Examining engagement in care of women living with HIV in South India

Abstract

HIV seropositive adherence-challenged women, who reported being on ART for at least four months were interviewed. Data on healthcare history, anti-retroviral therapy, clinic visits, doctor communication, disclosure and fear of stigma were collected. Better engagement in care was significantly more likely among older women, ≥ 10 years of education, higher income, HIV status disclosure to family, with higher community stigma fears and fewer healthcare access barriers. To promote retention, women may be encouraged to consider disclosing their HIV serostatus to supportive household members. A variety of possible interventions to overcome the prevalent barriers to care are provided.

Sustained engagement in care (EiC) of HIV infected persons is the key to the success of ART (Antiretroviral Therapy) programs. Women living with HIV face several barriers to EiC. There is an urgent need to address these barriers and form sustainable solutions to the common issues which can be adapted to any setting. Hence, this work was conceived by the authors with the specific aim of probing into psychosocial, cultural and female gender-specific issues pertaining to EiC of women living with HIV in a low/middle income setting irrespective of their reproductive status or age. Prior research has identified that women with challenges to medication adherence (Renju et al., 2017) are more likely to disengage from care and these women constituted our sample population. They were interviewed using a structured survey instrument. In this paper, we present the results of our study on challenges to EiC among HIV+ women seeking HIV care from government ART centers and National AIDS control organization (NACO) aided private ART centers in South India and discuss practical strategies for their redressal.

ART is currently recommended for all individuals infected with HIV, regardless of their CD4 cell count. Soon after the WHO’s (WHO, 2015) global policy of “test and treat” all people with HIV was adopted by the Government of India in 2017 (NACO, 2017) there has been an emphasis on ART expansion with a focus on early initiation, making the need for efficient retention in HIV care greatly relevant to attain the 90–90-90 goal (UNAIDS, 2014). However, complex individual, interpersonal, contextual and community factors interfere with initiation and retention of individuals in HIV care (Hodgson et al., 2014). Commonly cited individual level retention barriers are younger age, low literacy, longer duration of disease, drug/alcohol use, depression and neurocognitive involvement, transgender identity or female sex, poor recent health, lack of health insurance, financial constraints including not owning a home and need of public transport (Hodgson et al., 2014; Howarth et al., 2017). Interpersonal level barriers include friends/household/partner stigma, health care provider and employment issues. Community barriers that have been reported include community blame, stigma and ostracism resulting in denial of rights and social participation (Hodgson et al., 2014; Howarth et al., 2017; Topp et al., 2018). Contextual factors include rural residence, distance from treatment facility, lack of transport or help with childcare, public hospital as place of treatment leading to fear of loss of confidentiality and long clinic waiting times, to mention a few (Hodgson et al., 2014; Howarth et al., 2017; Topp et al., 2018).

Many of these barriers are more acute in women, due to their dependent status in many societies and reproductive role. Though there has been a wealth of evidence on retention barriers of women of childbearing age, (Chadambuka et al.,2017; Besada et al.,2018; Erlwanger et al.,2017; Knettel et al.,2018; Katirayi et al.,2016; Parker et al.,2015; Vrazo et al., 2018) female gender specific issues with engagement in care (EiC) of HIV infected women unrelated to pregnancy and childbirth are sparse in the literature and no such studies have been reported on women living with HIV (WLWH) in India. Even in pregnant and postpartum women, delays in ART initiation and disengagement from care are common, both in low/middle income countries and high income countries with robust ART services (Bolsewicz, Debattista, Vallely, Whittaker, & Fitzgerald, 2015; French, Thorne, Byrne, Cortina-Borja, & Tookey, 2017; Howarth et al., 2017). A systematic review of individual and contextual factors affecting ART initiation and adherence in pregnant and post-partum women showed that younger age, poor knowledge of PMTCT (prevention of maternal to child transmission) and ART, fear of domestic violence leading to non-disclosure, spouse/partner dependence, and presentation to antenatal care late in pregnancy were the barriers, while desire to remain healthy and protect the child, early enrollment in antenatal and HIV care right from the first pregnancy, successful completion of PMTCT, and community health worker involvement were enablers to engagement in HIV care (Hodgson et al., 2014; Vrazo et al., 2018). We report here our findings pertaining to EiC in women living with HIV in South India.

Methods

Study Design and Sample

This cross-sectional analysis includes the female participants enrolled in the Chetana study, which is a cluster randomized trial (CRT) to evaluate an intervention designed to improve adherence to ART among people living with HIV (PLWH) seeking health care in public and government-aided private ART centers in South India.

Participants and Procedures

Potential participants were referred to the study interviewers by their physician, the outpatient clinic clerk or ART center personnel. After eligibility confirmation and a written informed consent, a face-to-face interview was administered by trained study staff using the standard survey instrument at a private location, distant from the ART clinic. All participants were 18 years of age or older, HIV seropositive, on ART and had challenges with adherence, defined as ≤ 90% adherence to ART or presence of treatment interruptions of ≥2 days during the past year. The interview lasted approximately one hour and covered a variety of topics, including demographics and health care history, date of HIV diagnosis, ART start date, their current and past ART regimens, MD communication, and psychosocial aspects like stigma fears and disclosure. Many of these measures have been used previously in members of this population in our former projects (Ekstrand et al., 2010). All the measures were developed in English and then translated into Kannada, the local language. All translations were independently back-translated into English to ensure semantic equivalence.

Participants were recruited from two types of settings: public (government ART centers) as well as public-private collaborative setting (St John’s Medical College Hospital) between Sept 2012 and March 2017. Both of these provide free point of care services and ART through the government-sponsored program per NACO protocol (NACO, 2017). According to the existent NACO guidelines (NACO, 2013) at the time of data collection, all individuals with CD4 cell count ≤ 350 cells/μl and symptomatic individuals of any CD4 count were eligible for ART. The most commonly prescribed antiretroviral therapy was a generic NRTI backbone based regimen, in a fixed dose combination along with one of the NNRTI’s (NACO, 2013).

Two hundred and seventy-five female participants were interviewed of which eleven were excluded because they had been on ART for three months or less, which we considered not long enough to reliably establish consistent EiC. Hence baseline data of 264 participants were used in these analyses.

Measures

Demographics

Age, marital status, number of children, living situation (alone, nuclear or extended family, other), employment, level of education, income and religion were measured.

HIV diagnosis and start of ART

Participants were queried on the date when they were first diagnosed and informed about their HIV infection and on the date of initiation of ART.

Time to ART initiation

Time in months from the date of HIV diagnosis to the start of ART was calculated as the time to initiation.

ART medications

Participants were asked about the ART medications prescribed, dose, schedule, length of time on the current regimen and special instructions, if any. We validated their self-report by either checking their drugs or doctor’s prescription.

ART side effects

Patients responded 0 (no) or 1 (yes) to 17 common symptoms and side effects related to ART medicines. The sum of side effects endorsed by each participant was calculated.

Health care access barriers

A four item list assessed barriers to health care access specifically prevalent among women (transportation, childcare, time off from household chores/work and long wait-times at the clinic). Participants were asked how often in the past 6 months these factors had made it difficult for them to go and see their doctor. Response options were noted on a four-point scale ranging from 0 never, to 3 most of the time. For analyses, a dichotomous variable was created, scored 0 if all items had been answered with ‘never’, and scored 1 otherwise.

Support in obtaining ART

Participants responded to an eight-item list opening with “In the past 6 months, have you received help from …to get your HIV medications?” Items included, e.g. spouse, siblings, friends. The response options, 0 (no) or 1 (yes), were summed. Again, a dichotomous variable was created to indicate no (sum=0) vs. any help (sum>0) in getting ART.

Disclosure

If HIV status was known to all household members with whom the person lived, other than children, HIV-status was considered disclosed.

MD communication

A ten item list assessed participants’ perceptions about their HIV doctor’s knowledge, attention, and problem-solving skills as well as satisfaction with the health care provider. Using a four-point scale from 0 (never) to 3 (every time), participants responded to questions such as, “Do you feel that you doctor really listens to your concerns?”. Items were averaged for scoring. Internal consistency of the scale was α = 0.95.

Fear of Stigma

Five subscales were used to assess fear of being stigmatized upon disclosure to friends (six items), family (six items), healthcare workers (five items), people at work (four items), and the community (twelve items). Participants responded to “How worried are you that the following (e.g., verbal or physical abuse, isolation, humiliation) might happen if you were to disclose your HIV status to (e.g., the friends) who don’t already know?”. Response options ranged from 0 (not at all worried) to 3 (very worried). For each sub-scale, the mean was calculated. Cronbach’s α ranged from 0.87 for stigma fears about people at work to 0.95 for stigma fears about health care workers.

Engagement in care (EiC)

Participants were queried about the number of times they visited HIV care services in the past 6 months. Since ART medications are dispensed monthly and patients counselled to attend follow-up visits on the exact dates for the succeeding month charted in their treatment folders, a count of six times was considered as fully engaged in care. Also, situations where the participants came once in two months to procure anti-retrovirals for two months, or where after requesting the treating physician relatives picked up ART on their behalf were included. Participants who had only been taking ART for four or five months were included and their calculation of EiC was adjusted accordingly, but those on ART for three months or less were excluded.

Data analysis

Baseline data from the CRT were used in the current analyses. We calculated descriptive statistics consisting of frequencies for categorical variables and mean plus standard deviation (SD) for continuous variables and assessed differences in these variables between those engaged vs. not engaged in care via χ2 and t-test, respectively. Variables associated with EiC in these bivariate analyses at a significance level of p<0.20 were included in a subsequent multivariate logistic regression model to identify which of them remained independently associated with EiC. All variables were entered in the initial model and non-significant variables were removed stepwise until only significant variables remained in the model. The logistic regression was performed in Stata 15, all other analyses in SPSS 25. All significance levels reported are two-sided.

Results

Of the 264 participants, 214 (80.7%) were consistently engaged in care. Table 1 shows the demographic characteristics. The mean (SD) age of the participants was 36.5 (7.4) years (range 20 – 65). Half of the sample reported being widowed (n=135), 37.9% (n=100) was married. The majority was Hindu (90.5 %, n=239), had between one and six children (88.6%, n=234), lived with their extended (44.3%, n=117) or “nuclear” (46.6%, n=123) family, and was employed (64.0%, n=169). Nearly half (n=130) had less than four years of education. Participants reported having been diagnosed with HIV for a median duration of 65.5 months (IQR: 37.3 – 96.8) and starting antiretroviral medication for a duration of 47 months (IQR: 24.3 – 71) before the interview (Table 2). The median time to initiation of ART was 2 months (IQR: 0 – 29). About half (45.8%, n=121) reported barriers and 61.4% (n=162) acknowledged support for procuring ART. In all, 89.3% (n=234) of participants had their status disclosed to their household members. All but five (98.1 %) of the participants were on NNRTI-based regimens, most commonly lamivudine/zidovudine/nevirapine (52.7%, n=139), followed by lamivudine/tenofovir/efavirenz (30.7 %, n=81). The other five participants were on protease inhibitors. The mean (SD) number of side effects reported was 4.6 (3.7). The mean scores (SD) for fear of stigma on a scale ranging from 0 to 3 were 2.3 (1.0) for friends, 1.0 (1.3) for colleagues, 1.0 (1.2) for health care workers, and 1.6 (1.0) for community.

Table 1.

Demographic description of the sample (n=264)

Age: mean (SD)	36.5	(7.4)
Marital Status: n (%)
Married	100	(37.9)
Widowed	135	(51.1)
Divorced/separated	24	(9.1)
Other	5	(1.9)
Number of children: n (%)
0	30	(11.4)
1	89	(33.7)
2	104	(39.4)
3–6	41	(15.5)
Living situation: n (%)
Nuclear family only	123	(46.6)
Extended family	117	(44.3)
Alone	18	(6.8)
Other	6	(2.3)
Hindu religion: n (%)	239	(90.5)
Education: n (%)
< 4 yrs	130	(49.2)
4 – 9 yrs	69	(26.1)
≥ 10 yrs	65	(24.6)
Personal income: n (%)
None (not employed)	95	(36.0)
1 – 5,000 Rupees	118	(44.7)
5,001 – 10,000 Rupees	48	(18.2)
>10,000 Rupees	3	(1.1)

Table 2.

HIV- and ART-related characteristics of the sample (n=264)

Regimen:
Zidovudine/Lamivudine/Nevirapine	139	(52.7)
Zidovudine/Lamivudine/Efavirenz	14	(5.3)
Tenofovir/Lamivudine/Nevirapine	25	(9.5)
Tenofovir /Lamivudine/ Efavirenz	81	(30.7)
Protease inhibitors	5	(1.9)
Months since HIV diagnosis: median (IQR)	65.5	(37.3 – 96.8)
Months on ART: median (IQR)	47	(24.3 – 71)
Months to initiation of ART: median (IQR)	2	(0 – 29)
Household members know HIV status: n (%)	234	89.3
Any barriers re. travel, child care, wait times: n (%)	121	45.8
Any support to get ART: n (%)	162	61.4
Number of side effects: mean (SD)	4.6	(3.7)
MD communication (0–3): mean (SD)	2.0	(0.9)
Stigma fears friends (0–3): mean (SD)a	2.3	(1.0)
Stigma fears family (0–3): mean (SD)a	1.0	(1.0)
Stigma fears colleagues (0–3): mean (SD)a	1.0	(1.3)
Stigma fears HCW (0–3): mean (SD)a	1.0	(1.2)
Stigma fears community (0–3): mean (SD)a	1.6	(1.0)

^an=252

IQR: Interquartile Range

Demographic, treatment related, Psychosocial

We initially explored bivariate associations of three sets of key factors pertaining to EiC of these women: demographic, HIV disease related and psychosocial. Results are presented in columns 2–4 of Table 3

Table 3.

Unadjusted and adjusted odds ratios of retention in care

	Bivariate			Initial multivariate modela			Final multivariate modelb
	OR	95% CI	p-value	AOR	95% CI	p-value	AOR	95% CI	p-value
Age	1.03	0.99 – 1.08	0.142	1.05	1.00 – 1.11	0.034	1.05	1.00 – 1.11	0.033
≥ 10 years education	2.35	1.00 – 5.52	0.049	4.49	1.45 – 13.89	0.009	4.70	1.54 – 14.36	0.007
Personal income: none	(ref)			(ref)			(ref)
1 – 5000	1.74	0.91 – 3.34	0.096	2.20	1.04 – 4.63	0.039	2.08	1.00 – 4.36	0.051
> 5000	4.43	1.45 – 13.5	0.009	6.14	1.57 – 24.02	0.009	5.89	1.54 – 22.62	0.010
Months on ART	1.01	1.00 – 1.02	0.092	1.00	0.99 – 1.02	0.537
Number of side effects	0.99	0.92 – 1.08	0.877
MD communication	0.85	0.59 – 1.22	0.371
Any barriers re. travel, child care, wait times	0.64	0.34 – 1.18	0.150	0.41	0.20 – 0.86	0.018	0.42	0.20 – 0.87	0.020
Any support to get ART	1.39	0.75 – 2.58	0.293
Household members know HIV status	2.17	0.92 – 5.12	0.078	3.00	1.08 – 8.32	0.035	2.96	1.08 – 8.10	0.035
Stigma fears friends	1.29	0.95 – 1.74	0.104	1.21	0.81 – 1.80	0.360
Stigma fears family	1.11	0.81 – 1.53	0.504
Stigma fears colleagues	1.04	0.81 – 1.34	0.741
Stigma fears HCW	1.21	0.92 – 1.60	0.180	1.09	0.79 – 1.50	0.586
Stigma fears community	1.46	1.05 – 2.04	0.025	1.58	1.01 – 2.47	0.043	1.81	1.24 – 2.64	0.002

^acontains all predictors with bivariate p<.20

^bafter backward elimination of non-significant predictors

Analysis of bivariate associations of age, education, having a job and/or income showed that ≥ 10 years of education (p=0.049) and a higher personal income (p=0.009) were positively associated with EiC. The bivariate association of time since initiation of ART (p=0.09) was marginally significant whereas drug side effects and MD communication issues showed no significant association with EiC. Bivariate analysis of psychosocial characteristics like disclosure of HIV status to household members (p=0.078) showed a trend to better EiC. Health care access barriers (transport, being busy with household chores and long waiting times at the clinic) showed no significant relationship. Likewise, stigma fears upon disclosure to health care workers, family members, colleagues and friends showed no significance, whereas community stigma fears were positively (p=0.025) associated with EiC.

To examine whether these bivariate associations remained independently associated, a subsequent multivariate logistic regression analysis was done including all variables with p-value < 0.20: education, income, age, time on ART, household members know HIV status, any health care access barriers and stigma fears from friends, health care workers and community. Non-significant variables were eliminated until all predictors left in the model were significant (Table 3). It was found that those with older age (AOR 1.05; CI [1.00–1.11]), ≥ 10 years of education (AOR 4.70; CI [1.54–14.36]), higher income (AOR 5.89; CI [1.54–22.62]), whose household members know their HIV status (AOR 2.96; CI [1.08–8.10]), and no health care access barriers (AOR 2.39; CI [1.15–4.96]), had significantly higher odds of being engaged in care than their respective counterparts. Stigma fears with respect to health care workers, friends and colleagues showed no significant association with those engaged in care. On the other hand, counterintuitively, we did find that those with greater community stigma fears had significantly (p=0.002) higher odds of engaging in care (AOR 1.81; CI [1.24–2.64]).

Discussion

To the best of our knowledge, there have been no published papers on female gender specific EiC among women living with HIV in India. We found a relatively high EiC rate (81%) over the past 6 months (defined as the proportion of patients regularly visiting the ART center to procure antiretrovirals for the past 6 months) in our cohort. Overall retention rates of PLWH in the US ranged between 60 to 87% (Rebeiro et al., 2016) with retention being defined as proportion of patients making at least 2 HIV medical visits at least 90 days apart during a one year period (Roscoe & Hachey, 2017), and retention rates using the same measure were around 58% in China (Yang, Yan, Liu, Huang, & Long, 2015) and between 50 and 67% in Sub-Saharan Africa (Babatunde et al., 2015; Haas et al., 2018). Retention rates of US women one year post-partum have been found to range between 37% and 75% (Momplaisir, Storm, Nkwihoreze, Jayeola, & Jemmott, 2018). Similar rate ranges are reported for post-partum women in countries within Africa and Asia (Hodgson et al., 2014; Knettel et al., 2018; Koirala, Deuba, Nampaisan, Marrone, & Ekstrom, 2017). By another indirect measure of retention i.e. the loss to follow-up (Matsumoto et al., 2015), defined as missing an appointed visit for more than three months, a high proportion of HIV-infected mothers and exposed babies were lost to follow-up (LTFU) in Mandalay, Myanmar as evidenced by a study reviewing routinely collected data of their Integrated HIV care program (Kyaw et al., 2017). LTFU rate was estimated to be 7 per 1000 person-years for pregnant women, and 10 per 1000 person-years for their babies.

Our results are in line with existing literature that older, more educated, employed and higher income participants with fewer health-care access barriers, whose household members knew HIV status were more likely to be consistently engaged in care (Hodgson et al., 2014; Johnson et al., 2015). We found that EiC was not significantly correlated with fears of stigma from friends, family, co-workers or health-care providers. However, there was a significant positive correlation between anticipated community stigma and EiC which initially seemed counterintuitive. The finding that participants with higher community stigma fears were more likely to be engaged in care is in contrast to existing literature where stigma has been found to be one of the most common barriers to both access to and retention in care (Ashaba et al., 2017; Chadambuka et al., 2017; Hodgson et al., 2014; Howarth et al., 2017; Johnson et al., 2015; Katirayi et al., 2016; Koirala et al., 2017; Kyaw et al., 2017; Paudel & Baral, 2015; Steward et al., 2008). There could have been a few reasons for this divergence. With more than half of our sample being widowed, it is likely that these women may have been infected by their spouses. The majority (89%) of the participants reported disclosing their status to household members and almost two thirds (61%) acknowledged receiving support to procure ART. Most of them are financially dependent and live with their extended families who form their support group. It is also likely that these women have concerns about the fate of their children in the event of their death or disability and want to remain healthy for the sake of their children (Kohli, Sane, Ghate, & Paranjape, 2016; Kumar, Mohanraj, Rao, Murray, & Manhart, 2015). By remaining healthy, which requires staying on ART and hence retained in care, they are more likely to delay the onset of symptoms which may allow them to keep their diagnoses private to help them avoid community stigma. Neither medication side effects nor communication issues with clinic physicians were found to be associated with EiC. It is known that a trusting relationship with the physician positively influences engagement (O’Brien et al., 2018), while relatively fewer studies show that drug side effects can trigger disengagement from care (Renju et al., 2017).

Limitations

Limitations of this study include being cross sectional, which precludes us from drawing conclusions regarding causality. Longitudinal studies are needed to confirm the associations and their directionality. The fact that participants were all recruited from South India also limits our ability to generalize to other regions. Other limitations are that visits to the ART centre was self-reported and hence may have been overestimated due to memory or social desirability effects, though great care was taken to assure respondent of the confidentiality of their responses. Further, since we recruited from public and non-profit private ART centers, we had limited access to those seeking care in for-profit private medical facilities.

Conclusions

To promote sustained EiC, disclosure of HIV serostatus to supportive household members may be encouraged judiciously, after carefully considering the existing domestic milieu to avoid potential harm. To this end, assistance of health care providers and counsellors may be utilised. Already existing ART centers need to be strengthened to include HIV prevalent rural areas. Effective strategies need to be undertaken to overcome health care access barriers by support networks which include peers, health care workers or family members (Buregyeya et al., 2017; Haas et al., 2018; Kohli et al., 2016; Kumar et al., 2015; Parker et al., 2015; Paudel & Baral, 2015; Vrazo et al., 2018). Given the level of community level stigma, stigma reduction efforts may need to include community HIV education. At a structural level, proven strategies like integration of maternal and child health care services (Erlwanger et al., 2017; Haas et al., 2018; Howarth et al., 2017; Momplaisir et al., 2018; Vrazo et al., 2018) and developing or adapting women-centric programs (Momplaisir et al., 2018) are needed to promote effective implementation of HIV care services for women in India. Also, identifying cultural and region-specific barriers by tracking and interviewing women who never access HIV health care services or are lost to follow-up by field healthcare workers and their redressal is critical. Accredited social health activists (ASHA) are such community health workers instituted by Government of India, to improve rural health. Vocational training and job placement assistance may be provided if feasible for unemployed women for income generation. Family centered approaches and use of ASHA (Nyamathi et al., 2018; Nyamathi et al., 2011) to specifically target women of reproductive age group may mitigate the barriers of travel, childcare, clinic wait times and stigma fears (Besada et al., 2018; Ruzagira, Baisley, Kamali, Biraro, & Grosskurth, 2017) whereas trained peer mentors (Momplaisir et al., 2018) who have been found to improve self-efficacy of HIV infected women, may be helpful in India as well.