Publications

BACKGROUND: Doulas, non-clinical professionals who provide support throughout the perinatal period, can positively impact patient experiences and clinical outcomes during birth. Doulas often support hospital-based births without being employed by the hospital system, resulting in varied relationships with hospitals and clinicians. Systems-level changes are needed to maximize collaboration between hospitals and doulas to ensure facilitation of, and not barriers to, doula support. We implemented and evaluated a new program, called the "Supportive Birth Collaborative," to maximize effectiveness of doula support in hospital settings.

METHODS: We conducted a single-site feasibility study of the use of implementation mapping to make systemic changes to clinician-doula collaboration for labor and delivery. Implementation mapping consisted of five steps: developing a collaborative of program implementers and knowledge holders, conducting a needs assessment, developing a logic model, applying implementation strategies, and evaluating changes in outcomes. To evaluate change, process data were collected throughout, and implementation outcomes were measured in 2022 and again after one year of implementation via online surveys to all clinicians who provided labor and delivery care. Descriptive statistics were calculated and change over time was analyzed in Stata using log-binomial regression models with clustering to account for respondents who completed both surveys.

RESULTS: The "Supportive Birth Collaborative" (SBC) was founded in November 2021. The first meeting included 19 people, who were obstetricians, anesthesiologists, nurses, doulas, students, social workers, administrators, researchers, and individuals who had given birth at the study hospital. From 2022-2023, the SBC adopted 11 implementation strategies and piloted or fully implemented 10 of them. Implementation strategies ranged from making training dynamic, to changes in the physical environment, to changes in formal policy. In 2022, 104 clinicians participated in the survey; 97 participated in 2023. There was significant improvement in clinician-reported trust in doulas (0.23, 95% CI: 0.12, 0.34) and doula-clinician communication (0.25, 95% CI: 0.12, 0.38). Clinicians had a limited understanding of the doula's role, and that understanding did not significantly improve.

CONCLUSIONS: Using implementation mapping as a guide to collaborative work can lead to meaningful health system changes. Regular review of implementation outcomes could allow for adaptation and tailoring of implementation strategies.

The provision of high-quality antenatal care (ANC) is important for preventing maternal and newborn mortality and morbidity, but only around half of pregnant women in Tanzania attended four or more ANC visits in 2019. Although there is emerging evidence on the benefit of community health worker (CHW) interventions on ANC uptake, few large-scale pragmatic trials have been conducted. This pragmatic cluster-randomized trial, implemented directly through the public sector health system, assessed the impact of an intervention that trained public sector CHWs to promote the uptake of ANC. We randomized 60 administrative wards in Dar es Salaam to either a targeted CHW intervention or a standard of care. The impact of the intervention was assessed using generalized estimating equations with an independent working correlation matrix to account for clustering within wards. A total of 243 908 women were included in the analysis of our primary outcome of four or more ANC visits. The intervention significantly increased the likelihood of attending four or more ANC visits [relative risk (RR): 1.42; 95% confidence interval (CI): 1.05, 1.92] and had a modest beneficial effect on the total number of ANC visits (percent change: 7.7%; 95% CI: 0.2%, 15.5%). While slightly more women in the intervention arm attended ANC in their first trimester compared with the standard-of-care arm (19% vs 18.7%), the difference was not significant (RR: 1.02; 95% CI: 0.84, 1.22). Our findings suggest that trained CHWs can increase attendance of ANC visits in Dar es Salaam and similar settings. However, additional interventions appear necessary to promote the early initiation of ANC. This study demonstrates that routine health system data can be leveraged for outcome assessment in trials and programme evaluation and that the results are likely superior, both in terms of bias and precision, to data that are collected specifically for science.

BACKGROUND: As awareness of perinatal health disparities grows, many birthing people of color are seeking racially and/or culturally concordant providers. We described preferences for, and perceptions of, racial and/or cultural concordance and cultural competence in the context of the doula-client relationship.

METHODS: Seven focus group discussions (FGDs) with a total of 27 participants were conducted to investigate the perspectives of patients and doulas across Massachusetts, United States. An interdisciplinary stakeholder group informed the data collection instrument content and design. Two coders achieved 0.89 Kappa for inter-rater reliability prior to coding the remaining transcripts. We used a modified grounded theory approach and Dedoose software for coding.

RESULTS: Two major themes emerged. First, cultural competency in doula care is a learning process, with definitions consistent with terms such as "cultural humility" and "structural competency." Doulas discussed listening to clients' needs rather than making assumptions, the importance of understanding privilege and power dynamics, and self-initiating relevant education beyond formal doula training. Second, trust was most frequently cited as an indicator of successful doula-patient relationships.

CONCLUSIONS: Most study participants specified the importance of cultural humility in doula-client relationships. Doulas approaching the relationship humbly with a willingness to learn and challenge their own assumptions-regardless of the level of concordance-can make a meaningful impact on the perinatal experience.

INTRODUCTION: The financial burden of pregnancy in the United States can be high and is associated with worse mental health and birth outcomes. Research on the financial burden of health care, such as the development of the COmprehensive Score for Financial Toxicity (COST) tool, has been conducted primarily among patients with cancer. This study aimed to validate the COST tool and use it to measure financial toxicity and its impacts among obstetric patients.

METHODS: We used survey and medical record data from obstetric patients at a large medical center in the United States. We validated the COST tool using common factor analysis. We used linear regression to identify risk factors for financial toxicity and to investigate associations between financial toxicity and patient outcomes including satisfaction, access, mental health, and birth outcomes.

RESULTS: The COST tool measured two distinct constructs of financial toxicity in this sample: current financial toxicity and concern over future financial toxicity. Racial/ethnic category, insurance, neighborhood deprivation, caregiving, and employment were associated with current financial toxicity (P < 0.05 for all). Only racial/ethnic category and caregiving were associated with concern over future financial toxicity (P < 0.05 for all). Both current and future financial toxicity were associated with worse patient-provider communication, depressive symptoms, and stress (P < 0.05 for all). Financial toxicity was not associated with birth outcomes or keeping obstetric visits.

CONCLUSIONS: The COST tool captures two constructs among obstetric patients, current and future financial toxicity, both of which are associated with worse mental health and patient-provider communication.

OBJECTIVE: The PNQIN (Perinatal-Neonatal Quality Improvement Network of Massachusetts) sought to adapt the Reduction of Peripartum Racial and Ethnic Disparities Conceptual Framework and Maternal Safety Consensus Bundle by selecting and defining measures to create a bundle to address maternal health inequities in Massachusetts. This study describes the process of developing consensus-based measures to implement the PNQIN Maternal Equity Bundle across Massachusetts hospitals participating in the Alliance for Innovation on Maternal Health Initiative.

METHODS: Our team used a mixed-methods approach to create the PNQIN Maternal Equity Bundle through consensus including a literature review, expert interviews, and a modified Delphi process to compile, define, and select measures to drive maternal equity-focused action. Stakeholders were identified by purposive and snowball sampling and included obstetrician-gynecologists, midwives, nurses, epidemiologists, and racial equity scholars. Dedoose 9.0 was used to complete an inductive analysis of interview transcripts. A modified Delphi method was used to reach consensus on recommendations and measures for the PNQIN Maternal Equity Bundle.

RESULTS: Twenty-five interviews were completed. Seven themes emerged, including the need for 1) data stratification by race, ethnicity and language; 2) performance of a readiness assessment; 3) culture shift toward equity; 4) inclusion of antiracism and bias training; 5) addressing challenges of nonacademic hospitals; 6) a life-course approach; and 7) selection of timing of implementation. Twenty initial quality measures (structure, process, and outcome) were identified through expert interviews. Group consensus supported 10 measures to be incorporated into the bundle.

CONCLUSION: Structure, process, and outcome quality measures were selected and defined for a maternal equity safety bundle that seeks to create an equity-focused infrastructure and equity-specific actions at birthing facilities. Implementation of an equity-focused safety bundle at birthing facilities may close racial gaps in maternal outcomes.

OBJECTIVE: To explore Spanish-speaking patients' experiences and preferences regarding communication during pregnancy care with specific attention to language barriers.

METHODS: Patients with a Spanish language preference who gave birth between July 2022 and February 2023 at an academic medical center were invited to participate in focus groups. Focus groups were held over Zoom, audio-recorded, transcribed in Spanish, translated into English, and reviewed for translation accuracy. Thematic analysis was conducted with deductive and inductive approaches. Three investigators double-coded all transcripts, and discrepancies were resolved through team consensus.

RESULTS: Seven focus groups (27 total participants, range 2-6 per group) were held. Three key themes emerged regarding patient experiences and communication preferences when seeking pregnancy care: 1) language concordance and discordance between patients and clinicians are not binary-they exist on a continuum; 2) language-discordant care is common and presents communication challenges, even with qualified interpreters present; and 3) language discordance can be overcome with positive interpersonal dynamics between clinicians and patients.

CONCLUSION: Our findings highlight the importance of relationship to overcome language discordance among patients with limited English proficiency during pregnancy care. These findings inform potential structural change and patient-clinician dyad interventions to better meet the communication needs of patients with limited English proficiency.

BACKGROUND: The WHO strategy for cervical cancer elimination strives to achieve 70% coverage with high-performance cervical screening. While few low- and middle-income countries have achieved this, high-risk human papillomavirus (hrHPV) self-testing creates the possibility to rapidly upscale access to high-performance cervical screening across resource settings. However, effective hrHPV screening requires linkage to follow-up, which has been variable in prior studies. This study developed and tested an implementation strategy aimed at improving screening and linkage to follow-up care in South East District in Botswana.

METHODS: This study performed primary hrHPV self-testing; those with positive results were referred for a triage visit. Withdrawals for any reason, loss-to follow-up between hrHPV test and triage visit, and number of call attempts to give hrHPV results were also documented. Acceptability of the program to patients was measured as the proportion of patients who completed a triage visit when indicated, meeting the a priori threshold of 80%. Feasibility was defined as the proportion of participants receiving the results and attending follow-up. To assess the associations between participant characteristics and loss-to-follow-up we used log-binomial regressions to estimate risk ratios and 95% confidence intervals (CI).

RESULTS: Enrollment of 3,000 women occurred from February 2021 to August 2022. In total, 10 participants withdrew and an additional 33 were determined ineligible after consent, leaving a final cohort of 2,957 participants who underwent self-swab hrHPV testing. Half (50%) of participants tested positive for hrHPV and nearly all (98%) of participants received their hrHPV results, primarily via telephone.  Few calls to participants were required to communicate results: 2,397 (82%) required one call, 386 (13%) required 2 calls, and only 151 (5%) required 3-5 calls. The median time from specimen collection to participant receiving results was 44 days (IQR, 27-65). Of all hrHPV positive participants, 1,328 (90%) attended a triage visit.

DISCUSSION: In a large cohort we had low loss-to-follow-up of 10%, indicating that the strategy is acceptable. Telephonic results reporting was associated with high screening completion, required few calls to participants, and supports the feasibility of hrHPV self-testing in primary care followed by interval triage.

OBJECTIVE: User experience is an important aspect of quality of care that is highly valued by patients. However, there are currently no validated tools for measuring user experience among caregivers of sick children in low- and middle-income countries. We aimed to develop and validate a measure of user experience in this population in primary healthcare facilities in rural Tanzania, where major quality improvement efforts to date have not included a large focus on user experience. We then aimed to describe variation in user experience between and within facilities.

METHODS: Informed by theory and formative qualitative research, we developed questions to measure user experience across three domains: prompt care, respect, and communication. We then conducted interviewer-administered surveys with caregivers of sick children. Using survey data, we conducted psychometric analyses to inform the development of a composite measure of user experience. Finally, we used multilevel models to describe variation in user experience and examine associations with facility, patient, and caregiver characteristics.

RESULTS: Surveys were completed by 1085 caregivers across 75 facilities. In exploratory factor analysis, user experience items did not group according to theoretical domains. We therefore assessed items individually and designed a single 8-item additive measure of user experience. Using this composite measure, and adjusting for differences in case mix across facilities, we found that 69% of variation in user experience was within facilities and 31% was between facilities. Smaller facility size and more caregiver education were positively associated with user experience.

CONCLUSIONS: We found that user experience varied significantly across health facilities in rural Tanzania, highlighting opportunities for improvement. Measurement tools are needed to inform efforts to improve user experience and monitor changes over time. The scale developed in this study could serve as a starting point for the measurement of user experience among caregivers in similar settings.

OBJECTIVE: To assess whether application of a standard algorithm to hospitalizations in the prenatal and postpartum (42 days) periods increases identification of severe maternal morbidity (SMM) beyond analysis of only the delivery event.

METHODS: We performed a retrospective cohort study using data from the PELL (Pregnancy to Early Life Longitudinal) database, a Massachusetts population-based data system that links records from birth certificates to delivery hospital discharge records and nonbirth hospital records for all birthing individuals. We included deliveries from January 1, 2009, to December 31, 2018, distinguishing between International Classification of Diseases Ninth (ICD-9) and Tenth Revision (ICD-10) coding. We applied the modified Centers for Disease Control and Prevention algorithm for SMM used by the Alliance for Innovation on Maternal Health to hospitalizations across the antenatal period through 42 days postpartum. Morbidity was examined both with and without blood transfusion.

RESULTS: Overall, 594,056 deliveries were included in the analysis, and 3,947 deliveries met criteria for SMM at delivery without transfusion and 9,593 with transfusion for aggregate rates of 150.1 (95% CI 146.7-153.5) using ICD-9 codes and 196.6 (95% CI 189.5-203.7) using ICD-10 codes per 10,000 deliveries. Severe maternal morbidity at birth increased steadily across both ICD-9 and ICD-10 from 129.4 in 2009 (95% CI 126.2-132.6) using ICD-9 to 214.3 per 10,000 (95% CI 206.9-221.8) in 2018 using ICD-10. Adding prenatal and postpartum hospitalizations increased cases by 21.9% under both ICD-9 and ICD-10, resulting in a 2018 rate of 258.7 per 10,000 (95% CI 250.5-266.9). The largest increase in detected morbidity in the prenatal or postpartum time period was attributed to sepsis cases.

CONCLUSION: Inclusion of prenatal and postpartum hospitalizations in the identification of SMM resulted in increased ascertainment of morbid events. These results suggest a need to ensure surveillance of care quality activities beyond the birth event.

HIV-related stigma represents a potent risk factor for a range of poor health outcomes, including mental health symptoms, treatment non-adherence, and substance use. Understanding the role of HIV-related stigma in promoting healthcare outcomes is critical for vulnerable populations, such as pregnant women living with HIV, in contexts with continued high rates of HIV and associated stigma, such as sub-Saharan Africa. The current study examined a range of risk and protective factors for HIV-related stigma with 742 pregnant women (M age = 29.6 years) living with depression and HIV accessing prevention of mother-to-child transmission of HIV (PMTCT) services in Dar es Salaam, Tanzania. Risk factors included depressive symptoms, ART non-adherence, intimate partner violence, food insecurity, and alcohol problems. Protective factors included disclosure of HIV status, social support, an appreciative relationship with their partner, hope, and self-efficacy. Findings highlight key psychosocial and behavioral determinants of HIV-related stigma for pregnant women living with HIV in Tanzania, and can inform perinatal care programming and interventions to optimize mental health and adherence outcomes.