Social support and spousal relationship quality improves responsiveness among anxious mothers

Abstract

Maternal responsiveness, a mother’s ability to consistently identify infant cues and then act on them, is critical for healthy child development. A woman’s social support and spousal relationship may affect responsiveness to an infant, especially among mothers with anxiety. We assessed how social support and spousal relationship quality is associated with responsiveness among anxious mothers, and if postpartum depression (PPD) moderated these associations. Cross-sectional data were collected from 2019 to 2022 in a public hospital in Pakistan from 701 women at six-weeks postpartum. Eligible women had at least mild anxiety in early- to mid- pregnancy. Linear regression analyses assessed if spousal relationship quality and social support from family and friends were associated with maternal responsiveness, measured using the Maternal Infant Responsiveness Instrument. Interaction terms were used to examine if PPD moderated these associations. Spousal relationship quality (B=2.49, 95% CI: 1.48, 3.50) and social support (B=1.07, 95% CI: 0.31, 1.83) were positively related to maternal responsiveness to the infant. Emotional support from a spouse was positively associated with responsiveness (B=1.08, 95% CI: 0.12, 2.03 depressed; B=2.96, 95% CI: 1.34, 4.58 non-depressed), and conflict with the spouse was negatively associated with responsiveness (B=−1.02, 95% CI: −1.94, −0.09 depressed; B=−2.87, 95% CI: −4.36, −1.37 non-depressed). However, social support was related to responsiveness only in non-depressed women (B=2.61, 95% CI: 1.14, 4.07). While spousal relationships and social support enhance maternal responsiveness, for depressed women, spousal relationships were particularly critical. In considering maternal-infant interventions to improve child development outcomes, our study indicates the importance of supportive relationships that foster effective responsiveness.

INTRODUCTION

Maternal responsiveness refers to a mother’s ability to identify infant cues about physical and emotional states consistently and then to act on those cues (1,2). Responsiveness to an infant is critical for promoting healthy physical and psychological child development (3,4) through advancing secure infant attachment (1,5). Such responsive interactions with an infant have been identified as critical for children’s cognitive and emotional development across cultures (1,5,6). Because symptoms of anxiety in mothers have been linked to poor maternal responsiveness (7,8), children of women with anxiety are at especially high risk of poor outcomes. Such data are particularly concerning, given that global prevalence of self-reported anxiety symptoms during pregnancy and the postpartum period has been estimated at 22.4% and 15.0%, respectively (9), while postpartum depression was found in 17.2% of the global population (10). These estimates are slightly higher in LMICs, as meta-analytic studies indicate the prevalence of antepartum anxiety to be 29.2%, postpartum anxiety to be 24.4% (11), and postpartum depression to be 19% (12,13). Given the high prevalence in LMICs including Pakistan, maternal responsiveness – and thus potentially child development – may be impacted, despite limited research in such contexts.

Anxiety is a strong predictor of postpartum depression (14,15). In one study, women with high antenatal anxiety were 2.6 times more likely than those with no or low symptoms to experience postnatal depression, after controlling for confounders including the level of antenatal depression (16). This suggests that anxiety alone can have a critical impact on depression amongst postnatal mothers. Although effective treatments exist, there is minimal evidence on preventative approaches that may reduce the prevalence of postnatal depression (17). Our research was part of a parent study that assessed anxiety-focused cognitive behavioral therapy delivered to pregnant women by non-specialists to prevent postpartum depression in Rawalpindi, Pakistan (18). Given the importance of prenatal anxiety, the parent study sought to investigate the sole effect of treatment of prenatal anxiety on PPD by excluding depressed pregnant women from the study. Due to this exclusion, the parent study was able to show that women receiving the intervention were much less likely to develop depression, with 11.6% of those in the intervention arm reporting depression, compared to 40.5% of controls (18).

The role of maternal responsiveness in child development has been studied in both LMIC and HIC contexts. Across contexts, maternal responsiveness and sensitivity are positively associated with child development (1,4,19–21). Depression, however, has been negatively associated with maternal responsiveness and child development globally (6,19,22), including in rural Pakistan (20,21). While evidence from HICs indicates that psychosocial risk factors are interrelated, research shows additional complexity in LMIC settings due to their interaction with biological risk factors for poor child development (that are less pervasive in HIC settings), such as child undernutrition or low birthweight (23). As conceptual and empirical research has principally occurred in HICs (24,25), some interventions have failed to account for cross-cultural dynamics in LMIC settings, such as differing maternal roles on maternal responsiveness. There are also fewer studies from LMICs, despite being the settings where they are most needed given the compounded risk factors for poor child development, and the fact that the majority of the world’s population is both rural and non-westernized (24).

Social support encompasses informational, emotional, and material assistance (26,27) and is a strong predictor of a healthy pregnancy, including in the Pakistani context (28–30). Inadequate support in this context has been connected to poor antenatal healthcare-seeking practices, poor mental health, physical health, and child neglect (31–33). Among anxious pregnant women in Pakistan, a good, close relationship with a spouse or others has been found to contribute to maternal well-being (34,35). Relationship problems, too many children and socioeconomic adversity are risk factors for anxiety and depressive disorders among women generally in Pakistan, while supportive family and friends may protect against such disorders (36). In other contexts, high quality relationships have been shown to moderate the effect of social support on a host of broad health outcomes (37–40). In Pakistan, one study found that perceived higher social support was positively associated with marital satisfaction, which was associated with reduced risk of depression (41). Given that good social relationships are protective of anxiety (42,43), support and high-quality relationships could be resilience factors that may lead to responsiveness, especially for women with anxiety. Therefore, it is important to understand these associations among anxious women, who are at higher risk of poor responsiveness (7,8).

Prior research has shown a positive relationship between social support and maternal responsiveness, often with reference to and focused on its association with postpartum depression (PPD) (26,44,45). The larger a mother’s social network and the more social support she has, the less likely it is that PPD will occur (46). Likewise, good relationship quality is important for maternal mental health; specifically, poor partner relationship quality is a risk factor for PPD (39,47–49). Further, higher levels of PPD symptoms are associated with lower levels of responsiveness (2,50), suggesting PPD is important in this relationship. PPD could potentially moderate the relationships between spousal relationship quality and/or social support with maternal responsiveness. A thorough understanding of the associations between social relations and maternal responsiveness is lacking, particularly in LMICs and among women with anxiety symptoms. Contextual factors including low income and material stress also appear to impact maternal responsiveness and sensitivity (51). Therefore, further research in LMICs can offer deeper insight into divergent settings. The primary aim of this study was to investigate if spousal relationship quality and social support were associated with maternal responsiveness among women who had prenatal symptoms of anxiety. Secondarily, we sought to examine if PPD moderated these associations. We hypothesized that spousal relationship quality and social support would be positively associated with better maternal responsiveness among anxious women, and that PPD would moderate these associations.

METHODS

Study setting and participants

Data were collected from Holy Family Hospital (HFH), in Rawalpindi, Pakistan. This is a public facility with a catchment population of more than 7 million (52). On average, 250-300 women from low-resource communities are recipients of free antenatal care each day at HFH. This study draws on data from a randomized controlled trial (National Library of Medicine identifier: NCT03880032) to test the Happy Mother - Healthy Baby (HMHB) program, which sought to reduce anxiety among pregnant women recruited from HFH (53). A group of female assessors at the outpatient Department of Gynecology and Obstetrics recruited the women during their first prenatal visit, which occurred at or prior to 22 weeks of gestation.

Two levels of screening criteria were utilized to enroll participants. For the first level, women had to be ≥18 years old, pregnant at ≤ 22 weeks of gestation, reside ≤20 kilometers from HFH, and have a basic knowledge of Urdu. The exclusion criteria included past or current significant learning disabilities, life-threatening health conditions, past or current psychiatric disorders, or severe maternal morbidity. After meeting these criteria and demonstrating willingness to participate, potential participants were asked to give informed consent to continue in the screening process. The second level of screening assessed potential participants using the Hospital Anxiety Depression Scale (HADS) for indications of at least mild anxiety (scoring ≥8 on the HADS anxiety subscale). Participants who at least had mild depression (scoring ≥8 on the HADS depression subscale) were screened by a psychiatrist who initiated a Structured Clinical Interview for DSM Diagnoses (SCID). Women with a diagnosis of major depressive episode (MDE) were excluded. We excluded participants with past or current psychiatric disorders because we were specifically interested in the role of anxiety in pregnancy for our intervention (for which this study is a secondary analysis) which was concerned with whether reducing anxiety symptoms could prevent postpartum depression. Based on these criteria, 1,200 participants were enrolled and randomized equally into the intervention and control arms. A total of 720 women gave birth at the study hospital and responded to the questionnaire administered at a six-week postpartum visit. These postpartum data were collected between April 16, 2019, and when the last postpartum follow-up was conducted on October 7, 2022. After excluding 19 women who had missingness on income, our final analytic sample size was 701 (97.4% of the available sample; Figure 1).

Figure 1.

Consort diagram for HMHB randomized control trial

Ethical approval for this study was obtained from the Johns Hopkins Bloomberg School of Public Health, Rawalpindi Medical University, Human Development Research Foundation, and a NIMH-appointed Data Safety Monitoring Board.

HMHB Intervention

The HMHB intervention was implemented for the intervention group, which involved a culturally adapted cognitive-behavioral therapy (CBT) treatment that sought to reduce anxiety. HMHB is based on a well-established Thinking Healthy Programme and details of its development are described elsewhere (52). Briefly, this CBT treatment applied a three-step approach: 1) learning to identify unhelpful or unhealthy thinking and behavior; 2) learning to replace unhealthy or unhelpful thinking and behavior with thinking and behavior that is helpful; and 3) practicing acting and thinking healthy (52). The treatment involved six core one-on-one sessions that could be supplemented with up to six optional booster sessions, depending on individual needs. The first five core sessions were delivered weekly, beginning as early as possible during pregnancy (followed by any booster sessions). The sixth and final core session was delivered during the third trimester of pregnancy. Treatment was delivered by six non-specialist therapists, who had at least a master’s degree, but who did not have prior clinical experience. These non-specialists received 42-50 hours of classroom training and additional field training to develop experience delivering the intervention (52).

Enhanced Usual Care (EUC) was delivered to the control group. They received the standard of prenatal care, which includes up to eight visits in which any concerns were discussed, health status was evaluated and routine exams reflective of the stage of pregnancy were conducted. The study enhanced care by additionally reimbursing all participants in both arms for transportation and as many ultrasounds during pregnancy as medically indicated at HFH, in order to enhance participation in the study.

The intervention assessments were conducted individually at the Obstetrics Department of the Holy Family Teaching Hospital, Rawalpindi. They usually lasted approximately 25 minutes and were carried out with each participant after two levels of written consent, in order to maintain confidentiality according to the research protocol. Assessments were conducted by a team of six research assistants who were trained psychologists in the study’s outcome measurements. Study questionnaires were administered orally and responses were recorded electronically on android tablets, using the online data collection tool, Open Data Kit (ODK).

Variables of interest

The exposure variables in this study were relationship quality with the spouse and social support from significant others, family and friends, both measured at six-weeks postpartum. Relationship quality with the spouse was adapted from the MacArthur Battery, originally a six-item scale developed to assess whether the social network ties (including spouse, family, or friends) of older adults provide emotional or instrumental support, or whether they are a source of conflict and/or demand (54). We adapted these items specifically for the spouse, and asked participants about the frequency with which the spouse provided emotional support (two items; by making the participant feel cared for or loved; or by listening to the participant’s problems or worries) and instrumental support (two items; by helping the participant with daily tasks; or by giving participant information or advice about problems). Two further items were used to assess conflict/demand with spouse, by asking whether they made too many demands of participants or were critical of participant behavior. Example items include: How often does your spouse make you feel loved and cared for?” (emotional support); “How often does your spouse help with daily tasks like shopping, giving you a ride or helping you with household tasks?” (instrumental support); and “How often does your spouse make too many demands on you? (conflict). We used a five-point Likert scale that ranged from 0 (never) to 4 (frequently) to respond to all six items (such that mean scores also ranged from 0 to 4). We reversed conflict/demand scores when calculating the full scale mean, such that high scores indicated stronger relationship quality with a spouse. For emotional support and instrumental support subscales, high means indicated stronger relationship quality with a spouse. This adapted scale showed good internal consistency in our study (Cronbach’s α = 0.88). Translation and adaptation of the MacArthur Battery occurred in three steps: (a) initial translation; (b) cross-cultural verification and adaptation; and (c) checking the psychometric properties of the instrument in our population. We conducted the initial review and assessment to see if the instrument was understandable in the Pakistani context. Two independent translators fluent both in original English and Urdu then translated the instrument. To ensure the accuracy and cultural appropriateness, the translated version was then back translated into English by a different translator. The original version of the MacArthur battery was compared with a back-translated version to help identify any discrepancies to make sure that translated version accurately conveyed the intended meaning. Piloting and cognitive debriefing followed and the MacArthur Battery was finalized once it was refined, after getting feedback from participants.

The Multidimensional Scale of Perceived Social Support (MSPSS), a validated 12-item measure with subscale measures of support received from family, friends, and significant others, was used to assess social support, though excluding the subscale of significant other (55). It has been effectively adapted to the Pakistani context (30). Participants were presented with a series of statements pertaining to social support from significant others, family, and friends and asked to rate their level of agreement on a Likert scoring scale, ranging from 0 (very strongly disagree) to 4 (very strongly agree) (34). Thus, mean scores on the whole scale also ranged from 0 to 4. Higher mean scores of subscales or the full scale indicated greater perceived social support. The MSPSS, with two subscales, showed strong internal consistency in this study (Cronbach’s α = 0.95).

To measure the outcome, we used the Maternal Infant Responsiveness Instrument (MIRI), a 22-item scale measuring mother’s feelings about the infant and an appraisal of the infant’s responses measured at six-weeks postpartum (56). It includes statements such as “I have seen my baby respond to my playing with him/her” and “I believe that I can comfort my baby when she/her cries.” We used a five-point Likert scale answer option, ranging from 1 (strongly disagree) to 5 (strongly agree). Total scores range from 22 to 110, where high scores indicate good responsiveness. MIRI showed good internal consistency in our study (Cronbach’s α = 0.88).

We measured postpartum depressive symptoms (PPD), an effect modifier, at six-weeks postpartum using the Patient Health Questionnaire (PHQ-9). The PHQ-9 is a multipurpose instrument that screens, diagnoses, monitors and measures symptoms of depression and their severity (57). There are ten questions overall, with responses options using a four-point Likert scale from 0 (not at all) to 3 (nearly every day). High scores indicated greater severity of depression. PHQ-9 had good internal consistency in our study (Cronbach’s α = 0.92). We dichotomized PPD scores from the PHQ-9 into none-mild (score 0-4) versus moderate-severe symptoms (score 5-27) (57).

Statistical analysis

To calculate means, standard deviations (SD), and the range of our exposure and outcome variables, as well as the correlation among these variables, we conducted univariate descriptive statistics. Then, we performed bivariate descriptive analyses to compare means and SDs of MIRI scores by categories of sociodemographic variables. We used Kruskal-Wallis equality of populations rank tests (for categorical variables) and Wilcoxon rank sum tests (for binary variables) as maternal responsiveness scores were not normally distributed. If socio-demographic variables were significantly associated with maternal responsiveness scores in the bivariate analyses (p<0.05), they were then included in further regression analytic models.

We conducted seven separate Ordinary Least Square (OLS) regression analyses in order to assess social support and spousal relationship quality (for overall and subscale scores) in relation to maternal responsiveness. In these analyses, we adjusted for education, preterm birth, low child birthweight, treatment group, PPD, and exposure to intimate partner violence (IPV) in the postpartum period. To account for multiple hypotheses testing, we calculated Romano-Wolf step-down p-values with 1,000 bootstrap replications (58).

Our final analyses included creating interaction terms between the overall and subscales of PPD and exposure variables (social support and spousal relationship quality), and tests of effect modification of PPD on the association of these exposure variables on maternal responsiveness. We used Wald F-tests to identify models with significant moderating effects (p<0.05), and the models with significant interaction terms were graphically depicted in order to illustrate how associations varied by levels of PPD. All analyses were conducted using STATA 15.1 (StataCorp, College Station, TX), and p<0.05 was considered statistically significant.

RESULTS

Descriptive statistics

The means of spousal relationship quality and social support were 2.8 (SD=0.9) and 2.4 (SD=1.1) respectively (Table 1). These two variables were significantly positively correlated (r=0.5, p<0.001). The mean maternal responsiveness score was 90.4 (SD=10.2), showing relatively high levels of responsiveness in the range between 22 and 110. The responsiveness score was positively correlated with both spousal relationship quality (r=0.3, p<0.001) and social support (r=0.2, p<0.001).

Table 1.

Mean, standard deviation (SD), and correlation of spousal relationship quality, social support, and maternal responsiveness (n=701)

	1.	2.	3.	Mean (SD)	Possible range
1. Spousal relationship quality	1.00			2.80 (0.89)	0-4
2. Social support	0.51***	1.00		2.35 (1.12)	0-4
3. Maternal responsiveness	0.30***	0.23***	1.00	90.40 (10.24)	22-110

Note: High scores in relationship quality and social support indicate high levels of spousal relationship quality and social support. High scores in maternal responsiveness indicate good responsiveness.

^***p<0.001.

Table 2 shows the bivariate relationships between demographic covariates in relation to maternal responsiveness at six-weeks postpartum. Participants in the intervention group scored significantly higher in maternal responsiveness than participants in the control group (mean=91.5, SD=9.9 for intervention; mean=89.33, SD=10.5 for control, p=0.008). Exposure to any postpartum IPV and experiencing moderate-severe PPD were significantly associated with lower maternal responsiveness scores (mean=86.2, SD=11.1 for exposed to IPV; mean=89.3, SD=10.6 for moderate-severe PPD), when compared to participants not exposed to IPV and with no or mild PPD (mean=92.0, SD=9.5 for non-exposed to IPV, p<0.001; mean=92.3, SD=9.3 for none-mild PPD, p<0.001). A monthly household income of >20,000 PKR was significantly associated with higher maternal responsiveness scores (mean=91.8, SD=9.1 for >20,000 income, mean=89.6, SD=10.8 for ≤20,000 income, p=0.02). Having a preterm birth was significantly associated with higher maternal responsiveness scores (mean=92.4, SD=9.2 for preterm, mean=89.9, SD=10.4 for full term, p=0.02).

Table 2.

Bivariate descriptive statistics for the relation of demographic variables with maternal responsiveness at six-weeks postpartum (n=701)

	n (%)	Maternal responsiveness, mean (SD)	p-value†
Participant characteristics
Age (years)			0.543
≤25	419 (59.77)	90.53 (10.17)
>25	282 (40.23)	90.21 (10.36)
Education (years)			0.721
≤Middle school (≤8 years)	287 (40.94)	90.12 (10.86)
>Middle school (>8 years)	414 (59.06)	90.60 (9.80)
Gravidity			0.549
Primigravida	201 (28.67)	90.71 (10.10)
Multigravida	500 (71.33)	90.28 (10.30)
Previous pregnancy loss			0.784
Yes	294 (41.94)	90.27 (10.36)
No	407 (58.06)	90.50 (10.17)
Treatment group			0.008
Intervention group	350 (49.93)	91.48 (9.92)
Control group	351 (50.07)	89.33 (10.45)
Exposure to IPV at postpartum			<0.001
Exposed to any IPV	190 (27.10)	86.18 (11.08)
Not exposed	511 (72.90)	91.97 (9.45)
Postpartum depressive symptoms			<0.001
None-mild (0-4)	258 (36.80)	92.33 (9.25)
Moderate-severe (5-27)	443 (63.20)	89.28 (10.63)
Household characteristics
Family structure			0.178
Nuclear	226 (32.24)	89.34 (10.67)
Joint (living with in-law parents)	234 (33.38)	90.65 (10.17)
Extended/multiple households	241 (34.38)	91.16 (9.85)
Monthly household income (PKR)			0.021
≤20,000	445 (63.48)	89.60 (10.79)
>20,000	256 (36.52)	91.79 (9.06)
Child characteristics
Sex			0.459
Male	341 (48.64)	90.72 (9.82)
Female	360 (51.36)	90.10 (10.63)
Birthweight			0.405
Low birthweight (<2.5kg)	97 (13.84)	90.03 (9.30)
Normal (≥2.5kg)	604 (86.16)	90.46 (10.39)
Preterm birth			0.017
Preterm (<37 weeks)	144 (20.54)	92.35 (9.23)
Full term (≥37 weeks)	557 (79.46)	89.90 (10.43)

^†Wilcoxon rank sum tests and Kruskal-Wallis equality of populations rank test were used to compare maternal responsiveness scores by the categories of covariates.

Note: High scores in the Maternal Infant Responsiveness Instrument (MIRI) indicate good responsiveness.

Spousal relationship quality and social support in relation to maternal responsiveness

Spousal relationship quality was associated with maternal responsiveness (Table 3). A one-point increase in spousal relationship quality was associated with a 2.5 increase in maternal responsiveness, adjusting for treatment group, monthly household income, preterm birth, exposure to IPV, and PPD (B=2.49, 95% CI: 1.48, 3.50). Having emotional and instrumental spousal support was significantly related to responsiveness (B=1.53, 95% CI: 0.68, 2.38 for emotional; and B=1.74, 95% CI: 0.98, 2.51 for instrumental). In contrast, a one-point increase in conflict/demand was significantly associated with a 1.5 decrease in responsiveness score, indicating a negative association between conflict and maternal responsiveness (B=−1.48, 95% CI: −2.31, −0.66).

Table 3.

Associations of relationship quality and social support with maternal responsiveness at six-weeks postpartum

	Crude coefficient (95% CI)	Adjusted coefficient (95% CI)†
Relationship quality ‡
Overall	3.49 (2.68, 4.30)**	2.49 (1.48, 3.50)**
Subscales:
Emotional support	2.49 (1.75, 3.22)**	1.53 (0.68, 2.38)**
Conflict/demand‡	−2.59 (−3.27, −1.90)**	−1.48 (−2.31, −0.66)**
Instrumental support	2.59 (1.90, 3.28)**	1.74 (0.98, 2.51)**
Social support
Overall	2.12 (1.46, 2.78)**	1.07 (0.31, 1.83)*
Subscales:
Support from family	1.75 (1.08, 2.43)**	0.50 (−0.26, 1.27)
Support from friends	1.64 (1.12, 2.18)**	0.98 (0.40, 1.57)**

^†Models were adjusted for treatment group, monthly household income, preterm birth, exposure to intimate partner violence (IPV) at postpartum, and postpartum depressive symptoms (PPD).

^‡High scores in conflict/demand subscale indicates high levels of conflict or demand. In the overall score, conflict/demand subscale was reversely scored. High scores in overall relationship quality indicate good relationship with spouse.

Note: Romano-Wolf stepdown p-values was calculated to take account for multiple hypothesis testing.

^*adjusted p<0.05

^**adjusted p<0.01

^***adjusted p<0.001.

Perceived social support from family and friends was positively associated overall with maternal responsiveness. After adjustment, a one-point increase in overall social support was positively associated with a 1.1-point increase in maternal responsiveness score (B=1.07, 95% CI: 0.31, 1.83). The perceived support from friends subscale showed a similar association to the overall social support scores (B=0.98, 95% CI: 0.40, 1.57).

Postpartum depression as an effect modifier

The associations between both spousal relationship quality and social support with maternal responsiveness were modified by PPD, adjusting for treatment group, low birthweight, preterm birth, and exposure to any IPV at six-weeks postpartum (Table 4). Interaction terms between emotional support from a spouse and PPD symptoms, as well as the interaction between conflict with a spouse and PPD symptoms were significant in relation to maternal responsiveness, indicating that PPD was an effect modifier (p=0.04 for emotional; and p=0.03 for conflict). The interaction term between overall perceived social support and PPD symptoms was significant in relation to maternal responsiveness (p=0.02), as was the support from family subscale (p=0.03).

Table 4.

Moderating effect of postpartum depressive symptoms (PPD) in the association of spousal relationship quality and social support with maternal responsiveness at six-weeks postpartum

Interaction terms	Coefficient (95% CI)†	p-value§
Spousal relationship quality ‡
Overall × PPD	−1.68 (−3.75, 0.38)	0.110
Emotional support × PPD	−1.88 (−3.70, −0.07)	0.042
Conflict × PPD‡	1.85 (0.18, 3.52)	0.030
Instrumental support × PPD	0.94 (−0.61, 2.48)	0.234
Social support
Overall × PPD	−2.01 (−3.66, −0.36)	0.017
Support from family × PPD	−2.01 (−3.89, −0.14)	0.035
Support from friends × PPD	−1.19 (−2.41, 0.04)	0.058

^†Adjusted for treatment group, monthly household income, preterm birth, and exposure to intimate partner violence (IPV) at postpartum.

^‡High scores in conflict/demand subscale indicates high levels of conflict or demand. In the overall score, conflict/demand subscale was reversely scored. High scores in overall relationship quality indicate high levels of relationship quality with spouse.

^§p-values were obtained from the Wald test for moderating effects

Figure 2 graphically displays analyses for effect modification of PPD for the associations between spousal relationship quality/social support and maternal responsiveness. Emotional support from a spouse was positively associated with maternal responsiveness, and the magnitudes of the relationships were greater in non-depressed women compared to depressed women (B=2.96, 95% CI: 1.34, 4.58 for non-depressed; and B=1.08, 95% CI: 0.12, 2.03 for depressed). Similarly, conflict with the spouse was negatively associated with maternal responsiveness with a greater decrease in non-depressed women than depressed women (B=−2.87, 95% CI: −4.36, −1.37 for non-depressed; and B=−1.02, 95% CI: −1.94, −0.09 for depressed). In contrast, overall social support was positively associated with maternal responsiveness in non-depressed women (B=2.61, 95% CI: 1.14, 4.07), whereas this association was not significant in depressed women (B=0.59, 95% CI: −0.26, 1.44). Similarly, a one-point increase in social support from family was associated with a 2.19 increase in responsiveness in women with none to mild PPD (B=2.19, 95% CI: 0.44, 3.93). The association was null among women with moderate-severe PPD symptoms (B=0.17, 95% CI: −0.65, 0.99).

Figure 2.

Potential effect modification of postpartum depressive symptoms (PPD) in the associations of spousal relationship quality and social support with maternal responsiveness

(A) Association between emotional support from spouse and maternal responsiveness; (B) Association between conflict with spouse and maternal responsiveness; (C) Association between overall social support from family and friends and maternal responsiveness; and (D) Association between social support from friends and maternal responsiveness.

Note: Models were adjusted for treatment group, monthly household income, preterm birth, and exposure to intimate partner violence (IPV) at postpartum.

Graphic created using STATA 15.1 (StataCorp, College Station, TX).

When we conducted sensitivity analyses including all socio-demographic variables (regardless of their associations with maternal responsiveness), the associations of spousal relationship quality and social support with maternal responsiveness remained consistent with the original analysis results, although the magnitude of the associations for the relationship quality and responsiveness slightly decreased (Supplementary Table 1). In these sensitivity analyses, the moderating effects of PPD remained consistent with the original analyses, except for the interaction terms between emotional support from the spouse and PPD, and between social support from family members and PPD (Supplementary Table 2). The p-values for these two interaction terms increased slightly and became non-significant.

DISCUSSION

Results of our study indicate that spousal relationship quality and social support are both positively associated with the responsiveness of a mother to her infant, supporting our original hypothesis. Among the components of our relationship quality measure, emotional and instrumental support from the spouse were associated with more responsiveness, while conflict or demands between the spouses was associated with less responsiveness. Our study also suggested that PPD modified the association of spousal relationship quality and social support with maternal responsiveness. For instance, the effects of social support and of relationship conflict on responsiveness were less pronounced among women with higher levels of PPD, compared to women with low levels. Likewise, while social support helped women who were not depressed to be more responsive to their infants, there was almost no change in responsiveness based on support levels of women who were depressed. However, our findings also indicate that among depressed women, spousal relationship quality (i.e., emotional support and less conflict) improved maternal responsiveness, whereas social support showed a null association with responsiveness, suggesting that good spousal relationship quality is particularly important to depressed women. These findings indicate complex linkages between relationship quality with a spouse, social support, PPD, and maternal-infant responsiveness that are a novel contribution to the literature from LMICs.

Our finding that spousal relationship quality was positively associated with maternal responsiveness supports prior research (including among pregnant women) showing the importance of quality of close relationships to positive interactions with infants. One meta-analytic review of US studies indicated that variations in spousal relationship quality can influence the quality of parent-child interactions (59). Another study conducted in the US found that married mothers were more sensitive to their children than stably cohabiting ones (60). Moreover, a study from Italy showed a significant association between couple relationship quality and maternal-fetal attachment (61). Our data suggests similar findings in Rawalpindi, Pakistan, a lower middle income setting. Moreover, despite the important role played by a partner in maternal well-being and child development during the postpartum period, discussion and research about men’s role in particular has been largely missing in LMICs (with the majority of similar studies being conducted in HICs) (62). However, in one qualitative study in Pakistan analyzing the agreement between non-anxious mothers’ and fathers’ stimulation of their child, the strength of agreement around engagement in stimulation appeared to be greater among couples with higher quality co-parenting relationships (63), suggesting the importance of such relationships for child development in the Pakistani context. Our study brings men into the framework of women’s mental well-being in the postpartum period, and quantitatively sheds light on what men can potentially do to improve the relationship between the mother and child. This implicates parents, irrespective of gender, as directly and indirectly critical to child developmental outcomes, and thus the role and input of both parents should be considered in clinical care when assessing child development dynamics. Our study contributes data to this knowledge gap, which is important in light of the fact that parenting practices, support received, and relationship and family dynamics vary across cultures. Research in LMICs on such topics is particularly vital since children in these settings are falling short of reaching developmental milestones (64,65).

Several studies in HICs have shown that social support reduces stress and raises self-esteem, consequently increasing maternal responsiveness (26,66,67). For instance, one US study of non-anxious women with preterm infants found that social support had a significant, positive relationship with maternal responsiveness (26). Likewise, a Korean study found that provision of social support was associated with increased maternal sensitivity and, along with several other factors, explained a large proportion of the variance in maternal sensitivity (68). Consistently, our findings indicate that social support plays an important role in maternal responsiveness among pregnant Pakistani women with at least some symptoms of anxiety. In Pakistani culture, chilla, a rest period of up to 40 days after childbirth (69), is typically a time when family members provide adequate social support, thereby, potentially improving mental health of postpartum women. Such traditional postpartum practices include a woman returning to her natal home to receive care, support, and relief from household responsibilities. Future intervention strategies aiming to promote the mental health of postpartum women could integrate provision of social support to mothers during the chilla period.

Moreover, wider cultural factors may influence the observed relationships. While conceptual and empirical research has principally occurred in HICs (24), prior research has emphasized the importance of accounting for cross-cultural dynamics in LMIC settings, such as the different roles or behaviors of mothers. Cultural norms and practices regarding spousal relationships vary across contexts. For example, Iqbal et al.’s study found that Pakistanis endorsed couple bond standards more than Westerners or Chinese, and family responsibility standards substantially more than Westerners or Chinese (70), while Qadir et al.’s study found that Pakistani women often view marriage as a social and familial obligation (71). The strength of family ties and gender dynamics in Pakistan may have a bearing on our results, especially when considering the significant positive associations of family social support among women with high levels of PPD, but not related to social support from friends. Perceived higher social support has been associated with marital satisfaction in Pakistan, and both of these have an important association with reduced risk of depression (35,41,52,72). These distinct cultural factors and their impact on health outcomes should be considered in programs when seeking to address mother-infant relationships, including family dynamics that may have a strong influence on both maternal responsiveness and maternal mental health in general.

Among depressed women, we found that spousal relationship quality (i.e., emotional support and less conflict) improved maternal responsiveness, whereas social support showed a null association with responsiveness, suggesting that spousal relationship quality is particularly important to depressed women. Our null findings that were observed among depressed women, could be a result of negative mood, depression, having a greater overall effect on maternal responsiveness than positive stimulators, e.g. social support and spousal relationship quality. This could imply that interventions in anxious women with symptoms of PPD should prioritize addressing depressive symptoms, unless those intervention strategies themselves involve the enhancement of social support and improving spousal relationships. However, additional research is required to more thoroughly understand these null associations, particularly given the importance of emotional support and less spousal conflict with spouse for maternal responsiveness among depressed women.

Overall, our findings indicate complex relationships between relationship quality with spouse, social support, PPD, and maternal-infant responsiveness that are a novel contribution to the literature from LMICs. In the Pakistani context among women with symptoms of anxiety, we observed that a mother’s close relationships to both the spouse and to wider social networks had an impact on responsiveness to their infant, only among the non-depressed, indicating positive maternal mental health (lack of depression) and maternal relationships are pathways to influencing responsiveness in this group. To our knowledge, this is the first study investigating these associations in an LMIC. Research from the HICs may shed some light on the mechanisms involved. One such study concentrated on the positive relationship between self-focus (i.e., use of “I” in speech) and depressive symptoms (73) and examined the associations between self-focus and responsiveness. That study, conducted in the UK, found that greater partner-inclusive talk – use of “we” in parents’ speech, as opposed to self-focus – was associated with reduced sensitivity to infants’ cues in mothers (74), possibly also indicating lower responsiveness. However, another study from Australia suggested that the relationships between child development outcomes and maternal mental health (i.e., depression and anxiety) were largely independent of social support (75). Our results indicate that more research is required to understand the multifaceted relationships between these variables, and thus the factors advance positive maternal health and child development in different settings.

In addition to being the first study in a LMIC to explore social support and spousal relationship quality with maternal responsiveness, we do not know of other research that has examined these relationships specifically among women with anxiety. Among the limitations of our study, firstly, data were collected at a single time point. Secondly, data collection occurred over a three-year period, April 2019 – October 2022, which may have introduced variability due to potential shifts in social, economic, or healthcare-related factors during this time, such as the COVID-19 pandemic. Thirdly, while the MSPSS and PHQ-9 were validated in Pakistan (30,76), the spousal relationship quality and MIRI scales were not previously used in our study population. It is not known if our results are generalizable to non-anxious women or a broader population, given we drew participants from one public hospital in Rawalpindi serving people with low and middle socioeconomic status; thus, more research is needed to establish if this is the case.

We found that less social support and weaker spousal relationship quality was directly related to maternal infant relationships after pregnancy, with potential implications for child development. The United Nations’ Sustainable Development Goal 3 advocates for healthy lives and well-being, including for infants, children, and parents. Our study contributes to this agenda, by providing insight into maternal responsiveness – critical to child development – and factors that support women in their parenting experience, including those with PPD. Our findings indicate that improving child developmental outcomes – of significant concern in LMIC settings – requires the attention of the wider family and social context. Maternal responsiveness, a key to child development, is enhanced through strong social support and high spousal relationship quality. The strong connections between maternal mental health, maternal-infant relationships, and the spousal and social environment, imply that policies and programs should include mothers, fathers and wider support networks in order to more effectively address child developmental concerns.

Prior research indicates the importance of accounting for cultural context when developing interventions. One study in Pakistan found that gender-sensitive and culturally-appropriate social support can promote the mental health of women from low socioeconomic backgrounds (77), while analysis of Pakistani couples suggested that couples therapy could be adjusted to specific relationship standards and expectations among such couples (78). For PPD, prior studies in Pakistan have highlighted the important role that nurses can play in identifying women at risk and connecting them to relevant treatment (79), and that found participation in chilla and similar traditional postpartum practices can be used as entry points for creating low-cost, community-based, sustainable interventions (69). Further research is required, however, to more deeply understand the associations between maternal responsiveness and partner dynamics in particular, given how it is much less studied than social support. Given the critical role responsiveness plays for child development and the high levels of depression among women globally as well as in LMIC settings, the fact we did not observe social support in general or from friends to enhance maternal responsiveness among women with high levels of depressive symptoms suggests that more research is needed to identify protective factors for these women.

SUMMARY

Our study offers unique insights into the associations between spousal relationship quality/ social support and maternal responsiveness to infants among anxious mothers, in the lower middle income setting of Pakistan. We found that social support from family and friends was related to responsiveness only in non-depressed women, while spousal relationship quality was related to responsiveness in both non-depressed and depressed women. This suggests that spousal relationship quality may be particularly important for maternal responsiveness to infants of depressed women. More research is needed to identify what aspects of social relationships are protective for anxious women, given their important impact on maternal responsiveness, which is in turn critical for child development.

Availability of data and materials

Data used in this study can be accessed at the US National Institutes of Health, National Institute of Mental Health (NIMH) Data Archive: https://nda.nih.gov/.