Evaluation of the Helping Babies Breathe programme in Migori County, Kenya Authors: Carly Schade Ratekin, Erick Auko, Meshack Malenya Wafula, Wycliffe Okoth Omwanda, Jane Wamae, Sandra A Mudhune, Ash Rogers, Julius Mbeya, Frederick Ochieng Ouma, Lillian Njoki, William Young, Joseph R Starnes Published: 2025 ABSTRACT AbstractIntroductionKenya’s 2021 neonatal mortality rate was 18.42 per 1000 live births, far exceeding the Sustainable Development Goal of 12 per 1000 live births. Helping Babies Breathe (HBB) was developed to equip birth attendants with necessary skills for neonatal resuscitation in low-resource settings. This study aims to evaluate the implementation and impact of HBB training in Migori County, Kenya.MethodsThis single-group pretest–posttest study was conducted at 26 healthcare centres in Migori County, Kenya. Preimplementation neonatal health data were collected between January 2022 and June 2023. HBB training was conducted in June 2023, and postimplementation neonatal health data were collected from then until January 2024. To evaluate providers’ retention of knowledge and skills, validated Knowledge Checks and Objective Structured Clinical Examination (OSCEs) were conducted immediately following training, at 4 months and at 9 months following training.ResultsThe proportion of neonates not breathing and not receiving bag and mask ventilation within 1 min of birth decreased significantly following HBB training (12.73% pretraining vs 6.90% post-training, p=0.01). However, in-facility neonatal mortality did not decrease significantly when a trained HBB provider was present (86.73% survival without a provider vs 87.92% with, p=0.64). Providers’ knowledge and hands-on skills deteriorated significantly within 9 months following initial training despite quarterly follow-up trainings (Knowledge Check score average 97.24% immediately post-training vs 92.67% at 9 months, p=0.01; OSCE B average 92.48% immediately post-training vs 87.10% at 9 months, p=0.04).ConclusionsHBB training significantly improved providers’ knowledge of HBB initially, but quarterly follow-up trainings were insufficient to sustain knowledge and resuscitation skills. No significant reduction in in-facility neonatal mortality was observed following implementation of HBB, although the study was not powered to detect this change. There was an increase in the appropriate use of bag and mask ventilation. As HBB training is expanded across Kenya, more frequent refresher trainings and provider engagement strategies are recommended to sustain providers’ knowledge and skills.Keywords: Low and Middle Income Countries, Neonatology, Mortality, Health Policy WHAT IS ALREADY KNOWN ON THIS TOPIC Implementation of the Helping Babies Breathe (HBB) programme is a strategy to reduce neonatal mortality in low-resource settings such as Migori County, Kenya. However, the evidence for mortality benefit is mixed, and the optimal timing of follow-up training has yet to be established. Implementation of the Helping Babies Breathe (HBB) programme is a strategy to reduce neonatal mortality in low-resource settings such as Migori County, Kenya. However, the evidence for mortality benefit is mixed, and the optimal timing of follow-up training has yet to be established. WHAT THIS STUDY ADDS HBB training was associated with an increase in the percentage of babies appropriately given bag and mask ventilation within 1 min of birth; however, having a trained HBB provider attend the birth was not associated with a reduction in in-facility neonatal mortality. HBB training initially improved providers’ knowledge of HBB, but quarterly follow-up trainings may be insufficient to sustain providers’ knowledge and resuscitation skills. HBB training was associated with an increase in the percentage of babies appropriately given bag and mask ventilation within 1 min of birth; however, having a trained HBB provider attend the birth was not associated with a reduction in in-facility neonatal mortality. HBB training initially improved providers’ knowledge of HBB, but quarterly follow-up trainings may be insufficient to sustain providers’ knowledge and resuscitation skills. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Given improvement in the percentage of neonates appropriately receiving key HBB interventions following training, we recommend expanding implementation of HBB; but with more frequent follow-up trainings so providers’ knowledge and skills are retained. Given improvement in the percentage of neonates appropriately receiving key HBB interventions following training, we recommend expanding implementation of HBB; but with more frequent follow-up trainings so providers’ knowledge and skills are retained. Introduction The neonatal period—defined by the United Nations (UN) as the first 28 days of life—remains the most vulnerable time for child survival.1 Worldwide, neonatal mortality accounted for 2.4 million deaths in 2020, or approximately 17 deaths per 1000 live births.2 3 However, the burden of neonatal mortality is not distributed evenly throughout the world. The African region has the highest neonatal mortality rate at 27 per 1000 live births, which lies in contrast to the World Bank High Income Economy country average of just 2.69 per 1000 live births.2 3 The UN Sustainable Development Goal (SDG) 3.2 aims to reduce every nation’s neonatal mortality rate to less than 12 per 1000 live births by 2030.4 Although global neonatal mortality has been declining since 1990, the decline in neonatal mortality has been outpaced by the decline in under-5 mortality, and the percentage of under-5 deaths accounted for by neonatal deaths has increased from 40% to 47% between 1990 and 2022.3 Without ‘rapid, urgent action’, 61 nations are estimated to fail to reduce their neonatal mortality rate to 12 per 1000 live births by 2030.3 A renewed focus on neonatal mortality is required if SDG 3.2 is to be met. In Kenya, the UN Inter-agency Group for Child Mortality estimated that the 2021 neonatal mortality rate was 18.42 per 1000 live births.2 This national average highlights the need for Kenya to continue working to reduce the neonatal mortality towards the SDG of 12 per 1000 live births,4 but it obscures the regional inequities within the country. The county of Marsabit has a neonatal mortality rate of just 4 per 1000 live births, while Migori county, Wajir county and Kirinyaga county each have a rate of 37 per 1000 live births.5 Though interventions to improve child health in Kenya have reduced under-5 mortality by 64% over the last 20 years (115 to 41 per 1000 live births), neonatal mortality has only been reduced by 36% (33 to 21 per 1000 live births) over the same time period.5 As of 2022, neonatal deaths account for 66% of under-1 deaths and 51% of under-5 deaths in Kenya.5 This parallels the worldwide trend of neonatal mortality improving less quickly than under-5 mortality and underscores the need for improved care during the neonatal period. Helping Babies Breathe (HBB) was developed by the American Academy of Pediatrics and launched in 2010 to equip birth attendants with necessary skills for neonatal resuscitation in low-resource settings.6 The curriculum emphasises providing adequate bag-mask ventilation within the first ‘Golden Minute’ after birth to neonates that are not breathing well at birth.7 8 This is key to neonatal survival given that 95% of newborns not breathing well at birth will start breathing independently if given adequate positive-pressure bag-mask ventilation within the first minute of life.9 The simplified visual decision tree, pedagogical focus on hands-on skills practice and provision of essential equipment following training has made HBB a revolutionary intervention.6 10 HBB was lauded as an innovative approach to reducing mortality in low-resource settings after a 2013 study in Tanzania demonstrated a 47% decline in 24-hour neonatal mortality after HBB was implemented.11 Two later systematic reviews found that HBB training was associated with improvements in 28-day neonatal mortality, intrapartum stillbirths, first-day neonatal mortality and first-week neonatal mortality.9 12 However, many studies have failed to demonstrate a reduction in mortality. A large 2020 systematic review of 94 articles published in Paediatrics found significant reduction in fresh stillbirth and 24-hour mortality rates with HBB training.13 However, they did not find any change in 7-day or 28-day mortality rates.13 They discussed the poor quality and availability of preintervention data as a barrier to true estimates of mortality reduction and recommended continuing to roll out the HBB programme despite inconsistent evidence for a mortality reduction.13 In sum, though there has been difficulty measuring mortality benefit after HBB training, it has been shown to be a cost-effective intervention that is likely to be an effective approach in reducing neonatal mortality in low-resource settings.14 15 Despite clear improvements in resuscitation skills after HBB training, performance using these skills has almost universally been shown to degrade with time.9 16 17 A well-studied solution to this is frequent refresher training for clinicians. Though the optimal frequency and structure of refresher training has not been established, many innovative solutions integrating quality improvement science, technology and peer mentorship have been implemented with wide success.18,23 An emphasis on high-frequency, low-duration refresher training was added to the second edition of the HBB training manual to assist birth attendants’ knowledge retention.8 Lwala Community Alliance (Lwala)—an organisation that works to build the capacity of rural communities in Kenya to advance their own comprehensive well-being—has been an advocate of skills-based training in conjunction with supportive supervision.24 Lwala began programming in the Rongo subcounty of Migori county and is working in partnership with the Kenya Ministry of Health (MoH) to expand programming throughout Migori county and to additional surrounding counties. Part of this expansion has included facilitating initial HBB training for healthcare workers and providing follow-up training. Approximately 4200 providers from 214 facilities in Migori county have been trained to date through this partnership. This pre–post study contributes to the array of pre-existing literature by evaluating the scale-up of HBB in 26 public and private delivery centres in Migori County, Kenya. The study aims to (1) evaluate for changes in the frequency of key HBB interventions provided to neonates not breathing at birth before and after HBB training, (2) identify if there is a change in in-facility neonatal mortality before vs after HBB training and (3) evaluate knowledge acquisition and retention among trained healthcare providers receiving quarterly follow-up training on HBB. We hypothesised that initial HBB training paired with quarterly follow-up trainings would increase the frequency of appropriate HBB interventions provided and sustain provider knowledge over time. The study was not powered to detect a change in neonatal mortality, but we hypothesised that improved provision of HBB interventions has the potential to improve neonatal outcomes. This is vital information as Kenya works to meet the 2030 SDGs for neonatal mortality by scaling up HBB training throughout Migori County and nationwide. Methods Setting This study was conducted at public and private healthcare centres located in Migori County, Kenya. Migori County is situated in western Kenya and has a population of 1.2 million people.25 The primary ethnic group in the region is Luo. The majority of residents live in rural areas, and the economy is primarily dependent on subsistence farming, mining and fishing. However, there are several urban centres within the county, with Migori town being the largest. Study design This study used a single-group pretest–posttest design. Preintervention neonatal health indicators and HBB training process outcomes were collected at healthcare centres between January 2022 and June 2023. Baseline provider knowledge and skills assessments were collected immediately prior to HBB training which occurred in June 2023. Postintervention neonatal health indicators and HBB training process outcomes were collected at healthcare centres between June 2023 and January 2024. Post-training provider knowledge and skills assessments were collected 4 months following initial HBB training in October 2023. Endline provider knowledge and skills assessments were collected 9 months post-training in March 2024. Intervention Study facilities identified one provider to coordinate data collection at their healthcare facility and attend the HBB training. At the start of the study, this provider was given a mobile phone and trained to input delivery data into REDCap.26 27 Study personnel conducted site visits quarterly, or more frequently as needed, to assist with data input and troubleshoot technologic issues. After all pretraining delivery data had been collected, all selected providers were brought to a central location for a 4-day HBB training. Training was conducted by the Master Trainer of Trainees who had previously been provided with the HBB training as well as training on the use of teaching tools. Providers included in the study were trained in accordance with the pre-established American Academy of Pediatrics HBB training curriculum.7 8 All trained providers were provided with basic HBB commodities, including a bag-valve-mask ventilator, masks size 0 and 1, Penguin suckers, fetal stethoscopes, action plan charts and a Neonatale for their respective facilities. High volume sites were provided with multiple commodities to accommodate their need, and commodities were replaced as needed during follow-up training sessions. At the end of the training, each provider was certified as a Trainer of Trainees and developed a work plan to train the remaining providers at their healthcare facilities to provide neonatal resuscitation in accordance with the HBB training and to submit delivery data through REDCap. In addition to the initial HBB training, Master Trainer of Trainees employed by the MoH conducted quarterly follow-up training with each healthcare facility. All providers present at the time of follow-up training participated. During these follow-up trainings, one of the six county-accredited Master Trainers would coach providers as they practised the steps of neonatal resuscitation using NeoNatalie Neonatal Simulators and the basic HBB commodities.28 These sessions ranged from 20 min to 1 hour depending on the number of participants. Throughout the post-training period, study personnel conducted site visits as needed for data quality checks and to troubleshoot technologic issues. Site selection This study was conducted at 26 healthcare facilities in Migori County, Kenya. Initially, 16 healthcare facilities were enrolled, but due to a low number of deliveries at these sites, an additional 10 sites were added. Sites were selected through convenience sampling by the MoH and Department of Child Health. Inclusion criteria included the ability to conduct deliveries, and exclusion criteria included prior HBB training at the facility. Within each selected healthcare facility, one healthcare provider was self-selected to participate in the HBB training and oversee data collection at their site. Providers were primarily nurses but also included clinical officers. All participants remained employed in clinical settings throughout the study, but exposure to deliveries varied based on the volume of deliveries at each facility. Providers were excluded if they did not regularly attend deliveries and if they had previously completed HBB training. Sampling and sample size All deliveries occurring at a study site during the study period where the neonate had an APGAR score less than seven or was not breathing at birth were included in the analysis. The study was powered to detect a change in the proportion of babies receiving appropriate HBB interventions before vs after HBB training. The length of time spent gathering pretraining and post-training delivery data was dependent on time required to reach our predefined sample size. Sample size was calculated based on previous data in Migori County reporting that, on average, 70 babies were born not breathing per month. There was a lack of data on the proportion of babies receiving appropriate resuscitative measures before implementation of the HBB programme, but previous programme data demonstrated that 76% of neonates received appropriate resuscitative interventions following HBB implementation. We calculated that we would be able to detect a 0.12 difference in the proportion of babies receiving appropriate HBB interventions before and after implementation of the HBB programme with approximately 80% power and a 5% significance level using data from 210 pretraining and 210 post-training births requiring resuscitation. However, given the uncertainty of the proportion of babies successfully resuscitated before implementation of the HBB programme, we used an inflation factor of 20% to yield a final calculated sample size of 250 pretraining births and 250 post-training births. This study was not powered to detect a change in in-facility mortality. Data collection Both pretraining and post-training delivery data were recorded by delivery attendants in a physical register at every healthcare facility and entered electronically into REDCap via a mobile phone provided to each healthcare facility. Data was stored in REDCap until the end of the study period when it was downloaded for analysis. The physical register was checked against data in REDCap at each study facility during site visits to ensure accuracy. Providers self-reported if key HBB tasks were completed within 1 min of delivery and were timed using existing tools to record time of birth. Provider knowledge and skills assessments were obtained using pre-established validated HBB assessments.7 29 The HBB Knowledge Check was used to evaluate providers’ knowledge of HBB. This assessment consists of 18 written multiple-choice questions and is included in the HBB training materials.7 These Knowledge Checks were administered just before HBB training, just after HBB training and 9 months following training. The standardised HBB Objective Structured Clinical Exam (OSCE) B was used to evaluate providers’ skills performing neonatal resuscitation. The OSCE B is an examination where participants simulate a neonatal resuscitation and are evaluated on their demonstration of 23 HBB skills.7 29 The OSCE B was administered directly following training, 4 months following training and 9 months following training. The OSCE A was not used for evaluation because the OSCE B covers all skills evaluated in OSCE A. Master Trainers of Trainers employed by the MoH administered the HBB assessments and were oriented on uniform assessment criteria. Each healthcare provider was evaluated concurrently by three Master Trainers of Trainers during the OSCEs and return demonstrations. Inter-rater agreement was strengthened through joint assessments and consensus discussions, thereby minimising subjectivity and enhancing reliability of the results. Data management and analysis At the end of the data collection period, all data were downloaded from REDCap and transferred into Stata V.17.0 for analysis.30 Data were manually inspected for errors, and erroneous entries were resolved by cross-checking the physical register. Pretraining versus post-training demographic characteristics of delivering mothers were analysed with unpaired t-tests. Paired t-tests were used to examine pretraining versus post-training provider knowledge and skills. Pretraining written Knowledge Checks were compared with Knowledge Checks administered directly following completion of HBB training to assess for a change in provider knowledge associated with HBB training. Additionally, Knowledge Checks administered directly following completion of training were compared with Knowledge Checks administered 9 months following HBB training to evaluate retention of knowledge following training. To evaluate providers’ skills over time, paired t-tests were run comparing providers’ OSCE B scores at 4 months post-training and 9 months post-training to that provider’s immediate post-training score. These analyses were also completed using an unpaired t-test which did not change the significance of results. All neonatal health data were analysed using unadjusted χ2 tests and adjusted logistic regression models. The logistic regression models were adjusted for birth weight, maternal age at delivery, number of antenatal clinic visits completed and the mother’s number of previous deliveries. Patient and public involvement The public was involved in the design and conduct of this research. Lwala Community Alliance organises community committees to launch their own holistic health initiatives and participate in the governance of community and primary levels of healthcare. These committees inform programming priorities, which in turn influence the priorities of this research. Results of this study will be shared with the Migori County Ministry of Health, in Lwala Community Alliance’s publicly available quarterly report, and with relevant community committees. Results Maternal demographics Delivery data from the birth of 557 neonates were included in this study. Prior to HBB training, 267 neonates were included, and 290 were included following HBB training. The mean birth weight, maternal age at delivery, and mother’s number of previous births were similar between pretraining and post-training groups (table 1). The mean number of antenatal clinic (ANC) visits attended prior to delivery was significantly higher in the post-training time period at 4.19 visits vs 3.90 visits in the pretraining time period. Table 1Maternal demographic characteristicsCovariatePre-HBB TrainingPost-HBB trainingTotalP valueMean birth weight at delivery3063.30 (41.92)2960.80 (40.35)3009.93 (29.12)0.08Mean maternal age at delivery24.42 (0.36)25.21 (0.40)24.83 (0.27)0.14Mean number of previous deliveries1.49 (0.12)1.60 (0.11)1.54 (0.08)0.51Mean number of ANC visits attended3.90 (0.09)4.19 (0.09)4.05 (0.06) 0.02 Values presented as mean (SD).Bolded values reached a significance level <0.05.ANC, antenatal clinic; HBB, Helping Babies Breathe. Maternal demographic characteristics Values presented as mean (SD). Bolded values reached a significance level <0.05. ANC, antenatal clinic; HBB, Helping Babies Breathe. Of the 26 facilities participating in the study, 21 recorded births during the study period (81%). Five facilities included in the study had no neonates with an APGAR score less than seven or born not breathing, and thus did not record any births during the study period. These five facilities had low birth rates during the study period, which reduced the likelihood of having a baby needing resuscitation. Capacity building data for providers from these five facilities was collected and analysed below. Healthcare provider knowledge and skills In June 2023, 26 healthcare providers participated in a 5-day HBB training as detailed above. Providers’ knowledge was assessed using the standardised written Knowledge Check.7 Providers’ HBB knowledge increased significantly from a mean score of 83.12% to 97.77% from immediately pre-training to immediately post-training (table 2). Due to staff attrition, only 21 of the original 26 providers trained were able to complete knowledge evaluations 9 months following training. Among these providers, Knowledge Check scores at 9 months following training decreased to 92.67% (p=0.01), meaning that their knowledge of HBB was not sustained. Table 2Paired t-test of provider knowledge and skills pre-HBB training versus post-HBB trainingTime point 1 per cent correctTime point 2 per cent correctP valuePre training vs immediate post-training Knowledge Check (n=26)83.12 (11.88)97.77 (3.63) 0.00 Immediate post-training vs 9-month post-training Knowledge Check (n=21)97.24 (3.86)92.67 (7.14) 0.01 Immediate post-training vs 4-month post-training OSCE B (n=24)93.21 (5.41)85.63 (9.92) 0.00 Immediate post-training vs 9-month post-training OSCE B (n=21)92.48 (5.32)87.10 (10.10) 0.04 Values presented as mean (SD).Bolded values reached a significance level <0.05.HBB, Helping Babies Breathe; OSCE, Objective Structured Clinical Examination. Paired t-test of provider knowledge and skills pre-HBB training versus post-HBB training Values presented as mean (SD). Bolded values reached a significance level <0.05. HBB, Helping Babies Breathe; OSCE, Objective Structured Clinical Examination. Providers’ skills performing HBB interventions were assessed using the standardised OSCE B.7 Of the 26 providers initially trained, 24 were able to be evaluated on skills 4 months post-training, and 21 were able to be evaluated on skills 9 months post-training. When comparing scores immediately following training to 4 months following training, and when comparing scores immediately following training to 9 months following training, mean scores decreased significantly, suggesting that follow-up training did not sustain providers’ HBB skills. Neonatal health indicators Several HBB process indicators were recorded to evaluate if completion of key HBB interventions changed from the pre-HBB training period to post-HBB training period. Both an unadjusted χ2 model and adjusted logistic regression model were used to analyse each process indicator. Adjusting the model for birth weight, maternal age at delivery, number of antenatal clinic (ANC) visits completed and mother’s number of previous deliveries did not change the significance of results for any of the process indicators examined. The percentage of deliveries that had an HBB trained provider at birth increased significantly from 24.34% (n=65) in the pretraining period to 54.48% (n=158) in the post-training period (table 3). Similarly, the percentage of neonates warmed at birth, suctioned at birth and provided bag and mask ventilation at birth all increased significantly from the pretraining period to the post-training period. The percentage of neonates dried at birth increased from 91.01% (n=243) to 100% (n=290) in the pretraining versus post-training periods. Of note, this change was significant using the unadjusted model, but it was not mathematically possible to use the adjusted logistic regression model to analyse this indicator because all neonates were dried at birth in the post-training period. The percentage of neonates stimulated at birth increased slightly from 89.89% (n=240) in the pretraining period to 93.79% (n=272) in the post-training period; however, this change did not reach statistical significance. Table 3Number of neonates receiving HBB interventions pre-HBB training versus post-HBB trainingPretraining number of neonates (%)Post-training number of neonates (%)OR95% CIAdjusted logistic regression model*P valueHaving an HBB trained provider at birth65 (24.34)158 (54.48)3.79(2.62 to 5.47) 0.00 Dried at birth243 (91.01)290 (100.00)–––Stimulated at birth240 (89.89)272 (93.79)1.66(0.88 to 3.13)0.11Warmed at birth239 (89.51)280 (96.55)3.24(1.53 to 6.86) 0.00 Suctioned at birth238 (89.14)280 (96.55)3.47(1.64 to 7.37) 0.00 Received bag and mask ventilation at birth169 (63.30)236 (81.38)2.72(1.83 to 4.04) 0.00 Percentage of infants not receiving bag and mask ventilation when not breathing at 1 min34 (12.73)20 (6.90)0.45(0.25 to 0.81) 0.01 Bolded values reached a significance level <0.05.*Adjusted for birth weight, maternal age at delivery, number of ANC visits completed and mother’s number of previous deliveries.ANC, antenatal clinic; HBB, Helping Babies Breathe. Number of neonates receiving HBB interventions pre-HBB training versus post-HBB training Bolded values reached a significance level <0.05. Adjusted for birth weight, maternal age at delivery, number of ANC visits completed and mother’s number of previous deliveries. ANC, antenatal clinic; HBB, Helping Babies Breathe. The percentage of neonates that were not breathing at 1 min postdelivery and not appropriately receiving bag and mask ventilation decreased significantly from 12.73% (n=34) in the pretraining period to 6.90% (n=20) in the post-training period. However, neonates delivered without a trained HBB provider throughout the study period were 9.09% more likely to receive bag and mask ventilation than those delivered with an HBB trained provider (n=255 without a trained provider, n=150 with a trained provider) (table 4). Table 4Number of neonates receiving bag and mask ventilation when delivered by an HBB trained provider versus delivered by a non-HBB trained providerNot having an HBB trained provider (%)Having an HBB trained provider (%)Adjusted logistic regression model*P value255 (76.35)150 (67.26) 0.03 Bolded values reached a significance level <0.05.*Adjusted for birth weight, maternal age at delivery, number of ANC visits completed and mother’s number of previous deliveries.ANC, antenatal clinic; HBB, Helping Babies Breathe. Number of neonates receiving bag and mask ventilation when delivered by an HBB trained provider versus delivered by a non-HBB trained provider Bolded values reached a significance level <0.05. Adjusted for birth weight, maternal age at delivery, number of ANC visits completed and mother’s number of previous deliveries. ANC, antenatal clinic; HBB, Helping Babies Breathe. Of the 553 births recorded, there were 47 neonatal deaths at study facilities. Neonates that required advanced care were referred to Migori County Referral Hospital for neonatal intensive care. There were 76 neonates transferred to a higher level of care, and of the 57 transferred neonates with recorded outcomes, there were 24 neonatal deaths. Mortality outcomes for these 57 transferred neonates were included in the overall mortality analysis. It was not possible to track outcomes for 19 of the 76 transferred neonates because the maternity register only captures the outcome of the primary facility and it was not possible to track those neonates across facilities (figure 1). Mortality outcomes of neonates included in the study. The only HBB process indicator significantly associated with in-facility neonatal mortality was receiving bag and mask ventilation. Neonates not receiving ventilation were 10.82% more likely to survive than those who did receive ventilation (95% CI (0.12 to 0.59)). Having an HBB trained provider attend the delivery was associated with a 13% higher odds of neonates living following delivery, but this change did not reach a significance level of 0.05 (95% CI (0.66 to 1.93)) (table 5). Moreover, for all neonates with a trained HBB provider at delivery, 43.96% more survived following HBB training versus prior to training. However, this difference did not reach statistical significance (p=0.87, 95% CI (0.33 to 2.54)) (table 6). The total percentage of neonates living remained essentially unchanged in the pretraining period versus the post-training period. There was an increase in the percentage of neonates living associated with neonates being dried, warmed and suctioned; however, none of these associations were statistically significant. Being stimulated at birth was associated with a 0.98% decrease in the percentage of neonates living following delivery; though again this change was not statistically significant (95% CI (0.36 to 2.57)). Neonatal survival was not significantly associated with a composite indicator of receiving drying, stimulation, suctioning or bag and mask ventilation. Table 5Number of neonates living with and without receiving HBB interventionsNumber living without indicator (%)Number living with indicator (%)OR95% CIAdjusted logistic regression model*P valueHaving an HBB trained provider at birth281 (86.73)182 (87.92)1.13(0.66 to 1.93)0.64Post-HBB training224 (89.24)239 (85.36)0.74(0.44 to 1.26)0.27Dried at birth19 (86.36)444 (87.23)1.15(0.33 to 4.01)0.83Stimulated at birth37 (88.10)426 (87.12)0.97(0.36 to 2.57)0.95Warmed at birth32 (86.49)431 (87.25)1.12(0.42 to 2.99)0.83Suctioned at birth30 (85.71)433 (87.30)1.17(0.43 to 3.18)0.76Received bag and mask ventilation at birth136 (95.10)327 (84.28)0.26(0.12 to 0.59) 0.00 Received drying, or stimulation, or warming, or suctioning, or bag and mask ventilation at birth--463 (87.19)0.43(0.07 to 2.49)0.34Bolded values reached a significance level <0.05.*Adjusted for birth weight, maternal age at delivery, number of ANC visits completed and mother’s number of previous deliveries.ANC, antenatal clinic; HBB, Helping Babies Breathe. Number of neonates living with and without receiving HBB interventions Bolded values reached a significance level <0.05. Adjusted for birth weight, maternal age at delivery, number of ANC visits completed and mother’s number of previous deliveries. ANC, antenatal clinic; HBB, Helping Babies Breathe. Table 6Number of neonates with an HBB trained provider at delivery living pre-HBB training versus post-HBB trainingNumber living pretraining (%)Number living post-training (%)OR95% CIAdjusted logistic regression model*P value51 (28.02)131 (71.98)0.92(0.33 to 2.54)0.87*Adjusted for birth weight, maternal age at delivery, number of ANC visits completed and mother’s number of previous deliveries.ANC, antenatal clinic; HBB, Helping Babies Breathe. Number of neonates with an HBB trained provider at delivery living pre-HBB training versus post-HBB training Adjusted for birth weight, maternal age at delivery, number of ANC visits completed and mother’s number of previous deliveries. ANC, antenatal clinic; HBB, Helping Babies Breathe. Discussion We report the successful implementation of HBB training at 26 facilities in Migori County, Kenya. Training was associated with an initial increase in providers’ knowledge and skills, although these improvements waned with time following training. Babies born in the postimplementation period were more likely to receive appropriate resuscitation, including bag and mask ventilation, when not breathing well. There was no observed change in in-facility mortality. We expect these findings to be broadly generalisable because the standardised HBB curriculum was used and the programme was implemented across 26 diverse, public and private, urban and rural health centres. Improving provider training HBB training was associated with a significant increase in providers’ Knowledge Check scores, and their OSCE B scores immediately post-training averaged over 90%. This suggests that HBB training in Migori County improved providers’ knowledge and skills as seen in previous studies.7 13 20 31 32 It has previously been established that a provider’s knowledge and skills decline over time without follow-up training.13 16 17 21 33 Previous literature suggests that short duration, high frequency follow-up training can sustain knowledge and skills obtained during training.1315,18 20 22 34 Following initial training, each study facility was visited quarterly by a Master Trainer of Trainees who conducted follow-up training. With these follow-up trainings, Knowledge Check scores declined significantly from 97.24% immediately post-training to 92.67% at 9 months post-training. The significant decline in providers’ Knowledge Check scores suggests that quarterly follow-up training was insufficient to sustain providers’ knowledge of HBB. Similarly, providers’ skills decreased significantly over time with the average OSCE B score immediately post-training being 92.48% and the 9 month post-training average being 87.10%. Data regarding OSCE B scores prior to training were not collected, so we are unable to determine if these scores remained increased over baseline. Although this suggests that quarterly follow-up training was insufficient to sustain providers’ skills at 92.48%, successful completion of the HBB simulation requires an OSCE B score of 73.91% which was reached by 19 of 21 providers tested at 9 months following training.7 This suggests that quarterly follow-up training may be sufficient to sustain skills above the minimum threshold. Tabangin et al’s 2018 study implementing HBB at a rural clinic and community hospital in Honduras found that providers undergoing monthly refresher trainings had a 2.9 times higher odds of passing the OSCE B on the first attempt than providers undergoing follow-up training at 3, 5 and 6 months post-training.17 KC et al’s 2017 study went beyond follow-up trainings and demonstrated that daily skills checks, preparing for resuscitation prior to every birth, using a self-evaluation after every delivery, and attending a weekly HBB review meeting was associated with a higher likelihood of retaining HBB skills 6 months following initial training.34 Given that neither providers’ knowledge nor skills were sustained with quarterly follow-up training 9 months following initial HBB training, intensifying provider follow-up training should be considered. The ideal frequency of follow-up training has not been established, but increasing training from quarterly to monthly is in alignment with Tabangin et al’s findings and may improve retention of providers’ HBB skills.17 The WHO’s Essential Newborn Care course recommends ongoing paired practice of skills, which may facilitate increased frequency of follow-up training without added cost of Master Trainers travelling to disparate facilities. Alternatively, adding additional interventions like KC et al’s daily skills checks, predelivery preparation and postdelivery self-evaluation may be more cost-effective and improve sustainment of providers’ skills.34 Structured monitoring of training and provider skill retention is recommended in future implementation given the lack of data on optimal follow-up training modalities and frequencies. Increasing HBB process measures Prior to HBB training, 12.73% (n=34) of neonates not breathing 1 min after birth were not receiving bag and mask ventilation, meaning that they were not receiving appropriate resuscitation. Following HBB training, only 6.90% (n=20) of neonates were not being appropriately resuscitated 1 min following birth. This significant increase in appropriate neonatal resuscitation following HBB training is congruent with many previous studies.1335,37 Interestingly, the percentage of neonates being given bag and mask ventilation was 76.35% (n=255) without a trained HBB provider, but only 67.26% (n=150) with a trained HBB provider. This reduction in bag and mask ventilation with HBB trained providers has been replicated by Kamath-Rayne et al and Mduma et al who suggest it is secondary to effective early steps in the HBB resuscitation pathway including drying, stimulation and suction.18 37 This study found that neonatal drying, warming and suctioning at birth were significantly increased following HBB training, which supports Kamath-Rayne et al and Mduma et al’s hypothesis. Taken together, these data suggest that HBB training improves providers’ early resuscitation skills and prepares them to provide bag and mask ventilation only when required. Reducing in-facility neonatal mortality This study did not demonstrate a statistically significant decrease in in-facility neonatal mortality associated with HBB training. Previous studies have conflicting evidence on the mortality benefit of HBB training. A 2020 systematic review of HBB training found several studies demonstrating a reduction in 24-hour neonatal mortality and fresh stillbirth.13 However, overall in-facility neonatal mortality has not been shown to decrease in other studies.13 Agudelo-Pérez et al’s meta-analysis found that implementation of HBB reduced intrapartum stillbirth, 24-hour neonatal mortality and first week neonatal mortality; but did not have an effect on late neonatal mortality.9 A change in in-facility neonatal mortality may have been detected in this study with a larger sample size, as this study was powered to detect an increase in appropriate resuscitation rather than a change in in-facility neonatal mortality. Additionally, change in in-facility neonatal mortality is potentially confounded by incomplete data for neonates that were transferred to higher levels of care. Of the 553 births recorded for this study, 19 were transferred to a different facility where mortality data was not available. These medically complex patients are most likely to benefit from HBB interventions, so the absence of this data may confound our results, but with such a limited number of unrecorded outcomes, the effect size is anticipated to be small. Of note, this study found that 95.10% of neonates lived without receiving bag and mask ventilation, while only 84.28% of neonates lived after receiving bag and mask ventilation. This significant increase in mortality potentially represents harm secondary to ventilation, but more likely reflects that more critically ill neonates requiring bag and mask ventilation are also at higher risk of mortality. Study limitations This pre–post study without a control group is limited because it cannot control for external changes in population health over time. It is possible that unmeasured changes in accessibility to healthcare, prenatal care provision and quality of neonatal care beyond HBB occurred during the study period and confound these results. Using a traditional control group was deemed unethical due to concern about limiting improvements in care. A stepped wedge design was not used due to concern for the increased burden of data collection, additional time required for recruitment and complexity of data analysis in addition to these ethical concerns. Additionally, this study is limited by its reliance on healthcare providers self-reporting delivery data. Providers were asked to self-report if they had completed each of the HBB interventions within 1 min of delivery, but due to disparate geography and infrequent deliveries at many centres these data were not able to be externally verified. Self-reporting may lead to an overreporting of completed interventions within 1 min of delivery, and future studies may be strengthened by including periodic direct observations. However, the potential overreporting in this study is expected to be present in the pretraining data collection period as well as the post-training data collection period and should not affect this pre–post analysis. Although providers were excluded from the study if they had previously completed HBB training, data are not available regarding trainees’ prior experience with deliveries and other resuscitation trainings. Non-random variations in providers’ skills and prior training may impact these results and were not able to be measured. Additionally, follow-up of providers’ knowledge and skills over 9 months was limited by staff attrition and transfer to other facilities. Of the 26 providers initially trained, 5 were not able to be administered a Knowledge Check or OSCE B 9 months following training. Improved follow-up over a longer time period would provide valuable information on sustainment of providers’ knowledge and skills following training. Staff transfer was also a challenge because a few previously HBB trained providers were transferred into study facilities during the pretraining delivery data collection period, and four of the study trained providers were transferred out of their health facility during the post-training data collection period. This may dilute our findings demonstrating an improvement in HBB process indicators following HBB training; however, this should make detecting a difference in the pre- vs post-training periods more difficult, and despite this dilution, we demonstrate a significant change in all but one of these process indicators. The percentage of neonates having an HBB trained provider at birth increased significantly from 24.34% (n=65) pretraining to 54.48% (n=158) post-training (p=0.00). This suggests that the implementation of HBB training contributed to more neonates having a trained provider at their delivery. However, following HBB training, 45.52% (n=132) of deliveries were still unattended by an HBB skilled provider. This may be secondary to high staff turnover, which is a frequent occurrence in Migori County. It may also indicate incomplete penetrance of the train-the-trainer method which led to incomplete coverage. Incomplete penetrance may contribute to a mortality benefit not being observed, and further study on this is warranted. This study was implemented throughout Migori County, and the large geographic area made more frequent follow-up training difficult. Increasing follow-up training to monthly or weekly occurrences may not be possible with the current administrative model, so different supervision structures that encourage more frequent follow-up training should be explored. Conclusions This study demonstrates three key findings: (1) Provider knowledge and skills are improved following initial HBB training but are not sustained after 9 months with quarterly follow-up training alone, (2) The percentage of neonates being appropriately resuscitated at 1 min following birth is increased following HBB training and (3) A change in in-facility neonatal mortality is not observed following HBB training, but the study was not powered to observe a change in mortality, and this outcome is likely confounded by several factors. As HBB training is scaled up throughout Kenya, we recommend an emphasis on more frequent follow-up training and additional incentives for providers to practise HBB skills on a regular basis. With this scale-up, we expect an improvement in the proportion of neonates being appropriately resuscitated at birth. This is likely to lead to improvement in in-facility neonatal mortality, though future research is needed to support this. Global partnership reflexivity statement This study assesses implementation of HBB as a strategy to reduce neonatal mortality, which is a priority of the Kenyan national government and Migori County local government. Local stakeholders from the Kenyan Ministry of Health and Kenyan researchers at Lwala Community Alliance designed this study with technical guidance from two US-based researchers. All research funding has come from Lwala Community Alliance to support local research teams. One US-based researcher receives compensation for part-time work, and the other volunteers their time. All research staff who conducted data collection are included as authors on this paper and have access to study data. Data analysis was discussed by all team members but completed by a US-based researcher. Data interpretation was conducted collaboratively with all team members during regular meetings. A US-based researcher completed an initial draft of the manuscript, but all research partners were involved in writing sections pertinent to their technical expertise and in editing. Data from this study will be shared with the MoH and Lwala collaborators so that key takeaways can be implemented. All Kenyan team members involved in data collection, analysis and manuscript writing are included as authors. This includes early career researchers and a mix of both male and female authors. Kenyan researchers were leaders in each phase of this research and received formal and informal training as necessary. This project has improved local infrastructure by providing all health centres with essential HBB commodities and strengthening the relationship between Lwala Community Alliance and the Migori County MoH. Neonates involved in the study were protected because no identifiable information was collected on them. Data on care providers were stored securely and deidentified for publication. Footnotes Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. Data availability free text: Deidentified participant data and statistical analysis plans are available on reasonable request from Carly Ratekin (ORCID ID 0009-0000-0512-9573). Patient consent for publication: Not applicable. Ethics approval: The protocol and study design were approved by the Ethics and Scientific Review Committee at Strathmore University (SU-IERC 1292/22) and the Institutional Review Board at Dartmouth College (STUDY00032621). A research licence was obtained from the Kenya National Commission for Science, Technology and Innovation (Licence Number 759407). All healthcare providers who participated in the study gave written informed consent. Consent was not required to collect delivery data for neonates included in this study as they were being provided with emergent treatment and no identifiable data were collected. Provenance and peer review: Not commissioned; externally peer reviewed. Patient and public involvement: Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details. Data availability statement Data are available on reasonable request. References Data Availability Statement Data are available on reasonable request.