Predictors of Non-Adherence to Combined Anti-Retroviral Therapy among Expectant and Breastfeeding Women Receiving Care through Test and Treat Model in Lusaka ()
1. Introduction
Non-adherence to combined Anti-Retroviral Therapy (cART) by any pregnant and breastfeeding woman requires accelerated action from institutional caregivers, communities, and individual clients. To achieve viral suppression and reduce vertical transmission of HIV, effective adherence counselling for lifelong cART should be a prerequisite [1]. Evidence showed that pregnant or breastfeeding women, who consistently used cART did not only have their risk of vertical transmission of HIV reduced but also had improved quality of life [2]. Anti-retroviral drug adherence was a major requirement for successful elimination of HIV vertical transmission world over [3].
ART adherence is a strategy that could work to ensure accelerated reduction in vertical transmission of HIV among pregnant and breastfeeding women [4]. The World Health Organisation proposed a supper track intervention called Universal Test and Treat model which most countries adopted in managing HIV clients. To this effect, various literature proved that Universal Test and Treat Model was a strategy in HIV management that linked several HIV clients to care and treatment on the same day of diagnosis; among these clients were pregnant and breastfeeding women [5], the strategy was formulated to assist HIV client’s gain access to lifelong ARVs in time. For pregnant and breastfeeding women, ARVs were also given to achieve HIV viral suppression by 95 percent and above which in turn reduced the risk of vertical transmission to their infants by up to less than two percent [2].
In most countries, HIV Test and Treat services were carried out in communities through outreach activities on selected days and on a daily basis in all Health facilities. These services included HIV testing, counselling and cART initiation. Women coming for antenatal, postnatal or family planning clinics could also choose to be access these services [3]. As the Model required ART initiation of lifelong treatment on the same day of diagnosis, some women needed to receive effective pre-ART initiation counselling in order for them to be prepared for lifelong cART retention [6].
In line with the Global HIV Strategy, in 2017 through the Ministry of Health, Zambia implemented the Test and Treat model as a strategy to reduce the number of new HIV infections from Mother to Child which was to be achieved by the year 2030 [4]. Following the launch of the Test and Treat model, the Zambian Government had been sustaining the strategy by training Health workers to manage HIV-positive pregnant and breastfeeding women. Furthermore, the Government has ensured no drug shortage by strengthening the medicine supply chain through Medical Stores Limited [7]. With this intervention, the Country had recorded 87 percent increment in the number of infected pregnant and postpartum women initiated into care compared with other previous years [8].
While Test and Treat Model is noticeably appreciated, Zambia had on the other hand seen sustainability challenges in some Hospitals where pregnant and nursing women failed to endure lifelong care. As a result, a number of exposed babies became infected with HIV infection.
At Chawama First-Level Hospital, statistics indicated that 344 pregnant and breastfeeding women enrolled into care from January 2017 to December 2018; and only 111 women were retained in MCH/ART by 2018. Pharmacy and clinical records show that 68 percent of women enrolled in care had not picked up drugs at this hospital consistently according to the way it should have been [9]. This data signified that only 32 percent of women adhered to cART which was below the national standard of 90 percent. The number of infants who were born from these women and tested HIV positive was reportedly high despite their mothers ever been initiated on cART. Statistics showed that from January to December 2018, 775 HIV exposed infants were tested using Early Infant Diagnosis (EID) testing protocols, 43 of them were positive with HIV Deoxyribo-Nucleic Acid (DNA) [8]. The figure represented a six percent infant positivity rate which was higher compared with the acceptable level of two percent or less as a national standard [10]. This positivity rate could have been because of ineffective cART administration by their mothers during pregnancy or while breastfeeding.
1.1. Study Aim
The aim of the study was to establish what causes non-adherence to cART by pregnant and breastfeeding women who started care through the Test and Treat model at Chawama Hospital in Lusaka province of Zambia.
1.2. Conceptual Framework
The conceptual framework used in the study was adapted from Social-Ecological Model ( SEM ) [11] and the Patient Health Engagement Model [12]. This study proposed this integration for a deeper understanding of link between individual characteristics and the Test and Treat counsellling services patients received in MNCH facilities.
Originally, the Social-Ecological Model ( SEM ) had four factors that were formulated in layers which were: Individual, Relationship, Community and Societal Factors. For this study, the researcher had modified these factors to suit the discussion and replaced societal factors with selected Health service attributes. This was because the study sought to understand the complex interplay between an individual, health service-related attributes and other factors. While the original SEM model suggested acting across multiple levels of the model at the same time, this study proposed acting on health services that may influence individual factors and improve pregnant and breastfeeding womens’ adherence to treatment.
As illustrated in Figure 1 below, the adapted conceptual framework had an arrow running from the Social-Ecological factor model integrating with the Patient Health Engagement Model. The blend of these two models had been suggested because as recommended earlier, in order to resolve most individual factors identified in the Social-Ecological models, Midwives in Maternal Neonatal and Child Health/Anti-Retroviral Treatment MNCH/ART units needed to effectively engage individual clients in their health care if effective life long cART adherence by expectant and breastfeeding women was to be realised.
Patient Health Engagement (PHE) Model is a psychosocial concept that creates an understanding of clients’ meaningful engagement in taking care of their ownhealth. Originally, the model had four pillars which included: Dissengangement stage, Arousal stage, Adhesion and Eudaimonic stages [12]. Asillustrated in
![]()
Figure 1. (SekMaiMay, 2021)’s Conceptual framework on women’s non-adherence to cART adapted from SEM [11] and patient health engagement model [12].
Figure 1, SekMaiMay (2021)’s Conceptual framework had these four pillars merged into one and which was later called the Pillar of Patients Health Engagement (PPHE). There were variables such as the patient’s beliefs about cART that an HIV diagnosed woman may have [6]. Using the Test and Treat model pregnant or breastfeeding, woman may feel overwhelmed and shocked upon being diagnosed with HIV and to be initiated on cART on the same day of diagnosis which may even worsen the client’s shock. The conceptual framework suggested the need for a professional psychotherapist or a Midwife who could offer adherence counselling effectively by using all the four pillars of health engagement. For example, effective engagement of clients was important for them to disengage from previous beliefs of non-adherence to lifelong cART which were earlier identified by use of the social-ecological individual factor model [11].
Other than individual factors, the study used SekMaiMay (2021)’s Conceptual framework to identify other factors of non-adherence that include health facility-related activities such as the counselling services that were offered to individual clients in MNCH units. Furthermore, interpersonal factors in relation to the client’s intimate partner’s attitude were other factors that were identified. In the communities where these women came from, factors such as stigma were also known to have an effect on non-adherence to cART by pregnant and breastfeeding women.
2. Methods and Study Design
This was a mixed study with both quantitative and qualitative components used. This publication focused on the quantitative aspect of non-adherence to cART among pregnant and breastfeeding women who were initiated through Test and Treat model at Chawama first Level Hospital in Lusaka, Zambia.
2.1. Study Population and Sampling
The sampling frame for the 92 participants who were involved in the quantitative aspect of the studyusing ART defaulter register. By using a computational systematic randomization sampling technique, selection biases were minimized. Selection criteria was set for participants who had earlier been initiated in cART but during the course of care stayed without drugs as they missed their scheduled appointment for drug pick up without notifying the hospital. Following the random selection, arrangements with the facility were made through phone notification for these defaulting clients to physically come to the facility for drug collection as well as to participate in the study.
2.2. Data Analysis
Quantitative data analysis, STATA Software Version 14.0 SE (STATA Corporation College Station, TX, USA) was used. The outcome variable was categorized into two possible levels of non-adherence namely poor and very poor level. Statistical tests included: Chi-square was used to determine the association of non-adherence on a continuum and client-related characteristics. Fisher’s exact test was used where observations were less than five. Univariate logistic regression was used to get unadjusted estimates. Multiple logistic regression was used to assess the relative contribution of each of the predictor variables. The variables with a large P-value greater than 0.05 were removed until only significant variables were left in the model.
2.3. Validity
Internal validity was maintained by ensuring that all aspects of variables relating to the study participants were included in the data collection instruments which were the Questionnaire. While random selection of participants for the quantitative component of the study also added to internal validity.
On the other hand, external validity in this study was ensured by selecting a sample group that was representative of the target population. Furthermore, the set inclusion and exclusion criteria were followed when selecting participants which assisted in getting the defined population from the sampling frame.
2.4. Reliability
To ensure the reliability of the study, a pilot before the main study was conducted.
Furthermore, the researcher ensured consistency by using the same questions to different respondents at different times. Additionally, an extensive literature search was done before designing the tools which were subsequently checked by Research Supervisors and HIV/AIDS Experts.
3. Results
3.1. Age Group
A total of 92 women were enrolled in the study. As illustrated in Table 1 above, the oldest woman was aged 43 while the youngest was 18, thus, the age range was thus 25. The mean age was 28 with a standard deviation of 6.3. Generally, most expectant, and breastfeeding women were in the age category of 25 to 31 years at 42 percent (39/92); followed by those in the age category of 18 - 24 at 33 percent (29/92). The other age category of 32 to 38 years had a proportion of 15 percent (14/92). Lastly, the least number of women were in the 39 to 45 years age
![]()
Table 1. Description of the sample by age group (n = 92).
category at 11 percent (10/92).
3.2. Level of Non-Adherence
The outcome variable was categorized into two possible levels of non-adherence namely poor and very poor level because the study wanted to detect the difference within the non-adherent group. The study established that 38 (41%) of women had poor adherence while 54 (59%) percent had very poor adherence to cART.
Figure 2 illustrates the level of non-adherence while using the Test and Treat model.
3.3. Client Demographic and Social Characteristics According to the Level of Non-Adherence to cART
As illustrated in Table 2, the study established that the majority of the respondents, 79.35 percent (73/92) were married while the remaining 20.65 percent (19/92) were single women. Concerning education, 57.6 percent (53/92) of the respondents had never been to school or went up to primary school level only and these were said to have been uneducated, while the remaining 42.4 percent (39/92) were educated respondents.
Regarding religious denominations, 40.2 percent (37/92) were Catholics, 34.8 percent (32/92) were protestants, while other denominations accounted for 16.3 percent (15/92) of the total respondents. Muslims accounted for 8.87 percent (8/92) of the respondents.
3.4. Client Related Predictors of Non-Adherence to cART
In the final model below, two client-related variables had significant associations with non-adherence because their P-values were less than 0.05 which was the cut-off point at 95 percent confidence interval. Firstly, client-related predictors of non-adherence to cART included the employment status of being a housewife, whereby, women who were housewives were 84 percent less likely to be non-adherent [AOR 0.16; 95% CI 0.12, 0.36] to cART compared to women who were formally employed. This difference was statistically significant with an associated P-value of 0.04 (Table 3).
![]()
Figure 2. Women’s level of non-adherence.
![]()
Table 2. Client demographic and social characteristics according to level of non-adherence to cART (n = 92).
*Fishers exact.
Secondly, the period that women stayed away from home and usual clinic for several months was a predictor of non-adherence to cART. In this study, it was found that women who were several months away from home and usual clinic were 84.9 percent more likely to be non-adherent to cART [AOR 15.11; 95% CI 13.9, 16. 4] compared to women who had been away from home for several days. The associated P-value was 0.03.
However, this study found that travelling away for several weeks was not a predictor of non-adherence to cART with the associated P-value of 0.22 which was way above the set point of 0.05.
In the regression model, the marital status; travel history; reasons for resuming visits; disclosure of status to partner; community perception and individual client’s feeling were not predictors of non-adherence to cART by women receiving care under the Test and Treat model of HIV care at Chawama First Level Hospital in Lusaka.
3.5. Health Service Attributes and Non-Adherence to cART
Results of health service attributes that predicted non-adherence to cART among pregnant and breastfeeding women under the study are displayed in Table 3. When asked what room was offered to them for counselling women who were attended to from an enclosed unit, individually, were 91 percent less likely to be non-adherent compared to those who were counselled in an open space as a
![]()
Table 3. Univariate and multiple logistic regression (n = 92).
group [AOR 0.09; 95% CI 0.02, 0.53] with associated P-value of 0.01.
3.6. Women’s Reasons for Non-Adherence to cART
The study established a varied number of reasons why women were non-adherent to cART. A summary of these responses by (n = 92) women was cited as reasons as depicted in Figure 3.
3.7. Summary of Quantitative Results
The study facilitated understanding the phenomenon of non-adherence among pregnant and breastfeeding women.
Summary of these findings was illustrated in Figure 4.
4. Discussion and Conclusion
Discussion and conclusion are based on the findings from the quantitative aspects of the study.
4.1. Client Related Predictors of Non-Adherence and Perceptions
The quantitative component of this study revealed that age was not a predictor of non-adherence to cART because of the P-value which was way above 0.05 and was removed from the final adjusted model. This was contrary to [13] conducted a study in America and found that the younger the age at the time of cART initiation the more they were likely to be non-adherent to cART. In the triangulation of results, the qualitative component of the study, health workers noted that age influenced women’s non-adherence to cART.
Concerning woman’s employment status, the study findings revealed that the women who were employed were non-adherent compared to housewives, these study findings were consistent with [14], who concluded that clients who were employed were more likely to be non-adherent to cART compared to those that
![]()
Figure 3. Women’s reasons for non-adherence to cART.
![]()
Figure 4. Summary of quantitative findings.
were fulltime housewives. However, these findings were contrary to the findings of [15] in a cohort study conducted in Lesotho who concluded that women who were full-time housewives were more non-adherent compared to working women. To triangulate these findings, the qualitative aspect of the study noted perceptions by health workers which revealed that working women were seen to be more non-adherent to cART because they devoted themselves to other job commitments. When compared with those that were unemployed health workers further, revealed that some working-class women were denied permissions to attend cART clinic by their employers.
Another individual client level predictor of non-adherence that was significant was being away from the usual home, which was consistent with a study reported by [16] who established that some clients had medication breaks due to travelling away for business obligations which resulted in them missing their scheduled clinic appointments.
4.2. Health Service Attributes and Non-Adherence to cART
Another predictor that was established from the study significant to cART non-adherence among women who were initiated in care through the Test and Treat model at Chawama level Hospital was that of the counselling rooms. Our findings showed that most patients did not favour the approach of receiving adherence counselling in open spaces considering that some people may have the fear of being identified as HIV positive by other clients present during the counselling session. These findings were supported by other studies like [17] [18] who conducted studies in Malawi and Uganda and concluded that lack of individualised counselling rooms was a barrier that stopped some women from continuing cART care. To echo these sentiments, health service providers acknowledged that limited counselling space was a factor that influenced women’s non-adherence to cART as they were counselled in an open space. Furthermore, health providers suggested the need for expansion to create more space for individualised adherence counselling.
4.3. Women’s Reasons for Being Non-Adherent to cART
In this study, a total of 92 women were asked to give reasons for missing their appointments. Their responses indicated barriers to access and utilisation of cART services. Individual factors were noticeable according to the responses that women gave and included the following sub-themes: patient other commitments, being unaware of the consequences of being non-adherent to cART which were like barriers identified by [16] who established that personal factors like being away for other commitments had an impact on one’s adherence to cART.
Some women gave responses that were perceived to be barriers to utilisation of cART services such as were unsupportive husbands. This was similar to findings made by [19] who discovered that intimate partner’s violence discouraged HIV-positive patient’s ability to comply with treatment recommendations.
The poor referral system was the subtheme that was established with some women respondents reportedly having missed their appointments according to clinic records, yet others said that they had received drugs from other facilities. These findings point to the fact that the patient’s electronic referral system had not been strengthened in most facilities offering cART Services.
Some clients had disrupted cART due to adverse pharmacological reactions that they had, when they were started on treatment. These women could not stand side effects of the drugs as they reportedly experienced hallucinations, tiredness, and weakness. This was similar to findings that were made by [20] who reported that some clients stopped therapy if counselling or management of side effects was not done upon cART initiation.
5. Conclusion
The study has provided information that revealed that there was a need to improve adherence counselling if we were to retain women that initiated care through the Test and Treat model.
6. Recommendations
Based on the study, recommendations implications for practice and policy levels were made as follows.
6.1. Implications of the Findings at the Practice Level Chawama Hospital Management and District Health Office
Staff attitude towards HIV-positive clients should be improved because a bad attitude could facilitate non-adherence to cART among pregnant and breastfeeding women initiated into care through the Test and Treat model.
Midwives and clinicians should provide adherence counselling while considering individual client-related factors identified through skillful counselling at every visit.
6.2. Implications of the Findings at Policymakers Level
Training and retraining of health workers in patient-centred Care model could be cardinal in facilitating effective patient health engagement and in turn, health workers will be able to provide confidential space for effective engagement with their clients when providing adherence counselling.
There is a need to expand counselling units in cART considering the growing population and the number of people accessing cART services.
Electronic referral systems should be established or improved to keep track of registered HIV-positive women accessing care to and from other facilities. This improved system could capture new clients who may be in care from other facilities while seeking care at Chawama First Level Hospital. Ultimately improved client tracking system may improve women’s access to cART as they could access care anywhere in the country even without physically carrying patients’ care cards with them.
6.3. Implications of the Findings to the Body of Researchers
Furthermore, the study recommends that the future researchers should address non-adherence targeting specific age groups of pregnant or breastfeeding women, additionally, researchers should consider biomedical parameters such as the viral load as an indicator of non-adherence.
7. Study Limitations
Ability to research within the planned time frame was impacted by the Covid-19 pandemic. However, to mitigate this setback, the researcher worked with the hospital staff in the scheduling of pregnant and breastfeeding women participants.
8. Informed Consent
The study was explained to each participant and those who agreed to take part were asked to fill and sign the consent form. They were also allowed to withdraw at any time if they so wished.
8.1. Ethics Statement
Permission was sought from ERES Converge IRB approval number 2019-Oct-006, National Health Research Authority, Ministry of Health and Chawama First Level Hospital Management. Voluntary participation was granted to all participants and was all accorded with written consent. Since this was a sensitive issue that could affect some respondents emotionally, throughout the study, registers and any other material with client’s information were kept confidential. Data collecting tools did not bear patient’s names but rather were identified with serial numbers.
8.2. Conflicts of Interest Declaration
The authors have declared no conflicts of interest that may inappropriately influence writing or publication of this article.
Acknowledgements
The study was conducted in the partial fulfilment of the requirement for the award of the master of science in midwifery women and Child’s health degree at the University of Zambia, School of Nursing Sciences, Lusaka, Zambia.
Author’s Contribution
MSK conceptualised the research idea, drafted the manuscript, collected and analysed data.
MCM was the principal supervisor who helped in the refining of the research topic, reviewed the final draft for its intellectual content.
SM co-supervised the entire research process, helped in the refining of the research topic, reviewed the final draft for its intellectual content.
All authors reviewed and approved the final version.
Appendix
SEMI-STRUCTURED QUESTIONAIRE FOR PREGNANT AND BREASTFEEDING WOMEN
TOPIC: Predictors of Non-Adherence to combined Anti-Retroviral therapy among expectant and breastfeeding women receiving care through Test and Treat model in Lusaka
DATE: ……………………………………
PLACE: ……………………………………
NAME OF THE RESEARCHER ……………………………………
PLACE OF INTERVIEW: ……………………………………
SERIAL NUMBER: ……………………………………
INSTRUCTIONS TO THE RESEARCH ASSISTANT:
1) Introduce your self.
2) Explain the reason for the interview following the information guide.
3) Allow the participant to sign the informed consent provided.
4) Assure the participant of anonymity and confidentiality.
5) Circle appropriate responses and fill in responses in the space provided.
6) Provide time for the respondents to ask questions at the end of the interview.
7) Thank the participant at the end of the interview.