Gender and Other Psychosocial Factors as Predictors of Adherence to Highly Active Antiretroviral Therapy (HAART) in Adults with Comorbid HIV/AIDS, Psychiatric and Substance-related Disorder
- First Online:
- Cite this article as:
- Applebaum, A.J., Richardson, M.A., Brady, S.M. et al. AIDS Behav (2009) 13: 60. doi:10.1007/s10461-008-9441-x
- 250 Views
This study assessed adherence to HAART among 67 HIV-infected adults, and the degree to which gender and psychological factors—including depression, drug and alcohol use, quality of life, and medication side effects—influenced adherence. Although overall adherence was greater than rates reported in similar studies, no significant difference in adherence was observed between men and women in the present sample. Medication side effects were a significant predictor of non-adherence in the sample at large and among women in particular, while alcohol dependence was a significant predictor of non-adherence only in women. Possible explanations are explored.
A growing body of literature points to a discrepancy in adherence to antiretroviral medications between HIV-infected men and women. This includes evidence that HIV-infected women face different barriers to adherence than do their male counterparts (Johnston and Mann 2000), and that certain variables, such as substance use and depression, influence this relationship between gender and adherence.
Women living with HIV appear to be particularly vulnerable to depressive symptomatology (Cook et al. 2002) and adherence to HAART is adversely affected by depression (Cook et al. 2007). Adherence is also undermined by substance abuse and dependence (for a review, see Uldall et al. 2004), and alcohol use may be especially hazardous to adherence in HIV-infected women (Berg et al. 2004). Moreover, depression is highly comorbid with illicit substance use in HIV-infected women (Cook et al. 2007). Therefore, among female HIV-infected substance abusers, adherence to HAART may be especially poor.
Furthermore, adherence to HAART is influenced by medication side effects, which may include pain, numbness, tingling, fevers, skin rashes, nausea, vomiting, and pain and bleeding at urination. There is evidence that such side effects are predictive of poor adherence to HAART, and that women are bothered by more unpleasant side effects than are men (e.g., Haug et al. 2005). Therefore, medication side effects may be a particularly important predictor of treatment adherence in women.
In addition, adherence appears to be influenced by quality of life (QOL), an indicator of one’s overall well being, including psychological, social and physical health status. Research indicates a positive correlation between QOL and adherence to HAART (e.g., Carrieri et al. 2003). Moreover, lower levels of QOL are reported in women than in men (Shor-Posner et al. 2000), regardless of HIV status, which may reflect the inverse relationship between QOL and depression. Therefore, differential rates in adherence between men and women may be related to QOL.
The purpose of the present study was to examine the influence of depression, substance use/dependence, medication side effects and QOL on adherence to HAART in HIV-infected men and women with one or more substance-related disorders and one or more mental health disorders. The following hypotheses were evaluated: (1) the rate of adherence to HAART is lower in women than in men; and (2) depression, drug and alcohol abuse/dependence, QOL and experience of medication side effects account for a greater amount of the variance in adherence in women than in men.
Sociodemographic characteristics of respondents by sex, valid percentage
AIDS diagnosis (yes)
Full time (36+ h/week)
Years of school completed
A modified version of the Structured Clinical Interview for DSM-IV (SCID) for Axis I Disorders (SCID-I/P with psychotic screen) was used, along with the SCID for Personality Disorders (SCID-II), to assess for psychiatric eligibility and to create dichotomous variables indicating the presence or absence during the past month (or within the past 12 months) of symptoms consistent with a diagnosis of a major depressive episode, alcohol abuse or dependence, and illicit drug abuse or dependence.
The Beck Depression Inventory-II (BDI-II), a 21-item self-report measure of depressive symptomatology, was used to determine the presence of current depressive symptoms (within the past 2 weeks), whereas the SCID measured the presence of a depressive episode in the last month.
The Addiction Severity Index Lite (ASI-Lite), a semi-structured interview, was used as one source of demographic information and as a measure of recent patterns of alcohol and drug use.
HIV-related Physical Health
The Health Status Questionnaire (HSQ), a brief inventory designed specifically for the multi-site HIV/AIDS Cost Study, captured information from both participants’ medical records and their self-report about stage of HIV disease, and HIV-or-AIDS related conditions and comorbidities (e.g., CD4 count, viral load, opportunistic infections, HIV-related hospitalization).
Adherence was assessed using the Medication Adherence Questionnaire, which is composed of questions from the Adult AIDS Clinical Trials Group (AACTG; Chesney et al. 2000) questionnaire used in national trials of factors impacting adherence. This questionnaire allowed participants to describe their medication regimen, general pattern of adherence over the prior 3 months, side effects of this regimen, and number of missed pills/doses for each medication participants reported taking during the 3 days immediately preceding baseline assessment. For the present study, adherence was described using an adherence ratio (1 − average ratio of pills missed to total pills expected across all medications during the past 3 days), yielding values from .00 to 1.00.
Medication Side Effects
The impact of medication side effects (experienced during the 3 months prior to interview) was evaluated using the following three items from the Medication Adherence Questionnaire: missed taking your medications because you wanted to avoid side effects; missed taking your medications because you felt like the drug was toxic; and missed taking your medications because you felt sick/ill from side effects. These items were scored on a 4-point scale from 0 (never) to 3 (often) and were not drug specific. Therefore, endorsement of each item may have referenced concerns about one or many medications within the participants’ overall treatment regimen.
Quality of Life
Quality of life was assessed using the Medical Outcomes Study Short Form Health Survey-36 (MOS SF-36). This 36-item survey assesses physical and mental health components through eight health domains: physical functions, social functions, role limitations due to emotional and physical problems, bodily pain, vitality, mental health, and general health perception. The Physical Health and Mental Health composite norm-based scores ranging from 0 to 100 were used for the present study as indicators of QOL, with higher numbers indicating better health.
Potential participants were prescreened in a brief telephone or face-to-face interview. Those who reported recent experiences consistent with inclusion criteria were invited to meet privately with study staff to receive an overview of the study design and to complete informed consent procedures. Once written consent was obtained, participants were scheduled for further evaluation to assess their eligibility for study entry and to gather baseline data. One hundred and thirty one potential participants completed informed consent procedures and 85 completed baseline assessment procedures and were found eligible for participation in the longitudinal study. The present findings relate to the 67 participants who were taking antiretroviral medications at the time of the baseline assessment. There were no significant differences across demographic factors, psychiatric diagnoses, and quality of life between participants included in the present analyses and those not on HAART at the time of the assessment.
Study staff verified participants’ HIV status through chart review or written confirmation from participants’ HIV/AIDS primary care provider. Absolute CD4 and viral load values were taken from medical records and when necessary, new tests were ordered to ensure that they were current. The comprehensive baseline assessment of physical health, mental health and psychosocial functioning included measures of adherence, quality of life, substance use, psychiatric symptomatology and overall health and functioning.
All study procedures were reviewed and approved by the appropriate institutional review boards. Participants received $10/h, to a maximum of $60, for completing the baseline assessment battery.
Exploratory analyses revealed that men in the study sample were significantly older and had completed significantly more years of education than the women. No other significant gender differences were observed on demographic variables (i.e., race, income). Moreover, there were no significant correlations between any of the demographic variables—including age and income—and the predictor (e.g., QOL, substance abuse) or outcome (i.e., adherence) variables of interest; therefore, the former were not controlled for in subsequent analyses.
The mean adherence ratio was .890 and .912 for women and men, respectively. An independent-samples t-test was conducted to evaluate the hypothesis that the rate of adherence to HAART was lower in women than in men. The test (t = −.403, P = .69) indicated no significant difference in adherence between women and men. Additionally, at least 95% adherence to medication regimens was reported by 68.2% of women and 71.1% of men. A two-way chi-square analysis was conducted to assess for differences in the proportion of women and men whose adherence ratio was equivalent to .95 or greater, and those who did not meet this standard criterion for optimal adherence. The results were again non-significant, χ2 (1, N = 67) = .384, P = .54.
To further evaluate the relationship between gender and adherence to HAART, a hierarchical regression analysis was conducted for the entire sample with the predictors entered as follows: gender was entered in the first step; depression, alcohol abuse, alcohol dependence, substance abuse and substance dependence in the second step; and QOL and medication side effects in the third. The linear combination of the predictors evidenced a positive predictive trend but was not significantly related to adherence (F[11, 48] = 1.78, P = .06), and accounted for about 35% of the variance in adherence (R2 = .346). Further, the only variable that was a significant predictor of adherence was an indicator of aversive side effects of these medication regimens (i.e., missed taking your medications because you felt sick/ill from side effects) (t = −2.269, P = .037), with non-endorsement of this item predicting higher rates of adherence.
Two-way chi-square analyses were conducted to examine gender differences on the hypothesized predictors of medication adherence (recent alcohol and drug abuse/dependence, depression, and medication side effects). A statistically significant difference was observed for only one of these factors. Specifically, a significantly greater proportion of women (77%) than men (44%) met criteria for recent major depression, χ2 (1, N = 67) = 7.896, P = .011. In addition, independent samples t-tests yielded non-significant findings on QOL measures. Responses yielded comparable scores on the SF-36 Physical Health Composite (male: X = 44.9, SD = 9.8; female: X = 43.4, SD = 7.9) and Mental Health Composite (male: X = 37.4, SD = 12.9; female: X = 32.2, SD = 9.7) scales, scores that are suggestive of slightly compromised general physical and mental health status (i.e., are indicative of slight medical limitations, decrements in physical well being and energy level and some psychological distress, Ware et al. 1995).
To test our second hypothesis that depression, drug and alcohol abuse/dependence, quality of life and experience of medication side effects would account for a greater amount of the variance in adherence in women than in men, hierarchical regression analyses (with depression, alcohol abuse, alcohol dependence, substance abuse and substance dependence entered in the first step and QOL and medication side effects in the second step) were conducted separately for men and women. Among men, the linear combination of the predictors was not statistically significant (F[10, 20] = 1.021, P = .461, R2 = .338), and none of the predictors accounted for a significant amount of the variance in adherence. Among women, there was a trend toward significance for the overall model (F[8, 9] = 2.987, P = .06), with the linear combination of the predictors accounting for 73% (R2 = .726) of the variance in adherence among women. This analysis also revealed that for women, an absence of alcohol dependence (t = −2.315, P < .05) and non-endorsement of specific concerns about medication side effects (i.e., missed taking your medications because you wanted to avoid side effects [t(9) = −2.324, P = .05] and missed taking your medications because you felt sick/ill from side effects [t(9) = −2.728, P = .02]) were significant predictors of adherence.
The hypothesis that the rate of adherence to HAART would be lower in women than men was not supported in the present study. There was no significant difference in the reported adherence ratio between men and women, or in the proportion of men and women meeting the optimal rate of adherence (i.e., adherence ratio of .95). Indeed, it must be noted that the reported rate of adherence in the recent past was quite high for the entire study sample (nearly 70% of participants reported such near perfect adherence in the recent past), a rate much higher than reported in prior studies of adherence involving similar samples (e.g., Berg et al. 2004; Haug et al. 2005; Turner et al. 2003). Our results also indicated that for the sample as a whole only illness as a side effect of HAART was (inversely) predictive of rates of adherence.
Our second hypothesis, that depression, drug and alcohol abuse/dependence, quality of life and experience of medication side effects, would be differentially predictive of adherence as a function of gender received partial support from the present findings. Hierarchical regression analyses revealed that among women only, alcohol dependence and specific concerns about medication side effects were significantly and inversely predictive of adherence. These findings are consistent with extant literature suggesting that among women, alcohol use (e.g., Berg et al. 2004) and medication side effects (Johnston and Mann 2000) may be independently associated with non-adherence. Although quite appropriately there was no difference in the proportion of men and women meeting criteria for recent substance-related disorders (required for study eligibility and verified via preliminary chi-square analyses), alcohol dependence was a significant predictor of adherence only in women, in part a reflection of the relative consistency of such patterns among male participants.
Somewhat surprisingly, depression did not emerge as a significant predictor of adherence in the present study. Despite prior evidence that HIV-infected depressed women are significantly less likely to be adherent to HAART than their non-depressed counterparts (Cook et al. 2002) and evidence of differential rates of depression as a function of gender among participants, this factor was secondary to alcohol and medication side effects as a predictor of adherence in the present study. These findings are also in contrast with prior evidence of poorer generalized quality of life (e.g., Shor-Posner et al. 2000) and its impact on chronic, life-affirming behaviors such as adherence to prescription medication regimens.
It is possible that these findings reflect other characteristics endemic to the present sample. Participants were recruited from clinics where they had routine access to integrated medical and mental health care that emphasized positive health practices; their comparatively high adherence rates may be at least partially attributable to their access to such comprehensive routine care. Reported rates of adherence may also result from characteristics of the participants themselves, who were willing to endure extensive assessments in return for the possibility of gaining access to experimental substance and mental health treatment designed to enhance adherence, over and above that routinely available to them. In short, these participants may have been exceptionally motivated and involved in their care relative to non-participants or to individuals enrolled in comparable prior studies. The present findings may reflect participants’ overall quality of general medical and psychiatric care, and may also suggest that this sample included more relatively marginalized men than found in prior studies. For example, in this sample, significant gender differences in income levels or rate of employment—traditionally demonstrated in similar samples—were not observed.
Self-reported adherence is viewed by some as more prone to distortion than are more “objective measures” (e.g., electronic pill bottles/MEMS caps), potentially increasing the probability of inflated adherence estimates. However, prior research demonstrated the validity of the adherence questionnaire employed in the present study as an indicator of recent behavior (Chesney et al. 2000). As such, the present findings appear to highlight a more optimistic appraisal of adherence in a sample of high-risk adults living with HIV/AIDS than found in earlier studies. Nonetheless, confidence in the present findings could have been enhanced by the use of more than one indicator of adherence (e.g., self-report measures and MEMS caps), as employed in some prior research on adherence.
It is also likely that the present findings underestimate the degree to which comorbid psychiatric disorders were present in study participants, limiting the degree to which effects of psychiatric morbidity on adherence could be explored. The version of the SCID utilized in the multi-site HIV/AIDS Cost study was designed to streamline screening for eligibility and general risk within diagnostic domains (e.g., Mood Disorders) in an effort to reduce overall assessment liability for any given participant, likely resulting in our failure to identify one or more comorbid conditions in several participants. Finally, this study’s relatively small sample and participants’ relatively restricted range of reported treatment adherence may have limited its sensitivity to factors affecting the latter.
On the other hand, this study illustrates the relative contribution of current psychiatric morbidity, substance use, treatment-related factors and well-being on treatment adherence among a sample of adults likely to be increasingly represented among people living with HIV in the near term. Present findings further emphasize the need for investigators and clinicians alike to attend to patterns of alcohol use (versus the use of other substances of abuse) and—in the absence of objective measures of adverse treatment effects—to carefully attend to subjective reports of such discomfort in relation to treatment adherence. Future research involving comparable target samples may consider participants’ motivation to enhance physical and mental health functioning and the nature of substance use in which they are engaged, as well as sources of instrumental assistance that may enhance adherence, even in the face of adverse effects.
This work was supported by the HIV/AIDS Treatment Adherence, Health Outcomes and Cost Study, a collaboration of six Federal entities within the U.S. Department of Health and Human Services (DHHS): The Center for Mental Health Services (CMHS), which had the lead administrative responsibility, and the Center for Substance Abuse Treatment (CSAT), both components of the Substance Abuse and Mental Health Services Administration (SAMHSA); the HIV/AIDS Bureau of the Health Resources and Services Administration (HRSA); the National Institute of Mental Health (NIMH), the National Institute on Drug Abuse (NIDA), and the National Institute on Alcohol Abuse and Alcoholism (NIAAA), all parts of the National Institutes of Health (NIH). The content of this publication does not necessarily reflect the views or policies of these or any other agencies of the DHHS.