Maternal depression and early childhood growth in developing countries: systematic review and meta-analysis
Pamela J Surkan, Caitlin E Kennedy, Kristen M Hurley & Maureen M Black
Volume 89, Number 8, August 2011, 608-615E
Table 1. Studies included in systematic review of maternal depression and child growth in developing countries, 1996–2010
| Geographical region and study | Country, study design and sample sizea | Depression measure and timing of assessment | Growth measure and timing of assessment | Findings | Notes |
|---|---|---|---|---|---|
| Southern or south-eastern Asia | |||||
| Anoop et al. 2004 |
India, case–control (matched); n = 144 (72 malnourished cases and 72 well nourished controls) | Measure: major depression diagnosis using the DSM III R (SCID). Timing: postpartum depression 1 month after birth. Current depression during the past month – after growth assessment | Measure: 50–80% versus > 80% of expected weight-for-age. Timing: 6–12 months postpartum | Significant finding: postpartum depression, underweight: adjusted OR: 7.4, 95% CI: 1.6–38.5. Non-significant finding: current major depression, underweight: adjusted OR: 3.1, 95% CI: 0.9–9.7 | Note: Interactions between current maternal depression and low birth weight and between postpartum depression and low maternal intelligence were statistically significant. The severity of malnutrition was significantly associated with postpartum depression and low maternal intelligence. Covariates included maternal intelligence, low birth weight, socioeconomic status, duration of exclusive breastfeeding, duration of supplementary breastfeeding, immunization status and mother’s literacy. |
| Black et al. 2009b, |
Bangladesh, cross-sectional, n = 221 | Measure: CES-D using a cut off of ≥ 16. Timing: 12 months postpartum | Measure: underweight, height-for-age z-score and weight-for-height z-score; stunting, height-for-age z-score < −2. Timing: 6 and 12 months postpartum | Significant findings: 12-month height-for-age z-score, adjusted β: –0·01 (P = 0·01); stunting, unadjusted OR: 2.2, 95% CI: 1.3–3.8. Non-significant finding: 12-month weight-for-height z-score, adjusted β: –0·01 (P > 0.05) | Note: The relationship between depressive symptoms and infant growth was partially mediated by caregiving. Covariates at 12 months included 6-month weight-for-height z-score, maternal education, poverty status, infant sex, birth order, receipt of zinc or iron, HOME score, maternal perceptions of infant temperament and months of breastfeeding (models were also presented without either infant temperament or the HOME score or both). |
| Harpham et al. 2005 |
India, cross-sectional, n = 1 823 | Measure: SRQ-20 using a cut off of 7–8. Timing: 6–18 months postpartum | Measure: underweight, weight-for-age z-score < −2; stunting, height-for-age z-score < −2. Timing: 6–18 months postpartum | Significant finding: stunting, adjusted OR: 1.4, 95% CI: 1.2–1.6. Non-significant finding: underweight, adjusted OR: 1.1, 95% CI: 0.9–1.4 | Note: Covariates included maternal age and education, child sex, age and birth weight, and household composition, wealth index and geographical location. |
| Harpham et al. 2005 |
Viet Nam, cross-sectional, n = 1 570 | Measure: SRQ-20 using a cut off of 7–8. Timing: 6–18 months postpartum | Measure: underweight, weight-for-age z-score < −2; stunting, height-for-age z-score < −2. Timing: 6–18 months postpartum | Significant finding: underweight, adjusted OR: 1.4, 95% CI: 1.1–1.8. Non-significant finding: stunting, adjusted OR: 1.3, 95% CI: 0.9–1.7 | Note: Covariates included maternal age and education, child sex, age and birth weight and household composition, wealth index and geographical location |
| Patel et al. 2003a, |
India, hospital-based cohort, n = 171 (23% of infants had depressed mothers) | Measure: EPDS using a cut off of 11–12. Timing: 6–8 weeks postpartum | Measure: underweight, weight-for-age < 5th percentile; stunting, height-for-age < 5th percentile. Timing: 6–8 weeks and 6 months postpartum | Significant findings: underweight, adjusted OR range: 2.7–3.6; stunting, adjusted OR range: 3.2–3.8 | Note: Significant findings remained after adjusting for the following confounding variables one by one: maternal and paternal education, infant birth weight, sex and feeding practice, infant illness in the first 6 weeks of life and prematurity. |
| Rahman et al. 2004 |
Pakistan, case–control, n = 172 (82 cases, 90 controls). Adjusted analyses: n = 107 (48 cases, 59 controls) due to missing data | Measure: SRQ-20 using a cut off of ≥ 10. Timing: infant at mean 9.7 months (SD: 0.9) postpartum | Measure: underweight, cases, weight-for-age < 3rd percentile; controls, weight-for-age > 10th percentile. Timing: mean 9.7 months (SD: 0.9) postpartum | Significant findings: underweight, unadjusted OR: 3.9, 95% CI: 1.9–7.8; adjusted OR: 2.8, 95% CI: 1.2–6.8 | Note: Covariates included infant birth weight, number of young children in the household and socioeconomic status. |
| Rahman et al. 2004 |
Pakistan, prospective cohort, n = 320 infants (160 with depressed mothers,160 with psychologically well mothers) | Measure: diagnosis of depressive disorder using SCAN. Timing: third trimester and 2, 6 and 12 months postpartum | Measure: underweight, weight-for-age z-score < −2; stunting, height-for-age z-score < −2. Timing: 2, 6 and 12 months postpartum | Significant findings: prenatal depression, underweight at 6 months, adjusted OR: 3.5, 95% CI: 1.5–8.6; stunted at 6 months, adjusted OR: 3.2, 95% CI: 1.1–9.9; underweight at 12 months, adjusted OR: 3.0, 95% CI: 1.5–6.0; stunted at 12 months, adjusted OR: 2.8, 95% CI: 1.3–6.1 | Note: Chronic depression carried a greater risk for a poor child growth outcome than episodic depression. Covariates included low birth weight, early breastfeeding cessation < 6 months, ≥ 5 diarrhoeal episodes per year, mother’s and father’s education, maternal financial empowerment and relative poverty. Other analyses included calculations of the relative risks of stunting and underweight in children of depressed mothers. Relative risks were also calculated for underweight and stunting in children aged 6 and 12 months of chronically depressed mothers (at four time points) versus depressed at no time point. |
| Africa | |||||
| Adewuya et al. 2007 |
Nigeria, longitudinal case–control, n = 242 (120 depressed cases and 122 non-depressed matched controls) | Measure: major depression diagnosis using the DSM-III-R (SCID-NP). Timing: 6-weeks postpartum | Measure: malnutrition, < 5th percentile of weight-for-age or height-for-age. Timing: 6 weeks and 3, 6 and 9 months postpartum | Significant findings: malnutrition, weight for age at 3 months, unadjusted OR: 3.2, 95% CI: 1.2–8.4; height for age at 3 months, unadjusted OR: 3.3, 95% CI: 1.03–10.5; weight for age at 6 months, unadjusted OR: 4.2, 95% CI: 1.4–13.2; height for age at 6 months, unadjusted OR: 3.3, 95% CI: 1.2–9.6. No statistically significant results at 6 weeks or 9 months | Note: Depressed mothers were more likely to stop breastfeeding early and their infants were more likely to have episodes of diarrhoea or other infectious illnesses. Covariates were not included in the models. |
| Harpham et al. 2005 |
Ethiopia, cross-sectional, n = 1 722 | Measure: SRQ-20 using a cut off of 7–8. Timing: 6–18 months postpartum | Measure: underweight, weight-for-age z-score < −2; stunting, height-for-age z-score < −2. Timing: 6–18 months postpartum | Non-significant findings: underweight, adjusted OR: 0.9, 95% CI: 0.6–1.2; stunting, adjusted OR: 1.1, 95% CI: 0.9–1.4 | Note: Unadjusted and adjusted associations between maternal mental health and infant growth were not significant. Covariates included maternal age and education, the child’s sex, age and birth weight, and household composition, wealth index and geographical location. |
| Stewart et al. 2008 |
Malawi, cross-sectional, n = 501 | Measure: SRQ-20 using a cut off of 7–8. Timing: median infant age at time of survey: 9.9 months postpartum | Measure: mean weight-for-age z-score, mean height-for-age z-score. Timing: depended on child | Significant findings: height-for-age z-score: adjusted β: −0.27 (P = 0.01). Non-significant findings: weight-for-age z-score: adjusted β: −0.02 (P = 0.87) | Note: Multivariate analyses adjusted for maternal height, maternal weight, wealth index, education to Standard 6 or above, ≥ 4 surviving children, maternal occupation, marital status, paternal occupation, mother able to confide in husband or relative, recent infant diarrhoea, recent infant fever, infant age, infant sex and weight-for-age z-score at first postnatal weighing. |
| Tomlinson et al. 2006 |
South Africa, prospective cohort, n = 147 at baseline; n = 98 at 18 months | Measure: major depression diagnosis using the DSM-IV (SCID). Timing: 2 and 18 months postpartum | Measure: height-for-age z-score < 10th percentile; weight-for-age z-score < 10th percentile; height z-score and weight z-score (using Tanner scales). Timing: 2 and 18 months postpartum | Non-significant findings: using depression at 2 months to predict infant growth at 18 months, height-for-age z-score < 10th percentile, unadjusted OR: 2.3, 95% CI: 0.9–6.0; weight-for-age z-score < 10th percentile, unadjusted OR: 2.5, 95% CI: 0.98–6.5; weight-for-age z-score adjusted for birth weight, β: 0.2, 95% CI: −0.21 to 0.78; height-for-age z-score, unadjusted β: 0·56, 95% CI: −0·16 to 1·3 | Note: The attrition rate was 33%. Birth weight was a covariate in one model. |
| South America and the Caribbean | |||||
| Baker-Henningham et al. 2003 |
Jamaica, case–control (matched), n = 210 (139 malnourished cases and 71 well nourished controls) | Measure: CES-D (adapted) used as a continuous variable. Timing: after initial growth assessment | Measure: underweight, cases, history of weight-for-age z-score < −2 plus current weight-for-age z-score < −1.5; controls, weight-for-age z-score > −1 plus no history of undernutrition. Timing: 9–30 months postpartum | Non-significant adjusted findings: underweight, mean = 26; normal weight, mean = 16.5; unadjusted comparison, P < 0.01 | Note: In crude analyses, maternal depression was related to underweight. The relationship became non-significant after adjusting for confounding variables. Covariates included maternal height, economic stress, work skill level, absent father and possessions. |
| Carvalhaes and Benício 2002 |
Brazil, case–control, n = 301 (101 cases and 200 controls) | Measure: selected SRQ-20 subset of items: (i) three depressed mood items (scored 0/1–2/3), and (ii) four depressive symptoms (scored 0/1–2/3–4). Timing: sometime after initial screening of children at 12–23 months postpartum | Measure: underweight, cases, < 5th percentile of weight-for-age; controls, > 25th percentile of weight-for-age. Timing: sometime after initial screening of children at 12–23 months postpartum | Underweight, significant or marginally significant findings: maternal depressive symptoms, unadjusted OR: 4.1 (P < 0.01), adjusted OR: 3.1, 95% CI: 1.0–10.3. Non-significant findings: maternal depressed mood, unadjusted OR range: 1.8–1.9 (P = 0.3); adjusted results not shown | Note: Models were adjusted for per capita household income, maternal education, presence of the partner, child’s age when the mother returned to work, hospitalization during gestation and alcoholism in the family. |
| De Miranda et al. 1996 |
Brazil, case–control, n = 139 for unadjusted analyses; n = 105 for adjusted analyses (60 controls and 45 cases) | Measure: QMPA score > 6. Timing: cases, mean: 10.9 months (SD: 6.9) postpartum; controls, mean: 8.4 months (SD: 4.8) postpartum | Measure: underweight, cases < 75% expected weight for age. Timing: cases, mean: 10.9 months (SD: 6.9); controls, mean: 8.4 months (SD: 4.8) postpartum | Significant findings: underweight, unadjusted OR: 2.8, 95% CI: 1.2–6.9; adjusted OR: 2.9, 95% CI: 1.3–6.8 | Note: Individual adjustments were made for the number of siblings, maternal age, infant birth weight, income and maternal education. The number of siblings, maternal age and infant birth weight were included in multivariate models. |
| Harpham et al. 2005 |
Peru, cross-sectional, n = 1949 | Measure: SRQ-20 using a cut off of 7–8. Timing: 6–18 months postpartum | Measure: underweight, weight-for-age z-score < −2; stunting, height-for-age z-score < −2. Timing: 6–18 months postpartum | Non-significant findings: underweight, adjusted OR: 0.9, 95% CI: 0.6–1.2; stunting, adjusted OR: 1.1, 95% CI: 0.9–1.4 | Note: Unadjusted and adjusted associations between maternal mental health and infant growth were not significant. Covariates included maternal age and education, the child’s sex, age and birth weight and household composition, wealth index and geographical location. |
| Santos et al. 2010 |
Brazil, longitudinal, n = 4 287 initially enrolled (3 748 remained at the 48-month follow-up) | Measure: EPDS using a cut off of ≥ 13. Timing: 12, 24 and 48 months postpartum. (Meta-analysis included depression at 1–2 time points in relation to anthropometry at 48 months) | Measure: underweight, weight-for-age z-score < −2; stunting, height-for-age z-score < −2. Timing: 48 months postpartum | Non-significant findings: underweight, adjusted OR: 1.5, 95% CI: 0.8–2.8; stunting, adjusted OR: 1.0, 95% CI: 0.6- 1.5 | Note: Models adjusted for family income, maternal skin colour, maternal schooling, parity, pre-pregancy body mass index, smoking during pregnancy, preterm birth and hospitalization in the first year of life. The underweight analysis also adjusted for the duration of breastfeeding and the stunting analysis adjusted for age. |
| Surkan et al. 2008 |
Brazil, cross-sectional, n = 595 | Measure: CES-D using a cut off of ≥ 16. Timing: 6–24 months postpartum | Measure: stunting, height-for-age z-score < −2. Timing: 6–24 months postpartum | Significant findings: stunting, adjusted OR: 1.8, 95% CI: 1.1–2.9 | Note: Covariates included the child’s gender, birth weight and age, breastfeeding duration, maternal education, sanitation score, socioeconomic status, living conditions, the number of children living in the household and participation in the Family Health Programme. |
CES-D, Center for Epidemiologic Studies Depression Scale; CI, confidence interval; DSM-III-R (SCID), Diagnostic and statistical manual of mental disorders, third edition, revised Structured Clinical Interview; DMS-III-R (SCID-NP), Diagnostic and statistical manual of mental disorders, third edition, revised Structured Clinical Interview – non-patient version; DSM-IV (SCID), Diagnostic and statistical manual of mental disorders, fourth edition, Structured Clinical Interview; EPDS, Edinburgh Post-natal Depression Scale; HOME, Home Observation for Measurement of the Environment; OR, odds ratio; QMPA, Adult Psychiatric Morbidity Questionnaire; SCAN, Schedules for Clinical Assessment in Neuropsychiatry; SD, standard deviation; SRQ-20, 20-item Self-Reporting Questionnaire.
a Each sample comprised a mother and child pair.
b Although the paper presented continuous outcomes and did not report adjusted results for underweight or stunting, we obtained the data from the authors and re-analysed it using multivariate logistic models to obtain the adjusted ORs for stunting and underweight included in the meta-analysis.