Background: The Health Extension Program (HEP) is a flagship program launched by the Ethiopian government in 2004, with the aim of expanding health care services nearer to the community in a cost efficient manner. Promotive, preventive and basic curative services are included in the program. HEP is provided by trained and salaried Health Extension Workers (HEWs). The HEP has been first launched in agrarian setting then expands to pastoralist and urban.  Recent years has seen a sharp increase in the number of HEWs leaving their job which could strain the program and could potentially rollback the gains made by the program so far. Attrition survey was made with Objectives of determining the prevalence and incidence rate of attrition, estimate the median time of attrition, ascertain the trend of attrition, identify the factors associated with attrition and the reasons for leaving.

Methods: The study was conducted in all regions and two city administrations. A retrospective cohort study (records of HEWs from 2004-2019) design was employed. Additionally, in depth interview with resigned HEWs were conducted to examine reasons for attrition.  All HEWs workers who were registered and deployed by Woreda Health Offices were eligible for inclusion. A random sample of implementing Woredas were taken proportionally from all the three types of Woredas. A cluster sampling method was used to include all HEWs in the selected Woredas. A total of 85 Woredas and 3,486 HEWs were included for the quantitative study whereas the sample size for the qualitative study was 18 HEWs who have left their job. A data extraction tool was developed by the study team and trained data collectors used tablet computers to extract data from personnel files of HEWs. Data quality was continuously monitored through daily calls with each teams and onsite supervision.  Before data analysis, a rigorous data cleaning was done to ensure implausible and inaccurate data be treated.  Descriptive statistics such as measures of central tendencies and variation were used to describe the study population, estimate the prevalence and trend of attrition. Additionally, analytic studies were conducted to estimate the incidence, time to attrition and the various factors associated with attrition. The analytic study was mainly conducted using Kaplan Meier estimation and cox regression. The qualitative data was analyzed using the thematic analysis method.

Results: Data collection was conducted in 68 Woredas and a total of 3,476 HEWs personnel files were reviewed. The gross magnitude of attrition was found to be 21.1% for the year 2004-2019. Least attrition rate was observed in Harari (6.1%) followed by Dire Dawa (8.4%). Highest attrition saw in Addis Ababa 38.5%) followed by Afar (36.1%). The HEWs left their job in variety of ways. About 27.32% of them had formally submitted a resignation letter while about 40% of them have simply disappeared without notifying. Additionally, 3% of attrition was due to death while 4.7% of the HEWs were discharged from employment due to various reasons including absenteeism from work and poor work performance. Attrition rate for agrarian was 18.7% but it was highest for pastoralist and   urban, 31.9% and 21.2% respectively.  Attrition started in the year 2005, just one year after HEP initiation. Trend of attrition kept increasing except in the 2014, where it showed a slight drop. The increasing trend peaked in 2013 at 975 per 10,000 HEWs. The median time of attrition of HEWs workers in this study was 5.9 years with an inter-quartile range of 2.9 to 8 years. However, the overall median time of service of the HEWs included in the study was 7 years [IQR = 3 – 11.5]. The overall incidence rate of attrition was 288 per 10,000 person years [95% CI 267–309]. It was 182 per 10,000 person years [95% CI, 142 - 235] in the first year. The highest incidence rate was seen during the 8th year of implementation, 520 per 10,000 person years [95% CI 421 - 642]. The factors significantly associated with attrition were HEWs age, place of birth, type of deployment, HEWs current and deployment certification level, perceived distance of HP from Woreda Health Office, having a child, and COC status were able to independently predict time to the HEWS attrition. Personal and work related causes were identified as thematic cause for  attrition.

Conclusion: In general, a high level attrition was observed which could hamper the program implementation if actions are not taken immediately. Ensuring job competitiveness in terms of payment and other benefits is mandatory to keep the experienced and high performing workers.  The amount of time served before leaving could sharply decrease. Factors such as HEWs age, marital status and the number of children, level of certification and other factors could help identify those who are going to stay long serving. Considering these factors during selection and deployment could be beneficial for the program.

Keywords: Attrition, HEW, Ethiopia