Male partner involvement and successful completion of the prevention of mother-to-child transmission continuum of care in Kenya

Abstract

Objective:

To examine the relationship between male partner involvement (MPI) in prevention of mother-to-child transmission (PMTCT) activities and successful completion of the PMTCT continuum of care, which remains sub-optimal in settings with high prevalence of HIV.

Methods:

A cross-sectional survey was administered in June–August 2017 to a sample of 200 postpartum Kenyan women with HIV enrolled in a parent trial. Composite PMTCT and MPI variables were created. Descriptive, simple and multivariable regression, and mediation analyses were performed.

Results:

Of the women, 54% reported successful completion of PMTCT. Depression and internalized HIV stigma were independently associated with lower likelihood of successful completion of PMTCT (adjusted risk ratio [aRR] 0.97; 95% confidence interval [CI] 0.94–0.99; aRR 0.92; 95% CI 0.88–0.98, respectively). Each MPI activity was associated with 10% greater likelihood of successful completion of PMTCT (P<0.05). The relationship between MPI and the successful completion of PMTCT was partially mediated through women’s reduced internalized HIV stigma (β −0.03; 95%CI −0.06 – −0.00).

Conclusion:

Greater MPI in PMTCT activities has direct and indirect effects on women’s successful completion of all necessary steps across the PMTCT continuum. Reduced internalized HIV stigma is likely a key mechanism in the relationship.

1. INTRODUCTION

Among pregnant and postpartum women living with HIV (WLWH), there are a series of critical medical interventions aimed at preventing transmission of HIV to infants and promoting maternal and child health, referred to as the prevention of mother-to-child transmission (PMTCT) continuum of care. Key interventions include: women’s engagement in HIV care; uptake and adherence to lifelong antiretroviral therapy (ART); health facility delivery; safe infant feeding; and pediatric testing for HIV [1]. Despite commendable achievements in PMTCT, critical gaps remain [2]. For instance, pooled estimates of the retention of women on ART across sub-Saharan Africa suggest that one in four pregnant women become lost to follow-up within 6 months of beginning ART [3] and only an estimated 76% of women achieve adequate adherence to ART during pregnancy, while even fewer (53%) achieve adequate adherence to ART postpartum [4]. In eastern Africa, almost 40% of new HIV infections among children in 2018 were attributable to lack of maternal retention on ART [5]. Studies also report extremely variable prevalence (<50% to >90%) of deliveries in health facilities in Africa among WLWH [6–9]. Lastly, less than half of HIV-exposed infants in low- and middle-income settings receive an HIV test within the recommended first 2 months of life [10].

There is increasing evidence that in order to maximize the benefits of PMTCT, a renewed emphasis on couples is needed [11]. Prior research indicates, for example, that male partner involvement (MPI) can improve uptake of PMTCT and decrease mother-to-child HIV infection [12, 13]. However, there is no consensus regarding the conceptualization or measurement of MPI in PMTCT [11]. Indeed, various definitions of MPI in PMTCT have been used across studies, including male attendance at antenatal care (ANC) and/or uptake of couple/male HIV testing, status disclosure, and financial support [14–16].

Moving forward, there is a need to expand the existing knowledge base and assess the breadth of activities involving male partners that may promote successful completion of the PMTCT continuum, including the potential use of composite indicators. It is believed that no research has yet to establish if a dose-response relationship exists between various activities involving male partners and successful completion of key steps across the PMTCT continuum of care among WLWH. Moreover, there is little understanding of the mechanisms linking MPI and PMTCT outcomes. The aim of the present study was to examine a wide range of activities involving male partners within a composite indicator of the level of MPI and the direct and indirect effects on WLWH’s successful completion of steps across the PMTCT continuum of care in Kenya. It was hypothesized that there was a significant relationship between activities involving men and successful completion of PMTCT, which was mediated through improvements in maternal mental health, including lower internal HIV stigma and fewer symptoms of depression.

2. MATERIALS AND METHODS

A sub-study was conducted on MPI in PMTCT by administering a cross-sectional survey to a convenience sample of postpartum WLWH enrolled in a cluster randomized parent study (the MOTIVATE! trial, Clinicaltrials.gov NCT02491177). The parent study tested the effects of community mentor mothers and mobile phone text messages to improve PMTCT in three counties of southwestern Kenya with a high prevalence of HIV: Kisumu; Homa Bay; and Migori [17]. The present study reports on the findings from the follow-up survey, which was verbally administered in-person by Kenyan research assistants in the local languages to a sub-sample of 200 postpartum WLWH who were enrolled in the parent study, taking ART, and at least 12 months postpartum (to prevent contamination into the outcomes of the parent study). All participants provided written consent after a verbally administered informed consent process in the local languages. Ethical approval for the present study was provided by the institutional review boards at the University of Colorado Denver, University of Alabama at Birmingham, and Kenya Medical Research Institute (KEMRI).

All WLWH were asked a series of self-reported PMTCT questions, including where the birth took place (facility or home), difficulty with taking ART, safe infant feeding practices (exclusive breastfeeding to 6 months followed by the introduction of complementary foods and continued breastfeeding), and the number of HIV tests taken by the index child by 12 months postpartum. Women who reported a male partner (n=180, 90%) also answered a series of questions regarding their relationship with the male partner and specific forms of MPI in PMTCT. A module was developed on MPI based on existing literature on the topic [13], including questions on whether the male partner attended healthcare visits, encouraged and reminded the woman about specific PMTCT interventions, and provided transportation money to go to the clinic. The survey also collected data on participant sociodemographics and other potential predictors of adherence to PMTCT, including HIV stigma (modified version of the People Living with HIV Self-Stigma Index) and symptoms of depression (Patient Health Questionnaire Depression Scale [PHQ-8]).

In STATA version 16 (StataCorp., College Station, TX, USA), descriptive, simple and multivariable Poisson regression, and structural equation modeling (SEM) with mediation analysis were conducted to establish associations between greater MPI and successful completion of the PMTCT continuum. A dichotomous composite variable was created to capture women who had successfully completed all five steps in the PMTCT continuum compared to those who did not, which was used as the outcome in our Poisson regression models. A composite continuous indicator was also used to capture the number of steps in the PMTCT continuum that women successfully completed by aggregating: (1) facility delivery; (2) no difficulty taking ART; (3) engagement in care (no missed clinic visits in the past year); (4) safe infant feeding (exclusive breastfeeding to 6 months followed by the introduction of complementary foods and continued breastfeeding); and (5) infant HIV testing (infant received at least two HIV tests by the age of 12 months). This continuous variable was used in the SEM analysis.

A composite indicator was also developed for MPI in PMTCT by capturing the aggregated number of activities involving male partners that women reported their male partners engaged in, including: (1) attendance at healthcare visits; (2) encouraging clinic attendance; (3) encouraging delivery at a facility; (4) medication reminders for the mother; (5) reminders to give the child HIV prophylaxis; (6) giving the child prophylaxis; (7) encouraging specific infant feeding; (8) collecting HIV medication from the clinic or dispensary; (9) providing money for transportation; and (10) encouraging HIV testing for the infant.

Generalized linear Poisson regression models were used to estimate unadjusted and adjusted “risk” of successful completion of PMTCT (yes/no). In the multivariable models, any variables as covariates that were associated with the successful completion of PMTCT in the unadjusted models at the P<0.20 level were included. Three separate multivariable Poisson regression models were run to examine the effect of activities involving male partners, symptoms of depression, and internal HIV stigma, on the likelihood of women’s successful completion of the PMTCT continuum due to high collinearity between these three indicators. SEM was then used to test whether less internal HIV stigma and/or fewer symptoms of depression mediated the relationship between activities involving male partners and the number of steps successfully completed in the PMTCT continuum. Bootstrap mediation tests were used to estimate the indirect effect in the sample. Lastly, the following model fit statistics for the final SEM was examined: model chi-square; root mean square error of approximation (RMSEA); comparative fit index (CFI); and Tucker-Lewis index (TLI).

3. RESULTS

Table 1 presents the descriptive results of the sample of postpartum WLWH who participated in the survey (n=200) stratified by successful completion of all steps in the PMTCT continuum: 101 (54%) women reported successfully completing all steps, while 90 (46%) were unsuccessful (i.e. failed to complete at least one step in the PMTCT continuum). The average age of participants was 29 years with a mean of four pregnancies. Just over half (57%) reported completing a primary education. Of the women, 18% were newly diagnosed with HIV, meaning they found out they were living with HIV during the most recent pregnancy. Symptoms of depression were relatively common with a mean score of 7.6 out of a possible score of 20. Internalized HIV stigma was also common, with women having a mean score of 9.1 out of a possible score of 21. The women in the sample were of a relatively low socioeconomic status, as indicated by only 20% of women reporting electricity in the home. The majority of women (90%) reported having a male partner, with 96% of them disclosing their HIV-positive status to the male partner. Of the women in a relationship, 14% reported their male partner’s HIV status was unknown, while 59% reported a seroconcordant relationship where the male partner is also living with HIV.

Table 1.

Sample characteristics stratified by successful completion of PMTCT.^a

Variable	Total	Successful completion of PMTCT	Unsuccessful completion of PMTCT	RR (95% CI)
Total	200 (100)	106 (54.1)	90 (45.9)
Demographic characteristics
Mother’s age (years)	28.5±5.2	28.4±4.7	28.6±5.6	1.00 (0.97–1.02)
Completed primary school (yes)	114 (57.0)	63 (59.4)	48 (53.3)	1.12 (0.86–1.46)
No. of times pregnant	3.8±1.7	3.7±1.6	3.9±1.8	0.95 (0.88–1.03)
Age of index child (years)	12.6±1.6	12.7±1.7	12.5±1.5	1.03 (0.96–1.11)
Newly diagnosed HIV positive (yes)b	36 (18.1)	45 (42.5)	40 (44.4)	0.96 (0.74–1.25)
Internalized HIV stigma scale c	9.1±3.8	8.2±3.3	10.1±3.9	0.93** (0.89–0.98)
Depression symptoms (PHQ-8) d	5.8±6.0	4.6±5.7	7.1±6.1	0.97** (0.94 – 0.99)
Household economics
Electricity in the home (yes)	40 (20.1)	22 (20.8)	16 (18.0)	1.08 (0.79–1.47)
Home has a cemented/tiled floor (yes)	53 (26.6)	33 (31.1)	18 (20.2)	1.28 (0.98–1.65)
House has more than one room (yes)	150 (75.4)	86 (81.9)	61 (67.8)	1.48* (1.01–2.15)
Food insecurity in the past month	86 (43.9)	41 (38.7)	45 (50.0)	0.81 (0.61–1.06)
Parent study intervention exposure (yes)	156 (79.6)	82 (77.4)	74 (82.2)	0.88 (0.65–1.18)
Relationship characteristics
In a relationship with a male partner (yes)	177 (92.7)	95 (92.2)	82 (93.2)	0.94 (0.58–1.51)
Disclosed HIV status to male partner (yes)d	171 (95.5)	92 (96.8)	76 (95.0)	1.28 (0.54–3.05)
Male partner HIV status e
Positive (yes)	105 (59.0)	59 (55.7)	46 (51.1)	ref.
Negative (yes)	48 (27.0)	25 (23.6)	23 (25.6)	0.93 (0.67–1.28)
Unknown (yes)	25 (14.0)	22 (20.8)	21 (23.3)	0.91 (0.65–1.28)
Male partner involvement in PMTCT e,f
Total number of activities involving male partners reported	7.6±2.5	8.2±2.0	7.0±2.9	1.11** (1.03–1.19)

Abbreviations: CI, confidence interval; PMTCT, prevention of mother-to-child transmission; RR, risk ratio.

^aValues are given as number (percentage) or mean ± standard deviation unless otherwise indicated.

^bReference group: knew status before pregnancy.

^cScore in the range of 6–21.

^dScore in the range of 0–20

^eDenominator: n=180 (women reporting a male partner).

^fScore in the range of 0–10

^*P<0.05.

^**P<0.01.

In simple regression models, the following were significantly associated with the likelihood of successful completion of PMTCT at the P<0.05 level (see Table 1): lower depression score (risk ratio [RR] 0.97; 95% confidence interval [CI] 0.95–0.998); less internalized stigma (RR 0.93; 95% CI 0.89–0.98); having a house with more than one room (RR 1.48; 95% CI 1.01–2.15); and greater MPI in PMTCT activities (RR 1.11; 95% CI 1.03–1.19).

Table 2 presents the multivariable Poisson regression results. Greater involvement from male partners in PMTCT activities was associated with a greater likelihood of successful completion of PMTCT: for each additional activity involving male partners reported, women had an increased likelihood of 10% of successfully completing all steps in the PMTCT continuum (adjusted RR [aRR] 1.10; 95% CI 1.02–1.18). Moreover, there appears to be a clear dose-response relationship between the number of activities involving male partners and successful completion of PMTCT (Fig. 1). Conversely, symptoms of depression and internal HIV stigma among women were associated with reduced likelihood of successfully completing all steps in the PMTCT continuum in the multivariable models (aRR 0.97; 95% CI 0.91–0.997 and aRR 0.92; 95% CI 0.88–0.98, respectively).

Table 2.

Multivariable results for the adjusted risk of successful completion of PMTCT by key predictors.^a,b

Independent variables	Successful completion of PMTCT
No. of activities involving male partners	1.10* (1.02–1.18)
Severity of women’s symptoms of depression	0.97* (0.91–0.997)
Severity of women’s internal HIV stigma	0.92** (0.88–0.98)

Abbreviations: aRR, adjusted risk ratio; CI, confidence interval; PMTCT, prevention of mother-to-child transmission.

^aValues are given as aRR (95% CI).

^bThree separate models adjusting for household wealth (house having more than one room, house having a cemented floor) and food insecurity (variables associated with successful completion of PMTCT at the P<0.20 level in simple models).

^*P<0.05.

^**P<0.01.

Figure 1.

Dose-response relationship between the number of male partner involvement activities and successful completion of PMTCT

Figure 2 displays the hypothesized relationship between MPI activities, internalized HIV stigma, symptoms of depression, and the number of PMTCT steps successfully completed. It was hypothesized that both symptoms of depression and internalized stigma would mediate the relationship between greater MPI and PMTCT steps successfully completed.

Figure 2.

hypothesized model of the relationship between male involvement, mediating effects of internalized HIV stigma and depression, and successful completion of PMTCT

As hypothesized, SEM analysis indicated that lower internalized stigma partially mediated the relationship between greater MPI and the number of PMTCT steps successfully completed (Fig. 3). There was both a significant direct effect (β 0.04; 95%CI 0.00 – 0.08) and indirect effect (β −0.03; 95%CI −0.06 – −0.00) of greater MPI on successful completion of PMTCT, which was partially mediated through lower internalized HIV stigma (β –0.04; 95% CI –0.07 to –0.01). Fewer symptoms of depression, on the other hand, did not mediate the relationship between MPI and PMTCT steps successfully completed in SEM analysis (β –0.01; 95% CI –0.02 to 0.011, not shown in Fig. 3). Fit indices indicated a good fit of our final SEM model with the data (model Chi-square = 0; RMSEA < 0.05; CFI=1; TLI=1).

Figure 3.

Mediation model of the relationship between male involvement, internalized HIV stigma, and successful completion of PMTCT

4. DISCUSSION

It is believed that this is the first study to establish a dose-response relationship between greater MPI in PMTCT activities and women’s successful completion of steps across the PMTCT continuum of care. Evidence was also found that lower internalized HIV stigma among WLWH partially mediates the relationship between MPI and successful completion of steps across the PMTCT continuum. These findings significantly expand the knowledge base of how MPI may affect the uptake and adherence to PMTCT among WLWH in a low-resource setting of sub-Saharan Africa with a high HIV prevalence.

The findings of the present study are in line with prior studies that report a significant association between MPI and PMTCT outcomes. A series of systematic reviews examining studies from sub-Saharan Africa by Takah et al. [13, 18–20] indicated that MPI in PMTCT—typically conceptualized as a single isolated behavior such as male attendance at ANC—is associated with women’s uptake of ART during pregnancy [13], facility delivery [18], safe infant feeding [19], and uptake of infant HIV prophylaxis [20]. In a prospective cohort study, Aluisio et al. [21] further found that MPI in PMTCT in the form of ANC attendance in Kenya resulted in increased HIV-free survival of infants throughout the first year of life.

Among prior studies, the conceptualization of MPI varies drastically with most focusing on male ANC attendance as a proxy for “male involvement” [13]. The present study emphasizes the benefits of examining MPI though a multi-faced lens, including male behaviors that occur both inside the home (e.g. reminders and encouragement) and outside of the home (e.g. attendance at health care), instead of a single isolated behavior. Indeed, the present analysis of MPI using a continuous composite measure provided evidence that the more involved male partners are in PMTCT activities, the more steps across the PMTCT continuum of care a WLWH is likely to complete. Two other published studies have also used ad-hoc indices to assess the level of MPI [22, 23]. Both these studies, however, used the MPI index as the outcome of interest, whereas the present study is believed to be the first to use a composite measure of MPI as the primary predictor of successful completion of steps across the PMTCT continuum.

The present hypothesis regarding mental health mediators in the relationship between MPI and positive PMTCT outcomes was partially supported. It was predicted that both lower internalized HIV stigma and fewer symptoms of depression would be mediators. Evidence was found that internalized HIV stigma does indeed partially mediate the relationship between greater MPI and successful completion of more steps in the PMTCT continuum, which it is believed is the first study to report. Conversely, maternal depression did not emerge as a mediator in the relationship between MPI and successful completion of PMTCT steps. However, similar to other studies [24–26], we did find that greater symptoms of depression were independently associated with unsuccessful completion of the PMTCT continuum.

Although not well studied within the context of PMTCT, there is some prior literature linking internalized HIV stigma with sub-optimal PMTCT outcomes, including decreased likelihood of maternal testing for HIV [27], missed clinic visits [28], difficulty taking ART drugs [28], not completing the full course of maternal ART [27], and lack of HIV testing for infants [27]. The present findings build on these studies and indicate that not only is internalized HIV stigma an important independent predictor of successful PMTCT but that it also serves as an underlying mechanism in the relationship between MPI and positive PMTCT outcomes. It appears that greater MPI may reduce some of the shame and guilt experienced by many WLWH, which enhances their ability to more optimally engage in PMTCT.

The present findings should be interpreted in light of several limitations. MPI in the present study was assessed from the perspective of the female partner, not from direct reports from the men themselves or objective measures. Surveys were conducted 12 months postpartum and used self-reports, and thus, responses might be subject to recall and/or social desirability bias. It is plausible that survey respondents may have over-reported both PMTCT behaviors and MPI activities due to social desirability, which could potentially result in underestimated associations. Further, all participants were engaged in HIV care and a PMTCT behavioral intervention, which may result in selection bias by overrepresenting women with high uptake of PMTCT. Due to the nature of cross-sectional data, casual relationships cannot be determined, but significant associations were observed. Participants for this sub-study were recruited from within the parent MOTIVATE study, which aimed to increase PMTCT outcomes. Depending on the study arm, the survey participants may have received text messages and/or community-based mentor mother intervention. Thus, the sample is likely not representative of all peripartum women, and the estimations of PMTCT health behaviors, and potentially MPI, are likely higher than the general population of WLWH. Yet, it was possible to uncover several important associations between greater MPI and successful completion of steps in the PMTCT continuum, as well as potential underlying mechanisms, within this unique sample. Further exploration in larger, more generalizable populations of peripartum WLWH, ideally using longitudinal study designs, is warranted.

Based on these findings, it is recommended that the conceptualization of MPI be expanded to include participation in numerous activities across the PMTCT continuum, including behaviors that occur both inside and outside of the home. Messaging to male partners should focus on them being as engaged as possible instead of participating in one single isolated event, such as ANC attendance. Presenting this message to male partners may help them feel more empowered to be involved since there are numerous barriers to the standard MPI activity of attendance to the ANC clinic, including prohibitive clinic hours that overlap with men’s work schedules, long wait times, and discouraging social norms or clinic environments [29].