Effect of a quality improvement package for intrapartum and immediate newborn care on fresh stillbirth and neonatal mortality among preterm and low-birthweight babies in Kenya and Uganda: a cluster-randomised facility-based trial

listen audio

Study Justification:
– Reductions in stillbirth and neonatal mortality have been slow in low-income and middle-income countries.
– Prematurity complications are a major driver of stillbirth and neonatal mortality.
– Evidence-based practices for intrapartum and immediate newborn care are often underutilized in Kenya and Uganda.
– The study aimed to assess the effect of a quality improvement package on stillbirth and preterm neonatal survival in these countries.
Study Highlights:
– The study was a cluster-randomized controlled trial conducted in Kenya and Uganda.
– Facilities providing 24-hour maternity care with at least 200 births per year were included.
– The intervention group received a quality improvement package in addition to maternity register data strengthening and a modified WHO Safe Childbirth Checklist.
– The control group received only maternity register data strengthening and the modified WHO Safe Childbirth Checklist.
– The primary outcome was fresh stillbirth and 28-day neonatal mortality.
– The intervention group had a significantly lower rate of stillbirth and neonatal mortality compared to the control group.
– No harm or adverse effects were found.
Recommendations for Lay Reader and Policy Maker:
– Implementing a quality improvement package for intrapartum and immediate newborn care can decrease fresh stillbirth and neonatal mortality among preterm and low-birthweight babies.
– Reinforcing evidence-based practices and investing in health system strengthening are crucial for improving outcomes.
– Policy makers should prioritize the adoption and implementation of the quality improvement package in healthcare facilities.
– Efforts should be made to ensure access to necessary equipment and supplies for quality care.
– Continuous monitoring and evaluation of outcomes should be conducted to assess the effectiveness of the interventions.
Key Role Players:
– Healthcare providers: Midwives, nurses, doctors, and other healthcare professionals involved in intrapartum and newborn care.
– Facility managers: Administrators responsible for overseeing the implementation of interventions and quality improvement measures.
– Policy makers: Government officials and policymakers responsible for making decisions regarding healthcare policies and resource allocation.
– Community health volunteers: Individuals who can assist in reaching out to mothers and providing support and education on intrapartum and newborn care.
Cost Items for Planning Recommendations:
– Training and mentoring programs for healthcare providers.
– Equipment and supplies for intrapartum and newborn care.
– Data strengthening and quality improvement initiatives.
– Monitoring and evaluation activities.
– Community engagement and education programs.
– Staffing and personnel costs for implementing and sustaining the interventions.
– Infrastructure improvements, if necessary, to support the quality improvement package.

The strength of evidence for this abstract is 8 out of 10.
The evidence in the abstract is strong, as it is based on a cluster-randomised controlled trial with a large sample size. The study design and methodology are clearly described, and the results show a significant reduction in fresh stillbirth and neonatal mortality. However, to improve the evidence, it would be beneficial to include more details on the specific interventions used in the quality improvement package and provide information on the statistical power calculations for the study.

Background: Although gains in newborn survival have been achieved in many low-income and middle-income countries, reductions in stillbirth and neonatal mortality have been slow. Prematurity complications are a major driver of stillbirth and neonatal mortality. We aimed to assess the effect of a quality improvement package for intrapartum and immediate newborn care on stillbirth and preterm neonatal survival in Kenya and Uganda, where evidence-based practices are often underutilised. Methods: This unblinded cluster-randomised controlled trial was done in western Kenya and eastern Uganda at facilities that provide 24-h maternity care with at least 200 births per year. The study assessed outcomes of low-birthweight and preterm babies. Eligible facilities were pair-matched and randomly assigned (1:1) into either the intervention group or the control group. All facilities received maternity register data strengthening and a modified WHO Safe Childbirth Checklist; facilities in the intervention group additionally received provider mentoring using PRONTO simulation and team training as well as quality improvement collaboratives. Liveborn or fresh stillborn babies who weighed between 1000 g and 2500 g, or less than 3000 g with a recorded gestational age of less than 37 weeks, were included in the analysis. We abstracted data from maternity registers for maternal and birth outcomes. Follow-up was done by phone or in person to identify the status of the infant at 28 days. The primary outcome was fresh stillbirth and 28-day neonatal mortality. This trial is registered with ClinicalTrials.gov, NCT03112018. Findings: Between Oct 1, 2016, and April 30, 2019, 20 facilities were randomly assigned to either the intervention group (n=10) or the control group (n=10). Among 5343 eligible babies in these facilities, we assessed outcomes of 2938 newborn and fresh stillborn babies (1447 in the intervention and 1491 in the control group). 347 (23%) of 1491 infants in the control group were stillborn or died in the neonatal period compared with 221 (15%) of 1447 infants in the intervention group at 28 days (odds ratio 0·66, 95% CI 0·54–0·81). No harm or adverse effects were found. Interpretation: Fresh stillbirth and neonatal mortality among low-birthweight and preterm babies can be decreased using a package of interventions that reinforces evidence-based practices and invests in health system strengthening. Funding: Bill & Melinda Gates Foundation.

An unblinded, pair-matched cluster-randomised controlled trial was implemented in public sector facilities in Uganda and public sector facilities in Kenya.20 All interventions were delivered at the facility level. The study was done in the Busoga region in eastern Uganda (with a population of 3 million) and in Migori County in western Kenya (with a population of 1 million).21 The regional Busoga and national neonatal mortality rates are similar (27 deaths per 1000 livebirths vs 28 deaths per 1000 livebirths, respectively), as are the stillbirth rates (17 stillbirths per 1000 pregnancies vs 16 stillbirths per 1000 pregnancies, respectively).22 The regional Migori County mortality rates are 19 deaths per 1000 livebirths and the national neonatal mortality rates are 22 deaths per 1000 livebirths; the regional Migori County stillbirth rates are 9·8 stillbirths per 1000 pregnancies and the national stillbirth rates are 13·2 stillbirths per 1000 pregnancies.23 The total fertility rate is 5·3 children per woman in Kenya and 6·1 children per woman in Uganda, with 24·3% of women in Kenya and 20·7% of women in Uganda having begun childbearing before the age of 19 years.22, 23 Women seeking care at government facilities in both regions predominantly live below the global poverty level. The trial gained ethical approval from the Kenya Medical Research Institute, Makerere University School of Public Health, and the University of California, San Francisco Institutional Review Boards. Mothers of eligible babies who were alive at discharge provided informed consent for 28-day follow-up. The trial is registered on ClinicalTrials.gov, {“type”:”clinical-trial”,”attrs”:{“text”:”NCT03112018″,”term_id”:”NCT03112018″}}NCT03112018. 23 rural and peri-urban facilities were assessed as potential clusters. Inclusion criteria were 24-h labour and delivery services, at least 200 births per year, and having a comparable facility for pairing in the same country. Tertiary referral facilities were excluded. One county referral hospital in Kenya, and one district referral and one regional referral hospital in Uganda were assessed but not included in the study matching or randomisation. These three hospitals did not have comparable hospitals with which they could be paired. The study clusters in Uganda were four district facilities (two public and two non-profit missionary facilities); all four facilities did caesarean sections and had a newborn care unit without capacity for continuous positive airway pressure and without an onsite paediatrician. Together, they had approximately 9000 deliveries per year (about 6% of the region’s births) between 2016 and 2018.21 In Kenya, the 16 study clusters included 14 public and two non-profit missionary facilities; only two did caesarean sections, none had functional designated newborn care units when the study began, and none had an onsite paediatrician. One control site added caesarean section capacity while the study was ongoing. The 16 Kenya study facilities had approximately 11 000 deliveries per year between 2016 and 2018, representing 23% of the county’s annual births.24 The study facilities in Uganda had between 1081 and 3142 deliveries per year with an average of two-to-three midwives covering each shift. The facilities in Kenya were smaller than those in Uganda, with a range of 310–1599 deliveries per year and an average of one-to-two midwives per shift. Deliveries were attended by one midwife with additional help called for when needed. The study did not add any additional clinical providers to the study sites during the course of the study. Intra-facility and inter-facility nursing staff rotations occur at regular intervals, approximately every 6 months in Kenya and approximately every 2 years in Uganda. First trimester ultrasound for pregnancy dating was not available at any of the 20 sites. Using indicators collected from maternity registers covering a 1-year pre-intervention period (June 1, 2015, to May 31, 2016), we applied a non-bipartite matching algorithm to match ten facility pairs on the basis of country, monthly deliveries, deliveries to staff ratio, stillbirth rate, low-birthweight rate, and pre-discharge neonatal death rate (appendix p 1). We also assessed facility readiness at the time of matching. The resultant pairs were reviewed with field teams and five of the ten pairs were re-matched on the basis of local knowledge of functional level and facility type. After pair matching was finalised, a study statistician based in the USA randomly assigned one of each pair to the intervention group using R software. No allocation concealment was possible given the nature of the intervention. Figure 1 describes the intervention elements, frequencies, and fidelities. The package aimed to strengthen provider skills and teamwork, emphasising uptake of evidence-based practices, which include but are not limited to the use of antenatal corticosteroids, immediate skin to skin, breastfeeding, newborn resuscitation, and preterm feeding. It also reinforced data use for clinical and administrative decision making, with a focus on data quality for accurate gestational age assessment and key quality improvement indicator tracking. The four component intervention package mSCC=modified WHO Safe Childbirth Checklist. *Including accurate gestational age assessment, use of magnesium sulphate and antenatal corticosteroids, immediate skin to skin and breastfeeding, newborn resuscitation, and preterm feeding guidelines. Before study initiation, sites were documented to have basic equipment and supplies, such as functional digital scales and neonatal bag valve masks for preterm and term newborn babies. If unavailable, PTBi-EA provided these supplies to standardise resource availability across all study sites, whereby these expenses did not exceed US$50 000 in either country. Introduction of data strengthening and the mSCC to all 20 clusters began before Sept 30, 2016, after which PRONTO and quality improvement collaboratives were added to the ten intervention facilities. The study obtained a waiver of consent to collect de-identified maternity register data. Birth outcome data were captured for all deliveries listed in maternity registers between Oct 1, 2016, and May 31, 2018, in Uganda and between Oct 1, 2016, and April 30, 2019, in Kenya. Pre-intervention data were also collected between March 1 and Sept 30, 2016, in Uganda, and between June 1 and Sept 30, 2016, in Kenya. In Kenya, the study was extended to compensate for a 5-month nurses’ strike and a 2-month doctors’ strike that greatly reduced delivery volume. The study team abstracted deidentified maternity register data on a monthly basis on all births to obtain overall denominators for facility rates. We excluded register entries for mothers admitted for antenatal complications or referred without delivery (ie, threatened preterm or false labour) and babies born outside the facility. Mothers with babies born alive weighing less than 2500 g or between 2500 g and 2999 g with a recorded gestational age of less than 37 weeks were approached to consent for follow-up to 28 days. In addition to capturing all babies with low birthweight, most of whom are either preterm or small for gestational age,25 we also estimated that the upper limit of 3000 g would capture 90–97% of infants younger than 34 weeks and 60–70% of infants younger than 37 weeks.20 We used this definition because measuring birthweight in addition to gestational age has greater reliability than estimating gestational age alone. Fresh stillborn babies meeting the same eligibility criteria were also included.26 Stillborn and liveborn babies weighing less than 1000 g were excluded from our primary analysis because they are considered previable in both countries, but were evaluated in secondary analyses. Infants not meeting these criteria or listed in the registers as abortions, macerated stillbirth, or without entries for birthweight and gestational age were excluded. Mothers of eligible live infants were approached for consent to be contacted by phone for 28-day follow-up. Trained facility providers (Kenya and Uganda) or community health volunteers (Kenya) obtained consent before the mother was discharged home. Providers or community health volunteers reviewed the maternity register daily to ascertain eligible deliveries. If the mother left the facility before she was able to consent, the study team used contact information provided in the facility records to reach her by telephone or in person through community health volunteers. Women agreeing to participate provided either written consent (Kenya and Uganda) or verbal consent if followed up retrospectively by telephone (Uganda). In Kenya, an additional 7-day follow-up call was made. Study staff made three attempts to reach a mother by phone before sending a study nurse (Uganda) or a community health volunteer (Kenya) to trace the mother at her home. Eligible stillbirths and pre-discharge deaths were included in analyses from register data, while mothers were not consented for the 28-day follow-up. Study staff entered data into an encrypted Open Data Kit database on tablets or laptops. Follow-up data were collected using paper forms, which was then delivered to each country’s central study office where the data were entered into an encrypted Django web-based database. Each mother and each infant was assigned unique separate identifiers that were linked. Unique identifiers were used to link 28-day outcomes with register data. The electronic data were maintained on secure systems with access limited to the study principal investigators, epidemiologists, and designated study staff. The primary outcome was the combined incidence of fresh stillbirth and 28-day neonatal mortality among eligible babies weighing between 1000 g and 2500 g, or between 2500 g and 2999 g with a recorded gestational age of less than 37 weeks. After publication of our protocol, we reworded our primary outcome to clarify that babies weighing 3000 g or above with a recorded gestational age of 37 weeks would be excluded. Babies under these cutoffs are included. Secondary outcomes included perinatal mortality (fresh stillbirth plus 7-day mortality), facility-based maternal mortality (captured in the maternity register), pre-discharge mortality (death of a liveborn baby before facility discharge, noted in the discharge status), and neonatal mortality for babies born alive and weighing less than 1000 g. Additional post-hoc analyses included caesarean section rates, individual components of the primary outcome (eg, fresh stillbirth and neonatal mortality), and fresh stillbirth and pre-discharge mortality for all registered births. As described in the protocol,20 with an estimated sample size of 4000 eligible births, an assumed intraclass correlation coefficient of 0·03, and a baseline incidence of fresh stillbirth plus neonatal mortality among eligible infants of 25%, the study would have 80% power at the 5% significance level to detect a 30% relative reduction in the primary outcome. We assumed that the intervention and control groups were balanced with respect to delivery volume. No effect sizes were hypothesised about secondary outcomes and the sample size was not adjusted for multiple comparisons. During the extended health worker strikes in Kenya, we reassessed our sample size. We accounted for the higher-than-expected birth volumes in Uganda, the need for a longer-than-expected data collection period in Kenya, data from the pre-intervention period estimate on primary outcomes, and a post-hoc one-tailed test on the primary outcome. Given a type I error of 0·05, power of 80%, a one-tailed test, and a balanced (1:1 for control and intervention group) sample, a sample size of 1133 preterm births was required per study group. The sample size was increased by 35% (to 1530 preterm births per study group) to account for clustering and loss to follow-up or missing information. Range and logic checks were applied to the maternity register and follow-up data using the mySQL data management and development software. Out-of-range data and data with discrepancies in eligibility or critical outcomes were sent to designated field staff to review and resolve. Biologically implausible or invalid data were recoded as missing before analysis. χ2 and t tests were used to compare the study groups by sociodemographic, reproductive health, and facility characteristics. An intention-to-treat analysis was done using logistic regression and generalised estimating equations with robust variance to account for clustering of births within facilities and to adjust for pairing of facilities. All primary and secondary outcomes, except neonatal mortality among babies weighing less than 1000 g, were assessed in the intention-to-treat population. Liveborn babies whose birthweight did not meet inclusion criteria (< 1000 g) were examined as a separate population for the secondary outcome only (because they fall outside viability definitions in Kenya and Uganda). We used an exchangeable correlation structure. At the facility level, we also estimated the difference in log odds of the outcomes within each matched pair and did a paired t-test on the average difference, weighted by the delivery volume. Although the study started with relatively balanced groups, there were only ten matched pairs, and adjustment for potential confounding factors was not possible at the facility level because the number of observations (n=10) was too small. Therefore, we did an individual-level analysis in which the unit of analysis was a delivery. We used a directed acyclic graph to identify potential confounders and mediators on which data were available (appendix pp 6–7). The main results were adjusted for pairing and clustering only. To examine robustness of the main results, we additionally adjusted for potential confounders and examined the effect of mediators. We also did sensitivity analyses to examine if the key results changed because of changes in conditions (eg, initiation of caesarean section at a single facility mid-study, but its paired facility did not have this capacity) or if any individual matched pair disproportionately influenced the overall results. Analogous analyses were done to assess secondary outcomes. Significance tests were two-tailed at the 5% level. To show that the results were not affected by multiple comparison, we corrected the p values using Bonferroni corrections. Analyses were done using SPSS version 23 and STATA version 15 (StataCorp). The trial is registered with ClinicalTrials.gov, {"type":"clinical-trial","attrs":{"text":"NCT03112018","term_id":"NCT03112018"}}NCT03112018. The funder of the study reviewed the study design, but had no role in data collection, data analysis, data interpretation, or writing of the report. National and community advisory boards provided input on intervention priorities. Health facility providers, managers, and local authorities were involved in implementation activities and influenced the focus and content of those activities on the basis of their roles and priorities. DW, PO, EB, CO, PM, RG, NLS, and PW had full access to all the data in the study. The corresponding author had final responsibility for the decision to submit for publication.

The study described is titled “Effect of a quality improvement package for intrapartum and immediate newborn care on fresh stillbirth and neonatal mortality among preterm and low-birthweight babies in Kenya and Uganda: a cluster-randomised facility-based trial.” The goal of the study was to assess the impact of a quality improvement package on stillbirth and preterm neonatal survival in Kenya and Uganda.

The study implemented several interventions to improve access to maternal health, including:

1. Provider mentoring using PRONTO simulation and team training: This intervention aimed to strengthen provider skills and teamwork, emphasizing the uptake of evidence-based practices for intrapartum and immediate newborn care.

2. Quality improvement collaboratives: Facilities in the intervention group participated in collaboratives to improve the quality of care provided. This involved data use for clinical and administrative decision-making, with a focus on data quality and key quality improvement indicator tracking.

3. Modified WHO Safe Childbirth Checklist: All facilities received a modified version of the WHO Safe Childbirth Checklist, which reinforced evidence-based practices for intrapartum and immediate newborn care.

These interventions were implemented at the facility level in public sector facilities in Kenya and Uganda. The study found that the quality improvement package significantly reduced fresh stillbirth and neonatal mortality among low-birthweight and preterm babies.

It is important to note that this study focused on a specific context and may not be directly applicable to other settings. However, the interventions implemented in this study can serve as potential recommendations for improving access to maternal health in other contexts.
AI Innovations Description
The recommendation from the study is to implement a quality improvement package for intrapartum and immediate newborn care in order to improve access to maternal health. This package includes interventions such as provider mentoring using PRONTO simulation and team training, as well as quality improvement collaboratives. By reinforcing evidence-based practices and investing in health system strengthening, the study found that fresh stillbirth and neonatal mortality among low-birthweight and preterm babies can be decreased. This recommendation can be used as an innovation to improve access to maternal health and reduce stillbirth and neonatal mortality rates in low-income and middle-income countries.
AI Innovations Methodology
The study described is a cluster-randomized controlled trial conducted in Kenya and Uganda to assess the effect of a quality improvement package for intrapartum and immediate newborn care on stillbirth and preterm neonatal survival. The intervention package included provider mentoring using PRONTO simulation and team training, as well as quality improvement collaboratives. The study aimed to reinforce evidence-based practices and invest in health system strengthening to improve outcomes for low-birthweight and preterm babies.

To simulate the impact of these recommendations on improving access to maternal health, a methodology was implemented. Here are the key steps of the methodology:

1. Selection of study clusters: Twenty facilities in Kenya and Uganda were randomly assigned to either the intervention group or the control group. Facilities were pair-matched based on various criteria, including country, monthly deliveries, deliveries to staff ratio, stillbirth rate, low-birthweight rate, and pre-discharge neonatal death rate.

2. Intervention implementation: The intervention group received the quality improvement package, which included provider mentoring using PRONTO simulation and team training, as well as quality improvement collaboratives. The control group received maternity register data strengthening and a modified WHO Safe Childbirth Checklist.

3. Data collection: Data on maternal and birth outcomes were abstracted from maternity registers for both the intervention and control groups. Follow-up was conducted by phone or in person to identify the status of the infant at 28 days.

4. Analysis: The primary outcome was fresh stillbirth and 28-day neonatal mortality. Logistic regression and generalized estimating equations were used to analyze the data, adjusting for clustering of births within facilities and pairing of facilities. Sensitivity analyses were conducted to examine the robustness of the results.

5. Results: The study found that the intervention group had a lower incidence of fresh stillbirth and neonatal mortality compared to the control group. No harm or adverse effects were found.

In summary, the methodology involved selecting study clusters, implementing the intervention, collecting data, analyzing the data using statistical methods, and reporting the results. This approach allowed for the evaluation of the impact of the recommendations on improving access to maternal health.

Share this:
Facebook
Twitter
LinkedIn
WhatsApp
Email