Effect of Maternal and Newborn Care Service Package on Perinatal and Newborn Mortality

Key Points Question Can packaged community-based interventions delivered through existing health care systems improve neonatal survival in resource-constrained regions? Findings This pragmatic, cluster randomized clinical trial in Pakistan included 15 615 births and an intervention package consisting of community mobilization through health education, provision of clean delivery kits to pregnant women, and training community- and facility-based health care professionals. Although the intervention showed no reduction in perinatal mortality, there was a statistically significant 25% reduction in neonatal mortality. Meaning This trial demonstrates a high level of acceptance for improved household practices around delivery and neonatal care by community members and supports the use and scale-up of community-based approaches to improve neonatal mortality in Pakistan.


Introduction
In 2020, the global neonatal mortality rate (NMR) stood at 17 deaths per 1000 live births, resulting in 2.4 million fatalities. 1 Despite extensive efforts to improve newborn survival, neonatal mortality remains a significant public health concern, contributing to nearly half of all deaths of children younger than the age of 5 years in 2020.Projections indicate that between 2020 and 2030, approximately 24 million newborn deaths are expected, with most concentrated in sub-Saharan Africa and South Asia. 2 At 40 deaths per 1000 live births, Pakistan has one of the highest NMRs in the world. 3Most newborn deaths are preventable through proper care during delivery and the postpartum period; however, barriers within health systems significantly impede the uptake of these essential services.
5][6][7][8] In Pakistan, 41% of women deliver at home in rural areas, and 37% of newborns are born without skilled birth attendants. 9Fewer than half the mothers in rural areas (49%) receive a postnatal visit by a skilled clinician, and only 52% of newborns are examined by a qualified clinician within 2 days of birth. 9Furthermore, nearly three-fourths (74.9%) of women in rural areas face at least 1 challenge in accessing health facilities, with reluctance to travel alone and distance to health facilities being the most commonly reported barriers. 91][12][13] At the same time, interventions encouraging mothers to seek care by educating them and their families on safe motherhood and facility births, 14,15 providing incentives, and improving referral systems can increase the uptake of antenatal and early postnatal care. 16,17This study aimed to determine if an integrated health services package that focused on demand creation within the community and delivery of lifesaving interventions through existing community-based health care professionals could improve neonatal mortality in rural Pakistan.

Study Design and Setting
enrollment and intervention delivery occurred between November 1, 2012, and December 31, 2013.
The predominantly rural Tehsil RYK has a population of 1.4 million people and is administratively divided into 40 union councils (UCs), each with a population of approximately 15 000 to 20 000 people and at least 1 basic health unit that provides primary care services to the UC.Each UC is also served by lady health workers (LHWs), community health workers employed by the government, who provide door-to-door primary, preventative, and curative care to men, women, and children, with a focus on reproductive, maternal, and child health, and act as liaisons between the community and health facilities.District RYK has a high NMR of 39 deaths per 1000 live births. 18The protocol for this study has been previously published (trial protocol in Supplement 1). 19The study was approved by the ethics review committee of Aga Khan University and the national bioethics committee of Pakistan.Verbal consent was obtained from respondents and heads of households before data collection.This study is reported according to the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.

Participants, Sample Size, Randomization, and Masking
Pregnant women, regardless of gestational age, residing in the study area, identified through an ongoing pregnancy surveillance system instituted in the study area and LHW records, were eligible to participate.A cluster was defined as an administrative UC.Assuming each cluster has a population of 15 000 to 20 000 people, a crude annual birth rate of 20 births per 1000 people, and an estimated perinatal mortality rate of 60 deaths per 1000 births, a coefficient of variation (κ) between clusters of 0.125 and an intracluster correlation coefficient of 0.05 were used.A total of 20 clusters were required to detect a difference of 20% in mortality rates between the intervention and control groups with 90% power.There were 10 clusters each in the intervention and control groups (eFigure 1 in Supplement 2).Blocked randomization of clusters was conducted after stratifying the clusters.As a result, 139 random allocations were identified that resulted in similar populations in the 2 groups (difference, <20 000), number of live births (difference, <1000), NMRs (difference, <10 deaths per 1000 live births), and proportions of women delivering in a hospital (difference, <5%).One scheme was selected from this list of balanced allocations using a computer-generated random number to allocate clusters to their respective groups.Due to the nature of the study (provision of intervention vs no intervention), blinding was not possible.However, to reduce bias in the study, independent data collection teams masked to cluster allocation were established to collect data on the outcomes.

Procedures
An evidence-based intervention package was developed comprising a maternal and newborn health pack, training for community health workers (including LHWs and community midwives), training for facility-based health care professionals, and community mobilization through health education.
The evidence taken into consideration for each component of the package is presented in eTable 2 in Supplement 2.
To assess the preintervention health indicators of the study area, tailor the health education component for the community, and develop training modules for clinicians, a comprehensive baseline survey was conducted.Sociodemographic characteristics, maternal mortality rates and NMRs within the past year, and health care practices during prior pregnancies, including the number of facility-based births, clean deliveries, and deliveries performed by skilled attendants, were obtained.Clean delivery was defined as clean hands, perineum, surface, cord cutting, tying, and avoiding introduction of unclean materials into the vagina. 20Nested within the baseline survey, a study on knowledge, attitudes, and practices explored the community's understanding, practices, and health-seeking behaviors related to maternal and newborn health and informed the development of the health education component.A health facility assessment was conducted to collect information on staffing, level of care provided, and availability of laboratory, blood bank, transportation facilities, labor and delivery room, postnatal wards, and nursery.These data informed the development of clinical training modules for the health care professionals tailored to the available services.These messages were also reiterated via text messages sent to pregnant women and their families throughout the pregnancy and postpartum period.

JAMA Network Open | Global Health
An emergency fund was also established with the help of LHWs using local resources and contributions from pregnant women's families.This fund was used to transport women at high risk or with other complications to health care facilities.

Study Teams
Twenty independent data collection teams were formed, 1 for each cluster, with each team comprising 2 enumerators.Enumerators were briefed on the study and its objectives and trained to identify eligible participants, obtain informed consent, conduct interviews, and accurately complete data collection tools.After training, all study instruments were pretested in a population outside the study area.
The data collection teams conducted quarterly pregnancy surveillance rounds and visited each household in both the intervention and control clusters to identify pregnant women.The teams were trained to collect information on in-and out-migrations, knowledge and practices related to maternal and newborn health, presence of LHWs at the delivery, and pregnancy outcomes including miscarriages, stillbirths, and neonatal mortality.These teams were blinded to cluster allocations to minimize bias.Data on neonatal and maternal outcomes were collected until the 28th day of life and 40th postpartum day, respectively.
Three independent study intervention teams were used to deliver the maternal and newborn health pack along with instructions on each component of the pack to the study participants in the intervention clusters during the third trimester.Each team consisted of 1 LHW and 2 community health workers.Participants who chose to deliver at home were advised to inform the LHW so that she could attend the birth and conduct timely postnatal visits.The intervention teams also maintained inventory records for the distribution and use of health packs.

Control Clusters
In the control clusters, the LHW program continued to function as usual.The LHWs conducted routine household visits while receiving regular refresher training according to the national LHW program.

JAMA Network Open | Global Health
Effect of Maternal and Newborn Care Service Package on Perinatal and Newborn Mortality

Outcomes
The primary outcome of the study was all-cause perinatal mortality, defined as the composite rate of stillbirth (pregnancy loss after 24 weeks per 1000 births per year) and early neonatal mortality (neonatal deaths in the first 7 days per 1000 live births per year).The secondary outcome was all-cause neonatal mortality, defined as neonatal deaths per 1000 live births per year.Late neonatal mortality was defined as deaths from day 8 to day 28 per 1000 live births per year.Additional exploratory outcomes included the rate of miscarriages (pregnancy loss before 24 weeks), community health practices, and postpartum complications.

Statistical Analysis
Statistical analysis was performed from January to May 2014.Before data entry, all forms were checked for completeness and consistency.Databases and entry screens were developed for data entry using Microsoft Visual FoxPro, version 9.0 (Microsoft Corp).Consistency checks and skips were conducted, and data were entered twice to minimize erroneous data entry.Special arrangements were made to enforce the referential integrity of the database.
The neonatal and perinatal mortality rates during the preceding year were established through data analysis.Descriptive analysis was used to describe the demographic characteristics, socioeconomic characteristics, antenatal care practices, and past pregnancies within the community.
The differences in mortalities were assessed using data collected through the baseline survey and surveillance rounds, and cluster-adjusted analysis was conducted to control for variability between clusters.
A modified intention-to-treat analysis was conducted using Stata, version 16 (StataCorp).For analysis of the primary outcome, due to the small number of clusters per group, we opted to follow the method as recommended by Hayes and Moulton. 22The perinatal mortality rate for each cluster was calculated, and the logarithm of the cluster-level perinatal mortality rate was used as the independent variable in a linear regression model to estimate the rate ratio associated with the intervention and its 95% CI while accounting for cluster randomization.To adjust the analysis for baseline (preintervention) values, the logarithm of the baseline cluster-level rate was included as a covariate.Furthermore, the regression was weighted based on the number of observations in each cluster.A similar approach was used to analyze the NMR.When analyzing reported practices, we used the svy commands in Stata to account for the clustered nature of the data.A Kaplan-Meier analysis was performed to determine and compare the survival of newborns in the intervention clusters with the survival of newborns in the control group.All P values were from 2-sided tests and results were deemed statistically significant at P < .05.

Results
Table  2).An analysis by surveillance round showed similar NMRs among the intervention and control clusters during the first round, with subsequent rounds showing a consistent decrease in NMRs among the intervention clusters (eFigure 2 in Supplement 2).The rates of miscarriages and stillbirths were similar between the intervention and control clusters (Table 2).
The numbers of facility births and home births attended by skilled birth attendants were similar among both the intervention and control clusters (

Discussion
This randomized clinical trial used an integrated community-based intervention package to address neonatal mortality in rural Pakistan.The trial did not demonstrate a significant decrease in stillbirth and perinatal mortality.There is evidence to suggest that fewer antenatal care visits and the absence of a skilled birth attendant are associated with stillbirth and perinatal mortality. 23,24Despite the inclusion of a community mobilization component in the present trial, there was no improvement in antenatal and intrapartum care seeking in the intervention group.No additional measures were taken to overcome barriers to accessing health care facilities aside from establishing an emergency  transportation system, which may be the reason for the failure of our intervention to increase antenatal and intrapartum health care service use and improve perinatal mortality.
However, this trial did show a reduction in the risk of early and overall neonatal mortality by 27% and 25%, respectively.These results are consistent with previous meta-analyses of trials assessing the effect of community-based interventions on neonatal mortality, demonstrating a 33% reduction in the risk of early neonatal mortality 25 and a 25% reduction in the risk of overall neonatal mortality. 25,26Given that the use of a CDK and chlorhexidine were the only significant improvements in postintervention health care practices, it can be postulated that these were the primary mechanisms for the reduction in neonatal mortality observed in this study.Another important reason for the decrease in neonatal mortality may be the capacity building of health care professionals in the intervention clusters.8][29][30] A meta-analysis reported that standardized formal training in newborn resuscitation in low-and middle-income countries can reduce the risk of neonatal mortality by 15%. 31It is challenging to attribute the results of the present trial to any specific mechanism due to the multicomponent nature of the study and the possibility of a change in health behaviors that were not captured in this study.
This effectiveness trial offers several crucial insights that should be considered in future scale-up efforts and provides guidance for further research.First, the use of chlorhexidine and CDKs was increased significantly in the intervention clusters.4][35] The results of our trial indicate that even in regions with high NMRs, detrimental household health practices, and low levels of care seeking, these interventions can be delivered through existing community health worker programs and improve neonatal survival.
Second, it was possible to closely engage the LHW program for the delivery of the intervention.
All LHW sessions were integrated into their regular training, and the LHW supervisors and district health officials consistently monitored LHW activities during the trial.However, the coverage of some components of the intervention remained low, including the presence of LHWs at delivery and postnatal visits within the first 2 days after birth.The operational challenges faced by the LHWs during implementation included difficulty reconciling additional tasks with routine responsibilities, delays in receiving birth notifications, and challenges traveling alone or at night, which demonstrate the limitations of the community outreach program.Last, while there was significant acceptance and uptake of the components of the intervention package at the household level, elements of the intervention package aiming to increase health facility use during pregnancy and delivery did not appear to be as effective, representing an area for further research.

Limitations
This study has some limitations.There was a possibility of contamination of health messages and information between intervention and control clusters due to the exchange of information between families.To minimize contamination, clusters were defined using existing administrative boundaries, and training sessions were conducted with health care professionals from facilities only in the intervention UCs.Most of the population living in a particular UC was likely to receive care from LHWs and facilities within that UC.The study could not be blinded because of the nature of the intervention.However, separate teams were used for data collection in the intervention and control clusters to minimize the risk of performance bias.All collected data were self-reported, which may have introduced reporting bias due to underreporting of poor pregnancy and newborn outcomes and overreporting of positive newborn care practices among intervention clusters.Moreover, even though data collectors were extensively trained to obtain accurate data, self-reporting may have led to misclassification in early and late neonatal mortality.
Effect of Maternal and Newborn Care Service Package on Perinatal and Newborn Mortality JAMA Network Open.2024;7(2):e2356609.doi:10.1001/jamanetworkopen.2023.56609(Reprinted) February 19, 2024 3/12 Downloaded from jamanetwork.comby guest on 02/27/2024 Intervention Package Maternal and Newborn Health Pack | The health pack contained a 4% chlorhexidine solution, sunflower oil emollient, and a clean delivery kit (CDK) with a soap bar, disposable gloves, a clean plastic sheet, a single-use razor blade, a cord clamp, and sterile thread.Pictorial brochures in the local language provided instructions on CDK use, key health messages on birth preparedness, maternal and newborn danger signs, acute obstetric and newborn emergencies, and immediate newborn care.Enhanced Training | A detailed description of the training provided to each group of health care workers is presented in eTable 1 in Supplement 2. In brief, LHWs and community midwives in the intervention clusters received refresher training on recognizing complications and danger signs among mothers and neoates, providing prompt referrals, and the importance and use of CDKs.Clinicians in primary and secondary health care facilities received refresher training on basic and comprehensive emergency obstetric and neonatal care. 21Community Mobilization | To foster maternal and newborn health awareness within the community, LHWs facilitated informational sessions at the household and community levels for pregnant women and their families in the intervention clusters.Participants learned about antenatal nutrition, vaccination, pregnancy complications, maternal and neonatal danger signs, and essential and immediate newborn care.Pregnant women were encouraged to seek antenatal and postnatal care and to deliver in a health care facility or with a skilled birth attendant in case of a home birth.

Figure 2 .
Figure 2. Kaplan-Meier Survival Estimates for Neonatal Mortality 1 presents the overall summary of the demographic characteristics and pregnancy outcomes Effect of Maternal and Newborn Care Service Package on Perinatal and Newborn Mortality perinatal and neonatal mortality, were self-reported in quarterly household surveillance visits.The perinatal mortality rate was 67.4 per 1000 live births in the intervention clusters and 79.5 per 1000 live births in the control clusters (rate ratio, 0.86; 95% CI, 0.69-1.08;P=.19)(Table2).The NMR was lower among the intervention clusters vs the control clusters (39.2 per 1000 live births vs 52.2 per 1000 live births; rate ratio, 0.75; 95% CI, 0.58-0.95;P = .02).Early NMR was significantly lower among the intervention clusters than the control clusters (32.2 per 1000 conducted 46.5% of births (7561 of 16 250).Clean delivery practices were followed in only 7.2% of home deliveries (594 of 8247).All baseline indicators were similar across the intervention and control groups.A total of 7943 and 7509 outcomes were captured in the intervention and control clusters, respectively (Figure1).During the intervention, pregnancy outcomes, including miscarriages,JAMA Network Open | Global HealthJAMA Network Open.2024;7(2):e2356609.doi:10.1001/jamanetworkopen.2023.56609(Reprinted) February 19, 2024 5/12 Downloaded from jamanetwork.comby guest on 02/27/2024 stillbirths, and

Table 3 )
. Lady health workers were more likely to conduct their first postnatal visit within 48 hours in the intervention group than in the control group (risk ratio, 2.20; 95% CI, 1.42-3.42).Clean delivery practices were more common in the intervention group than in the control group (63.2% [2284 of 3616] vs 13.2% [455 of 3458]; risk ratio, 4.80; 95% CI, 3.71-6.22).An analysis by surveillance rounds showed that by the fourth round, 90.2% of

Table 2 .
Summary Outcomes a All parameter estimates, 95% CIs, and P values were estimated by (weighted) analysis of variance at the cluster level; dependent variable = log (rate); log (baseline neonatal mortality rate) included as a covariate; and the weights used were based on the number of events reported in each cluster.

Table 3 .
Maternal, Neonatal, and Child Health Practices Before, During, and After Delivery From Surveillance Rounds 1 to 4

SUPPLEMENT 2. eFigure 1.
Map of Tehsil RYK eTable 1.Evidence for Each Component of the Intervention Package eTable 2. Enhanced Training for Healthcare Providers eFigure 2. Neonatal Mortality Rate by Arm and Surveillance Round eTable 3. Postpartum Complications eReferences.