Background: Disparity in resource allocation is an issue among various health delivery units in Ethiopia. To sufficiently address this problem decision-makers require evidence on efficient allocation of resources. It has an implication the neonatal health service and mortality, which accounts for nearly half of the under-five mortality in Ethiopia. Therefore, the purpose of this study was to assess the technical efficiency of primary health care units providing neonatal health services in Southwest Ethiopia.

Methods: Two-stage data envelopment analysis was conducted based on one year (2016/17) data from 69 health posts and 23 health centers in Southwest Ethiopia. Health facilities were selected following the world health organization’s guideline for assessment of district health system. Input and output data were collected from each of the facility, respective district health offices and finance, and economic cooperation offices. Technical efficiency scores were calculated using data envelopment analysis software version 2.1. Tobit regression was then applied to identify determinants of technical efficiency. STATA version 14 was used in the regression model and for descriptive statistics.

Results: By utilizing the best combination of inputs, eight health posts (11.59%) and eight health centers (34.78%) were found to be technically efficient in delivering neonatal health services. The mean technical efficiency score of the health posts was 0.42 (± 0.31) with 17.39% pure technical and 11.59% scale efficiency. On the other hand, mean technical efficiency score for the health centers was 0.75 (± 0.26). Out of the total health centers, 65.22% were pure technically efficient and 34.78% were scale efficient.  Compared with others included in the analysis, inefficient health delivery units were using more human and non-salary recurrent resources. The regression model indicated that there was a positive association between efficiency and the health facility’s size of catchment population [β = 0.00002, 95% CI, 2.13E-06, 2.98E-05]. Waiting time at the health posts [β = -0.011, 95% CI, -0.018, -0.004] was found to negatively affect efficiency.

Conclusions: Most of the health posts and the majority of health centers were found to be technically inefficient in delivering neonatal health services. This indicates issues with the performance of these facilities with regards to the utilization of inputs to produce the current outputs. Either the existing resources could be used to serve additional neonates, or the inefficient facilities could reduce their inputs without affecting the neonates cared for.