| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
a Mandel School of Applied Social Sciences, Case Western Reserve University, Cleveland, Ohio
b Department of Sociology, University of Kansas, Lawrence
c College of Nursing, University of Maryland, Baltimore
d Department of Community Medicine, Wayne State University, Detroit, Michigan
Baila Miller, Mandel School of Applied Social Sciences, Case Western Reserve University, 1746 N. Larrabee, Chicago IL 60614 or 11235 Bellflower Road, Cleveland, OH 44106-7164 E-mail: bhm4{at}po.cwru.edu.
 |
Abstract |
|---|
 TOP Abstract Models of Social Support...MethodsResultsDiscussionReferences |
|---|
Methods. The authors analyzed 3 models specifying patterns of relationship between social support and depression (main effect, mediation effect, and moderation effect) separately within data sets using hierarchical ordinary least squares regression. Results were compared across data sets.
Results. The replication analysis confirmed the robustness of behavior problems and caregiver health as important contributors to caregiver distress. Results of hypotheses examining the pattern of relationship between social support and distress were inconsistent, however. Only 1 type of social support was associated with distress in the expected direction: Less emotional support was associated with higher levels of distress in 2 of the 4 data sets.
Discussion. More complex theoretical models that incorporate common measures to represent the linkages between types of stressor, types of support, and their interactions are needed to foster replicability and generalizability of research results.
THE stress process model of caregiver distress delineates the role of such resources as social support in buffering and/or mediating the effect of stressors on distress (
Lawton, Kleban, Moss, Rovine, and Glicksman 1989
;
Lazarus and Folkman 1984;
Pearlin, Mullan, Semple, and Skaff 1990). A plethora of empirical studies examining the stress-support-distress process using a variety of clinical and community populations have shown inconsistent results. Informal support is believed to lessen the negative consequences of caregiving (
George and Gwyther 1986;
Matthews, Werkner, and Delaney 1989;
Pearlin, et al. 1990), but the nonsupportive aspects of personal relationships have also been noted (
Barusch and Spaid 1989;
Bass, Tausig, and Noelker 1988–89). Informal support networks alleviate such negative aspects of caregiving as emotional distress, health concerns, and economic strain (
Clipp and George 1990;
Thompson, Futterman, Gallagher-Thompson, Rose, and Lovett 1993) but also demonstrate mixed, and sometimes unexpected, relationships to stressors and distress (
Franks and Stephens 1996). Which type of social support is most effective is also in dispute (
Aneshensel, Pearlin, Mullan, Zarit, and Whitlatch 1995;
Li, Seltzer, and Greenberg 1997;
Thompson et al. 1993). The results of the large number of caregiving support studies cannot be easily integrated because, in addition to different sample selection procedures, different measures of stressors, support, and distress outcomes have been used. As a result, although these studies provide breadth of knowledge, the likelihood of replication of results, a crucial element of scientific validation in the logical positivist framework of social research, is hampered.
One approach to distinguishing the effect of sampling differences from measurement variation is to develop and replicate a common set of measures across different samples of known characteristics. We report the results of a cross-sectional secondary analysis of four caregiving community-based studies representative of many conducted in the past 10 years. These studies contained many common measures of selected stressors, support resources, and caregiver distress. Two of the studies examined caregivers of persons with dementia and two examined caregivers of older persons with a range of heterogeneous frailties. Although the data sets were developed for other purposes, they shared sufficient common indicators for us to control for conceptual and operational equivalence of measures. Our goal was to identify areas of consistency and inconsistency in findings across the data sets controlling, as many studies cannot, for variation in definition of measures.
|
Models of Social Support and Caregiver Distress |
|---|
|
TOPAbstractModels of Social Support...MethodsResultsDiscussionReferences |
|---|
;
Stull, Kosloski, and Kercher 1994). We interpreted distress as a reflection of negative consequences among caregivers who continue to provide care in the community, using depressive symptomatology as an outcome indicator of a mental health consequence of caregiving.
Stressors are experiences or events that create a state of arousal and that motivate persons to cope with behavioral demands and with the emotional reactions that are evoked by them (
Lazarus and Folkman 1984). The mechanisms that link stressors to distress outcomes are difficult to specify. Some stressors may affect distress directly, whereas others have primarily indirect effects, depending on the mobilization of support and other resources. Research has suggested that increases in functional impairments are associated with increased distress among caregivers of the physically impaired, whereas behavior problems are more consistent predictors of distress among caregivers of persons with dementia (
Miller and McFall 1991;
Pruchno and Resch 1989). Our underlying assumption was that the stressors of behavior problems, functional limitations of the care receiver, and caregiver health would be associated with higher levels of depressive symptomatology, controlling for the demographic and caregiver context variables often associated with depressive symptomatology (e.g., gender, race, income, years caregiving, and caregiver relationship).
Our study focused on a key resource for caregivers, that is, the social support they receive from their informal network of family, friends, and neighbors and the formal network of paid service providers. Social support is a metaconstruct representing the instrumental, emotional, and informational assistance provided to individuals by others (see
House, Umberson, and Landis 1988;
Sarason, Pierce, and Sarason 1990;
Thoits 1995, for summaries). Because social support is a multidimensional construct, we included two types of informal support (emotional support and instrumental support) and one type of formal support (instrumental support) in this study. Informal emotional and instrumental functions of support differ in degree and purpose. Informal and formal support share the ability to provide concrete or instrumental aid, to assist in evaluating problems, to help in formulating and implementing plans of action, to give feedback, and to affirm a person's capacity to deal with a situation (
Caplan 1981;
Krause 1987). Their main difference is that formal support is governed by contractual or paid arrangements rather than the affiliation and obligation norms that represent provision of informal care (
Litwak 1985).
Theoretical interpretations of the mechanisms by which social support influences the distress process vary. Social relationships are believed to enhance the feelings of self-worth, self-esteem, and the sense of well-being that comes from feeling valued by meaningful others and to aid in the resolution of designated problems or losses (
Antonucci 1990;
Cutrona and Russell 1990;
Krause, Liang, and Yatomi 1989;
Thoits 1995). Many researchers have suggested that the links between stressors and support may be strongest in their effects when there is a match between the benefits provided by specific types of supportive behaviors from specific sources and the needs created by particular events in an older person's life (
Cohen and McKay 1984;
Cutrona and Russell 1990;
Krause 1990). Studies in the caregiving literature illustrate the relevance of matching between type of support and type of stressor and/or type of caregiver burden/distress outcome (
Clipp and George 1990;
Cohen and McKay 1984;
Cutrona and Russell 1990;
Hays et al. 1998;
Li et al. 1997). But little consensus has developed about which are the most important linkages. In this study, we examined linkages between different stressors and different sources and types of support in their relationship with caregiver distress.
Three models of the ways in which social support influences distress levels guided our choice of hypotheses (
Cohen and Wills 1985). Main effect models propose that there is a relationship between social support and caregiver distress regardless of the stressor condition. Each type of support may have differential effects on distress but in general can be expected to have a negative relationship with caregiver distress. Hypothesis 1 stated that higher levels of informal emotional support would be associated with lower distress, because perceived emotional support is associated with higher self-worth and a sense of acceptance by others, and thus is effective in reducing emotional distress (
Krause and Borawski-Clark 1994;
Sarason et al. 1990). Hypothesis 2 stated that informal instrumental support may lower caregiver distress through the concrete aid and respite provided to the caregiver by family and friends. The association between formal support and caregiver distress is less easy to predict. The few studies that have examined the main effects of formal support on well-being (
Krause 1990) or caregiver distress (
Aneshensel et al. 1995) identified contradictory results. In accordance with the general stress model, and similar to the assistance provided by instrumental support, Hypothesis 3 stated that formal support may also lower caregiver distress by the provision of aid and respite.
Moderating or buffering models assume that social support changes the way in which caregivers appraise the seriousness of the care receivers' impairment and caregiver activities, or that support is effective mainly under conditions of high levels of stressors (i.e., social support affects the direction and/or strength of the relationship between stressors and caregiver distress;
Baron and Kenny 1986). The relatively few studies of the moderating effects of informal and formal social support in the caregiving literature have identified study-specific, nonuniform effects of social support that depend on different sources or different types of support at different points at the stress process (
Aneshensel et al. 1995;
Bass, Noelker, and Rechlin 1996;
Li et al. 1997). We did not propose any specific moderation hypotheses at the outset of the study given the ambiguity in the literature, but we systematically assessed the likelihood of combinations of stressors and support moderating caregiver distress.
Whereas moderator models specify when certain effects of social support will occur, mediating models examine how or why such effects occur (
Baron and Kenny 1986). The effect of stressors on mobilization of support has been discussed in terms of the formation and content of helper networks (
Cantor 1983;
Litwak 1985;
Stone, Cafferata, and Sangl 1987) or substitution, supplementation, and linking mechanisms between formal and informal supports in changing contexts of need (
Cantor 1983;
Chappell and Blandford 1991;
Greene 1983;
Miner 1995;
Noelker and Bass 1989). As the number of functional limitations increases, caregivers may seek to increase their support networks and turn to increased contact with family and friends or use of formal services (
George 1996;
Miller and McFall 1991;
Stoller and Pugliesi 1988). We thus hypothesized (Hypothesis 4) that higher levels of stressors would lead to mobilization of formal and informal support (i.e., a positive relationship between stressors and support) and, in turn for reasons noted in the discussion of the main effect model, a negative relationship between support and distress.
In sum, we proposed four hypotheses testing potential main and mediating effects of the relationship between stressors, social support, and distress and systematically examined the moderation model. By assessing these patterns of relationship across four data sets, we assessed consistency and replication of findings.
|
Methods |
|---|
|
TOPAbstractModels of Social Support...MethodsResultsDiscussionReferences |
|---|
|
Respite care study (RCS).
This sample (n = 412) was drawn from Study NIMH-RO1 MH45840, Targeting Respite to Promote Mental Health of Alzheimer's Families, 1991–94 (Rhonda J. V. Montgomery, principal investigator). The original purpose of this study was to identify factors that influence initial respite use and respite use over time. The nonprobability sample of caregivers was recruited from referrals from agencies providing respite or professionals who routinely come in contact with Alzheimer's disease patients. Caregivers were eligible if they were providing primary care for a person with dementia (unspecified) and thus were eligible for respite services. The sampling strategy selected users of 26 county-based respite programs in Michigan and nonusers of respite services, that is, those who had inquired about services but not yet used them or caregivers who had never inquired about such services. Caregivers in this sample were 79% female, 12% African American, and 57% spouses; had a mean age of 63 and a mean educational level of high school graduate; and were caregivers for an average of 6 years. The average age of care receivers was 77, and 92% resided with their caregivers.
Health care study (HCS).
This sample (n = 293) was drawn from Study NINR-RO1 3381, Caregivers Formal/Informal Service Use, 1992–96 (May Wykle, principal investigator). The purpose of this study was to examine differences by race in access to formal services and attitudes about the mix of formal and informal service use. A probability sample of primary caregivers in northeast Ohio was identified by random digit dialing. Caregivers were eligible if they provided a least 5 hr of care per week to the care recipient and lived either in the same household or within 1 hr's travel. Caregivers in this sample were 84% female, 35% African American, and 10% spouses; had a mean age of 52; and had an average of 5 years caregiving. The average age of care receivers was 79, and 48% resided with their caregivers.
Elder care study (ECS).
This sample (n = 155) was drawn from Study NIA R29 AG07587, Caring for Elders: Impact of Social Support and Burden, 1988–93 (Donald Stull, principal investigator). The original purpose of this study was to investigate diversity of family caregiving arrangements and the extent to which these change over time. The nonprobability sample of primary caregiving families was recruited through newspaper and radio ads, notices in doctors' offices, church bulletins, and referrals from the local Area Agency of Aging in northeast Ohio. Families were eligible to participate if the older person was aged 65 years and older, residing in the community, and receiving help with at least one activity of daily living (ADL) from a family member. Caregivers in this sample were 85% female, 4% African American, and 7% spouses; had a mean age of 54 and a mean educational level of high school graduate; and had an average of 7 years caregiving. The average age of care receivers was 82, and 62% resided with their caregivers.
The data set samples varied on key dimensions. The dementia samples, ACS and RCS, had substantial proportions of spouse caregivers, and thus a larger proportion of resident caregivers, whereas the heterogeneous samples, ECS and HCS, primarily comprised adult child caregivers. Data set distribution of gender was confounded with caregiver relationship, but all samples were largely female. ACS and HCS had approximately one third African American caregivers compared with much smaller proportions in the other two data sets. As with most caregiving samples, respondents in all four data sets were relatively well educated.
Description of Measures
In Table 1 , we present a brief description of the measures, means, and standard deviations by data set. Control measures were selected on the basis of their association with caregiver distress in previous literature (
Schulz et al. 1995). The demographic control measures were straightforward and will not be described further. Rather, we focus on the conceptual and operational description of measures of stressors and support. Most of these were identical (common) across data sets, but a few were unique to a data set, sharing conceptual but not operational definitions.
Stressors common to all data sets.
Functional limitations refer to the care receiver's ability to independently perform basic personal care activities and was operationalized by a combined summary index of 10 ADL and instrumental activities of daily living (IADL) items (i.e., bathing, toileting, dressing, transferring, eating, using the telephone, food shopping, preparing meals, managing money, and doing light housework;
Katz et al. 1963;
Lawton and Brody 1969;
Spector, Katz, Murphy, and Fulton 1987). Caregivers' physical health was measured by a single item that asked caregivers to give a subjective appraisal of their overall physical health status (compared in ACS, ECS, and RCS to others of their age group and in ACS and RCS to others of their gender).
Stressors unique to each data set.
Cognitive impairment refers to difficulties that a care receiver may demonstrate with memory, judgment, and problem solving. In RCS and ACS, a summary index was computed on the basis of the caregiver's assessment of the care receiver's cognitive functioning in eight areas (remembering recent events, knowing the day of the week, remembering the home address, remembering words, understanding simple instructions, finding his or her way around the house, speaking sentences, and recognizing people). In HCS, a summary index was computed of three dichotomous items (problems with his or her memory, periods where his or her judgment seemed impaired, or difficulty solving day-to-day problems). In ECS, only one item was available that asked caregivers (0 = no and 1 = yes) if their care receivers had dementia. Behavior problems refer to the care receiver's disruptive behaviors. In ACS, caregivers were asked how many days in the past week care receivers had kept them up at night, repeated questions, dressed the wrong way, had bowel or bladder accidents, hidden belongings and then forgotten about them, cried easily, gotten depressed or downhearted, clung to the caregiver or followed him or her around, become restless or agitated, sworn or used foul language, become suspicious or believed someone was going to do him or her harm, threatened people, or shown sexual behavior at inappropriate times or places. For purposes of this analysis, a summary scale was computed on the basis of the total number of days a caregiver spent dealing with these problems. Behavior problems were measured in RCS in a manner similar to ACS, with an additional item asking caregivers about irritable or angry behaviors on the part of the care receiver. In HCS, only one item asked caregivers if their care receivers demonstrated behavior problems affecting the care receiver's safety or that of the family. ECS did not include a measure of behavior problems.
Support measures common to all data sets.
Informal instrumental support refers to the amount of help a caregiver received from family and/or friends with specific instrumental tasks (e.g., household chores, shopping, or running errands) that are associated with caregiving. A three-item summary scale was computed that reflected the number of types of instrumental supports received (i.e., help with money, transportation, or meals). Alpha reliabilities were .91 (ACS), .76 (RCS), .52 (HCS), and .69 (ECS). Formal support refers to the amount of assistance a caregiver received from community service providers. A four-item summary scale was computed that was based on a count of the number of services (i.e., meals on wheels, home health aides or visiting nurses, chore or homemaker services, and short-term nursing home respite or use of adult day care) a caregiver had used during the previous 6 months.
Support measures unique to each data set.
Informal emo-tional support refers to the affective support provided to the caregiver by family and friends through companionship and comfort. In ACS, this concept was operationalized by asking caregivers how often in the preceding 6 months the caregiver had confided in someone, been comforted by someone, or been offered assistance (
= .66). In RCS, informal emotional support was operationalized by asking caregivers how often in the preceding 4 months someone had comforted them, told them they were OK, showed physical affection, offered assistance, or told the caregiver that he or she was loved ( = .88). In HCS, informal emotional support was operationalized by asking caregivers whether family and friends had provided companionship, listened to the caregiver's problems, or given the caregiver advice on dealing with life's problems or ways to care for the care receiver ( = .64). In ECS, only one dichotomous item was comparable to the informal emotional supports operationalized in the other data set: Caregivers were asked if they had confided in or talked with someone within the prior 6 months about their problems, including concerns about caregiving.
Caregiver distress.
Caregiver distress, operationalized as depressive symptomatology, reflects the psychological distress experienced by the caregiver. The Center for Epidemiologic Studies-Depression Scale (CES-D;
Radloff 1977) was used in all four data sets. Because of the present study's focus on social support, however, two items in the CES-D that dealt with negative interpersonal interactions were dropped from the analysis. A summary scale was computed from the remaining 18 CES-D items. Scores above 16 indicate probable clinical depression on the original 20-item CES-D. As noted in Table 1 , the average depressive symptomatology score was below the clinical cutoff in all the samples, but respondents in the two dementia samples (ACS and RCS) had higher levels of depressive symptomatology than those in the heterogeneous samples (ECS and HCS).
Analytic Strategy
The proportion of missing data on predictors was relatively small with the exception of income. We used the EM algorithm within the SPSS Missing Values Analysis package to impute data on all predictor variables with missing data but deleted cases missing on our dependent variable, caregiver depressive symptomatology. The Missing Completely at Random assumption (
Little and Rubin 1987) was not met in all instances (details available from first author), but differences in estimates between analyses based on listwise deletion and imputed models were minor.
Our strategy was to carry out all analyses separately within data sets, comparing results across data sets for consistency. We examined the four hypotheses in two stages. We first examined ordinary least squares (OLS) regression models using only those measures that were common to all data sets (i.e., omitting measures of behavior problems, cognitive impairment, and emotional support). We then performed the same analyses including the measures that were conceptually similar, but measured uniquely in each data set. Because of the similarity in results, we present only the results of the final model containing common and unique variables. (Common variable results are available from the first author.)
|
Results |
|---|
|
TOPAbstractModels of Social Support...MethodsResultsDiscussionReferences |
|---|
Table 2 presents the OLS regression results of caregiver distress on caregiving stressors controlling for demographic and caregiver context. Similar to findings from other caregiving studies (
Schulz et al. 1995), the consistent stressors contributing to higher levels of distress in all the data sets were caregiver's self-rating of health and care receiver's behavior problems. Among the demographic and contextual variables, African Americans had lower distress levels in three data sets (ACS, RCS, and HCS); women and those with lower incomes had higher distress levels in two data sets (ACS and HCS); and spouses had lower distress levels in one data set (ACS). Table 2 addresses the first three hypotheses, assessing the relationship between each type of social support and caregiver distress, controlling for demographic and caregiver context. The association between each type of support and distress varied by data set. In ACS, less informal instrumental support and greater use of formal support were associated with higher levels of caregiver distress. Lower levels of emotional support were significantly associated with higher levels of distress in RCS and HCS.
|
To evaluate the moderating role of social support, we centered the stressor and support variables at their respective means. Multiplicative interaction effects between these variables then were added to the hierarchical regressions and systematically cycled through a series of hierarchical regressions. Because of limited power, only two-way interactions were considered, and the moderation analyses were performed in three sets between four stressors and three types of support. One set included interaction between each type of support and the two significant stressors, caregiver health and care recipient behavior problems; a separate set included interactions between the remaining combinations of support and cognitive impairment and functional limitations. A final set included all interactions. The moderating role of social support was not supported in any of these hierarchical analyses, because the additional variance explained by each new set was not significant and the individual coefficients of each interaction effect were not significant (results available from first author).
Hypothesis 4, which suggested a mediating role for social support, therefore was not supported. The regression analyses indicated that the assumptions of the mediation model of associations between stressors and support and between support and distress were not met (results available from the first author). As shown in Table 2 , there were few significant patterns of association between stressors and our measures of support. Different types of support were associated with caregiver distress in different data sets. For example, in ACS informal instrumental support and use of formal services were linked to caregiver distress, whereas in HCS informal instrumental support and informal emotional support were associated with distress, and in RCS only informal emotional support had a relationship to caregiver distress. ECS showed no significant bivariate associations between support measures and caregiver distress. The bivariate results were confirmed in the multivariate analyses (Table 2 ).
|
Discussion |
|---|
|
TOPAbstractModels of Social Support...MethodsResultsDiscussionReferences |
|---|
). In ACS, less informal instrumental support and greater use of formal support are associated with higher levels of distress. In addition, there are few common patterns differentiating the two samples of caregivers of persons with dementia from the more heterogeneous samples. The lack of consistent findings across the data sets with respect to the associations between social support and caregiver distress suggests that similar concepts and common measures may produce mixed results, challenging some of our assumptions about the role of social support in the caregiver distress process. A major limitation of this study is the use of cross-sectional data to represent processes that occur over time. The lack of support for the moderation and mediation models of social support may be due to both measurement issues and theoretical concerns. In our effort to identify and use common measures, we had to rely on simplified measures (some with low reliabilities) to represent complex concepts, thus potentially attenuating the correlations. In addition, insufficient variation among the support measures may have constrained the estimates of association. Overall, the caregivers in these samples did report relatively low levels of support. It would have been difficult for us to develop more complex measures in some of the data sets because the measures differed from one another and would not have represented the same latent dimensions of stressors and support. We suspect that many of the measurement difficulties we encountered represent ambiguous theoretical interpretations of social support constructs. We do not yet know the most relevant dimensions and the best measurement strategies needed to understand the role of social support in the caregiving and psychological distress process.
Our results support the argument that an important aspect in understanding the causal processes linking stressors, support, and distress is the specificity of the relationship between these elements of the support process. The specificity hypothesis of social support suggests that types of social support that do not match the defined need of the caregiver may not be highly valued and thus will have little effect in reducing depressive symptoms (
Cohen and McKay 1984;
Cutrona and Russell 1990;
Krause 1987). But this principle leaves unanswered the many complexities that may influence specificity of match between supportive benefits and needs. For example, the timing of support during the course of a caregiving career may be a key factor in reducing distress (
Jacobson 1986;
Pearlin, Aneshensel, Mullan, and Whitlatch 1995). Similarly, the proportion or intensity of negative interactions with others may influence psychological responses to stressors even more than supportive interactions (
Pagel, Erdly, and Becker 1987;
Rivera, Rose, Futterman, Lovett, and Gallagher-Thompson 1991). Our measures of social support in the caregiving literature rarely capture what is communicated in specific observable social interactions (
Coyne, Ellard, and Smith 1990).
In addition, the robust relationship between behavior problems and distress and caregiver health and distress suggests the possibility that, because of their intractability, these stressors are not easily amenable to the influence of various types of social support. For example, poor caregiver health can lead to restricted activities and social relationships, which in turn may be associated with higher levels of caregiver distress. Similarly, problem behaviors of the older person may make it difficult for the caregiver to leave the care recipient with other informal supporters for a sufficient amount of time to make a difference in levels of distress. Clearly, more complex theoretical models and longitudinal analyses that incorporate linkages between types of stressors, types of support, and their interaction are needed to further knowledge about the interrelationships between stressors, social support, and caregiver distress. Yet, as this research shows, greater replicability of measurements across studies would also help clarify the generalizability of findings.
|
Acknowledgments |
|---|
Received for publication April 18, 2000. Accepted for publication February 5, 2001.
|
References |
|---|
|
TOPAbstractModels of Social Support...MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. L. Douglas, B. J. Daly, C. G. Kelley, E. O'Toole, and H. Montenegro Impact of a Disease Management Program Upon Caregivers of Chronically Critically Ill Patients Chest, December 1, 2005; 128(6): 3925 - 3936. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. B. Adams, K. A. Smyth, and M. J. McClendon Psychosocial Resources as Moderators of the Impact of Spousal Dementia Caregiving on Depression Journal of Applied Gerontology, November 1, 2005; 24(5): 475 - 489. [Abstract] [PDF]
|
|
|
|
 |
|
 W. L. Collins, T. A. Holt, S. E. Moore, and L. K. Bledsoe Long-distance caregiving: A case study of an African-American family American Journal of Alzheimer's Disease and Other Dementias, September 1, 2003; 18(5): 309 - 316. [Abstract] [PDF]
|
|
|
|
|
|
S. L. Douglas and B. J. Daly Caregivers of Long-term Ventilator Patients: Physical and Psychological Outcomes Chest, April 1, 2003; 123(4): 1073 - 1081. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
