| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychiatry (J.G.R., S.J.F., M.S., M.M.), Weill College of Medicine, Cornell University, Ithaca; and New York State Psychiatric Institute (J.G.R., S.-H.L.), New York, NY.
Address reprint requests to: Judith G. Rabkin, PhD, New York State Psychiatric Institute, Unit 51, 1051 Riverside Dr., New York, NY 10032. Email: jgr1{at}Columbia.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: One hundred seventy-three HIV+ gay or bisexual men with symptomatic HIV illness (40% nonwhite) were evaluated semiannually in a university-affiliated research program between July 1995 and December 1997. The primary outcome measures were the Structured Clinical Interview for DSM-IV, Beck Depression Inventory, Endicott Quality of Life Enjoyment and Satisfaction Questionnaire, and Beck Hopelessness Scale.
RESULTS: Psychological distress in this sample was mild to moderate at baseline. During the first 2 years that highly active antiretroviral therapy became widely available, we observed a statistically significant but clinically modest reduction in distress in the sample as a whole, with significant covariates of CD4 cell count, HIV symptoms, and social support in a mixed-effects model. Rates of clinical depression declined. However, this generalized mental health improvement was not related to individual medical improvement of markers of HIV illness progression; those classified as improved were no more likely than those who remained unimproved to report greater declines in measures of distress and hopelessness. Number of self-reported physical symptoms were directly related to distress levels.
CONCLUSIONS: A cohort effect was observed, with overall psychological improvement. Physical symptoms were more strongly related to psychological distress than were laboratory markers. Consequently, those whose CD4 cell count and HIV RNA viral load reflected successful treatment were no more likely than others to be relieved of the psychological burdens of illness.
Key Words: psychiatry • psychosocial • highly active antiretroviral therapy • naturalhistory • quality of life
Abbreviations: AIDS = acquired immune deficiency syndrome; BDI = BeckDepression Inventory; BHS = Beck Hopelessness Scale; CD4 =subset of lymphocytes used as a marker of immunosuppression in HIVdisease; DSM-IV = Diagnostic and Statistical Manual ofMental Disorders, fourth edition; HAART = highly activeantiretroviral therapy; HAM-D = Hamilton Rating Scale forDepression; HIV = human immunodeficiency virus; HIV RNA = HIVribonucleic acid; the assay measures number of copies of virus per unitof blood (plasma); QOL = Endicott Quality of Life Enjoyment andSatisfaction Questionnaire; SCID = Structured Clinical Interviewfor DSM-IV.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
A decade ago, several studies of community volunteers were initiated to assess psychological adaptation to HIV illness over time, with the assumption that the inevitably progressive nature of HIV/AIDS was likely to serve as a natural experiment in the unfolding of psychopathology (1). Surprisingly, this was not generally observed (see Ref. 2 for review). Among cross-sectional studies comparing HIV+ subjects with no or mild medical symptoms and HIV-seronegative subjects from the same communities, no systematic differences in rates of major psychiatric disorders were found (3–8). Some cross-sectional studies that included subjects with AIDS reported higher rates of major depression in that subgroup compared with asymptomatic HIV+ or HIV-seronegative subjects (9–11), but others did not (12). Overall, measures of distress (although not syndromal disorders) increased in the presence of HIV symptoms but not in relation to laboratory markers of illness stage.
In longitudinal studies of HIV+ samples, several investigators found stable rates of distress and depressive disorder over time despite illness progression (13–17), although others (eg, Ref. 18) observed increased distress in the year before AIDS-defining conditions developed. With the exception of the latter study, however, the HIV+ study participants were either asymptomatic or had mild to moderate symptoms at study entry, and many did not experience late-stage illness during the periods of observation. In July 1995, our group at Cornell sought to clarify and extend these findings by initiating a longitudinal study of HIV+ men with symptomatic HIV illness and AIDS-defining conditions at baseline, using a similar measurement paradigm. The aim was to determine whether distress increased over time in a sicker cohort facing the likelihood of death in the next 2 or 3 years.
Saquinavir, the first of a new class of HIV-1 protease inhibitor medications, became available on the market in December 1995 and was swiftly followed by two other protease inhibitors, ritonavir and indinavir. These agents, combined with previously available and newer nucleoside reverse transcriptase inhibitors (eg, zidovudine and lamivudine) and nonnucleoside reverse transcriptase inhibitors (eg, nevirapine), had a prompt and profound effect on HIV-related morbidity and mortality rates (19, 20). In the ensuing months, it became apparent that, instead of facing death, many HIV+ patients now faced the prospect of extended survival and perhaps even a "second life" (21). Accordingly, our aim shifted to comparing the psychological consequences of improved health for the majority of study participants whose viral load declined, CD4 cell counts increased, and whose opportunistic infections remitted with changes in mood status among those whose health did not improve. Therefore, our study might be considered an investigation into the mood effects of illness "regression," in contrast to our earlier studies of the effects of illness progression.
The original studies were designed to investigate the widely held but undocumented assumption, of both patients and physicians, that people with terminal illnesses are apt to become clinically depressed, especially as death approaches. A corollary belief, heard anecdotally in the past 2 years, is that people will feel happier with improved health, especially those who make dramatic recoveries from formerly life-threatening conditions, sometimes referred to as the "Lazarus effect." Serendipitously, we found ourselves in a position to evaluate this question.
The current report is based on five semiannual assessments of medical, psychosocial, and psychiatric parameters in a group of gay and bisexual men with symptomatic HIV/AIDS who enrolled in this longitudinal study in 1995. We addressed the following questions: 1) Do rates of psychiatric disorder and distress and severity of depressive symptoms change over time for the cohort as a whole? 2) Do rates of distress differentially decline over time among men who experience improving health compared with those whose health remains unimproved? To address the latter question, we used three strategies, first using standard medical definitions of improvement based on magnitude of change in CD4 cell count and HIV RNA viral load. Next, we compared subjects whose viral load had become undetectable with those whose viral load remained detectable, on the assumption that this definition of improvement might be more powerful psychologically. Finally, we reversed strategies and compared the trajectory of change in these biological markers for men who were most and least distressed and hopeless at their final study visit.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Measures
The outcome variables of interest were measures of depression, life satisfaction, and hope. Predictor variables included in these analyses were shown in previous research or hypothesized to be associated with these outcome variables and represent the following categories: social support; laboratory markers of disease progression, and HIV medical symptoms.
Psychiatric assessments.
Lifetime and current (past month) axis I disorders were assessed using the SCID (22), which was administered at visits 1, 3, and 5. We included depression, anxiety, and substance use disorder modules and the psychotic screen.
Severity of depressive symptoms was assessed with the 21-item self-reported BDI (23) and the clinician-rated HAM-D, both of which are considered the "gold standard" among scales of depressive severity in their class. On the BDI, scores range from 0 to 63. A score of 13+ is recommended for research purposes as a cutoff point to indicate significant depressive symptoms (24). The 17-item structured version of the HAM-D (25) was administered as an independent assessment of depressive severity. Scores range from 0 to 52, with scores of 8 or less generally considered to signify the absence of depression and scores of 14 or more considered to reflect significant depressive symptoms.
To avoid confounding of somatic depressive symptoms, such as fatigue, with those induced by either HIV or medications used to treat HIV-related conditions, we calculated separately the affective and somatic subscales as well as the total scores of these two measures of depressive severity.
Two other self-reported scales addressing psychological dimensions were used. The BHS (26) is a 20-item true/false questionnaire that elicits outlook toward the future (score range: 0–20, with higher scores indicating greater hopelessness). Scores of 9 or more indicate moderate or greater hopelessness (27), whereas scores of 3 or less indicate absence of hopelessness. The short-form QOL (28) asks the respondent to rate 14 domains of everyday living (eg, leisure time activities and social relationships) on a five-point satisfaction scale (1 = very poor, 5 = very good). The time frame is 2 weeks, and the total score is the sum of items.
Finally, clinicians rated social support on a five-point scale based on the model of axis IV of DSM-IV. This was done at the end of each visit, incorporating all information elicited during the interview. This scale was found to be significantly correlated (p < .001) with scores on the 18-item self-reported Wortman Social Support Scale (29), which was administered simultaneously at one study visit and has been used in other studies of HIV+ subjects.
Laboratory markers.
CD4 cell count and HIV RNA viral load (determined by polymerase chain reaction) were measured after every study visit at a local commercial laboratory (Quest, formerly Metpath). For the HIV RNA assay, the lower limit of detection was 400 copies per milliliter of serum. Undetectable test results were conservatively scored as 399 (2.6 log) copies.
HIV-related symptoms.
HIV-related symptoms were assessed with a 14-item scale of signs and symptoms commonly associated with HIV illness, such as fatigue, night sweats, unintended weight loss, and oral candidiasis. This scale has been used in other HIV+ cohort studies (16, 30). In addition, because restricted mobility often characterizes symptomatic HIV illness, we administered the Physical Limitations Scale from the RAND Medical Outcomes Study (31), which lists 10 activities in descending order of difficulty. Subjects respond "Yes, I can do this"; "yes, but only slowly"; or "no, I can’t do this." Items are summed to obtain a total score, with higher scores indicating more physical limitations.
Procedure
Each participant was seen for about 3 hours, during which clinician-administered interviews were conducted and self-reported measures were completed. (After the baseline visit, self-report forms were mailed to participants the week before their next visit and brought to the visit.) Subjects then went to a nearby commercial laboratory for blood tests. Interviewers were postdoctoral fellows in the National Institute of Mental Health–sponsored AIDS Research Training Program. Reliability of psychiatric diagnoses was established in biweekly meetings with an expert diagnostician (L. Jacobsberg, MD, PhD), and medical information was elicited under the supervision of a physician who worked in the hospital HIV/AIDS clinic (S.J.F.). The study protocol was approved by the institutional review board of Cornell University Medical College (now Weill), and all subjects gave written informed consent. They were paid $30 for each visit.
Statistical Analyses
We examined the variable distributions with respect to skewness and kurtosis and found that four of the six major variables were significantly skewed. The following transformations were therefore performed: CD4 cell count and BDI and HAM-D scores were converted to square roots, and log base 10 was used in calculations of HIV RNA viral load. With these transformations, all scales were normalized. The skew ranged from -0.1 to 0.578, and kurtosis ranged from -0.03 to -1.19 for the lowest and highest values. The QOL scores were unchanged, as were the covariates.
Cross-sectional comparisons were made using t tests, 
2 tests, and analysis of variance. The longitudinal data were fitted to mixed-effect models (32) by using the SAS Institute’s mixed procedure, which is composed of fixed effect and random effect (33). The data were fitted to the model with random effects of intercept and time and unstructured covariance of random coefficients and constant error variances. The maximum likelihood method was used for the estimation. A Type 3 F test was used in significance testing (34).
In these analyses, all available data were used, with no estimations for missing data. Subjects with missing data were retained in the analyses.
The explanatory variables used in the first set of analyses (N = 173), in which the dependent variables were depressive symptoms, hopelessness, and quality of life, include time elapsed since baseline, CD4 cell count, HIV RNA viral load, physical limitations, HIV symptoms, and social support. Age and ethnicity had no effect on outcomes and therefore are not included in the data presentations.
The first set of analyses was performed to determine whether there were significant changes over time for the sample as a whole. The second set of analyses evaluated change in mood measures for subjects classified as either improved or unimproved over time (N = 161). Improvement was defined as either an increase in CD4 cell count of at least 100 cells/mm3 or a decrease in HIV RNA viral load of at least 0.5 log (adapted from antiretroviral guidelines, Ref. 35). Using these criteria, 105 (65%) were classified as improved. The other 56 men (35%) either showed little or no improvement or worsening of these markers. Covariates included in these analyses were baseline measures of physical manifestations of HIV illness (eg, mobility limitations and symptoms) and social support.
The next analysis used a more stringent definition of improvement, requiring HIV RNA viral load to have reached an undetectable level (<400 copies) at their last visit. In this classification, only 50 men (31%) were improved (viral load undetectable), whereas 111 men (69%) continued to have detectable virus.
Finally, we grouped men into quartiles on the basis of level of depressive symptoms or hopelessness at their final study visit and then compared the upper and lower quartiles in terms of their immunologic markers at that visit and across all prior visits. By using the highest quartile of depression, we identified subjects whose BDI score was 13 or higher, indicating significant depressive symptoms. For hopelessness, the highest quartile scores were 7.5 or higher, all in the range of moderate or greater hopelessness.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Medical characteristics.
At baseline, mean CD4 cell count was 242 cells/mm3 (SD = 223, range = 3–1226). Mean number of viral copies was 398,727 (SD = 1,018,238, range = undetectable to >7 million), and mean log viral load was 4.89 (SD = 0.85). Six men had an undetectable viral load (<400 copies). Sixty-one percent had an AIDS diagnosis (based on 1993 Centers for Disease Control and Prevention criteria). Ten men were taking a protease inhibitor (in clinical trials) at baseline, and 51 additional men were taking two antiretroviral medications not including a protease inhibitor.
Over time, the proportion receiving HAART (defined as a protease inhibitor plus two or more other antivirals) rose to 67%. Another 26 men (15%) were taking two or more nucleoside or nonnucleoside reverse transcriptase inhibitors excluding protease inhibitors. Mean CD4 cell counts rose from 242 to 333 cells/mm3, and mean values for viral load declined, as did number of deaths between visits, as shown on Table 1. By visit 5, 39% had an undetectable viral load (<400 copies). The incidence of opportunistic infections or AIDS-related cancers, including recurrences, decreased by 50% from the first year to the second (49 in year 1, 25 in year 2).
|
In the assessment of psychological status, we compared BDI and BHS scores for dropouts at their last study visit with those of remaining subjects at the same visit. Although the disparity in sample sizes detracted from statistical power, no statistically significant differences were found (data not shown). With respect to axis I current depression, 4 of the 13 men who died had been diagnosed with dysthymia at an earlier visit; of those who left the study for other reasons, 1 had been diagnosed with dysthymia and 1 with current major depression. Overall, 21% of dropouts (6 of 29) had a diagnosis on one or more occasions, compared with 31% of remaining subjects (45 of 144; 2 = 1.29, df = 1, p = NS).
Psychiatric Status
Lifetime disorders.
As shown in Table 2, 51% had lifetime (including current at baseline) depressive disorders, 15% had lifetime (including current) anxiety disorders (panic disorder, social phobia, or generalized anxiety disorder) and 1% had bipolar disorder. Lifetime substance use disorders were reported by 45%.
|
Question 1: Does the Sample as a Whole Show Change in Mood Measures Over Time?
On average, all measures of distress showed a statistically significant (but clinically modest) decline over time. Because the results were essentially the same for our two measures of depressive severity (BDI and HAM-D), only BDI data are reported in following data analyses to simplify the presentation.
The decline was linear for BDI measures, whereas the BHS and QOL showed both linear and quadratic trends, with the greatest decline occurring between visits 1 and 2 (when the first protease inhibitors were marketed). We conducted the same analyses using the separate affective and somatic subscales of the BDI with the same results; therefore, only data using the total score are presented.
Regression estimates and their significance are shown in Table 3; the outcome measures were the BDI, QOL, and BHS. The time-variant covariates in this mixed model were laboratory markers (CD4 cell count and HIV RNA viral load), physical limitations, HIV symptoms, and social support. In terms of changes on the BDI, all but viral load were statistically significant. On all three outcome measures, the passage of time made a significant contribution to distress reduction. Physical limitations were consistently associated with greater distress, whereas greater social support was consistently associated with greater reduction of distress on all three measures. HIV symptoms were significantly associated with changes on the BDI and QOL. Viral load was not significantly associated with change on any measure, whereas declines in CD4 cell count were associated with less reduction of distress on the BDI and QOL. Direction of effect of the covariates was the same in the other analyses as well.
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This generalized mental health improvement was not related to individual improvement on medical markers of illness progression; those classified as medically improved (using two different definitions of improvement) showed no greater decline in hopelessness or increase in quality of life than those classified as medically improved. Moreover, on our measure of depressive symptoms, the medically improved group showed no change, whereas the medically unimproved group reported fewer symptoms over time. When we looked at those who were most and least depressed and hopeless at visit 5, no difference in medical markers was found over time: for both extreme groups, CD4 cell count rose and viral load declined. However, the presence and number of physical symptoms was directly related to distress levels in all analyses.
Other studies of patients receiving combination antiretroviral therapy also have failed to find improvement in perceived quality of life despite laboratory indices of positive change (36, 37). In contrast, alleviation of symptoms (pain, fatigue, and role function) was associated with numerically better quality-of-life scores in another report (38).
How can we account for the apparent paradox that medically unimproved subjects were, if anything, more likely to show positive psychological change over time compared with medically improved subjects? First, it is important to recall that the improved subjects were significantly more immunosuppressed at baseline. Probably as a result, significantly more of them started protease inhibitor therapy during the course of the study. Although their objective indices of health showed improvement, they had been closer to death and may have remained cautious about their personal chances and futures.
Second, greater change may create a sense of greater uncertainty about the probable magnitude and duration of improvement. In addition, formerly relinquished responsibilities, such as work and financial solvency, become plausible again, possibly raising concerns about goals and life direction that earlier had been set aside. In short, the unanticipated prospect of a "second life" brings with it complex issues that seem largely irrelevant when there seems to be little time left. Prolonged but uncertain survival may also be difficult.
Substantially more improved than unimproved subjects were taking combination antiretroviral therapy, which itself imposes burdens in terms of the unremittingly onerous schedule, the widespread and intermittent side effects, and anxiety about lapses in adherence even among those who make no mistakes. Finally, the literature has consistently shown that depressive symptoms are associated with physical symptoms, not laboratory markers, so that feeling better is associated with improved mood regardless of what the numbers say.
Limitations of this study, apart from the obvious sampling restrictions to non–intravenous-drug-using gay and bisexual men, include the lack of precision inevitably associated with measures conducted only at 6-month intervals. In addition, rates of current syndromal depressive and anxiety disorders were low throughout the period of observation, and scores assessing depressive severity were consequently truncated; perhaps different patterns would emerge among people with more pervasive psychiatric disorders.
In conclusion, our data indicate that the prospects of restored health and extended survival have positively affected mood and outlook among HIV+ people in a general sense. However, those whose laboratory markers reflect successful treatment are no more likely than others to be relieved of the psychological burdens of illness.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication May 3, 1999.
Revision received September 10, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. J. Repetto and J. M. Petitto Psychopharmacology in HIV-Infected Patients Psychosom Med, June 1, 2008; 70(5): 585 - 592. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. W. Burgoyne and D. H. S. Tan Prolongation and quality of life for HIV-infected adults treated with highly active antiretroviral therapy (HAART): a balancing act J. Antimicrob. Chemother., March 1, 2008; 61(3): 469 - 473. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. H. Antoni, A. W. Carrico, R. E. Duran, S. Spitzer, F. Penedo, G. Ironson, M. A. Fletcher, N. Klimas, and N. Schneiderman Randomized Clinical Trial of Cognitive Behavioral Stress Management on Human Immunodeficiency Virus Viral Load in Gay Men Treated With Highly Active Antiretroviral Therapy Psychosom Med, January 1, 2006; 68(1): 143 - 151. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. W. Carrico, M. H. Antoni, K. E. Weaver, S. C. Lechner, and N. Schneiderman Cognitive--behavioural stress management with HIV-positive homosexual men: mechanisms of sustained reductions in depressive symptoms Chronic Illness, September 1, 2005; 1(3): 207 - 215. [Abstract] [PDF]
|
|
|
|
 |
|
 P. N. Halkitis, M. T. Shrem, D. D. Zade, and L. Wilton The Physical, Emotional and Interpersonal Impact of HAART: Exploring the Realities of HIV Seropositive Individuals on Combination Therapy J Health Psychol, May 1, 2005; 10(3): 345 - 358. [Abstract] [PDF]
|
|
|
|
 |
|
 D. S. Storm, M. G. Boland, S. L. Gortmaker, Y. He, J. Skurnick, L. Howland, J. M. Oleske, and for the Pediatric AIDS Clinical Trials Group Proto Protease Inhibitor Combination Therapy, Severity of Illness, and Quality of Life Among Children With Perinatally Acquired HIV-1 Infection Pediatrics, February 1, 2005; 115(2): e173 - e182. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Siegel, D. Karus, and L. Dean Psychosocial Characteristics of New York City HIV-Infected Women Before and After the Advent of HAART Am J Public Health, July 1, 2004; 94(7): 1127 - 1132. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. G. Rabkin, M. McElhiney, S. J. Ferrando, W. Van Gorp, and S. H. Lin Predictors of Employment of Men With HIV/AIDS: A Longitudinal Study Psychosom Med, January 1, 2004; 66(1): 72 - 78. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. F. Morrison, J. M. Petitto, T. T. Have, D. R. Gettes, M. S. Chiappini, A. L. Weber, P. Brinker-Spence, R. M. Bauer, S. D. Douglas, and D. L. Evans Depressive and Anxiety Disorders in Women With HIV Infection Am J Psychiatry, May 1, 2002; 159(5): 789 - 796. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Scalera and N. Shear Use and Citation of Beck Depression Inventory to Assess Depression in HIV Infection Psychosomatics, February 1, 2002; 43(1): 88 - 88. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
