Using a composite adherence tool to assess ART response and risk factors of poor adherence in pregnant and breastfeeding HIV-positive Cameroonian women at 6 and 12 months after initiating option B+

Adherence to antiretroviral therapy (ART) for HIV infection is essential for plasma viral load suppression. This is key to treatment success as it reduces both morbidity and mortality but most importantly improves the quality of life and reduces drug resistance development [1‐3]. Non-adherence is the most important factor that leads to viral resistance [4]. As test and treat is being adopted worldwide including Option B+ procedures for newly diagnosed HIV-positive pregnant and breastfeeding women, additional adherence and retention support are needed as most of the patients initiating ART present with asymptomatic HIV infection [5, 6]. Since the introduction of Option B+ in Malawi in 2011, ART uptake has significantly improved for this target group but there are still concerns that women with asymptomatic HIV infection may not be adequately retained or poorly adhere to treatment [7‐9]. Recent studies since the introduction of Option B+ have focused on retention in care [8, 10, 11] and very little data is available assessing patients’ adherence to ART. Evidence also shows that retention, which is often used as a surrogate for adherence, may fail to account for the pattern of clinic attendance and drug taking behaviour [12]. The few studies evaluating adherence to Option B+ used a single adherence measure such as self-report or pharmacy refill which both have been shown to overestimate adherence [13, 14]. Overestimated adherence could result in patient misclassification and lead to inaccurate targeting of adherence-improving interventions or delays in addressing adherence problems. Self-reported adherence is subjective and its reliability drops overtime as patients get acquainted with the assessment tool used [15]. Pharmacy refills and pill counts on their part, though objective and not easily affected by recall and social desirability biases, may result in ‘pill-dumping’, thus overestimating adherence [16, 17]. Viral load (VL) suppression data can provide additional information to improve adherence assessment conducted using any subjective measure, though it could also be affected by pre-treatment HIV drug resistance [18]. To optimise the Option B+ programme and reduce the risk of treatment failure, timely and adequate assessment of adherence in HIV-positive pregnant and breastfeeding women with a simple but robust tool is critical in resource limited settings where second line medication options are limited and third line regimens are virtually non-existent or are still very expensive [2, 5]. Sub-optimal adherence needs to be identified and addressed early prior to treatment failure and the development of viral drug resistance. This is possible only if we understand the risk factors of poor adherence and those enablers of proper medication taking behaviour. These factors have been studied in other settings but are poorly understood in Cameroon [19, 20]. In order to meet the UNAIDS 90–90-90 objective by 2020 [21], treatment programmes should not only focus on HIV testing, ART initiation and retention in care, but also on adherence to lifelong ART which is key to reaching the 3rd 90. However, notwithstanding the importance of near perfect adherence on viral suppression, recent research findings are showing that newer ART regimens may only require moderate adherence levels to achieve viral suppression [22‐24].

In our previous analysis of this cohort of pregnant or breastfeeding women, we observed good uptake of HIV testing and counselling, ART and retention in care, which however, declined over time. We analysed treatment discontinuation taking into account women lost to follow up, transferred out to other ART clinics, intentionally stopped medication or died. Discontinuation from Option B+ was highest at small sites with a high staff turnover [8]. Based on those findings, we now specifically evaluated information on women who remained in care with respect to ART adherence in association with viral suppression using a multiple method tool. We further investigated risk factors associated with poor adherence alongside enablers of proper drug taking behaviour.

Using a composite adherence score (CAS) we determined adherence in association with viral suppression and risk factors of poor adherence in a cohort of HIV-positive pregnant and breastfeeding women who remained in care at 6 and 12 months after initiating ART as part of Option B+. The study demonstrated that of those women who remained in care after Option B+ initiation, 59.0 and 31.4% were either highly or moderately adhering to their treatment as per our CAS tool at 6 month, and 62.7 and 24.9% at month 12, respectively. The predictive accuracy of the CAS tool was reflected by virological suppression rates, which were 100% in women who scored high, 89.5% who scored moderate, and 52.9% who scored low with the CAS tool at month 12. Low CAS adherence as compared to moderate adherence was significantly associated with virological failure (OR 7.6, p = 0.005). In our cohort, moderate to high adherence scores by the CAS tool were sufficient to reach more than 90% of viral suppression (< 1000 copies/mL) as defined by the UNAIDS targets, thus 90.8% of women at 6 months and 87.6% at 12 months were considered with good adherence to reach durable treatment response. Our adherence levels were comparable with findings of earlier reports of good adherence with Option B+ from Eastern Africa [13, 32], but slightly higher than adherence reported for the same target group from an Option A study [33]. Adequate adherence assessment is important to better understand and optimise the success of the Option B+ programme.

A key strength of this study was the use of a CAS to assess adherence in this population. To our knowledge this is the first study to measure adherence to ART in the context of Option B+ using a CAS. We found a high level of correlation between all single adherence measures (pharmacy refill, self-reported adherence, and VAS) and the CAS which were also all associated with the 12 months viral suppression indicating the validity of each of the measures. It has been argued that the use of a CAS to measure adherence may be cumbersome in routine clinical settings [34]. Our observed close correspondence between the VAS and the CAS suggests that a single and less cumbersome tool like the VAS could be more conveniently used in clinical settings, and this observation was also reported in other older studies [30, 31, 35]. Furthermore, in a recent review of adherence and retention beyond Option B+ the authors’ claimed that the paucity of literature on adherence in this population was due to the lack of adherence measure systematically used in routine care [36]. This thus justifies the importance of a simple tool like the VAS for routine adherence assessment at the point of care. Among clients with virologic failure who reported a high adherence, 50% came from self-reported adherence indicating its high vulnerability to social desirability and recall biases [29]. Despite its mediocre performance as an adherence assessment tool in predicting virologic failure, we would still argue that self-reported questionnaires remains useful in providing additional information necessary to provide adequate adherence support to clients with adherence challenges.

Viral suppression rates in women who remained in care at 12 months in our study (92.7%) outreached the UNAIDS 2020 target [21], suggesting that viral suppression is achievable both in pregnancy and/or breastfeeding. However, 8.5% of our women had low replicating viraemia between 40 and 999 copies/mL, and given the low genetic resistance barrier of non-nucleotide reverse transcriptase inhibitor containing regimens, close monitoring is required to promptly identify treatment failure and switch clients to a second line therapy.

Risk factors for low adherence to ART were younger age, low level of education and employment in the informal sector. Younger age and low level of education had been reported as predictors of lost to follow-up and treatment discontinuation [8, 20]. Older women with more experience and better self-care skills/abilities may be more responsible towards their health compared to younger women [20]. Employees in the informal sector have their own unique work challenges and stresses, which can adversely affect their ability to concentrate on health and medication adherence. A South African study showed that conflict with work commitment and the difficulty of disclosing ones HIV status to an employer affected ARV adherence in women on Option B+ [37]. Potential strategies to mitigate these challenges include flexible clinics opening hours like evenings or weekends when most workers are free and do not need employer’s permission or a day off just to attend clinic.

Like other studies from sub-Saharan Africa, most participants in this study cited forgetfulness, being away from home and lack of transport to the health facility as possible reasons for missing doses [32, 38, 39]. Our study participants also indicated that reminder aids like cell phones, alarm clocks and reminder from family members helped improved their adherence. A good proportion of the respondents (36.6%) claimed that medication taking had become a routine for them so they needed no reminders. Of these, 88.2% had viral suppression showing that they had in fact adapted this routinely. This claim for those not virally suppressed may just covey their good intentions to continue taking their medication or methodological problems such as social desirability concerns. Recent studies have shown that the use of cell phones short message reminders can greatly improve adherence [40‐42]. The role of family members like husbands and sisters to remind clients take their medication could improve adherence [13, 33].

During the first year of the Option B+ implementation in Cameroon we used a novel adherence tool, which effectively provided an adherence score that correlated well with viral suppression. However, younger, less educated and informal sector employed pregnant and breastfeeding women may need more attention for optimal adherence to help reduce the risk of virological failure.