The study sample was the 50,000 population living in Maseno. This is a rural agricultural area of western Kenya, on the edge of Lake Victoria. A one in fifty random sample was drawn of all 50,000 households in Maseno to give a projected sample of 1,000 households, using the 1999 Kenya Census conducted by the Government of Kenya in which all households had been enumerated. Each of the sampled households was then visited, all members recorded and on the same day one eligible person aged 16–65 selected at random for interview.

Efforts were made to use local capacity. Pencil and paper administration of the survey was coordinated by the Ministry of Health section for surveys. Interviews were conducted in 2004 by lay volunteer community health workers linked to primary care centres in Maseno. Interviewers received a brief orientation to the instrument and were trained in its use. Responses were recorded verbatim, scored and entered into an SPSS database. The CIS-R was administered by community health workers, supervised by a public health nurse working in Chulaimbo rural health training centre in Maseno. Informed consent was obtained from all respondents. 876 interviews were completed and the response rate was 87.6%.

The survey gathered information about demographic and socioeconomic factors, common mental disorders (using the Clinical Interview Schedule–Revised (CIS-R)) [20], psychotic symptoms (using the Psychosis Screening Questionnaire (PSQ)) [21], and use of alcohol, drugs and tobacco [22,23]. Related articles will report the prevalence of psychotic symptoms and substance misuse in the population. All diagnostic categories of mental disorder included in the current paper were based on ICD 10 [24]. These ICD 10 diagnoses were derived by computer algorithms based on the combination of individual symptom scores, and were not allocated by the field interviewers. The CIS-R is a gold standard instrument for assessing psychopathology in community settings, which has been widely used in both rich and poor countries [25], and is designed to be used by lay interviewers. The instrument had not been used in Kenya before, so it was subjected to scrutiny by local mental health professionals and researchers and Ministry of Health survey officers to determine its suitability. It has been used recently in Tanzania [26]. The CIS-R uses an initial filter question about symptoms in the last month, and then asks frequency, duration and severity of each symptom in the last seven days. Calculation of prevalence is based on the last seven days. It provides diagnoses of depressive episode (mild, moderate or severe), obsessive compulsive disorder, panic disorder, phobic disorder, generalised anxiety disorder and mixed anxiety/depressive disorder. These diagnoses were the basis for an overall category of common mental disorder (otherwise non-psychotic disorder or neurosis). The numbers were not large enough for separate analyses of risk factors for individual diagnostic categories.

Data was analysed using SPSS software for Windows Version 15. Chi squared (χ²) tests examined demographic and socio-economic differences between the areas. The raw data were weighted. The weights were calculated to take account of selection bias due to household size and to correct for the oversampling of Head of Household (HoH) and spouse, by weighting down those with a status of HoH/spouse. The final weighting variable did adjust for patterns of non-response in terms of differential probabilities of selection within households (i.e., respondents living in households with more residents were weighted up due to the fact that they had a lower chance of selection), and adjusting the profile of the achieved sample to match that from Census. Odd ratios (ORs) with 95% confidence intervals (CIs) were calculated to determine significant associations with the presence of any CMD (any CMD is slightly different rate to 12+ score) and then “any CMD” was examined as the dependent variable in multi-variate logistic regression.

Approval was granted by Mathari National Mental Hospital, Ministry of Health, Kenya, and Maudsley (SLaM), National Health Service (NHS) Foundation Trust.

In the current study, there was no difference in the prevalence of CMD between genders. In Nigeria, Gureje and colleagues also failed to demonstrate a difference between genders and, although consistent with Britain [29] and other parts of the world such as Brazil [30], higher rates among women have been found in Africa including Zimbabwe [31], Ethiopia [32], and South Africa [33,34,35]. Similar to findings in Tanzania [26] older age (30+ years) was associated with higher odds of CMD compared to people aged 16–24, while in Ethiopia rates were highest in people aged 35–44 [32], and 50–64 in Nigeria [36]. Patel and colleagues [37], in four low- and middle- income countries, and Lima and colleagues [30] in Brazil also found higher rates in older age groups. In Britain, the peak age groups were 50–54 for men and 45–49 for women [38].

Marital status was not significantly associated with disorder. The lack of significant association is consistent with recent reports from Nigeria [36] and South Africa [33], although previous studies have found higher rates in the widowed, separated and divorced in a number of developing countries [37].

Co-morbidity between mental illness and physical illness is a common finding in epidemiological surveys in rich, middle and low income countries, and Kenya is no exception [39].