Effectiveness of Neonatal Resuscitation Training Programs, Implementation, and Scale-Up in Low- and Middle-Income Countries

Abstract

Introduction

To describe recent evidence regarding the most effective neonatal resuscitation training program and scale-up of these programs in low- and middle-income countries (LMICs), which has contributed to the upcoming Lancet Global Newborn Care Series 2025, and forms part of a supplement describing an extensive synthesis on effective newborn interventions in LMICs.

Methods

We included relevant studies from Medline, Embase, CINAHL, Cochrane CENTRAL and Global Index Medicus databases on the effectiveness and scale-up of Neonatal Resuscitation Training Programs (NRTP), with searches run August 2022. Data extraction and quality assessments were completed independently and in duplicate.

Results

A total of 93 unique records met the eligibility criteria and were included in our analyses across the reviews. NRTPs improved most knowledge and skill-based outcomes but impact on mortality varied. Included studies identified knowledge and skill retention, standardized training protocols, and limited training opportunities for health care providers as challenges to current NRTPs.

Conclusion

Reported knowledge, skills, and mortality outcomes were similar across NRTPs. The Helping Babies Breathe (HBB) program was found to be cost-effective in Tanzania, suggesting that the HBB program or elements thereof are low-cost and scalable in LMICs. Future research across diverse settings should evaluate the cost-effectiveness of other NRTPs. To scale-up current NRTPs, programs should focus on improving long-term retention outcomes and improving training material accessibility.

Introduction

In 2020, 2.4 million newborns died globally [1], illustrating the vulnerability newborns experience within the first 28 days of life. While the rate of under-5 deaths has decreased since 1990, neonatal mortality has failed to decrease at the same rate and accounts for 47% of under-5 deaths [1]. Intrapartum-related complications, such as birth asphyxia or the inability to breathe at birth, account for one of the leading causes of these global newborn deaths. Specifically, approximately 5–10% of newborns require assistance at birth to begin breathing [2–7] and approximately 25% of all neonatal deaths are caused by asphyxia [8]. The majority of deaths caused by asphyxia occur in low- and middle-income countries (LMICs), where access to obstetric care is limited [3, 7]. For example, Sub-Saharan Africa, and Central and Southern Asia account for the highest neonatal mortality rates globally at 43% (27 deaths per 1,000 live births) and 36% (23 deaths per 1,000 live births), respectively [1]. In comparison, high-income countries have an average mortality rate of three deaths per 1,000 live births [9].

To improve newborn survival, neonatal resuscitation (NR) has been found to be effective in preventing a large proportion of these deaths, reducing mortality due to birth asphyxia by 30% [2–4, 8, 10]. A neonatal resuscitation training program (NRTP) involves a set of interventions delivered at birth to support the initiation of breathing [2]. Basic resuscitation involves airway clearing (i.e., suctioning), head positioning, positive-pressure ventilation (i.e., bag-and-mask ventilation), stimulation, and drying and is effective for most neonates with secondary apnoea [2, 3, 11]. Advanced resuscitation is necessary for approximately 1% of births and involves chest compression, intubation, or medications [5]. In instances where advanced resuscitation is needed, the baby will often require ongoing support and intensive care.

NRTPs generally involve multiple health providers with extensive training and access to advanced technology. As such, the currently accepted standard of NRTP took decades to become a reality for high-income countries and continues to be a challenge to implement for low-resource countries [12, 13]. To improve the situation, numerous educational programs have been designed to train health care providers in NR. The standardized American Academy of Pediatrics (AAP) Neonatal Resuscitation Program (NRP) was the first such programme developed in 1987 in the USA. The NRP is responsible for standardizing the approach to resuscitation through training that provides both the knowledge and the skills required for resuscitation [12, 14]. Globally used, the programme is taught in over 140 countries and is based upon the best currently available evidence [12]. However, as the rates of neonatal mortality in LMICs continued to soar despite the widespread use of the NRP, an alternative standardized approach was deemed necessary to improve implementation of learned skills within clinical practice, specifically in low-resource settings.

To address the need for a NRTP that was tailored to low-resource settings, the AAP developed the 2010 Helping Babies Breathe (HBB) program [12, 13]. The approval of the HBB program sparked at a time when the International Liaison Committee on Resuscitation (ILCOR) changed its recommendation of using 100% oxygen for NR to room air since 100% oxygen during resuscitation was proven to cause oxidative stress, leading to increased risk of mortality or neurological damage. This was an important shift, as supplying 100% oxygen in NR care settings would have been a barrier to the scale-up and implementation of the HBB program in LMICs. HBB emphasizes knowledge building with pictures and low-cost inflatable mannequins and is focused on establishing ventilation in newborns within the first 60 s of life [13, 15]. However, evidence on the effectiveness of the HBB program in reducing neonatal mortality has been mixed [16], and it has been well documented that skills and knowledge retention after training tend to decrease with time [13]. As such, there is a need to determine which NRTPs are the most effective in LMICs using more recent evidence regarding neonatal mortality and morbidity, in addition to outcomes related to retention of knowledge and skills by birth attendants.

Consonant with determining the most effective and cost-effective NRTPs for LMIC settings, the challenge to achieving impact becomes scaling-up training to provide universal coverage for basic resuscitation in these resource-limited settings. There has been a transition in place of births and hence location of resuscitation needs globally. While the majority of births in LMICs previously occurred in home settings, approximately 80% of births now occur in facility-based settings [1]. Previous studies have reported that reductions in mortality outcomes are more common in secondary care facilities, compared to lower level facilities, as they are better equipped to provide more advanced care and have more staff per bed [17]. Therefore, to achieve universal coverage, a better understanding of the strengths and challenges of current NRTPs is needed to inform scale-up across low-resource settings.

Objective

The purpose of this paper was to synthesize the current evidence regarding NRTPs and scale-up of these programs, which has contributed to the upcoming Lancet Global Newborn Care Series 2025, and forms part of a supplement describing an extensive synthesis of the evidence on effective newborn interventions in LMICs. This manuscript will contribute to a forthcoming Lancet Global Newborn Care Series 2025s key message and strategies regarding effectiveness of interventions for asphyxia management, which will help progress goals in reducing neonatal mortality.

The following two review questions were assessed:

• 1.What are the most effective NRTPs for improving mortality and morbidity outcomes in newborns, and knowledge and skill retention in health care providers in LMICs?
• 2.What are effective implementation strategies for the roll-out and scale-up of NRTPs in LMICs?

Methods

For topics regarding NRTP effectiveness and scale-up, we were unable to identify any recent systematic reviews of relevance, thus new systematic reviews were conducted. Herein, we provide a high-level summary of the review methodology per review topic; however, a detailed summary of all methods can be referenced here [18].

Search Strategy

NRTP effectiveness and scale-up studies were searched in Medline, Embase, CINAHL, Cochrane CENTRAL, and Global Index Medicus databases in August 2022. Additional studies were found and added through snowballing and grey literature searches. There were no limitations on geographical setting or publication language for both reviews. All search strategies can be found in supplementary material A (for all online suppl. material, see https://doi.org/10.1159/000542539).

Search Selection

This review was limited to studies conducted in LMICs only. The target population was primarily focused on newborns; however, we were also interested in understanding the impact of NRTPs on healthcare professionals’ knowledge and skill acquisition. The first research question was strictly focused on effectiveness and therefore only included randomized controlled trials (RCTs) or quasi-experimental studies, whereas the NRTP scale-up question included a broader scope of study designs to ensure the capture of implementation and cost-analysis research. Studies reporting on non-NR specific training programs where the effect of NR training outcomes could not be disaggregated from outcomes associated with broad, integrated packages on maternal and child health, were excluded. See Table 1 for inclusion criteria per topic.

Table 1.

Inclusion criteria per included review

Topic	Participant	Intervention	Comparator	Primary outcomes	Study design
1. NRTP Effectiveness	Health care professionals and newborns in LMICs	Any NR training programa	Any other NR program, basic newborn care, or no resuscitation training	Neonatal mortality and morbidity (HIE, seizures, hypothermia, meconium aspiration syndrome, neurodevelopmental outcomes)	Trials (randomized or quasi-experimental allocation)
				Change in healthcare professional behaviour; decrease in resuscitation time; knowledge and/or skill acquisition and retention
2. NRTP Scale-up	Any individual, organization or community involved in the delivery or receipt of NR that was the target of the implementation or scale-up of an NRTP in LMICs	Any component of a strategy used to implement or scale-up a NRTP	No restrictions	Any outcome, including measures associated with the implementation or scaling-up process	Trials
				Provider or patient reported outcomes regarding the effect of the NRTP	Implementation studies
					Program evaluations
					Observational studies
					Qualitative studies

NRTP, neonatal resuscitation training program; LMICs, low- and middle-income countries; HIE, hypoxic-ischaemic encephalopathy.

^aStudies reporting on non-NR specific training programs where NR training outcomes could not be disaggregated were excluded (i.e., Emergency Obstetric and Early Newborn Care [EmONC], Essential Care for Every Newborn [ECEB], and Advanced Life Support in Obstetrics [ALSO]).

Statistical Analyses

For the topic of NRTP effectiveness, we performed standard meta-analyses to generate a summary risk ratio (RR) and 95% confidence interval (CI) for each dichotomous outcome. For continuous outcomes, a mean difference was calculated if studies used the same scale to assess the outcome or a standardized mean difference (SMD) if studies used different scales to assess the outcome and 95% CI. Due to heterogeneity across studies, a random-effects model was used for all pooled estimates. The inverse variance method was used as the default for all analyses, except when dichotomous data were sparse in terms of low event rate or small study size, in which case we used the Mantel-Haenszel method. When a study had multiple arms that received different interventions of interest, they were both included in the meta-analyses as separate intervention-control comparisons. Findings from non-randomized controlled trials, controlled pre-post studies, and interrupted time-series designs were analysed and reported separately. Studies that randomized participants at the cluster level were adjusted for clustering by reducing the size of the trial to its effective sample size or by inflating the standard error. We did not make these adjustments if the study authors reported they had adjusted for clustering already. Studies reporting a standard error were converted to standard deviation using Cochrane methods [19]. Knowledge, skill, and self-efficacy outcomes were pooled across studies by converting numerical scales and mean values to percent values for analysis. Self-efficacy was defined as a healthcare provider’s level of confidence in their ability to effectively resuscitate a newborn according to NRTP guidance. Very early neonatal and perinatal mortality were defined as death within 24 h, whereas early neonatal and perinatal mortality were defined as death after 24 h to 7 days. Additional, non-time dependent mortality outcomes included stillbirths, fresh stillbirths, macerated stillbirths, perinatal mortality, mortality due to asphyxia, and all-cause neonatal mortality. Sensitivity analysis by quality assessment was considered, such that any study assessed as high risk using the ROB tool 2.0, serious risk using the ROBINS-I tool, or poor quality using the NIH tool or CASP tool were excluded in the effect estimate. Subgroup analyses were conducted for primary outcomes according to a priori defined sources of potential clinical and methodological heterogeneity for NRTP effectiveness: study design (i.e., pre-post data from non-randomized one-arm trial and pre-post data from one or more arms of a randomized trial); World Health Organization (WHO) region (i.e., Africa, the Americas, South-East Asia, Europe, Eastern-Mediterranean, and Western Pacific); simulation (i.e., simulation and no simulation); standardization (i.e., standardized and no standardization); duration of program (e.g., 2-day training program); and type of provider (i.e., nurses, midwives, birth attendants, physicians, and healthcare students). Given the shift from home-based to facility-based births, a growing interest in the impact of NRTPs at different levels of care has emerged. Thus, subgroup analyses were conducted by level of care (i.e., community, primary, secondary, and tertiary facility settings). Given the plethora of data, we report all pooled estimates for totals and subgroup analyses by level of care for three comparisons assessing NRTP effectiveness (i.e., Any NRTP vs. no NRTP; HBB vs. no HBB; NRP vs. no NRP) in the main text while all other subgroup analyses and comparisons can be found in online supplementary material B.

For the topic of NRTP scale-up review, data from included studies were inappropriate for meta-analysis. Therefore, results were narratively summarized and tabulated. See the methods paper for further detail on data extraction, methodological quality assessment, synthesis, and analysis methods [18].

Results

Study Inclusion

Both reviews generated a total of 2,224 records. After removing duplicates, 1,684 records were screened, of which 358 records were included for full-text analysis. Of these, 65 records met the eligibility criteria for NRTP effectiveness and 39 records for NRTP scale-up. These totals include 11 overlapping studies which met the eligibility criteria for NRTP effectiveness and scale-up. Therefore, only 93 unique records were included in our analyses across both reviews (Fig. 1).

Fig. 1.

PRISMA flow diagram, NRTP effectiveness and scale-up. *Studies reporting on non-NR specific training programs where NR training outcomes could not be disaggregated were excluded (i.e., Emergency Obstetric and Early Newborn Care [EmOC], Essential Care for Every Newborn [ECEB], Advanced Life Support in Obstetrics [ALSO]).

Study Characteristics of Included Studies

The 93 unique studies included 50 quasi-experimental studies, 18 randomized controlled studies, 11 observational and cross-sectional studies, seven mixed methods studies, three qualitative studies, two implementation studies, and two cost-analysis studies (Table 2). Of the 93 included studies, five studies were implemented at a community-level, 12 studies were implemented at a primary-level, 29 studies were implemented at a secondary-level, 19 studies were implemented at a tertiary-level, and 28 studies were implemented at a mixed-level. Of the 65 included studies for NRTP effectiveness, 29 studies assessed the HBB program compared to no training, 26 studies assessed the NRP compared to no training, and 10 assessed other training programs (see online suppl. material C for a list of all included comparisons; see online suppl. material D for all standardized resuscitation training programs identified within the included studies). The publication year of the included studies ranged from 2001 to 2023. A detailed COIS table can be found for each included study in online supplementary material E. Global maps illustrating the number of studies that implemented NRTP(s) per LMIC can be found in online supplementary material F.

Table 2.

Summary of study characteristics

Topic	Number of studies	Study design	Type of population	Type of intervention	Level of implementation	Publication date range
NRTP effectiveness	65a	18 RCT	Health care professionals and newborns in LMIC	29 HBB versus no training	3 community	2001–2023
		47 quasi-experimental		26 NRP versus no training	9 primary
				10 other training versus no training	17 secondary
					17 tertiary
					19 mixed
NRTP scale-up	39a	2 RCT	Any individual, organization, or community involved in the delivery or receipt of NR	All scale-up or implementation	3 community	2009–2022
		2 cost analysis			4 primary
		2 implementation study			16 secondary
		3 qualitative			3 tertiary
		7 mixed methods			13 mixed
		11 observational cohort or cross-sectional
		12 quasi-experimental

NRTP, neonatal resuscitation training program; RCT, randomized controlled trial.

^a11 overlapping studies between NRTP scale-up and effectiveness topics.

Methodological Quality of Included Studies

Of the overall 18 RCTs included in this review, one (5.5%) was rated as having a high risk of bias, 10 (55.6%) were rated as having some risk of bias, and seven (38.9%) were rated as having a low risk of bias. Of 50 quasi-experimental included studies, 30 (60%) were rated as having a serious concern, 19 (38%) were rated as having a moderate concern, and one (2%) study was rated as having a low concern. Of 13 observational cohort and cross-sectional included studies, seven (54%) were rated as poor quality, four (31%) were rated as fair quality, and two (15%) were rated as good quality. Of the three qualitative studies included, one (33%) was rated as poor quality, one (33%) was rated as fair quality, and one (33%) was rated as good quality. All nine mixed methods studies were of high quality. See online supplementary material G for a detailed quality assessment rating for each included study.

NRTP Effectiveness

Comparison 1: Post-NR Training Program versus Pre-NR Training Program

Knowledge and Skill-Based Outcomes

Post-NRTP, participants experienced improvements in mean knowledge test scores, more healthcare providers passed knowledge tests with a score of 75% or higher, and improved self-efficacy compared to pre-NRTP training. However, healthcare providers had borderline significantly higher mean knowledge post-test scores immediately following NR training compared to 3 months later (p = 0.05). Further, NRTPs also improved both mean composite practical skills and specialized skills (i.e., bag-mask ventilation skills, objective structured clinical examination (OSCE) A and OSCE B skills); however, providers had significantly lower practical skill scores 3- and 6-months later compared to immediately after training (Table 3). OSCE A and B refer to validated knowledge check questions and simulation tests available in NRTP educational materials. Sensitivity analyses by study quality assessment revealed no significant changes in effect estimates for knowledge- and skill-based outcomes compared to analyses with all included studies (online suppl. material B Table B1).

Table 3.

Pooled effect estimates for any NR training with pre-post data and any NR training versus no NR training with intervention-control data, totals, and by level of care

Outcome	Group	# Studies (# participants)	Effect estimate (95% CI)	Heterogeneity (I2)	Test for subgroup differences
Pre-post data
Mean knowledge test score	Total	21 (1,778a)	SMD 1.75 [1.36, 2.15]	96%	–
Passed knowledge test (achieved score of ≥75%)	Total	8 (2,069)	RR 1.48 [1.24, 1.77]	93%	–
Mean knowledge retention test scores, 3-month follow-up	Total	3 (264)	SMD −1.34 [−2.65, −0.03]	98%	–
Mean knowledge retention test scores, 6-month follow-up	Total	3 (370)	SMD −0.55 [−2.03, 0.94]	99%	–
Passed knowledge test refresher (achieved score of ≥75%)	Total	3 (1,090)	RR 1.19 [0.95, 1.49]	94%	–
Mean practical skills test score (composite)	Total	5 (271a)	SMD 4.39 [3.02, 5.76]	95%	–
Passed skills test score (composite) (achieved score of ≥75%)	Total	2 (157)	RR 9.09 [0.12, 663.64]	95%	–
Mean practical skills retention scores (composite), 3-month follow-up	Total	2 (224)	SMD −0.91 [−1.80, −0.02]	95%	–
Mean practical skills retention scores (composite), 6-month follow-up	Total	2 (233)	SMD −1.12 [−1.99, −0.25]	95%	–
Mean practical bag-mask ventilation skills test score	Total	3 (217a)	SMD 5.22 [2.36, 8.09]	98%	–
Mean OSCE A skills test score	Total	4 (324a)	SMD 3.02 [1.19, 4.86]	98%	–
Mean OSCE B skills test score	Total	5 (406a)	SMD 4.94 [1.98, 7.90]	99%	–
Passed OSCE B skills test score refresher (achieved score of ≥75%)	Total	2 (749)	RR 1.24 [1.12, 1.36]	15%	–
Mean self-efficacy score	Total	2 (268)	SMD 1.36 [1.17, 1.55]	0%	–
Stillbirths	Total	6 (144,380)	RR 0.79 [0.63, 1.00]	84%	–
Stillbirthsb	Primary facility	1 (902)	RR 0.95 [0.34, 2.64]	–	p = 0.02
	Secondary facility	2 (29,955)	RR 0.54 [0.22, 1.28]	93%
	Tertiary facility	1 (24,977)	RR 1.18 [0.98, 1.42]	–
	Primary and secondary facility	1 (9,598)	RR 0.77 [0.60, 0.98]	–
	Primary, secondary, and tertiary facility	1 (78,948)	RR 0.86 [0.78, 0.95]	–
Fresh stillbirths	Total	9 (594,549)	RR 0.89 [0.79, 1.02]	76%	–
Fresh stillbirthsb	Secondary facility	5 (524,412)	RR 0.98 [0.85, 1.13]	77%	p = 0.002
	Tertiary facility	1 (24,977)	RR 0.93 [0.67, 1.29]	–
	Primary and secondary facility	1 (9,598)	RR 0.54 [0.38, 0.78]	–
	Primary, secondary, and tertiary facility	1 (31,217)	RR 0.76 [0.61, 0.94]	–
	Community	1 (4,345)	RR 0.31 [0.14, 0.69]	–
Macerated stillbirths	Total	4 (48,628)	RR 1.33 [1.04, 1.68]	36%	–
Macerated stillbirthsb	Secondary facility	1 (9,708)	RR 1.35 [1.06, 1.73]	–	p = 0.20
	Tertiary facility	1 (24,977)	RR 1.61 [1.26, 2.04]	–
	Primary and secondary facility	1 (9,598)	RR 1.06 [0.75, 1.50]	–
	Community	1 (4,345)	RR 0.75 [0.22, 2.56]	–
Perinatal mortality	Total	6 (369,854)	RR 0.68 [0.52, 0.88]	94%	–
Perinatal mortalityb	Secondary facility	3 (234,750)	RR 0.54 [0.39, 0.74]	87%	p < 0.00001
	Tertiary facility	1 (24,977)	RR 1.60 [1.37, 1.87]	–
	Primary, secondary, and tertiary facility	2 (110,127)	RR 0.87 [0.80, 0.94]	11%
Very early perinatal mortality	Total	4 (313,698)	RR 0.59 [0.46, 0.76]	90%	–
Very early perinatal mortalityb	Secondary facility	3 (234,750)	RR 0.55 [0.40, 0.75]	86%	p = 0.01
	Primary, secondary, and tertiary facility	1 (78,948)	RR 0.84 [0.77, 0.92]	–
Very early neonatal mortality	Total	7 (275,006)	RR 0.62 [0.46, 0.83]	86%	–
Very early neonatal mortalityb	Secondary facility	4 (124,801)	RR 0.51 [0.33, 0.78]	74%	p = 0.01
	Tertiary facility	1 (2,228)	RR 0.29 [0.09, 0.87]	–
	Primary, secondary, and tertiary facility	2 (147,977)	RR 0.88 [0.74, 1.05]	63%
Early neonatal mortality	Total	3 (94,423)	RR 0.73 [0.43, 1.24]	83%	–
Early neonatal mortalityb	Secondary facility	1 (20,186)	RR 0.87 [0.58, 1.31]	–	p = 0.003
	Primary, secondary, and tertiary facility	1 (69,892)	RR 1.05 [0.92, 1.19]	–
	Community	1 (4,345)	RR 0.31 [0.15, 0.63]	–
Mortality due to asphyxia	Total	3 (35,898)	RR 0.65 [0.24, 1.71]	88%	–
Mortality due to asphyxiab	Primary facility	1 (887)	RR 1.83 [1.00, 3.37]	–	p = 0.05
	Tertiary facility	2 (35,011)	RR 0.36 [0.08, 1.57]	87%
All-cause neonatal mortality	Total	3 (119,655)	RR 0.68 [0.40, 1.15]	91%	–
All-cause neonatal mortalityb	Secondary facility	1 (86,624)	RR 0.53 [0.43, 0.66]	–	p < 0.0001
	Tertiary facility	1 (32,783)	RR 0.94 [0.82, 1.07]	–
	Primary and secondary facility	1 (248)	RR 0.27 [0.03, 2.39]	–
Neonates successfully resuscitated	Total	4 (27,795)	RR 1.34 [0.60, 3.03]	99%	–
Neonates successfully resuscitatedb	Secondary facility	2 (15,969)	RR 1.37 [0.90, 2.08]	95%	p < 0.00001
	Tertiary facility	1 (2,228)	RR 249.89 [15.56, 4,012.49]	–
	Primary and secondary facility	1 (9,598)	RR 0.41 [0.38, 0.45]	–
Intervention-control data
Mean knowledge test score	Total	2 (140a)	SMD 0.65 [0.31, 0.99]	0%	–
Mean self-efficacy score	Total	2 (180a)	SMD 0.31 [0.01, 0.60]	0%	–
Stillbirths	Total	2 (1,490a)	RR 1.04 [0.50, 2.19]	0%	–
Stillbirthsa	Primary and secondary facility	1 (37a)	Not estimablec	–	Not applicable
	Community	1 (1,453)	RR 1.04 [0.50, 2.19]	–
Mortality due to asphyxia	Total	3 (3,931a)	RR 0.25 [0.11, 0.59]	0%	–
Mortality due to asphyxiab	Secondary facility	1 (159a)	Not estimablec	–	p = 0.48
	Tertiary facility	1 (2,342)	RR 0.19 [0.06, 0.62]	–
	Community	1 (1,430a)	RR 0.34 [0.11, 1.11]	–
All-cause neonatal mortality	Total	2 (3,417a)	RR 0.58 [0.39, 0.85]	0%	–
All-cause neonatal mortalityb	Primary and secondary facility	1 (37a)	RR 2.09 [0.09, 48.04]	–	p = 0.42
	Community	1 (3,380a)	RR 0.57 [0.38, 0.83]	–

Bolded values represent significant effect estimates and subgroup findings defined as p < 0.05.

#, number; CI, confidence interval; I², heterogeneity; SMD, standardized mean difference; RR, risk ratio.

^aAdjusted for clustering.

^bNewborn outcome assessed by the subgroup level of care.

^c0 events were found for experimental and control after adjusting for clustering.

Subgroup analyses by study design, WHO region, level of care (Table 3), duration of program, and type of healthcare provider revealed very heterogeneous results for knowledge- and skill-based outcomes (see online suppl. material B Table B1 for results of all remaining subgroup analyses) for which no conclusions could be drawn. The test for subgroup differences by standardization and simulation did not reach statistical significance.

Perinatal and Newborn Outcomes

Findings were discordant across newborn outcomes. For example, post-NRTP revealed a borderline reduction in risk (p = 0.05) for stillbirths when compared to pre-NRTP. However, when disaggregated by fresh or macerated stillbirths, there was a 33% (95% CI: 4–68%) increased risk for macerated stillbirths, while fresh stillbirths had no significant risk reduction following the NRTP. This could represent improved reporting and disaggregation of stillbirths. NRTP led to a significant reduction in perinatal mortality (RR = 68%, 95% CI: 52–88%), but no impact on all-cause mortality, mortality due to asphyxia or neonates successfully resuscitated. However, when disaggregated by time, a significant risk reduction was found for very early perinatal mortality (RR = 59%, 95% CI: 46–76%) and very early neonatal mortality (RR = 62%, 95% CI: 46–83%) for post-NRTP compared to pre-training (Table 3). Due to the number of forest plots, only those with robust mortality outcomes (i.e., with a substantial number of participants) can be found in online supplementary material H. Sensitivity analyses by quality assessment revealed a 31% (95% CI: 5–50%) reduction in stillbirths, while revealing a nonsignificant effect estimates for macerated stillbirths. All other newborn outcomes had effect estimates with similar significance and effect direction compared to analyses with all included studies (i.e., all pooled studies with no sensitivity analysis) (online suppl. material B Table B1).

Subgroup analyses by study design, WHO region, level of care (Table 3), duration of program, and type of healthcare provider varied widely by subgroup (see online suppl. material B Table B1 for results of all remaining subgroup analyses) for which no conclusions could be drawn. The test for subgroup differences by standardization did not reach statistical significance.

Comparison 2: Any NR Training Program Intervention versus Control without NR Training Program

When comparing intervention-control data, the intervention group which underwent any NRTP had improved mean knowledge test scores and self-efficacy compared to control. Further, the intervention group had a 75% (95% CI: 41–89%) reduced risk for mortality due to asphyxia and a 42% (95% CI: 15–61%) reduced risk of all-cause neonatal mortality, when compared to control (see Table 3). No sensitivity analysis by quality assessment was possible for any NRTP versus no NRTP with intervention and control data. This is because all included studies were either low risk and included in the analysis or high risk, and therefore no studies would have been included in the analysis. The test for subgroup differences by study design, WHO region, duration of program, type of healthcare provider and standardization did not reach statistical significance (online suppl. material B Table B1).

Comparison 3: Post-HBB Training versus Pre-HBB Training

Knowledge- and Skill-Based Outcomes

When looking specifically at HBB training, post-HBB training was associated with improved mean knowledge test scores, and more healthcare providers passed knowledge tests with a score of 75% or higher compared to pre-training. Further, skill-based outcomes (i.e., mean composite practical skills, bag-mask ventilation skills, OSCE A, and OSCE B skills) also improved post-HBB training compared to pre-training (Table 4). Sensitivity analyses by quality assessment revealed no significant changes in effect estimates for knowledge- and skill-based outcomes compared to analyses including all studies.

Table 4.

Pooled effect estimates for HBB training with pre-post data, totals, and by level of care

Outcome	Group	# Studies (# participants)	Effect estimate (95% CI)	Heterogeneity (I2)	Test for subgroup differences
Pre-post data
Mean knowledge test score	Total	13 (1,250)	SMD 1.28 [0.93, 1.62]	92%	–
Passed knowledge test (achieved score of ≥75%)	Total	5 (1,881)	RR 1.50 [1.26, 1.77]	93%	–
Mean knowledge retention test scores, 3-month follow-up	Total	2 (158)	SMD −1.17 [−3.28, 0.95]	98%	–
Passed knowledge test refresher (achieved score of ≥75%)	Total	3 (1,090)	RR 1.19 [0.95, 1.49]	94%	–
Mean practical skills test (composite)	Total	3 (151a)	SMD 5.42 [3.37, 7.48]	96%	–
Mean practical bag-mask ventilation skills test	Total	3 (217a)	SMD 5.22 [2.36, 8.09]	98%	–
Mean OSCE A skills test	Total	4 (324a)	SMD 3.02 [1.19, 4.86]	98%	–
Mean OSCE A retention skills test, 6-month follow-up	Total	2 (137)	MD −0.80 [−2.35, 0.75]	0%	–
Mean OSCE B skills test	Total	5 (406a)	SMD 4.94 [1.98, 7.90]	99%	–
Passed OSCE B test refresher (achieved score of ≥75%)	Total	2 (749)	RR 1.24 [1.12, 1.36]	15%	–
Stillbirths	Total	6 (144,380)	RR 0.79 [0.63, 1.00]	84%	–
Stillbirthsb	Primary facility	1 (902)	RR 0.95 [0.34, 2.64]	–	p = 0.02
	Secondary facility	2 (29,955)	RR 0.54 [0.22, 1.28]	93%
	Tertiary facility	1 (24,977)	RR 1.18 [0.98, 1.42]	–
	Primary and secondary facility	1 (9,598)	RR 0.77 [0.60, 0.98]	–
	Primary, secondary, and tertiary facility	1 (78,948)	RR 0.86 [0.78, 0.95]	–
Fresh stillbirths	Total	7 (186,655)	RR 0.75 [0.64, 0.87]	44%	–
Fresh stillbirthsb	Secondary facility	3 (116,518)	RR 0.78 [0.69, 0.90]	0%	p = 0.04
	Tertiary facility	1 (24,977)	RR 0.93 [0.67, 1.29]	–
	Primary and secondary facility	1 (9,598)	RR 0.54 [0.38, 0.78]	–
	Primary, secondary, and tertiary facility	1 (31,217)	RR 0.76 [0.61, 0.94]	–
	Community	1 (4,345)	RR 0.31 [0.14, 0.69]	–
Macerated stillbirths	Total	4 (48,628)	RR 1.33 [1.04, 1.68]	36%	–
Macerated stillbirthsb	Secondary	1 (9,708)	RR 1.35 [0.93, 1.95]	–	p = 0.20
	Tertiary	1 (24,977)	RR 1.61 [1.26, 2.04]	–
	Primary and secondary facility	1 (9,598)	RR 1.06 [0.75, 1.50]	–
	Community	1 (4,345)	RR 0.75 [0.22, 2.56]	–
Perinatal mortality	Total	5 (231,497)	RR 0.77 [0.55, 1.08]	96%	–
Perinatal mortalityb	Secondary facility	2 (96,393)	RR 0.44 [0.19, 1.03]	94%	p < 0.00001
	Tertiary facility	1 (24,977)	RR 1.60 [1.37, 1.87]	–
	Primary, secondary, and tertiary facility	2 (110,127)	RR 0.87 [0.80, 0.94]	11%
Very early perinatal mortality	Total	3 (175,341)	RR 0.57 [0.40, 0.83]	94%	–
Very early perinatal mortalityb	Secondary facility	2 (96,393)	RR 0.44 [0.19, 1.03]	94%	p = 0.14
	Primary, secondary, and tertiary facility	1 (78,948)	RR 0.84 [0.77, 0.92]	–
Very early neonatal mortality	Total	7 (275,006)	RR 0.62 [0.46, 0.83]	86%	–
Very early neonatal mortalityb	Secondary facility	4 (124,801)	RR 0.51 [0.33, 0.78]	74%	p = 0.01
	Tertiary facility	1 (2,228)	RR 0.29 [0.09, 0.87]	–
	Primary, secondary, and tertiary facility	2 (147,977)	RR 0.88 [0.74, 1.05]	63%
Early neonatal mortality	Total	3 (94,423)	RR 0.73 [0.43, 1.24]	83%	–
Early neonatal mortalityb	Secondary facility	1 (20,186)	RR 0.87 [0.58, 1.31]	–	p = 0.003
	Primary, secondary, and tertiary facility	1 (69,892)	RR 1.05 [0.92, 1.19]	–
	Community	1 (4,345)	RR 0.31 [0.15, 0.63]	–
Mortality due to asphyxia (livebirths)	Total	2 (3,115)	RR 0.55 [0.05, 6.23]	94%	–
Mortality due to asphyxia (livebirths)b	Primary facility	1 (887)	RR 1.83 [1.00, 3.37]	–	p < 0.0001
	Tertiary facility	1 (2,228)	RR 0.15 [0.05, 0.44]	–
Neonates successfully resuscitated	Total	4 (27,795)	RR 1.34 [0.60, 3.03]	99%	–
Neonates successfully resuscitatedb	Secondary facility	2 (15,969)	RR 1.37 [0.90, 2.08]	95%	p < 0.00001
	Tertiary facility	1 (2,228)	RR 249.89 [15.56, 4,012.49]	–
	Primary and secondary facility	1 (9,598)	RR 0.41 [0.38, 0.45]	–

Bolded values represent significant effect estimates and subgroup findings defined as p < 0.05.

#, number; CI, confidence interval; I², heterogeneity; SMD, standardized mean difference; RR, risk ratio.

^aAdjusted for clustering.

^bNewborn outcome assessed by the subgroup level of care.

The test for subgroup differences by study design did not reach statistical significance for any knowledge- or skill-based outcomes, while subgroups by WHO region and duration of program were significant for some knowledge and skill outcomes but had no impact on others (online suppl. material B Table B2). Although the test for subgroup differences by type of healthcare provider was significant across all knowledge and skill-based outcomes, the impact varied by outcome for which type of provider led to the greatest improvements (online suppl. material B Table B2).

Perinatal and Newborn Outcomes

Results for post versus pre-HBB training, for the most part, remain the same as for comparison 1 looking at any NRTP given only two studies compared pre-post data for non-HBB interventions on newborn outcomes, while the remaining were all HBB studies. We report here a difference in results for newborn outcomes after removing the non-HBB studies, where post-HBB training reduced the risk of fresh stillbirths (RR = 75%, 95% CI: 64–87%), perinatal mortality (RR = 77%, 95% CI: 55–108%), and very early perinatal mortality (RR = 57%, 95% CI: 40–83%) when compared to pre-HBB training (Table 4). See online supplementary material H for forest plots on all robust mortality outcomes. Sensitivity analyses revealed a significant reduction in stillbirths (RR = 66%, 95% CI: 48–92%) and significant improvement in neonates successfully resuscitated (RR = 1.76%, 95% CI: 0.74–4.20%) while revealing non-significant effect estimates for fresh stillbirths, macerated stillbirths, and very early perinatal mortality. All other newborn outcomes had effect estimates with similar significance and effect direction when compared to analysis with all included studies (i.e., all pooled studies with no sensitivity analysis).

Again, most findings from subgroup analyses (i.e., WHO region, level of care, duration of program, and type of healthcare provider) varied widely by subgroup and thus did not allow for robust conclusions (online suppl. material B Table B2). Studies with randomized intervention-control data for newborn and healthcare provider outcomes on HBB training versus control could not be pooled due to insufficient number of studies [20, 21].

Comparison 4: Post-NRP Training versus Pre-NRP Training

Knowledge- and Skill-Based Outcomes

Post-NRP training significantly improved mean knowledge and practical composite skills compared to pre-NRP training. Consistent with other training programs, healthcare providers had a higher mean practical composite skill score immediately after training compared to 6 months later (Table 5). Sensitivity analyses by quality assessment revealed no significant changes in effect estimates for knowledge- and skill-based outcomes compared to analysis with all included studies.

Table 5.

Pooled effect estimates for NRP training with pre-post data and NRP training versus no NRP training with intervention-control data, totals, and by level of care

Outcome	Group	# Studies (# participants)	Effect estimate (95% CI)	Heterogeneity (I2)	Test for subgroup differences
Pre-post data
Mean knowledge test score	Total	7 (389)	SMD 2.71 [1.79, 3.63]	94%	–
Passed knowledge test (achieved score of ≥75%)	Total	3 (188)	RR 6.86 [0.80, 58.49]	92%	–
Mean knowledge retention test scores, 6-month follow-up	Total	2 (233)	SMD −0.86 [−3.62, 1.90]	99%	–
Mean practical skills test (composite)	Total	2 (120)	SMD 2.90 [0.22, 5.57]	97%	–
Passed skills test (composite) (achieved score of ≥75%)	Total	2 (157)	RR 9.09 [0.12, 663.64]	95%	–
Mean practical skills retention scores (composite), 6-month follow-up	Total	2 (233)	SMD −1.12 [−1.99, −0.25]	95%	–
Intervention-control data
Mean self-efficacy score	Total	2 (180a)	SMD 0.31 [0.01, 0.60]	0%	–
Mortality due to asphyxia	Total	2 (1,589a)	RR 0.34 [0.11, 1.11]	0%	–
Mortality due to asphyxiab	Secondary facility	1 (159a)	Not estimablec	–	Not applicable
	Community	1 (1,430)	RR 0.34 [0.11, 1.11]	0%

#, number; CI, confidence interval; I², heterogeneity; SMD, standardized mean difference; RR, risk ratio.

Bolded values represent significant effect estimates and subgroup findings defined as p < 0.05.

^aAdjusted for clustering.

^bNewborn outcome assessed by the subgroup level of care.

^c0 events were found for experimental and control after adjusting for clustering.

Subgroup analyses by WHO region, level of implementation of care (Table 5), duration of the program, and type of healthcare provider varied widely across subgroups; therefore, no conclusions could be drawn (see online suppl. material B Table B3). The test for subgroup differences by study design did not reach statistical significance. No studies compared NRP training with pre-post data on perinatal or newborn outcomes.

Comparison 5: Any NRP Training Intervention versus Control without Specific NRP Training

Pooled analyses, comparing NRP training as the intervention group versus a control group with no NRP training (presumably existing stands of care), revealed improved mean self-efficacy scores for those who underwent training (Table 5). No sensitivity analysis by the quality of study was possible for intervention and control data. The test for subgroup differences by level of care (Table 5), WHO region, duration of program, and type of healthcare provider did not reach statistical significance (online suppl. material B Table B3).

Additional Comparisons

See online supplementary material I for a summary analysis of additional comparisons.

NRTP Scale-Up

Program Strengths

Various programs (i.e., HBB, NRP, Emergency Obstetric and Early Newborn Care (EmONC), and the WHO-adapted Emergency Triage Assessment and Treatment guidelines on newborn resuscitation) found that most healthcare providers reported greater knowledge, skills, and confidence in their practice after NRTP implementation or scale-up [22–38], while only healthcare providers who partook in HBB programs reported follow-up visits as being useful refreshers [22, 39]. HBB studies also reported that mentorship and real-time advice played a key role in improving confidence, ultimately allowing providers to improve on-site decision-making skills [35, 40]. In addition to mentorship, HBB studies also found that small group and peer-learning experiences allowed for greater interaction with facilitators who were able to provide genuine feedback and promote a collaborative teaching environment [33, 41]. Providers from HBB studies valued hands-on learning experiences over traditional didactic lectures [22, 34, 41]. This was especially helpful when clean and reusable equipment was readily available to providers, as reported by HBB and NRP studies [22, 26, 34, 39, 42]. One EmONC study also noted that there was a large availability of equipment and supplies in regional hospitals at baseline; however, this improved in lower level health facilities after program implementation [29]. One NRP study found that improving accessibility (i.e., flexible training delivery methods) and incentivizing training (i.e., providing snacks) helped increase attendance [42].

Program Challenges or Barriers

HBB studies reported a lack of knowledge and skill retention, especially if providers were moved to a new site or employment opportunity [40, 43]. HBB and NRP trainings were frequently reported as being too short with insufficient time for hands-on practice and knowledge gaps [22, 39, 42], while one HBB study reported that the training was too long [33]. Additionally, one HBB study reported trainees having inadequate time to read and review materials before the course, especially with the high complexity of content for NRTP [41]. Both HBB and NRP studies reported that programs often had unclear and inconsistent instructions, advocating for standardized and clear language during training instructions [33, 41, 44]. One NRP study also found that due to suboptimal training, equipment was often torn, misplaced, or improperly maintained [30]. Staff shortages [22, 30, 39, 44] and limited space [22, 33, 39] prevented providers from adequately practicing NR in HBB and NRP studies. One NRP study found that there was poor collaboration among providers and a lack of incentives to attend meetings and trainings [42]. Lastly, one HBB study noted that older midwives were more resistant to change and overcame this by facilitating separate training sessions dedicated to this group [26].

Cost Analysis

One study revealed the total cost of implementing a full HBB program in a rural Tanzanian hospital as 4,431 United States dollars (USD) [45]. The cost for 1-day training was 2,084 USD and 1-day refresher training was 1,515 USD [45]. One-day accommodations were estimated at USD 20 and training material was 812 USD [45]. The cost per life saved was 233 USD, while achieving a 4.21 USD per life year gain [45]. Another study reported on the proportional cost of implementing a large-scale HBB program in Tanzania and found that the highest proportion of cost was from initial training and equipment practices (45.2%) [46], followed by central program administration (37.2%), and follow-up visits (17.6%) [46]. The estimated cost for NRTP implementation was 600 USD per facility [46]. For all key findings of included scale-up studies, see Table 6 [22–60].

Table 6.

Key findings for scaling-up NRTPs

Study (study design)	Population	Intervention	Level of implementation	Key theme/outcome	Key findings
Arlington et al. [22] (2017) (Mixed Methods)	13,167 health care providers: majority nurses/nurse-midwives	Implementation of HBB	Secondary	What providers liked	Providers largely pleased with the HBB program in Tanzania as a whole. Improved provider-level newborn resuscitation knowledge, skills, and confidence. Simple HBB equipment and poster. Follow-up visits are good refresher. Practical hands-on training and ability to practice at their facility is useful. HBB trainee manual and handouts useful for future reference
				What providers disliked	Perceived small per diems during training. Training too short. A lot covered in 1 day, and sometimes not enough time for hands-on practice. Some HBB attendees were not birth attendants, and some birth attendants were not able to attend the trainings
				Barriers	Regular rotation of HBB-trained providers away from labour ward. Not enough time to practice HBB at work because of staff shortages and large patient volumes. Limited space for newborn resuscitation in some smaller facilities and operating theatres. Many providers still struggling with using bag-mask device
				Recommendations	Additional training desired. Train the staff who were not able to attend original training. Frequent follow-up visits and supportive supervision. Integrate HBB training into medical and nursing preservice curricula
				OSCE passing rate	87% of providers passed OSCE immediately after HBB however dropped to 56% at 4–6 months (p < 0.001)
				Equipment availability and functionality	>90% of facilities had bag-mask devices available to all beds in the labour and delivery ward, and 96% were functional
Bellad et al. [47] (2016) (Quasi-experimental)	70,704 registry births ≥1,500 g, 2,227 birth attendants trained	Multi-country scale-up of HBB + refresher program	Primary, secondary, and tertiary	Perinatal mortality rate	No significant differences in perinatal mortality rate among all registry births (p = 0.39) and fresh stillbirths (p = 0.06). Significant improvement noticed in perinatal mortality rate when stratified by <2,500 g birthweight in Belgaum only (p < 0.001) while no improvement in Nagpur or Kenya
Chalise et al. [48] (2022) (Prospective cohort)	77,688 newborns; 88 facility-based trainers	Safa Sunaulo Nepal model, scale-up, and skill retention program based on HBB	Primary, secondary, and tertiary	Very early mortality and stillbirth rate	Overall decreases in <24 h neonatal deaths (p < 0.001) and intrapartum stillbirths (p < 0.001) before training compared to after training
				Skill retention	Facility-based trainers taught resuscitation skills to 231 medical personnel to support skill retention
Charafeddine et al. [24] (2016) (Cross-sectional)	256 health care providers: doctors, nurses, midwives	Strategic assessment of NRTP implementation at a national level	Mixed	Knowledge	Average post-training test was significantly (p = 0.000) higher compared with pre-training test
				Mortality	Mortality before hospital discharge decreased significantly after implementation of NRP (p < 0.05)
Chaudhury et al. [46] (2016) (Cost analysis)	Not applicable	Cost analysis HBB program	Secondary	Cost analysis	Highest proportion of cost includes initial training and equipment (45.2%), followed by central program administration (37.2%), and follow-up visits (17.6%). Estimated cost is 600 USD per facility. The study found that nationwide expansion of this program across the range of health provision levels and regions of Tanzania would be feasible
Chaulagain et al. [49] (2021a) (Quasi-experimental)	All providers working in perinatal care were included (380 for knowledge, 286 for skill evaluation)	HBB-QI initiatives and refresher training	Secondary	Knowledge and skills refresher evaluation	Knowledge significantly increased post-basic training compared to pre-basic training (p < 0.001). Knowledge and skill decreased by 6-month pre-refresher training (p < 0.001)
Chaulagain et al. [25] (2021b) (Prospective observational cohort)	3,031 newborns	HBB-QI initiative	Secondary	Performance of health care workers on basic NR	For all non-crying newborns, there was a lower proportion given additional stimulation (aOR 0.18; 95% CI: 0.13–0.26), increase in the introduction of quality initiative interventions (aOR 1.23; 95% CI: 1.03–1.46), and greater number of newborns that were initiated on BMV (aOR 1.28, 95% CI: 1.04–1.57) during intervention compared to control
Chiamaka et al. [40] (2021) (Prospective cohort)	72 health care providers	HBB mentorship program to inform the national scale	Primary and Secondary	Knowledge and skills	Most health care workers reported using HBB daily, with the most common method including suctioning (89.5%), and drying and positioning (86%), however 0% use of BMV. 98% of providers shared knowledge and skills with colleagues. 53% of providers moved to a new employment and did not use their knowledge and skills in the new site
Close et al. [26] 2016 (Program implementation evaluation)	42 health care providers: doctors, nurses, midwives, community workers, dispensaries, community members	HBB knowledge translation and implementation program	Primary	Translation of knowledge and skills into routine personal practice	The three most common responses to the question, “As a result of the training have you seen or initiated any change in your personal practice or your health centre?” w ere, in order of frequency
					1. Use of the self-inflating bag to give respirations; 2. Rubbing and drying the baby to give stimulation instead of foot tapping or back slapping; 3. Preparation before delivery: of the room and equipment and having a helper available
				Equipment	The presence of resuscitation equipment in the delivery room was prioritized due to training
				Changes in patient outcome	When participants were asked, “do you think the training made any difference to patient outcome?” they reported feeling more confident in resuscitating a baby not breathing at birth because of their training which provided them with the necessary equipment and response plan
				Identification of inhibitors to change	Resistance to the new training by older midwives was the primary inhibitor to change. Attempted to overcome this by holding a separate training session for those midwives and then asking the midwives to help them train the other less-educated healthcare workers. This was helpful and suggest doing the same if introducing another similar concept
				Knowledge and skills	PCVs and rural healthcare workers showed significant increases in knowledge and skills (p < 0.001)
Das et al. [50] (2019) (Quasi-experimental)	779 birth attendants 4,431 newborns required resuscitation at birth	3-day ENC + NRP + skill laboratories, better documentation, and supervision implementation project	Secondary	Fresh stillbirth	There was 15.6% reduction in fresh stillbirth rate in the three districts included after intervention compared to no intervention (p < 0.001)
Disu et al. [51] (2015) (Cross-sectional)	727 health care workers: doctors and nurses/midwives	Nation-wide NRP training	Secondary and Tertiary	Attendance rate and use of bag and mask	The delivery attendance rate increased after training compared to before training. 88–94% of doctors and 72–93% of nurses successfully conducted bag and mask in post-training period. Nurses used bag and mask for resuscitation more commonly than doctors. 87–94% of the doctors and 92–97% of the nurses/midwives trained other birth attendants
Drake et al. [43] (2019) (National cohort study)	8,391 health care providers, Initial training: 3,592, Modified training: 4,799	HBB implementation study (initial vs. modified on-the-job training guidance)	Secondary	Skill score and retention	Similar post-training average skill score (p = 0.07) between initial and modified training, and both groups experienced significant drop in skills over time. Though, modified training had significantly higher skill score 4–6 months post-training compared to initial training (p < 0.0001). Medical attendant cadre had greatest skill retention
Eblovi et al. [52] (2017) (Quasi-experimental)	48 midwives	HBB implementation and refresher course	Primary	OSCE score	Significant decrease in OSCE scores immediately after training compared to 4 months later (p < 0.00001), however, attained a similar level to immediate post-training scores after a 4-month refresher course (p = 0.0013). Very early neonatal mortality rate was 0.71% compared to nationwide Ghana average of 1.7%
Enweronu-Laryea et al. [27] (2009) (Quasi-experimental)	271 health providers: midwives, nurses, residents in obstetrics and paediatrics, medical officers, and nurse anaesthetists	AAP/AHA NRP implementation project	Primary, secondary, and tertiary	Knowledge and skills	Significant improvement (p < 0.001) in post-training test scores compared to pre-training test scores for knowledge and skill outcome among all health providers (nurse anaesthetics, nurses, physicians, midwives)
				Knowledge and skill by level of care and health care provider	Midwives in primary health care facilities were less likely to achieve a passing post-test score than midwives at secondary and tertiary facilities (p = 0.004). Passing rate was 85% for physicians and 60% for non-physicians
Ersdal et al. [53] (2013) (Observational cohort)	53 health care providers: midwives, anaesthetic, operating, or student nurses, and ward attendant	HBB implementation	Secondary	NR testing performance	Significant increase in testing performance for NR scenarios post-HBB training compared to pre-HBB training (p < 0.0001). Suctioning (p = 0.782) and ventilation (p = 0.185) techniques at birth had no significant changes, however use of stimulation in the delivery room significantly decreased after HBB training (p < 0.0001)
Findlay et al. [54] (2022) (Retrospective cohort)	452 birth attendant-mother pair	Adapted HBB implementation	Community	Mortality	Neonatal mortality was not significantly different between those that had adapted HBB implementation versus those that did not (aOR = 0.48; 95% CI: 0.16–1.44)
				Implementation of HBB-trained steps	The following outcomes were all statistically significant post-adapted HBB (p < 0.001): drying baby and removal of cloth, finding clean flat place, cutting cord after 3 min, preparation, use of a sterile razor, stimulating baby if no crying, sucking liquid from mouth and then nose, suctioning, tying cord, having clean and dry sheets, and washing hands before putting on gloves
Gill et al. [55] (2011) (Cluster RCT)	3,497 newborns, intervention: 1,961, control: 1,536	Modified version of the NRP implementation	Modified version of the NRP	Mortality and stillbirth	Birth attendants assigned to intervention versus control had reduction in very early neonatal mortality (RR = 0.40; 95% CI: 0.19–0.83), and reduction in mortality at day 28 (RR = 0.55; 95% CI: 0.33–0.90). Stillbirth rates were similar between the intervention and control groups
Gupta et al. [28] (2014) (Quasi-experimental)	392 health care providers: medical assistant, clinical officer, registered midwife, enrolled nurse/midwife, nurse/midwife technician, other	HBB scale-up initiative	Secondary	Feedback from supervisor	39.7% and 41.5% in round 1 and 2, respectively, had received written comments from supervisors, most of which were discussed between supervisees and supervisor (86% and 90% in round 1 and 2 with verbal feedback)
				Knowledge and skills	Mean number of bag and mask ventilation steps done correctly were higher in round 2 (mean 7.1; SD 2.0) versus round 1 (mean 6.2; SD 2.4). Mean number of steps performed correctly when delivering a term baby without complications was higher in round 2 (mean 11.0; SD 2.9) compared to round 1 (mean 10.1; SD 3.3)
				Availability of Supplies, Equipment and Guidelines	Guidelines were more commonly available in round 2 compared to round 1. Generally, equipment and supplies were indifferent between round 1 and 2
Isangula et al. [39] (2018) (Longitudinal prospective)	599 health care providers	HBB focus group discussion program implemented at scale in Tanzania	Primary and secondary	What providers liked	Improved newborn resuscitation knowledge, skills, and confidence; largely very pleased with the programme overall; HBB equipment and poster easy to use; follow-up visits were useful refreshers; practical hands-on training and ability to practice at their facility; and HBB trainee manual and handouts useful for future reference
				What providers disliked	Fewer reported dislikes than likes; perceived small per diems; some HBB attendees were not birth attendants while some birth attendants were not able to attend; and training length too short with not enough time for hands-on practice
				Barriers	Rotation of HBB-trained providers off labour ward; limited space for resuscitation in smaller facilities and operating theatres; not enough time to practice HBB at work owing to staff shortages; and many providers still struggling with using bag-and-mask device
				Recommendations	Expand the training time; train staff who were not able to attend the original training; provide frequent follow-up visits and supportive supervision; and integrate HBB training into medical and nursing curricula
Kamath-Rayne et al. [56] (2017) (Quasi-experimental)	250 newborns, 70 health care providers: physicians and nurses	HBB-QI	Primary	Implementation of HBB-trained steps	More babies received basic NR (i.e., drying and stimulation) after HBB training, however remained below ideal goals. These practices, however, led fewer babies to require BMV
Kamath-Rayne et al. [41] (2018) (Qualitative program evaluation)	24 health care providers	HBB implementation	Mixed/Unclear	Facilitators	The participants commented that the facilitators were able to handle multiple groups at once, provoke discussion, and receive constructive feedback
				Workshop structure	Enjoyed hands-on learning over lectures. Small groups helped with practicing skills and allowed peer-learning experience. Small groups also allowed for personalized interaction with facilitators who were then able to provide participants with genuine feedback
				Quality improvement	New addition of instruction on quality improvement challenged participants to consider how they could improve or what they could due to bring change to their facility
				Effective content	Content was presented and displayed in a clear and concise manner, and that the flip chart contained more guidance for facilitation and more instruction on how to organize and conduct a workshop
				Ineffective content	Use of suggestive or ambiguous language such as “may” or “could,” insufficient time to read and review material prior to the course and high complexity of content for peripheral clinics
				Ineffective integration	Require continuous practice or refresher course
				Unrealistic expectation	Vague instructions on skin-to-skin scare. Lack of resources and space
KC et al. [23] (2017) (Quasi-experimental)	137 nurses	HBB-QI cycle	Tertiary	Knowledge and skills	Mean knowledge, skills scores for preparation for resuscitation before every birth, bag-and-mask skills, OSCE A, and OSCE B were improved immediately after training (p < 0.001)
				Knowledge and skills retention	Mean knowledge (p = 0.6), mean skills scores for preparation for resuscitation before every birth (p = 0.1), bag-and-mask skills (p = 0.815), OSCE A (p = 0.07), and OSCE B (p = 0.1) was retained after 6 months compared to immediately post-training
Kebaya et al. [42] (2018) (Implementation study)	55 health care providers	NRP based on the WHO ETAT+ course and audit/feedback tool for quality improvement, implementation project	Secondary	Barriers	Key barriers included providers who were resistant to change, lack of incentive to attend meetings/trainings, poor collaboration among providers, knowledge gaps, and time constraints within provided education and training
				Strategies	Hold meetings to understand providers perspective on out-dated practices and using evidence from best practices. Additional strategies include creative incentives (e.g., snacks), plan join meetings, hold education sessions, flexibility and convenience in training and identify champions to ensure best practice maintains
				Resources deployed and outcome	Used evidence from studies done in similar settings led health care providers to embrace skills taught and increase their familiarity with best practice. Held monthly meetings with snacks which improved staff collaboration. Printed notes, projectors, and laptops available by hospital administration also improved staff’s familiarity with current best practice. Increased flexibility assisted in increased education and training opportunities
				Program compliance	Significant improvement in all criteria audited. Specifically, 100% compliance for healthcare providers attending deliveries having sufficient NR training, availability of equipment and supplies, availability of evidence-based protocol in labour ward, and all newborns were assessed accurately and assigned an APGAR score. 90% compliance was found for newborns with birth asphyxia who were correctly assessed and managed according to WHO guidelines
Makene et al. [29] (2014) (Quasi-experimental)	1,049 newborns total (489 newborns in 2010 and 560 in 2012)	Basic emergency obstetric and newborn care quality improvement program at scale	Primary and secondary	Quality care outcomes	Overall improvement in quality-of-care intervention from 39% pre-training to 73% post-training (p < 0.0001) (i.e., immediately placing newborn on mother’s abdomen, discarding wet towel and covering with dry towel, help initiate breastfeeding within 1 h, tie or clamp cord when pulsation stop)
				Knowledge and skills	Knowledge significantly improved from 23% pre-training to 41% post-training (p < 0.0001); however, no improvement noticed in skills
				Equipment and supplies	High availability of equipment and supplies for essential newborn care at baseline for regional hospitals; however, this improved at lower level health facilities
Mildenberger et al. [30] (2017) (Mixed methods)	44 health care providers: midwives and intern doctors	NR implementation course	Secondary	Knowledge and skills	Solidified NR steps, improved staff's skills and helped them be better prepared, helped clear misunderstandings regarding methods/equipment. Successful training improving hands-on practice, knowledge retention, and more babies surviving. Did not have a continuation or follow-up training, or resources/materials that could be used long term. Improved confidence and helped midwives continue practicing NR long-term
				Equipment and supplies	Would have appreciated sufficient equipment (i.e., bag and masks). Often equipment is torn and misplaced. Due to improper training, the items are not maintained well
				Staffing	Inadequate staffing needs to be addressed
				Rural insight	Highlights the need for more reference material, updated protocols to reach rural and remote providers in a convenient manner (i.e., through email), more equipment supply, and opportunities for further training and meetings with neighbouring rural providers to discuss NR cases
Munyaw et al. [31] (2022) (Quasi-experimental)	107 health care workers: nurse/midwives and doctors	Scaled-up electronic-device heart rate assisted HBB training	Secondary	Knowledge and skills	Knowledge and skills (OSCE A and OSCE B) improved significantly (p < 0.001) post-training compared to pre-training
				Stillbirth and early neonatal death	Fresh stillbirths rate decreased from 9 to 5 deaths per 1,000 total births and early neonatal deaths at 7 days from 5 to 3 deaths per 1,000 live births (p < 0.05) post-training compared to pre-training
Naresh et al. [57] (2022) (Before and after)	1,785 midwives, nurses, and physicians; 15,947 births were assessed	SSN package includes HBB training-of-trainer scale-up and retention training program	Primary, secondary and tertiary	Mortality, morbidity, and stillbirth	Compared to the beginning (the first 3 months of assessment), at follow-up (last 3 months of the program) there was a significant reduction in mortality (p = 0.01), morbidity (p = 0.01), and intrapartum stillbirth (p = 0.001)
Naz et al. [33] (2022) (Qualitative)	39 health care providers	HBB (low dose, high frequency approach) implementation	Community	Effectiveness of training	Good teaching pace, effective small groups, collaborative teaching environment, taught according to mental capacity, patient when things written in English, friendly and respectful learning atmosphere, continuous supervision was helpful, improved knowledge and skills, and enhanced confidence in performing skills
				Challenges	Discrepancies in theoretical concepts of trainers and facilitators, poor weather, long hours of training, lack of transportation, and lack of opportunities to attend deliveries due to the large number of maternal and child healthcare providers in the same vicinity
				Suggestions	Consider spreading training to more birth attendants as level of patient intake is increasing and consider integrating HBB training into midwifery curriculum because all midwives need to learn and practice these concepts
Nvonako et al. [32] (2022) (Mixed method)	30 health care providers	Implementation of the WHO-adapted ETAT	Tertiary	Challenges and solutions	Challenge and solution 1: inadequate knowledge on newborn resuscitation in HCWs, equipment in theatre, training on newborn resuscitation, and teamwork between theatre staff and midwives. Proposed solution is to provide appropriate equipment by hospital
					Challenge and solution 2: need more trained midwives. Proposed solution is to revise intake criteria by hospital management of staff who are to work in the maternity ward and frequently providing training and refresher courses on newborn resuscitation for maternity staff
					Challenge and solution 3: due to lack of knowledge, often the anaesthetist will take over. Proposed solution is to have in-hospital drills with mannequins, as well as training of midwives and anaesthetists on newborn resuscitation to increase collaboration
				Skills	Skills significant increased for real-life observations (p < 0.00054) and newborn resuscitation scenario simulations (p = 0.0004)
Patel et al. [58] (2019) (Quasi-experimental)	78,948 births, pre-HBB: 38,078, post-HBB: 40,870	HBB implementation	Primary, secondary, tertiary	Mortality and stillbirth	Compared to pre-training, after post-training there was a significant reduction in stillbirth rate (p = 0.002), very early neonatal mortality rate (p = 0.006), and very early perinatal mortality (p = 0.0001)
Rule et al. [59] (2017) (Quasi-experimental)	96 health care providers	HBB-QI	Tertiary	Suspected hypoxic-ischaemic encephalopathy (SHIE)	Significant reduction in SHIE post-HBB compared to pre-HBB (p = 0.01)
Singhal et al. [34] (2012) (Mixed method)	102 health care providers	HBB multinational program evaluation	Primary and secondary	What providers liked	Enjoyed graphic linkage among materials, hands-on learning, having cleanable and reusable bag-and-mask and suction devices, and use of appropriate assessments. Acknowledged the importance and understandability of The Golden Minute
				Suggestions	Suggest having additional time for hands-on learning and skills, incorporate real-life resuscitation experiences, greater practice opportunities after the course, have video demonstration of skills and patient assessments, larger illustration in print material, have translation of print material into local language, and conduct assessment in local language
				Knowledge and skills	After training, there was a significant improvement in knowledge and skills compared to no training (p < 0.05), although no significant findings for bag and mask ventilation (p > 0.05)
Umunyana et al. [35] (2020) (Mixed method)	712 health care providers	HBB (low dose, high-frequency training) with clinical mentorship and QI and implementation program	Primary and secondary	Provider perspectives on mentorship	Healthcare providers found that mentorship improved confidence, skills, and led to better quality of data collection, ultimately allowing for better decision making
				Skill score	After receiving two or more visits, 60% of providers received an equal to or greater than 80% on OSCE B for their 1st visit and 100% by the 6th visit
				Mortality and stillbirth	Significant findings comparing post-training to pre-training for neonatal admissions due to asphyxia (annual reduction of 12.9% [95% CI = 8.3–17.5%]), fresh stillbirths (a 6.2% annual reduction [95% CI = 2.1–11.4%]); and neonatal deaths due to asphyxia (an annual reduction of 4.2% [95% CI = 0.2–6.3%]). Borderline insignificance for death within 30 min of birth (19% reduction, p = 0.06)
VanHeerden et al. [44] (2016) (Action research qualitative study)	69 nurses (questionnaire), 8 doctors (focus groups)	Strategies to sustain a quality improvement initiative in NR	Secondary	Challenges prior to training	Need for more training, staff, resources, and training on mannequins. Often felt overworked and had a lack of self-confidence. There was a lack of operational equipment. Protocols were inconsistent and unavailable in some units
				Evaluation of training	Providers had opportunity to practice NR. Improved confidence and knowledge. Improved equipment availability which contributed to the functionality of maternity units. Overall improvement in staff attitude, teamwork, and collaboration
				Areas for improvement	Communicating training and practice sessions were not optimal leading to under-attendance. Limited financial resources meant certain equipment could not be supplied and the maintenance of equipment could not be achieved. Staff shortages and high turnover rates need to be addressed to keep up the standard of care. Better management and communication required for transport of critically ill neonates. Protocols should be standardized across hospital units
Vossius et al. [45] (2014) (Cost-analysis)	Birth attendants	Cost-effectiveness of HBB program	Secondary	Mortality	Very early neonatal mortality decreased significantly (p = 0.047); however, fresh stillbirth did not decrease significantly after training
				Overall cost	Total costs for the full HBB implementation were USD 4431. Costs for the initial 1-day HBB training in April 2010 were USD 2084, re-training was USD 1515 for the 1-day HBB, 1-day accommodations were USD 20 and costs for training material were USD 812. The HBB program is a low-cost intervention. Implementation in a very rural faith-based hospital like is highly cost-effective. To facilitate further global implementation of HBB a cost-effectiveness analysis including government owned institutions, urban hospitals and district facilities are beneficial for a diverse analysis to explore cost-driving factors and predictors of enhanced cost-effectiveness
				Cost per life saved and per life gained	Costs per live saved were USD 233, while it was USD 4.21 per life year gained
Wilson et al. [37] (2017) (Mixed method)	33 midwives	HBB implementation	Primary	What providers liked	Simple and understandable training made practice easier, as well as the provision of the necessary equipment
				Improvement	Duration of trainings was sufficient, but trainings need to be recurring
				Knowledge and skills	New knowledge learned and increased provider confidence to deal with emergency situations. Knowledge scores and resuscitation skills significantly improved (p < 0.05)
Wilson et al. [36] (2020) (Mixed method)	59 midwives	HBB implementation	Secondary	Most impactful parts of training	Most impactful parts of training included usefulness of training in practice. Providers felt proud after completing training
				Areas of improvement	Areas of improvement included expanded training to include care of delivering mother
				Knowledge and skills	Knowledge scores and resuscitation skills significantly improved and were sustained over a 6 months (p < 0.001)
Xu et al. [60] (2012) (Pre-post evaluation program)	9,905 health care providers: obstetricians, paediatricians, midwives	Nationwide NRP training program	Secondary	Availability of equipment	90% of delivery rooms had appropriate resuscitation equipment (i.e., bags, infant masks, laryngoscopes, and radiant warmers), however, was generally more complete in delivery than operating rooms. Generally, a higher proportion of provincial-level hospitals had supplies for intubation and suction compared to prefecture and country hospitals
				Intrapartum death and Apgar	Comparing pre-NRP to post-NRP, intrapartum-related deaths in the delivery room decreased from 7.5 to 3.4 per 10,000 from, and the incidence of Apgar ≤7 at 1 min decreased from 6.3% to 2.9%
Xu et al. [38] (2014) (Cluster RCT)	Intervention: 28,456 newborns, 97 health care providers, Control: 26,182 newborns, 87 health care providers	NRP implementation	Secondary	Knowledge	Knowledge and self-confidence were significantly higher in intervention group compared to control group (p < 0.001)
				Mortality	The intervention group had significant reduction in incidence of birth asphyxia (defined as 1-min Apgar score ≤7) (p < 0.001) and asphyxia-related deaths (p = 0.076) over time; however, control counties had no difference over time in asphyxia rates

HBB, helping babies breathe; NRP, neonatal resuscitation program; NRTP, neonatal resuscitation training program; NR, neonatal resuscitation; ENC, essential newborn care; AAP, American Academy of Paediatrics; AHA, American Heart Association; QI, quality improvement; ETAT, Emergency Triage Assessment and Treatment; APGAR, Appearance, Pulse, Grimace, Activity, and Retention; OSCE, Objective Structured Clinical Examination; SSN, Safa Sanaulo Nepal; PCV, Peace Corp Volunteers; WHO, World Health Organization; RCT, randomized control trial; aOR, adjusted odds ratio; RR, risk ratio; SD, standard deviation; CI, confidence interval; BMV, bag-mask ventilation.

Laryngeal Mask Airway

The topic of laryngeal mask airway (LMA) versus face mask ventilation (FMV) or endotracheal intubation (ETT) was of interest for the purpose of NR practices. During the initial scoping exercise, we identified and leveraged data from an existing recent, high-quality review [61]. Since this manuscript is focused on the de novo findings of the effectiveness and scale-up of NRTPs, we have reported the study authors’ results in online supplementary material J and further contextualized the findings as a future option in LMICs [62–64].

Discussion

To increase knowledge about newborn survival in LMICs, we conducted a de novo systematic review to synthesize the most up-to-date evidence on NR strategies and training. This synthesis includes a review of the effectiveness and scale-up of NRTPs, and the effectiveness of LMA versus FMV or ETT (Fig. 2). Findings suggest that the NRP and HBB training programs had a similar impact across all outcomes. Both programs improved knowledge- and skill-based outcomes but had discordant newborn outcomes (e.g., mortality). Implementation of the HBB program showed promising cost-effectiveness in LMIC settings. To scale-up current NRTPs, programs should focus on improving long-term retention outcomes and improving training material accessibility.

Fig. 2.

Laryngeal mask airway.

NRTP effectiveness studies, evaluating both NRP and HBB training programs, reported improved knowledge- and skill-based outcomes immediately post-training, but these results were not maintained at follow-up assessment. These findings align with the literature on retention, where both the NRP and HBB programs show retention is low with declining knowledge and skill post-training over time [13, 65]. This is concerning as healthcare providers, although satisfactory to provide NR immediately post-training, may not qualify as having the minimum knowledge and skill required to perform resuscitation at follow-up. This would suggest that one-off training is not suitable in these settings and training programs should prioritize routine refresher courses instead. Many studies have found that a low-dose high-frequency training program can improve retention of knowledge and skills of healthcare providers [66–68]. Current recommendations from the NRP 8th Edition suggest implementation of Resuscitation Quality Improvement (RQI) with quarterly low-dose, high-frequency sessions for 10–15 min to maintain and improve resuscitation skills [14]. Further research also suggests a Just-In-Time (JIT) training program which offers on-time training for those required may provide promising simulation-based refresher opportunities compared to scheduled training [69]. Additional research should consider the optimal length and frequency of refresher courses. No consistent differences were observed when considering the knowledge- and skill-based outcomes by level of care, study design, WHO region, simulation, standardization, duration of program, or type of healthcare provider. Moreover, no major differences in effect size were identified for knowledge and skill outcomes post-sensitivity analysis by quality assessment.

We found mixed results for NRTPs on the outcomes of newborns. No pre-post pooled estimates, and only two intervention-control studies leading to pooled estimate, evaluated the effectiveness of NRP on newborn outcomes. This is likely because the NRP has previously been validated as effective for reducing the risk of neonatal mortality in the literature [2, 4, 70], and thus more recent studies have focused on healthcare provider knowledge and skills. Additionally, NRP is less likely to be evaluated in LMICs as the resources required to provide NRP may not be available based on the level of care within the country. Studies evaluating pre-post HBB training programs showed a reduced risk of stillbirth and fresh stillbirth but an increased risk of macerated stillbirth. However, findings from a sensitivity analysis by quality assessment revealed a nonsignificant effect estimate for macerated stillbirths after studies appraised as high risk were removed. While studies including macerated births generally had similar demographic findings between control and intervention periods, additional factors such as antenatal care, enrolling women with higher risk pregnancies in the post-intervention group, or improved classification may have increased the heterogeneity within this outcome [71]. When disaggregating mortality by time, HBB post-training also indicated a reduced risk of very early perinatal mortality and very early neonatal mortality and, a systematic review by Patel et al. [4] assessing all NRTPs also reported a reduced risk of stillbirths, early neonatal mortality, and perinatal mortality among LMICs. According to the literature, stillbirths as commonly observed within perinatal programs are sometimes misclassified as early neonatal deaths, and perhaps more commonly, asphyxiated newborns are often misclassified as fresh stillbirths post hoc [16, 72–76]. Misclassification of stillbirths in low-resource settings is common, as infrequent access to heart rate monitors and where signs of life are not rigorously checked often cause misclassification [77, 78]. These findings also lend credibility to considering intrapartum or fresh stillbirths as an extension of early neonatal mortality and resuscitation programs as a strong strategy to reduce both. Moreover, it is important to underscore that the terms fresh and macerated stillbirth are not synonymous with intrapartum stillbirths and could underestimate intrapartum stillbirths; therefore, current recommendations suggest reporting disaggregated data for fresh and macerated stillbirth with a focus on foetal heart rate documentation at admission and beyond [79]. It is also important to capture longer term developmental outcomes as such resuscitation programs are scaled up. Subgroup analyses by study design, WHO region, level of care, duration of program, standardization, and type of healthcare provider tended to reveal discordant results across outcomes considered.

To effectively implement and scale-up NR strategies in LMICs, our review suggests improving the frequency of refresher training [22, 30, 39, 41], implementing dedicated hands-on learning and skill opportunities [34], updating outdated NRTP protocols [30], including handouts and large training posters [22, 34], and improving the availability of equipment in training settings [30, 32, 34, 44]. Studies also suggest training more and healthcare providers diverse in specialization [32, 33] to help with staff shortages. These recommendations are aimed to address barriers currently identified in training programs such as improving knowledge and skill retention, standardizing training protocols, and increasing training opportunities for healthcare providers in LMICs. These findings are in accordance with Wall et al. [80] who suggest refresher training should be provided on a standard and frequent basis, as well as the need for available, functioning, and clean equipment to improve training accessibility. These findings align with the WHO Essential Newborn Care Course which emphasizes clinical and ongoing practice, ultimately improving knowledge and skill retention, as well as improving access to materials, which are all important goals of scaling-up a resuscitation program.

While comprehensive asphyxia care is vital to newborn care, studies reporting on non-NR specific training programs where the effect of NR training outcomes could not be disaggregated from outcomes associated with broad, integrated packages on maternal and child health, were excluded (online suppl. material K). No studies directly compared NRP to HBB, limiting ability to provide conclusions about which program may be better to implement and scale-up in LMIC settings. However, NRP is generally best applied in facilities with qualified providers and settings with access to specialized equipment, whereas HBB may be implemented in primary care settings [81]. HBB also appears to be cost-effective in low-resource settings, though this evidence originates from a single country in Africa and more research is necessary. Two included studies discussed the implementation and cost-effectiveness of the HBB program in Tanzania. One study reported the total cost of implementing an HBB program in rural Tanzania as 4431 USD [45], while another study reported implementing HBB at large-scale implementation as 600 USD [46]. Although both studies conclude the HBB program is cost-effective, the discrepancy in costs between the two studies may be attributed to the level of implementation (i.e., one rural hospital vs. several facility-level centres). The cost-effectiveness of the NRP in LMIC settings remains unclear and future research is also warranted. Although publications are forthcoming, the Helping Mothers and Babies Survive (HMBS) program may provide promising training opportunities as they target remote and online training while bringing down associated costs to implement and scale-up resuscitation training programs.

Strengths and Limitations

Our study has several strengths. First, we conducted a rigorous literature search and performed data extraction and quality assessments independently and in duplicate to reduce human error. Second, our review conducted an extensive list of subgroup analyses to identify consistent themes across healthcare providers and newborn outcomes. Third, our project was largely conceptualized and informed by a technical advisory group (TAG), who provided strategic and expert advice to improve the overall quality of our review.

We acknowledge several important limitations of this study. First, our review is limited by the quality of included studies, where studies reported as high risk, serious risk, or poor quality depending on the study design, potentially threaten the internal validity of findings. Second, most outcomes had substantially high heterogeneity between included studies, which may be due to the lack of standardized testing and evaluation approach used across studies. Moreover, since included studies were conducted in settings with diverse resources, modified NRTP protocols, and varied experience among providers on resuscitation practices may have led to the increased heterogeneity across findings. Third, the current meta-analyses consist largely of pre-post studies without concurrent control groups, limiting our ability to measure an effect estimate from resuscitation training alone compared to other changes at health facilities or in communities during the same time. Lastly, since both cost-analysis studies were limited to HBB programs in Tanzania, conclusions on cost-effectiveness lack generalizability to all LMICs and training programs, necessitating further research in this area.

Conclusion

NRTPs, including standard NRP and HBB programs, improved knowledge and skill scores immediately post-training that were retained during longer follow-ups, suggesting the need for refresher courses. Both the NRP and HBB appear to be effective; however, two studies reported that the HBB program was cost-effective in Tanzania, suggesting that the HBB program or elements thereof are low-cost and scalable in LMICs. Future research across diverse settings should evaluate the cost-effectiveness of the NRP. Our review could be improved upon with studies directly comparing HBB to NRP, thereby allowing conclusions about which program is most effective in different contexts. To scale-up NRTPs, programs should focus on increasing the frequency of refresher training and improving training material accessibility.

Acknowledgments

We gratefully acknowledge the asphyxia TAG of experts for their contributions to the conceptualization of the study. We also gratefully acknowledge the work of Li Jiang who assisted in creating forests plots by level of care for the review topic LMA versus FMV or ETT.

Statement of Ethics

An ethics statement is not applicable because this study is based exclusively on published studies.

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

This work is supported by funding received from the Bill and Melinda Gates Foundation (#INV-042789). The funders had no role in the conceptualization, analysis or writing of this article.

Author Contributions

L.H. and Z.A.B. conceptualized the study. D.S., R.L.H., and L.H. designed the study and carried out the search and study selection. D.S. and R.L.H. completed data collection and critical appraisal. D.S. and L.H. led data analysis. D.S. completed writing of manuscript with substantial contributions from R.L.H., G.D., L.H., T.V., and Z.A.B.

Funding Statement

This work is supported by funding received from the Bill and Melinda Gates Foundation (#INV-042789). The funders had no role in the conceptualization, analysis or writing of this article.

Data Availability Statement

The data used to support the findings of this study are included within the article and its supplementary material. Further enquiries can be directed to the corresponding author.