Mental health symptoms and their relations with dietary diversity and nutritional status among mothers of young children in eastern Democratic Republic of the Congo

Background: In developing countries, maternal mental health problems have been linked to sub-optimal child feeding practices and child underweight and stunting, but little is known about how maternal mental health is associated with mothers’ own diets and nutritional status. The objective of the study was to investigate the association between mental health symptoms and diet and nutritional status of mothers of young children in South Kivu, DR Congo. Methods: Participants were 828 mothers of young children enrolled in a larger, quasi-experimental study evaluating a multi-year food security and nutrition project. The present analysis was conducted with cross-sectional data collected from 2015 to 2016. We assessed symptoms of anxiety and depression using the Hopkins Symptom Checklist-25 (HSCL-25) and post-traumatic stress disorder (PTSD) with the Harvard Trauma Questionnaire (HTQ), using a four-point Likert scale. Mean scale scores were calculated ranging from one to four. A variable was created for high distress (participants scoring in the upper quartile of both measures). Dietary diversity scores were calculated from the number of food groups (range zero to ten) consumed the previous day, identified from an open recall. Nutritional status was measured by body mass index (BMI) and underweight (BMI < 18.5 kg/m2, or mid-upper arm circumference < 23 cm for pregnant women). Bivariate and multivariate (adjusting for parent study intervention group, education, age, health, parity, livelihoods zone, and territory of origin) regression analyses were conducted. Results: Maternal mental health measures were positively and statistically significantly associated with higher dietary diversity scores in adjusted analyses (HSCL-25: ß= 0.18, p = 0.002, HTQ: ß= 0.12, p = 0.029, High Distress: ß= 0.47, p < 0.001). Mental health symptoms were not significantly associated with BMI (HSCL-25: ß = - 0.04, p = 0.824; HTQ: ß = 0.02, p = 0.913; High distress: ß= - 0.02, p = 0.938) or underweight (HSCL 25: OR = 0.91, p = 0.640; HTQ: OR = 1.03, p = 0.866; High distress: OR = 0.78, p = 0.489). Conclusions: More severe maternal mental health symptoms were associated with higher dietary diversity but not nutritional status, and the reasons for these findings are not clear from available data. More research is needed to identify underlying factors that could influence mental health symptomatology and diet quality among food insecure and extremely resource-limited populations. The present study is a sub-study of a larger, quasi-experimental evaluation of a United States Agency for International Development (USAID) food assistance program called Jenga Jamaa II [30]. Jenga Jamaa II was designed to improve household food security and child nutrition in Uvira and Fizi territories in South Kivu through four distinct nutrition-specific and nutrition-sensitive interventions, and was implemented by the non-governmental organizations (NGOs) Adventist Development and Relief Agency (ADRA) and World Vision International from 2011 to 2016. Enrollment of project beneficiaries and control group participants for the parent study occurred from August to October 2012, with 1820 households enrolled and followed for three and a half years. Data were collected via eight cross-sectional surveys occurring in August/September and February/March of each year to account for seasonal variation in food security. More details related to the parent study can be found elsewhere [30]. The present study is a cross-sectional sub-study utilizing data collected during the last two Jenga Jamaa II surveys conducted in September 2015 and March 2016. Participants from the Jenga Jamaa II parent study who were mothers of children under five also enrolled were eligible for the sub-study. Individuals were excluded from the study if they were not enrolled in the parent study or were not the biological mother of a child also enrolled in the parent study. Maternal mental health was assessed at one time point (September 2015 for the majority of the participants), but maternal diet and weight data were utilized from both the September 2015 and March 2016 surveys (among participants present at both) to account for seasonal variation. Data were collected electronically using the mobile data collection application Magpi and Android tablets provided by ADRA [31]. The Institutional Review Board of the Johns Hopkins Bloomberg School of Public Health approved all data collection instruments and procedures. At every data collection encounter for the parent study, study staff obtained oral consent in Swahili, the predominant local language, and participants were reminded that they could decline to participate at any time. Additional oral consent was requested for participation in the sub-study, in which the mental health questionnaire was described as well as the potential risks and benefits of participation. ADRA field agents who served as study enumerators received training on the questionnaire and on research ethics prior to data collection. Independent variables assessed were depression/anxiety symptoms, post-traumatic stress disorder (PTSD) symptoms, and a variable (high mental distress) constructed to identify participants with high levels of both depression/anxiety and PTSD symptoms, having mean item scale scores in the upper quartile of both measures. The Hopkins Symptom Checklist (HSCL-25) was used to assess depression symptoms (15 items) and anxiety symptoms (10 items), and the Harvard Trauma Questionnaire (HTQ) was used to assess PTSD symptoms (16 items) [32, 33]. One item on suicidality was dropped from the HSCL-25 due to ethical considerations. Participants rated the frequency of each symptom in the prior four weeks on a four-point Likert scale, and mean item scores were calculated for depression/anxiety (HSCL-25) and PTSD (HTQ), with a possible range of one to four. The mental health questionnaire that included the HSCL-25 and HTQ tools was adapted from an earlier study in South Kivu that evaluated the impact of a cognitive behavioral therapy intervention among female survivors of sexual violence [34]. The questionnaire was administered in Swahili, the predominant local language. The HSCL-25 is frequently used in a variety of cultural settings [7, 35–37], and the validity of a Swahili version of the scale has been evaluated in a sample of Tanzanian women using content and construct validation methods [36]. Additionally, it has been used to assess depression symptoms among Congolese refugees in the United States [38]. The validity of the Harvard Trauma Questionnaire has been assessed in multiple settings and is often used among populations who have experienced conflict and displacement, such as refugees [33, 39–41]. After data collection was complete, the internal consistency reliability of mental health measures was assessed using Cronbach’s alpha [42]. The HSCL-25 items had a scale reliability coefficient of 0.92 and the HTQ items had a scale reliability coefficient of 0.91. The anxiety and depression subscales of the HSCL-25 had a correlation coefficient of 0.72 (r(826), p < 0.001). The correlation coefficient of the HSCL-25 and HTQ was 0.82 (r(826), p < 0.001). Maternal BMI and underweight were used as measures of maternal nutritional status. Maternal height, weight, and mid-upper arm circumference (MUAC) were measured by trained ADRA field agents using standard protocols [43]. BMI was calculated for non-pregnant participants, and MUAC was used to assess nutritional status of pregnant participants [44]. Weight was averaged for the two data collection periods if participants were present at both surveys, or a sole weight measure was used for participants who were only present at one survey. Pregnant mothers with MUAC < 23 cm and non-pregnant mothers with BMI < 18.5 kg/m2 were classified as underweight [44]. Participants were measured using a Model 1582 Tanita Mommy and Baby Infant Scale (Arlington Heights, IL) and a Shorr Productions (Olney, MD) height board. Dietary diversity scores were used to measure diet quality, using the Minimum Dietary Diversity for Women (MDD-W) tool [45]. Enumerators asked participants to list all foods consumed the previous day and night. When composite dishes were mentioned, they asked for a list of ingredients and probed for additional items. All of the food items were recorded in Swahili, and then translated to English and classified into one of ten possible food groups: 1) starchy staples (grains, white roots, tubers, and plantains); 2) pulses (beans, peas, lentils); 3) nuts and seeds; 4) dairy; 5) meat, poultry, and fish; 6) eggs; 7) dark green leafy vegetables; 8) other vitamin A-rich fruits and vegetables; 9) other vegetables; and 10) other fruit. The number of food groups consumed was summed to create a dietary diversity score, with a possible range of zero to ten, with higher scores indicating greater diversity. Maternal dietary data were collected at two time points six months apart, and an average score was calculated, in order to address seasonal variation. The mental health questionnaire included a section on background and demographic characteristics: age, years of education obtained, ethnic group, living in territory of origin, currently pregnant, number of children, and marital status. Education was recoded as a categorical variable with three categories: no education, completing at least some primary school, and completing at least some secondary school or higher education. Participants were also asked if they had a child that died, and to rate their physical health status, using a scale that ranged from excellent to poor. Household-level data, including household size, income in the past month, and food insecurity, were collected as part of the parent study survey questionnaire. Food insecurity was measured using the Household Food Insecurity Access Scale (HFIAS) [46]. Households were classified in categories ranging from food secure to severely food insecure based on HFIAS score. Participants were also asked their income in the past month. Household income was not included in the final analysis because it was not necessarily reflective of socioeconomic status; in this context it may have represented the sale of assets due to hardship or food insecurity. Indicator variables for intervention groups were created for the four parent study interventions and the control group. Indicator variables for geographic region (Uvira or Fizi territory) and livelihoods zone (plains, mountains, or lakeside) were also included. Data were analyzed using Stata 13 [47]. Distributions of continuous variables and frequencies of categorical variables were explored, and outlying values were identified. The three dependent variables assessed were dietary diversity score, BMI, and underweight. Dietary diversity score was a continuous variable. BMI was continuous and limited to non-pregnant participants. Underweight was constructed as a binary variable. Independent variables for maternal mental health were mean item HSCL-25 score (measuring depression and anxiety symptoms), mean item HTQ score (measuring PTSD symptoms), and a binary variable for high psychological distress (upper quartile of both measures). Separate analyses were conducted for each of the three independent variables due to multicollinearity. Bivariate regression analyses were conducted between demographic/socioeconomic measures and independent and dependent variables (Additional file 1: Tables 4 and 5). Potential confounding variables were selected for inclusion in the model if they were associated with both independent and dependent variables, or if they had a conceptual relationship to both. Multivariate linear and logistic regression analyses were conducted, adjusting for potential confounding variables.