Accelerated Course of Human Immunodeficiency Virus

Accelerated Course of Human Immunodeficiency Virus Infection in Gay Men
Who Conceal Their Homosexual Identity
STEVE W. COLE, PHD, MARGARET E. KEMENY, PHD, SHELLEY E. TAYLOR, PHD,
BARBARA R. VISSCHER, MD, AND JOHN L. FAHEY, MD
Research linking psychological inhibition to physical illness led us to examine whether human immunodeficiency virus (HIV) infection might progress more rapidly among gay men who conceal their homosexual
identity than among those who do not. We also sought to determine whether any accelerated course of HIV
infection among "closeted" gay men might be attributable to differences in health-relevant behavior (e.g.,
health practices, sexual behavior) or psychosocial characteristics (e.g., depression, anxiety, social support,
repressive coping style). Data came from a longitudinal psychosocial study associated with the Los Angeles
site of the Multicenter AIDS Cohort Study. Eighty gay men, HIV-seropositive but otherwise healthy at study
entry (CD4 T lymphocytes = 30-60% of total lymphocytes), were examined at 6-month intervals for 9 years.
Indicators of HIV progression included time to a critically low CD4 T lymphocyte level (15% of total
peripheral blood lymphocytes), time to AIDS diagnosis, and time to AIDS mortality. On all measures, HIV
infection advanced more rapidly in a dose-response relationship to the degree participants concealed their
homosexual identity. Sample characteristics and statistical controls ruled out explanations based on
demographic characteristics, health practices, sexual behavior, and antiretroviral therapy. Mediational
analyses indicated that observed effects were not attributable to differences in depression, anxiety, social
support, or repressive coping style. HIV infection appears to progress more rapidly in gay men who conceal
their homosexual identity. These results are consistent with hypotheses about the health effects of psychological inhibition, but further research is required to definitively identify the psychosocial, behavioral, and
physiological mechanisms underlying these findings.
Key words: HIV/AIDS, CD4 T lymphocytes, disease progression, homosexuality, psychological inhibition,
psychosocial characteristics.
INTRODUCTION
Infection with the human immunodeficiency virus
1 (HIV-lj has a highly variable course, with some
individuals progressing rapidly to acquired immune
disease syndrome (AIDS) and others remaining
asymptomatic and immunologically healthy for long
periods of time (1-5). Although some variation in
disease progression can be attributed to characteristics of the virus (e.g., viral strain) and the host
immune system (e.g., human histocompatability leukocyte antigen genotype), these characteristics alone
do not appear to fully explain the substantial individual differences in HIV progression (1). As a result,
a great deal of research has concentrated on identi-
From the Departments of Psychology (S.W.C., S.E.T), Psychiatry and Biobehavioral Sciences (M.E.K.), Microbiology and Immunology (J.L.F.), and the School of Public Health (B.R.V.), University of California, Los Angeles, Los Angeles, California.
Address reprint requests to: Steve W. Cole, PhD, UCLA Department of Psychology, 405 Hilgard Ave., Los Angeles, CA 900241563.
Received for publication February 15, 1995; revision received
September 18, 1995.
Psychosomatic Medicine 58:219-231 (1996)
0033-3174/96/5803-0219$03.00/0
Copyright © 1996 by the American Psychosi
tying other factors that might influence the course of
HIV infection (1, 5). Psychosocial characteristics of
the host are one type of cofactor that has received a
great deal of attention in this regard (6-7).
Perhaps the most classic of all psychosomatic
hypotheses suggests that inhibiting the expression of
thoughts, feelings, or social behaviors can be harmful to one's physical health. The roots of this hypothesis are often traced back to the second century
Greek physician Galen, who described a heightened
incidence of cancer among those with inhibited,
"melancholic" personalities (8). Galen believed that
differences in both personality and physical health
stemmed from their common basis in the balance of
bodily humors (e.g., an excess of black bile among
the cancer-prone melancholic).
Later researchers proposed that psychological inhibition is actually physiologically harmful (9). Experimental studies have shown that inhibiting the
expression of emotions and self-threatening knowledge can increase activity in the sympathetic division of the autonomic nervous system (10-13), and
high levels of sympathetic activity are capable of
suppressing some aspects of immune function (e.g.,
in vivo: levels of circulating lymphocytes and monocytes, antibody production, immediate and delayed219
S. W. COLE et al.
type hypersensitivity reactions, and activity in secondary lymphoid organs, and in vitro: cytokine
production, lymphocyte maturation, motility, proliferation, and cytolysis, and phagocyte chemotaxis
and lysosomal activity) (14-19). Studies in which
participants disinhibited psychological material by
expressing thoughts and feelings associated with
traumatic personal experiences have demonstrated
short term improvements in some measures of physical health (e.g., physician visits) and immune function (e.g., latent virus control, antibody response,
and mitogen-induced lymphoproliferation) as well
as reductions in sympathetic nervous system activity
(20-25). In addition to these experimental results,
some epidemiological studies have documented associations between inhibited psychosocial characteristics (e.g., reduced emotional expression) and the
incidence and progression of cancer (26-27), hypertension (28-29), and some types of autoimmune
disease (30). These epidemiological claims are controversial, and the evidence is uneven, particularly
in the case of cancer (26-27). However, such findings do converge with the experimental evidence in
suggesting a possible link between psychological
inhibition and physical illness.
Relationships between psychological inhibition
and physical health may have important implications for the course of HIV infection in gay men.
Many gay men inhibit the expression of their homosexual identity to avoid stigmatization, ostracism,
and physical assault (31-33). To determine whether
inhibition-linked alterations in immune function
might potentially hasten the course of HIV infection,
we examined the relationship between concealing
homosexual identity and HIV progression in data
from a 9-year longitudinal study of 80 (initially
healthy) seropositive gay men participating in a
psychosocial study associated with the University of
California of Los Angeles (UCLA) site of the Multicenter AIDS Cohort Study (MACS) (34). HIV progression was measured by time from MACS entry to
three end points: a critically low CD4 level, diagnosis of an AIDS-defining condition [under the 1987
Centers for Disease Control (CDC) case definition],
and death due to HIV-related pathology. Times to
each end point were analyzed using standard survival analytic statistical models, and all analyses
controlled for any potential differences in age, disease progression at study entry (marked by CD4 level
at MACS entry), antiretroviral use, unsafe sexual
practices, and several other biobehavioral characteristics that could conceivably influence immune
function or HIV progression (e.g., recreational drug
use, sleep disturbance, exercise) (6, 35). Anxiety,
220
depression, repressive coping style, and social support were also examined as psychosocial variables
that might potentially mediate concealment-related
differences in disease progression. Results of these
analyses indicate that HIV infection did indeed
progress more rapidly among those participants who
concealed their homosexual identity. Moreover,
these progression differences could not be attributed
to differences in demographic characteristics,
health-relevant behavioral habits, or psychosocial
characteristics such as anxiety, depression, repressive coping style, or social support.
METHOD
Participants
The MACS/Natural History of AIDS Psychosocial Study. Between 1984 and 1986, 1759 gay and bisexual men were recruited
into the Los Angeles sample of MACS (34). No participant knew
his HIV serostatus at study entry (HIV antibody testing was not yet
available), all were over 18 years of age, and none had been
diagnosed with cancer or any AIDS-defining condition. The
present analyses focused on the 988 MACS participants who were
subsequently found to be HIV-seropositive at study entry. At
6-month intervals, blood samples were taken, participants were
examined for signs and symptoms of AIDS, and interviewers
assessed variables potentially relevant to the course of HIV
infection (e.g., demographic characteristics, drug use, health practices, sexual behavior).
Beginning in 1987, 333 HIV-positive MACS participants volunteered to participate in the Natural History of AIDS Psychosocial Study (NHAPS), a project focusing on cognitive, emotional,
social, and behavioral factors associated with HIV infection (3637). Participants completed psychosocial questionnaires at
6-month intervals coinciding with MACS examinations. In 1987
and 1988,192 of these participants completed an additional set of
questionnaires that solicited information about psychosocial
characteristics, including homosexual identity and AIDS-related
social support. To ensure that advanced HIV infection did not
influence any of the psychosocial variables measured in this
study, analysis was restricted to individuals with normal CD4
levels (30-60% of peripheral blood lymphocytes) at MACS entry.
This resulted in a group of 80 HIV-seropositive MACS/NHAPS
participants in the earliest stages of infection (median CD4 level =
686 cells/mm3, interquartile range = 524-829). Median age was
34 years (range 23-50), 96% were Caucasian (3% Hispanic, 1%
Asian), 86% attended college, and 60% held jobs in the managerial/professional sector of the U.S. Census occupational classification (22% technical/sales, 12% service, 6% other). This group
of 80 subjects was representative of the entire sample of seropositive Los Angeles MACS participants on each MACS-assessed
variable examined in this analysis (except initial CD4 level) and
was representative of the remaining 253 HIV-seropositive NHAPS
participants on each NHAPS-assessed variable (all differences <
.2 SD). Between entry into MACS and September 1993, 34% of
these individuals received an AIDS-related diagnosis (1987 CDC
case definition), and 26% died of an AIDS-related disorder. All
data were collected in accord with Institutional Review Board
procedures at UCLA.
Psychosomatic Medicine 58:219-231 (1996)
CONCEALED HOMOSEXUALITY
Measures
Four types of variables were assessed: psychosocial characteristics (including degree of concealed homosexual identity), HIVrelevant diagnostic events (AIDS diagnosis and AIDS-related
mortality), HIV-relevant immune parameters (CD4 and CD8 T
lymphocyte levels), and demographic and biobehavioral characteristics that could potentially affect immune parameters or HIV
progression (e.g., health practices and sexual behavior). Unless
otherwise indicated, all psychosocial, biobehavioral, and demographic variables were measured at entry into the NHAPS (i.e.,
either through NHAPS instruments or the immediately preceding
MACS interview).
General deficit in perceived social support was summarized by
averaging deficit scores across sources.
Repressive Coping Style. Individuals who tended not to experience the negative or threatening aspects of an event were
identified by scores below the sample median (5.5) on the Bendig
short form of the TMAS and above the sample 75th percentile
(13.0) on Strahan and Gerbasi's 20-item revision of the MarloweCrowne Scale of Social Desirability (44). [For more on the definition and measurement of repressive coping style, see Weinberg et
al. (45).]
Diagnostic Measures
Psychosocial Measures
Concealed Homosexual Identity. All participants indicated
being either "exclusively homosexual," or "predominantly homosexual," and "only insignificantly heterosexual." As in previous
studies (32-33, 38), the degree to which respondents concealed
their homosexual identity was measured by a 5-category index
asking respondents to judge themselves as being ("relative to
other gay men"): "definitely in the closet, in the closet most of the
time, half in and half out, out of the closet most of the time, or
completely out of the closet."
Affect and Mood. Depression was measured by the Center for
Epidemiologic Studies Depression scale (CES-D) (39), a 20-item
scale measuring affective and vegetative symptoms of depression,
and by the depression-dejection subscale of the Profile of Mood
States (POMS) (40), a 65-item instrument measuring the intensity
of six moods (depression-dejection, tension-anxiety, anger-hostility, fatigue-inertia, confusion-bewilderment, and vigor-activity)
over the previous week. Anxiety was measured by the 20-item
Bendig short form of the Taylor Manifest Anxiety Scale (TMAS)
(41) and the POMS tension-anxiety subscale.
Social Support. The presence of a primary partner was assessed
by responses to the question "Are you currently in an intimate
relationship (with a spouse or partner)?" and, if yes, "How long
have you been in this relationship?" Relationships of at least a
6-month duration were counted as stable, intimate relationships,
and all participants who reported such relationships indicated
that their partner was a male. Satisfaction with social relationships was measured by the 20-item UCLA Loneliness scale (42).
Subjective isolation was measured by responses to the question
"Is there someone you can talk to about things that are important
to you—someone you can count on for understanding and support?" Response options were "No, no one" (scored 0), "Yes, one
person" (scored 1), "2 or 3 people like that" (scored 2.5), "4 or 5
people like that" (scored 4.5), and "6 or more people like that"
(scored 6). AIDS-specific social support was assessed by a modified version of the UCLA Social Support Inventory (43). This
15-item instrument assesses the degree to which respondents
want, receive, and are satisfied with, the support they get from
intimate partners, gay friends, straight friends, and parents as
respondents face the threat of AIDS. Each dimension of subjective
social support (desired/received/satisfaction) is assessed for each
of five sources (partner/gay friends/straight friends/mother/father). All responses were given on 5-point scales with verbal
anchors ranging from "never" to "all the time" (desired and
received) or from "very dissatisfied" to "very satisfied" (satisfaction). Summary measures for each dimension were obtained by
averaging across responses for each source. Discrepancies between desired and received support may also be relevant, so a
deficit score (received - desired) was computed for each source.
Psychosomatic Medicine 58:219-231 (1996)
Diagnosis of HIV Infection. HIV serostatus was determined by
enzyme-linked immunosorbent assay for anti-HIV-1 antibodies
with positive results confirmed by Western blot for HIV-i gene
products (46).
Diagnosis of AIDS. All diagnoses of AIDS-defining conditions
were made by physicians associated with MACS or were based on
medical records supplied to MACS by the participant's primary
physician. Case definition followed the 1987 CDC criteria (47).
AIDS-Related Mortality. Cause and date of death information
was based on death certificate searches routinely conducted by
MACS. All deaths occurring in this sample were AIDS-related,
and the present analyses were based on death certificates available through August, 1993.
Immunological Measures
T Lymphocyte Levels. Blood was drawn at each MACS visit,
and levels of CD4 and CD8 T lymphocytes were measured by
two-color (CD4: Leu-3+ Leu-4+; CD8: Leu-2+ Leu-4 + ) flow
cytometry on whole blood (48). CD4 levels were reported as both
number (per cubic millimeter of peripheral blood) and percentage
of total lymphocytes. Percentage of CD4 was used as the primary
measure of CD4 loss because previous research has shown it to be
a more reliable indicator of HIV progression (49-50).
Measures of Immune-Relevant Background
Characteristics
Demographic and biobehavioral characteristics that may influence immune function and HIV progression included age, cigarette, drug, and alcohol use, exercise, and sleep disturbance (6,
35). In addition, the course of HIV infection may also be affected
by the use of antiretroviral therapies (e.g., AZT/zidovudine) (51)
and by rates of unprotected anal-receptive intercourse (2). The
number of anonymous sexual contacts may also relate to the
number or virulence of HIV strains contracted. The MACS interview assesses age (birth date), antiretroviral use (yes/no), the
number of partners in unprotected anal-receptive intercourse (per
6-month period between visits), the number of anonymous sexual
partners ("you would not be able to find this person again"; per
6-month period between visits), as well as a number of other
forms of sexual contact (e.g., anal-insertive, oral-genital, and
oral-anal contact). Anal-insertive contact, oral-genital contact,
and oral-anal contact are not discussed in detail here because
these behaviors were not significantly associated with either
concealment or HIV progression. Incidence of unprotected analreceptive intercourse was summed over the entire study period to
221
S. W. COLE et al.
stabilize substantial within-individual variation on this measure
(intraclass correlation = .12). Other biobehavioral measures
showed greater stability over time (intraclass correlations ranging
between .40 and .72), and assessment of those variables was based
on Ihe 6-month period preceding NHAPS entry. Any use of
antiretrovirals before or during the course of the study was
counted, and analyses treating antiretroviral use as a time-varying
covariate produced results similar to those reported here. The
MACS interview also assesses cigarette smoking, alcohol consumption, and the use of marijuana/hashish, nitrite inhalants
("poppers"), cocaine, barbiturates/tranquilizers/sedatives, and
amphetamines. Use of each class of drug was scored dichotomously (none during previous 6 months vs. some), smoking
responses were absent, present, or occasional (less than 1 cigarette
per day), and alcohol consumption was measured dichotomously
(no drinks containing alcohol during previous 6 months vs.
some), with drinkers then classified as consuming a maximum of
3 or more drinks in any given 24-hour period vs. 2 or fewer
(analyses based on average number of drinks per 24-hour period
produced similar results, but average consumption levels were
less predictive of HIV progression than maximum consumption
levels). Dichotomous scores were created by grouping occasional
smokers with nonsmokers and consumers of 2 or fewer drinks per
24-hour period with nondrinkers. NHAPS questionnaires include
items assessing sleep disruption ("Over the last month, how many
nights in an average week did you get less sleep than you
needed?"; responses ranging from 0 to 7) and aerobic and anaerobic exercise ("Over the last month, how many days in an average
week did you engage in aerobic exercise: vigorous and continuous
activity such as running, swimming, bicycling?" and "Over the
last month, how many days in an average week did you engage in
anaerobic exercise: short bursts of activity such as tennis, walking, yoga, baseball, stretching?"; both with responses ranging from
0 to 7). Dichotomous scores were created by grouping those who
reported an average of 2 or more nights of insufficient sleep per
week (vs. 0 or 1) and those who reported either aerobic or
anaerobic exercise an average of 1 day a week or more (vs. those
who were sedentary).
participants indicated that the risk of AIDS morbidity and mortality increased dramatically at this point (Fig. 1) (54). Similar
results emerged using either 10% CD4 or 200 CD4 cells/mm3 as an
immunological end point. Event times were censored at September 1993 if individuals failed to reach an end point by then (55).
Time to critically low CD4 was censored at date of death for
individuals who died before reaching this end point.
Analyses
Our general analytical strategy involved determining whether
the inclusion of a measure of concealed homosexual identity
significantly improved the predictive power of a statistical model
of HIV progression times based on biobehavioral variables identified in previous research as significant predictors of disease
progression or immune function. We then sought to determine
whether any incremental predictive power associated with concealment could be attributed instead to differences in psychosocial characteristics such as anxiety, depression, repressive coping
style, or social support.
Accelerated failure-time survival analysis (55) was used to
model times to 15% CD4, AIDS diagnosis, and death. Specification checks showed the data to be consistent with a log linear
model of (possibly censored) survival times with Weibull-distributed residuals (55). To ensure the robustness of results, each
analysis was repeated using a) Cox's semiparametric proportional
hazards model (55), b) a linear accelerated failure time model with
Weibull residuals, and c) linear Weibull, log linear Weibull, and
Cox models of age at event (date of event - date of birth) rather
than time from study entry (date of event — date of MACS entry).
In every case, the six models produced substantively identical
CD4 endpoint
AIDS diagnosis
AIDS mortality
Measures of Disease Progression
HIV infects cells expressing the CD4 surface protein (1). AIDS
is characterized by the onset of opportunistic infections and
neoplastic diseases after a selective decline in levels of CD4+ T
lymphocytes ("helper T cells") (1). The rate of CD4 T cell decline
varies across individuals (1-6); those who show more rapid loss
rates are more likely to develop an AIDS-defining condition
(52-53). We assessed the progress of HIV infection by a) time from
MACS entry to a critically low CD4 level ( s 15% of total
lymphocytes), b) time from MACS entry to first diagnosis of an
AIDS-defining condition, and c) time from MACS entry to death
from HIV-related pathology. CD4 T cell levels were expressed as a
percentage of total lymphocytes. Similar results emerged from
analyses based on arcsin-transformed percentages, log- and fourth
root-transformed CD4 numbers, and log-transformed CD4/CD8
ratios. Results for percentage of CD4 are reported here because
previous research has indicated that CD4 percentages are the best
T cell marker of AIDS progression (49-50), and our results
concur; log likelihoods for percentage-based models of AIDS
progression and mortality exceeded those for count-based models
by 9 to 11 units on equivalent degrees of freedom. Fifteen percent
CD4 was selected as an immunological end point because preliminary analysis of data from 318 HIV-seropositive MACS/NHAPS
222
Q
<
o
E
CO
60
50
40
30
20
10
CD4 level (% total lymphocytes)
Fig. 1. Six-month risk of AIDS diagnosis and mortality as a
function of CD4 level.
Psychosomatic Medicine 58:219-231 (1996)
CONCEALED HOMOSEXUALITY
results. Log linear Weibull results are reported here because
clinical event-times are conventionally represented in a log linear
format (55) and parametric results are likely to be more accurate
than semiparametric results when the data follow a known hazard
function (56). To aid interpretation, parameter estimates are
expressed as a percentage change in expected event time (calculated as eh - 1, with b representing the log linear parameter
estimate).
Our basic model of HIV progression times included variables
known to influence disease progression, including age, CD4 level
at MACS entry1 (as a measure of disease progression at study
entry; analyses using CD4 level at NHAPS entry produced similar
results), antiretroviral use, and frequency of unprotected analreceptive intercourse (1, 2, 6, 51). This model also included
behavioral variables that previous researchers have suggested be
controlled in all analyses relating psychosocial factors to immune
function (6, 35): cigarette smoking, heavy alcohol consumption,
sleep disruption, exercise, and the use of five types of recreational
drugs. MACS participants were predominantly white (96%), well
educated (86% attended college), and affluent (68% with incomes > $30,000/year, 60% managerial/professional, 22% technical/sales, 12% service), so confounding by socioeconomic status and ethnicity could be ruled out.
Responses to the concealment measure were analyzed as scores
on a 5-point ordinal scale (1 = "completely out", 5 = "definitely
in the closet"). We also assessed the dose-response relationship
between concealment and disease progression by treating concealment scores as a 4-level grouping variable and by conducting
trend analyses. (Few participants indicated being "definitely in
the closet," so we grouped them with participants who indicated
being "in the closet most of the time" to increase group size and to
stabilize sampling variance. Substantively identical results
emerged from analysis as a 5-level categorical variable, but the SE
of parameter estimates for the fully closeted group was substantially inflated because of small group size.) To simplify the
graphical presentation of results, concealment scores were also
collapsed into two groups representing individuals who were
relatively high in concealment (at least "half in the closet") and
those who were relatively low ("mostly" or "completely out").
1
All analyses relating concealment to disease progression
controlled for CD4 level at study entry as a measure of disease
progression at study entry. Regardless of whether initial CD4
levels were expressed as numbers or percentages, coefficients for
concealment indicated similar reductions in times to each end
point (e.g., in analyses of concealment scale scores, acceleration
values ranged between —4.4 and -13.1% per scale point for
number CD4 vs. -7.7 and -14.9% for percentage of CD4).
However, analyses using CD4 number produced less significant p
values because SE for all coefficients were inflated by the increased random variability associated with CD4 numbers. Relative to CD4 percentages (which depend only on relatively accurate
flow cytometry results), CD4 numbers are known to show increased measurement error (due to measurement error in the
white blood cell differential). As a result, CD4 numbers are known
to be poorer predictors of HIV progression than CD4 percentage,
and studies comparing the two suggest that percentage of CD4 be
used instead of number of CD4 in analyses of HIV progression
(49-50).
Psychosomatic Medicine 58:219-231 (1996)
RESULTS
Concealed Homosexuality and Biobehavioral
Characteristics
All members of this sample reported being either
exclusively or predominantly homosexual. Few (4%)
of our participants reported being "definitely in the
closet" or "in the closet most of the time." A larger
number (14%) reported being "half in and half out,"
with the remainder being "mostly" (38%) or "completely out of the closet" (45%). This distribution of
concealment scores is similar to that found in other
recent studies of gay men in urban settings (38).
Table 1 reports descriptive statistics for the distribution of potential biobehavioral confounders along
with the correlation between each characteristic and
the 5-level ordinal measure of concealment. Concealment was associated with increased rates of
sleep disturbance and unprotected anal-receptive
intercourse and decreased rates of physical exercise.
Concealment was not significantly correlated with
either CD4 level or HIV-related symptomatology at
either MACS or NHAPS entry (both -.20 < r < .20,
p > .40).
Biobehavioral Characteristics and HIV
Progression
As in previous studies (2), we found that individual differences in times to each HIV-relevant end
TABLE 1. Biobehavioral Characteristics
Mean (SD) or
Percentage
Age (yr)
Initial CD4 (% total lymphocytes)
Antiretroviral use (during entire
study)
Frequency of unprotected analreceptive intercourse (mean/6
months during entire study)
Exercise (average a 1 day/wk)
Sleep disturbance (average a 2
nights/wk)
Cigarette smoking (any in previous
6 mo)
Alcohol consumption (£3 drinks
max/24 h during previous 6
mo)
Drug use (any in previous 6 mo)
Marijuana/hashish
Nitrite inhalants
Cocaine
Amphetamines
Barbiturates/tranquilizers/sedatives
Correlation
P
with
Concealment
34.4 (6.13)
-.032
.775
40.0 (7.17)
.181
-.151
53.8
.195
.146
1.12 (1.72)
.211
.060
75.0
57.5
-.287
.257
.010
.021
38.8
.030
.791
67.5
.159
.159
63.8
41.3
20.0
11.3
17.5
.049
.123
-.136
-.096
-.068
.664
.276
.228
.395
.550
223
S. W. COLE et al.
point could be predicted from age, disease progression at MACS entry (indexed by CD4 level), and
frequency of unprotected anal-receptive intercourse
(all p < .05 in multivariate models controlling for all
biobehavioral factors listed in Table 1). We also
found, as have previous studies (57), that antiretroviral use was associated with significant differences
in times to critically low CD4 level and times to
AIDS diagnosis (both p < .05) but not times to AIDS
mortality [p > .20). In addition, cocaine use, cigarette smoking, and use of barbiturates/tranquilizers/
sedatives each independently predicted times to all
end points (all p < .05). None of the other biobehavioral characteristics listed in Table 1 predicted disease progression across all three end points, although several predicted times to one or two. As a
set, biobehavioral variables accounted for 33, 54, and
85% of the total deviance in times to 15% CD4, AIDS
diagnosis, and AIDS mortality, respectively (all p <
.001). In each case, variations in age and CD4 level at
study entry were the strongest predictors.
Concealment and HIV Progression
In every analysis conducted, adding measures of
concealed homosexual identity significantly improved the predictive power of a basic model including initial CD4 level, age, antiretroviral use, unprotected anal-receptive intercourse, exercise, sleep
disturbance, cigarette smoking, alcohol consumption, and recreational drug use (p < .05 for each
likelihood ratio increment-to-fit test). Relative to
participants who indicated being "mostly" or "completely out of the closet," those who reported being
"half" or more "in the closet" experienced a 40%
reduction in times to a critically low CD4 T lymphocyte level (p = .007), a 38% reduction in times to
AIDS diagnosis (p = .002), and a 21% reduction in
times to death (p = .012). Figures 2 and 3 portray the
magnitude of these effects for CD4 T lymphocyte
levels and for times to AIDS diagnosis and mortality.
Parameter estimates from analyses treating degree of
concealment as a 5-level score indicated a 7 to 14%
reduction in time to each HIV-relevant end point
with each successive degree of concealment (all p <
.05). Trend analyses treating degree of concealment
as a 4-level categorical score (Table 2) indicated a
significant linear decline in times to each HIVrelevant end point as degree of concealment increased (all p < .05). After accounting for biobehavioral variables, concealment accounted for an
additional 3 to 6% of the total deviance in times to
each end point.
224
CD4 decline
0
2
4
6
8
Time since MACS entry (years)
Fig. 2. CD4 loss rate. Median CD4 percentage ± bootstrap SE
(78). CD4 percentages did not differ at study entry (p =
AIDS and HIV mortality
i
Closet:
AIDS Death
° U t : "AIDS "Delth"
2
4
6
8
10
Time since MACS entry (years).
Fig. 3. Times to AIDS diagnosis and mortality. Estimates based
on model controlling for biobehavioral characteristics
listed in Table 1. Survivor functions are truncated at
median event time to facilitate comparison and to avoid
extrapolation beyond observed data.
Across all three HIV-relevant end points, the lowest concealment score ("completely out") was associated with slightly shorter progression times than
the next lowest score ("out most of the time"), but
this difference was not statistically significant in any
of the analyses conducted (all p > .20). The appearance of a statistically significant quadratic trend in
AIDS mortality times (Table 2) suggested that the
dose-response function may level off at the lowest
levels of concealment. However, this hypothesis was
not supported by significant quadratic trends in
times to critical CD4 level or AIDS diagnosis (Table
2). Although these data offer partial support for the
notion that HIV progression times may not differ for
Psychosomatic Medicine 58:219-231 (1996)
CONCEALED HOMOSEXUALITY
TABLE 2. HIV Progression as a Function of Degree of Concealment
Acceleration Coefficient
(% change relative to mean event time)
15%CD4
AIDS diagnosis
AIDS mortality
Trend Analysis (p values)
Completely
Out
Mostly
Out
Half Out
Half In
Less than
Half Out
Omnibus
Linear
Trend
Quadratic
Trend
Cubic
Trend
+27.8
+ 22.2
+ 10.3
+ 37.7
+ 38.7
+ 15.5
-17.3
-9.5
+ 3.8
-31.3
-34.8
-24.4
.001
.006
.002
.027
.001
.001
.533
.112
.031
.245
.250
.878
those who are "mostly" or "completely out of the
closet," they indicate a consistent linear dose-dependent decrease in HIV progression times as concealment increases through moderate to high levels (i.e.,
from "mostly out" to "half in and half out" to "in the
closet most of the time" and "definitely in the
closet").
Concealment was not significantly associated with
either time from study entry to, or CD4 level at the
commencement of antiretroviral therapy (both -.20
< r < .20, p > .40). However, to ensure that variations in the initiation and duration of antiretroviral
treatment could not explain the observed progression differences, we also fitted Cox proportional
hazards regression models treating antiretroviral use
as a time-varying covariate. Again, degree of concealment (no matter how it was coded) significantly
predicted times to each HIV-relevant end point (all
p < .05).
Psychosocial and Behavioral Mediators
If concealing homosexual identity affects HIV progression only indirectly through its impact on mediating variables (e.g., social support, affective states),
then a) putative mediators should correlate with
degree of concealment, and b) statistical adjustment
for the effects of these mediators should eliminate
concealment-related differences in disease progression. We considered 17 psychosocial characteristics
that might potentially mediate relationships between concealment and HIV progression, including
individual differences in depression (CES-D, POMS
depression-dejection), anxiety (TMAS, POMS tension-anxiety), and other dimensions of affect (POMS
fatigue-inertia, anger-hostility, vigor-activity, confusion-bewilderment), social support (UCLA Loneliness, number of confidants, four dimensions of
AIDS-specific social support), repressive coping
style, presence of a primary partner, and frequency
of anonymous sexual contact. Table 3 reports descriptive statistics and correlations with concealment for each potential mediator. This sample
Psychosomatic Medicine 58:219-231 (1996)
showed relatively low levels of depression (e.g., 18%
with CES-D scores ^ 16), anxiety, and social isolation and relatively high levels of social support. Of
the 17 potential mediators considered, only the
presence of repressive coping style correlated significantly with degree of concealment (point biserial
r = .346, p < .01; for all other mediators, -.20 < r <
.20, p > .05). However, none of these potential
mediators consistently predicted HIV progression2,
and estimates of concealment-related differences in
HIV progression were not significantly reduced by
control for any of these variables (including repressive coping style). In models controlling for each
potential mediator or for all potential mediators
simultaneously, ordinal concealment scores and
trend analyses continued to indicate a significant
negative dose-response relationship between degree
of concealment and times to each HIV-relevant end
point (all p < .05). As a result, the accelerated course
of HIV infection in closeted participants cannot be
attributed to differences in anxiety, depression, so-
2
Several potential mediators did predict times to one end
point but not others. In analyses controlling for all biobehavioral
predictors (Table 1), deficits in AIDS-specific social support
(desired - received) were associated with longer times to a
critically low CD4 level ( + 18.3%, p = .048) but not with significant differences in times to AIDS diagnosis or mortality. High
levels of both desired and received AIDS-specific social support
were associated with accelerated progression to AIDS mortality
(-8.2% and -7.6%, respectively, p = .025 and .013) but no
significant differences in times to AIDS diagnosis or critical CD4.
In contrast to these results associating social support with accelerated disease progression, presence of a primary partner predicted longer times to AIDS diagnosis (+22.2%, p = .039) and
AIDS mortality (+17.9%, p = .003) but not critical CD4. No
significant predictions were made by anxiety, depression, POMS
affective states, loneliness, satisfaction with AIDS-specific social
support, repressive coping style, or rates of anonymous sexual
contact. The small ratio of significant results to the total number
of mediational analyses (5/51: 17 mediators X 3 end points) and
the inconsistency of results across end points suggest that the
scattered pattern of significant mediational predictions may reflect Type 1 errors rather than substantive effects.
225
S. W. COLE et al.
TABLE 3. Potential Psychosocial Mediators
Anxiety (20-point Bendig/TMAS)
Depression
Total CES-D
CESDs 16
Mood (POMS) (mean on 0 - 4 scale)
Depression-dejection
Tension-anxiety
Anger-hostility
Fatigue-inertia
Confusion-bewilderment
Vigor-activity
Repressive coping style (present)
Primary partner (present a 6 mo)
Number of close confidants
Loneliness (total UCLA score)
AIDS-specific social support (0-5 scale)
Desired
Received
Satisfaction
Deficit (Desired - Received)
Anonymous sexual partners (6 mo)
Mean (SD) or
Percentage
Correlation
with
Concealment
6.08 (4.23)
-.06
.60
9.17(9.96)
18.4
-.01
-.03
.90
.77
0.69 (0.77)
1.20(0.84)
0.85 (0.75)
1.17 (0.94)
0.90 (0.78)
2.34 (0.79)
11.3
43.7
3.28(1.65)
37.2 (12.0)
-.03
-.08
-.05
-.07
.00
-.20
.35
.01
-.18
.16
.76
.47
.68
.56
.98
.08
<.O1
.91
.10
.16
2.14(0.95)
2.12(0.92)
2.56 (0.99)
0.03 (0.66)
3.37(11.4)
-.09
-.17
-.19
.11
-.05
.41
.12
.10
.34
.67
cial support, repressive coping style, primary partner status, or anonymous sexual contact.
DISCUSSION
In this sample of gay men, HIV infection advanced
more rapidly among those who concealed their homosexual identity than among those who did not.
Times to a critically low level of CD4 T lymphocytes,
times to AIDS diagnosis, and times to AIDS-related
mortality all declined significantly as degree of concealment increased from "mostly out of the closet" to
"definitely in the closet." HIV infection did not
progress significantly more rapidly among those
who were "mostly out of the closet" than among
those who were "completely out of the closet,"
suggesting that only moderate to high levels of concealment may be associated with significant accelerations in disease progression.
The accelerated course of HIV infection in relatively closeted participants could not be attributed to
differences in age, disease progression at study entry, medical treatment (e.g., antiretroviral use),
health-relevant biobehavioral characteristics (e.g.,
sexual practices, alcohol and recreational drug use,
cigarette smoking, exercise, sleep disturbance), or
psychosocial characteristics (e.g., anxiety, depression, repressive coping style, social support). In
226
P
addition, the present sample was homogeneous on
other demographic dimensions that have been
linked to disease progression (for example, gender,
ethnicity, risk group, e.g., hemophiliac vs. homosexual), so these characteristics cannot explain the
progression differences observed. As a set, biobehavioral characteristics predicted 33 to 85% of total
deviance in times to each end point, with differences
in age and CD4 level at study entry representing the
strongest predictors. Degree of concealed homosexual identity predicted an additional 3 to 6% of total
deviance in times to each end point, an effect that
amounted to a 1.5- to 2-year difference in median
HIV progression times for those "half" or "more" in
the closet relative to those "mostly" or "completely"
out. The magnitude of concealment-related differences was similar to those associated with most
biobehavioral variables considered individually
(e.g., antiretroviral use was associated with an approximate 1-1.5-year difference in times to critically
low CD4 and AIDS diagnosis, as was a 10-year age
differential, or a difference in initial CD4 level of 160
cell/mm3 or 10% of total lymphocytes).
The present results are consistent with data from
another study of 222 HIV-seronegative gay men in
which we found an increased incidence of several
infectious and neoplastic diseases as degree of concealment increased (58). Previous research also indicated increased rates of psychosomatic symptomPsychosomatic Medicine 58:219-231 (1996)
CONCEALED HOMOSEXUALITY
atology among closeted gay men (32). Thus, other
data converge with the present findings in linking
concealed homosexuality to heightened physical
health risk. Other studies using different models of
psychological inhibition (i.e., contexts unrelated to
homosexual identity) have also shown effects on
physical health outcomes (9, 23-24). Within the
domain of HIV infection, these results are also consistent with other longitudinal studies linking HIV
progression to psychosocial factors (e.g., bereavement, depression, denial, cognitive behavioral interventions, and expectancies about survival) (6-7, 37,
59-70), although other studies have failed to find
such effects (e.g., for depression) (6-7, 71).
When considering the health differences associated with concealed homosexuality, it is important
to note that "passing" as a heterosexual is not a
discrete proposition. Gay men may be "in" vs. "out"
to various degrees (e.g., known vs. suspected vs. not
suspected) and to various audiences (e.g., friends,
family, co-workers, other gay men) (32-33). Most of
our respondents reported intermediate levels of concealment, and ancillary analyses (not reported here)
suggested that these individuals were in the closet to
a greater extent at work and in public than they were
among friends or family members. However, even
partial concealment would require gay men to inhibit the expression of homosexual identity over
significant periods of time in highly consequential
settings (e.g., 8 hours a day at work). Although this
sample contained few individuals at the most closeted end of the concealment continuum, the doseresponse relationship observed over intermediate
levels of concealment supports the fundamental hypothesis relating degree of concealment to physical
health risk.
There are a variety of reasons why concealing
homosexual identity might be associated with an
accelerated course of HIV infection. These analyses
were motivated by experimental research indicating
that psychological inhibition can influence healthrelevant physiological parameters (e.g., specific aspects of sympathetic nervous system activity and
immune function) (9-13, 20-22, 24-25). To the
extent that living in the closet requires gay men to
inhibit the expression of subjectively significant
thoughts, feelings, and social behaviors, the present
results are consistent with hypotheses suggesting a
causal influence of psychological inhibition on
physical health. However, the present data come
from a nonexperimental analysis and are thus
equally consistent with other possible hypotheses
that do not imply any causal influence of concealment on physical health.
Psychosomatic Medicine 58:219-231 (1996)
For example, gay men who are in the closet may
differ from those who are out by possessing a syndrome of neurophysiological and psychosocial characteristics known as "inhibited temperament" (7273). Previous research on the inhibited temperament
syndrome has identified components that could influence both immune function (e.g., heightened activity in the sympathetic nervous system) and the
social presentation of homosexual identity (e.g.,
heightened sensitivity to others and their evaluation
of the self) (72-73). Both of these characteristics are
believed to be associated with individual differences
in activity in specific central nervous system structures (e.g., the amygdala and septo-hippocampal
system) (72-74). Under this hypothesis, concealing
vs. expressing a homosexual identity would not be
expected to influence physical health outcomes unless it was accompanied by changes in central nervous system activity. This neurophysiological-temperamental hypothesis is clearly reminiscent of
Galen's constitutional explanation for the association between psychological inhibition and physical
illness (8, 73).
Closeted gay men may also differ from those who
are out in aspects of health-relevant behavior that
could potentially influence HIV progression (e.g.,
drug use, sleep or exercise habits, sexual activities).
At study entry, concealment was not confounded by
antiretroviral treatment or recreational drug use (including cigarette smoking and heavy alcohol consumption). Concealment was weakly associated with
higher rates of unprotected anal-receptive intercourse, higher rates of sleep disturbance, and lower
rates of exercise. However, these variables were
controlled in all disease progression analyses, and
concealment remained a significant predictor of HIV
progression despite such controls. In ancillary analyses (not reported here), we also considered the
possibility that changes in behavior over the course
of the study might have differed for those who were
in vs. out of the closet. Random coefficients regression models examining patterns of behavior over
time (75) suggested no significant relationship between concealment and changes in recreational drug
use, sexual behavior, sleep disruption, or exercise.
Thus differential changes in health-relevant behavior do not appear to account for the present results.
Variations in social support are known to correlate
with many physical health outcomes (76), and concealing homosexual identity might reduce the opportunity for, or effectiveness of, various types of
social support. In this sample, most participants
reported moderate to high levels of social support,
and degree of concealment did not correlate signifi227
S. W. COLE et al.
cantly with primary partner status, social isolation,
subjective loneliness, or any dimension of AIDSspecific subjective social support (desired, received,
desired — received deficit, or satisfaction with received support). Although some aspects of social
support were inconsistently associated with times to
some studied end points,2 concealment continued to
predict disease progression despite control for all
measures of social support. Caumartin and colleagues [77] have suggested that low levels of involvement in the gay community may predict accelerated mortality among gay men who have already
developed AIDS. However, ancillary analyses of the
present data failed to indicate any association between concealment and measures of involvement in
the gay community or the number or fraction of
friends who were homosexual or bisexual. These
results were not surprising given the partial nature of
concealment in this sample and given that this
sample was recruited from the gay community in
west Los Angeles. In other samples, variations in
social support may be more marked than in this
sample and thus more likely to correlate with physical health outcomes. In addition, aspects of social
support not measured in this study (e.g., antagonistic
"negative social support," participation in support
groups for HIV-infected individuals) could potentially relate to the effects observed here.
Some evidence has also suggested that affective
states such as depression may relate to indices of
HIV progression (e.g., declining CD4 levels) (70),
although other studies have failed to find such
effects (71). Levels of depression and anxiety were
not high in this sample, and they did not correlate
with either degree of concealment or HIV progression. Concealment was associated with the presence
of a repressive coping style, but repressive style itself
did not predict disease progression, and concealment continued to predict HIV progression despite
control for differences in repressive style.
Closeted gay men may be more reluctant than
those who are out to engage in behaviors that might
identify their homosexual orientation to others. This
could bias decisions to participate in a study of a
homosexuality-linked disease such as HIV infection.
Although such biases might reduce the overall rate
of participation among closeted individuals, this
alone would not induce a correlation between concealment and disease progression among those who
do elect to participate. Selective nonparticipation by
those who subsequently showed relatively slow disease progression could induce such a correlation.
However, such an effect would require participants
to have information about their likely disease course
228
at study entry, and no such information was available to the individuals analyzed here (all were AIDSfree with normal CD4 levels at study entry, and none
was aware of his HIV serostatus).
Reluctance to be identified with a homosexualitylinked disease might also lead closeted participants
to delay or forego medical treatments linked to HIV
infection (e.g., antiretroviral therapy). Concealment
was not significantly associated with either time
from study entry to, or CD4 level at the commencement of antiretroviral therapy, and concealment continued to predict HIV progression despite control for
antiretroviral use as a time-varying covariate. Potential delays in implementing other forms of treatment
(e.g., prophylaxis against opportunistic infections,
dietary changes, homeopathic or other "alternative
therapies") could not be evaluated because these
treatments were not measured in the present study.
However, the lack of correlation between concealment and antiretroviral use suggests that closeted
participants did not experience a generalized reluctance to seek appropriate medical treatment.
Advancing HIV infection could conceivably influence a gay man's ability to conceal his homosexuality. However, such an effect would induce a negative
relationship between disease progression and concealment, rather than the positive association observed in these data. In addition, concealment was
measured prospectively in a concurrently healthy
group of individuals, rendering it unlikely that overt
symptomatology would influence degree of concealment in this analysis. Consistent with this conclusion, we found no significant relationship between
concealment and either CD4 levels or HIV-related
symptomatology at NHAPS entry (when concealment was assessed)3.
The structure of this sample and the variables
controlled ruled out a variety of potential explanations for the accelerated course of HIV infection
observed in closeted participants (e.g., differences in
anxiety, depression, repressive coping, positive social support, unsafe sexual practices, demographic
characteristics, health practices, recreational drug
use, antiretroviral treatment, participation biases
stemming from reluctance to be identified with a
3
Negligible correlations between concealment and CD4 levels
at study entry are not surprising because analyses were restricted
to individuals who were healthy at study entry. However, the
differential advance of HIV infection did induce significant negative correlations between concealment and CD4 levels during the
later periods of the study.
Psychosomatic Medicine 58:219-231 (1996)
CONCEALED HOMOSEXUALITY
homosexuality-linked disease, and causal influences
of disease progression on concealment). Several
other hypotheses could not be fully evaluated because these data contained no measures of putatively
influential constructs (e.g., concealment-induced
psychological inhibition, neurophysiological characteristics of "inhibited temperament," "negative
social support," support group utilization). As a
result, the mechanisms linking concealed homosexual identity to accelerated HIV progression remain
unclear. Until such mechanisms are identified and
the present findings replicated, these results should
be interpreted with great caution. In particular, two
important limitations inherent in the epidemiological design of the present study need to be borne in
mind.
First, these results are based on a study of relatively affluent, predominantly white, and initially
healthy volunteers from the west Los Angeles metropolitan area. Although this sample is representative of all HIV-seropositive Los Angeles MACS participants on MACS-assessed variables and is
representative of healthy NHAPS participants on
NHAPS-assessed variables, it is unknown whether
Los Angeles MACS/NHAPS participants are representative of gay men in general. Many researchers
have questioned whether there even exists a population of "gay men in general" (as opposed to a
multitude of subpopulations defined by geographic
and demographic criteria). Even if such a population
does exist, it would be impossible to evaluate sampling representativeness because stigmatization prevents the establishment of any valid sampling frame
for homosexual men (38). Although the distribution
of concealment scores in this sample was comparable to that found in other recent samples of gay men
in urban settings (38), it still remains to be determined how these results might generalize to other
groups of gay men.
Second, the observational nature of this study
does not permit any causal conclusions about the
mechanism relating concealed homosexual identity
to accelerated HIV progression. In particular, these
results do not imply that "coming out of the closet"
will improve physical health. Differences between
individuals cannot inform us about the effects of a
single individual's change in status. Even if changes
in the public expression of homosexual identity
were shown to affect health within individuals, any
potential benefit would constitute only one of many
factors to consider in determining when and to
whom one should express aspects of homosexual
identity. However, these data do raise the possibility
that gay men may suffer significant health costs as
Psychosomatic Medicine 58:219-231 (1996)
they attempt to appease societal prejudice by concealing their sexual orientation. Whether such costs
might be offset by potential benefits of concealment
(e.g., sparing gay men stigmatization, ostracism, or
physical assault) is an important consideration and a
potentially fruitful topic for further research. However, these data suggest that the psychological, behavioral, and biological correlates of social identity
may be relevant to understanding individual differences in physical health and illness.
SUMMARY
On the basis of hypotheses linking psychological
inhibition to physical illness, we examined the rate
at which HIV infection progressed in 80 gay men
who varied in the extent to which they publicly
expressed their homosexual identity. Those who
were "half" or "more" in the closet reached a critically low level of CD4 T lymphocytes, AIDS diagnosis, and AIDS mortality significantly faster than did
those who were "mostly" or "completely out of the
closet." The accelerated course of HIV infection in
closeted gay men could not be explained by differences in disease progression at study entry, medical
treatment, HIV-relevant sexual practices, health-relevant behavioral practices (e.g., sleep disruption,
exercise, recreational drug use), anxiety, depression,
social support, or repressive coping style. These
results parallel those of a similar study of 222 HIVseronegative gay men documenting a heightened
incidence of infectious and neoplastic diseases as
degree of concealment increased. Both findings are
consistent with previous research linking psychological inhibition to alterations in measures of sympathetic nervous system activation, immune function,
and physical health.
This research was supported by funding from the
National Institute of Mental Health (MH42918) and
the National Institute of Allergy and Infectious Diseases (N01-A1-72631). Steve W. Cole was supported
in part by NIMH Grant MH 15750, and Margaret E.
Kemeny was supported in part by the NIMH Research Scientist Development Award MH 00820.
Special thanks to Joanie Chung, MPH, and Mei Ling
Shen, MPH, for their help in data management and
to all MACS/NHAPS participants for their continual
support and contribution.
REFERENCES
1. Levy JA: Pathogenesis of human immunodeficiency virus
infection. Microbiol Rev 57:183-289, 1993
229
S. W. COLE et al.
Phair J, Jacobson L, Detels R, et al: Acquired Immune Deficiency Syndrome within 5 years of infection with Human
Immunodeficiency Virus Type-1: The Multicenter AIDS Cohort Study. J Acquir Immun Defic Syndrome 5:490-496,1992
Sheppard HW, Lang W, Ascher MS, et al: The characterization
of non-progressors: Long term HIV-1 infection with stable
CD4+ T-cell levels. AIDS 7:1159-1166, 1993
Munoz A, Wang MC, Bass S, et al: Acquired Immunodeficiency Syndrome (AIDS)-free time after Human Immunodeficiency Syndrome Virus Type 1 (HIV-1) seroconversion in
homosexual men. Multicenter ADDS Cohort Study Group. Am
J Epidemiol 130:530-539, 1989
Detels R, English PA, Giorgi JV, et al: Patterns of CD4+ cell
changes after HIV-1 infection indicate the existence of a
codeterminant of AIDS. J Acquir Immun Defic Syndrome
1:390-395, 1988
Kemeny ME: Stressful events, psychological responses, and
progression of HIV infection. In Glaser R, Kiecolt-Glaser J
(eds), Handbook of Human Stress and Immunity. New York,
Academic Press, 1994
Kemeny ME: Psychoneuroimmunology of HIV infection. Psychiatr Clin North Am 17:55-68, 1994
Siegel RE: Galen's System of Physiology and Medicine. New
York, Karger, 1968
Pennebaker JW: Overcoming inhibition: Rethinking the roles
of personality, cognition, and social behavior. In Traue HC,
Pennebaker JW (eds), Emotion, Inhibition and Disease. Kirkland, WA, Hogrefe & Huber, 1993, 100-115
Gross JJ, Levinson RW: Emotional suppression: Physiology,
self-report, and expressive behavior. J Pers Soc Psychol 64:
970-986, 1993
Koriat A, Melkman R, Averill JA, Lazarus R: The self-control
of emotional reactions to a stressful film. ] Pers 40:601-619,
1972
Pennebaker JW, Chew CH: Behavioral inhibition and electrodermal activity during deception. J Pers Soc Psychol 49:14271433, 1985
Wegner DM, Shortt JW, Blake AW, et al: The suppression of
exciting thoughts. J Pers Soc Psychol 58:409-418, 1990
Crary B, Hauser SL, Borysenko M, et al: Epinephrine-induced
changes in the distribution of lymphocyte subsets in the
peripheral blood of humans. J Immunol 131:1178-1181,1983
Felten DL, Felten SY, Bellinger DL, et al: Noradrenergic
sympathetic neural interactions with the immune system:
Structure and function. Immunol Rev 100:225-260, 1987
Ottaway CA, Husband AJ: Central nervous system influences
on lymphocyte migration. Brain Behav Immun 6:97-116,
1992
Roszman TL, Carlson SL: Neurotransmitters and molecular
signals in the immune response. In Ader R, Felten DL, Cohen
N (eds), Psychoneuroimmunology, 2nd Edition. San Diego,
Academic Press, 1991, 311-338
Sanders VM, Munson AE: Norepinephrine and the antibody
response. Pharmacol Rev 37:229-248, 1985
Stein M, Schliefer SJ, Keller SE: Hypothalamic influences on
immune response. In Ader R (ed), Psychoneuroimmunology.
Orlando, FL, Academic Press, 1981, 429-448
Pennebaker JW, Hughes C, O'Heeron RC: The psychophysiology of confession: Linking inhibitory and psychosomatic
processes. J Pers Soc Psychol 52:781-793, 1987
Pennebaker JW, Kiecolt-Glaser JK, Glaser R: Disclosure of
traumas and immune function: Health implications for psychotherapy. J Consult Clin Psychol 56:239-245, 1988
Esterling BA, Antoni MH, Fletcher MA, et al: Emotional
230
disclosure through writing or speaking modulates latent
Epstein-Barr virus reactivation. J Consult Clin Psychol 62:
130-140, 1994
23. Francis ME, Pennebaker JW: Putting stress into words: The
impact of writing on physiological, absentee, and self-reported well-being measures. Am J Health Promo 6:280-287,
1992
24. Pennebaker JW, Beall SK: Confronting a traumatic event:
Toward an understanding of inhibition and disease. J Abnorm
Psychol 95:274-281, 1986
25. Booth RJ, Petrie KJ, Pennebaker JW, et al: Disclosure of trauma
and the immune response to hepatitis B vaccination. Abstract
presented at the Meeting of the Psychoneuroimmunology
Research Society, Key Biscane, FL, 1994
26. Contrada RJ, Leventhal H, O'Leary A: Personality and health.
In Pervin LA (ed), Handbook of Personality: Theory and
Research. New York, Guilford Press, 1990, 638-669
27. Gross J: Emotional expression in cancer onset and progression. Soc Sci Med 28:1239-1248, 1989
28. James SA: Psychosocial precursors of hypertension: A review
of the epidemiologic literature. Circulation 76(Suppl l):I60166, 1987
29. Sommers-Flanagan J, Greenberg RP: Psychosocial variables
and hypertension: A new look at an old controversy. J Nerv
Ment Dis 177:15-24, 1989
30. Solomon GF, Moos RH: Emotions, immunity, and disease: A
speculative theoretical integration. Arch Gen Psychiatry 11:
657-674, 1964
31. Herek GM: Hate crimes against lesbians and gay men. Am
Psychol 44:948-955, 1989
32. Weinberg MS, Williams CJ: Male Homosexuals: Their Problems and Adaptations. New York, Oxford University Press,
1974
33. Bell AP, Weinberg MS: Homosexualities: A Study of Diversity
Among Men and Women. New York, Simon and Schuster,
1978
34. Kaslow RA, Ostrow DG, Detels R, et al: The multicenter AIDS
cohort study: Rationale, organization, and selected characteristics of the participants. Am J Epidemiol 126:310-318, 1987
35. Kiecolt-Glaser JK, Glaser R: Methodological issues in behavioral immunology research with humans. Brain Behav Immun
2:67-78, 1988
36. Taylor SE, Kemeny ME, Aspinwall LG, et al: Optimism,
coping, psychological distress, and high-risk sexual behavior
among men at risk for acquired immunodeficiency syndrome
(AIDS). J Pers Soc Psychol 63460-473, 1992
37. Kemeny ME, Weiner H, Taylor SE, et al: Repeated bereavement, depressed mood, and immune parameters in HIV seropositive and seronegative homosexual men. Health Psychol
13:14-24, 1994
38. Martin JL, Dean L: Developing a community sample of gay
men for an epidemiologic study of AIDS. Am Behav Sci
33:546-561, 1990
39. Radloff LS: The CES-D scale: A self-report depression scale
for research in the general population. Appl Psychol Meas
1:386-401, 1977
40. McNair DM, Lorr M, Droppleman LF: Profile of Mood States.
San Diego, Educational and Industrial Testing Service, 1980
41. Bendig AW: The development of a short form of the Manifest
Anxiety Scale. J Consult Psychol 20:384, 1956
42. Russell D, Peplau LA, Cutrona CE: The revised UCLA Loneliness Scale: Concurrent and discriminant validity evidence. J
Pers Soc Psychol 39:472-480, 1980
43. Schwarzer R, Dunkel-Schetter C, Kemeny ME: The multidi-
Psychosomatic Medicine 58:219-231 (1996)
CONCEALED HOMOSEXUALITY
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
mensional nature of received social support in gay men at risk
of HIV infection and AIDS. Am J Commun Psychol 22:319339, 1994
Strahan R, Gerbasi KC: Short, homogeneous versions of the
Marlowe-Crowne Social Desirability Scale. J Clin Psychol
28:191-193, 1972
Weinberger DA, Schwartz GE, Davidson RE: Low anxious,
high anxious and repressive coping styles: Psychometric
patterns and behavioral and physiological responses to stress.
J Abnorm Psychol 88:369-380, 1979
Nishanian P, Taylor JMG, Korns E, et al: Significance of
quantitative enzyme-linked immunosorbent assay (ELISA)
results in evaluation of three ELISAs and Western blot tests
for detection of antibodies to human immune deficiency virus
in a high risk population. J Clin Microb 25:395-400, 1987
Centers for Disease Control: Revision of the CDC surveillance
case definition for Acquired Immunodeficiency Syndrome.
MMWR Morb Mortal Wkly Rep 36:lS-15S, 1987
Giorgi JV, Cheng H-L, Margolick JB, et al: Quality control in
the flow cytometric measurement of T-lymphocyte subsets:
The Multicenter Aids Cohort Study (MACS) experience. Clin
Immunol Immunopathol 55:173-186, 1990
Taylor JMG, Fahey JL, Detels R, et al: CD4 percentage, CD4
number, and CD4: CD8 ratio in HIV infection: Which to
choose and how to use. ] Acquir Immun Defic Syndrome
2:114-124, 1989
Burcham J, Marmor M, Dubin N, et al: CD4% is the best
predictor of development of AIDS in a cohort of HIV-infected
homosexual men. AIDS 5:365-372, 1991
Saah AJ, Hoover DR, He Y, et al: Factors influencing survival
after AIDS: Report from the Multicenter AIDS Cohort Study
(MACS). J Acquir Immun Defic Syndrome 7:287-295, 1994
Fahey JL, Taylor JMG, Detels R, et al: The prognostic value of
cellular and serologic markers in infection with human immunodeficiency virus type 1. N Engl J Med 322:166-172,
1990
Phillips AN, Lee CA, Elford J, et al: Serial CD4 lymphocyte
counts and development of AIDS. Lancet 337:389-392, 1991
Phillips AN, Lee CA, Elford J, et al: The cumulative risk of
AIDS as the CD4 lymphocyte count declines. J Acquir Immun
Defic Syndrome 5:148-152, 1992
Kalbfleisch JD, Prentice RL: The Statistical Analysis of Failure
Time Data. New York, Wiley, 1980
Miller RG: What price Kaplan-Meier? Biometrics 39:10771081, 1983
Aboulker JR, Swart AM: Preliminary analysis of the Concorde
Trial. Lancet 341:889-890, 1993
Cole SW, Kemeny ME, Taylor SE, Visscher BR: Elevated
physical health risk among gay men who conceal their homosexual identity. Health Psychol, in press
Solano L, Costa M, Salvati S, et al: Psychosocial factors and
clinical evolution in HIV-l infection: A longitudinal study.
J Psychosom Res 37:39-51, 1993
Kemeny ME, Weiner H, Duran R, et al: Immune system
changes following the death of a partner in HIV positive gay
men. Psychosom Med, in press
Goodkin K, Blaney NT, Feaster D, et al: Active coping style is
associated with natural killer cell cytotoxicity in asymptom-
Psychosomatic Medicine 58:219-231 (1996)
atic HIV-l seropositve homosexual men. J Psychosom Res
36:635-650, 1992
62. Goodkin K, Fuchs I, Feaster D, et al: Life stressors and coping
style are associated with immune measures in HIV-l infection: A preliminary report. Int J Psychiatry Med 22:155-172,
1992
63. Antoni MH, Baggett L, Ironson G, et al: Cognitive-behavioral
stress management intervention buffers distress responses
and immunologic changes following notification of HIV-l
seropositivity. J Consult Clin Psychol 59:906-915, 1991
64. Esterling BA, Antoni MH, Schneiderman N, et al: Psychosocial modulation of antibody to Epstein-Barr viral capsid
antigen and human herpesvirus type-6 in HIV-l infected and
at-risk gay men. Psychosom Med 54:354-371, 1992
65. Kemeny ME, Reed GM, Taylor S, et al: Negative HIV-specific
expectancies predict immunologic evidence of HIV progression. Health Psychol, in press
66. Reed GM, Kemeny ME, Taylor SE, et al: Negative HIV-specific
expectancies and AIDS-related bereavement as predictors of
symptom onset in asymptomatic HIV seropositive gay men.
Health Psychol, in press
67. Reed GM, Kemeny ME, Taylor SE, et al: "Realistic acceptance" as a predictor of survival time in gay men with AIDS.
Health Psychol 13:299-307, 1994
68. Soloman GF, Temoshok L, O'Leary A, et al: An intensive
psychoimmunologic study of long-surviving persons with
AIDS. Ann N Y Acad Sci 496:647-655, 1987
69. Ironson G, Friedman A, Klimas N, et al: Distress, denial, and
low adherence to behavioral interventions predict faster disease progression in gay men infected with Human Immunodeficiency Virus. Int J Behav Med 1:90-105, 1994
70. Burack JH, Barrett DC, Stall RD, et al: Depressive symptoms
and CD4 lymphocyte decline among HIV-infected men. J Am
Med Assoc 270:2568-2573, 1993
71. Lyketsos CG, Hoover DR, Guccione M, et al: Depressive
symptoms as predictors of medical outcomes in HIV infection. J Am Med Assoc 270:2563-2567, 1993
72. Kagan J: The concept of behavioral inhibition to the unfamiliar. In Reznick JS (ed), Perspectives on Behavioral Inhibition.
Chicago, University of Chicago Press, 1989, 1-23
73. Kagan J: Galen's Prophecy: Temperament in Human Nature.
New York, Basic Books, 1994
74. Gray JA: The neuropsychology of temperament. In Strelau J,
Angleitner A (eds), Explorations in Temperament: International Perspectives on Theory and Measurement. New York,
Plenum Press, 1991, 105-128
75. Gibbons RD, Hedeker D, Elkin I, et al: Some conceptual and
statistical issues in analysis of longitudinal psychiatric data.
Arch Gen Psychiatry 50:739-750, 1993
76. Berkman LF: The role of social relations in health promotion.
Psychosom Med 57:245-254, 1995
77. Caumartin S, Joseph JG, Chmiel J: Premorbid psychosocial
factors associated with differential survival time in AIDS
patients. Abstract presented at the VII International Conference on AIDS, Florence, Italy, 1991
78. Efron B, Tibshirani RJ: An Introduction to the Bootstrap. New
York, Chapman and Hall, 1993
231