Design: The study design is population-based cross-sectional analytical survey.
Setting: six sub-metros of Accra, Ghana - Ablekuma, Ashieduketeke, Osuklottey, Kpeshie, Ayawaso and Okaikoi.
Participants and study size: 2,814 randomly-sampled adult women (aged 18 and older). The population was stratified by age and social position, so as to ensure representation across all types of people. Older women (over the age of 55) were purposively over-sampled so as to achieve statistical power.
A complete description of the sampling methods has been described elsewhere.21
Variables: The primary outcome was symptoms of common mental disorder, as measured by two interview-administered self-report instruments: the short-form 36 (SF36) and the Kessler 6 (K6). Secondary outcomes include an exploratory analysis of the validity of the K6, prevalence of symptoms of psychosis and use of healthcare services. The exposure variables were age, education, wealth, ethnicity, region of birth, occupation, religion, number of pregnancies, and physical health.
The SF36 is composed of 36 questions divided across eight domains. Four of these domains are considered to relate to mental health, namely: 1) role limitations due to personal or emotional problems (“role-emotional”);, 2) emotional wellbeing, 3) social functioning and 4) energy and fatigue (see appendix 1 for full list of parameters corresponding to each of the mental health sub-scales). Higher scores indicate better health on each of the sub-scales. Together, these four scores form a single mental component scale, while the other four sub-scales aggregate into a physical component scale. Full details of the scales and the psychometrics of the SF36 are presented by Ware, Snow and colleagues.22
The tool and its scoring code book are available free of charge from the RAND Corporation (rand.org/health/surveys_tools/mos/mos_core_36item.html
) making it a good tool for use in resource-constrained settings. It is also widely used in health care studies.
The K6 is a six-question abbreviated form of the K10 scale, designed in the United States to estimate the prevalence of mental distress and disorder in the general population. Scores range from 0 to 30 with higher scores indicating better outcomes. The instrument has been used in 14 countries through the WHO's World Mental Health Survey, including South Africa (n=4,315) and Nigeria (n=2,143). It is easy to administer and easy to score, requiring a simple, unweighted sum of the responses. The predictive probability of a DSM-IV diagnosed mental disorder using the K6 was found to be 0.82 in Nigeria, the country most resembling Ghana in the WHO survey, meaning that 82% of people with a mental disorder were correctly screened positive.23
Specificity of the tool is not reported in the literature.
The K6 is not yet validated in Ghana, so no clinical threshold has been established to distinguish between mental distress and disorder.
Despite this drawback, the K6 offers the advantage of being a broader screening tool than some of the other locally validated mental health screens (eg. the PHQ-9, and the Edinburgh Postnatal Depression Rating Scale), because it is not specific to a single disorder. Since the SF36 and the K6 capture predominantly common mental disorders, four additional questions were asked about symptoms of psychosis experienced in the last month. The questions gauged the following symptoms: 1) feeling of strangeness; 2) paranoia; 3) thought control; and 4) hallucinations and are drawn from standardized instruments. Finally, epilepsy was assessed via a self-report question about seizures in the past year.
Wealth was measured by the Wealth Index, a composite scale taking into consideration household characteristics and durable goods. Physical health was captured by medication prescription, as well as by common somatic complaints of headaches and sleep disturbance.
Response bias was mitigated by choosing all-female interviewers who were native speakers of the local language. Interviews were conducted by ten women, aged between 25 and 45 years old who spoke the three main languages in Accra: Ga, Twi and Ewe. Care was taken to ensure the validity of the translation of the tool into these three languages by means of a focus group discussion with all the interviewers, all of whom were bilingual speakers of English and at least one of the local languages. The translations were then typed up as a reference document for the interviewers, while the data were entered on the original English language form. No back translation was conducted, because the translation had been done via group consensus; and group translation has been deemed more effective at identifying and addressing culturally ambiguous terms than individual translation and back translation.24
Individual translation service was offered to speakers of Hausa.
Statistical Methods: The internal consistency of the SF36 was examined by means of a Pearson correlation between the different sub-scales. On the basis of high levels of correlation in the mental health sub-scales, a principle component analysis was performed on the four mental health domains and a composite mental health factor score was produced for each woman in the sample. The K6 was correlated with the SF36 composite mental health factor by means of a Pearson correlation.
Risk factors for the SF36 and the K6 were calculated using a multivariate linear regression with categorical correlates. The analysis was done using the SAS statistical software.
Ethics: The interview was administered in the women's homes with their signed informed consent. Ethical approval was received from Noguchi Memorial Institute for Medical Research at the University of Ghanaand the Harvard School of Public Health .