Time series analysis of sexual assault case characteristics and the 2007-2008 period of post-election violence in Kenya

Abstract

BACKGROUND: Following the declaration that President Mwai Kibaki was the winner of the Kenyan presidential election held on December 27, 2007, a period of post-election violence (PEV) took place. In this study, we aimed to identify whether the period of PEV in Kenya was associated with systematic changes in sexual assault case characteristics. METHODS AND FINDINGS: Medical records of 1,615 patients diagnosed with sexual assault between 2007 and 2011 at healthcare facilities in Eldoret (n = 569), Naivasha (n = 534), and Nakuru (n = 512) were retrospectively reviewed to examine characteristics of sexual assault cases over time. Time series and linear regression were used to examine temporal variation in case characteristics relative to the period of post-election violence in Kenya. Key informant interviews with healthcare workers at the sites were employed to triangulate findings. The time series of sexual assault case characteristics at these facilities were examined, with a specific focus on the December 2007-February 2008 period of post-election violence. Prais-Winsten estimates indicated that the three-month period of post-election violence was associated with a 22 percentage-point increase in cases where survivors did not know the perpetrator, a 20 percentage-point increase in cases with more than one perpetrator, and a 4 percentage-point increase in cases that had evidence of abdominal injury. The post-election violence period was also associated with an 18 percentage-point increase in survivors waiting >1 month to report to a healthcare facility. Sensitivity analyses confirmed that these characteristics were specific to the post-election violence time period. CONCLUSION: These results demonstrate systematic patterns in sexual assault characteristics during the PEV period in Kenya