Outcomes of the MAMA Training: A Simulation and Experiential Learning Intervention for Labor and Delivery Providers to Improve Respectful Maternity Care for Women Living with HIV in Tanzania

Abstract

Respectful maternity care (RMC) for women living with HIV (WLHIV) improves birth outcomes and may influence women’s long-term commitment to HIV care. In this study, we evaluated the MAMA training, a team-based simulation training for labor and delivery (L&D) providers to improve RMC and reduce stigma in caring for WLHIV. The study was conducted in six clinical sites in the Kilimanjaro Region of Tanzania. 60 L&D providers participated in the MAMA training, which included a two-and-a-half-day workshop followed by a half-day on-site refresher. We assessed the impact of the MAMA training using a pre-post quasi-experimental design. To assess provider impacts, participants completed assessments at baseline and post-intervention periods, measuring RMC practices, HIV stigma, and self-efficacy to provide care. To evaluate patient impacts, we enrolled birthing women at the study facilities in the pre- (n = 229) and post- (n = 214) intervention periods and assessed self-reported RMC and perceptions of provider HIV stigma. We also collected facility-level data on the proportion of patients who gave birth by cesarean section, disaggregated by HIV status. The intervention had a positive impact on all provider outcomes; providers reported using more RMC practices, lower levels of HIV stigma, and greater self-efficacy to provide care for WLHIV. We did not observe differences in self-reported patient outcomes. In facility-level data, we observed a trend in reduction in cesarean section rates for WLHIV (33.0% vs. 24.1%, p = 0.14). The findings suggest that the MAMA training may improve providers’ attitudes and practices in caring for WLHIV giving birth and should be considered for scale-up.

Background

The perinatal period is critical for identifying individuals living with HIV, both to prevent HIV transmission to their children and to engage birthing individuals and their partners in HIV care. Clinical and behavioral interventions in the antenatal and postnatal clinical settings have produced extraordinary gains in the prevention of mother-to-child transmission (PMTCT) [1] and have improved the psycho-social needs and well-being of women living with HIV (WLHIV) [2]. A focus on building a healthcare workforce that can care for pregnant and postpartum WLHIV in PMTCT clinics has resulted in a workforce that is trained on the clinical and psychosocial needs of WLHIV. However, the intrapartum period (i.e., labor and delivery) has largely been neglected in PMTCT provider training; in particular, provider-focused interventions that build a culture of respectful and non-stigmatizing care for WLHIV in labor and delivery (L&D) are lacking [3, 4].

HIV-related stigma has a significant impact on the mental health and well-being of WLHIV [5]. There is evidence that WLHIV may receive suboptimal and stigmatizing care when they deliver. [6, 7] Stigma has been consistently identified as a reason why women disengage from HIV care [8–11], so disrespectful and stigmatizing experiences during childbirth may impact long-term engagement in postpartum HIV care. Stigmatizing attitudes and behaviors on the part of healthcare providers may stem from a fear of occupational exposure to HIV or personal disapproval of disease-associated behaviors, resulting in neglect, differential treatment, blame, or judgment [3, 12]. HIV-related stigma impacts patient engagement in PMTCT programs, medication adherence, and continued commitment to HIV care [8–11]. In addition, a lack of self-efficacy and clinical knowledge on the management of L&D for WLHIV may lead to suboptimal care through neglect, unnecessary interventions, or referrals that are not medically indicated [13].

Respectful and non-stigmatizing care by L&D providers improves the childbirth experience and can promote women’s continued commitment to HIV care [7, 14]. The World Health Organization defines respectful maternity care (RMC) as care that is free from mistreatment, maintains dignity and confidentiality, and enables informed choice and continuous support during childbirth [15]. Such practices not only improve outcomes for women and newborns but also reduce experiences of stigma, disrespect, and abuse, resulting in greater engagement and trust in the healthcare system [16, 17]. PRONTO International developed an evidence-based model for provider simulation training and, through a pilot study in Ghana, showed that the intervention significantly improved person-centered RMC as well as provider knowledge and self-efficacy [18, 19].

In this study, we adapted the PRONTO model, using the ADAPT-ITT Model [20] for evidence-based HIV Interventions, to develop MAMA (Mradi wa Afya ya Mama Mzazi, Project to Support the Health of Women Giving Birth), a simulation team-training curriculum for L&D providers that addresses provider stigma and promotes the delivery of evidence-based RMC for WLHIV. We pilot-tested the training in November 2022 in Moshi, Tanzania, with 60 L&D providers from primary care clinics in the Kilimanjaro region [21]. The goal of this paper is to assess the impact of the MAMA training on: 1) providers’ attitudes and practices, and 2) patients’ experiences of care. We also assess the feasibility and acceptability of the MAMA training to inform future scale-up.

Methods

This study used a quasi-experimental pre-post design to evaluate the impact of the MAMA training on outcomes in two populations: 1) L&D providers (n = 60) from six study sites who participated in the training, and 2) individuals giving birth at the six study sites, comparing individuals enrolled in the pre-training period (n = 229) and the post-training period (n = 214). The full study protocol has been described elsewhere [21].

The study was approved by the ethical review committees at the University of Utah (Protocol 00143918), Kilimanjaro Christian Medical University College (Protocol 2516), and the National Institute for Medical Research in Tanzania (Protocol 3853). The trial is registered at clinicaltrials.gov (NCT05271903).

Description of the MAMA Training

Using participatory methods and the ADAPT-ITT model for adapting evidence-based interventions in HIV prevention and care, we developed the MAMA training to promote the delivery of evidence-based respectful maternity care and address HIV stigma among L&D providers. The MAMA training is based on a simulation training model developed by PRONTO International, a global non-governmental organization, which uses low-tech and low-cost tools to deliver highly realistic birth simulations. Through applications and evaluations in multiple international settings, including East Africa, the PRONTO model has been shown to reduce morbidity and mortality [22–25], improve provider skills [26–28], and improve the use of RMC practices [19].

The MAMA training was a two-and-a-half-day training, followed by a two-hour in-situ refresher session in the providers’ clinical setting. It included case-based learning and knowledge review sessions, simulations, and interactive activities on teamwork and communication, HIV stigma, mindfulness, and provider burnout. The training was anchored around three simulation scenarios to apply and contextualize knowledge. The three simulation scenarios covered (1) a normal birth, (2) postpartum hemorrhage in a WLHIV, and (3) pre-term placenta previa in a newly diagnosed WLHIV. At the end of the training, each clinical team developed an action plan laying out their goals and desired changes for their facility. The in-situ follow-up training was an opportunity to reinforce objectives and training content, perform a simulation in the participants’ clinical environment, and review the facility-level action plan.

Procedures for Provider Participants

Study Participants

Eligible providers were doctors, clinical officers, nurses, and midwives working in L&D departments in six study clinics. The clinics were all classified as primary healthcare facilities and included three urban clinics in the Moshi district and three rural clinics in the Rombo district, of the 6 clinics 4 had the capacity for cesarean section. The sample size was based on clinical workload and provider availability, which enabled us to deliver the MAMA training to 60 providers, representing 62% of all eligible providers across the study clinics, and ranging from 48 to 83% in individual clinical sites.

Intervention Delivery

In cooperation with the leadership from the six study clinics, we invited ten providers from each clinic to participate in the MAMA training. The 2.5-day training was conducted at a conference center in Moshi, with overnight accommodation and meals provided to all participants. We offered two different trainings, each with 30 providers (5 per clinic) in order to minimize the impact on clinic staffing. The training was facilitated in Kiswahili by PRONTO expert trainers from Kenya who followed the MAMA training curriculum. Some content and debriefing sessions were conducted by local Tanzanian staff (in Kiswahili) and U.S. trainers/investigators (in English with simultaneous translation). Approximately one month after the training, the Tanzanian based- staff visited all six study sites to conduct the in-situ refresher training.

Data Collection

Participants completed surveys at four time points: immediate pre-training, immediate post-training, one-month post-training (at the in-situ visit), and three-months post-training. All surveys were completed in Kiswahili and self-administered on paper. All participants were assigned a unique study ID that linked their data across time points. The document linking the study ID with participant contact information was kept separate from the data and accessible only by the study team.

Measures

The surveys were developed using validated measures. Robust procedures were used to translate and back-translate measures and the surveys were pilot-tested with providers prior to use. Table 1 describes when each measure was administered.

Table 1.

Data collection timepoints for each measure

Measure	Pre-training	Post-training	1 month post-training	3 months post-training
Demographics	X
Practices of RMC	X		X	X
HIV stigma: precautions	X		X	X
HIV stigma: fear of acquisition	X	X		X
HIV stigma: attitudes	X	X		X
Self-efficacy	X	X		X

Primary Outcome

Practices of RMC were assessed using a measure of person-centered maternity care validated in Kenya and Ghana [29] including nine questions that asked how often they had used person-centered practices in the past month. Items had four response options (never to always) and were summed for a possible range of 0–27 (α = 0.73).

Secondary Outcomes

HIV stigma was assessed in three domains: extra precautions in providing care for WLHIV, fear of acquiring HIV in caring for WLHIV, and attitudes towards WLHIV.

To measure extra precautions taken in caring for WLHIV, we adapted a measure from previous work evaluating HIV stigma in healthcare in Ghana [30], with modifications specific to the L&D setting. The measure included seven questions asking providers how often they took additional precautions (e.g., avoiding physical contact, wearing double gloves) when caring for a WLHIV. Items had four available responses (never to most of the time) and were summed for a possible range of 0–21 (α = 0.71).

To measure fear of acquisition, we adapted a measure from previous work evaluating HIV stigma in healthcare in Ghana [30], with modifications specific to the L&D setting. The measure included nine questions asking providers how worried they would be about getting HIV if they had certain contact (e.g., taking blood pressure, dressing wounds) with a WLHIV who was not taking antiretroviral therapy (ART). Items had four available responses (not worried to very worried) for a possible range of 0–27 (α = 0.85).

To measure stigmatizing attitudes, we adapted a measure of provider stigma from previous work tested in six countries [31]. The measure included six items that asked providers how much they disagreed or agreed with statements about people living with HIV (e.g., HIV is a punishment for bad behavior). Items had four available responses (strongly disagree to strongly agree) and were summed for a possible range of 0–18 (α = 0.79).

Self-efficacy in obstetric care was measured using items adapted from previous PRONTO evaluations [18]. The items asked providers to rate their degree of confidence on a scale from 0 to 100 for several clinical actions (e.g., identifying the cause of bleeding in a case of postpartum hemorrhage, providing emotional support and guidance for a woman during labor and birth). Providers were asked to first rate their self-efficacy on each item when caring for a woman without HIV; the items were then repeated for caring for a WLHIV. We created a mean score (0–100) separately for self-efficacy in caring for a woman without HIV (α = 0.90) and caring for a woman with HIV (α = 0.91).

Co-variates

We included demographic questions on age, gender, clinical training, years since clinical schooling, years working in L&D, years at current hospital, and number of previous trainings on L&D for WLHIV in the past year.

Procedures for Birthing Women Participants

Study Participants

Eligible participants were women who gave birth in the six study clinics during the pre-training period (March to July 2022) and post-training period (January to May 2023). All WLHIV who gave birth in either period were eligible for enrollment. For each enrolled WLHIV, we enrolled one to two HIV-negative women who delivered in the same facility during the same time period, with matching on parity (primiparas or multiparas), mode of delivery (cesarean or vaginal birth), and age (± 3 years). Individuals were excluded if they had mental or physical disabilities (e.g., difficulty seeing or hearing) that limited their ability to provide informed consent and complete the survey assessment. An a priori power calculation suggested that in order to detect a moderate intervention effect (d = 0.4) with 80% power and a 2-sided significance level of 5%, we needed to enroll 103 WLHIV per time period. To allow comparison, a comparable sample of HIV-negative women with some similar characteristics with WLHIV was recruited.

We also conducted a review of clinical ledgers at the facilities during the same periods when we enrolled study participants. The rationale was to explore broader facility-level trends that could not be captured in our enrolled sample. We recorded the number of individuals who gave birth at the study clinics, the proportion of those who were documented as living with HIV, and the mode of delivery (vaginal vs. cesarean section).

Data Collection

When a WLHIV gave birth at any study site, L&D staff introduced the study and asked the patient if she was interested in participating in the study. If so, the study team was called, and a team member would travel to the clinic to enroll the patient after birth and prior to discharge. Informed consent was conducted orally and did not include information about HIV status to ensure their HIV status was not involuntarily disclosed to others. This same process was repeated for women who were HIV-negative and had recently given birth.

After being consented, participating women completed a structured survey using audio computer-assisted self-interview (ACASI) technology in Kiswahili with headphones on tablets running Questionnaire Development System (QDS) software. The ACASI modality was used to ensure patient privacy and minimize social desirability bias to improve overall data validity [32]. The survey took approximately 30–45 min, and research staff were present to aid participants if needed. Medical record data were collected by a research staff on the same day. A locator form was completed to allow for a brief follow-up survey approximately three weeks after the birth. The follow-up survey was orally administered over the phone and took approximately 10 min to complete. The follow-up survey focused on the health and well-being of the woman and her child, including details about engagement in post-partum clinical care. The researcher attempted follow-up with women for three weeks before classifying a participant as lost to follow-up.

Measures

The survey for birthing women was developed using validated measures. Robust procedures were used to translate and back-translate measures. The survey was pilot-tested with study team members and nursing students prior to use.

Primary Outcome

Perceptions of RMC were assessed using a measure of person-centered maternity care developed by Afulani and colleagues [33, 34], including 30 items that had participants rate the care they had received. Items were standardized on a scale of 0–100 for both the full RMC scale (α = 0.89) and the sub-scales of dignity and respect (6 items, α = 0.70), communication and autonomy (9 items, α = 0.76), and supportive care (15 items, 0.78).

Secondary Outcomes

Secondary outcomes were measured among the sub-set of the full sample who were living with HIV.

Postpartum HIV medication adherence was assessed during the follow-up phone survey three-weeks post-discharge. Participants living with HIV were asked to self-report their adherence to ART in the previous four days. Poor adherence was defined as missing any ART medication in the past four days.

Perceptions of HIV stigma in the facility were assessed with an 11-item measure among WLHIV. This measure has been adapted from two measures that identify HIV stigmatizing behaviors among healthcare workers [35, 36]. Participants were asked to rate their perceptions of stigma by the healthcare provider (e.g., Do you feel you received poorer quality care because of your HIV status?). Items had four available responses (never to always) and were summed for a possible range of 0–33 (α = 0.80).

Internalized HIV stigma was assessed using Scale A of the HIV and Abuse Related Shame Inventory (HARSI) [37]. The 13-item measure asked participants about experiences of internalized stigma (e.g., I put myself down for getting HIV). Items had four available responses (not at all to very much) and were summed for a possible range of 0–52 (α = 0.90).

Co-variates

The survey included demographic questions about age, education, marital status, religion, literacy, and household wealth. Literacy was measured with a single question: “How would you describe your ability to read and write?” with three response options (1 = “I cannot read and write”, 2 = ”I can read and write with some difficulty”, 3 = ”I can read and write very well”). Options one and two were combined to capture low literacy. Household wealth was measured in quintiles and calculated from the wealth index based on ten questions about household assets [38]. Data collected from the medical records included parity, number of antenatal care visits, and presence of complications (e.g., hemorrhage, eclampsia, obstructed labor, episiotomy, third/fourth degree tear, retained placenta).

Data Processing and Analysis

Provider data were entered into REDCap software and exported to R for data analysis. We first used summary statistics to describe participant characteristics and visualize trends in each measure over time. Mixed effect regressions were then used to assess the significance of observed changes from baseline to each follow-up timepoint, using means and 95% confidence intervals (CI).

Patient data were either transferred from QDS software (ACASI survey) or entered into REDCap software (medical record data and follow-up telephone survey) and exported to R for data analysis. We first used summary statistics to describe participant demographic and birth characteristics. Comparison between pre-intervention and post-intervention full RMC scores was made using a linear regression model while controlling for demographic and clinical characteristics. Similar models were also used to assess changes in each of the RMC subscales. To assess whether HIV status moderates the impact of the intervention on RMC, an interaction term for HIV status and timepoint was added to the models. To assess the secondary outcomes, we evaluated similar models for the sub-set of the sample living with HIV.

Acceptability, Feasibility, and Fidelity of the MAMA Training

Acceptability of the MAMA training from the perspective of the provider participants was operationalized as satisfaction and assessed at two time points: (1) in the post-training survey immediately following the 2.5-day training, and (2) in the survey immediately following the in-situ training 1 month later. Participants were asked on a 4-point Likert scale how much they disagreed or agreed with eight statements (e.g., the content of the MAMA training was useful to me; I felt like the training was a good use of my time). We also included three open-ended questions, eliciting feedback from providers on aspects of the training they most enjoyed, features they would like changed, and ways they felt the training will impact their practice.

Feasibility was evaluated based on the proportion of invited providers who attended the training and the proportion of the training time attended, using attendance rosters requiring sign-in for each half-day of the 2.5-day workshop and the in-situ training. Our criteria for feasibility were 65% of participants attending for a minimum of 75% of the training time. Fidelity was assessed through semi-structured observations of whether the content was covered as intended in our MAMA training manual.

Results

Provider Results

Description of the Sample

Table 2 describes baseline characteristics of our sample of 60 providers.

Table 2.

Description of the provider sample (n = 60)

	n (%)
Gender
Female	29 (48%)
Male	31 (52%)
Clinical training
Nurse midwife	40 (66.7%)
Clinical officer	4 (6.7%)
Medical doctor	16 (26.7%)
Number of trainings on L&D for WLHIV in the past year
None	46 (77%)
1 or more	14 (23%)
	Median (Q1, Q3)
Age	32 (27, 37)
Years since clinical training	8 (4, 12)
Years providing care in L&D	5 (2, 8)
Years working at current clinic	5 (1, 9)

Intervention Impacts

Positive changes were observed across all the outcomes of interest (Table 3).

Table 3.

Primary and secondary intervention outcomes with providers (n = 60)

	Pre-intervention	Immediate post-intervention			1-month post-intervention			3-months post-intervention
	Mean (SD)	Mean (SD)	Regression coefficient (95% CI)	p value	Mean (SD)	Regression coefficient (95% CI)	p value	Mean (SD)	Regression coefficient (95% CI)	p value
RMC	17.1 (2.9)				20.9 (3.0)	3.8 (2.9, 4.7)	<.001	21.8 (3.1)	4.7 (3.8, 5.6)	<.001
Stigma: precautions	9.9 (3.8)				6.5 (3.4)	−3.2 (−4.1, −2.3)	<.001	6.3 (3.5)	−3.4 (−4.3, −2.6)	<.001
Stigma: fear of acquisition	6.5 (5.7)	1.9 (2.8)	−4.6 (−5.9, −3.4)	<.001				3.2 (3.8)	−3.2 (−4.5, −1.9)	<.001
Stigma: attitudes	4.6 (2.8)	2.8 (2.7)	−1.8 (−2.4, −1.2)	<.001				3.0 (3.2)	−1.5 (−2.1, −0.9)	<.001
Self-efficacy: General	84.6 (11.8)	95.5 (5.9)	10.9 (8.2, 13.6)	<.001				94.1 (7.4)	9.6 (6.8, 12.3)	<.001
Self-efficacy: WLHIV	82.4 (13.1)	95.2 (6.8)	12.8 (9.7, 15.8)	<.001				93.4 (8.4)	11.0 (7.9, 14.1)	<.001

All regression coefficients are compared to baseline

At baseline, the mean RMC score reported by providers was 17.1 on a scale of 0–27. At one-month post-intervention, the mean RMC score increased to 20.9 with a regression coefficient of 3.8 (95% CI 2.9, 4.7, p < 0.001). At three months, the mean RMC score was 21.8 with a regression coefficient of 4.7 (95% CI 3.8, 5.6, p < 0.001) compared to the baseline (Fig. 1).

Fig. 1.

Self-reported RMC practices pre- and post-MAMA training (n = 60)

Each stigma subscale (extra precautions, fear of acquisition, attitudes towards WLHIV) was analyzed separately. For increased precautions taken in caring for WLHIV, scores significantly decreased from 9.8 on a 0–21 scale at baseline to 6.5 (p < 0.001) and 6.3 (p < 0.001) at one- and three-months post-intervention, respectively. For fear of acquisition, scores significantly decreased from 6.4 on a 0–27 scale at baseline to 1.8 (p < 0.001) and 3.2 (p < 0.001) immediately post- and three-months post-intervention, respectively. For stigmatizing attitudes towards WLHIV, scores significantly decreased from 4.6 on a 0–18 scale at baseline to 2.8 (p < 0.001) and 3.0 (p < 0.001) immediately post- and three-month post-training, respectively.

The median score at baseline for self-efficacy in providing obstetric care to all women was 84.6 on a scale of 0–100. After the training, the median score significantly increased to 95.5 (p < 0.001) and 94.1 (p < 0.001) immediately post- and three-months post-intervention, respectively. In caring specifically for WLHIV, the median reported score for self-efficacy also significantly increased from 82.4 at baseline to 95.2 (p < 0.001) and 93.4 (p < 0.001) immediate post- and three-months post-intervention, respectively.

Patient Results

Description of the Sample

A sample of 443 birthing women was enrolled, 229 in the period before the intervention and 214 in the period after the intervention. Apart from age, the demographic and birthing characteristics of participants in both time points were overall similar (Table 4). Women enrolled after the intervention were slightly older than women enrolled before the intervention [Mean(sd): 27.6 (5.6) vs 29.1 (6.2); p = 0.018].

Table 4.

Characteristics of participants (n = 443)

Characteristic	Pre-intervention n = 229	Post-intervention n = 214	p value
Demographics
Age	27.62 (5.60)	29.08 (6.19)	0.018
Missing	1	2
Level of education			0.3
Some primary education	108 (47.2%)	116 (54.5%)
Some secondary education	102 (44.5%)	79 (37.1%)
Complete university/college	19 (8.3%)	18 (8.5%)
Missing	0	1
Relationship status			0.7
Married/in a relationship	211 (92.5%)	196 (91.6%)
Single/separated/divorced/widow	17 (7.5%)	18 (8.4%)
Missing	1	0
Literacy			0.13
Can’t or read and write with some difficulty	43 (18.8%)	53 (24.8%)
Read and write very well	186 (81.2%)	161 (75.2%)
Wealth categories			0.10
Lowest	14 (6.2%)	5 (2.4%)
Medium	31 (13.7%)	24 (11.3%)
Highest	182 (80.2%)	183 (86.3%)
Missing	2	2
Religion			0.8
Christian	171 (74.7%)	154 (72.0%)
Muslim	57 (24.9%)	59 (27.6%)
Other	1 (0.4%)	1 (0.5%)
Type of facility			0.6
Public	160 (69.9%)	144 (67.3%)
Private	69 (30.1%)	70 (32.7%)
Birth history
Parity			0.14
1	63 (27.5)	46 (21.5)
> 1	166 (72.5)	168 (78.5)
Number of ANC visits during pregnancy			0.5
< 4 visits	55 (24.3%)	57 (26.9%)
> = 4 visits	171 (75.7%)	155 (73.1%)
Missing	3	2
Experienced any complications?			0.9
No complication	117 (51.1%)	111 (51.9%)
Had complications	112 (48.9%)	103 (48.1%)

Intervention Impacts

The RMC scores of participants recruited in the period after the intervention did not differ significantly from the scores of patients recruited in the period before the intervention, considering both overall score and subscale scores (Table 5). Observed mean scores before and after intervention were 73.7 vs. 71.7 overall, 79.8 vs. 79.6 for dignity and respect subscale, 63.5 vs. 61.5 for communication and autonomy, and 75.3 vs. 74.6 for supportive care. Similarly, there were no significant differences in RMC scores by time period in regression models after controlling for relevant variables (Table 5). Also, the interaction term of HIV status and timepoint was not significant, indicating that HIV status did not moderate the impact of the intervention. Among WLHIV only (n = 206), perception of stigma in the health care facility and internalized stigma did not change after the intervention (Table 5). Adherence to ART and Nevirapine use for HIV exposed baby were both high, and did not change over time (Table 6).

Table 5.

Intervention effect (n = 443)

	Mean (SD)	Regression coefficient (95% CI)	p value
RMC full scale
Pre-intervention	73.7 (12.3)	1
Post-intervention	71.7 (13.1)	−1.1 (−3.3, 1.0)	0.299
Dignity and respect
Pre-intervention	79.8 (17.9)	1
Post-intervention	79.6 (16.9)	−0.1 (−0.7, 0.5)	0.701
Communication and autonomy
Pre-intervention	63.5 (16.8)	1
Post-intervention	61.5 (17.3)	−0.6 (−1.5, 0.3)	0.173
Supportive care
Pre-intervention	75.3 (11.8)	1
Post-intervention	74.6 (13.2)	−0.4 (−1.4, 0.6)	0.458
Women living with HIV only (n = 206)
Perception of HIV Stigma at health facility
Pre-intervention	4.2 (5.2)	1
Post-intervention	3.9 (4.7)	−0.5 (−1.8, 0.8)	0.419
Internalized stigma
Pre-intervention	23.2 (12.4)	1
Post-intervention	20.4 (11.9)	−2.5 (−5.9, 0.9)	0.148

Regression coefficient have adjusted for participant’s age, level of education, relationship status, literacy, wealth category, experience of complications, number of ANC visits, type of facility and HIV status

Table 6.

Clinical outcomes and evidence-based practices

Characteristic	Pre-intervention n = 229 n (%)	Post-intervention n = 214 n (%)	p value
How was the baby delivered?			> 0.9
Vaginal	177 (77.3)	166 (77.6)
C-Section	52 (22.7)	48 (22.4)
Reasons for C-section noted
No	2 (96.2)	1 (2.0)	> 0.9
Yes	50 (3.8)	47 (98.0)
Completion of the 3 steps of AMSTL			0.024
No	55 (24.0)	70 (32.7)
Yes	165 (72.1)	142 (66.4)
Missing	9 (3.9)	2 (0.9)
Use and timing of oxytocin
No	1 (0.4)	2 (0.9)	0.6
Yes, before labor	0 (0.0)	1 (0.5)	0.5
Yes, during 1st stage of labor	0 (0.0)	2 (0.9)	0.2
Yes, during 3rd stage of labor	200 (87.3)	192 (89.7)	0.4
Yes, Postpartum	2 (0.9)	1 (0.5)	> 0.9
Missing (not in record)	24 (10.5)	17 (7.9)	0.4
Transferred in			0.11
No	205 (89.5)	202 (94.4)
Yes	23 (10.0)	12 (5.6)
Missing	1 (0.5)	0 (0.0)
Episiotomy			0.5
No	200 (87.3)	182 (85.0)
Yes	29 (12.7)	32 (15.0)
WLHIV only	n = 103	n = 105
How was the baby delivered?
Vaginal	75 (72.8)	79 (75.2)	0.8
C-Section	28 (27.2)	26 (24.8)
Baby currently using Nevirapine?
No	1 (1.2)	0 (0)	0.3
Yes	83 (97.7)	85 (94.4)
Not applicable	1 (1.2)	4 (4.4)
Missing	0 (0)	1 (1.1)
ART adherence
No	1 (1.2)	0 (0)	0.5
Yes	84 (98.8)	89 (98.8)
Missing	0 (0)	1 (1.2)

When examining clinical ledgers in the study facilities during the participant enrollment periods, we identified women giving birth in the pre- (n = 3645) and post- (n = 3797) intervention periods. The proportion of births where the patient was identified as living with HIV was similar in the two time periods (2.99% vs. 3.05%, p = 0.892). In the four clinics that had the capacity to perform cesarean sections, the proportion of all births (n = 7,442) that were cesarean sections did not differ in the two time periods (28.4% vs. 29.2%, p = 0.41). Considering only WLHIV (n = 208), the proportion of births that were cesarean sections decreased by 9% following the intervention, but this difference was not significant (33.0% vs. 24.1%, p = 0.14).

Intervention Feasibility, Fidelity, and Acceptability

All participating providers agreed with all the statements regarding a positive training experience, useful content, and effective facilitation (Fig. 2). The qualitative feedback was also overwhelmingly positive, with providers requesting additional sessions and longer training.

Fig. 2.

Ratings of the MAMA training (n = 60)

One hundred percent of invited providers attended the entire 2.5-day MAMA training, and 91.7% (55/60 providers) attended the in-situ training, indicating a high feasibility and acceptability of the MAMA training. The content was covered as intended in the training manual, with no deletions and only minor modifications for time management, indicating high fidelity to the intervention curriculum.

Discussion

This study evaluated the MAMA training, a simulation team-training curriculum for L&D providers that addresses providers’ instrumental and attitudinal stigma toward WLHIV and promotes the delivery of evidence-based RMC for WLHIV. In our quasi-experimental evaluation of the MAMA training with 60 L&D providers across six clinics, we saw positive changes among providers on RMC practices, HIV stigma, and self-efficacy to care for birthing WLHIV. We also observed that the proportion of cesarean section deliveries among WLHIV decreased following the intervention, although the difference was not statistically significant. The MAMA training was highly rated among provider participants and was observed to have high feasibility and fidelity. The training impact, combined with the positive ratings of the training modality, suggests that a provider intervention that uses low-tech and hands-on simulation activities is an effective way to simultaneously address clinical and interpersonal skills to improve the birth experience for WLHIV and supports the justification for future implementation research of the MAMA training.

The MAMA training used hands-on simulation to refine clinical skills and improve interpersonal competencies, such as patient-provider communication. The simulations included cognitive, technical, and behavioral objectives, building on shared experiences and facilitating reflection and debriefing. Simulation learning methodology has been shown to improve the delivery of RMC, change providers’ attitudes and behaviors, and enhance clinical self-efficacy [39]. In our formative research, we noted that instrumental HIV stigma was driven largely by fear of occupational exposure to HIV. The simulations helped providers develop the skills and self-efficacy to manage routine births and obstetric emergencies in WLHIV while protecting themselves. The simulations further helped providers maintain the principles of RMC and be responsive to the unique needs of WLHIV while dealing with high-stress clinical emergencies.

The MAMA training had a significant impact on providers’ self-reported practices of RMC, their HIV stigmatizing attitudes, and their self-efficacy to care for birthing women living with HIV. These changes in practices and attitudes likely emanated from the unique curriculum that allowed providers to review RMC practices, apply new skills through simulations, and develop and practice their clinical empathy and communication skills with dedication to patient privacy, confidentiality, and autonomy. Doing this through a shared experience with their fellow colleagues likely strengthened the cooperation and shared values in the clinical environment and sustained the impacts beyond the intervention period. Healthcare providers in low-and middle-income country (LMIC) settings, and L&D clinics in particular, are often overworked and under-valued [14, 40–42]. Having a training model that centers the providers and addresses both their challenges (e.g., burnout, resource constraints) and their strengths (e.g., dedication to their profession and patients) is essential to facilitate sustained changes in clinical practice.

In the facility-level data, we observed a decrease in rates of cesarean sections among WLHIV in our study clinic. Although this change was not statistically significant (p = 0.14), it does represent a 9% decrease from 33 to 24% of births. While cesarean sections can be lifesaving when medically indicated, global trends suggest that unnecessary cesarean sections have increased significantly in the last two decades, with a deleterious impact on maternal morbidity and mortality [43]. A recent meta-analysis concluded that elective cesarean sections for WLHIV were associated with a three-fold increase in maternal morbidity [44]. The World Health Organization has long determined that elective cesareans for WLHIV are not medically indicated because the harms outweigh the benefits [45]. In light of universal treatment for HIV and evidence that maternal viral suppression nearly eliminates the risk of maternal-to-fetal transmission of HIV [46, 47], the guidance that WLHIV should have a vaginal birth, if possible, becomes even more important. Supporting a vaginal birth takes time, attention, and patient-centered support. The MAMA training provided participants with skills of respectful and non-stigmatizing maternity care to support birthing WLHIV through labor and delivery, emphasized the value of a vaginal birth, helped providers identify obstetric emergencies that require cesarean sections, and addressed provider fear of occupational exposure to HIV. This training content likely made providers more willing to see birthing women through to a successful vaginal birth and less likely to opt prematurely for a cesarean section. Although we do not know whether the cesarean sections that were done were medically indicated, we did observe that the cesarean rate for WLHIV got closer to the 15% threshold that is typically considered an indicator of appropriate cesarean section rates [48], and that the gap in cesarean section rates between women with and without HIV closed following the study intervention. This is an important finding that deserves additional investigation.

We did not observe patient-reported impacts following the MAMA training. Our assessment of the pre-intervention data showed that RMC was generally high across all domains; however, patients with low socio-economic status (specifically, low literacy) had much lower ratings of RMC [49]. The high ratings of RMC prior to intervention delivery may have precluded us from seeing any longitudinal change. Additionally, our sample was well engaged in care, as noted in the high levels of postpartum medication adherence at both the pre- and post-intervention time points, which may indicate that it is a unique patient population. Finally, the training did not reach all the providers in the six study clinics, which may have prevented a broader shift in clinic culture and practices to reach all birthing individuals.

The study must be interpreted in light of its limitations. First, the study design was not randomized and did not have a comparison group to control for the effect of time and regression to the mean. Second, the providers were nominated by clinic leadership and not randomly selected from the study clinics. As a result, the providers selected as participants may have been more amenable to change than providers who were not selected. Third, the outcome measures of interest were self-reported and were therefore subject to social desirability bias. We did our best to minimize this with self-completed surveys by providers and the use of ACASI technology with patients. Still, participants may have felt compelled to provide positive reports of their experiences, behaviors, and attitudes. Fourth, participant feedback on satisfaction of the intervention was limited, and may not be comprehensive enough to inform the scale-up of the intervention. Participant feedback was given on the MAMA intervention as a whole, and not on the individual components of the intervention, which precludes us from making specific recommendations for changes to the intervention content. It is also possible that participant feedback reflected positive feelings about for the logistics of the training (e.g., time away from training, group meals), rather than the specific intervention content. Finally, this study did not assess the opportunities and barriers related to scale-up of the intervention. Future research should include measures for cost and time, assess administrative investment, and explore train-the-trainer cascade models to reach more providers.

In conclusion, this study provides data on the feasibility, acceptability, and potential impact of a simulation team-training curriculum for L&D providers that aims to improve the delivery of respectful, non-stigmatizing care for birthing WLHIV. L&D providers who deliver respectful and intentionally non-stigmatizing care to WLHIV can improve maternal and child health outcomes [16, 17], deepen women’s trust in the health care system, and motivate them to remain in HIV care in the postpartum period [7]. The impact of the training curriculum on provider outcomes justifies a more robust evaluation and consideration for scale-up to reach more L&D providers in Tanzania and beyond.