Religious Elderly,
Among Harold
and
Ill Men
M.D., M.Sc., Harvey J. Cohen, M.D., Dan G. Blazer, M.D., D.P.H., Keith G. Meador, M.D., M.P.H., Frank Sheip, M.D., Veeraindar Goli, M.D., and Bob DiPasquale, M.S.
G. Koenig, Carl Pieper,
Objective: cal inpatients,
Coping and Depression Hospitalized Medically
The investigators the characteristics
depression.
Method:
The
Ph.D.,
examined the frequency ofreligious coping among older mediof those who use it, and the relation between this behavior subjects
were
850
men
aged
65 years
and
over,
without
psychi-
atric diagnoses, who were consecutively admitted to the medical or neurological services of a southern Veterans Administration medical center. Religious coping was assessed with a threeitem index. Depressive symptoms were assessed by self-rating (the Geriatric Depression Scale) and observer five patients used to cope
rating (the Hamilton reported that religious with illness. Variables
Rating Scale for Depression). Results: One out of every thought and/or activity was the most important strategy that were associated with religious coping included black
race, older age, being retired, religious affiliation, alcohol use, a prior history ofpsychiatric problems, sive symptoms were inversely related to religious other sociodemographic and health correlates ated during their subsequent hospitaladmissions was
the
only
sions:
These
related (Am
J
baseline
variable
findings
suggest
to depression Psychiatry
that that
predicted religious
in hospitalized elderly 1992; 149:1693-1700)
sometimes
despair.
During
physical
illness
coping resources are seriously tested and frequently overwhelmed, as evidenced by the fact that 40% or more of older hospitalized patients experience some form of clinical depression (1, 2). Religious beliefs and behavior, in turn, are prevalent among older persons (3,
Received
March
July
26, 1991; revision From the Department Psychiatry, and the Center
I 9, 1 992.
Geriatric
received Feb. 5, 1992; accepted of Psychiatry, the Division of for the Study of Aging and Hu-
man Development, Research Education
Duke University Medical Center, and the Geriatric and Clinical Center, VA Medical Center, Dur-
ham,
reprint
N.C.
Address
requests
to Dr.
Koenig,
GRECC,
Box
I 82,
VA Medical Center, 508 Fulton St., Durham, NC 27705. Supported by the Mellon Foundation, Sandoz Pharmaceutical Corporation, grant AG-00371 from the Center for the Study ofAging and Human Development, Duke University Medical Center, and NIMH grant MH-40159.
Am
J
Psychiatry
I 49:1
2, December
1992
When months
lower
scores
depression is a common
support, infrequent functioning. Depreswhich persisted after
202 men were later, religious at follow-up.
behavior
that
reevalucoping Conclu-
is inversely
men.
edical illness that precipitates hospitalization is a stressful experience that can interrupt social and work routines, drain finances, separate families, reverse canegiver roles, and create situations of forced dependency. Physical illness also brings with it the threat of pain and long-term disability, disfiguration, the prospect of approaching death, and feelings of existential and
were controlled. an average of6 coping
M
anxiety
high level of social and higher cognitive coping, an association
4) and are reported to serve as a coping strategy to help manage emotional distress (5-8). The extent to which older medical inpatients use religion for this purpose, the characteristics of those who do so, and the effectiveness of this strategy are largely unknown. National samples of Americans of all ages indicate that persons who find personal comfort and support from religion are more likely to be older, female, black, less educated, widowed, employed in manual or unskilled occupations, more economically deprived, and affiliated with conservative Protestant religious denominations (3). These sociological findings suggest that persons with fewer health, social, and financial resources, when facing situations over which they have little control (such as acute hospitalization), might turn to religion for solace. Whether religious beliefs and behavior actually help to prevent or relieve emotional distress is far from clear. Psychiatric illness may be even more common among the religious (9-11), perhaps pnedisposing them to greater problems in later life when they are faced with the stress of physical illness and/on approaching death. Thus fan, systematic research on the relation between religious coping and depression among olden adults in
1693
RELIGIOUS
COPING
AND
DEPRESSION
clinical settings has been limited. Previous studies have been hampered by small sample sizes, less than rigorous sampling methods, and unequal sex distribution (few males) (6-8). This report emanates from the Durham Veterans Administration (VA) Mental Health Survey (12, 13), a cross-sectional and longitudinal epidemiologic study of depression among hospitalized medically ill men. We examine religious coping, its sociodemographic and health correlates, and its relation to depression both cross-sectionally and over time. Four major questions guided this investigation. 1 ) How common is religious coping in this population? 2) Is this coping behavior more likely among those with fewer socioeconomic, physical, on mental health resources? 3) When other factors are controlled, is there an association between religious coping and depression? 4) If such a relationship exists, is it especially strong in any subgroup of the population, such as those with more severe medical illness, low social support, prior mental health problems, on other attributes?
METhOD Sept. 1 , I 987, and Jan. 1 , 1 989, all men aged over who had been admitted to medical or neurological services at the Durham VA Medical Centen were screened for depression. For inclusion in the study, patients were required to scone 15 on higher on the Mini-Mental State examination (14) and to be physically capable of undergoing a psychiatric evaluation. Psychiatric patients were excluded. Patients were evaluated within 48-72 hours of admission by a social worker and/or a Fellow in geriatric medicine (H.G.K.). Data were collected on demographic characteristics and social and economic resources, including age, race, education, prior occupation (IS), retirement status, current living situation, marital status, yearly income, and social support. Social support was measured with a three-item index that explored size of the support network (16), frequency of interaction, and perceived adequacy of support ( 1 7) (values for each item ranged from 1 to 5; Cronbach’s alpha=0.57). Over 40 religious denominations were represented in the study group; these were categorized into nine general religious groups according to a schema provided by Roof and McKinney (18). The physical health of the participants, including medical diagnoses, functional status, and cognitive status, was assessed. Functional status was determined by measuring physical (19) and instrumental (20) activities of daily living. Physical activities of daily living included bathing, dressing, toileting, problems with incontinence, transfer from bed to chair or vice versa, and feeding (each rated 0-2); instrumental activities of daily Between
65 years
living
and
involved
ability
to travel,
shop,
prepare
meals,
do
housework, and handle finances (rated 0 or 1). These two types of ratings together produced an 1 1 -item scale with a possible range of scores from 0 (low functioning)
1694
to 17 (high functioning). As we have mentioned, cognitive status was measured with Mini-Mental State examination scores, which ranged from 15 to 30. The assessment of mental health included information on alcohol use, prior psychiatric problems, and family history of psychiatric problems. Depression was assessed by self-rated and observer-rated scales. The self-rated 30-item Geriatric Depression Scale (21 ) was administered by the social worker to all patients; this scale has been validated as a measure of depression in older medical inpatients (22). The men aged 70 years and olden were also assessed by one of us (H.G.K.) with the Hamilton Rating Scale for Depression (23), a measure used in other studies of the elderly (24) and medical inpatients (25). Depression scales were administered early in the interview, before the assessment of religious coping. Religious coping was assessed with a three-item index. Each item measured how much the patient relied upon religion to help manage the emotional stress associated with his illness. In item 1, the patient was asked an open-ended question about how he coped. This item was chosen in order to identify the coping behavior that the patient himself felt was most helpful, and the question was asked before questions on religion that might bias
responses.
If the
response
was
religious
in nature
(e.g., faith in God on Jesus, prayer, church), the patient received a scone of 1 0; if the spontaneous response was not religious (e.g., stay busy, family support), a score of 0 was assigned. A value of 10 for religious responses was chosen in order to give this item equal weight with the others in the index. In item 2, patients were asked to rate on a visual analog scale the extent to which they found religious beliefs or activities helpful in coping with their situation. The scale was numbered from 0 (“not much on not at all”) to 10 (“the most important thing that keeps me going”). In approaching patients with the rating scale, the investigators allowed patients to define for themselves the meaning of the term “religion” but made clear that this could involve personal belief alone or include neligious activity such as prayer or church attendance. In item 3, the interviewer rated the patient on a scale of 0-10 on the basis of an overall assessment of how much the patient used religion to cope. This judgment was based on the patient’s further elaboration on religious coping themes during answers to items 1 and 2 and on a separate discussion about how religion was helpful. The scores from the three items were then summed, and an index with values ranging from 0 to 30 was obtained. Cronbach’s alpha (reliability) for the index in the overall sample (N=850) was acceptable (0.82). Test-retest/internaten reliability for the religious coping index was also determined for a subgroup of 188 consecutively admitted men. The religious coping index was administered twice to these patients, each time by a different rater; ratings were separated by 12-36 hours. The Pearson correlation between scores obtamed on the religious coping index at the two administrations was 0.81. The interraten agreement for the ob-
Am
J
Psychiatry
149:12,
December
1992
KOENIG,
server-rated religious coping item (item 3) was surpnisingly high (Pearson’s n=0.87) given that the raters came from markedly different religious backgrounds (secular humanist versus conservative Protestant). All participants in the baseline study who were readmitted to the medical or neurological services during the 16-month study period and S months thereafter (designated “time 2”) were reevaluated by the social worker with the Geriatric Depression Scale and the religious
coping
index.
Again,
psychiatric
admissions
were excluded. If more than one neadmission occurred, the results from the last interview during the 21-month observation period was used for time 2. This provided information on change in depressive symptoms and religious coping oven time among patients whose medical illness prompted rehospitalization during the project. Analyses were performed with the SAS statistical package (26). Simple statistics were used to determine the frequency of spontaneously reported, self-rated, and observer-rated religious coping. Bivaniate relations of covaniates with religious coping (religious coping indcx score) and depression (Geriatric Depression Scale and Hamilton depression scale scores) were examined with Pearson correlations. Hierarchical regression was used to examine the strength of relationships, controlling for the effects of other covaniates. To determine whether patients who used religious coping had fewer socioeconomic or health resources than other patients, the religious coping index scone was regressed on a series of 16 demographic, socioeconomic, and health variables (excluding depression). A backward stepwise regression method was used to eliminate nonsignificant variables (p>O.OS ); missing values were dealt with by listwise deletion. The regression
was
performed
in
five
stages.
In
the
first
coping
in the
following
manner.
First,
a model
for depressive symptoms was constructed by regressing it on the 15 sociodemographic and health variables (cxcluding religious variables). Once a final model had been obtained, we added religious coping and religious affiliation variables. Interactions between religious coping and all other variables in the model were tested and included in the final model if alpha was less than or equal to 0.05. Regression models were developed for both self-rated symptoms (Geriatric Depression Scale) and observer-rated symptoms (Hamilton Rating Scale for Depression). We analyzed the longitudinal data as follows. Pearson correlations determined bivaniate relations between baseline (time 1 ) group characteristics and follow-up (time 2) Geriatric Depression Scale score. Hierarchical regression was then used to assess the relation between time 1 religious coping index score and time 2 Geriatric Depression
Am
J
Psychiatry
I 49:1
2, December
1992
Cha racteristi
and Health
Mean
Characteristic
BLAZER,
SD
ET AL.
cs of 85 0 EldN
%
Demographic Age (years) Black race Education (years) Unskilled occupation
69.8
4.9 240
28.3
333
39.2
589
69.3
577 160
67.9 18.8
History of psychiatric problems
216
25.7
Family history problems Alcoholuse
153
18.2
186
22.1
49
14.7
Retired
more
8.9
than
3.8
S years
Social/economic Married Livingalone
Income (dollars per year) Social support ratinga Mental health
Geriatric
76
Depression
score
disease
Neurological
disease disease
Renal disease Cardiovascular Other Functional status daily living)’ Cognitive status
Mental
State
of possible
9.1
Scale
Cancer Gastrointestinal
aRange
3,316 1.8
of psychiatric
score11 Hamilton depression 15 Physical health Medical diagnosis
Respiratory
8,582 10.6
disease (activities
188 125
22.1 14.7
I 22 87 40 216 72
14.4 10.2 4.7 25.4 8.5
of
(Miniscore)C
14.1
4.0
26.4
2.8
scores=3-1S.
bRange
of possible
scores=0-1
cRange
of possible
scores=0-30.
7.
stage,
demographic variables (age, race, and occupational and retirement status) were entered into the model. In subsequent stages, religious affiliation, socioeconomic resources, and mental and physical health variables were added. The fifth stage produced a final model that contamed the significant correlates. We examined the relation between depression and religious
TABLE 1. Sociodemographic erly Male Medical Inpatients
COHEN,
Scale score, controlling Scale score and other
for time significant
I Geriatric Depression baseline correlates.
RESULTS There were I , 1 1 0 consecutively admitted new patients during the study period. Eight hundred fifty men (77%) underwent comprehensive social, psychological, and physical health examinations; 260 did not participate because of advanced dementia or delirium (12% with Mini-Mental State scores less than IS), communication problems (2%), refusal or discharge before being seen (5%), or other reasons (4%). Nonparticipants were more likely to be older, black, and residents of nursing homes and to have diagnoses of neurological or respiratory illness. The sociodemognaphic and health characteristics of the subjects are presented in table I . The distributions of race, marital status, and medical diagnoses were similar to those among elderly male patients discharged from VA hospitals in District 8, which covers most of North Carolina and parts of Virginia, Kentucky, Tennessee, and South Carolina (12). Table 2 presents the
I 695
RELIGIOUS
COPING
AND
DEPRESSION
TABLE 2. Distribution of Religious Affiliations Among 850 Elderly Male Medical Inpatients, Elderly Men in Central North Carolina, and in the United States as a Whole Percent Percent Elderly
Religious
Male Medical Inpatients
Groupa
Central North Carolina’
DepressionL Percent in the
U.S. Populationa
8.0
7.1
8.7
Lutheran, of Christ,
Reformed) Conservative Protestant (White Baptist, Church Christ,
of
Elderly Men in
Nazarene,
12.5
17.2
19.9
Seventh-
40.8
S4.7’
22.2
I 1.3 5.8
5.1
4.5
Protestant (unspecified) Catholic Jewish Nontraditional Christian (Mormon, Jehovah’s Witnesses, Christian Science, Unitarian)
3.3 2.6 0.0
5.9 1.5 0.3
25.0 2.3
3.2
3.3
8.0
No
2.4
5.1
6.9
bMen
preference
Roofand
aged
demiological Clncludes
McKinney
55 years
(18).
and over who
Catchment conservative
Area Protestants
participated
study and
(central black
in the Piedmont North
Carolina)
Epi(27).
Protestants.
religious affiliations of the participants. The distnibution was comparable to that of elderly men living in central North Carolina; compared to the population of the nation as a whole, however, a disproportionate number of men came from conservative on black Protestant denominations (63% in this study, 54% in central North Carolina, and 17% nationally). In response to the open-ended question directed at how they coped, 20% (N=167) of the subjects (24% of those aged 70 years and over) spontaneously replied that religion was a primary factor. Religion in this sense typically involved having trust or faith in God, praying, reading the Bible on other religious literature, listening to religious programs on the radio or watching religious programs on television, participating in church services or other related activity, and receiving emotional support from church members or a pastor. On the visual analog scale, which ranged from 0 to 10, the mean rating by patients was 6.5 (SD=3.1). More than half of the subjects (56%, N=471 ) rated themselves 7.5 or higher, and 21 % gave themselves a rating of 10 (religion being “the most important thing that keeps me going”). Observer ratings of religious coping for the group ranged from 0 to 10, with a mean 5.7 (SD=3.2). The religious coping index scones for the group ranged from 0 to 30; the mean was 14.3 (SD=8.7).
1696
Age Race Education Occupation Retirement Social/economic
Depression Scale
Depression Scale
(N=842)
(N=841)
(N=333)
-0.02
-0.02 -0.1
0.00
-0.o9’ 0.06
008e
-O.ll’
-0.02
0.05 0.1Sc
status resources
-0.01
status
Social
support
Mental health History of psychiatric problems Family history of psyAlcohol
-0.04
-0.00
situation
chiatric 5.1
Hamilton
Religious Copinga
characteristics
Marital Income
Assemblies of God, Church of God)
religious
Covariate
Living
.
of
Day Adventist) Black Protestant (Methodist, Baptist) Fundamentalist/evangelical (Pentecostal, Holiness,
aSce
Geriatric
Demographic
Liberal Protestant (Episcopal, United Church of Christ, Presbyterian) Moderate Protestant (Methodist, Disciples
of
TABLE 3. Bivariate Relation of Covariates to Religious Coping and Depression (Pearson correlations) Among 850 Elderly Male Medical Inpatients
0.03
0.03
0.06 0.05
-0.lO’ -.o9’
-0.10 .o.13e
0.12c
0.24c
O.21c
0.03
problems
0.30c
0.08e
use
Ic
0.05
0.27c
0.13c
0.l6c
0.1 Ic
O.08e
0.1Sc
0.26c 0.19c
017d 0.1
-0.00 0.01 -0.03 0.16c 0.02 0.08e
-0.01 0.02 -0.01
-0.02
-0.03
0.12e
Liberal Protestant Moderate Protestant Conservative Protestant Black Protestant
0.07c -0.06
-0.00
Fundamentalist/evangelical
0.17’ -0.05 -0.1 1d 0.04 0.1Sc
Physical health Functional status (activities ofdaily living) Cognitive status
0.01 0.05
Medical diagnosis Cancer Gastrointestinal disease Neurological disease Respiratory disease Renal disease Cardiac disease Miscellaneous Religious
-0.03
0.03 0.04
0#{149}14d
-0.02
0.00
affiliation
Protestant
aMeasured
with on the
the religious Hamilton
coping
depression
O.I2
0.1 I
0.12c
Christian preference
0.07
-0.06
-0.02
(unspecified)
Catholic Nontraditional No religious bScores
-0.01 0.02 -0.02
IC
0.02
o1od
-0.04
0.00 0.04 -0.06 -0.01 0.06
0.07 -0.10
0.09 -0.07 0.l4’
index. scale
were
available
only
aged 70 years and over. For categorical variables, point relations are reported. The correlation between Geriatric Scale and Hamilton depression scale scores was 0.66.
for
men
biserial corDepression
CP
Among Harold
and
Ill Men
M.D., M.Sc., Harvey J. Cohen, M.D., Dan G. Blazer, M.D., D.P.H., Keith G. Meador, M.D., M.P.H., Frank Sheip, M.D., Veeraindar Goli, M.D., and Bob DiPasquale, M.S.
G. Koenig, Carl Pieper,
Objective: cal inpatients,
Coping and Depression Hospitalized Medically
The investigators the characteristics
depression.
Method:
The
Ph.D.,
examined the frequency ofreligious coping among older mediof those who use it, and the relation between this behavior subjects
were
850
men
aged
65 years
and
over,
without
psychi-
atric diagnoses, who were consecutively admitted to the medical or neurological services of a southern Veterans Administration medical center. Religious coping was assessed with a threeitem index. Depressive symptoms were assessed by self-rating (the Geriatric Depression Scale) and observer five patients used to cope
rating (the Hamilton reported that religious with illness. Variables
Rating Scale for Depression). Results: One out of every thought and/or activity was the most important strategy that were associated with religious coping included black
race, older age, being retired, religious affiliation, alcohol use, a prior history ofpsychiatric problems, sive symptoms were inversely related to religious other sociodemographic and health correlates ated during their subsequent hospitaladmissions was
the
only
sions:
These
related (Am
J
baseline
variable
findings
suggest
to depression Psychiatry
that that
predicted religious
in hospitalized elderly 1992; 149:1693-1700)
sometimes
despair.
During
physical
illness
coping resources are seriously tested and frequently overwhelmed, as evidenced by the fact that 40% or more of older hospitalized patients experience some form of clinical depression (1, 2). Religious beliefs and behavior, in turn, are prevalent among older persons (3,
Received
March
July
26, 1991; revision From the Department Psychiatry, and the Center
I 9, 1 992.
Geriatric
received Feb. 5, 1992; accepted of Psychiatry, the Division of for the Study of Aging and Hu-
man Development, Research Education
Duke University Medical Center, and the Geriatric and Clinical Center, VA Medical Center, Dur-
ham,
reprint
N.C.
Address
requests
to Dr.
Koenig,
GRECC,
Box
I 82,
VA Medical Center, 508 Fulton St., Durham, NC 27705. Supported by the Mellon Foundation, Sandoz Pharmaceutical Corporation, grant AG-00371 from the Center for the Study ofAging and Human Development, Duke University Medical Center, and NIMH grant MH-40159.
Am
J
Psychiatry
I 49:1
2, December
1992
When months
lower
scores
depression is a common
support, infrequent functioning. Depreswhich persisted after
202 men were later, religious at follow-up.
behavior
that
reevalucoping Conclu-
is inversely
men.
edical illness that precipitates hospitalization is a stressful experience that can interrupt social and work routines, drain finances, separate families, reverse canegiver roles, and create situations of forced dependency. Physical illness also brings with it the threat of pain and long-term disability, disfiguration, the prospect of approaching death, and feelings of existential and
were controlled. an average of6 coping
M
anxiety
high level of social and higher cognitive coping, an association
4) and are reported to serve as a coping strategy to help manage emotional distress (5-8). The extent to which older medical inpatients use religion for this purpose, the characteristics of those who do so, and the effectiveness of this strategy are largely unknown. National samples of Americans of all ages indicate that persons who find personal comfort and support from religion are more likely to be older, female, black, less educated, widowed, employed in manual or unskilled occupations, more economically deprived, and affiliated with conservative Protestant religious denominations (3). These sociological findings suggest that persons with fewer health, social, and financial resources, when facing situations over which they have little control (such as acute hospitalization), might turn to religion for solace. Whether religious beliefs and behavior actually help to prevent or relieve emotional distress is far from clear. Psychiatric illness may be even more common among the religious (9-11), perhaps pnedisposing them to greater problems in later life when they are faced with the stress of physical illness and/on approaching death. Thus fan, systematic research on the relation between religious coping and depression among olden adults in
1693
RELIGIOUS
COPING
AND
DEPRESSION
clinical settings has been limited. Previous studies have been hampered by small sample sizes, less than rigorous sampling methods, and unequal sex distribution (few males) (6-8). This report emanates from the Durham Veterans Administration (VA) Mental Health Survey (12, 13), a cross-sectional and longitudinal epidemiologic study of depression among hospitalized medically ill men. We examine religious coping, its sociodemographic and health correlates, and its relation to depression both cross-sectionally and over time. Four major questions guided this investigation. 1 ) How common is religious coping in this population? 2) Is this coping behavior more likely among those with fewer socioeconomic, physical, on mental health resources? 3) When other factors are controlled, is there an association between religious coping and depression? 4) If such a relationship exists, is it especially strong in any subgroup of the population, such as those with more severe medical illness, low social support, prior mental health problems, on other attributes?
METhOD Sept. 1 , I 987, and Jan. 1 , 1 989, all men aged over who had been admitted to medical or neurological services at the Durham VA Medical Centen were screened for depression. For inclusion in the study, patients were required to scone 15 on higher on the Mini-Mental State examination (14) and to be physically capable of undergoing a psychiatric evaluation. Psychiatric patients were excluded. Patients were evaluated within 48-72 hours of admission by a social worker and/or a Fellow in geriatric medicine (H.G.K.). Data were collected on demographic characteristics and social and economic resources, including age, race, education, prior occupation (IS), retirement status, current living situation, marital status, yearly income, and social support. Social support was measured with a three-item index that explored size of the support network (16), frequency of interaction, and perceived adequacy of support ( 1 7) (values for each item ranged from 1 to 5; Cronbach’s alpha=0.57). Over 40 religious denominations were represented in the study group; these were categorized into nine general religious groups according to a schema provided by Roof and McKinney (18). The physical health of the participants, including medical diagnoses, functional status, and cognitive status, was assessed. Functional status was determined by measuring physical (19) and instrumental (20) activities of daily living. Physical activities of daily living included bathing, dressing, toileting, problems with incontinence, transfer from bed to chair or vice versa, and feeding (each rated 0-2); instrumental activities of daily Between
65 years
living
and
involved
ability
to travel,
shop,
prepare
meals,
do
housework, and handle finances (rated 0 or 1). These two types of ratings together produced an 1 1 -item scale with a possible range of scores from 0 (low functioning)
1694
to 17 (high functioning). As we have mentioned, cognitive status was measured with Mini-Mental State examination scores, which ranged from 15 to 30. The assessment of mental health included information on alcohol use, prior psychiatric problems, and family history of psychiatric problems. Depression was assessed by self-rated and observer-rated scales. The self-rated 30-item Geriatric Depression Scale (21 ) was administered by the social worker to all patients; this scale has been validated as a measure of depression in older medical inpatients (22). The men aged 70 years and olden were also assessed by one of us (H.G.K.) with the Hamilton Rating Scale for Depression (23), a measure used in other studies of the elderly (24) and medical inpatients (25). Depression scales were administered early in the interview, before the assessment of religious coping. Religious coping was assessed with a three-item index. Each item measured how much the patient relied upon religion to help manage the emotional stress associated with his illness. In item 1, the patient was asked an open-ended question about how he coped. This item was chosen in order to identify the coping behavior that the patient himself felt was most helpful, and the question was asked before questions on religion that might bias
responses.
If the
response
was
religious
in nature
(e.g., faith in God on Jesus, prayer, church), the patient received a scone of 1 0; if the spontaneous response was not religious (e.g., stay busy, family support), a score of 0 was assigned. A value of 10 for religious responses was chosen in order to give this item equal weight with the others in the index. In item 2, patients were asked to rate on a visual analog scale the extent to which they found religious beliefs or activities helpful in coping with their situation. The scale was numbered from 0 (“not much on not at all”) to 10 (“the most important thing that keeps me going”). In approaching patients with the rating scale, the investigators allowed patients to define for themselves the meaning of the term “religion” but made clear that this could involve personal belief alone or include neligious activity such as prayer or church attendance. In item 3, the interviewer rated the patient on a scale of 0-10 on the basis of an overall assessment of how much the patient used religion to cope. This judgment was based on the patient’s further elaboration on religious coping themes during answers to items 1 and 2 and on a separate discussion about how religion was helpful. The scores from the three items were then summed, and an index with values ranging from 0 to 30 was obtained. Cronbach’s alpha (reliability) for the index in the overall sample (N=850) was acceptable (0.82). Test-retest/internaten reliability for the religious coping index was also determined for a subgroup of 188 consecutively admitted men. The religious coping index was administered twice to these patients, each time by a different rater; ratings were separated by 12-36 hours. The Pearson correlation between scores obtamed on the religious coping index at the two administrations was 0.81. The interraten agreement for the ob-
Am
J
Psychiatry
149:12,
December
1992
KOENIG,
server-rated religious coping item (item 3) was surpnisingly high (Pearson’s n=0.87) given that the raters came from markedly different religious backgrounds (secular humanist versus conservative Protestant). All participants in the baseline study who were readmitted to the medical or neurological services during the 16-month study period and S months thereafter (designated “time 2”) were reevaluated by the social worker with the Geriatric Depression Scale and the religious
coping
index.
Again,
psychiatric
admissions
were excluded. If more than one neadmission occurred, the results from the last interview during the 21-month observation period was used for time 2. This provided information on change in depressive symptoms and religious coping oven time among patients whose medical illness prompted rehospitalization during the project. Analyses were performed with the SAS statistical package (26). Simple statistics were used to determine the frequency of spontaneously reported, self-rated, and observer-rated religious coping. Bivaniate relations of covaniates with religious coping (religious coping indcx score) and depression (Geriatric Depression Scale and Hamilton depression scale scores) were examined with Pearson correlations. Hierarchical regression was used to examine the strength of relationships, controlling for the effects of other covaniates. To determine whether patients who used religious coping had fewer socioeconomic or health resources than other patients, the religious coping index scone was regressed on a series of 16 demographic, socioeconomic, and health variables (excluding depression). A backward stepwise regression method was used to eliminate nonsignificant variables (p>O.OS ); missing values were dealt with by listwise deletion. The regression
was
performed
in
five
stages.
In
the
first
coping
in the
following
manner.
First,
a model
for depressive symptoms was constructed by regressing it on the 15 sociodemographic and health variables (cxcluding religious variables). Once a final model had been obtained, we added religious coping and religious affiliation variables. Interactions between religious coping and all other variables in the model were tested and included in the final model if alpha was less than or equal to 0.05. Regression models were developed for both self-rated symptoms (Geriatric Depression Scale) and observer-rated symptoms (Hamilton Rating Scale for Depression). We analyzed the longitudinal data as follows. Pearson correlations determined bivaniate relations between baseline (time 1 ) group characteristics and follow-up (time 2) Geriatric Depression Scale score. Hierarchical regression was then used to assess the relation between time 1 religious coping index score and time 2 Geriatric Depression
Am
J
Psychiatry
I 49:1
2, December
1992
Cha racteristi
and Health
Mean
Characteristic
BLAZER,
SD
ET AL.
cs of 85 0 EldN
%
Demographic Age (years) Black race Education (years) Unskilled occupation
69.8
4.9 240
28.3
333
39.2
589
69.3
577 160
67.9 18.8
History of psychiatric problems
216
25.7
Family history problems Alcoholuse
153
18.2
186
22.1
49
14.7
Retired
more
8.9
than
3.8
S years
Social/economic Married Livingalone
Income (dollars per year) Social support ratinga Mental health
Geriatric
76
Depression
score
disease
Neurological
disease disease
Renal disease Cardiovascular Other Functional status daily living)’ Cognitive status
Mental
State
of possible
9.1
Scale
Cancer Gastrointestinal
aRange
3,316 1.8
of psychiatric
score11 Hamilton depression 15 Physical health Medical diagnosis
Respiratory
8,582 10.6
disease (activities
188 125
22.1 14.7
I 22 87 40 216 72
14.4 10.2 4.7 25.4 8.5
of
(Miniscore)C
14.1
4.0
26.4
2.8
scores=3-1S.
bRange
of possible
scores=0-1
cRange
of possible
scores=0-30.
7.
stage,
demographic variables (age, race, and occupational and retirement status) were entered into the model. In subsequent stages, religious affiliation, socioeconomic resources, and mental and physical health variables were added. The fifth stage produced a final model that contamed the significant correlates. We examined the relation between depression and religious
TABLE 1. Sociodemographic erly Male Medical Inpatients
COHEN,
Scale score, controlling Scale score and other
for time significant
I Geriatric Depression baseline correlates.
RESULTS There were I , 1 1 0 consecutively admitted new patients during the study period. Eight hundred fifty men (77%) underwent comprehensive social, psychological, and physical health examinations; 260 did not participate because of advanced dementia or delirium (12% with Mini-Mental State scores less than IS), communication problems (2%), refusal or discharge before being seen (5%), or other reasons (4%). Nonparticipants were more likely to be older, black, and residents of nursing homes and to have diagnoses of neurological or respiratory illness. The sociodemognaphic and health characteristics of the subjects are presented in table I . The distributions of race, marital status, and medical diagnoses were similar to those among elderly male patients discharged from VA hospitals in District 8, which covers most of North Carolina and parts of Virginia, Kentucky, Tennessee, and South Carolina (12). Table 2 presents the
I 695
RELIGIOUS
COPING
AND
DEPRESSION
TABLE 2. Distribution of Religious Affiliations Among 850 Elderly Male Medical Inpatients, Elderly Men in Central North Carolina, and in the United States as a Whole Percent Percent Elderly
Religious
Male Medical Inpatients
Groupa
Central North Carolina’
DepressionL Percent in the
U.S. Populationa
8.0
7.1
8.7
Lutheran, of Christ,
Reformed) Conservative Protestant (White Baptist, Church Christ,
of
Elderly Men in
Nazarene,
12.5
17.2
19.9
Seventh-
40.8
S4.7’
22.2
I 1.3 5.8
5.1
4.5
Protestant (unspecified) Catholic Jewish Nontraditional Christian (Mormon, Jehovah’s Witnesses, Christian Science, Unitarian)
3.3 2.6 0.0
5.9 1.5 0.3
25.0 2.3
3.2
3.3
8.0
No
2.4
5.1
6.9
bMen
preference
Roofand
aged
demiological Clncludes
McKinney
55 years
(18).
and over who
Catchment conservative
Area Protestants
participated
study and
(central black
in the Piedmont North
Carolina)
Epi(27).
Protestants.
religious affiliations of the participants. The distnibution was comparable to that of elderly men living in central North Carolina; compared to the population of the nation as a whole, however, a disproportionate number of men came from conservative on black Protestant denominations (63% in this study, 54% in central North Carolina, and 17% nationally). In response to the open-ended question directed at how they coped, 20% (N=167) of the subjects (24% of those aged 70 years and over) spontaneously replied that religion was a primary factor. Religion in this sense typically involved having trust or faith in God, praying, reading the Bible on other religious literature, listening to religious programs on the radio or watching religious programs on television, participating in church services or other related activity, and receiving emotional support from church members or a pastor. On the visual analog scale, which ranged from 0 to 10, the mean rating by patients was 6.5 (SD=3.1). More than half of the subjects (56%, N=471 ) rated themselves 7.5 or higher, and 21 % gave themselves a rating of 10 (religion being “the most important thing that keeps me going”). Observer ratings of religious coping for the group ranged from 0 to 10, with a mean 5.7 (SD=3.2). The religious coping index scones for the group ranged from 0 to 30; the mean was 14.3 (SD=8.7).
1696
Age Race Education Occupation Retirement Social/economic
Depression Scale
Depression Scale
(N=842)
(N=841)
(N=333)
-0.02
-0.02 -0.1
0.00
-0.o9’ 0.06
008e
-O.ll’
-0.02
0.05 0.1Sc
status resources
-0.01
status
Social
support
Mental health History of psychiatric problems Family history of psyAlcohol
-0.04
-0.00
situation
chiatric 5.1
Hamilton
Religious Copinga
characteristics
Marital Income
Assemblies of God, Church of God)
religious
Covariate
Living
.
of
Day Adventist) Black Protestant (Methodist, Baptist) Fundamentalist/evangelical (Pentecostal, Holiness,
aSce
Geriatric
Demographic
Liberal Protestant (Episcopal, United Church of Christ, Presbyterian) Moderate Protestant (Methodist, Disciples
of
TABLE 3. Bivariate Relation of Covariates to Religious Coping and Depression (Pearson correlations) Among 850 Elderly Male Medical Inpatients
0.03
0.03
0.06 0.05
-0.lO’ -.o9’
-0.10 .o.13e
0.12c
0.24c
O.21c
0.03
problems
0.30c
0.08e
use
Ic
0.05
0.27c
0.13c
0.l6c
0.1 Ic
O.08e
0.1Sc
0.26c 0.19c
017d 0.1
-0.00 0.01 -0.03 0.16c 0.02 0.08e
-0.01 0.02 -0.01
-0.02
-0.03
0.12e
Liberal Protestant Moderate Protestant Conservative Protestant Black Protestant
0.07c -0.06
-0.00
Fundamentalist/evangelical
0.17’ -0.05 -0.1 1d 0.04 0.1Sc
Physical health Functional status (activities ofdaily living) Cognitive status
0.01 0.05
Medical diagnosis Cancer Gastrointestinal disease Neurological disease Respiratory disease Renal disease Cardiac disease Miscellaneous Religious
-0.03
0.03 0.04
0#{149}14d
-0.02
0.00
affiliation
Protestant
aMeasured
with on the
the religious Hamilton
coping
depression
O.I2
0.1 I
0.12c
Christian preference
0.07
-0.06
-0.02
(unspecified)
Catholic Nontraditional No religious bScores
-0.01 0.02 -0.02
IC
0.02
o1od
-0.04
0.00 0.04 -0.06 -0.01 0.06
0.07 -0.10
0.09 -0.07 0.l4’
index. scale
were
available
only
aged 70 years and over. For categorical variables, point relations are reported. The correlation between Geriatric Scale and Hamilton depression scale scores was 0.66.
for
men
biserial corDepression
CP
