Correlates of and barriers to ART adherence among adherence-challenged people living with HIV in southern India

Abstract

Suboptimal adherence to Antiretroviral Therapy (ART) regimens can lead to the development of drug resistance, virologic and clinical failure, and, on the community level, the spread of drug-resistant HIV. To design effective interventions, it is crucial to understand locally specific barriers to optimal adherence. Self-report data from a cross-sectional sample of 527 adherence-challenged people living with HIV (PLWH) in the South-Indian state of Karnataka showed that they took on average 68% of prescribed doses in the past month. Large majorities of participants encountered individual (95%), social/structural (88%), and clinic/regimen (80%) adherence barriers. Multivariate linear regression analyses of past month adherence showed that disclosure to all adults in the household was positively related to adherence, as was employing a larger number of adherence strategies, perceiving more benefits of ART, and having been on ART for longer. Fears of stigmatization upon disclosure of HIV-status to friends and people at work were negatively related to adherence. These results suggest that some barriers, especially individual-level barriers like forgetfulness are very common and can be targeted with relatively simple individual-level strategies. Other barriers, related to fear of stigma and lack of disclosure may require family- or community-level interventions.

Introduction

Adherence to antiretroviral therapy (ART) is crucial for its success in controlling HIV-infection. Suboptimal adherence can lead to development of drug resistance, virologic and clinical failure (Gardner et al., 2008). On a community level, it contributes to the spread of HIV (Tanser, Bärnighausen, Grapsa, Zaidi, & Newell, 2013).

Given the limited availability of second-line ART in low and middle income countries (LMIC), it is crucial to understand barriers to optimal adherence in order to maximize the effectiveness of first-line therapy. Successful interventions need to target locally-specific adherence barriers, including stigma (Ekstrand et al., 2018; Simoni, Amico, Smith, & Nelson, 2010). According to a 2016 meta-analytic review (Shubber et al., 2016), the most frequently reported ART adherence barriers by HIV+ adults from 38 countries were forgetting, travel, being busy, changes in routine, and being asleep (pooled frequencies: 41-25%). Depression and substance misuse were reported by 16% and 13%, respectively. Common contextual barriers were stigma/secrecy (14%) and food insecurity (13%). Health-service-related barriers included distance to clinic (18%) and stock-outs (16%). Other recent reviews from global research found suboptimal adherence was also related to social/structural factors such as intimate partner violence (Hatcher, Smout, Turan, Christofides, & Stockl, 2015), unemployment (Costa, Torres, Coelho, & Luz, 2018; Nachega et al., 2015), financial constraints (Langebeek et al., 2014), male gender (Heestermans, Browne, Aitken, Vervoort, & Klipstein-Grobusch, 2016), and younger age (Gari et al., 2013; Ghidei et al., 2013). Healthcare- and ART-related barriers include using traditional medicine (Heestermans et al., 2016), dissatisfaction with healthcare facility/workers (Heestermans et al., 2016; Langebeek et al., 2014), concerns about the utility/necessity of ART (Langebeek et al., 2014), and lack of adherence self-efficacy (Gari et al., 2013; Langebeek et al., 2014). Anxiety (Wykowski, Kemp, Velloza, Rao, & Drain, 2019), and poor social support (Gari et al., 2013; Heestermans et al., 2016; Langebeek et al., 2014) are common psychosocial adherence barriers.

Many of these barriers exist in India as well. A review (Mhaskar et al., 2013) of eight studies before 2010 showed medication cost was the most commonly reported adherence barrier. After free first-line ART became available, financial problems were still reported as a barrier (Joshi et al., 2014; Shukla et al., 2016). Other perceived barriers included running out of medication (Mhaskar et al., 2013; Pina et al., 2018), side effects or opportunistic infections (Joshi et al., 2014; Mhaskar et al., 2013; Nyamathi et al., 2018), forgetting (Banagi Yathiraj et al., 2016; Joshi et al., 2014; Pina et al., 2018) or being busy/away from home (Shukla et al., 2016). Furthermore, suboptimal adherence in India has been associated with illiteracy (Joshi et al., 2014; Mehta, Baxi, Patel, & Parmar, 2016), male gender (Joshi et al., 2014), younger age (Pina et al., 2018), alcohol use (Banagi Yathiraj et al., 2016; Pina et al., 2018; Schensul et al., 2017), dissatisfaction with healthcare facilities (Shukla et al., 2016), negative beliefs about ART (Pina et al., 2018; Shukla et al., 2016), anxiety and depression (Joshi et al., 2014; Kleinman et al., 2015; Schensul et al., 2017; Shukla et al., 2016), non-disclosure (Joshi et al., 2014; Pahari, Roy, Mandal, Kuila, & Panda, 2015), poor social support (Joshi et al., 2014; Vallabhaneni, Chandy, Heylen, & Ekstrand, 2012), and stigma (Ekstrand et al., 2018).

The purpose of this paper is to describe suboptimal adherence and its correlates in a sample of adherence-challenged people living with HIV (PLWH) in Karnataka, the state with the sixth highest prevalence of PLWH in India in 2017 (National Aids Control Organisation, 2018). By focusing on individuals with suboptimal adherence, a deeper understanding of local adherence patterns and barriers can be achieved, which in turn allows for the development of more successful interventions.

Methods

Setting and Sample

Data are from baseline interviews of a cluster-randomized controlled efficacy trial of a comprehensive adherence intervention (Ekstrand et al., 2020). Participants were recruited from NGOs and ART centers in Bengaluru and three smaller towns 60-160 km away (Mysuru, Kolar, Chikkaballapura) in the Southern Indian state of Karnataka. Eligibility criteria were being ≥18 years of age, speaking Kannada, residing within 40km of an intervention site and self-reporting <90% adherence to ART over the past 30 days or more than two treatment interruptions of ≥48 hours in the past year. Data were collected in face-to-face interviews between November 2013 and March 2017.

This study was approved by the Indian Health Ministry Screening Committee and the Ethics Committees at the University of California, San Francisco, and St. John’s Medical College and Hospital, Bengaluru, India.

Measures

Demographics and regimen

Demographics included age, religion, gender, education, employment and household income. We also asked current ART regimen and when the participant had started ART.

Adherence

Adherence in the past month was assessed using a visual analogue scale (VAS) (Giordano, Guzman, Clark, Charlebois, & Bangsberg, 2004). The participant indicates a point on a line running from 0 to 100 that best corresponds with the percentage of pills he/she has taken. This measure has previously been found to predict viral load in this setting (Ekstrand, Chandy, Heylen, Steward, & Singh, 2010).

Adherence and Clinic-Attendance Barriers

Barriers to adherence in the past six months consisted of individual (seven items, e.g. being busy), family (three items, e.g. financial difficulties), social/structural (seven items, e.g. perceived societal stigma), and regimen/clinic-related barriers (nine items, e.g. side effects, drug stock-outs). Four additional items measured barriers to clinic attendance (e.g. clinic wait time). Responses measured on a 4-point scale from “never” to “most of the time” were dichotomized into never vs. ever and summed per category.

Adherence Strategies

Thirteen items assessed strategies participants used to remember to take their ART medication in the past six months (e.g. using a pill organizer). The number of strategies endorsed was tallied.

Adherence Support

We assessed social support to take ART from eight types of people, including spouse, family members, and friends. “Yes” responses were summed and dichotomized into 0 vs. ≥1.

Perceived Symptoms/Side Effects

Participants were asked if in the past six months they had experienced any of 17 symptoms identified as side effects of the available ART medications (e.g. diarrhea, rashes). An index was created based on the number of symptoms endorsed.

Perceived ART Benefits

Participants reported whether they had experienced any of 13 positive effects of ART in the past six months (e.g. feeling more alert). The number of perceived benefits endorsed was summed.

Disclosure of HIV-status

We assessed whether participants had disclosed their HIV-status to their spouse, parents, and other relatives. Combined with information regarding with whom the participant lived, a disclosure variable was constructed and scored 1 if the participant had disclosed to all adults with whom he/she lived or if the participant was the only adult in the household, and scored 0 otherwise.

Internalized Stigma

We assessed whether participants had internalized stigmatizing social norms regarding HIV (e.g. bringing shame onto one’s family) in a 10-item measure (Steward et al., 2008). Responses were measured on a 4-point scale (0 “not at all” to 3 “a great deal”), and summarized by taking their mean (Cronbach’s alpha = 0.83).

Disclosure Avoidance Strategies

We assessed strategies used to avoid disclosure of HIV-status using 15 items (Steward et al., 2008), e.g. describing one’s illness as tuberculosis instead of HIV. Responses were measured on a 4-point scale (0 “never” to 3 “often”). The number of items with scores ≥1 was summed.

Stigma Fears

Participants were asked to report on a 0 “not at all” to 3 “very worried” scale to what extent they feared stigmatizing reactions (e.g. verbal or physical abuse, job loss) upon disclosure of their HIV-status to family members (six items), friends (six items), colleagues (four items), health care workers (HCW, five items), and community members (twelve items) who did not yet know their status. Subscales were created for each of these categories by averaging over all items per category. Cronbach’s alphas for the subscales ranged from 0.88 to 0.95. If a participant had already disclosed to all members of a particular category, the questions were skipped and a subscale score of 0 was assigned.

Depression

We measured depression severity using the Patient Health Questionnaire (PHQ-9) (Kroenke, Spitzer, & Williams, 2001). Participants reported how often they had experienced nine problems in the last two weeks on a 4-point scale (1 “not at all” to 3 “every day”). Scores were summed. Standard cut offs for levels of distress ranging from none to severe were used (Cronbach’s alpha = 0.84).

Viral load (VL)

HIV plasma VL was determined via real-time PCR assay with a fluorescein-labeled TaqMan probe for quantification of HIV particles (Palmer et al., 2003). Detectable VL was defined as ≥100 copies/ml of blood.

Analyses

Descriptive analyses consisted of frequencies for categorical variables and means and standard deviations (SD) for continuous variables. Bivariate associations with VAS adherence were assessed using linear regression. We explored the association between VAS and demographic, psychosocial, and ART-related variables found to be significantly associated in previous studies. Variables associated at the p<0.20 level bivariately were added to a multivariate regression model. To maximize the available sample size for the variable that assessed stigma fears with respect to people at work, we scored participants who were not employed as 0. We did the same for stigma fears from HCW if those questions were not applicable because participants indicated all HCW knew their HIV-status. In sensitivity analyses, we reran the regressions without these participants. The issue did not occur for the other stigma fear variables.

Results

Demographic Characteristics

The mean age of the 527 participants was 38.7 (SD=8.4) and half were male (Table 1). Over half were currently married. The majority were Hindu, two-thirds of participants had less than 10 years of education, and three-quarters were employed.

Table 1:

Demographic description of the sample (n=527)

	N	%
Male	263	49.9
Marital status
Married a	294	55.8
Widowed	149	28.3
Divorced/separated	37	7.0
Single	47	8.9
Living situation
Alone	30	5.7
Nuclear family	245	46.5
Extended family	235	44.6
Other	17	3.2
Education (years)
<4	214	40.6
4-9	149	28.3
≥10	164	31.1
Employed	396	75.1
Hindu religion	486	92.2
Age, mean (SD)	38.7	(8.4)

^aIncludes those living together as married and one male participant married to his male partner

Adherence and ART-related Variables

All but 9 participants were on a first-line regimen of 2 NNRTIs + 1 NRTI. Participants had been on ART for an average of over 3.5 years. Mean percent of pills taken in the past month (VAS) was 68.1% (SD=24.6; see Table 2 for a categorical breakdown). Over half had detectable VL, and these 277 participants were on average 11.4% less adherent than virally suppressed participants (mean VAS 62.7 vs. 74.1, p<0.001). Most participants endorsed individual (n=501, 95.1%), social/structural (n=466, 88.4%), and clinic/regimen (n=420, 79.7%) adherence barriers in the past six months. Family-level barriers were reported by 24.3% (n=128) and 41.0% (n=216) had experienced clinic-attendance barriers. Mean number of barriers endorsed in each category is reported in Table 2. The most endorsed items were being away from home (85.0%) being busy with other things (81.8%), and forgetting to take the medication (68.3%).

Table 2:

ART, adherence and psychosocial characteristics of the sample (n=527):

	N	%
VAS, categorical
0 – 50 %	98	18.6
51 – 60 %	18	3.4
61 – 70 %	58	11.0
71 – 80 %	206	39.1
81 – 90%	147	27.9
ART regimen
Zidovudine + Lamivudine + Nevirapine	257	48.8
Tenofovir + Lamivudine + Efavirenz	179	34.0
Other 2 NRTI + 1 NNRTI combination	82	15.6
Protease inhibitors	9	1.7
Detectable viral load (≥ 100 copies/ml), n=526	277	52.7
Any social support to remember to take ART	439	83.3
At least moderately depressed (PHQ ≥ 10), n = 526	63	12.0
Disclosed to all adult household members, n=524	433	82.6
	Mean	SD
VAS	68.1	24.6
Time on ART, in months	45.2	34.2
No. of perceived side effects of ART (0 – 17)	3.9	3.6
No. of perceived benefits ART (0 – 13)	10.2	3.7
Internalized stigma(0 – 3)	0.9	0.7
Stigma fears re. family (0 – 3)	0.9	1.0
Stigma fears re. friends (0 – 3)	2.3	1.0
Stigma fears re. people at work (0 – 3)	1.3 / 1.7 a	1.3 /1.3a
Stigma fears re. HCW (0 – 3)	0.9 / 1.1b	1.2 / 1.2b
Stigma fears re. community members (0 – 3)	1.4 /	1.0
Avoidant coping index (0 – 15)	4.4	2.6
No. of individual adherence barriers (0 – 7)	2.5	1.2
No. of family adherence barriers (0 – 3)	0.3	0.5
No. of clinic and regimen adherence barriers (0 – 9)	1.4	1.1
No. of social and structural adherence barriers (0 – 7)	1.5	1.0
No. of clinic-attendance barriers (0 – 4)	0.7	1.0
No. of adherence strategies used (0 – 13)	4.1	1.6

HCW, health care workers; NRTI, Nucleoside Reverse Transcriptase Inhibitor; NNRTI, Non-Nucleoside Reverse Transcriptase Inhibitor

^aFirst result is based on all participants; those not-employed scored 0. Second result is based on employed participants (n=396) only.

^bFirst result is based on all participants; those who had disclosed to all HCW scored 0. Second result is based on only participants who had not fully disclosed to HCW (n=411).

To facilitate adherence, most participants (83.3%) reported having someone to remind them to take their ART medication. They used a mean of 4.1 (SD=1.6) strategies to help them adhere. The most common strategies were keeping medications in a convenient location (n=468, 88.8 %), carrying medications while away from home (n=428, 81.2%), and refilling prescriptions before running out (n=389, 73.8%).

Participants reported a mean of 3.9 (SD=3.6) ART side effects, most commonly fevers (n=254 48.2%), fatigue (n=248, 47.1%), and joint/leg pain (n=194, 36.8%). They endorsed on average 10.3 (SD=3.7) out of 13 perceived ART benefits, especially feeling healthier (n=442, 83.9%), enjoying social life more (n=436, 82.7%), and feeling more hopeful (n=432, 82.0%).

Disclosure, Stigma and Depression

A majority (82.6%) had disclosed their HIV-status to all the adults in their household. Based on their PHQ-scores, 12% of participants had moderate to severe depression. The mean internalized stigma score was 0.9 out of 3 (SD=0.7). Ninety-one percent of respondents reported some level of fear of stigma after HIV-status disclosure from their friends (n=478) and community (n=479), 61.1% (n=322) feared stigma from family members. Nearly three quarters (284/396) of employed participants expressed some fear of stigma from people at work. Of the 411 participants who visited any HCWs who did not know their HIV-status, 54.0% (n=222) feared stigma from HCW. The mean level of feared stigma for each group is reported in Table 2. Participants used on average 4.4 strategies (SD=2.6) to avoid disclosure. The most common strategies were making up reasons for HIV-related medical visits (n=338, 64.1%) and hiding ART pills (n=321, 60.9%).

Correlates of Adherence

We examined the association between past month adherence (VAS) and several demographic and other key measures, first bivariately and then multivariately for variables with a bivariate significance level of p<.20. As shown in Table 3, three ART-related measures were positively related to adherence when controlling for other variables in the model: employing more adherence strategies (b=3.08; p<0.001), perceiving a greater number of ART benefits (b=0.75; p=0.013), and time on ART (b=3.14; p=0.002).

Table 3:

Results of linear regression of % adherence past month (VAS) (n=524)

	Unadjusted		Adjusted
	b	p-value	b	p-value
Female	3.33	0.121	3.24	0.148
Age	0.20	0.128	0.20	0.123
Employed	3.62	0.145	4.16	0.090
Household income		0.233
≤5000	Ref
5001-10000	4.15
>10000	1.77
Education: >10 years	1.39	0.232
Any adherence support	6.03	0.036	0.65	0.821
No. of adherence strategies	3.87	<.001	3.08	<.001
No. of ART side effects	0.06	0.850
No. of perceived ART benefits	1.25	<.001	0.75	0.013
Months on ART, ln transformed	4.51	<.001	3.14	0.002
Disclosure to adults in household	8.08	0.004	5.67	0.048
Moderate-severe depression	−1.24	0.709
Internalized stigma	−0.81	0.588
No. of avoidant coping strategies	0.56	0.166	0.87	0.029
Stigma fears friends	−1.80	0.096	−2.82	0.007
Stigma fears family	0.10	0.929
Stigma fears people at work	−0.85	0.293
Stigma fears HCW	0.03	0.970
Stigma fears community	−0.78	0.486

b, linear regression coefficient; HCW, health care workers; ln, natural log

Of the psychosocial variables, disclosure to adult household members was positively associated with adherence (b=5.67; p=0.048). Depression and internalized stigma showed no association with adherence. Of the stigma fear variables, only fear of stigma from friends was included in the multivariate model and was negatively associated with adherence (b=−2.82; p=0.007). Sensitivity analyses with only those participants who had not disclosed to all HCW (n=411), showed that the effect of HCW-stigma fears on adherence remained non-significant (p=0.643). Using only the sample of 396 employed participants, fear of stigma at work was significantly negatively related to adherence (b=−2.25; p=0.029).

Finally, using more avoidant coping strategies was related to better adherence (b=0.87; p=0.029), contrary to our expectations.

Discussion

In this study of adherence-challenged PLWH from mostly urban/peri-urban southern India, self-reported mean adherence in the past month was 68.1%. The most commonly reported adherence barriers correspond to those found to be prevalent in other international and Indian studies: being away from home, busy with other things, and simply forgetting to take the medication (Banagi Yathiraj et al., 2016; Joshi et al., 2014; Shubber et al., 2016; Shukla et al., 2016). These individual-level barriers can be addressed by teaching PLWH strategies like setting reminders on one’s private mobile phone or learning to associate daily events (e.g. a TV program) with time for pill-taking. Employing more adherence strategies was associated with better adherence in the present analyses.

Also positively associated with adherence was the perception of more benefits from ART, a finding in line with several previous studies (Croome, Ahluwalia, Hughes, & Abas, 2017; Langebeek et al., 2014). Along with other strategies to improve adherence, HCW could (re-)emphasize benefits that may not be apparent to a patient struggling to see the use of taking daily medication and staying in care for a still stigmatized disease. This is all the more relevant as some studies found that feeling healthy was seen as a reason to stop ART (Croome et al., 2017; Mills et al., 2006).

As expected based on earlier studies (Heestermans et al., 2016; Joshi et al., 2014; Pahari et al., 2015; Shubber et al., 2016), disclosure at home was positively related to ART adherence. The proportion of participants who reported disclosure of their HIV-status to all adult household members was over 80% in our sample, which was within the range of 65-90% found by others in India (Joshi et al., 2014; Pahari et al., 2015; Ramchandani et al., 2007). This high number is perhaps not surprising, though, given the National Aids Control Organisation’s (NACO, 2013) requirement that PLWH bring a support person to the ART clinic before being prescribed ART, which requires disclosing. While this support person does not have to be a household member, it usually is. After disclosure to one household member, disclosure to others in the home may be easier. Full disclosure at home eliminates the potential need to hide pills and daily pill ingestion from family members, which is difficult in a setting with very little privacy such as the typical Indian home.

Such privacy worries are also a likely explanation for the observed levels of feared stigma from friends and co-workers if the PLWH’s status were disclosed, and for the negative association of these stigma fears with adherence. Disclosure to friends and to people at work was low in our sample – post hoc analyses showed about 70% had not disclosed to any of their friends or anyone at work. But contact with friends and especially people at work is frequent and privacy limited, causing PLWH to skip pills when in the presence of friends or co-workers who do not know their status.

Depression was not significantly related to adherence in our study, which may be due to the low number of depressed participants in our sample. Many of these PLWH had been on ART for years and they and their families may have come to terms with the HIV-infection and stopped internalizing depressing HIV stigma norms and fearing stigma by relatives (Steward et al., 2011).

This study had limitations that need to be acknowledged. Data were cross-sectional, so no causality can be inferred. Our sample consisted of largely urban/peri-urban PLWH from southern India and results may not necessarily generalize to PLWH from rural areas or other regions. We also relied on self-report and hence social desirability or recall bias may have led to misrepresentation of actual adherence. However, in our earlier studies these self-reported adherence measures have been found to be associated with viral load and resistance mutations (Ekstrand et al., 2010; Ekstrand et al., 2011).

Conclusion

This adherence-challenged sample of PLWH reported mostly individual-level adherence barriers. In line with other recent studies, drug supply problems, costs, and long travel times to the clinic were less of an issue, likely due to the fact that this was an area where the government has had time to build local ART centers and adequate supply networks of free first-line regimens. The results largely agree with other global studies on ART adherence barriers. However, seemingly similar barriers may have different underlying reasons and require different solutions depending on the local context. Adherence counsellors, peer navigators/support groups and local NGOs are all valuable sources of information and support. They can share and encourage the use of multiple adherence strategies, and if deemed safe, help with disclosure to family, as well as mitigate the fear of stigmatization and its negative consequences that many PLWH experience.