Community suicide rates and related factors within a surveillance platform in Western Kenya

Suicide is an important contributor to the burden of mental health disorders, but community-based suicide data are scarce in many low- and middle-income countries (LMIC) including Kenya. Available data on suicide underestimates the true burden due to underreporting related to stigma and legal restrictions, and under-representation of those not utilizing health facilities.

We estimated the cumulative incidence of suicide via verbal autopsies from the Health and Demographic Surveillance System (HDSS) in Kisumu County, Kenya. We then used content analysis of open history forms among deaths coded as accidents to identify those who likely died by suicide but were not coded as suicide deaths. We finally conducted a case-control study of suicides (both verbal autopsy confirmed and likely suicides) compared to accident-caused deaths to assess factors associated with suicide in this HDSS.

A total of 33 out of 4306 verbal autopsies confirmed suicide as the cause of death. Content analysis of a further 228 deaths originally attributed to accidents identified 39 additional likely suicides. The best estimate of suicide-specific mortality rate was 14.7 per 100,000 population per year (credibility window = 11.3 - 18.0). The most common reported method of death was self-poisoning (54%). From the case-control study interpersonal difficulties and stressful life events were associated with increased odds of suicide in both confirmed suicides and confirmed combined with suspected suicides. Other pertinent factors such as age and being male differed depending upon which outcome was used.

Suicide is common in this area, and interventions are needed to address drivers. The twofold increase in the suicide-specific mortality rate following incorporation of misattributed suicide deaths exemplify underreporting and misclassification of suicide cases at community level. Further, verbal autopsies may underreport suicide specifically among older and female populations.