The study was carried out in Oyo State one of the 36 states in Nigeria. This community based study is cross sectional in design and aimed at collecting data on mental health in rural and urban Oyo State, Nigeria.
Sampling procedure
Multistage sampling technique was used to obtain a representative sample of the communities in Oyo state. The communities where the study was carried out were chosen as follows:
Stage 1
A sampling frame of all the local government areas in Oyo state was drawn and stratified into urban and rural areas based on World bank classification [9]. One rural and two urban local government areas was obtained by simple random sampling (balloting). This is based on the fact that two thirds of Oyo state is urbanized. Ibadan North-West, Egbeda and Saki-East local government areas were selected.
Stage 2
Sampling frame of all the communities in the selected local government areas was drawn. The communities where the study was carried out were randomly selected by simple random sample (balloting). The communities selected were Idikan in Ibadan North-west LGA, Olubadan Estate in Egbeda LGA and Ago-amodu in Saki-East LGA.
Stage 3
Using the PHC house numbering where available (in places where it has not been done the houses were numbered for the purpose of the study). Systematic sampling technique was employed to select the houses that were visited in the chosen communities. Seventy-four houses were selected in Idikan, eighty-five houses in Olubadan Estate and ninety-eight houses in Ago-amodu.
Stage 4
One household in each of the houses selected were recruited into the study.
Stage 5
Every resident aged 15 years and above who has resided in the area for at least 6 months was interviewed in the households selected. A total of one thousand, one hundred and five subjects were recruited into the study.
A sample size formulae for comparing two proportions was used to obtain the sample size. Prevalence of 12.0% of poor mental health using GHQ 12 questionnaire among clinical students of University of Ibadan was used as estimate for urban community, while the prevalence of 21.3% among rural primary health care patients in Nigeria was used for rural community [4, 5]. A precision of 95% is desired with a power of 90%. The calculated sample size was 610 while this was doubled to 1220 with response rate of 90.6% (1105 responses).
Data collection
The study was conducted using an interviewer administered structured questionnaire. The GHQ-12 was used to assess mental health status of the respondents. Scores were calculated with a 0-1-1 scale with a maximum score of 1 and a minimum score of 0 for each item. A score of three or more was used as cut-off to classify into good and poor mental health. WHO quality of life questionnaire is a five point scale with items which ranged in rating from "very poor", "not at all" or "very dissatisfied" (1 point) to "very good", "extreme amount" or "very satisfied" (5 points). For items with reverse scores "not at all" was scored 5 and "extreme amount" was scored 1. The score for both mentally healthy and mentally ill was computed to asses the effect of psychiatric morbidity on quality of life.
This questionnaire was translated into the local language for easy administration and translated back to English to ensure accuracy of translation. The GHQ (12) and WHO quality of life questionnaires were administered by research assistants after adequate training and the author's supervision. Research assistants were recruited from the communities and were trained to administer the questionnaires. The research assistants were recruited based on minimum qualification of OND certificate. They were trained on how to extract the information on psychiatric symptoms and their relevance to the research work. They were also guided through a good number of questionnaires until a reasonable level of competence had been attained before being left to do it themselves.
Data generated in the study were manually cleaned and then entered into the Computer using SPSS 10 statistical software for analysis. Logistic regression analysis was done to determine factors associated with mental ill health in rural and urban Oyo State and also to remove the effect of confounding variables. The dependent variable was psychiatric morbidity as a dichotomous variable with a score of less than 3 being an option indicating good mental health while a score of 3 and more being another option indicating abnormal or psychiatric morbidity. All the variables which were significant in the bivariable analysis with a p < 0.08 were fed into the model. Odd ratios were adjusted and p values of <0.05 were taken as significant for the study.