Depression and its psychosocial risk factors in pregnant Kenyan adolescents: a cross-sectional study in a community health Centre of Nairobi

Being an adolescent: Risk factor for both perinatal health and depression

Pregnancy is ordinarily a natural developmental process however evidence shows that a significant portion of adolescent fertility is unintended–either unwanted or mistimed–across countries in SSA [9]. Adolescent pregnancy is associated with poor health outcomes, including maternal deaths and injuries and adverse infant outcomes [11]. Eleven percent of all births occur in women between 15 and 19 years [12] with lower-and-middle income countries (LMICs) accounting for 95% of births before 18 years. Almost 10% of girls in LMICs become mothers by the age of 16, with the highest prevalence rates of early pregnancies seen in Sub-Saharan Africa (SSA) and in South Central and Southeast Asian regions [13] In Kenya, 26% of adolescents become mothers between 15 to 19 years of age [14] About 41% of these pregnancies are unintended, 26% are mistimed and 15% are unwanted [9]. These unintended and early pregnancies also make adolescents prone to STIs and HIV, adolescents experiencing an unwanted pregnancy are more likely to resort to abortions, which are often illegal and done by unskilled attendants leading to several pregnancy and birth-related complications [15]. Adolescent pregnancy predisposes young girls to adverse mental health [3] with depression and anxiety being the most common mental health disorders [16, 17]. Worldwide prevalence of depression during pregnancy is estimated to be between 11 and 18% [4, 18]. A higher prevalence is reported in women in LMICs with point prevalence of 15.6% during pregnancy and 19.8% post-partum [19, 20]. Major depressive disorder (MDD) accounts for approximately 40% of YLDs yet mental health services in Sub-Saharan Africa are mostly restricted to tertiary psychiatric facilities [21–23]. Recent studies have shown that the prevalence of depression across sub-Saharan Africa to be high with antenatal depression reported to be 27 and 33% respectively in Ghana and Côte d’Ivoire [24]. In Ethiopia, the prevalence of antenatal depression was 24.94% [25] and 39.5% in Tanzania [26]. In Kenya, a longitudinal study of women attending maternal and child health clinics in two major public hospitals in Nairobi (using Kiswahili version of EPDS with a cut off of 13 or more) found a prevalence of 18% for antepartum depression [27]. Given this context, and how rampant adolescent pregnancies are in LMICs like Kenya, the clear implementation gap has been to provide timely integrated sexual and reproductive health, maternity and mental health services in these informal settlements. For this implementation gap to be addressed we need robust epidemiological studies that look at depression prevalence, its severity and the associated risk factors that aggravate mental illness in this vulnerable group.

Depression and mental health problems in pregnant adolescents

In Kenya and other SSA countries, adolescent pregnancy is widely prevalent public health problem which poses a significant mental health burden. Despite this only a few studies on adolescent mental health needs during perinatal context can be found. Women living in LMICs are exposed to multiple risk factors for depression as their adverse life circumstances commonly include poverty and lack of family structure, low level of education, high rates of HIV/AIDS diagnosis, stigma and lack of social support [28–31]. There is a huge evidence now which suggests that the risk of depression is heightened when there is a previous adolescent pregnancy, low partner age, and adolescents whose mothers themselves had early unintended pregnancies [32, 33]. Furthermore, there is evidence of an association between a mood disorder and concomitant alcohol and substance use in adolescents in SSA which presents additional pregnancy and birth related risks [34]. One study reported that 43.7% of adolescents attending public secondary schools sample in Nairobi had clinically significant depression scores; nearly the same proportion were detected with moderate to severe anxiety along with depression symptoms [35]. Our study aims to determine the prevalence of depression in pregnant adolescents living in a resource-poor urban settlement of Nairobi County and an associated objective is to identify related psychosocial risk factors accompanying depression in this vulnerable population.

Study design, population and setting

We carried out a cross-sectional study assessing depression and associated psychosocial risk factors in pregnant adolescents attending a Maternal Child Health clinic at a Nairobi community health care center located within the informal settlement. The facility is operated through the County Council of Nairobi giving free maternity services and caters for low-and-middle income wage earners from nearby informal settlements. It receives between 12 and 15 pregnant women every day and operates every weekday. We recruited 176 participants between ages 15-18 years using a prevalence rate of 13% from a study [36]; using Cochran sample size estimation (1977) for a cross-sectional study keeping alpha at p < 0.05.

Study procedures

This study was approved by Kenyatta National Hospital and University of Nairobi Ethics Review Committee (approval no. P499/07/2015). We collected data between November, 2015 to January, 2016. The study’s purpose was explained to the participants and their caregivers. A written informed consent was signed based on willingness to participate in the study and for future uses of data, such as publication, preservation and long-term use of research data. About 6.8% (n = 12) of pregnant adolescents we approached during data collection declined to participate in the study and despite our assurance of anonymity and confidentiality, it was mostly due to a strong feeling of stigma associated with their pregnancy. We suspect that their partners or caregivers prohibited them from participation. Adolescent pregnancy has a significant social stigma associated with it in the community context where we were working.

Once data was collected, confidentiality was assured by other means such as by de-identifying the data by use of serial numbers and the paper data was kept in sealed boxes in the Department of Psychiatry office. The study was conducted in a small cubicle within the ANC clinic in the health facility. Participants completed the researcher-designed demographic and psychosocial risk factors questionnaire and PHQ-9 Kiswahili version. There was no incentive given for participation. The lead author, who carried out this study as part of her postgraduate research, is a trained clinical psychologist with wide experience in working with adolescents and young children in the informal settlements of Kenya.

The study team discussed the data collection, referral mechanisms and use of the tools with the lead researcher who collected all the data herself. The participants recording high scores with a cut off of 15+ on the PHQ-9 were referred to one of two places: (a) Psychiatric clinic operated on Wednesdays within the facility or (b) provided alternative of the Youth/Reproductive Health Centre at Kenyatta National Hospital where they could be seen free of charge. Within this inquiry, we had a nested small qualitative study exploring participants’ interpersonal, practical and cultural challenges and barriers to accessing depression and general mental health care. The substantive findings of this qualitative inquiry are under process [37].

Measures

Statistical analysis

SPSS version 22 [43] was used in data analysis. The association between depression and its psychosocial correlates was determined in two ways. Firstly, we divided our sample into two groups (depressed and non-depressed according to PHQ-9 cut-off score 15+ or more) and compared these groups using chi-square test. Secondly, we assessed each potential correlates with the PHQ-9 score using independent samples t-test and ANOVA. Hierarchical multivariate linear regression analyses were performed to determine the independent predictors of depression from the psychosocial factors that were significantly associated with depression at the univariate analyses. We ran regression analysis by entering the participants’ socio-demographical variables into Block 1, followed by other characteristics/ conditions in Block 2. There were no missing data for all the independent and dependent variables. Prior to running the analysis, all assumptions were checked including univariate/multivariate normality, linearity, homoscedasticity and diagnostic testing for multi-collinearity and independence of errors. After checking for univariate normality, the PHQ depression scores was transformed by a two-step approach using inverse distribution function (IDF) using maximum likelihood estimator (MLE) in which we retained the original series mean and standard deviation to improve the interpretation of results. The level of statistical significance was kept at P < 0.05, all tests were two sided.

Sample characteristics

Depression prevalence

The EPDS scale was used to screen the risk of depression in our pregnant adolescent sample and we found 58% falling under the category of a likelihood of depression. This was confirmed with the PHQ-9 diagnostic and severity assessment tool when a total of 33% (n = 58) tested positive for clinical depression. Out of 176 participants, 38 (21.6%) had no depression, 42 (23.9%) had mild depression, 38 (21.6%) had moderate depression, 30 (17%) had moderate severe depression while 28 participants (15.9%) were found to have severe depression (see Fig. 1).

Associations between socio-demographic factors and adolescent depression

We used PHQ-9 continuous scores as our main outcome to test associations with socio-demographic factors. On chi-square test of association, the age of the participants was significantly associated with their depression status, χ²_{(1, N = 176)} = 8.84; P = 0.012. All of the 27 younger participants (ages 15-16) were depressed, as were the majority of older participants. Of the 176 pregnant adolescents of ages 15-18 years sampled in the study, 32.9% (n = 58) tested positive for a depression diagnosis using PHQ-9 cut-off of 15+. We detected a significant difference those who were depressed and not depressed using PHQ-9 cut off on educational level, marital status, living with parents, monthly income (where we found an inverse relationship, adolescents who were in households earning more than 15,000 kes/per month had higher depression, χ² _{(1, N = 176)} = 27.59; P < 0.001), housing type, receiving social support, experienced stressful event or substance abuse, attending ANC clinic, experiencing domestic violence and having a HIV positive status (see Table 1).

Overall psychosocial risk factors model using multivariate linear regression

The results of multivariate linear regression using PHQ-9 scores as the dependent variable and 14 predictors in two blocks are shown in Table 2. The overall model with all the predictors were statistically significant and explained 55.4% of the variance in depression among the participants with F (17, 157) =11.46, P < 0.01. In Block 1, the respondents socio-demographic characteristics i.e. age, occupation, marital status, living arrangements and income explained 37.2% of the variance in depression scores, which was statistically significant with F (8, 166) = 12.31, P < 0.01), in which age, type of housing and marital status were statistically significant P < 0.05. In Block 2, other participants characteristics uniquely explained a statistically significant amount 18.1%, of the variance of the of participants depression after controlling for socio-demographical factors in Block 1, with F (9, 157) =7.10, P < 0.01. Therefore, the two blocks of variables significantly contributed to the prediction of the participant’s depression. After various iterations, when individual predictors using standardized beta scores were examined, having experienced a stressful life event (B = 3.27, P = 0.001, β =0.25), the absence of social support for pregnant adolescents (B = − 2.76, P = 0.008, β = − 0.19), being diagnosed of HIV/AIDS (B = 3.81, P = 0.004, β =0.17) and being of younger age (B = 2.46, P = 0.038, β =0.14) were the independent correlates of depression.