Partner-based HIV treatment for seroconcordant couples attending antenatal and postnatal care in rural Mozambique: A cluster randomized controlled trial

Abstract

Introduction:

There is evidence that a supportive male partner facilitates maternal HIV testing during pregnancy, increases maternal ART initiation and adherence, and increases HIV-free infant survival. Most male partner engagement clinical strategies have focused on increasing uptake of couple-based HIV testing and counseling. We delivered a couple-based care and treatment intervention to improve of ART adherence in expectant couples living with HIV.

Methods:

We implemented a cluster randomized controlled trial for seroconcordant couples living with HIV, comparing retention (using a patient’s medication possession ratio) in HIV care for a couple-based care and treatment intervention versus standard of care services in rural Mozambique. The intervention included couple-based treatment, couple-based education and skills building, and couple-peer educator support.

Results:

We recruited 1080 couples to participate in the study. Using a linear mixed effect model with a random effect for clinic, the intervention had no impact on the medication possession ratio among women at 12 months. However, the intervention increased men’s medication ratio by 8.77%. Our unadjusted logistic regression model found the odds of an infant seroconverting in the intervention group was 30% less than in the control group, but the results were not statistically significant.

Discussion:

Our study found no difference in maternal outcomes by study arm, but our intervention resulted in an improved medication possession ratio among male partners. We provide a community/clinic-based treatment framework that can improve outcomes among male partners. Further work needs to be done to improve social support for pregnant women and to facilitate prevention of vertical transmission to infants among couples living with HIV.

Background

In countries in Africa with a high HIV prevalence, everyone with HIV, including pregnant women, receives free life-long combination antiretroviral therapy (ART).^1–4 However, retention in care throughout pregnancy and the postpartum period remains sub-optimal.^5–10 Poor medication adherence, risk of vertical HIV transmission and ART resistance hampers the global effort to end the HIV pandemic.^11,12 Factors influencing poor adherence to ART among mothers and their infants in the region include a lack of male partner/extended family encouragement, support, and assistance (e.g., watching children while the mother goes to pick up medications, transportation fare), poor knowledge about the importance of ART, challenges with collecting and taking stigmatized medications at home, and a distrust of health care services/providers.^5,8,13–35

There is strong evidence that a supportive, involved male partner facilitates maternal HIV testing during pregnancy, increases maternal ART initiation and adherence, and increases HIV-free infant survival.^36–40 Uptake of antenatal care (ANC) attendance and couple-based HIV testing and counseling can lead to improved maternal and infant outcomes among pregnant women. Couple-based services facilitate disclosure of a person’s HIV status, allows the counselor time to address issues of concern for the couple (e.g., trust), and creates a time where both people can be educated about the necessary treatment required for the person who is found to have HIV.^41–48

Male partner strategies and interventions have been implemented in more than 14 African countries to increase maternal/couple-based HIV testing, HIV treatment initiation and adherence, and general pregnancy support.^43,49 The primary focus of these has been to facilitate couple-based HIV testing and counseling during ANC. Facility-based interventions have included the couple-friendly services at antenatal clinics such as counselor-supported disclosure assistance⁵⁰ and training peer educators and providers to offer and deliver couple-based testing.⁵¹ Community-based strategies have included community health worker-led counseling and testing and male counseling by a male champion.^49,52

In Mozambique, ANC services at public health care facilities have historically excluded male partners despite the importance of their approval in facilitating clinical care. This changed in the past decade, when couple-based testing became standard of care at most facilities through a community-based advocacy program Homens para Saude (HoPS, Men for Health).⁴¹ Despite this positive change, if a couple were found to be HIV-positive, the man would be referred to the adult HIV care clinic for treatment while the woman would continue ANC until delivery and subsequently receive care with her infant in the child-at-risk clinic. We built on the success of the community-based HoPS program and implemented a couple-based care and treatment intervention within the pre- and post-natal care clinics (HoPS+), comparing maternal and paternal retention in HIV care with standard of care services through a cluster randomized controlled trial at 24 health facilities in rural Mozambique.⁵³

Materials and Methods

Participants, Intervention, and Outcomes

Study Setting

Twenty-four health care facilities located in rural Zambézia Province, Mozambique, were selected to participate in this clinic-based unblinded, two-arm, cluster randomized control trial that compares usual care to couple-based HIV care and treatment among expectant and post-natal seroconcordant couples. The clinics represented both the northern and southern regions of the province, provided ANC for at least 50 women per month, and male partner accompaniment to ANC was documented for at least 30% of pregnant women. These facilities are owned and operated by the Ministry of Health (MOH) in Mozambique but receive technical and logistical support from Friends in Global Health (FGH), a non-governmental organization and wholly-owned affiliate of Vanderbilt University Medical Center (VUMC). Study information is described on ClinicalTrials.gov (Study ID#: NCT03149237), and the study protocol was published.⁵³

Eligibility Criteria

Couples (one pregnant woman and her male partner) were eligible to participate if: both partners were HIV-positive, due date was >2 weeks from the time of study enrollment, and both persons were 18 years or older, able to give informed consent, willing to consent to an infant medical record search, and willing to enroll in ART together. Participants did not necessarily need to be newly diagnosed with HIV, but they had to either: test and receive positive results together at ANC visit, or clinical staff had to be able to confirm their HIV-positive status from clinical records. Participants did not necessarily need to be treatment naïve, but they could not be active in clinical care (i.e., they had not had any HIV-related clinical services for at least six months) at the time of study recruitment. At each study visit the couple was screened for intimate partner violence. None was reported but if any was identified they would have been removed from the study and referred for support services through the health system.

Intervention and Control Conditions

The 12 clinics randomized to control continued providing standard of care ANC HIV services which included: (1) invitation of male partner to ANC, (2) opt-out rapid HIV testing of all pregnant women and/or couples testing, (3) HIV-specific counseling and support for all pregnant women who test positive (including support from a mentor mother), (4) provision of cotrimoxazole prophylaxis, and universal ART, as per World Health Organization Option B+ guidelines.² Male partners living with HIV received separate clinical and counseling services as well as relevant medications free of charge in the adult HIV/ART clinics. Postpartum women and their exposed infants continued prevention of maternal to child HIV transmission (PMTCT) follow-up after delivery at the child at risk clinic in a “one-stop point of care” model. In the child at risk clinic, infants had access to HIV testing by dried blood spot polymerase chain reaction test as early as 4 weeks after birth. Follow-up at the child-at-risk clinic for both mother and infant continued for up to 18 months, unless the infant was diagnosed HIV-positive and the dyad is referred to respective adult and pediatric HIV treatment services.

Instead of separate care for the female and male partners, the 12 clinics randomized to HoPS+ intervention received a combination of community and clinical vertical prevention services, including: (1) couple-based enrollment and treatment in ANC for seroconcordant HIV positive expectant couples, (2) couple-centered treatment (along with the child/children) in the postpartum and breastfeeding periods at the clinic for children at risk, (3) six couple-based education and skills building during the ANC and lactating/breastfeeding periods, and (4) nine treatment continuity support sessions led by expert patient (peer) educators selected among couples who have successfully navigated PMTCT. HoPS+ Couples Counselors employed cognitive-behavioral strategies to promote couple connectedness and strengthen couples’ communication, prevent marital problems, and equip couples with effective and accessible coping strategies. Peer sessions focused primarily on barriers to treatment adherence, with a focus on encouraging effective strategies that worked for the couple. Sessions were delivered at a time and location of the participants choosing, with most opting to receive sessions at home. Details of the interventions can be found in our study protocol.⁵³

Outcome Measurement

Our primary outcome was patient retention in care (at 12 months post-initiation), defined as proportion of time on medication. We refer to this as the medication possession ratio. Specifically, every patient was prescribed ART medication pick-ups every 30, 60, or 90 days, depending on previous treatment adherence (i.e., patients in care for 6 months that are virally suppressed would have a longer period between pickups than someone initiating care or someone who missed medication pickups in the past). If a patient picked up medication after the indicated date, they were considered “non-adherent” for each day after their personal scheduled pick-up date; this was assessed over the course of the one-year follow-up period. We planned to assess the impact of the intervention on viral suppression but high levels of missingness made this impossible.

Data Collection, Management, and Analysis

Data Collection

The outcome measure was derived from the MOH’s HIV clinical database, securely stored in the OpenMRS^™ electronic patient tracking system. Study specific data were collected by study staff, including survey and demographic data. Survey data included HIV knowledge, HIV-related stigma, depression screening.^54–56

Data Management

All non-clinical participant data were collected on paper forms and entered into a REDCap database.⁵⁷ Study staff conducted data quality checks on all forms to ensure data were recorded accurately. Clinical data were entered by FGH data managers into an OpenMRS^™ database owned by the MOH in Mozambique.

Statistical Analysis

To assess the medication possession ratio, we fit a linear mixed effects model with random effect for the clinics. Specifically, the expectation of the proportion, E(Y_ij), for subject j in clinic i was be modeled as

$$E\left(Y_{ij}\right)={\beta }_0+{\beta }_1{\text{Intervention}}_{\text{i}}+{\beta }_{\text{X}}{X}_j+b_{\text{i}}$$

where b_i is a random effect for clinic i which accounts for clustering of patients within clinics, due perhaps to similarities in health-seeking behavior and/or access to care. Intervention_i is an indicator of intervention, which is th⁵⁸e same for all patients in cluster i, and β₁ is the intervention effect (i.e., difference in mean medication possession ratio). We tested the null hypothesis, H₀: β₁ = 0. We provided estimates of β₁, 95% Wald confidence intervals, and two-sided p-values. The mixed effects model was fit using restricted maximum likelihood estimation. It included baseline covariates, X_j, to allow for adjustment of important subject-level variables including age, education, marital status, ART regimen (dolutegravir based vs. other), employment status, depression score, HIV knowledge, and HIV stigma. Missing data were multiply imputed with 20 imputations and estimates were pooled together using Rubin’s rules.⁵⁸ Separate analyses were performed for women and men. Additional analyses that included partner characteristics in the model were conducted to determine if partner factors impacted the effect of the intervention.

We conducted sensitivity analyses to assess the impact of the COVID-19 pandemic (specifically, the prevention and mitigation measures implemented by the MOH as community-based study activities were paused for approximately six months) on retention. We fit the same primary linear mixed effects model as described above, but with a COVID-19 indicator to differentiate participants who experienced any COVID-19 mitigation efforts during their data collection vs. none at all.

We performed a series of secondary analyses to investigate the robustness of our findings. Secondary analyses included fitting semiparametric cumulative probability models to address non-normality of the primary outcome,⁵⁸ and fitting generalized linear mixed effects models to secondary definitions of retention (retention coded as yes/no at each month based on participant medication pick-up and retention coded as yes/no at 6 and 12 months based on MOH guidelines). These latter analyses included separate random effects for health clinics and for participants. Data were multiply imputed and pooled as described in the primary analysis. More details of these secondary analyses are reported in the Supplementary Material.

All comparative analyses were performed in R version 4.2.2.⁵⁹

Sample Size

From a review of FGH-supported HIV program data at study design (2016), 55% of pregnant women living with HIV initiated ART and 52% were retained after 12 months with comparable rates for men. We estimated the intracluster correlation coefficient for retention to be in the range 0.02–0.07 and the correlation due to clinic matching as 0.23–0.71. With our proposed study with 24 clusters and 45 couples per cluster, we anticipated over 80% power to detect a 20% increase in adult ART retention, and 40% reduction in infant HIV positivity. Women, men, and infants were evaluated as separate study populations.

Results

Participant Recruitment/Flow

We recruited 1080 couples to participate in the study from November 20, 2017 to December 5, 2020. Uptake among eligible participants approached and invited to enroll in the study was 97% in the intervention arm and 99% in the control arm. One couple was removed from the study after it was identified that the male partner had previously enrolled with a different female partner at a different study site. All participants were followed for 18 months after enrollment to obtain infant seroconversion data (Figure 1).

Figure 1:

Study Enrollment – Adult Cohort.

Baseline Data

Of the 2160 enrolled adult participants, data were available from 2151. Participants in control and intervention sites were largely similar in terms of age and marital status. However, some differences could be seen in education attainment, where control sites were more likely to have no formal education (women: 23.9% vs. 17.3% and men: 10.1% vs. 5.6%). Differences were also seen in employment status, with women more likely to work in farming in control sites (58.2% vs. 50.9%) (Table 1).

Table 1.

Baseline demographic data (n=2151)

Variable	Control		Intervention		Overall
	Female (N=552)	Male (N=552)	Female (N=525)	Male (N=522)	Female (N=1077)	Male (N=1074)
Age at enrollment (years)
Mean (SD)	24.01 (5.03)	28.80 (6.94)	23.95 (5.2)	28.37 (6.45)	23.98 (5.11)	28.59 (6.7)
Missing	0 (0%)	0 (0%)	10 (1.9%)	0 (0%)	10 (0.9%)	0 (0%)
District
Gile	39 (7.1%)	39 (7.1%)	62 (11.8%)	61 (11.7%)	101 (9.4%)	100 (9.3%)
Inhassunge	160 (29%)	159 (28.8%)	31 (5.9%)	31 (5.9%)	191 (17.7%)	190 (17.7%)
Maganja da Costa	0 (0%)	0 (0%)	114 (21.7%)	113 (21.6%)	114 (10.6%)	113 (10.5%)
Mocubela	116 (21%)	116 (21%)	65 (12.4%)	65 (12.5%)	181 (16.8%)	181 (16.9%)
Namacurra	104 (18.8%)	104 (18.8%)	75 (14.3%)	75 (14.4%)	179 (16.6%)	179 (16.7%)
Pebane	133 (24.1%)	134 (24.3%)	157 (29.9%)	157 (30.1%)	290 (26.9%)	291 (27.1%)
Quelimane	0 (0%)	0 (0%)	21 (4%)	20 (3.8%)	21 (1.9%)	20 (1.9%)
Education level
None	132 (23.9%)	56 (10.1%)	91 (17.3%)	29 (5.6%)	223 (20.7%)	85 (7.9%)
Some primary (1st-7th)	329 (59.6%)	281 (50.9%)	354 (67.4%)	296 (56.7%)	683 (63.4%)	577 (53.7%)
Completed primary (finished 7th)	38 (6.9%)	69 (12.5%)	27 (5.1%)	49 (9.4%)	65 (6%)	118 (11%)
Some secondary (8th-10th)	32 (5.8%)	83 (15%)	34 (6.5%)	86 (16.5%)	66 (6.1%)	169 (15.7%)
Completed secondary (finished 10th)	12 (2.2%)	35 (6.3%)	11 (2.1%)	32 (6.1%)	23 (2.1%)	67 (6.2%)
Basic technical school	5 (0.9%)	17 (3.1%)	4 (0.8%)	20 (3.8%)	9 (0.8%)	37 (3.4%)
Advanced technical school	4 (0.7%)	9 (1.6%)	4 (0.8%)	9 (1.7%)	8 (0.7%)	18 (1.7%)
University	0 (0%)	2 (0.4%)	0 (0%)	0 (0%)	0 (0%)	2 (0.2%)
Other	0 (0%)	0 (0%)	0 (0%)	1 (0.2%)	0 (0%)	1 (0.1%)
Marital status
Married	102 (18.5%)	105 (19%)	96 (18.3%)	95 (18.2%)	198 (18.4%)	200 (18.6%)
Single	169 (30.6%)	166 (30.1%)	160 (30.5%)	159 (30.5%)	329 (30.5%)	325 (30.3%)
Common law/live together	281 (50.9%)	280 (50.7%)	268 (51%)	267 (51.1%)	549 (51%)	547 (50.9%)
Missing	0 (0%)	1 (0.2%)	1 (0.2%)	1 (0.2%)	1 (0.1%)	2 (0.2%)
Employment status
Agriculture	321 (58.2%)	148 (26.8%)	267 (50.9%)	168 (32.2%)	588 (54.6%)	316 (29.4%)
Domestic worker	137 (24.8%)	56 (10.1%)	189 (36%)	35 (6.7%)	326 (30.3%)	91 (8.5%)
Sales/Trader	3 (0.5%)	84 (15.2%)	3 (0.6%)	57 (10.9%)	6 (0.6%)	141 (13.1%)
Fisherman	0 (0%)	44 (8%)	0 (0%)	86 (16.5%)	0 (0%)	130 (12.1%)
Other	5 (0.9%)	74 (13.4%)	6 (1.1%)	49 (9.4%)	11 (1%)	123 (11.5%)
Missing	86 (15.6%)	146 (26.4%)	60 (11.4%)	127 (24.3%)	146 (13.6%)	273 (25.4%)

Primary Analysis among Women

We assessed the impact of the intervention on the proportion of time with medication among women and men at 12 months after study enrollment. Women in the intervention arm had a median 12-month ART medication possession ratio of 74% [interquartile range (IQR): 49, 88] vs 74% [IQR: 53,87] in the control arm. Using a linear mixed effect model with a random effect for clinic, for women, we were unable to find evidence of an impact of the intervention on the medication possession ratio (Table 2). Those with higher levels of education (more than primary school), those living with their partner, and those on dolutegravir (DTG) as first-line ART treatment had higher proportions of time with medication. We re-fit the model to include their male partners’ characteristics as co-variates but were still unable to find evidence of an impact of the intervention on female medication possession ratio (Table 3).

Table 2:

Female Participant Outcomes

Variable	Level	Estimate	(95% CI)	P-value
Study group	Intervention	−0.003	(−0.072, 0.065)	0.93
Age at enrollment (years)		0.002	(−0.001, 0.005)	0.12
Education	None	−0.069	(−0.132, −0.006)	0.032
	Some-completed primary	−0.058	(−0.110, −0.005)	0.031
Marital status	Single	−0.002	(−0.055, 0.051)	0.93
	Common law/live together	0.056	(0.007, 0.106)	0.025
ART Regimen (enrollment)	DTG	0.075	(0.033, 0.116)	<0.001
Depression score		−0.003	(−0.007, 0.000)	0.067
HIV Knowledge score		0.002	(−0.002, 0.006)	0.45
HIV Stigma score		−0.000	(−0.002, 0.001)	0.67

Reference levels: Study group (control); Education (some secondary or beyond); Marital status (married); ART regimen (other).

ART = Antiretroviral therapy; DTG = Dolutegravir.

Table 3:

Female Participant Outcomes including Male Characteristics in the Model

Variable	Level	Estimate	95% CI	P-value
Study group	Intervention	−0.0065	(−0.0745, 0.0616)	0.8519
Age at enrollment (years)		0.0024	(−0.0014, 0.0062)	0.2101
Education	None	−0.0518	(−0.1179, 0.0144)	0.1253
	Some-completed primary	−0.0373	(−0.0914, 0.0167)	0.1761
Marital status	Single	−0.0012	(−0.1881, 0.1858)	0.9902
	Common law/live together	0.1085	(−0.1536, 0.3707)	0.4171
ART Regimen (enrollment)	DTG	−0.0678	(−0.1445, 0.0089)	0.0832
Depression score		−0.0039	(−0.0088, 0.001)	0.1160
HIV score (knowledge)		0.0017	(−0.0028, 0.0062)	0.4585
HIV score (stigma)		−0.0004	(−0.0021, 0.0014)	0.6907
Age at enrollment (years), male		0.0013	(−0.0015, 0.0041)	0.3495
Education, male	None	−0.0282	(−0.0981, 0.0417)	0.4297
	Some-completed primary	−0.0307	(−0.0682, 0.0067)	0.1079
Marital status, male	Single	0.0026	(−0.1835, 0.1888)	0.9781
	Common law/live together	−0.0572	(−0.3193, 0.2049)	0.6688
Employment, male	Agriculture/Domestic	−0.0413	(−0.0812, −0.0015)	0.0429
ART Regimen (enrollment), male	DTG	0.1572	(0.0822, 0.2322)	0.0000
Depression score, male		0.0018	(−0.0033, 0.0069)	0.4897
HIV score (knowledge), male		0.0002	(−0.0065, 0.0069)	0.9484
HIV score (stigma), male		0.0002	(−0.0016, 0.002)	0.8208

Reference levels: Study group (control); Education (some secondary or beyond); Marital status (married); ART regimen (other); Employment status (other).

ART = Antiretroviral therapy; DTG = Dolutegravir.

Results were similar in secondary analyses that considered other definitions for adherence (see Supplementary Material). In a sensitivity analysis, women who enrolled during the COVID-19 pandemic had, on average, an 8% increase in their medication possession ratio compared to those who did not experience any COVID-19 mitigation measures during the data collection phase (see Supplementary Material).

Primary Analyses among Men

Among men, the intervention showed a positive impact on their medication possession ratio. Men in the intervention arm had a 12-month ART retention (median medication possession ratio) of 58% [IQR: 32, 81] vs. 48% [IQR: 17, 75] in the control. Using a linear mixed effect model with a random effect for clinic and adjusting for baseline covariates, men in the intervention arm had, on average, a medication possession ratio that was 8.8% higher than in the control arm (p=0.015) (Table 4). Men who were older, living with their partner, those on DTG as first-line ART treatment, and those who increased HIV knowledge had higher proportions of time with medication. When we re-fit the model to include their female partner’s characteristics as co-variates, the effort of the intervention remained. Men in the intervention group showed a medication possession ratio that was 9.3% higher than those in the control arm (Table 5).

Table 4:

Male Participant Outcomes

Variable	Level	Estimate	(95% CI)	P-value
Study group	Intervention	0.088	(0.017, 0.158)	0.015
Age at enrollment (years)		0.004	(0.002, 0.007)	0.001
Education	None	−0.038	(−0.110, 0.035)	0.31
	Some-completed primary	−0.014	(−0.054, 0.026)	0.49
Marital status	Single	0.004	(−0.054, 0.062)	0.90
	Common law/live together	0.081	(0.026, 0.135)	0.004
ART Regimen (at enrollment)	DTG	0.187	(0.143, 0.232)	<0.001
Employment status	Agriculture/Domestic	0.011	(−0.030, 0.051)	0.61
Depression score		−0.002	(−0.006, 0.002)	0.42
HIV Knowledge score		0.006	(0.001, 0.010)	0.012
HIV Stigma score		0.001	(−0.001, 0.002)	0.54

Reference levels: Study group (control); Education (some secondary or beyond); Marital status (married); ART regimen (other); Employment status (other).

ART = Antiretroviral therapy; DTG = Dolutegravir.

Table 5:

Male Participant Outcomes including Female Characteristics in the Model

Variable	Level	Estimate	95% CI	P-value
Study group	Intervention	0.0927	(0.022, 0.1635)	0.0102
Age at enrollment (years)		0.0024	(−0.0007, 0.0055)	0.1296
Education	None	−0.0316	(−0.1087, 0.0455)	0.4216
	Some-completed primary	−0.0157	(−0.0571, 0.0257)	0.4577
Marital status	Single	0.1002	(−0.106, 0.3064)	0.3407
	Common law/live together	−0.0873	(−0.3775, 0.2029)	0.5554
ART Regimen (enrollment)	DTG	0.0686	(−0.013, 0.1501)	0.0993
Employment status	Agriculture/Domestic	0.0069	(−0.0326, 0.0464)	0.7332
Depression score		−0.0009	(−0.0066, 0.0048)	0.7623
HIV score (knowledge)		0.0037	(−0.0017, 0.0092)	0.1784
HIV score (stigma)		0.0010	(−0.0009, 0.003)	0.2981
Age at enrollment (years), female		0.0038	(−0.0003, 0.008)	0.0679
Education, female	None	−0.0121	(−0.0851, 0.061)	0.7462
	Some-completed primary	0.0075	(−0.0521, 0.0672)	0.8043
Marital status, female	Single	−0.0905	(−0.2976, 0.1166)	0.3916
	Common law/live together	0.1682	(−0.1223, 0.4587)	0.2563
ART Regimen (enrollment), female	DTG	0.1323	(0.0494, 0.2151)	0.0018
Depression score, female		−0.0010	(−0.0061, 0.0042)	0.7141
HIV score (knowledge), female		0.0035	(−0.0018, 0.0088)	0.1976
HIV score (stigma), female		−0.0008	(−0.0028, 0.0012)	0.4340

Reference levels: Study group (control); Education (some secondary or beyond); Marital status (married); ART regimen (other); Employment status (other).

ART = Antiretroviral therapy; DTG = Dolutegravir.

Results were similar in secondary analyses that considered alternative definitions of retention (see Supplementary Material). In a sensitivity analysis, men who experienced COVID-19 mitigation measures on average had an almost 12% increase in their medication possession ratio compared with those who did not experience any COVID-19 mitigation measures during the data collection phase (see Supplementary Material).

Infant outcomes

We identified 940 live births among our study participants, 87% and 88% in the control and intervention arms, respectively (p=0.84). Of those, 825 infants had a documented confirmatory HIV test result (Figure 2).

Figure 2:

Infant Cohort.

Overall, 8.5% (80/940) of infants in our study received an HIV-positive test result by 18 months of age. In the control arm, 9.8% (47/479) received a positive test vs. 7.2% (33/461) in the intervention arm; HIV status was missing for 12.7% (61/479) and 11.4% (54/461) infants in control and intervention arms, respectively. Limiting analyses to the 825 infants with known HIV status, the odds of infant seroconversion were 30% lower in the intervention arm, but this was not statistically significant (unadjusted odds ratio [OR] = 0.70; 95% confidence interval [CI] 0.44, 1.11; p=0.13). When adjusting for mother’s age, mother’s education level, mother’s depression screening score at baseline, father’s HIV stigma score at baseline, and father’s ART regimen at baseline using a propensity score, the intervention continued to show a non-significant protective effect (adjusted OR 0.71; 95% CI 0.43, 1.16; p=0.17).

Intervention Completion Data

All health facility-based couples counseling and skills-building session activities were tracked for on-time completion (defined as within 45 days of scheduled session date) and attendance of both partners. Overall, female HoPS+ participants in the intervention group were more likely attend their scheduled counseling and skills sessions (median 100% of scheduled sessions vs. 50% among male participants) and peer sessions (89% vs. 56%) (Table 6). While we encouraged both partners to attend all sessions, we continued with conducting sessions if one partner attended and requested to continue.

Table 6.

Intervention Engagement by Sex

	Female Participants (n = 385)	Male Participants (n = 385)
Completed Counseling & Skills Sessions
0	6 (1.6%)	51 (13.2%)
1	9 (2.3%)	64 (16.6%)
2	23 (6.0%)	54 (14.0%)
3	26 (6.8%)	52 (13.5%)
4	33 (8.6%)	48 (12.5%)
5	61 (15.8%)	51 (13.2%)
6	227 (59.0%)	65 (16.9%)
Proportion of Scheduled Attended
Median [Q1, Q3]	1.00 [0.833, 1.00]	0.500 [0.200, 0.833]
Completed Peer Sessions
0	36 (9.4%)	58 (15.1%)
1	15 (3.9%)	42 (10.9%)
2	12 (3.1%)	36 (9.4%)
3	16 (4.2%)	43 (11.2%)
4	15 (3.9%)	24 (6.2%)
5	17 (4.4%)	43 (11.2%)
6	33 (8.6%)	57 (14.8%)
7	45 (11.7%)	28 (7.3%)
8	81 (21.0%)	21 (5.5%)
9	115 (29.9%)	33 (8.6%)
Proportion of Scheduled Attended
Median [Q1, Q3]	0.889 [0.667, 1.00]	0.556 [0.222, 0.667]
Total Attendance of Scheduled Sessions (Proportion)
Median [Q1, Q3]	0.917 [0.786, 1.00]	0.533 [0.333, 0.733]

Discussion

Our study examined the impact couple-based care and treatment for seroconcordant couples living with HIV within pre- and post-natal care services, comparing both maternal and paternal retention in HIV care at 24 health facilities in rural Zambézia Province, Mozambique. We hypothesized that couples who received care and treatment together during the prenatal and postpartum periods would have improved rates of retention vs. those who received care separately. Our study found no difference in maternal outcomes by study arm, but couple-based care and treatment resulted in improved medication possession ratios among male partners at 12 months. Infants born to couples in the intervention arm were less likely to seroconvert at 18 months, but the effect did not reach statistical significance.

Male partners in the intervention arm experienced a moderate improvement in their medication possession ratio, although their outcomes remained overall poorer than women. Men have historically had low levels of testing uptake and treatment retention in high-prevalence HIV countries in Africa,^60–62 leading to higher levels of mortality and morbidity among men.^63,64 Historically, HIV programs have been gendered, with a strong focus on testing and treating pregnant women to prevent vertical transmission.^63,64 Our program provided services jointly to men and women that were traditionally limited to only women (e.g., frequent home visits from peer couples, similar to mentor mothers,^65,66 throughout the pregnancy and postpartum periods) that were described positively by couples.⁶⁷ Despite the positive reception to these services, male partners frequently did not attend counselor or peer-led sessions. In some cases, men reported that they had work obligations during the scheduled session time and requested that sessions be moved to evenings or weekends so they could attend. In other cases, women complained that their partners were simply not interested in joining, and requested that the health system somehow require them to attend.⁶⁷ Clinical appointments were held jointly, allowing participants to pick up their partner’s medication if one or the other could not attend. Qualitative data suggest this was appreciated by both men and women in the study.⁶⁷ Despite this, men struggled to collect their medication regularly.

We originally postulated that by engaging male partners in care, women would experience greater support from their partners for treatment initiation and retention, leading to improved HIV-related outcomes. Unfortunately, we were not able to find evidence of improved treatment adherence among women in the prenatal period in our intervention This may be due, in part, to the low levels of program participation, particularly among men. In addition, previous studies assessing the impact of male partner engagement largely looked at the differences between men who participated in interventions designed to increase uptake of couple-based HIV testing and counseling vs. those who did not.⁴⁹ In our study, all the men who participated had voluntarily attended at least one ANC visit and completed couple HIV counseling and testing. This behavior has become more common in the region after community mobilization efforts to normalize male partner attendance at ANC appointments.⁴¹ It is possible if we worked to reach men who were not otherwise choosing to attend the ANC appointment, our intervention would have seen a greater effect.

For both men and women, retention was higher after emergence of COVID-19. This is likely due to new short-term regulations which allowed health facility-based clinicians to prescribe three-month drug dispensation (vs. one-month supply typically provided) to eligible post-partum patients to limit crowding and visit frequency at health facilities. Hence, this may not reflect improved adherence but simply a greater proportion of time that participants had ART on hand.

Infant seroconversion was slightly higher in our population than other countries in Southern Africa, with 8.5% of infants born alive testing positive for HIV by 18 months of age.^68,69 Infants born to couples in the intervention group trended towards lower risk of HIV seroconversion but the results were not statistically significant. Researchers have found that a supportive male partner can reduce the risk of vertical transmission by providing logistical assistance (e.g., medication pick-up as needed), encouraging adherence (e.g., reminding their partner to take their medication), and by providing social support in instances of experienced or anticipated discrimination due to their HIV status.^{46,50,70–74} Women and men participating in our study described in qualitative interviews the support their partner provided,⁶⁷ however, this did not translate to an impact on infant HIV outcomes.

Strengths and Limitations

Our large cluster randomized controlled study allowed us to compare the impact of a couple-based HIV care and treatment program on long-term clinical outcomes, including medication possession ratio after 12 months and infant seroconversion at 18 months of age. This study contributes to the literature by including couples who are both living with HIV and providing a clinic- and community-based intervention to improve treatment outcomes. Our study was limited to assessing the impact on medication pick-up given the sizable missingness around viral load (VL) data. While VL testing should be completed for all women on ART at entry to antenatal care or at three months after ART initiation for pregnant women initiating ART care during pregnancy, and at least annually for adults (i.e., non-pregnant/lactating persons) as standard of care, we suspect that stock outs in VL testing supplies and incomplete documentation in medical records resulted in insufficient VL data for analysis. Future studies in this area either need to allow for VL testing and result collection that is independent of or supplemental to the national health service system – an expensive and logistically challenging option – or use a point of care adherence assay (e.g., urine assay) to assess adherence.⁷⁵ Additionally, our study effects were likely blunted by men’s non-participation in approximately half of the community-based sessions designed to develop increased empathy and communication skills among the couples. Further studies to assess the impact among those who participated in interventional sessions vs. those who did not will be forthcoming.

Conclusions

Our study provides a community-clinic support framework that improved HIV retention outcomes among male partners. Further work needs to be done to improve social support for pregnant women and to facilitate prevention of vertical transmission to infants among couples living with HIV.