Effect of Community-Initiated Kangaroo Mother Care on Postpartum Depressive Symptoms and Stress Among Mothers of Low-Birth-Weight Infants

Key Points Question Does the practice of community-initiated kangaroo mother care (ciKMC), an intervention encompassing skin-to-skin-contact and exclusive breastfeeding, during the neonatal period reduce the risk of moderate-to-severe postpartum depressive symptoms among mothers of low-birth-weight (LBW) infants? Findings In a randomized clinical trial that included 1950 mothers of stable LBW infants from low-income areas in India, the practice of ciKMC resulted in a 25% reduction in the relative risk of moderate-to-severe depression at 4 weeks after delivery. The analysis estimated that supporting 36 mothers to perform KMC at home would prevent 1 mother from experiencing moderate-to-severe postpartum depressive symptoms. Meaning These findings suggest that ciKMC can have substantial benefits for maternal mental health, beyond improving the survival of LBW infants.


Introduction
Postpartum depression 1,2 affects quality of life and long-term psychological health in mothers and can also adversely affect mother-child interaction, breastfeeding, infant growth, and development. [3][4][5][6][7] Pooled estimates from 53 studies in 23 low-and middle-income countries reported the prevalence of postpartum depression to be 19% (95% CI, 16%-23%). 8 A meta-analysis 9 that included 38 studies in India estimated the prevalence of postpartum depression to be 22% (95% CI, 19%-25%). The prevalence of depression is reported to be higher among mothers of preterm compared with full-term infants in the first 12 weeks after birth. 10,11 Several psychosocial and psychological interventions have been evaluated for their effect on the risk of postpartum depression. These include postpartum home visits by professionals, telephone-based peer support, and interpersonal psychotherapy. Although promising, they are not widely implemented in the Indian public health system. 12 Kangaroo mother care (KMC), an intervention encompassing skin-to-skin contact (SSC) and exclusive breastfeeding, reduces the risk of death and severe infection in low-birth-weight (LBW) infants and is recommended by the World Health Organization and the Government of India. [13][14][15] In a large randomized clinical trial 16 in India among 8402 LBW infants, promotion of community-initiated KMC (ciKMC) improved postenrollment neonatal survival by 30% and infant 6-month survival by 25%. KMC could reduce the risk of postpartum depressive symptoms through mother-infant bonding and possibly via the release of maternal oxytocin and lowering of cortisol secretion. [17][18][19] Data from observational and quasiexperimental studies [20][21][22][23] suggest such a beneficial effect on mothers, but conclusive evidence is lacking.
Our primary objective was to test the hypothesis that the practice of ciKMC during the neonatal period can reduce the risk of moderate-to-severe maternal postpartum depressive symptoms. As a secondary objective, we estimated the effect of ciKMC on maternal salivary cortisol concentration, a biomarker of stress, at the end of neonatal period. Supplement 1). The study is reported as per the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline. 25 As part of the primary trial, pregnant women were followed up by a surveillance team until delivery. Newborn infants weighing 1500 to 2250 g and their mothers screened within 72 hours of birth were eligible to be included in the trial, unless KMC had already been initiated in a birth facility or infants were unable to feed, had breathing problems, had gross congenital malformations, or had less than normal movements. 26 Mothers not living with their infants and those intending to move away over the next 6 months were excluded. In our substudy, we excluded mothers of twins or triplets. For feasibility reasons, we made an a priori decision to restrict enrollments to a maximum of 6 per day, randomly selected from those enrolled in the primary trial. Between April 2017 and March 2018, 1950 of 3326 mothers enrolled in the primary trial were included in this substudy for evaluation of postpartum depressive symptoms (Figure 1).

Randomization and Masking
The randomization list was prepared by an independent statistician using random permuted blocks of variable size. Allocation of participant identification number was done by an off-site randomization coordinator using serially numbered opaque sealed envelopes. To minimize contamination between trial groups, if a previously enrolled infant was allocated to the intervention group, the next eligible infant belonging to the same household was also allocated to intervention. If a previously enrolled infant was allocated to the control group, the next eligible infant in the same household was assigned  to intervention or control as per the randomization sequence. Further details of methods of the primary trial have been published elsewhere. 16,24 Intervention and Usual Care The ciKMC intervention comprised promotion and support of SSC and exclusive breastfeeding.
Infant-mother dyads allocated to the ciKMC group were visited at home by an intervention delivery team consisting of a pair of trained workers to initiate and support KMC. The team observed breastfeeding and promoted SSC by using photographs and the local term chaati se chipkaana, which means sticking the baby to the chest. Mothers were counseled to practice SSC as long as possible during day and night with the assistance of other family members and were taught to daily record the duration of SSC. The team visited the families on days 1, 2, 3, 5, 7, 10, 14, 21, and 28 after birth to observe and address any problems related to SSC and/or breastfeeding; the duration of the visits ranged from 30 to 45 minutes. Visits continued until the infant wriggled out from the KMC position and no longer accepted SSC or until 28 days of age, whichever was earlier. 24,26 Referral of ill infants in both trial groups was facilitated through government-accredited social health activists. All infants in the intervention and control groups received usual care (ie, home-based postnatal care visits by accredited social health activists as implemented through the health system). 27

Outcomes
Our primary outcome was moderate-to-severe postpartum depressive symptoms at the end of the neonatal period (day 28 after birth) measured using a pretested Hindi version of the validated Patient Health Questionnaire-9 (PHQ-9). 28,29 Secondary outcomes were any depressive symptoms, PHQ-9 score, and maternal salivary cortisol concentration at the end of neonatal period. Postpartum depressive symptoms were defined as PHQ-9 scores of 10 or higher for moderate-to-severe symptoms, 5 to 9 for mild symptoms, and 0 to 4 for no or minimal symptoms. 28 A PHQ-9 score of 5 or higher defined any depressive symptoms. The sensitivity and specificity of a PHQ-9 score 10 or higher to diagnose major depression in the postpartum period are reported to be approximately 80% and greater than 90%, respectively. 29,30 Assessments Study workers were trained in good clinical practice. 31 Reported duration of SSC and exclusive breastfeeding were captured by an independent outcome ascertainment team at the end of the neonatal period for all study participants. For the PHQ-9 interviews, an independent team of health workers was trained and assessed by a clinical psychologist. The PHQ-9 assessment was done after day 28, but not later than day 42 after birth. Supervisors conducted random quality checks in 1% of the assessments. Women with PHQ-9 scores 10 or higher were referred to health facilities.
We collected salivary specimens from the first 550 mothers on day 28 after birth to measure salivary cortisol. To account for diurnal fluctuations and variability related to breastfeeding, the specimens were collected before and after breastfeeding, not later than noon.

Statistical Analysis
Assuming 90% power, 95% confidence, and 10% attrition and to be able to identify a minimum of 30% relative reduction from a 19% proportion of mothers with moderate-to-severe postpartum depressive symptoms in the control group, 9  conducted on an intent-to-treat basis using Stata statistical software version 16 (StataCorp). We estimated gestational age from ultrasonography reports, hospital records, or maternal recall, whichever were available, in the given order of preference.
We estimated relative risks (RRs) with 95% CIs for maternal moderate-to-severe or any depressive symptoms between the study groups using generalized linear models of the binomial family with a log-link. The efficacy of the intervention against maternal moderate-to-severe depressive symptoms was calculated as (1 − RR) × 100. We conducted multinomial logistic regression to estimate the effect of ciKMC on the risk of different categories of maternal depressive symptoms with no or minimal depressive symptoms (PHQ-9 scores of 0-4) as the reference. To compare salivary cortisol levels, we used the Wilcoxon rank-sum test.
Following CONSORT 2010 guidelines, 32 our approach was to present both unadjusted and adjusted results while avoiding overadjustment. Wealth quintile, mother age, mother education, and birth order were the potentially confounding covariates associated with the primary outcome at P < .10 in univariable analysis. To avoid overadjustment, 33 we included the potential confounding factors in the final multivariable regression model only if they were unequally distributed between the study groups at baseline, a priori defined as a relative difference of greater than 10%. 34 We used the likelihood ratio test based on deviance statistics to assess goodness-of-fit of the model 35 Design effects of infant-mother dyads enrolled from a single household were accounted for using Stata's robust variance estimator (vce) option. To quantify any biological interaction between ciKMC and preterm birth, we estimated the relative excess risk due to interaction 36 using the icp command.
We conducted exploratory subgroup analyses to estimate whether the effect of ciKMC on postpartum depressive symptoms was different in mothers with preterm birth (<37 weeks gestation) compared with full-term birth (Ն37 weeks gestation). Data analysis was performed from January to July 2020.

Results
Of  difference for moderate-to-severe postpartum depressive symptoms was 2.8% (95% CI, 0.1%-5.8%), corresponding to a number needed to treat of 36 mother-infant dyads (95% CI, 17-1000 dyads). The median (interquartile range) PHQ-9 scores were 2 (0-5) in the intervention group and 2 (0-6) in the control group. The cumulative frequency plot showed a left-shift of the PHQ-9 scores among ciKMC group compared with the control group mothers (Figure 2), suggesting that at any  given PHQ-9 score cutoff, a lower proportion of mothers in the ciKMC group were above the cutoffs (eTable 3 in Supplement 2).

JAMA Network Open | Psychiatry
In the subgroup of mothers with preterm births, the RR of moderate-to-severe postpartum depressive symptoms in the ciKMC group was 0.71 (95% CI, 0.52 to 0.96) vs 0.86 (95% CI, 0.56 to 1.34) among mothers with full-term infants. Preterm birth had a relative excess risk due to interaction with ciKMC for the study outcome of 0.19 (95% CI, −0.20 to 0.58).
Salivary cortisol measurements were done in 92.4% of the mothers (508 of 550 mothers) ( Figure 1). The median (interquartile range) day-28 maternal salivary cortisol concentrations in both the ciKMC group and control group were similar before and after breastfeeding ( Table 3). Abbreviations: ciKMC, community-initiated kangaroo mother care; RR, relative risk.
a Adjusted for birth order categories and accounting for household clustering.

Discussion
We found a 25% lower risk of moderate-to-severe postpartum depressive symptoms among mothers in the ciKMC group than among those in the control group. The RR reduction was somewhat higher among mothers of preterm infants. Our analysis indicated that ciKMC practice for approximately 36 stable LBW infant-mother dyads (ie, the number needed to treat) would prevent 1 case of moderate-to-severe postpartum depressive symptoms. We found no difference in the day-28 maternal salivary cortisol concentrations between the study groups.
Our study was done as a substudy within the primary ciKMC efficacy trial. 16 Selection and follow-up biases are not likely because of effective randomization, allocation concealment, and very low and balanced attrition. Nonetheless, we cannot rule out the possibility of recall bias. However, in the intervention group, information on the duration of SSC captured by the outcome assessment team based on mothers' history was similar to that obtained by the intervention delivery team during their scheduled visits, suggesting a low likelihood of such bias. 16 To minimize the possibility of conscious underreporting of symptoms, the outcome assessment team was extensively trained to conduct PHQ-9 assessment based on a comprehensive interview process in a comfortable home environment and not by asking the questions directly. Despite having an independent outcome assessment team, it might have been possible to guess the group allocation for those who practiced KMC beyond the 28-day period during PHQ-9 interview. However, such a possibility was small because only approximately 2% of mothers continued SSC beyond the 28-day period. Given the low risk of bias, the results are likely to be reliable and representative of the target population, suggesting that ciKMC can substantially reduce the risk of moderate-to-severe maternal postpartum depressive symptoms.
To the best of our knowledge, this is the first randomized clinical trial to estimate the effect of KMC on maternal depressive symptoms, and the findings substantiate those of previous reports from less rigorous study designs. In a quasi-experimental study in Canada, 20  Screening Scale decreased from 37% after delivery to 17% at hospital discharge with KMC practice.
Our trial provides valid evidence of the efficacy of KMC on the risk of postpartum depression in mothers with LBW infants and substantiates the observations from the earlier studies.
KMC may lessen depressive feelings by empowering women in their mothering role and improving infant-mother bonding. 19,22,37 The substantial effect of ciKMC that we observed on postpartum depressive symptoms is likely to be a result of a multitude of complex psychological mechanisms, potentially supported and intensified by enhanced oxytocin release as a result of SSC and exclusive breastfeeding. 17,38 Because of its clinical implications, we chose moderate-to-severe postpartum depressive symptoms as the primary outcome instead of PHQ-9 score. Women with PHQ-9 scores of 10 or higher have a high probability of major clinical depression that needs clinical attention, whereas the milder symptoms are often self-limiting. 28 In this context, our findings that ciKMC may reduce the risk of moderate-to-severe postpartum depressive symptoms are pertinent.
The potentially greater benefit of ciKMC in mothers with preterm infants in reducing moderate-tosevere postpartum depressive symptoms is relevant given their higher risk of postpartum depression. 11 However, we acknowledge the limitations of our subgroup analysis, in that we did not stratify our randomization on whether the birth was preterm and because the statistical precision of the interaction was moderate.
Cortisol is a hormone produced by the activation of hypothalamic-pituitary-adrenocortical axis in response to physiological stress. 20,39 Salivary cortisol is widely used to capture short-term fluctuations in physiological stress. 40 The observed null effect of KMC on day-28 maternal salivary cortisol is similar to that of a quasi-experimental study in Canada, 20 where the mean cortisol concentrations among intervention and control mothers at the end of neonatal period were 0.23 and 0.24 μg/dL, respectively. These findings suggest a possible dissociation between acute stress and postpartum depression, as far as the effect of ciKMC is concerned. Further research on the effect of KMC on acute and chronic stress (using markers such as hair cortisol 40,41 ) may be warranted.

Limitations
Although this is a large randomized clinical trial where almost all baseline characteristics were wellbalanced between the trial groups, a baseline PHQ-9 assessment would have been valuable. Our study population was limited to mothers with stable LBW infants weighing 1500 to 2500 g, and the findings may not be applicable to mothers with unstable or very-LBW (ie, <1500 g) infants. Neonatal survival programs would benefit from further research that also includes such infants.

Conclusions
The findings of our study support ciKMC as an intervention to prevent maternal depressive symptoms in the early postpartum period. Research on long-term benefits of the intervention on maternal psychosocial health and child development would be helpful. The study findings together with previous literature 13,16,20,23 suggest substantial benefits of home-based KMC for mothers, beyond improving health and survival of LBW infants. Developing a focused LBW program is identified as one of the important agendas of the Indian Government 42 and maybe relevant in other low-and middle-income countries where a high proportion of infants are born with LBW. This evidence supports the integration of KMC into essential newborn care programs for LBW infantmother dyads.