaction term suggests these same case managers, when assessing elders who are polite but are reluctant to answer questions implying ADL deficits, assign less generous care plans relative to their non–social worker peers.
|
Discussion |
|---|
|
TOPAbstractConceptual ModelMethodResultsDiscussionReferences |
|---|
). Observed physical and psychological indicators of ADL and IADL performance, primary medical diagnosis, and cognitive impairment, as well as information about marital status and living arrangements, figured prominently in the care plan eligibility determination process. Moreover, the differential strengths of particular predictors may readily be explained as in direct accordance with MHCP agency directives concurrent at the time of survey data collection. One of the strongest predictors of a more generous care plan, for example, was inadequate client nutrition. Indeed, at the time of data collection, the MHCP had implemented a special initiative to identify and treat malnourishment in home care clients. These findings illustrate case management response to in-home signs of inadequate nutrition that are in direct accordance with agency policy. A second example may be found in the predictive ability of signs of cognitive impairment. Case managers responded with increased care plan eligibility to assist clients with signs of cognitive impairment relative to cognitively intact clients. This finding suggests that resource allocation patterns are congruent with initiatives to increase service provider awareness of home care service needs of cognitively impaired elders who reside in the community, either alone or with caregivers.
Findings also indicate that MHCP case managers have developed systematic patterns of response to client characteristics that may only approximate agency directives. For example, undergoing rehabilitation and having a terminal illness significantly predicted increased care plans relative to having a primary diagnosis of chronic illness. How MHCP case managers respond to information about a client's diagnosis of rehabilitative, terminal, or chronic illness reflects front-line perceptions about the relative responsibility of the MHCP in providing care. Consider shifts in Medicare home health care funding in providing rehabilitative care services and the resultant service demands placed on state-level home care programs. Case managers in this sample who assessed higher care plans to rehabilitation clients illustrate a systematic pattern of response that may only approximate agency intentions to provide modest home help to frail, chronically ill elders. An additional example is found in the client characteristics of mental health difficulties, including substance abuse and bipolar disorder. In this instance, MHCP case managers systematically allocated increased care plans to these clients, establishing a front-line MHCP scope of responsibility for mental health in frail elders that may only approximate agency intent.
Of most importance to the central research question of this study, there was only one significant predictor among client descriptors, including sociodemographic measures of race or immigration status or measures of attitude to case manager. Thus, our findings of limited support for the third research hypothesis do not indicate widespread inclusion of nonstandardized client characteristics in systematic resource allocations representing overt violations of agency intent.
However, this limited ability to identify significant predictors that have been documented in the qualitative literature (Clemens and Hayes 1997; Corazzini 2000) in support of the third research hypothesis may simply reflect an important limitation of study—that of the use of vignette clients. On the one hand, the methodological advantages of using factorial vignettes allow the researcher to test the independent effects of multiple characteristics and avoid problems of multicollinearity among client characteristics commonly occurring together in real life. On the other hand, manipulated client characteristics are readily identified by respondents, introducing measurement error through social desirability. A possible alternative for additional research may be to construct computerized video clients, where the methodological advantages of the factorial vignette are retained but a more lifelike simulation of the assessment process is possible. The significance of the F test for the block of client descriptors provides additional support that, despite methodological limitations, these characteristics remain important in the client–case manager dynamic and ultimately affect service allocation decisions.
Although the majority of the nonstandardized descriptors did not significantly predict care plan eligibility level, one critical exception was the client's receptivity to help. Elders who denied needing assistance from the home care program were assigned less generous care plans relative to those elders who were grateful for MHCP assistance. The respect for client autonomy in stating care preferences is a critical development in home care policy and reflects a substantial body of literature on ethical issues in home care (Kane 1995). Thus, at one level this finding suggests important case management implementation of respect for client care preferences. The interactive effects of social work licensure with client denial of services as tested in the final research hypothesis yield additional support for this perspective. Case managers licensed as social workers provided, on average, more generous care plans to clients. When assessing clients who denied needing services, however, social worker case managers assessed less generous care plans than their non–social worker case manager peers. If reducing services for clients who deny needing help is indeed a reflection of a more client-centered approach to care plan development, we would expect that social worker case managers would respond in this manner owing to the client-centered orientation central in the social worker professional code of conduct.
However, at another level, the implementation of respect for client autonomy is complex. For example, what responsibility does the case manager have to assist clients who do not desire assistance but who are cognitively impaired or who show signs of self-neglect? The findings of this study suggest that case managers, on average, assess less generous care plans to clients denying the need for assistance, controlling for all other information included in the vignettes. In other words, the case management response to denial persists even with cognitive impairment or measures of severe functional limitations.
The ability to better understand why case managers respond to client denial of services by reducing care plan eligibility levels and why social work licensure augments this effect is beyond the scope of the vignettes included in this study. For example, no information is provided to the case manager about the referral source for the initial assessment. One might expect that elders who deny needing home care services do not initiate contact to the home care program for an assessment. Rather, referrals may be initiated by family members who themselves are in need of home care services. Further research in this area to refine our understanding of case management response to client receptivity to services, and in the context of social work licensure in particular, is needed.
These findings demonstrate the utility of moving toward a more dynamic conceptual model of the home care assessment process that explicitly integrates standardized and nonstandardized client cues. Although the majority of explained variance in care plan was attributed to standardized cues of client functioning and resources, one nonstandardized client descriptor, receptivity to services, significantly predicted care plan level in direct and interactive ways. This finding points to the vital need for research positing a means of moving toward more valid and reliable tools to assess client care preferences, such as the work of Degenholtz and colleagues 1997. More broadly, these findings indicate the importance of addressing the ethical complexities of the assessment and care-planning process in the home care arena.
|
Acknowledgments |
|---|
I wish to thank Francis G. Caro for critical reading of the manuscript.
Received for publication September 27, 2001. Accepted for publication May 14, 2002.
|
References |
|---|
|
TOPAbstractConceptual ModelMethodResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  K. D. Fraser and C. Estabrooks What Factors Influence Case Managers' Resource Allocation Decisions? A Systematic Review of the Literature Med Decis Making, June 1, 2008; 28(3): 394 - 410. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
