Health workforce attrition in the public sector in Kenya: a look at the reasons

BACKGROUND Kenya, like many other countries in sub-Saharan Africa, has been affected by shortages of health workers in the public sector. Data on the rates and leading reasons for health workers attrition in the public sector are key in developing effective, evidence-based planning and policy on human resources for health. METHODS This study analysed data from a human resources health facility survey conducted in 2005 in 52 health centres and 22 public hospitals (including all provincial hospitals) across all eight provinces in Kenya. The study looked into the status of attrition rates and the proportion of attrition due to retirement, resignation or death among doctors, clinical officers, nurses and laboratory and pharmacy specialists in surveyed facilities. RESULTS Overall health workers attrition rates from 2004 to 2005 were similar across type of health facility: provincial hospitals lost on average 4% of their health workers, compared to 3% for district hospitals and 5% for health centres. However, there are differences in the patterns of attrition rates by cadre. Attrition among doctors and registered nurses was much higher at the provincial hospitals than at district hospitals or health centres, whereas the opposite pattern was observed for laboratory and pharmacy staff (lost at a higher rate in lower-level facilities). In provincial hospitals, doctors had higher attrition rates than clinical officers, and registered nurses had higher attrition rates than enrolled nurses. In contrast, attrition of enrolled and registered nurses in district hospitals and health centres was similar. The main reason for health worker attrition (all cadres combined) at each level of facility was retirement, followed by resignation and death. However, resignation drives attrition among doctors and clinical officers; retirement accounts for the main share of attrition among nurses and pharmacy staff; and death is the primary reason for attrition among laboratory staff, particularly in district hospitals. One limitation of the data is that sampling of health centres was non-random and the results may thus not be representative of all health centres. CONCLUSION Our findings indicate that appropriate policies to retain staff in the public health sector may need to be tailored for different cadres and level of health facility. Further studies, perhaps employing qualitative research, need to investigate the importance of different factors in the decision of health workers to resign.