METHODS

The study hospital was a district hospital in the state of Goa in India, with a catchment population of urban women living in the town of Mapusa and rural women from neighbouring villages. As with other public hospitals in India, patients are drawn from the lower socioeconomic strata. A cohort study design was used to test the growth outcome hypothesis; nested in the cohort was a case-control study of developmental outcomes. The sample consisted of consecutive babies whose mothers were participating in a study of PND and who were brought to the hospital immunisation clinic at 6–8 weeks.⁶ Of the 171 babies recruited in this manner, 37 had mothers who were suffering from PND (22%) and 134 had mothers who were not depressed (78%). Babies were weighed and measured on recruitment.

The following data were collected from mothers at recruitment: sociodemographic data, including maternal and paternal education; infant data (prematurity, gender, birth weight, hospital admissions, sickness episodes); and feeding history. Mothers were also interviewed using the Edinburgh Postnatal Depression Scale (EPDS), a 10 item questionnaire specifically developed to measure maternal mood in the postnatal period.⁷ Versions of the EPDS in Konkani, the most widely spoken language of Goa, were developed by translation, backtranslation, consensus and field testing. A cut off score of 11/12 was found to detect caseness of PND with acceptable sensitivity and specificity.⁶ Using the EPDS, 23% of mothers were found to be suffering from depression.

Outcomes were measured at 6 months. Growth outcomes (weight and length) were carried out by clinical psychologists who had been trained by the second author using interobserver measurements in an immunisation clinic.⁸ Weight was recorded on a spring balance to the nearest 50 g. Length was measured using an infantometer, to the nearest 0.1 cm. Weight for age and length for age percentiles were computed using Epinut anthropometry software (World Health Organisation, 1994). Babies whose weight and length fell below the 5th centile for age were considered to be underweight and short for age respectively.⁹ Corrected age was used for the assessment of growth when the baby was premature.¹⁰

Babies whose mothers were depressed (cases) and a subsample of controls (selected as the next infant recruited after a case) were invited to participate in an examination with the Developmental Assessment Scale for Indian Infants (DASII)¹¹ at 6 months. This is a standardised developmental test that has been derived from Indian norms, based on the Bayley Scales of Infant Development.¹² The DASII can be used for the assessment of motor and mental development of babies from 0 to 30 months of age. Independent motor and mental developmental quotients are generated based on the performance of the baby during testing. The DASII was administered blind to the mental health status of the mother.

Univariate analyses were conducted to examine the association between growth outcomes and PND. Odds ratios and 95% confidence intervals were then computed for this association after adjustment for seven variables recognised to be associated with poor growth outcomes: prematurity; gender of the baby; paternal and maternal education (school completer or not); birth weight (below or above 5th centile); breast feeding practice (exclusive or not); and infant ill health (history of hospital admission or an episode of fever, diarrhoea, or vomiting in the first 6–8 weeks) (table 1⇓). Multiple logistic regression was carried out using Stata version 6 (Stata Corporation, USA). In the nested case-control study of developmental outcomes, the motor (MoDQ) and mental (MeDQ) developmental quotients were calculated for each baby and analysed as categorical and continuous variables (table 2⇓). For categorical outcomes, a developmental quotient below 85 was indicative of developmental delay in an individual baby following norms used in other Indian studies.¹³ All significance tests are two tailed.

RESULTS

DISCUSSION

This study shows that PND was associated with significantly poorer growth outcomes at 6 months, even after adjustment for other determinants of poor growth such as low birth weight and low parental education. Delayed mental development was also associated with PND. Thus, both the study hypotheses were found to be correct. To the best of our knowledge, this is the first study showing a linkage between PND and adverse infant growth outcomes in a low income country. The sample was drawn from a catchment area population of women attending a district maternity service. The case-control study of developmental outcomes was nested in a cohort, reducing the risk of bias imposed by retrospective recall in cross sectional designs. Infant developmental assessments were undertaken using a test standardised for Indian infants, by experienced examiners who were blind to maternal mental health status. The limitation of the study is that the population sampled was not representative of all mothers in the study area; in particular, it excludes mothers whose pregnancies were complicated and would have been referred to a tertiary medical centre. Mothers choosing private health care were also not included. Another limitation was that maternal IQ, which has a direct bearing on developmental outcomes, was not studied. However, maternal educational achievement was determined and controlled for in the analyses. Despite these limitations and the relatively small sample size, the null hypothesis in relation to growth and development outcomes was rejected. However, it is possible that there could be an independent, unmeasured, variable which may cofound the association we have reported

There is a growing body of evidence which explains how PND may impair infant growth. It is well recognised that depression is a profoundly disabling disorder, which is often unrecognised and untreated in low income countries.¹⁴ Typically, features of depression include fatigue and tiredness, sleep difficulties, loss of appetite and weight, poor concentration, and a loss of interest in daily activities. Such features may seriously impair the mothering role; for example, by raising the probability that the mother may stop nursing the infant. Breast feeding problems were significantly more commonly reported by depressed mothers in a cohort study of mothers; top feeding before the 6–8 week review was more common in this group.⁶ Early cessation of exclusive breast feeding has been identified as a key factor in infant undernutrition,¹⁵ and may play the mediating role in explaining why infants of depressed mothers have poorer growth outcomes. Depressed mothers are also emotionally unavailable to their babies, and this can lead to psychosocial deprivation and non-organic failure to thrive.

While several studies from developed countries have shown the relation between PND and adverse infant mental development, there is only one other published study from a developing country.¹⁶ A variety of factors have been suggested as mediating mechanisms between PND and adverse infant development.¹⁷ These can be grouped in three causal pathways: first, poorer development may be the result of direct exposure to mother’s depressive symptoms; second, development may be adversely affected by parenting difficulties, such as mother’s interactive style, associated with maternal depression; and third, risk factors for PND, such as poverty may also be independent risk factors for adverse infant development. It is not possible to study the relative role of the first two causal mechanisms in the present study. The third causal pathway was investigated by including, with PND, hunger, maternal literacy, and measures of marital discord (for example, violence) in logistic regression analyses. PND remained independently associated with poorer mental development in these analyses, suggesting a direct relation through either of the first two causal pathways.

The implications of the findings of this study are that PND, a potentially treatable illness, may be a cause for poor growth outcomes and poor infant development in South Asia. Keeping in mind the limitations of this study, there is a need to replicate the study in different populations. There is a growing body of evidence that brief counselling interventions can help prevent and treat PND.^18,19 Transferring evidence of the efficacy of such interventions in the considerably different environment and cultures of low income countries may not be possible because of a number of factors.²⁰ Thus, there is a need to conduct controlled trials of interventions delivered to parents to evaluate their efficacy in preventing PND and associated adverse infant outcomes. Such a trial is currently in progress in Goa.

This study shows a clear linkage between depression, a mental disorder of global public health significance,¹⁴ with poor infant growth, a major public health priority in many developing countries. It provides a rationale for integrating mental health, largely ignored by public health policy in developing countries, within the scope of maternal and child health services. Providing training on the recognition and management of PND may now serve the twin purposes of alleviating distress in mothers and reducing the rates of poor growth.