Abstract
Little is known about how engagement with healthcare providers mediates the relationship between psychosocial factors (anxiety, depression, stigma) and medication adherence among persons living with HIV (PLWH). Moreover, little research has investigated potential biological sex differences in this relationship. We conducted a secondary analysis of data collected from four projects (N =281) focused on improving health outcomes in PLWH. Males displayed (a) negative association between depression and engagement with healthcare providers (β = − 0.02, z= − 3.20, p = 0.001) and (b) positive association between engagement with healthcare providers and medication adherence (β = 0.55, OR = 1.73, z= 2.62, p = 0.009). Females showed no association between any of these factors. Anxiety and stigma were not significantly associated with medication adherence. Path analysis modeling for males had a very good fit (CFI = 1, TLI = 1, RMSEA = 0); none of the regression coefficients was significant for females. The significant relationship between depression and medication adherence among males was fully mediated by engagement with healthcare providers. Findings suggest that adherence interventions for PLWH should be tailored by biological sex.
Keywords: Patient–provider engagement, Sex differences, Medication adherence, Depression, HIV/AIDS
Résumé
Poco se sabe cómo la interacción con proveedores médicos funciona en la relación entre factores psicosociales (ansiedad, depresión, estigma) y adherencia a medicamentos en personas viviendo con VIH (PVV). Además, pocos estudios han inves- tigado posibles diferencias en el sexo biológico en estas relacione. Dirigimos un análisis secundario de cuatro proyectos (N = 281) enfocados en PVV. Hombres demostraron (a) una asociación negativa entre depresión e interacción con proveedores médicos (β=−0.02, z=−3.20, p= 0.001) y (b) una asociación positive entre interacción con proveedores médicos y adherencia a medicamentos (β= 0.55, OR= 1.73, z= 2.62, p= 0.009). Mujeres no demostraron asociación entre estos factores. Ansiedad y estigma no fueron asociados significativamente con la adherencia a medicamentos. El modelo de análisis del camino para hombres tuvo un muy ajuste (CFI = 1, TLI = 1, RMSEA = 0); ninguna de los coeficientes de regresión fue significativa para mujeres. La relación significativa entre depresión y adherencia a medicamentos para hombres fue completamente mediada por la interacción con proveedores médicos. Las recomendaciones sugieren que las intervenciones de adherencia para PVV deberían ser ajustado por sexo biológico.
Keywords: interacción entre paciente y proveedor, diferencias de sexo, adherencia a medicamento, depression, VIH/ SIDA
Introduction
In much of the developed world, human immunodeficiency virus (HIV) is considered a chronic manageable disease [1]. Disease management can be achieved only through adherence to antiretroviral therapy (ART), which is critical to achieve viral suppression [2, 3]. However, current U.S. estimates suggest that more than half of all persons living with HIV (PLWH) do not adequately adhere to ART [4].
Patient-healthcare provider engagement is a critical component for ensuring adherence to ART and quality of life among PLWH [1, 5, 6]. Research has shown that PLWH who were more engaged with their healthcare providers were more likely to follow the providers’ advice and less likely to miss appointments [7]. PLWH who reported high levels of provider engagement also reported higher self-efficacy for medication adherence, fewer missed medication doses, and better quality of life [8]. A systematic review of the determinants of HIV-related medication adherence reported that unpleasant provider-associated experiences (e.g., rudeness, condemnation, fatigue) were significantly associated with non-adherence to medications among PLWH [1].
Studies have shown that psychosocial distress (e.g., stigma, depression, and anxiety) is significantly associated with patient-healthcare provider engagement [6, 9, 10]. For example, PLWH who experience HIV-related stigma are more likely to abstain from getting support from their health- care providers and adequate medical care [10]. Mitchell and colleagues found that PLWH who experienced depressive symptoms reported poorer patient-provider engagement [6]. Bankoff et al. revealed that depression and anxiety were significantly and negatively correlated with PLWH perceptions of the quality of information offered by their healthcare providers, perceptions of their providers’ interactional style, and perceptions of their providers’ ability to conduct psychosocial status screenings [9].
There have been numerous HIV/AIDS studies focusing on (a) how HIV/AIDS affects men and women differently[11] and (b) how men and women may respond to medication and therapy differently [12]. While the terms sex and gender are often used interchangeably in research [13], sex refers to the biological distinction between men and women assigned at birth, and gender refers to socially created characteristics [13, 14].
Biological sex-based differences in the relationship between engagement with healthcare providers and medication adherence are not well understood [15]. Some studies have found that there are differences between male and female PLWH in terms of (a) psychosocial distress and engagement with healthcare providers and (b) the effect of psychosocial distress/engagement with healthcare providers on symptom management and medication adherence [14-16]. For example, women are more likely than men to have psychosocial distress (e.g., depression) and are consequently less likely to follow up on medical care [14-16]. Other studies have reported no significant difference between sex groups in engaging in medical care with healthcare providers [7] and adhering to HIV-related medication [17]. There is no consensus on how biological sex differences in PLWH affect the interactions among psychosocial distress, engagement with healthcare providers, and medication adherence [14]. An understanding of how these components interact will allow researchers and clinicians to design interventions that are more targeted to the individual needs of PLWH.
As described above, many studies have reported that PLWH-healthcare provider engagement plays a pivotal role in improving adherence to ART as well as decreasing psychosocial distress. However, little is known about the relationship between PLWH-healthcare provider engagement, psychosocial distress (e.g., anxiety, depression, stigma), and medication adherence in PLWH. Specifically, there are few studies identifying the directions of relationships among predictors (psychosocial distress) and the outcome (medication adherence) through the mediating role of engagement with healthcare providers. Thus, we sought to fill these knowledge gaps and test our hypothesis that engagement with healthcare providers mediates the relationship between psychosocial factors (stigma, depression, anxiety) and medication adherence (Fig. 1). Moreover, given that findings from previous studies on biological sex differences in the level of engagement with healthcare providers and medication adherence are less clear [1, 15, 16], the purpose of this study is to assess if there are biological sex-based differences among PLWH in the relationship between psychosocial distress engagement with healthcare providers, and medication adherence. We chose to focus on biological sex instead of gender because being designated a female as sex at birth has been associated with more vulnerability to infection and specifically HIV/AIDS [17].
Fig. 1.
Proposed conceptual model
Methods
Study Design and Sample
We conducted a secondary analysis using baseline data from four HIV-related research projects, including the WISE App usability study [18], the WISE App trial [19], the Video Information Provider for HIV-Associated Non- AIDS (VIP-HANA) trial [20, 21] and the VIP-HANA sex/gender supplement [11, 22, 23]. Each of these studies aimed to develop and test mobile health applications to improve self-care symptom management and medication adherence for PLWH.
Data collection for the four studies had different time- lines, which has been reported elsewhere [18, 22, 23]. In- person recruitment was done through clinic-and community-based advertising, and online recruitment through social media sites was also conducted. To ensure that participants were cognitively intact for study participation, we administered the Mini Mental State Examination [24]. We included participants if they were (a) diagnosed with HIV/AIDS, (b) aged 18 years or older, (c) able to speak and understand English or Spanish, (d) in possession of a smartphone, and (e) taking ART medications. Exclusion criteria included (a) pregnancy, (b) cognitive inability to participate in the study, and/or (c) any clinical problems (e.g., meeting the criteria for dementia) that would preclude someone from being able to use a cell phone or fulfill study procedures. Prior to participating in any study procedures, all potential participants voluntarily provided informed consent. Each of the study participants from the four studies was given between $10 and $40 compensation for their participation (i.e., $10 for the VIP-HANA women supplement, $25 for the VIP-HANA trial at baseline, $25 for the WISE App usability testing, $40 for the WISE App trial at baseline). All studies were approved by the Columbia University Irving Medical Center Institutional Review Board.
Measures
Each of the four parent studies included sociodemographic questions and measurements of stigma [25], depression [26], anxiety [27], engagement with healthcare provider [7], and medication adherence [28].
Stigma
We used Berger’s HIV stigma scale comprised of 40 items rated on a four-point Likert scale, ranging from strongly disagree (1) to strongly agree (4) [25]. This stigma scale has four subscales: personalized stigma, disclosure, negative self- image, and public attitudes. In this study, we chose to include the personalized stigma subscale defined as “experiences of rejection for having HIV or fears of rejection” [25] because it matches our intentions to both “capture perceived consequences of other people knowing about one’s HIV infection”[29] and identify how personalized stigma affects engagement with healthcare provider and medication adherence. The personalized stigma subscale consists of the following three items: “I have been hurt by how people reacted to learning I have HIV”; “I have stopped socializing with some people because of their reactions to my having HIV”; and “I have lost friends by telling them I have HIV” [30]. Internal consistency reliability was established by measuring Cronbach’s alpha of personalized stigma, which was 0.85 in this current study.
Psychological Symptoms: Depression, Anxiety
To measure depression and anxiety, we used the Patient-Reported Outcome Measurement Information System (PROMIS) short-form questionnaires [31]: Depression v1.0 [26] and Anxiety v1.0 [27]. Each PROMIS short form has four items, with response options ranging from never (1) to always (5). The PROMIS Depression short-form items are as follows: “I felt worthless”; “I felt helpless”; I felt depressed”; and “I felt hopeless.” The PROMIS Anxiety short-form items are as follows: “I felt fearful”; “I found it hard to focus on anything other than my anxiety”; “My worries overwhelmed me”; and “I felt uneasy.” Cronbach’s alpha reliability scores for depression and anxiety in this current study were 0.92 and 0.90, respectively. The PROMIS measurement was validated among PLWH [32].
Engagement with Healthcare Provider
We used the Engagement with Health Care Provider scale for PLWH developed by Bakken et al. [7]. Engagement with Healthcare Provider includes the following dimensions: access to healthcare providers, information sharing, involvement of PLWH in decision making and self-care activities, respect and support of the healthcare provider for PLWH’s choices, and management of PLWH concerns [7]. The scale has 13 items (e.g., “My primary healthcare provider listens to me,” “My primary healthcare provider respects me,” and “My primary healthcare provider provides me with information”) with four-point Likert scale responses ranging from always true (1) to never true (4). Cronbach’s alpha reliability of engagement with healthcare provider in this current study was 0.96.
Medication Adherence
The Center for Adherence Support Evaluation (CASE) Adherence Index was used to assess medication adherence among PLWH [28]. The CASE Adherence Index consists of the following three items: A1, which is “frequency of difficulty taking HIV medications on time” (i.e., no more than two hours before or two hours after the time the provider instructed that the medication be taken); A2, which is “average number of days per week at least one dose of HIV medications was missed”; and A3, which is “last time patient missed at least one dose of HIV medications.” A1 responses range from 1 to 4 points, and both A2 and A3 range from 1 to 6 points; composite scores range from 3 to16. A score of 10 or greater indicates “good adherence”; a score of less than 10 indicates “poor adherence” [28]. Cronbach’s alpha reliability of the CASE Adherence Index in this current study was 0.77.
Sociodemographic Questionnaire
Self-reported sociodemographic data were collected to describe participants’ age, sex assigned at birth, gender, race, ethnicity, sexual orientation, relationship status, education, employment status, annual income, and health insurance.
Data Analysis
R software was used to conduct all statistical analyses [33]. Descriptive statistics (i.e., mean, median, standard deviation, frequency, and percentages) were calculated to characterize study participants. Cronbach’s alpha was assessed for the scales with multiple items to ensure internal consistency reliability and was considered acceptable if > 0.7 [34].
Based on a four-step approach proposed by Baron and Kenny [35], we first tested the mediation effect, in which (a) independent variables (personalized stigma, depression, anxiety) led to (b) the outcome [medication adherence measured by the CASE Adherence Index (i.e., good adherence vs. poor adherence)] through (c) the mediator (engagement with healthcare provider). This approach tests the mediation effect by conducting several simple or multiple regression analyses and then examining the significance of the coefficients at each step. The first step is to conduct simple regression analysis with each predictor predicting outcome; the second step is to conduct simple regression with each predictor predicting mediator; the third step is to conduct simple regression with mediator predicting outcome; the final, or the fourth step, is to conduct multiple regression with mediator and each predictor predicting outcome if all the regression coefficients are significant at steps 1–3. In step 4, some form of mediation effect is supported if the effect of the mediator remains significant after controlling for the predictor. If the predictor is no longer significant when the mediator is controlled, full mediation is supported. If the predictor is still significant, partial mediation is supported. In step 4, some form of mediation effect is supported if the effect of the mediator remains significant after controlling for the predictor. If the predictor is no longer significant when the mediator is controlled, full mediation is supported. If the predictor is still significant, partial mediation is supported. For each of the steps, logistic regression was used when the dependent variable at the step was dichotomized, such as medication adherence, and linear regression was used when the dependent variable was a continuous variable such as engagement with healthcare provider.
Next, we conducted path analysis to confirm the mediation effect and to test the significance of the indirect effect. We used the Comparative Fit Index (CFI), Tucker and Lewis’s Index (TLI), and Root Mean Square Error of Approximation (RMSEA) to assess the model fit. Good fit is indicated by CFI ≥ 0.90, TLI ≥ 0.95, and RMSEA ≤ 0.08 [36]. We assessed all of the above models separately for males and females in order to determine if biological sex differences exist in the relationships between the factors. Odds ratio (OR) was reported if the dependent variable was dichotomized when using the four-step approach or performing the path analysis. The overall two-sided type I error throughout the study was 0.05.
Results
Sociodemographic Characteristics of Study Participants
Table 1 represents the sociodemographic characteristics of all participants in this study. The sample (N = 281) had a mean age of 51.0 (± 10.1) years; 52% of participants were assigned female sex at birth; and 51% currently identified as female. Overall, 73% of participants self- identified as Black/African American and 28% as His- panic/Latino. Half of the participants (50%) were single; most (57%) indicated an education level of high school or less; more than a third (34%) were unemployed; and 38% were disabled and unable to work. Nearly half (43%) of the participants reported an annual income of less than $10,000, and most (87%) had Medicaid/Medicare health insurance. Most participants (95%) reported having a healthcare provider in charge of overall HIV health care, and more than four-fifths (89%) reported having had an HIV-related appointment within the last three months with their healthcare provider.
Table 1.
Sociodemographic characteristics of persons living with HIV (n = 281)
| Characteristics | M (SD) or n (%) | |
|---|---|---|
| Age (year) | 51.0 (10.1) | |
| Sex at birth | ||
| Male | 136 (48.4) | |
| Female | 145 (51.6) | |
| Gender | ||
| Male | 129 (45.9) | |
| Female | 144 (51.2) | |
| Transgender male/transman/FTM | 1 (0.4) | |
| Transgender female/transwoman/MTF | 6 (2.1) | |
| Genderqueer | 1 (0.4) | |
| Race | ||
| Black/African American | 204 (72.6) | |
| American Indian and others | 52 (18.5) | |
| White | 25 (8.9) | |
| Ethnicity | ||
| Hispanic/Latino | 203 (72.2) | |
| Non-Hispanic/Latino | 78 (27.8) | |
| Reading/speaking Spanish | 71 (25.2) | |
| Relationship status | ||
| Single | 139 (49.5) | |
| In a relationship with a man | 64 (22.8) | |
| In a relationship with a woman | 21 (7.5) | |
| Legally married to a man or in a registered civil union/domestic partnership with a man | 22 (7.8) | |
| Legally married to a woman or in a registered civil union/domestic partnership with a woman | 4 (1.4) | |
| Divorced/separated from a man | 9 (3.2) | |
| Divorced/separated from a woman | 7 (2.5) | |
| Widowed (male partner) | 5 (1.8) | |
| Widowed (female partner) | 6 (2.1) | |
| Other | 4 (1.4) | |
| Education | ||
| Less than high school/HS diploma or GED | 159 (56.6) | |
| Some college | 75 (26.7) | |
| College | 32 (11.3) | |
| Graduate school/professional school | 13 (4.6) | |
| Employment status | ||
| Working full-time | 10 (3.6) | |
| Working part-time | 47 (16.7) | |
| Working off the books | 7 (2.5) | |
| Unemployed | 96 (33.8) | |
| Retired | 22 (7.8) | |
| Student | 14 (5) | |
| Disabled | 108 (38.4) | |
| Income | ||
| Less than $10,000 | 120 (42.7) | |
| $10,000–$19,999 | 75 (26.7) | |
| $20,000–$39,999 | 40 (14.2) | |
| $40,000–$59,999 | 4 (1.4) | |
| $60,000–$79,999 | 4 (1.4) | |
| $80,000–$99,999 | 1 (0.4) | |
| $150,000 or more | 1 (0.4) | |
| Health insurance | ||
| Medicaid/Medicare | 243 (86.5) | |
| Through affordable care act | 16 (5.7) | |
| Through job | 10 (3.6) | |
| Through someone else’ job | 2 (0.7) | |
| Other | 49 (17.4) | |
| Healthcare provider caring for HIV | ||
| Yes | 268 (95.4) | |
| No/prefer not to answer | 13 (4.6) | |
| Last time to make an appointment with healthcare provider | ||
| Last 3 months | 250 (89) | |
| 3–6 months ago | 17 (6) | |
| More than 6 months ago/don’t know | 14 (5) | |
HS high school, GED general education diploma
Psychosocial Distress, Engagement with Healthcare Provider, and Medication Adherence
Table 2 shows the regression models from the four-step approach [35]. First, medication adherence measured by the CASE Adherence Index was regressed separately by three predictors (depression, anxiety, personalized stigma). For males, higher depression scores were associated with poorer medication adherence scores (β = − 0.03, OR = 0.97, t = − 2.72, p = 0.0065). For females, depression was not significantly associated with medication adherence. Second, engagement with healthcare provider was regressed separately by three predictors (depression, anxiety, personalized stigma). For males, higher depression scores were associated with lower engagement with healthcare provider scores (β = − 0.02, t = − 3.41, p = 0.0006). Higher anxiety scores were associated with lower engagement with health- care provider scores (β = − 0.01, t = − 2.78, p = 0.0054). In addition, higher personalized stigma scores were associated with lower engagement with healthcare provider scores (β = − 0.22, t = − 3.96, p < 0.0001). For females, higher anxiety scores were associated with lower engagement with healthcare provider scores (β = − 0.01, t= − 2.04, p = 0.041). As the third step, medication adherence was regressed by engagement with healthcare provider. For males, higher engagement with healthcare provider scores were associated with better medication adherence scores (β = 0.69, OR = 1.99, t = 3.12, p = 0.0018). For females, engagement with healthcare provider was not significantly associated with medication adherence. In the fourth step, medication adherence was regressed by depression and engagement with healthcare provider. For males, higher engagement with healthcare provider scores were associated with better medication adherence scores (β = 0.58, OR = 1.79, t = 2.57, p = 0.0103), which indicates that some form of mediation effect is supported. The significant relationship between depression and medication adherence at step 1 was no longer significant at step 4 (β = − 0.02, OR = 0.98, t = − 1.93, p = 0.053), which means that the effect of depression on medication adherence was fully mediated by engagement with healthcare provider. For females, engagement with healthcare provider was not significantly associated with medication adherence. Thus, sex at birth modifies the relationship between factors.
Table 2.
Regression models from four step approach proposed by Baron and Kenny [35]
| Construct | Male (n = 135) | Female (n = 145) | ||||
|---|---|---|---|---|---|---|
| β | t-value | p-value | β | t-value | p-value | |
| Step 1: predictor to outcome | ||||||
| Depression to medication adherence | − 0.03 | − 2.72 | 0.0065 | − 0.01 | − 0.96 | 0.3383 |
| Anxiety to medication adherence | − 0.01 | − 1 | 0.3155 | − 0.01 | − 0.62 | 0.5313 |
| Personalized stigma to medication adherence | 0.01 | 0.09 | 0.9247 | − 0.04 | − 0.29 | 0.7705 |
| Step 2: predictor to mediator | ||||||
| Depression to engagement with healthcare provider | − 0.02 | − 3.41 | 0.0006 | − 0.01 | − 1.04 | 0.294 |
| Anxiety to engagement with healthcare provider | − 0.01 | − 2.78 | 0.0054 | − 0.01 | − 2.04 | 0.041 |
| Personalized stigma to engagement with healthcare provider | − 0.22 | − 3.96 | <.0001 | − 0.02 | − 0.33 | 0.7438 |
| Step 3. mediator to outcome | ||||||
| Engagement with healthcare provider to medication adherence | 0.69 | 3.12 | 0.0018 | 0.18 | 0.93 | 0.3516 |
| Step 4: predictor and mediator to outcome | ||||||
| Depression to medication adherence | − 0.02 | − 1.93 | 0.0528 | − 0.01 | − 1.17 | 0.2442 |
| Engagement with healthcare provider to medication adherence | 0.58 | 2.57 | 0.0103 | 0.16 | 0.82 | 0.4129 |
Engagement with Healthcare Provider as Mediator of the Relationship Between Depression and Medication Adherence
The regression analysis showed that, among males, the significant relationship between depression and medication adherence was fully mediated by engagement with healthcare provider. As a next step, we conducted path analysis to confirm the mediation effect of engagement with healthcare provider and test the significance of the indirect effect between variables by sex at birth. Model 1 from path analysis for males had a very good fit (CFI = 1, TLI = 1, RMSEA = 0).
Table 3 shows that, among males, depression and engagement with healthcare provider were negatively associated (β = − 0.02, z= − 3.20, p = 0.001) and that engagement with healthcare provider and medication adherence were positively associated (β = 0.55, OR = 1.73, z = 2.62, p = 0.009). Depression had a significant negative association with medication adherence when not including engagement with healthcare provider as a mediator (β = − 0.03, OR = 0.97, z= − 2.68, p = 0.007). However, after including engagement with healthcare provider as a mediator, depression no longer had a significant association with medication adherence (β = − 0.02, z = − 1.858, p = 0.063). The indirect effect of depression on medication adherence was significantly associated (β = − 0.01, z= − 1.99, p < 0.05). We then built a path analysis model 2 for males by removing the direct path from depression to medication adherence. The model fit was poor (CFI = 0.59, TLI = 0.59, RMSEA = 0.13); thus, Model 1 was the final model for males (Fig. 2).
Table 3.
Models from path analysis assessing the effect of depression on medication adherence mediated through engagement with healthcare provider by gender
| Relationship between construct | Male (n = 135) |
Female (n = 145) |
||||
|---|---|---|---|---|---|---|
| β | z-value | p-value | β | z-value | p-value | |
| Depression to engagement with healthcare provider | − 0.02 | − 3.20 | 0.001 | − 0.01 | − 0.96 | 0.339 |
| Engagement with healthcare provider to medication adherence | 0.55 | 2.62 | 0.009 | 0.16 | 0.83 | 0.406 |
| Depression to medication adherence | − 0.02 | − 1.89 | 0.063 | − 0.01 | − 1.20 | 0.231 |
| Indirect effect of depression to medication adherence through | − 0.01 | − 2.00 | 0.046 | − 0.001 | − 0.63 | 0.532 |
| Total effect of depression to medication adherence | − 0.03 | − 2.68 | 0.007 | − 0.02 | − 1.27 | 0.204 |
Fig. 2.
Path analysis model for males
As Fig. 3 and Table 3 illustrate, none of the regression coefficients from the path analysis model for females was significant, which is supported by the findings from the four-step approach regression analysis. To summarize, our findings from the path analysis indicate that the significant effect of depression on medication adherence among males was fully mediated by engagement with healthcare provider. No such effect was found for females.
Fig. 3.
Path analysis model for females
Discussion
Despite increasing interest in enhancing engagement with healthcare providers to maintain viral suppression among PLWH through adherence to ART [1, 5, 6, 9, 10], there has been little research on mechanisms for achieving optimal medication adherence among PLWH who experience psychosocial distress. To our knowledge, this is the first study to (a) identify the role of PLWH-healthcare provider engagement in the relationship between psychosocial dis- tress and medication adherence among a racially and ethnically diverse, urban sample of PLWH and (b) assess whether biological sex differences affect the relationships between these factors. Findings from this study show that engagement with healthcare providers fully mediated the relation- ship between depression and medication adherence in males but not females.
The findings regarding relationships between depression, engagement with healthcare providers, and medication adherence are partially similar to previous studies of PLWH [6] and other chronic diseases such as diabetes [37]. Mitchell and colleagues [6] found that depression was significantly associated with PLWH-healthcare provider engagement and that PLWH-healthcare provider engagement was significantly associated with medication adherence. How- ever, the path between depression and medication adherence was non-significant, and thus, the role of engagement with healthcare provider as a mediator did not exist in that study [6]. The findings were not consistent with our findings, which indicated that the path between depression and medication adherence among males was, in fact, significant. This discrepancy may be partially explained because of the difference in our study participants; Mitchell et al.’s study targeted and recruited PLWH who were current or former intravenous drug users [6].
On the other hand, in a study of persons living with diabetes, Tovar and colleagues found that, when compared with support from family/friends/significant others, support from healthcare providers was the most significant mediator between depression and diabetes medication adherence [37]. As such, the role of healthcare providers is important for improving medication adherence among persons who experience depressive symptoms, but limited evidence exists in PLWH. Thus, future studies that examine effective strategies for improving PLWH engagement with their healthcare providers are needed to enhance long-term outcomes including medication adherence in PLWH.
In this study, we found that only males showed significant associations between depression and medication adherence, between depression and engagement with healthcare providers, and between engagement with healthcare providers and medication adherence. This is interesting given that women in the general population are more likely than men both to be depressed [38] and visit healthcare providers and receive medical care services [39, 40]. However, HIV/AIDS research suggests that women living with HIV have different attitudes toward the healthcare system and their healthcare providers compared to women in the general population [15, 41]. Women living with HIV are less likely than men to have optimal HIV healthcare-service utilization, including lower use of HIV primary care services and greater use of the emergency department [41], less likely to receive care from their healthcare providers [15, 42], and less likely to adhere to their ART regimens [15]. Possible reasons that women living with HIV are more vulnerable than men to ineffective management of their HIV-related symptoms might include sex disparities in education, employment, and health insurance [41]. Future research should consider sex differences when examining medication adherence; factors associated with PLWH–provider engagement and medication adherence in both males and females should be identified. Moreover, sex-specific medication adherence interventions might be necessary to achieve viral suppression.
Findings from this study highlight the need to tailor healthcare delivery based on the sex of PLWH. Sex as a biological variable has also been highlighted as a priority area for the National Institute of Health initiative to enhance reproducibility through rigor and transparency [43]. The policy requiring that researchers factor biological sex into the design, analysis, and reporting of human studies was implemented as it has become increasingly clear that sex differences extend beyond reproductive issues, and this is further supported through the findings in our study [44].
Limitations
This study has several limitations. First, the study participants were predominantly African American and Hispanic, which limits the generalizability of these findings to other racial/ethnic minority groups. In addition, this was a secondary analysis of baseline data from four different projects and given the cross-sectional nature of these data, causality in the relationships between depression, engagement with healthcare providers, and medication adherence is limited. Moreover, focusing on biological sex instead of gender might have potential limitations. For example, findings from our study might not reflect potential differences in how those who acquire gender identity (e.g., transgender people) use healthcare services and adhere to medication regimens.
Conclusions
Despite the limitations, this study highlights the importance of the mediating role of engagement with healthcare provider in the relationship between psychosocial distress and medication adherence among PLWH. Findings from this study indicate that the significant effect of depression on medication adherence among males living with HIV was fully mediated by engagement with healthcare provider; however, no such effect was found for females. Study findings suggest that adherence interventions need to be tailored according to biological sex differences in PLWH and account for psychosocial factors such as depression.
Acknowledgements
This study was funded by the Agency for Health- care Research and Quality (R01HS025071), the National Institute of Nursing Research of the National Institute of Health (R01NR015737, R01NR015737-02S1, and K24NR018621). DB was funded by the Reducing Health Disparities through Informatics (RHeaDI) training grant funded by the National Institute of Nursing Research (T32 NR007969). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Agency for Healthcare Research and Quality.
Footnotes
Conflict of interest: The authors declare that they have no conflicts of interest.
References
- 1.Heestermans T, Browne JL, Aitken SC, Vervoort SC, Klipstein-Grobusch K. Determinants of adherence to antiretroviral therapy among HIV-positive adults in sub-Saharan Africa: a systematic review. BMJ Glob Health. 2016;1(4):e000125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Selik RM, Lindegren ML. Changes in deaths reported with human immunodeficiency virus infection among United States children less than thirteen years old, 1987 through 1999. Pediatr Infect Dis J. 2003;22(7):635–41. [DOI] [PubMed] [Google Scholar]
- 3.Deeks SG, Lewin SR, Havlir DV. The end of AIDS: HIV infection as a chronic disease. Lancet. 2013;382(9903):1525–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Centers for Disease Control and Prevention. More people with HIV have the virus under control. 2017. https://www.cdc.gov/nchhstp/newsroom/2017/2017-HIV-Continuum-Press-Release.html. Accessed 27 July 2017
- 5.Langebeek N, Gisolf EH, Reiss P, Vervoort SC, Hafsteinsdottir TB, Richter C, et al. Predictors and correlates of adherence to combination antiretroviral therapy (ART) for chronic HIV infection: a meta-analysis. BMC Med. 2014;12:142. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Mitchell MM, Nguyen TQ, Maragh-Bass AC, Isenberg SR, Beach MC, Knowlton AR. Patient–provider engagement and chronic pain in drug-using, primarily african american persons living with HIV/AIDS. AIDS Behav. 2017;21(6):1768–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Bakken S, Holzemer WL, Brown MA, Powell-Cope GM, Turner JG, Inouye J, et al. Relationships between perception of engagement with health care provider and demographic characteristics, health status, and adherence to therapeutic regimen in persons with HIV/AIDS. AIDS Patient Care STDs. 2000;14(4):189–97. [DOI] [PubMed] [Google Scholar]
- 8.Chen WT, Wantland D, Reid P, Corless IB, Eller LS, Iipinge S, et al. Engagement with health care providers affects self-efficacy, self-esteem, medication adherence and quality of life in people living with HIV. J AIDS Clin Res. 2013;4(11):256. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Bankoff SM, McCullough MB, Pantalone DW. Patient–provider relationship predicts mental and physical health indicators for HIV-positive men who have sex with men. J Health Psychol. 2013;18(6):762–72. [DOI] [PubMed] [Google Scholar]
- 10.Huang L, Li L, Zhang Y, Li H, Li X, Wang H. Self-efficacy, medication adherence, and quality of life among people living with HIV in Hunan Province of China: a questionnaire survey. J Assoc Nurses AIDS Care. 2013;24(2):145–53. [DOI] [PubMed] [Google Scholar]
- 11.Schnall R, Jia H, Olender S, Gradilla M, Reame N. In people living with HIV (PLWH), menopause (natural or surgical) contributes to the greater symptom burden in women: results from an online US survey. Menopause. 2018;25(7):744–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Scully EP. Sex differences in HIV infection. Curr HIV/AIDS Rep. 2018;15(2):136–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.World Health Organization. 2019. https://www.who.int/genomics/gender/en/.
- 14.Neil RC. Psychology: the science of behaviour. London: Pearson; 2010. [Google Scholar]
- 15.Puskas CM, Forrest JI, Parashar S, Salters KA, Cescon AM, Kaida A, et al. Women and vulnerability to HAART non-adherence: a literature review of treatment adherence by gender from 2000 to 2011. Curr HIV/AIDS Rep. 2011;8(4):277–87. [DOI] [PubMed] [Google Scholar]
- 16.Manteuffel M, Williams S, Chen W, Verbrugge RR, Pittman DG, Steinkellner A. Influence of patient sex and gender on medication use, adherence, and prescribing alignment with guidelines. J Women Health. 2002;23(2):112–9. [DOI] [PubMed] [Google Scholar]
- 17.Turmen T. Gender and HIV/AIDS. Int J Gynaecol Obstet. 2003;82(3):411–8. [DOI] [PubMed] [Google Scholar]
- 18.Beauchemin M, Gradilla M, Baik D, Cho H, Schnall R. A multi- step usability evaluation of a self-management app to support medication adherence in persons living with HIV. Int J Med Inform. 2019;122:37–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.ClinicalTrial.gov. The wise app trial for improving health out- comes in PLWH (WiseApp). Identifier NCT03205982. 2018. https://clinicaltrials.gov/ct2/show/NCT03205982.
- 20.ClinicalTrials.gov. Video information provider for HIV-associated non-AIDS (VIP-HANA) symptoms (VIP-HANA). Identifier NCT03205982. 2019. https://clinicaltrials.gov/ct2/show/NCT03182738.
- 21.Schnall R, Liu J, Mohr DC, Bakken S, Hirshfield S, Siegel K, et al. Multi-modal methodology for adapting digital health tools to new populations: adaptation of the video information provider (VIP) for persons living with HIV with HIV-associated non-AIDS (HANA) conditions. Stud Health Technol Inform. 2019;264:1347–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Schnall R, Jia H, Reame N. Association between HIV symptom burden and inflammatory cytokines: an analysis by sex and menopause stage. J Women Health. 2019;29(1):119–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Schnall R, Liu J, Reame N. Ecological momentary assessment of HIV versus reproductive health symptoms in women of differing reproductive stages living with HIV. Menopause. 2019;26(12):1375–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3):189–98. [DOI] [PubMed] [Google Scholar]
- 25.Berger BE, Ferrans CE, Lashley FR. Measuring stigma in people with HIV: psychometric assessment of the HIV stigma scale. Res Nurs Health. 2001;24(6):518–29. [DOI] [PubMed] [Google Scholar]
- 26.PROMIS. A brief guide to the PROMIS depression instruments. 2017. https://www.healthmeasures.net/images/PROMIS/manuals/PROMIS_Depression_Scoring_Manual_02222017.pdf.
- 27.PROMIS. A brief guide to the PROMIS anxiety instruments. 2019. https://www.healthmeasures.net/images/PROMIS/manuals/PROMIS_Anxiety_Scoring_Manual.pdf.
- 28.Mannheimer SB, Mukherjee R, Hirschhorn LR, Dougherty J, Celano SA, Ciccarone D, et al. The CASE adherence index: a novel method for measuring adherence to antiretroviral therapy. AIDS Care. 2006;18(7):853–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Reinius M, Rao D, Manhart LE, Wiklander M, Svedhem V, Pryor J, et al. Differential item functioning for items in Berger’s HIV stigma scale: an analysis of cohorts from the Indian, Swedish, and US contexts. Qual Life Res. 2018;27(6):1647–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Wright K, Naar-King S, Lam P, Templin T, Frey M. Stigma scale revised: reliability and validity of a brief measure of stigma for HIV+ youth. J Adolesc Health. 2007;40(1):96–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.HealthMeasures. PROMIS overview.2019. www.nihpromis.org/about/overview.
- 32.Schnall R, Liu J, Cho H, Hirshfield S, Siegel K, Olender S. A health-related quality-of-life measure for use in patients with HIV: a validation study. AIDS Patient Care STDs. 2017;31(2):43–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.R Development Core Team. R: a language and environment for statistical computing. Vienna: R Development Core Team; 2006. [Google Scholar]
- 34.Taber KS. The use of Cronbach’s alpha when developing and reporting research instruments in science education. Res Sci Educ. 2018;48(6):1273–96. [Google Scholar]
- 35.Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol. 1986;51(6):1173–82. [DOI] [PubMed] [Google Scholar]
- 36.Hopper D, Coughlan J, Mullen MR. Structural equation modeling: guidelines for determining model fit. Electron J Bus Res Methods. 2008;6:53–60. [Google Scholar]
- 37.Tovar E, Rayens MK, Gokun Y, Clark M. Mediators of adherence among adults with comorbid diabetes and depression: The role of self-efficacy and social support. J Health Psychol. 2015;20(11):1405–15. [DOI] [PubMed] [Google Scholar]
- 38.Albert PR. Why is depression more prevalent in women? J Psychiatry Neurosci. 2015;40(4):219–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Bertakis KD, Azari R, Helms LJ, Callahan EJ, Robbins JA. Gen- der differences in the utilization of health care services. J family Pract. 2000;49(2):147–52. [PubMed] [Google Scholar]
- 40.Redondo-Sendino A, Guallar-Castillon P, Banegas JR, Rodriguez- Artalejo F. Gender differences in the utilization of health-care services among the older adult population of Spain. BMC Public Health. 2006;6:155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Sohler NL, Li X, Cunningham CO. Gender disparities in HIV health care utilization among the severely disadvantaged: can we determine the reasons? AIDS Patient Care STDs. 2009;23(9):775–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Turner BJ, Laine C, Cosler L, Hauck WW. Relationship of gender, depression, and health care delivery with antiretroviral adherence in HIV-infected drug users. J Gen Intern Med. 2003;18(4):248–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Clayton JA. Applying the new SABV (sex as a biological variable) policy to research and clinical care. Physiol Behav. 2018;187:2–5. [DOI] [PubMed] [Google Scholar]
- 44.National Institutes of Health. Consideration of sex as a biological variable in NIH-funded research. 2015. https://grants.nih.gov/grants/guide/notice-files/NOT-OD-15-102.html