Quality of Kangaroo Mother Care services in Ethiopia: Implications for policy and practice

Background Providing high-quality kangaroo mother care (KMC) is a strategy proven to improve outcomes in premature babies. However, whether KMC is consistently and appropriately provided in Ethiopia is unclear. This study assesses the quality of KMC services in Ethiopia and the factors associated with its appropriate initiation among low birth weight neonates. Methods We used data from the 2016 national Emergency Obstetric and Newborn Care (EmONC) assessment which contains data on all health facilities providing delivery care services in Ethiopia (N = 3,804). We described the quality of KMC services provided to low-birth weight (LBW) babies in terms of infrastructure, processes and outcomes (survival status at discharge). We also explored the factors associated with appropriate KMC initiation using multivariable logistic regression models. Results The quality of KMC services in Ethiopia was poor. The facilities included scored only 59.0% on average on a basic index of service readiness. KMC was initiated for only 46.4% of all LBW babies included in the sample. Among those who received KMC, 66.7% survived, 13.3% died and 20.4% had no data on survival status at discharge. LBW babies born in health centers were twice more likely to receive KMC compared to those born in hospitals (AOR = 2.0, 95% CI: 1.3–3.0). Public facilities, those with a staff rotation policy in place for newborn care, and those with separate newborn corners were also more likely to initiate KMC for LBW babies. Conclusions We found low levels of appropriate KMC initiation, inadequate infrastructure and staffing, and poor survival among LBW babies in Ethiopia. Efforts must be made to improve the adoption of this life saving technique, particularly in hospitals and in the private sector where KMC remains underutilized. Facilities should also dedicate specific spaces for newborn care that enables mothers to provide KMC. In addition, improving record keeping and data quality for routine health data is a priority. We used data from the 2016 Ethiopian Emergency Obstetrics and Newborn care (EmONC) assessment [12]. The EmONC assessment was a national cross-sectional survey of all public hospitals, health centers and private facilities (higher clinics and above) that provided maternal and newborn health services and reported attending births in the past 12 months. The EmONC assessment did not include health posts or medium and small private clinics because these facilities are not expected to attend deliveries. Of the eligible 4,385 facilities in all nine regions and two city administrations in Ethiopia, 3,804 facilities were assessed (including 293 hospitals, 3,459 health centers and 52 clinics). A total of 11 facilities were not accessible due to political unrest or staff refusal. The survey used 13 questionnaires including 12 health facility assessment modules and one health system assessment module. These were adapted from the Averting Maternal Death and Disability program [12]. The survey collected data on EmONC signal functions, facility readiness (including the availability of equipment, guidelines, human resources, infection prevention measures, etc.), volume of services and maternal and newborn outcomes [12–13]. A module on newborn complications was designed to collect information on premature babies weighing less than 2000 grams. Trained data collectors extracted information from charts identified through facility registries or from the staff. In each facility, interviewers were expected to review charts for the last three LBW babies born in the past 12 months. Data on treatments provided and survival status were extracted. In most facilities, only one LBW baby chart was reviewed. We measured KMC quality using the three domains of quality defined by Donabedian: infrastructure, processes and outcomes [14]. Structural quality was assessed using all facilities included in the national EmONC survey. We used three indicators of service availability (facility density, maternity bed density and core health workforce density) and by one index for service readiness. These indicators were adapted from the WHO Service Availability and Readiness Assessment (SARA) manual. Facility density was calculated by the number of health facilities providing maternity care per 10,000 population. Maternity bed density was calculated by the number of maternity beds per 1,000 pregnant women and the health work force density was calculated by the number of core medical professionals per 10,000 population [11, 15–16]. The service readiness index was based the availability of a series of items necessary for the provision of maternal and newborn care including specific indicators related to KMC. The index covered four domains: infrastructure and equipment, essential medicine and commodities, core staffing and guidelines, job aids and documentation. The items and calculations are described in S1 Appendix. Process quality was assessed using a binary variable for whether KMC was initiated for each LBW baby. This is a measure of appropriate treatment and competent care [17]. Because most facilities had data on only one LBW baby, we selected only the last LBW baby per facility. Because KMC cannot be initiated until the baby is stable, we also looked at the proportion of LBW babies that were initially put in incubators. Outcomes were assessed based on the survival status at discharge for all LBW babies who received KMC. Process quality and outcomes were only assessed in the subset of facilities with data on LBW babies. We selected a series of facility- and provider-level covariates that may be associated with quality of care. These were selected based on prior literature on the factors affecting health care provider performance and quality in low income countries [17]. Facility characteristics included facility type (hospitals or maternal and child health (MCH) specialty centers, health centers and higher clinics), urban location, managing authority (public or private), whether the facility had a separate newborn corner, a newborn intensive care unit (NICU) and a policy in place for staff rotation. Facility types were based on definitions by the Federal Ministry of Health. Hospitals and MCH specialty centers generally have operating theaters while health centers and clinics do not. Provider characteristics included cadre, work experience in years, age, gender, and whether the provider had a written job description. Infrastructure processes and outcomes were assessed using descriptive statistics. We also looked at associations between each of the facility- and provider-level covariates and appropriate KMC initiation using bivariable logistic regressions. Covariates with a p-value of ≤0.25 were considered for inclusion in the multivariable logistic regression model with the forward likelihood ratio method. In the final multivariable model, a p-value<0.05 was used to determine statistical significance. All analyzes were performed using SPSS version 21TM software. Ethical approval for the original survey was granted by the Scientific and Ethical Review Office of the Ethiopian Public Health Institute (EPHI). The Federal Ministry of Health of Ethiopia granted access to the data for this analysis. The institutional review board of Mekelle University considers this analysis as exempt from ethical review as it is a secondary analysis of de identified data.