Psychiatric disorders are common among those infected with HIV. Previous studies of HIV-infected patients have shown prevalence rates of psychiatric disorders, including substance abuse disorders, to be nearly 50%,1 with nearly 30% of these patients reporting regularly taking psychotropic medication.2 The most commonly reported psychiatric disorder among HIV clinic populations is major depression, with prevalence estimates ranging between 4% and 18.4%.3 A recent meta-analysis documented that HIV-infected patients may have nearly twice the rate of major depression as HIV-negative patients. The findings of this meta-analysis may be limited by the low rates of depression reported in some of the studies.4
Despite the clear survival benefit associated with its use, highly active antiretroviral therapy (HAART) is not prescribed equally to all HIV-infected subpopulations.5,6 Previous studies have shown that HIV-infected patients with a history of depression were significantly more likely to have a delay in initiation of protease inhibitor (PI) therapy as compared with those without depression.7 Furthermore, clinical concerns about adherence8,9 and the perceived public health risk of the spread of resistant strains of HIV8,10,11 may limit the prescription of these potentially life-saving antiretroviral medications to HIV patients with psychiatric disorders.12
We initially hypothesized that patients with AIDS and a comorbid psychiatric disorder would be less likely to be prescribed HAART and thus at increased risk of mortality as compared with those with AIDS and no comorbid psychiatric disorder. The goal of this study was to investigate, among a group of patients diagnosed with AIDS, whether or not a diagnosis of a psychiatric disorder (1) affected the time to initiation of HAART, (2) predicted the likelihood of being prescribed HAART for at least 6 months and (3) affected survival.
Study Design/Study Population
This was a retrospective cohort study of patients with AIDS who were enrolled and followed at the Johns Hopkins University HIV Clinic between January 1996 and January 2002. All patients had clinically confirmed AIDS on enrollment based on the 1993 revised surveillance case definition issued by the Centers for Disease Control and Prevention (CDC). Patients who were 18 years of age or older with a baseline CD4 count <200 cells/mm3 and who were antiretroviral treatment naive were included in the study. Comprehensive baseline evaluations were performed on all patients on enrollment. Detailed demographic, social, behavioral, and clinical data were collected via structured interviews using standardized forms. The Structured Clinical Interview for Diagnostic and Statistical Manual (DSM-IV) was not used to generate psychiatric diagnoses. Data were abstracted from patients’ charts and from the hospital’s automated databases at baseline and every 6 months by trained monitors using standard data collection instruments employed in our clinic. Items in the data abstraction included all baseline information and information on follow-up diagnoses, treatments, hospitalizations, and death. Medical records from other institutions where patients may have received care were routinely sought. The clinic-based medical record maintains a section for each visit to document prescribed therapy by treatment name, dose, and number of refills. The records are also updated when prescriptions are filled over the telephone or mailed to patients. Information on death was obtained from patients’ charts and from a separate death registry maintained by the clinic. The names of those patients whose vital status was unknown for more than 12 months were searched for in death records of the Maryland Bureau of Vital Records and the National Death Index. Systematic problems were identified and corrected for all patients. Maintenance of our database and use of its contents for analysis of patient outcomes are approved by the Institutional Review Board of the Johns Hopkins University School of Medicine.
Patients were defined as having a current “psychiatric disorder” if they had all the following characteristics: (1) baseline patient report of a history of a psychiatric disorder recorded during the structured enrollment interview; (2) baseline indication and use of psychotropic medication specifically prescribed for psychiatric conditions recorded during the structured enrollment interview; and (3) a psychiatric evaluation, including a nonstructured diagnostic interview, performed by on-site psychiatric consultants affiliated with the clinic. The nonstructured diagnostic interviews by a psychiatrist were only completed if a patient was referred to a psychiatrist. The clinic has 2 psychiatrists who provide on-site delivery of care using collaborative care techniques to enhance interprovider communication. The treatment model used by the on-site psychiatric team has been previously described.13,14
Patients were classified as having “no psychiatric disorder” if they had none of the 3 criteria listed previously. Patients with 1 or 2 of the listed criteria were excluded from the initial analysis but included in subsequent sensitivity analysis. Chart review by a board-certified psychiatrist (S.H.) of a 10% random sample of patient charts based on the previously described classification of psychiatric disorders revealed excellent agreement between patient-reported psychiatric diagnosis and diagnosis confirmed independently by the psychiatric evaluation (κ statistic = 0.92).
For our analysis, HIV transmission risk factors included injection drug use (IDU), men who have sex with men (MSM), and heterosexual transmission, which was defined as heterosexual activity with a partner at high risk for HIV or sex with an HIV-infected patient. An HIV risk factor of “other” was defined when a patient had no known risk factor or a risk factor other than IDU, MSM, or heterosexual transmission. Risk factor assignment was not mutually exclusive, because patients could have multiple HIV risk factors. The definition of HAART was defined as receipt of a multidrug regimen that included 2 nucleoside agents and either a PI or a nonnucleoside reverse transcriptase inhibitor (NNRTI), an NNRT-PI combination, or 3 or more nucleoside agents. Prophylaxis for Pneumocystis jiroveci (formerly Pneumocystis carinii) pneumonia (PCP) was defined as receipt of any anti-Pneumocystis drugs within 6 months of enrollment. Prophylaxis for Mycobacterium avium complex (MAC) was defined as receipt of clarithromycin, azithromycin, or rifabutin within 6 months of enrollment. Enrollment CD4 counts were dichotomized to <50 cells/mm3 or between 50 and 200 cells/mm3.
Univariate distributions included percentages for dichotomous variables, medians for ordinal variables, and means for normally distributed continuous variables. Comparison of means was made using 2-sided t tests, whereas comparison of percentages was made using the χ2 method. The Kaplan-Meier method was used to estimate curves for the length of time to an event, and the log-rank test was used to compare the cumulative incidence curves in the psychiatric disorder and nonpsychiatric disorder groups. Cox proportional hazards regression models were used to estimate relative risks and 95% confidence intervals (CIs) and to adjust for differences in baseline outcome predictors. The Cox proportional hazards model suggests that when the hazard functions are proportional, their ratio is constant over time. Time 0 for all time-dependent analyses was time of enrollment.
Logistic regression models were used to examine the relation between history of a psychiatric disorder and treatment with HAART for ≥6 months. These models were used to estimate relative odds ratios and 95% CIs and to adjust for differences in baseline outcome predictors. Adjusted probabilities were then computed by back-transformation on the basis of the average values of the characteristics of the outcome predictors.
All regression models were sequentially built using the variable selection method described by Hosmer and Lemeshow.15 Although heroin or cocaine dependence is considered a DSM-IV diagnosis, we chose not include them in our definition of a psychiatric disorder, because IDU is a strong predictor of mortality. We did, however, explore the possible interaction between a psychiatric disorder and IDU. We also explored the possible interaction between a psychiatric disorder and gender, because depressed HIV-positive women have been shown to be at a higher risk for mortality.16 Although previous analyses have not found interactions between race and IDU,17 we also chose to explore this as a potential interaction. We found no statistically significant interactions in any of these analyses.
To evaluate the possibility that our definition of a psychiatric disorder, which included being treated with psychotropic medication, selected for those most willing to engage in HIV treatment itself, we performed a sensitivity analysis to evaluate patients that met 2 of the 3 criteria of a psychiatric disorder (a baseline diagnosis of a psychiatric disorder and baseline use of a psychotropic medication). All reported P values are 2-sided.
During the study period, 549 patients with AIDS and no prior antiretroviral treatment enrolled in the clinic. Eighteen percent (n = 100) were defined as having a mental disorder, 39% (n = 215) were defined as having no mental disorder, and 43% (n = 234) who had 1 or 2 criteria were indeterminate and were excluded.
Of those with a psychiatric disorder, 84% were reported to have a depressive disorder, 9% were reported to have an anxiety disorder, and 7% were reported to have either schizophrenia or mania. Of note, among the total eligible sample (549 patients with AIDS and no prior antiretroviral treatment enrolled in the clinic), only 15% reported having a diagnosis of a depressive disorder, whereas among those in our final study sample, 27% had a diagnosis of a depressive disorder. Those with a current psychiatric disorder were significantly more likely to be younger (35.4 years vs. 37.3 years; P = 0.05), white (22.0% vs. 10.7%; P = 0.008), female (44.0% vs. 32.3%; P < 0.001), to have a history of IDU (51.0% vs. 32.6%; P = 0.002), and to be on PCP prophylaxis within 6 months of enrollment (98.9% vs. 92.6%; P = 0.05) than those without a psychiatric disorder (Table 1).
Time From Enrollment to Receipt of HAART
Kaplan-Meier estimates of time to HAART from time of enrollment demonstrated that compared with those without a psychiatric disorder, those with a psychiatric disorder were significantly more likely to receive HAART (P = 0.05, log-rank test). After adjustment for race, IDU, baseline CD4 count, and receipt of PCP prophylaxis within 6 months of enrollment, the adjusted Cox model revealed that patients with a psychiatric disorder were 37% more likely to receive HAART compared with those without a psychiatric disorder (Cox adjusted hazard ratio [95% CI]: 1.37 [1.01-1.87]; Table 2).
Time on HAART
Compared with those without a psychiatric disorder, those with a psychiatric disorder had greater than twice the crude odds of being prescribed HAART for at least 6 months (odds ratio [95% CI]: 2.07 [1.25-3.40]. After adjustment for race, IDU, baseline CD4 count, and receipt of PCP prophylaxis within 6 months of enrollment, the adjusted odds of being prescribed HAART for at least 6 months continued to be greater than 2 times as great for those with a psychiatric disorder compared with those without a psychiatric disorder (adjusted odds ratio [95% CI]: 2.14 [1.24-3.69]; Table 3).
Time From Enrollment to Death
Kaplan-Meier survival estimates from time of enrollment revealed that compared with those without a psychiatric disorder, those with a psychiatric disorder were more likely to survive (P = 0.10, log-rank test). After adjusting for race, gender, IDU, baseline CD4 count, and receipt of PCP prophylaxis, the adjusted Cox model demonstrated that patients with a psychiatric disorder had nearly a 40% reduction in mortality as compared with those without a psychiatric disorder (Cox adjusted hazard ratio [95% CI]: 0.61 [0.37-0.99]; Table 4).
We subsequently examined the effect of HAART on survival. After adjustment for HAART, those with a psychiatric disorder no longer had a significant difference in survival as compared with those without a psychiatric disorder (Cox adjusted hazard ratio [95% CI]: 0.87 [0.53-1.45]).
Survival Analysis of Patients With 2 Criteria for a Psychiatric Disorder
We repeated the Kaplan-Meier survival estimates to assess whether selecting for those treated with psychiatric medication at enrollment may have biased the results in favor of survival. To do this, we created an alternative definition of a psychiatric disorder that included 2 of the 3 criteria of a psychiatric disorder, namely, baseline diagnosis of a psychiatric disorder and baseline use of a psychotropic medication. Kaplan-Meier survival estimates of time from enrollment to death comparing those with the 2 of 3 criteria definition of a psychiatric disorder to those without a psychiatric disorder revealed no significant difference in survival (P > 0.20).
Time to HAART Analysis Comparing Those With a Baseline Diagnosis of Psychiatric Disorder With and Without Baseline Use of Psychotropic Medication: Sensitivity Analysis
To investigate further whether receiving baseline psychotropic medication selected for those most likely to be prescribed and remain on HAART, we compared those individuals with a baseline diagnosis of a psychiatric disorder who also had baseline use of psychotropic medication with those individuals with a baseline diagnosis of a psychiatric disorder only. Those with a baseline diagnosis of a psychiatric disorder and baseline use of psychotropic medication were as likely to receive HAART (Cox hazard ratio [95% CI]: 1.07 [0.65-1.76]) and to remain on HAART for 6 months (odds ratio [95% CI]: 1.53 [0.70-3.33] as compared with those with a baseline diagnosis of a psychiatric disorder only.
Our study has several important findings. First, in a clinic offering on-site psychiatric services, among a cohort of patients with AIDS, those receiving treatment of their psychiatric disorder were 37% more likely to receive HAART and had 2.5 times the odds of being prescribed HAART for at least 6 months compared with those without a psychiatric disorder. In addition, we found that patients with AIDS who were receiving treatment of their psychiatric disorder had nearly a 40% reduction in mortality compared with those without a psychiatric disorder. These striking findings suggest that in a clinic offering on-site psychiatric services, patients with AIDS receiving treatment of their psychiatric disorders may not only be more likely to be prescribed HAART but may be more likely to reap the survival benefit by remaining on it.
There are several possible explanations as to why patients with psychiatric disorders in this study may be more likely to receive HAART. On-site psychiatric care in an HIV clinic such as was provided in our study has been previously shown to reduce psychiatric symptoms and improve abstinence from alcohol and drugs.13 Although this prior study did not report antiretroviral medication outcomes, it is possible that increased communication between HIV and psychiatric health providers facilitates treatment of both conditions. Several randomized controlled trials conducted in primary care clinics have demonstrated that primary care providers can attain better depressive outcomes with support of on-site psychiatric teams.18-20 These improved outcomes may be at least a partial result of enhanced attention to process measures such as better provider communication. Of note, a study by Turner et al21 reported, among HIV-positive patients with a psychiatric disorder, that the presence of care from a psychiatric health provider was significantly associated with a 50% increase in the odds of receiving HAART compared with those patients without a psychiatric health provider. Improved outcomes may also be the result of providers feeling reassured that their patients with psychiatric disorders are receiving consistent psychiatric care before starting them on antiretroviral treatment. Our study suggests that on-site psychiatric services may improve access to HAART and prolong survival in HIV patients. The exact component of integrated on-site psychiatric care in HIV clinics responsible for this survival benefit needs to be studied further.
Additionally, it is possible that patients treated with psychotropic medication may be more willing to take other types of prescribed medication, including HAART. A previous analysis of New Jersey Medicaid recipients from the pre-HAART era found that depressed HIV-infected patients treated with antidepressants were almost twice as likely to be on antiretroviral therapy as depressed HIV-positive patients who were not prescribed antidepressants.22 In contrast, our study found that among those with a baseline diagnosis of a psychiatric disorder, receipt of psychotropic medication at baseline was not significantly associated with time to receipt of HAART or being prescribed HAART for 6 months or more. Furthermore, as noted previously, those with a baseline diagnosis of a psychiatric disorder and baseline use of psychotropic medication were no more likely to survive compared with those without a psychiatric disorder, suggesting that selecting for those more likely to take medication may not fully account for the resulting reduction in survival. It is also noteworthy that although some patients and clinicians may be concerned about potential drug-to-drug interactions that may occur between antiretroviral drugs and psychotropic medication, this does not seem to be limiting patients with psychiatric disorders from being prescribed HAART in our study.
It may also be possible that patients with psychiatric disorders have worse AIDS-associated medical outcomes as compared with those without psychiatric disorders and thus have more to gain by receiving treatment. Some studies suggest that depressive symptoms do not predict poorer medical outcomes,23,24 whereas other studies suggest the contrary.16,25 Recent data from the HIV Epidemiology Research Study (HERS) report that HIV-infected women with depressive symptoms had faster disease progression than those without depressive symptoms.16,26 Even after adjusting for antiretroviral use, women with chronic depressive symptoms had twice the odds of dying as compared with women with no depressive symptoms. In our study, however, stratified analysis comparing women with psychiatric disorders with those without psychiatric disorders did not reveal any difference in mortality. Furthermore, even after adjusting for gender, those with a psychiatric disorder remained significantly more likely to have a reduction in mortality as compared with those without a psychiatric disorder.
Finally, previous work27 has shown that receipt of PCP prophylaxis is associated with increased survival in HIV patients. The increase in survival in patients with a psychiatric disorder in our study, however, is most likely a result of receipt of HAART itself, because adjustment for PCP prophylaxis did not change the survival results.
There are several potential limitations to this study. First, there is the issue of selection bias. A selected sample may produce results that may be incorrect, nongeneralizable, or both.28 We acknowledge that our sample is unique in that it was derived from only 1 site, which offers on-site psychiatric services as part of its comprehensive HIV care, and because of this, generalizability may be limited. Nevertheless, we believe that these findings are important and suggest that an integrated care system may serve as a model for HIV care for patients with psychiatric disorders. We have attempted to investigate other possible reasons for our findings, including the possibility that those who have a history of taking psychiatric medication are more likely to take other types of medication, and found that this did not fully explain our results. We also carefully adjusted for possible confounders, including PCP prophylaxis, IDU, and race. Future research needs to evaluate which components of integrated care may lead to improved outcomes. It is also important to replicate this study in another treatment setting.
Although all patients identified as having a psychiatric disorder were clinically confirmed by a board-certified psychiatrist, we did not use a validated measure to screen for and identify those with and without a psychiatric disorder. Because of this, we may have underestimated the true number of patients with psychiatric disorders in the clinic and may have selected for those most likely to adhere to HAART treatment. A chart review of 10% of the charts by a board-certified psychiatrist revealed excellent agreement between the presence or absence of a psychiatric disorder, however. In addition, patients in the psychiatric disorder group were more likely to be women and to have a history of IDU, which has been associated with increased mortality. We were also limited by our inability to identify patients who chose not to take psychotropic medicine and thus may be less likely to take other types of prescribed therapy such as HAART.
Third, we were only able to assess whether patients returned for prescription refills and not whether patients actually took their prescribed medication. The finding that patients with psychiatric disorders were 40% more likely to survive after adjusting for other independent predictors, including PCP prophylaxis, may suggest that, on average, the group with psychiatric disorders benefited from HAART.
Our study has several important implications. Perhaps most importantly, the presence of a psychiatric disorder should not automatically disqualify a patient from receiving HAART. Those with psychiatric disorders who have previously demonstrated adherence to psychiatric treatment are good candidates for HAART. For those patients with psychiatric disorders who have previously demonstrated nonadherence to psychiatric treatment or those patients with AIDS and a psychiatric disorder who do not receive care in an HIV clinic, more information is needed to help guide treatment decisions regarding HAART.
1. Bing EG, Burnam MA, Longshore D, et al. Psychiatric disorders
and drug use among human immunodeficiency virus-infected adults in the United States. Arch Gen Psychiatry
2. Burnam MA, Bing EG, Morton SC, et al. Use of mental health and substance abuse treatment services among adults with HIV in the United States. Arch Gen Psychiatry
3. Maj M, Satz P, Janssen R, et al. WHO Neuropsychiatric AIDS
study, cross-sectional phase II. Neuropsychological and neurological findings. Arch Gen Psychiatry
4. Ciesla JA, Roberts JE. Meta-analysis of the relationship between HIV infection and risk for depressive disorders. Am J Psychiatry
5. Palella FJ, Delaney KM, Moorman AC. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. N Engl J Med
6. Shapiro M, Morton SC, McCaffrey DF, et al. Variations in the care of HIV-infected adults in the United States. JAMA
7. Fairfield KM, Libman H, Davis RB, et al. Delays in protease inhibitor use in clinical practice. J Gen Intern Med
8. Sollitto S, Mehlman M, Youngner S, et al. Should physicians withhold highly active antiretroviral therapies from HIV-AIDS
patients who are thought to be poorly adherent to treatment? AIDS
9. Bassetti S, Battegay M, Furrer H, et al. Why is highly active antiretroviral therapy
(HAART) not prescribed or discontinued? Swiss HIV Cohort Study. J Acquir Immune Defic Syndr
10. Bogart LM, Catz SL, Kelly JA, et al. Factors influencing physicians’ judgments of adherence and treatment decisions for patients with HIV disease. Med Decis Making
11. Maisels L, Steinberg J, Tobias C. An investigation of why eligible patients do not receive HAART. AIDS Patient Care STDS
12. Bangsberg DR, Moss A. When should we delay highly active antiretroviral therapy
? J Gen Intern Med
13. Lyketsos CG, Fishman M, Hutton H, et al. The effectiveness of psychiatric treatment for HIV-infected patients. Psychosomatics
14. Treisman GJ, Lyketsos CG, Fishman M, et al. Psychiatric care for patients with HIV infection. The varying perspectives. Psychosomatics
15. Hosmer DW, Lemeshow S. Model development. In: Applied Survival Analysis
. New York: John Wiley & Sons; 1999:158-195.
16. Ickovics JR, Hamburger ME, Vlahov D, et al. Mortality, CD4 cell count decline, and depressive symptoms among HIV-seropositive women: longitudinal analysis from the HIV Epidemiology Research Study. JAMA
17. Gebo KA, Diener-West M, Moore RD. Hospitalization rates in an urban cohort after the introduction of highly active antiretroviral therapy
. J Acquir Immune Defic Syndr
18. Katon W, Von Korff M, Lin E, et al. Collaborative management to achieve treatment guidelines. Impact on depression in primary care. JAMA
19. Unutzer J, Katon W, Callahan CM, et al, for the IMPACT Investigators. Collaborative care management of late-life depression in the primary care setting: a randomized controlled trial. JAMA
20. Wells KB, Sherbourne C, Schoenbaum M, et al. Impact of disseminating quality improvement programs for depression in managed primary care: a randomized controlled trial. JAMA
21. Turner BJ, Fleishman JA, Wenger N, et al. Effects of drug abuse and mental disorders on use and type of antiretroviral therapy in HIV-infected persons. J Gen Intern Med
22. Sambamoorthi U, Walkup J, Olfson M, et al. Antidepressant treatment and health services utilization among HIV-infected Medicaid patients diagnosed with depression. J Gen Intern Med
23. Kilbourne AM, Justice AC, Rollman BL, et al. Clinical importance of HIV and depressive symptoms among veterans with HIV infection. J Gen Intern Med
24. Lyketsos CG, Hoover DR, Guccione M, et al. Depressive symptoms as predictors of medical outcomes in HIV infection. Multicenter AIDS
Cohort Study. JAMA
25. Leserman J, Jackson ED, Petitto JM, et al. Progression to AIDS
: the effects of stress, depressive symptoms, and social support. Psychosom Med
26. The NIMH Multisite HIV Prevention Trial. Reducing HIV sexual risk behavior. The National Institute of Mental Health (NIMH) Multisite HIV Prevention Trial Group. Science
27. Stansell JD, Osmond DH, Charlebois E, et al. Predictors of Pneumocystis carinii pneumonia in HIV-infected persons. Pulmonary Complications of HIV Infection Study Group. Am J Respir Crit Care Med
28. Ellenberg J. Selection bias in observational and experimental studies. Stat Med