Sociodemographic correlates of depressive symptoms: a cross-sectional analytic study among healthy urban Ghanaian women

Depression is one of the most debilitating mental health disorders in the world. According to the World Health Organization (WHO), 350 million people worldwide suffer from depression, and it is a major contributor to the overall global burden of disease [1]. Unipolar depression has been identified as the second highest illness burden, accounting for as much as 8.6% of disability-adjusted life years lost, and is the single most important cause of disease-related disability among women in the reproductive age-group [2].

The literature is replete with data on the higher prevalence of depression among women [3]. Several factors explain this phenomenon, including sociodemographic and economic factors. Poverty, problems of conflict, and violence are higher among females and are associated with mental health problems and depression. Whilst some studies in Ghana have suggested a lack of association between marital status and depression [4, 5] others have reported that a significant proportion of the depressed are married women. Women in either abusive relationships or married to controlling husbands have higher odds of psychological distress [6]. There are also age correlates in reported depression. Among younger women, levels of depression are lower than among older women [7]. Although depression occurs at a much younger age among African women (between 20 and 40 years of age) compared to Europeans (between 35 and 55 years of age), depression is the most common mental health disorder among older men and women in Ghana aged 60 years and above [5]. These differences extend beyond menopause [8].

Education and employment are protective against depression and mental disorders in general. The risk of a common mental disorder is higher for the poor, the unemployed, individuals with low education, and the physically ill [3–5, 9, 10], although this finding has not always been consistent [11]. According to Canavan and colleagues [10], the lack of consistency among studies is largely due to dissimilarities in study designs within low and middle income countries (LMIC). This, in part, has accounted for our limited knowledge about the extent of the association between mental health and income. Few studies have been conducted on the prevalence of mental disorders in LMIC, and even fewer studies exist on specific mental disorders, such as depression. The majority of studies largely focus on identifying predictors or the treatment of depression [12, 13]. Most prevalence studies on mental health have used a small number of study subjects or are based at the facility level [14].

In most studies, researchers study clinical depression and are therefore likely to miss individuals who are burdened or who show minor signs of distress and depression that are not significant enough to be debilitating or to report to the hospital. Our primary objective was to measure the prevalence of self-reported depressive symptoms within the previous 30 days and to identify socioeconomic and sociodemographic factors associated with depressive symptoms among women in Accra in a large-scale study within an urban population. This study explores depressive symptoms among healthy women with diverse characteristics, allowing us to examine several background characteristics, including education and employment, as well as less examined characteristics, such as ethnicity and religion.

The data for this paper were obtained from the Women’s Health Study of Accra, Wave 1 (WHSA-1), a large-scale, cross-sectional study conducted in Accra, Ghana between 2003 and 2007 involving 3183 women aged 18 years and over. The study design is a population-based cross-sectional survey. Respondents were selected from the six sub-metropolitan areas that make up urban Accra. Accra is the capital of Ghana and is part of the Greater Accra Region, which includes the sub-metropolitan areas of Ablekuma, Ashiedu Keteke, Osu Klottey, Kpeshie, Ayawaso, and Okaikoi.

The sample consists of women 18 years and older who were randomly selected from enumeration areas categorized by socioeconomic status. Sampling was stratified by socioeconomic status and then by age. Older women above 55 years were over-sampled to increase their representation and to improve statistical power in the study. The selected women were administered a comprehensive household survey that included questions on general health (including sleep and feelings), risk factors for non-communicable diseases, physical activity, food security, reproductive health, sexual behaviour, health seeking behaviour, and family planning. A detailed description of the study variables and sampling procedures has been published elsewhere [15].

Measurement

The main outcome variable—the presence or absence of depressive symptoms—was derived from three primary questions that asked about common symptoms of depression: feelings of sadness or depression, sleep disturbances, and excessive worry. These were part of a set of items that measured mental health and well-being among the study participants. The items selected from the WHSA-1 have not been validated as a measure of depression for use in Ghana. Therefore, no clinical benchmark has been established on this scale to distinguish the severity of symptoms. While we are mindful of this limitation, it is important to note that the selected items are commonly associated with depression. Our interest was to obtain a measure that reflected presence or absence of depressive symptoms and therefore recalculated a composite score from the selected items. Each item was originally rated on a 4-point Likert scale: 0 (no symptoms), 1 (mild symptoms), 2 (moderate symptoms) and 3 (severe symptoms). The aggregated score of all three questions was calculated, and the mean was subsequently calculated to represent the score for depressive symptoms. Thus, an individual who reported severe symptoms on all three variables scored a total of 9 and a maximum mean of 3. The mean symptom score was re-coded into a binary outcome. Individuals with a mean symptom score of less than or equal to 1.0 were coded as not having depressive symptoms. Individuals with a mean symptom score of more than 1.0 were coded as having depressive symptoms. This binary outcome variable was used for chi square tests and logistic regression analyses. The distribution of scores is presented in Table 1 and in Fig. 1 below.

*Depressive Symptoms Total Scores	Frequency	Percent
0	1078	33.9
1	395	12.4
2	494	15.5
3	372	11.7
4	327	10.3
5	186	5.8
6	186	5.8
7	65	2.0
8	54	1.7
9	26	0.8
Total	3183	100

*Total scores were obtained from aggregate scores on three depressive symptoms. Sample questions include: Q1-How much of a problem did you have with sleeping, such as falling asleep, waking up frequently during the night, or waking up too early in the morning? (0 = None, 1 = Mild, 2 = Moderate, 3 = Severe or extreme). Q2-In the last 30 days, how much of a problem did you have with feeling sad, low or depressed? (0 = None, 1 = Mild, 2 = Moderate, 3 = Severe or extreme). Q3-In the last 30 days, how much of a problem did you have with worry or anxiety? (0 = None, 1 = Mild, 2 = Moderate, 3 = Severe or extreme)

A high prevalence of depressive symptoms was observed among the healthy participants, and depressive symptoms increased with age. Self-employed women or women engaged as students or apprentices had reduced odds of reporting depressive symptoms compared to unemployed women. Being an orthodox Christian appeared to offer protection against depressive symptoms. Despite the strong association between age, employment, and depressive symptoms, we caution against the assumption of causality. Proving causality in cross-sectional studies is statistically and empirically difficult.

We found a relatively higher prevalence of depressive symptoms in our study than has previously been reported [6, 11, 16–19]. The only exception we are aware of is the 41.1% figure reported for Vietnam [18]. One reason for this may have been the relatively few items used in our assessment of depressive symptoms. We distinguished depressive symptoms from clinical depression that would require a stricter criterion, as specified in the DSM 5 or ICD-10. Even so, the assessment of depressed mood, sleep disturbance, and worry or anxiety are prominent symptoms of depression and are usually adversely affected during depression. Therefore, the proportion of people who might show clinically relevant signs of depression and yet would not consider the symptoms serious enough to seek treatment is interesting. These early signs of depression may be due to high levels of stress associated with the hassles of urban living. Much evidence exists showing strong links between poor mental health and stress or urban living [20, 21]. Other studies have shown an increase in nuclear families in urban areas, which in turn is associated with increased violence among women and the mental health of poor women [19]. In other words, the social safety net provided by the extended family in less urban areas is either diminished or removed, exposing vulnerable individuals to high stress levels, possibly due to challenging economic circumstances.

Our results support earlier studies on the association between age and depression. For example, Deyessa and colleagues [11] showed that the odds of experiencing depression increase with age. Compared to younger women, older individuals have more anxieties stemming from multiple sources: marriage, children, stable employment, and income. Older unmarried women have more worries about marriage and childbirth, motivated by societal pressures on marriage, [22] particularly in Africa where marriage and motherhood are expected of every woman.

In the crude analysis—compared to women who never married—unmarried women (divorced, separated, or widowed) had significantly higher odds of reporting depressive symptoms. This finding supports earlier reports by others [11]. Arguably, these life changes, in the absence of adequate social support, will present substantial stress that can induce depression.

Contrary to what has been reported previously, we did not find an association between socioeconomic status and depression, despite adequate global reporting of this effect [4, 5, 23]. For example, a large scale longitudinal study in Belgium reported that increases in financial strain, poverty, and the end of cohabitation were associated with increased depressive symptoms [23]. We found some of these in our study, but our measure of socioeconomic status was based on the respective locality or census enumeration areas within which we selected our respondents. A lack of variability in the enumeration areas likely limited the different characteristics that might contribute to socioeconomic status. In Nigeria, a higher prevalence has been reported among rural dwelling people than among urban inhabitants [24]. While we did not compare between rural and urban populations, rural societies are associated with lower education and income, both of which are indicators of lower socioeconomic status. The measure of socioeconomic status is not consistent in many instances and that should be noted in future research.

We found a limited effect of ethnicity and religion on depressive symptoms; that is, the effect was reported only for specific groups, such as Akans and non–orthodox Christians. Compared to indigenous Ga women, the reduced odds of depression observed among women of Akan extraction in both the crude and the adjusted model are interesting and present an interesting subgroup worth exploring in future research.

The major finding of this study is the realization that more self-reported depressive symptoms, or mild forms of depression, might exist in supposedly healthy populations than has previously been reported. This has important implications for public health policy. In routine examinations, we should begin including an assessment of mood disorders among selected groups, especially among older adults or individuals under economic strain. The usefulness of early detection of mental health problems cannot be overemphasized. With this in mind, we can begin to institute referrals and interventions following early detection and diagnosis of depression. Counselling and psychological services are limited and expensive globally. In many countries, particularly LMICs, they are not covered by health insurance schemes, thereby making them inaccessible. When counselling and psychological services are placed in the public health domain, provision of care is easily accessible.

The prevalence of depressive symptoms among urban Ghanaian women is relatively high. This means that the reported risk burden of women’s mental health is substantially underestimated. This finding is also true for individuals who are older, unemployed, or for those with low education. We found a higher prevalence among older individuals than among younger individuals. Among specific groups, the unemployed and those with lower education also showed susceptibility to depression. We advocate further research on high risk groups, such as the elderly, those under the strain of poverty, the unemployed, and the lowly educated. We expect that increased knowledge on this would provide new directions in public health policy that could facilitate early detection and referral of people found to be at risk.