Aims Infection remains the leading cause of neonatal mortality in developing countries. In Kenya, about 20% of neonatal deaths are attributable to sepsis. We aim to look at the epidemiological pattern of neonatal sepsis in a county referral hospital in Kenya.

Method Retrospective data was collected for all admissions to the Newborn unit between 2011 to 2014 in a county referral hospital in central Kenya. We calculated monthly rates of neonatal sepsis cases, mortalities, and case fatality rates for all admissions. We then plotted a monthly time series of sepsis cases and mortalities to determine if there was a seasonal trend over the four-year period. The epidemic time series was plotted and smoothed using a seasonal moving average estimator in Stata 12.1. The study was carried out during a Global Links RCPCH placement.

Results There were 1262 admissions to the Newborn Unit during the 4 year period. 23.9% of admissions had a diagnosis of neonatal sepsis. The overall mortality rate of admissions was 24.7%, whereas mortality attributed to sepsis was 18.2%. We observed a strong biannual peak in sepsis cases, with peaks in July 2012 and July 2014. Case fatality rates were highest in March 2012 (66.6%), July 2012 (50%), July 2014 (50%) and August 2014 (50%).

Conclusion The overall rate of neonatal mortality due to sepsis in this hospital is comparable to the national average. Our study indicates that sepsis cases correspond to a strong bi-annual pattern rather than a yearly one, with intermediate years yielding few sepsis cases. From this, we predict low sepsis rates in 2015, with a peak of cases in July 2016. This is the first time a bi-annual trend has been demonstrated for neonatal sepsis. We suggest further work should be done to analyse possible causes, including socio-political factors, for this bi-annual pattern for neonatal sepsis and mortality, and whether the same pattern can also be seen in other areas of Kenya over the same time period.