Impact analysis and evaluation of the Ethiopian Neonatal Network

Abstract

The Ethiopian Neonatal Network (ENN) represents one of the first low-income country neonatal quality improvement (QI) communities. To determine if changes in structure, process and outcome measures were associated with formation of the ENN and to learn from teams, quantitative and qualitative analyses were completed. All infants discharged during 2018–2022 from 11 hospitals with consistent data collection over the 5-year period were included in infant-level analyses. Trends by year were evaluated using Cochran-Armitage tests. Annual surveys captured facility-level data. Nurse and physician leads at ENN hospitals participated in focus groups in 2023. Inductive and deductive approaches were used to extract themes and findings. Overall, 38 049 infants were discharged. Participating sites reduced nurse-to-patient ratios, increased newborn beds and implemented continuous positive airway pressure (CPAP) with blended oxygen. There were significant increases in antenatal steroid exposure, kangaroo mother care and receipt of oxygen or CPAP among infants with respiratory distress (p<0.0001). Admission hypothermia among inborn infants decreased (p<0.0001). Overall survival decreased (p<0.0001). Mortality due to prematurity-related complications decreased (p=0.0089) while mortality due to infection increased (p=0.0016). Three themes were determined from focus groups: positive changes to data utilisation and patient care following ENN membership, data entry, technical issues and buy-in as barriers to participation and recommendations on additional support. Development of the ENN was associated with adopting a culture of data-driven improvement, positive changes in measurable quality of care and improved patient outcomes. Sustaining, spreading and evaluating multidisciplinary neonatal QI communities are important components to global efforts targeting mortality reduction.

Summary box.

• Standardised patient-level neonatal databases and quality improvement have been instrumental in improving quality of care and outcomes for infants in middle- and high-income country settings; however, the feasibility of a multidisciplinary neonatal quality improvement and learning community in a low-income country was unknown.

• Ethiopian Neonatal Network (ENN) teams and neonatal leaders demonstrated a visionary grassroots effort of implementing high quality data collection and use, embracing the responsibility to improve care through evidence-based practice and quality improvement.

• ENN teams led substantial improvements in structure, process and outcome measures, including mortality related to prematurity.

• The Sustainable Development Goals call for a reduction in neonatal mortality, which necessitates improving the quality of care for mothers and newborns.

• The formation or expansion and empowerment of multidisciplinary neonatal communities grounded in evidence-based and data-driven quality improvement should be prioritised and supported with investments.

Introduction

Ethiopia, a low-income country in sub-Saharan Africa, has made great strides in reducing mortality of children under the age of 5, meeting this Millennium Development Goal ahead of the targeted timeline and being one of 12 low-income countries reaching the two-thirds reduction target by 2015.¹ Neonatal mortality, deaths occurring within 28 days of birth, has improved at a slower pace and is currently 27 per 1000 live births in 2023.² The Sustainable Development Goals (SDGs) target for all countries is to reduce neonatal mortality to at least as low as 12 per 1000 live births.³ 64 countries are off track to meet this goal, calling for increased investment and attention to global inequities in newborn care and priorities.⁴

Vermont Oxford Network (VON) has partnered with Addis Ababa University and Tikur Anbessa Specialized Hospital in Ethiopia since 2008, when the Tikur Anbessa Neonatal Intensive Care Unit (NICU) project was jointly established and supported the fellowships of the first three Ethiopian-trained neonatologists.⁵ Building on that strong foundation and meeting the demand for locally relevant and actionable data paired with quality improvement (QI) initiatives tailored for resource-limited settings, VON partnered and collaborated with the Ministry of Health, Ethiopia and the Ethiopian Pediatrics Society to establish the Ethiopian Neonatal Network (ENN) in 2018. The ENN has a membership of 22 public hospitals with level 2–3 inpatient neonatal units caring for small and sick newborns. Teams collect data on all neonatal admissions and work on QI projects aimed at reducing preventable causes of neonatal mortality.

The formation of the ENN represents one of the first low-income country neonatal QI and learning communities. A key tenet of this community is to have frontline providers and stakeholders involved in the development of data items, improving the quality of data collected on newborns and therefore trusting it as a marker of quality of care and continually using it to learn and share within the newborn community. Using the Donabedian framework for evaluating the quality of medical care, data elements were selected within the categories of structure, process and outcome.⁶

Objectives

• Determine if changes in structure, process and outcome were observed after formation of the ENN.

• Learn from the experiences of nurses and physicians throughout the first 5 years of the ENN community.

Quantitative methods

ENN members submitted standardised data on infants discharged from a neonatal unit. Each hospital identified a data collector who did not receive remuneration from VON or the ENN for that role. All study data were collected and managed using REDCap electronic data capture tools hosted at VON.⁷

Infant-level data item definitions and a data collection form were provided in a manual of operations co-developed by the ENN and VON and did not change during the study period.⁸ Definitions for all infant-level measures can be found in the manual of operations except small for gestational age, which was defined as less than the 10th percentile for sex and gestational age on the 2013 Fenton growth chart.⁹ The 2013 revision to the Fenton reference harmonised the preterm references created by meta-analysis of population-based birth cohorts from six high-income countries with WHO growth standards to provide smoothed reference curves covering from 22 to 50 weeks of postmenstrual age. Determination of cause of death is based on the best available evidence, noting that autopsies or postmortem examinations are rare. All infants discharged from 1 January 2018 to 31 December 2022 from 11 hospitals with consistent data collection over the 5-year period were included in infant-level analyses (tables1 2, and figure 1). The 5-year interval was selected a priori, reflecting the initial 5 years of the ENN. Trends by year at the infant level in process and outcome measures were evaluated using Cochran-Armitage tests with a significance level of α<0.05.

Table 1. Demographics.

	n
Any antenatal care (%)	37 198	97.3
Location of birth (%)
Inborn	38 015	78.6
Outborn—other health facility	38 015	19.5
Outborn—home	38 015	1.9
Multiple gestation (%)	38 032	7.6
Small for gestational age (%)	33 565	35
Birth weight, g (median, IQR)	38 034	2800 (2080–3250)
Birth weight categories (%)
≤1000 g	38 034	1.4
1001–1500 g	38 034	8.0
1501–2000 g	38 034	15.0
2001–2500 g	38 034	15.6
2501–3000 g	38 034	25.2
3001–3500 g	38 034	22.1
>3500 g	38 034	12.9
Gestational age, weeks (median, IQR)	38 049	38 (36–39)
Gestational age categories (%)
<28 weeks	38 049	0.5
28–32 weeks	38 049	8.3
33–34 weeks	38 049	9.0
35–36 weeks	38 049	10.6
37–38 weeks	38 049	29.4
39–40 weeks	38 049	31.9
>40 weeks	38 049	10.3

Table 2. Process and outcome measures.

	2018		2019		2020		2021		2022		P value
	n	%	n	%	n	%	n	%	n	%
Process measures
Infants born at gestational age <34 weeks with exposure to antenatal steroids	335	21.2	334	34.1	614	37.5	827	44.0	825	46.4	<0.0001
Temperature measured on admission	5986	93.6	5303	97.6	7722	99.2	9243	99.5	8475	99.5	<0.0001
Infants with birth weight <2 kg that received KMC	1185	44.6	1065	64.3	1571	64.9	2033	63.8	1773	68.8	<0.0001
Infants with suspected or confirmed sepsis that received antibiotics	3878	95.3	3535	96.8	5426	97.8	6899	95.0	6880	96.1	0.3295
Infants with respiratory distress that received oxygen or CPAP	2481	91.1	2299	89.8	3676	94.2	4607	94.1	4384	95.0	<0.0001
Outcome measures
Admission hypothermia among inborn admissions	4457	15.1	4097	9.6	6197	8.4	7137	7.3	6634	6.0	<0.0001
Discharge weight <10th percentile among survivors with length of stay >2 weeks	611	80.0	493	74	672	71.7	827	74.1	765	78.8	0.7656
Survival among those with final disposition	6182	93.2	5520	89.9	7901	88.9	9420	88.2	8384	90.3	<0.0001
Causes of mortality
Prematurity-related complications	400	53.0	535	48.6	825	45.7	1042	43.3	775	46.2	0.0089
Infection	400	15.0	535	13.6	825	17.9	1042	23.3	775	17.8	0.0016
Intrapartum-related cause or asphyxia	400	25.8	535	32.5	825	29.8	1042	27.0	775	28.4	0.5852
Congenital anomaly	400	6.3	535	5.2	825	6.5	1042	6.4	775	7.6	0.1826

CPAP, continuous positive airway pressure; KMC, kangaroo mother care.

Figure 1. Survival of infants discharged from participating neonatal units of the Ethiopian Neonatal Network from 2018 to 2022, stratified by birth weight categories.

Members completed an annual facility survey. 14 hospitals were included in the facility-level analysis (table 3). Trends were not tested due to sample size. All quantitative analyses were conducted using SAS V.9.4.

Table 3. Structural measures.

	2018 (N=14)		2022 (N=14)
	n		n
Newborn intensive and special care bed capacity, median (range)		22 (15–47)		29 (10–56)
Average number of patients per bed, median (range)		2 (1–4)		1 (1–2)
Dedicated space for KMC/presence of KMC beds (%)	13	92.9	14	100
KMC beds, median (range)		4 (0–8)		6 (1–10)
KMC follow-up clinic (%)	11	78.6	11	78.6
High-risk infant (neurodevelopment) follow-up clinic (%)	12	85.7	13	92.9
Transport available with ambulance and staff (%)	12	85.7	12	92.3
Newborn transfers dependent on family income/resources (%)	8	57.1	11	78.6
Rotation of nursing staff (%)	9	64.3	7	50.0
Neonatal nursing considered a specialty (%)	11	78.6	11	78.6
Nurse:patients ratio, median (range), days		1:6 (1:2–1:8)		1:4 (1:2–1:10)
Nights		1:7 (1:4–1:12)		1:5 (1:1–1:38)
Weekends		1:7 (1:4–1:12)		1:5 (1:2–1:14)
Neonatologist on staff (%)	3	21.4	4	28.6
Number of paediatricians, median (range)		1 (0–4)		2 (1–8)
Number of general practitioners, median (range)		2 (0–4)		1 (0–10)
Paediatric surgery on site (%)	5	35.7	7	50.0
Blankets available for drying in the delivery rooms (%)	11	78.6	10	76.9
Functional sinks with running water available in newborn unit (%)	12	85.7	13	92.9
Alcohol rub at patient bedsides (%)	13	92.9	13	92.9
Thermometers functional and available (%)	14	100	12	85.7
Self-inflating bags with both term and preterm masks available in the delivery room (%)	10	71.4	11	78.6
CPAP with 100% oxygen available in the newborn unit (%)	11	78.6	13	92.9
CPAP with blended oxygen available in the newborn unit (%)	0	0	10	71.4
Mechanical ventilation available in newborn unit (%)	1	7.1	4	28.6
Surfactant available (%)	1	7.1	0	0
Caffeine available (%)	0	0	1	7.1
ROP screening available in newborn unit (%)	0	0	9	64.3
IV nutrition/ PN available (%)	1	7.1	2	14.3
IV pain medication available (%)	7	50.0	7	50.0

CPAP, continuous positive airway pressure; IV, intravenous; KMC, kangaroo mother care; PN, parenteral nutrition; ROP, retinopathy of prematurity.

Qualitative methods

Nurse and physician leads at ENN hospitals participated in focus groups to assess the perceived impact of the ENN, implementation challenges and recommendations to improve the ENN. We used a phenomenology approach to better understand the experiences of nurses and physician leads participating in the first 5 years of the ENN, with focus groups chosen to allow for participant interaction and open discussion on alignment of experiences. Two main researchers led the qualitative evaluation and analysis (HS and RW). The main researcher (HS) has been employed by VON since 2022 and had limited interactions with the participants and implementation of ENN activities within participating NICUs at the time of the focus groups and interview as well as during the data analysis process. This researcher has a maternal, newborn and child health background in public health and is not a practising clinician. The supporting qualitative researcher (RW) has previously served as a neonatal nurse and nurse educator at a NICU participating in the ENN; however, not in the nurse lead role. Following focus groups and the interview, the qualitative research team (HS and RW) held discussions and completed debrief forms to address reflexivity.

Team leads were divided by profession (nurse or physician) and by their hospital’s rate of data submission (high submission or low submission). The study team elected to separate nurses and physicians due to the potential bias or hesitancy to share information arising from power dynamics that may be present. Grouping by rate of infant data submission was intended to facilitate richer conversation as experiences may be more aligned within these domains. Ideally, teams entered infant data for every patient admitted to their neonatal unit. However, practically, some teams experienced challenges with data entry or data submission that led to data gaps. Although NICUs with inconsistent data collection were excluded from the quantitative analysis to minimise sampling error and bias, the experiences from these team leads were included in qualitative analyses. The threshold to be considered high or low data submission was database participation for ≥9 months during January–December 2022. Individual nurse or physician leads who had participated in the ENN for 1 year or less were excluded from focus groups to ensure participants had sufficient experience with the ENN to contribute to the conversation. Of 27 eligible participants, 26 were included in focus groups or were interviewed individually.

The study team developed semistructured interview guides. Focus groups were conducted in person in English and Amharic, one of the national languages of Ethiopia. Qualitative research team members (HS and RW) are fluent in both languages. HS conducted one interview, two focus groups, took notes for two focus groups and completed transcriptions. RW conducted two focus groups, took notes for two focus groups and spot-checked transcripts.

The focus groups and interviews were recorded using a voice recorder and recordings were transcribed verbatim into English. Prior to starting the recording, the participants were given the option to decline participation and assured of confidentiality. Access to the recordings and qualitative data was limited to the two qualitative researchers. Identifiers were not used. Recordings and qualitative data were saved with ID numbers. The main qualitative researcher (HS) transcribed the recordings and removed names of individuals and hospitals due to this researcher having the most neutral perspective. The supporting researcher (RW) listened and transcribed all recording segments with non-perfect audio clarity and compared results with the main researcher to ensure accuracy. The data were coded by both qualitative researchers using MAXQDA 24.4.1. Both researchers met to resolve discrepancies and completed the thematic analysis using methods as described by Braun and Clarke.¹⁰ The Standards for Reporting Qualitative Research and corresponding checklist were used to draft and edit this manuscript.¹¹

Patient and public involvement

Public stakeholders, including the Ethiopian Federal Ministry of Health, Ethiopian Pediatrics Society, and neonatal nursing and physician leaders were engaged in the co-development of data items, design and implementation of the ENN. Participating Ethiopian NICU teams were encouraged to involve families in local QI teams and projects as part of the development of the ENN. Public stakeholders and ENN teams, including families, will participate in dissemination and ongoing feedback.

Quantitative results

Overall, 38 049 infants were discharged from 2018 to 2022 of which 78.6% were inborn at the reporting facility (table 1). The median (IQR) birth weight was 2800 g (2080–3250) and the median (IQR) gestational age was 38 weeks (36–39).

The facility survey (table 3) revealed decreases in the average number of patients per bed (from 2 to 1) and nurse-to-patient ratios for day (from 1:6 to 1:4), night (from 1:7 to 1:5) and weekend (from 1:7 to 1:5). The median number of newborn intensive and special care beds increased (from 22 to 29), as did the number of units with continuous positive airway pressure (CPAP) devices with the ability to blend oxygen (from 0 to 10), provide mechanical ventilation (from 1 to 4) and screen for retinopathy of prematurity (from 0 to 9).

At the infant level (table 2), there were significant increases in infants born at less than 34 weeks’ gestation exposed to antenatal steroids (p<0.0001), temperature measured on admission to the neonatal unit (p<0.0001), infants born with birth weight <2000 g who received kangaroo mother care (KMC) (p<0.0001), and infants with respiratory distress who received oxygen or CPAP (p<0.0001). There was a significant decrease in admission hypothermia among inborn admissions (p<0.0001). Overall survival among infants admitted to participating neonatal units has decreased (2018: 93.2%, 2019: 89.9%, 2020: 88.9%, 2021: 88.2%, 2022: 90.3%; p<0.0001). Survival by birth weight categories is shown in figure 1. Mortality due to prematurity-related complications decreased (p=0.0089) while mortality due to infection increased (p=0.0016).

Qualitative results

Four focus groups and one interview were conducted in November 2023 with 26 participants representing 19 ENN hospitals. Three of the 22 ENN hospitals had nursing and physicians leads that were ineligible for the focus groups due to working in their role for less than 1 year. 50% of participants were nurses and 46% of participants were female. The focus groups were nurses at low data submission units (n=6), nurses at high data submission units (n=7), physicians at low data submission units (n=4) and physicians at high data submission units (n=8). The average focus group length was 39 min 19 s. The interview was 37 min 55 s and was conducted with a participant that met the eligibility criteria for the physician at high data submission unit but could not attend their assigned focus group. Three themes were determined: (1) positive changes to data utilisation and patient care following ENN membership, (2) data entry, technical issues and buy-in as barriers to ENN participation and (3) opportunities to better support ENN initiatives at an institutional and national level (box 1B).

Box 1. Major themes, subthemes and exemplary quotes on perceived impact of ENN membership.

A. Theme 1: Changes to data utilisation and patient care following ENN membership

Data utilisation

“Many parts of it (ENN) help us for the next steps. Having all the data at the year end and seeing the graphs was very helpful; everyone looks at it. I don’t think something like this exists in all hospitals, to see the work that has been done all year, to put it on a graph and have the opportunity to look at it and say this is where we are, it’s great… Now, at the NICU there wasn’t really something like this, we would do this and present to the staff, let’s say monthly, and there’s a chart we look at quarterly, and I took what I learned from that and did work based on it.” (male, nurse, low submission)

Training

“They would show us, and we would watch. We would sit and follow everything they were doing and saying. They would show us how to assemble the stuff, how to put it together. Before, we would wash the incubator. Every week, we would wash it, all the smaller parts like the glass and the sponge, and things like that, we would just clean surface parts. We didn’t know how to take apart the inside parts. Dr. X came one time, and Dr. X and they came together, and they took apart everything and showed us! They took it apart into pieces. We were so in shock we just wanted to disappear! So, after that, me, when I go to other places, when I go to primary hospitals, even if I go for another unrelated reason, I completely take apart the incubator and show them, so that they know to wash all the parts. [I show them] How the CPAP works, how to shut it off, before returning we would show them things like that when we go. That is very beneficial.” (female, nurse, low submission)

“The use of Downes Score has gone way up, it’s used by nurses and physicians. The Downes Score is basically now just considered like vital sign paperwork basically and it’s recorded. Using it to make changes to the care, upgrade, downgrade oxygen or decrease, etc. based on Downes Score is great. I consider the standardized Downes score utilization as a very high benefit.” (male, physician, high submission)

Network

“One thing that makes me happy is that you know, you say Ethiopia NICU Network, and we are actually a network… we exchanged our thoughts and ideas, meeting together, brainstorming and thinking through ideas… A lot of group channels were started on telegram, one with the whole group… You hear about each hospital’s activities, and if you hear of something that is better at another hospital, you bring it to your own hospital. We’re like a team; we became a team. We became one, we’re tied together.” (female, nurse, high submission)

Patient care

“At my hospital before ENN, which was 2.5 years ago, the neonatal mortality rate was 25%, but now it is about 5-10% with the use of CPAP and other things. That is after starting the ENN project, after improved use of CPAP, after training the nurses. Showed significant improvement for our hospital.” (male, physician, high submission)

B. Theme 2: Barriers to ENN implementation

Data entry

“Yes, with us there are reasons why it stopped, meaning the entry. Number one, I think it is the admission rate. From Ethiopia we have one of the highest admissions, I mean it’s large, we admit from 300 up to 400…So then the nurses we have are few. Patient ratio is [high]. Doing patient care and also completing those forms…it just becomes arguments. Before, I used to be the one that would enter and send on the tablet and I’m also doing work. When it became hard for me, I asked if the staff could enter data, so it was said that the staff will enter data too, but then they said, “This is research, I’m not completing this data entry”. And that was the end, that created a gap, and we stopped completing the entry… We want to do this work, but because we have a lot of patients, is hard…But we fully understand that this benefits us, 100 percent.” (female, nurse, low submission)

Technical issues

“Wi-Fi is another challenge. At our hospital the only place there is Wi-Fi is in the main administration building. To [upload] data [we] need internet and have to run back and forth to do that. [We] would sometimes miss meetings and things. We asked the hospital to fix this for us, the NICU is on the third floor, but we haven’t gotten Wi-Fi” (male, physician, high submission)

Buy-in (leadership and staff)

“The other big challenge is that when you consider the resources needed for this, the people above us, the leadership, they put us lower priority, they don’t accept us. When we ask them about it, they consider our work to be additional, like the hospital wouldn’t benefit from it, there are people that think like that. At times like that, it’s challenging.” (male, nurse, high submission)

“In the unit there are a lot of other activities so a lot of it depends on the willingness of the head nurse or the staff that would be entering [data]. It is a burden on them, usually the head nurse or someone else, and other people in the unit aren’t willing to help them.” (male, physician, high submission)

C. Opportunities to better support ENN initiatives

Solutions to data entry barriers

“What ENN can do, and also with EPS [Ethiopian Pediatric Society], together. I don’t think they [ENN and EPS] have the ability to compensate, but what they can do is they can communicate with the Ministry to communicate with the hospital administration to provide compensation with the person that is entering data and communicating with ENN. The hospital administration doesn’t want to pay, but the Ministry can convince them that this is a problem.” (male, physician, low submission)

Institutional and national support

“For ENN to be very effective and work correctly how it should, the hospital’s medical director and/or leadership need to be involved, need to be a team member. Just as everyone working in the NICU is on the team, they also should be on the team.” (male, physician, high submission)

Participants unanimously reported the benefits ENN membership had on patient care by nurses and physicians, and across settings from labour and delivery to the newborn unit. Specifically, they noted that membership increased their data utilisation, formed a network of neonatal care professionals that supports clinical practice through knowledge sharing and provided training in CPAP, using the Downes score¹² (an objective and standardised measure of respiratory distress, used for multidisciplinary assessments and communication of patient status and clinical progress), documentation, infection prevention, resuscitation, the warm chain and KMC. Participants agreed that the improvement in quality of data and its accessibility gave them the power to implement targeted data-driven approaches to address issues in their unit, including QI projects, and using ENN data to advocate for their unit. Some participants reported feeling galvanised, motivated and a sense of ownership from seeing their unit’s data, seeing themselves as change agents. Training provided through the ENN led to vast improvements in the patient care provided, especially the training for nurses, and an increase in the use of the Downes score. One benefit noted by the participants in both high and low data submission groups was training to use CPAP as a respiratory support modality. All participants agreed that prior to the ENN, even if CPAP machines were available, they were not being used, or, in the rare case they were being used, they were not being used, maintained and cleaned properly. Additionally, at that time, if patients were treated with CPAP, they were not properly or uniformly monitored while on CPAP. Following ENN membership, participants reported increased use, successful use, proper cleaning/maintenance of CPAPs and benefits to patients.¹³ Participants also found the network between clinicians at their centre and clinicians at other centres fostered through the ENN to be a major benefit. Forming a network of neonatal care professionals allowed participants to access one another for support in the form of formal and informal knowledge sharing, hands-on trainings and trainings through a CPAP tele-mentoring project. Exemplary quotes are listed in box 1A.

While all participants described the benefits of ENN membership, they all also acknowledged that there were considerable systemic barriers with implementing data collection and entry as well as making improvements in their patient care. Specifically, they noted a combination of individual, hospital and national level barriers including high staff turnover, shortage of staff, lack of staff buy-in, insufficient compensation for data entry personnel and internet challenges. Likewise, participants reported lacking the equipment and supplies needed to provide high quality patient care, particularly CPAP devices and necessary CPAP consumable supplies. Exemplary quotes are listed in box 1B.

Participants suggested opportunities to mitigate the barriers they identified and optimise engagement with the ENN at the individual, hospital and national level. First, they indicated that buy-in from the staff responsible for data entry is critical to success and suggested a bottom-up approach where all levels of staff are informed and involved in the ENN at the inception of membership at their unit and fully understanding the purpose and goals. Likewise, they described how support from hospital leadership and from the Ministry of Health is integral in addressing their challenges by creating changes that can address the systematic barriers. Participants suggested that the hiring of a data clerk or providing compensation for staff entering data could improve motivation for data entry locally. Similarly, increasing buy-in with hospital leaders and the Ministry of Health could assist in circumventing resource challenges by encouraging fair allocation of equipment and supplies and bolstering efforts to increase availability of needed equipment and supplies in markets accessed by hospitals. Exemplary quotes are listed in box 1C.

All participants (nurses and physicians from low and high submission units) were aligned in the themes and sub-themes that were reported. The level of detail and amount of time spent discussing themes, however, were subtly different between focus groups, which may correlate with level of importance or meaningfulness. Nurses spent a greater amount of time and were more detailed in their discussions on changes due to CPAP use, indicating an increased level of self-efficacy for nurses administering CPAP. One nurse said, “Before we would only administer CPAP when a physician told us to, but now we do it on our own based on protocols.” Another commented, “Before we didn’t administer CPAP without being told to because we would rather switch to manual [bag-mask ventilation] instead of risking doing CPAP incorrectly and having the baby die on us.” Physicians included more quantitative data in their discussions, including specific data points related to their units and improvements to patient care, likely due to physicians having more involvement in reporting QI projects. Challenges with data entry were discussed with equal emphasis in the high submission and low submission focus groups, which could mean that high data submission units encountered challenges with data entry but were able to overcome them.

Discussion

To our knowledge, the ENN is one of the first low-income country multidisciplinary neonatal network and community of practice built on the foundation of a standardised QI database. ENN teams and neonatal leaders demonstrated a visionary grassroots effort of implementing high-quality data collection and use, embracing the responsibility to improve care through evidence-based practice and QI. This shared vision of the ENN continues to be realised with culture change, creating and empowering community and developing voices driven by data, and the strength of collective frontline healthcare workers advocating for the patients and families they serve. The formation of the ENN itself is a tremendous accomplishment interprofessionally and for Ethiopia.

The importance of the ENN is amplified by the substantial improvements associated with the first 5 years of its existence. ENN teams learning from each other drove important structural changes, notably improving nurse-to-patient ratios, decreasing rotation of nursing staff, decreasing number of patients per bed and increasing KMC spaces and beds. Focusing on evidence-based practice and patient safety, and with necessary stakeholder collaboration, there were significant increases in availability of CPAP with blended oxygen and retinopathy of prematurity screening. ENN teams showed significant improvement in processes linked with mortality reduction: increasing receipt of antenatal steroids among preterm infants, temperature measurement and hypothermia on admission, receipt of KMC for infants with birth weight <2000 g, and increased use of oxygen or CPAP for respiratory distress. Although there was a significant reduction in mortality due to prematurity-related complications, mortality overall and specifically from infection increased during this 5-year period. It is notable, however, that during this time interval the number of infants admitted to participating ENN neonatal units significantly increased, from approximately 5000–6000 patients/year in 2018–2019 to 8000–9000 patients/year in 2021–2022. It is possible that an increased number of smaller and sicker babies are receiving care and are being counted in the denominator over time, contributing to the relative increase in calculated inpatient mortality.

This study highlights the ongoing challenges of providing high-quality care to small and/or sick newborns in resource-limited settings. While mortality due to prematurity-related causes decreased, overall mortality increased. The WHO identifies antenatal steroids, KMC, CPAP and surfactant therapy as high-impact interventions that improve outcomes for preterm infants in low- and middle-income countries (LMICs).¹⁴ The realised impact of the more intensive interventions likely requires a foundation of good thermoregulation, nutrition and infection prevention. A study from South Africa demonstrated that CPAP is the respiratory intervention with the largest impact on neonatal mortality in LMICs as compared with mechanical ventilation and surfactant.¹⁵ Our findings align with this, showing decreased mortality from prematurity-related causes over 5 years, coinciding with increased access to CPAP, improved use of antenatal steroids and increased rates of KMC. In a recent publication from the ENN specifically evaluating the implementation of a telementoring programme for CPAP use from September 2021 to September 2022, 19 participating ENN teams showed an increase in number of preterm infants receiving CPAP per month and reduction in overall preterm mortality (from 28% preintervention to 21.6% postintervention; p<0.0001).¹³

Strikingly, in this 5-year period, overall mortality and specifically mortality due to infections increased. The ratio of neonatal healthcare-associated infections to total neonatal infections in LMICs remains unknown.¹⁶ Additionally, sepsis and sepsis-related mortality are at times clinically diagnosed, and we lack true rates of culture-positive infections and antimicrobial resistance. Although the standard in all ENN sites is to obtain blood cultures prior to antibiotic administration and treat based on antibiotic sensitivities, at times there are stock outs of blood culture bottles, lab reagents and antibiotics, significant delays in receipt of results or infants born in settings within the healthcare system in which this ideal pathway cannot be followed. Challenges identified in this study—the lack of basic infrastructure such as running water, insufficient consumable supplies, inadequate decontamination of equipment, high staff turnover and rotation policies—place vulnerable neonates at heightened risk for sepsis. To reduce infection-related complications and mortality, stakeholders must invest in the healthcare system and implement multipronged strategies to improve infection prevention and control practices.¹⁷

In high-income countries, standardised patient-level databases for small and/or sick newborns—such as VON^{5 18 19} and the National Institute of Child Health and Human Development Neonatal Research Network²⁰ in the USA, the Canadian Neonatal Network,²¹ and the Neonatal Critical Care Minimum Dataset in the UK²²—have been instrumental in improving quality of care. In LMICs, however, such databases outside of research settings are relatively rare. Beyond the VON Global Database used in the ENN and now African Neonatal Network,²³ examples include the Indian Facility-Based Newborn Care Database,²⁴ the Kenyan Clinical Information Network,^{25 26} the Neonatal Nutrition Network,²⁷ Neotree in Malawi and Zimbabwe²⁸ and the NEST360 Neonatal Inpatient Dataset.²⁹ As care continues to advance in service of improving outcomes for infants and families, opportunities remain to align data definitions across platforms and reduce data burdens by minimising duplication and creating synergy by using high quality data for multiple purposes.

Our study identifies several challenges in data collection, including unreliable internet connectivity, high staff turnover, staff shortages, lack of buy-in from staff and hospital leadership and insufficient compensation for data entry personnel. Recommendations from team leads to address these challenges include hiring dedicated data clerks and increasing support from hospital leaders, Regional Health Bureaus and the Ministry of Health. The current structure does not provide external compensation for data entry and encourages locally developed solutions to incorporating data collection and use into routine workflows, emphasising the culture of local data use rather than the ENN as a time-limited external project. Recognising this effort with leadership, staff and stakeholders as part of the neonatal care team and reasonable workloads, as some ENN units have done, is essential for this model’s long-term success. Lessons from other data systems in LMICs—such as the Indian Facility-Based Newborn Care Database,²⁴ which employs government-funded data collectors equipped with desktop computers for real-time data entry, and the NEST360 Alliance in Malawi,²⁹ which deploys government-employed data collectors across 38 facilities—can also guide efforts and provide perspective to strengthen and sustain the ENN.

Accurate and high-quality data entry is essential for leveraging this information to improve care, advocate for resources and develop policies across the healthcare system. Ethiopia’s Health Sector Transformation Plan II (2015/2016–2019/2020) emphasises preventing data falsification as a key priority.³⁰ Abiy Seifu Estifanos et al have documented concerns about data falsification in Ethiopia, particularly in reporting neonatal outcomes.³¹ A key strength of this study is the use of a standardised patient-level database co-developed with ENN team leaders collected longitudinally and reviewed in real-time, which helps mitigate such concerns.

Additional strengths of this study include quantitative and qualitative methodology to deepen our understanding of the impact of the ENN and share lessons learnt and recommendations as the community continues to expand. The focus groups for this study were conducted in both Amharic (the national language) and English, with professional groupings that facilitated open and honest discussions. Unlike the aggregated data and indicator tracking in routine health management information systems, which is mandated by Ethiopia’s Ministry of Health but can appear removed from clinical care, the ENN infant-level database is collected by team members and is directly linked to documentation of patient clinical care. This approach not only improves the quality of care provided to patients but also fosters a sense of ownership and empowers healthcare professionals to advocate for their patients. A challenge for the future is thoughtful integration of these data streams and workflows. Furthermore, building QI capacity at the facility level generates quantifiable benefits, strengthening the healthcare system overall.

Limitations include restricted participation in the ENN due to COVID-19 and civil unrest within the country as well as capturing a time period overlapping a pandemic. Using a minimum of 1 year of ENN participation as an inclusion criterion for the focus groups of nurses and physicians meant some teams and historical knowledge were excluded, given high staff turnover. Furthermore, with only one country and 19 facilities represented, findings may lack generalisability. However, the expansion of the VON Global Database into the African Neonatal Network, encompassing select facilities in Ethiopia, Rwanda, Nigeria, Uganda, Zambia and Zimbabwe, promises broader representation and impact. Future work will examine the impact as the African Neonatal Network QI, leadership development and learning community continues to expand in sub-Saharan Africa. Additionally, multiple local, national and international changes and projects occurred during this 5-year timeframe in addition to the development of the ENN. It is possible that focus group members may have conflated co-occurring projects with the impact of the ENN itself. Although we report significant findings, these are associations and causations cannot be inferred.

The ENN has shown that data-driven improvement is possible in a low-income country setting. Following the success of the ENN, this community expanded into the African Neonatal Network, with continued opportunities to build on and collaborate with colleagues in Kenya’s Clinical Information Network, the Neonatal Nutrition Network, Neotree, NEST360 Alliance and others. International platforms such as the VON Global Neonatal Database facilitate African teams to learn, share and contribute to the global neonatal community. Everyone benefits from this enriched community with diverse practice settings, lessons learnt and innovations. To realise the infant mortality target within the SDGs, investment in our neonatal community and leaders is required to efficiently focus on gaps while retaining and empowering a specialised workforce. As ENN teams and leaders continue their improvement journey and welcome others to join the African Neonatal Network, the opportunity for collaboration and shared learning will increase exponentially, and with strategic investment directed by data and clinical leaders, the pace of improvement has the potential to soar.

Conclusion

The grassroots development of a multidisciplinary neonatal community in a low-income country with a foundation of evidence-based and data-driven QI was associated with positive change in measurable quality of care, patient outcomes and culture of data use and collaboration in Ethiopia. Continued success and increased pace of improvement will require listening to clinical leaders and data, acting on existing challenges and strengthening the shared and growing voice of our families and multidisciplinary teams.

Data availability statement

Consistent with membership agreements, Vermont Oxford Network data are not publicly available.