Despite remarkable gains, improving neonatal survival globally remains a challenge. Several innovative service-delivery packages have been developed for delivery by Community Health Workers (CHWs) to maximize system efficiency and increase the reach of services to communities in need.

However, embedding these in health systems is complex owing to the considerable structural and procedural alignments intrinsic to system-thinking. There is limited evidence from evaluations of such innovative packages implemented at scale within a health system.

In response to high neonatal mortality in Ethiopia, the Community-based Newborn Care (CBNC) programme was planned as the main strategy to maximize CHWs’ engagement. Key aspects included simplified treatment regimens for neonatal illness, integrating outreach services, and task-shifting.

The CBNC evaluation plan was co-designed by the Ethiopian Ministry of Health to fulfil evidence needs to improve service delivery. This research aims to present an evaluation of health system response and quality-of-care provided through the CBNC service package, based on WHO’s Community Health Worker functionality model.

Methods
There were three survey-assessment time-points: baseline (November 2013 – before CBNC implementation), quality-of-care (December 2015), and follow-up (December 2017- four years after implementation). Data were collected in 100 districts across four regions of the country, covering assessments of health facilities and 10,000 households in each round. Our analysis is framed by an adaptation of WHO’s CHW functionality model, and explores the CBNC innovation from two perspectives: a) health system response, in terms of tangible support and supportive supervision linkages, and b) quality-of-care, through CHWs’ capacity to deliver CBNC services, and their performance in doing so.

Results
From the health system perspective, the four-year evaluation of CBNC implementation shows significant gains in tangible aspects of service for young infants, including above 90% availability of antibiotics and job aids. However, in terms of human resource capacity, CHWs’ correct diagnosis of key conditions in infants was only 37%. There was a relative reduction of 70-80% in CHWs’ functional health literacy on fundamental aspects of case management over the time period of the study. The frequency of quarterly supportive supervision visits was above 80% during the evaluation period. However, there was relative neglect of supervision elements to support the management of young infants’ health conditions.

Conclusion
There was some degree of success in infrastructure and physical resources. However, embedding and scaling up innovative packages like CBNC lacks both the system-thinking and attention to health system building-blocks needed to optimize their potential.