Infant social withdrawal and rapid infant weight gain

Abstract

Background

Rapid infant weight gain (RIWG) and infant social withdrawal are important early markers of risk of subsequent overweight/obesity and mental health problems, respectively. While overweight and mental health problems are linked later in life, it remains unclear if this association is present already in infancy.

Objective

We examine if infant social withdrawal is associated with subsequent RIWG independently of an array of known risk factors (gestational age, birth weight, sex, parity, breastfeeding, maternal postpartum depressive symptoms, maternal smoking during pregnancy, maternal age at birth and pre-pregnancy BMI, family educational level, and employment status).

Methods

In a cohort of 12,468 children born 2015-19 in Copenhagen, we analyzed the association between infant social withdrawal around age 3 months, measured with the Alarm Distress Baby Scale (ADBB)—a validated screening tool, and subsequent RIWG. We retrieved information on social withdrawal and weight during infancy from health visitors’ records and linked these to register data. Multivariate logistic regression was used to estimate adjusted odds ratios (aOR) and 95% confidence intervals (CI) for RIWG, adjusting for the known risk factors outlined above.

Results

Children showing signs of social withdrawal had 1.19 higher odds of RIWG compared to children showing no signs of social withdrawal (aOR 1.19; 95% CI: 1.03, 1.37, p = 0.017). RIWG was also more likely among boys, firstborns, children born small-for-gestational age, and those with mothers who smoked during pregnancy or had higher pre-pregnancy BMI. In contrast, higher birth weight and breastfeeding >4 months were protective factors. No associations were found with maternal postpartum depression, low education, parental labor market status, or maternal age.

Conclusions

We found an association between infant social withdrawal and subsequent rapid infant weight gain independently of an array of known risk factors, which may expand our understanding of the developmental origins of excess weight gain and its association with early childhood mental health.

Introduction

Obesity and mental health problems in children are both significant and growing public health challenges [1, 2]. In adults and youths, a bidirectional link between excess weight and mental health difficulties is well established [3], yet the developmental origins and mechanisms driving this association remain unclear. Early childhood represents a critical developmental window, but few studies have explored how early distress may influence weight trajectories in infancy [4–6].

Rapid infant weight gain (RIWG), typically defined as an increase of >0.67 in weight-for-age z-scores (WAZ), is a strong predictor of later overweight and obesity [7–11]. Several biological, social, and behavioral factors contribute to RIWG. For example, children born prematurely or small-for-gestational-age often experience catch-up growth, partly due to suboptimal in utero environments and prenatal stress exposure [4, 12, 13]. Maternal smoking during pregnancy further increases the risk of preterm birth, low birth weight, and subsequent RIWG [14–16]. Postnatal feeding practices also affect growth; infants who are formula fed, as opposed to breastfed, are more likely to experience RIWG [17, 18]. Moreover, the timing of introduction of food and feeding practices have been linked to weight development [5, 19]. Certain background characteristics also influence whether infants grow disproportionately. First-born children are more likely to experience RIWG, possibly due to breastfeeding difficulties being more common among first-time mothers or because first-born children more often have a lower birth weight [16, 20]. Boys also have a higher likelihood of RIWG than girls [20], which may reflect sex-specific differences in placental function and biomarkers affecting fetal and infant growth [21]. Moreover, children from lower socio-economic households face a higher risk of obesity [22].

Although stress, both psychological and biological, has been associated with excess body weight in adolescents and adults [23–26], its role in infant growth remains poorly understood because of difficulties in measuring infant stress directly [5]. Infants are shaped by both the prenatal and postnatal environment, where early stress exposure may influence growth through programming of the hypothalamic–pituitary–adrenal (HPA) axis, affecting physiological stress regulation as well as the infant’s behavioral responses to stress [27]. Maternal stress and caregiving behavior can further reinforce these stress patterns after birth, shaping both emotion regulation and feeding behaviors that may contribute to RIWG. Prenatal oxidative stress, for example, has been linked to RIWG [12], potentially because mothers exposed to greater psychobiological stress are more likely to engage in obesogenic feeding practices such as formula feeding, early introduction of complementary foods, and feeding to soothe [5]. While stressful life events experienced during pregnancy have been shown to increase the odds of RIWG [4], findings regarding maternal perceived stress and postnatal depressive symptoms are inconsistent [4, 28, 29]. Thus, how a stressful early environment influences infant growth remains unclear, underscoring the need for studies that directly measure infant stress.

This study addresses this gap by examining whether infant social withdrawal—assessed with the validated instrument Alarm Distress Baby Scale (ADBB)—is associated with rapid infant weight gain during the first year of life, independent of known biological, social, and behavioral risk factors. Using data from the Danish home visiting program, where health professionals routinely employ the ADBB to assess infants aged 2–24 months [30], we test whether early social withdrawal signals elevated risk for disproportionate weight gain. Social withdrawal reflects early dysregulation in stress and attachment systems and is associated with congenital conditions, adverse caregiving, and later developmental and behavioral disorders [6, 30–34]. We hypothesize that infants displaying higher levels of social withdrawal during early infancy have increased odds of experiencing rapid weight gain during the first year of life, even after accounting for established biological, social, and behavioral risk factors.

Methods

The Danish ADBB infant cohort

In this study, we used data from the Danish ADBB Infant Cohort (DAIC) to examine infant social withdrawal and rapid infant growth. DAIC contained data from health visitors’ records. All families in Copenhagen with a newborn child were offered routine home visits by a public health visitor as part of the national social security program [35]. The health visitors were automatically informed about all deliveries and visited the families several times during the first year. The first two visits occurred during the first two weeks after birth. The subsequent visits occurred when the infant was 3–4 weeks, 2–3 months, and 8–10 months old. First-time parents were also offered a visit when the child was 4–6 months old. Each visit included anthropometry, evaluation of motor and speech development, guidance of infants’ emotional and developmental needs and feeding, support for the parents in their new roles. From age 2 months, the visits also included screening for maternal depressive symptoms using the Edinburgh Postnatal Depression Scale (EPDS) and screening for infant social withdrawal using the Alarm Distress Baby Scale (ADBB). All visits were reported in electronic health visitor records using a unique identifier for each individual. Based on this unique identifier, we linked the health visitors’ records with additional information from the Danish registers. From the Danish registers, we retrieved information on birth weight, gestational age, pre-pregnancy BMI, mother’s age at birth, smoking status during pregnancy, parental educational level and job status.

Rapid infant weight gain

Our outcome of interest was rapid infant weight gain (RIWG), defined as an increase in weight-for-age z-scores (WAZ) above 0.67 of a standard deviation between two time points in infancy following the ADBB screening. We calculated the weight gain as the difference in WAZ-scores between the weight measure from the visit where the ADBB was conducted (at 2–3 months old) and the weight measure from the last visit during the first year of life (at 8–10 months old). We used WHO age- and gender-specific growth standards to calculate the WAZ-scores using the lambda-mu-sigma formula [36]:

$$Z=\frac{{\left[\frac{W_t}{M_t}\right]}^{L_t}-1}{S_t{L}_t}$$

Where each individual weight measure (W) at age t (measured in days) was transformed using gender-specific parameters from WHO: median weight (M) at age t, skewness parameter (L) at age t, and variability parameter (S) at age t. To account for differences in timing between measurements, we controlled for the age gap between the two visits where weights were recorded, as previous research has shown that the association between RIWG and later obesity risk depended on the length of the period of rapid weight gain exposure [9].

Infant social withdrawal

Social withdrawal was measured with the ADBB scale when children were 2–3 months old. The ADBB is an observer-rated tool that assesses infant social withdrawal in relation to a stranger (i.e., the health visitor) on 8 items: facial expression, eye contact, general level of activity, self-stimulating gestures, vocalizations, response to stimulation, relationship with the observer, and capacity to attract and maintain attention. Infant social withdrawal is characterized by less frequent eye contact, fewer emotional displays, less vocalizations, a decreased level of activity, increased level of self-stimulation and delayed reaction time [30, 32]. The items are rated on a Likert scale from 0 to 4, with higher scores indicating more social withdrawal. International studies using the ADBB in pediatric settings have validated a cut-off of 5, though a Norwegian study suggested a lower cut-off of 3 may be more appropriate in home visiting contexts like Denmark [37]. Pediatric visits, where the ADBB has been used internationally, are often 15 min long, whereas home visits often last 1 h, allowing the health visitor more time to interact with the infant and observe social engagement. Thus, we used a dichotomous variable taking the value 1 if children scored 3 or above on the ADBB and 0 otherwise. Supplementary Table 1 shows the number of children scoring above the cut-off in the current sample using different cut-offs on the ADBB score.

Control variables and confounders

We controlled for socio-economic differences by including dichotomous variables for mothers and fathers with lower education, defined as no more than high school, and dichotomous variables for mothers and fathers without a job in the year before the child is born (including unemployed and people on public benefits). Moreover, we included a dichotomous control for maternal smoking during pregnancy and a control for the mother’s pre-pregnancy BMI (weight(kg)/height(m)^2). Mother’s age at birth was calculated subtracting the birthday of the child from the mother’s birthday. Maternal postpartum depressive symptoms was assessed using the Edinburgh Postpartum Depression Scale (EPDS) by Cox, Holden [38]. The EPDS is a 10 item self-report questionnaire ranging from 0 to 30 designed to screen for possible maternal depression postpartum. We included a dichotomous control for postpartum depressive symptoms using a cutoff of 11 as validated in a Danish context [39].

We controlled for the sex of the child by including a dichotomous variable taking the value 1 if the child was a boy and 0 if the child was a girl. Likewise, we controlled for parity by including a dichotomous variable for firstborn children. We constructed a dichotomous variable indicating children who were born small-for-gestational age (SGA), i.e., below the 10th percentile, using individual information on birth weight and gestational age from the registers in combination with the gender-specific fourth degree polynomial by Maršál, Persson [40]. Moreover, we added controls for gestational age and birth weight, with birth weight standardized to z-scores based on WHO reference data [41]. We constructed a variable taking the value 1 for children who were breastfed at least 4 months and 0 for children who were never breastfed or breastfed for less than 4 months. The longitudinal nature of the registers and the health records allowed us to measure the confounding variables before the ADBB screening and the period where we observed the weight gain (except breastfeeding duration). A timeline of when the different variables are measured is displayed in Fig. 1 and a correlation matrix is displayed in Supplementary Table 2.

Fig. 1. Timeline of Measurement of Included Variables.

Data was collected at pre-pregnancy, pregnancy, birth, and around 3 and 8 months postpartum.

Sample

We gathered information on children born 2015–2019 living in the municipality of Copenhagen. We included all children for whom we had a measure of ADBB and two weight measures, where the first weight measure was allowed to be concurrent with the ADBB, and the last weight measure was after the ADBB and before the child turned 1 year old. Moreover, we required that we had (valid) information on the child’s weight at birth, gestational age, the mother’s pre-pregnancy BMI, and an EPDS screening of the mother. Only children with gestational age >=35 weeks were included, resulting in a final sample of 12,468 children, including 156 twins. The sample selection process is presented in Fig. 2. Potential selection bias was evaluated by comparing characteristics of infants included and excluded from the analytic sample according to ADBB screening status. Supplementary Table 3 shows no evidence that children with an ADBB score were systematically positively or negatively selected.

Fig. 2.

Flowchart of the Study Population.

Statistical analysis

We first compared key characteristics between infants with and without social withdrawal using two-sample t-tests. Next, we examined unadjusted associations between RIWG and potential risk factors using logistic regression.

The primary analysis assessed the association between infant social withdrawal and RIWG using a multivariate logistic regression model. RIWG was defined as a binary outcome taking the value 1 if the infant’s weight-for-age z-score increased by more than 0.67 standard deviations between the two weight measurements, and 0 otherwise. Infant social withdrawal was defined as scoring 3 or above on the ADBB screening.

The model was adjusted for a range of confounding variables previously identified as relevant to both infant social withdrawal and RIWG: gestational age, small-for-gestational age, family socioeconomic position, maternal postpartum depressive symptoms, birth weight, infant sex, and breastfeeding status. Additional control variables included the time interval between weight measurements, maternal pre-pregnancy BMI, maternal smoking during pregnancy, firstborn status, and maternal age at birth, as these factors were considered related to RIWG but not to ADBB scores [13, 20, 32, 34, 42, 43].

Figure 3 provides an overview of the analytical framework, illustrating the hypothesized association between infant social withdrawal and RIWG, as well as the variables treated as confounders and control variables in the analysis.

Fig. 3.

Graphical representation of the associations between included variables in the statistical models.

To assess the robustness of our results, we first applied alternative cut-offs on the ADBB scale and used the continuous ADBB score as the main explanatory variable to test the sensitivity of the findings to different definitions of social withdrawal. We further examined the potential for overadjustment bias by excluding gestational age, birth weight, and SGA one at a time. Because information on breastfeeding duration was partly collected after the ADBB screening, we also re-estimated the models excluding this variable. Finally, to ensure that results were not driven by low birth weight infants or twins, we repeated the analyses after excluding them from the sample.

Results

Summary statistics

Table 1 shows the mean and standard deviation for continues variables and percentage and number of infants for dichotomous variables of variables used in the analysis for the total sample in column 1 and separately for children showing no signs of social withdrawal and for children showing signs of social withdrawal in columns 2 and 3, respectively. There were 1034 (or 8.3%) children in the sample with an ADBB score of 3 or above and thus showing signs of social withdrawal. Overall, 29.1% of the children experienced RIWG from around age 3 months to around age 8 months. More children with social withdrawal symptoms experienced RIWG (37.2%) compared to children without social withdrawal symptoms (28.4%). Children with social withdrawal were more often boys, had a lower birth weight, had lower gestational age, were more likely to be small-for-gestational age, and more likely to be large-for-gestational age. They were also less likely to be breastfed for more than 4 months. The age gap between the two weight measures was on average 5 days larger for children with social withdrawal symptoms. The mothers of children with social withdrawal were more likely to had smoked during pregnancy, had postpartum depressive symptoms, and more likely unemployed. Moreover, the fathers of children with social withdrawal were more likely to be low educated and unemployed. There were no significant differences in parity, mothers’ pre-pregnancy BMI, mothers’ age at birth, and the fraction of mothers with low education between children with and without social withdrawal symptoms.

Table 1.

Summary statistics for the total sample and by social withdrawal.

	All	No social withdrawal	Social withdrawal
	Mean (SD) or % (n)	Mean (SD) or % (n)	Mean (SD) or % (n)	Difference	P-value
RIWG (0/1)	29.10 (3628)	28.36 (3243)	37.23 (385)	−8.87	<0.001
Boy (0/1)	50.22 (6261)	49.83 (563)	54.45 (5698)	−4.62	0.005
Time between the two weight measures (days)	156.9 (49.65)	156.4 (49.32)	161.5 (52.99)	−5.08	0.002
Birth weight (z-score)	0.0395 (0.974)	0.0573 (0.965)	−0.156 (1.054)	0.21	<0.001
Gestational age (days)	279.9 (9.721)	280.2 (9.533)	276.1 (10.91)	4.11	<0.001
Small-for-gestational age (0/1)	11.94 (1 489)	11.77 (1346)	13.83 (143)	−2.06	0.051
Average-for-gestational age (0/1)	81.22 (10 126)	81.55 (9324)	77.56 (802)	3.99	0.002
Large-for-gestational age (0/1)	6.84 (853)	6.68 (764)	8.61 [89]	−1,93	0.019
Firstborn (0/1)	70.68 (8 812)	70.66 (8 079)	70.89 (733)	−0.23	0.876
Mother’s pre-pregnancy BMI	23.03 (3.873)	23.01 (3.840)	23.24 (4.213)	−0.23	0.069
Breastfed >= 4 mth. (0/1)	54.36 (6 777)	55.02 (6291)	47.00 (486)	8.02	<0.001
Smoking during pregnancy (0/1)	6.93 (864)	6.76 (773)	8.80 [91]	−2.04	0.013
Mother depressive symptoms (0/1)	8.21 (1 024)	7.75 (886)	13.35 (138)	−5.60	<0.001
Mother’s age at birth (years)	31.90 (4.590)	31.88 (4.568)	32.15 (4.820)	−0.27	0.072
Mother low education (0/1)	5.70 (711)	5.66 (647)	6.19 [64]	−0.53	0.481
Mother no job (0/1)	17.49 (2 181)	17.18 (1964)	20.99 (217)	−3.81	0.002
Father low education (0/1)	8.17 (1 019)	8.01 (916)	9.96 (103)	−1.95	0.028
Father no job (0/1)	15.86 (1 978)	15.63 (1787)	18.47 (191)	−2.84	0.017
Observations	12,468	11,434	1034

Column 1 shows baseline characteristics of the sample. Columns 2 and 3 show baseline characteristics by social withdrawal status. For continuous variables the columns report mean coefficients and standard deviations (SD) in parentheses. For all dichotomous variables indicated by (0/1) the columns report percentages of the sample (%) and number of children (n) with the given characteristic. Column 4 shows the difference in means between column 2 and 3, and column 5 shows the corresponding p-values from a t-test.

Regression results

Table 2 displays the odds ratios for RIWG of all the included confounders and control variables. Column 1 shows the unadjusted odds ratios between RIWG and each of the risk factors with corresponding confidence intervals in column 2. In the unadjusted model, children showing signs of social withdrawal had 1.50 higher odds for RIWG. Similarly, increasing the ADBB score by one unit increased the odds for RIWG by 1.11. In line with previous research our study confirmed that many of the known risk factors were related to RIWG: Children who were boys, had more time to grow, born SGA, and firstborn were more likely to experience RIWG. Moreover, we found higher odds for RIWG for children whose mothers had a higher pre-pregnancy BMI, smoked during pregnancy, and whose parents had low education. Birth weight, gestational age, and breastfeeding for more than 4 months were negatively associated with RIWG. We found no association between RIWG and mother’s depressive symptoms, mother’s age at birth, and mothers and fathers without a job.

Table 2.

Odds ratios for rapid infant weight gain.

	Unadjusted OR		Adjusted OR
	OR	95% CI	aOR	95% CI	aOR	95% CI
Social withdrawal	1.498***	[1.312,1.711]	1.190**	[1.033,1.371]
Total ADBB score	1.111***	[1.077,1.146]			1.049**	[1.011,1.088]
Boy	1.145***	[1.060,1.237]	1.165***	[1.074,1.263]	1.165***	[1.074,1.263]
Time between the two weight measures	1.008***	[1.007,1.009]	1.009***	[1.008,1.010]	1.009***	[1.008,1.010]
Birth weight (z-score)	0.660***	[0.633,0.688]	0.794***	[0.745,0.846]	0.794***	[0.745,0.847]
Gestational age (days)	0.963***	[0.960,0.967]	0.975***	[0.970,0.981]	0.976***	[0.971,0.981]
Small-for-gestational age	1.878***	[1.681,2.100]	1.256***	[1.083,1.457]	1.255***	[1.082,1.456]
Firstborn	1.241***	[1.138,1.354]	1.439***	[1.304,1.588]	1.438***	[1.303,1.586]
Mother’s pre-pregnancy BMI	1.031***	[1.021,1.041]	1.035***	[1.024,1.045]	1.035***	[1.024,1.045]
Breastfed >= 4 mth.	0.700***	[0.648,0.757]	0.676***	[0.623,0.734]	0.677***	[0.623,0.735]
Smoking during pregnancy	1.405***	[1.216,1.624]	1.297***	[1.108,1.519]	1.295***	[1.107,1.516]
Mother depressive symptoms	1112	[0.969,1.277]	1025	[0.885,1.187]	1022	[0.882,1.183]
Mother’s age at birth	1006	[0.998,1.015]	1.008*	[0.999,1.018]	1.008*	[0.999,1.018]
Mother low education	1.293***	[1.102,1.517]	1102	[0.917,1.325]	1101	[0.916,1.323]
Mother no job	1059	[0.957,1.171]	1012	[0.906,1.131]	1010	[0.904,1.128]
Father low education	1.180**	[1.028,1.354]	0.968	[0.831,1.128]	0.967	[0.830,1.127]
Father no job	1049	[0.944,1.165]	0.947	[0.846,1.061]	0.947	[0.845,1.061]
Observations	12,468		12,468		12,468

The table presents odds ratios (OR) and adjusted odds ratios (aOR) with 95% confidence intervals in brackets for the association between infant social withdrawal and rapid infant weight gain (RIWG). Columns 1–2 show unadjusted ORs; columns 3–6 show aORs, with dichotomous social withdrawal as the main predictor in column 3 and continuous ADBB score in column 5.

*p < 0.10, **p < 0.05, ***p <0.01.

In our preferred model in column 3, children, who showed signs of social withdrawal, had 1.19 higher odds for RIWG compared to children not showing signs of social withdrawal, when adjusting for all the other risk factors and control variables. As in the unadjusted model, children who were boys, firstborn, born SGA, and where the mothers had smoked during pregnancy were more likely to experience RIWG. Moreover, children born to mothers with higher pre-pregnancy BMI had higher odds for RIWG. Birth weight and breastfeeding for more than 4 months were negatively associated with RIWG. We found no statistically significant associations between RIWG and mother’s postpartum depressive symptoms, the mother’s age at birth, low education and labor market status of the parents.

To test the robustness of our results based on the cutoff of 3 on the ADBB scale, Table 2 reports OR and aOR for RIWG using the continuous ADBB score as the explanatory variable of interest in columns 1 and 5, respectively. In the adjusted model, a one-point increase in the ADBB score was associated with 1.05 higher odds of RIWG. Moreover, Supplementary Table 4 shows results using alternative cutoffs on the ADBB scale. The findings were robust for cutoffs between 1 and 4, whereas the cutoff of 5 yielded an aOR of 0.88 (95% CI: 0.602–1.293), suggesting limited statistical power at this threshold. Supplementary Table 5 presents aOR from models excluding birth weight, gestational age, and SGA one at a time, as well as a model including SGA and LGA but excluding birth weight and gestational age. Overadjustment bias did not seem to affect our results, as the association between social withdrawal and RIWG remained positively statistically significant across all model specifications. Supplementary Table 6 shows that the results were robust to excluding breastfeeding duration from the model. Finally, the results remained consistent when low birth weight infants and twins were excluded from the sample (Supplementary Table 7 and 8).

Discussion

This study examined the association between infant social withdrawal and rapid infant weight gain between 3 and 8 months of age. We found that infants showing signs of social withdrawal at 3 months had higher odds of experiencing RIWG even after adjusting for a wide range of known biological, behavioral, and socioeconomic risk factors. Consistent with previous research, our results confirmed that male sex, time between weight measures, lower birth weight, lower gestational age, being small-for-gestational age, and being firstborn were associated with higher odds of RIWG. Our results also confirmed that maternal smoking during pregnancy, pre-pregnancy BMI, and breastfeeding for less than four months increased the odds of RIWG. No association was observed with parental education and unemployment status, mother’s age at birth, and maternal postpartum depressive symptoms once covariates were adjusted for. The persistence of the association between social withdrawal and RIWG after adjustment for known correlates of RIWG suggests that early indicators of infant distress may independently contribute to shaping growth trajectories.

Our findings contribute to the existing literature linking mental health and excess body weight by uncovering that infant social withdrawal—an early warning sign of infant distress and depression—is linked with RIWG in the first year of life. This extends existing research on the relationship between high body weight and poorer mental health observed in adolescents and adults [23–26] to the very earliest stages of human development. The presence of this association already in infancy suggests that biological and behavioral mechanisms linking emotional distress and weight gain may emerge far earlier than previously documented.

Two broad explanatory pathways linked mental health and weight development have been proposed in the literature: biological and behavioral. Among the biological mechanisms, dysregulation of the hypothalamic-pituitary-adrenocortical (HPA) axis represents a key candidate. The HPA-axis regulates the body’s response to stress and affects neuroendocrine and metabolic processes [26]. HPA-axis hyperactivity, reflected in elevated cortisol levels, has been linked to depression and overweight [26]. In infancy, sustained distress might trigger similar dysregulation, altering the energy metabolism and promoting fat accumulation. Supporting this idea, a recent theory posits that social disruptions or perceived threats of food scarcity can activate an adaptive mechanism that promotes gradual fat accumulation as protection against perceived future scarcity, leading to increased energy storage even under conditions of food abundance [44]. Complementary to this, recent research suggests that during childhood, brain development and body growth compete for limited metabolic resources. When more energy is directed toward brain development—particularly in energetically costly areas such as the prefrontal cortex—less energy may remain for physical growth and fat deposition [45, 46], creating a perceived threat of food scarcity. In this context, infant social withdrawal may act as a neurobiological stress signal that promotes excess weight gain through such adaptive, yet maladaptive, mechanisms.

Inflammation provides another possible biological link. Overweight is well established to be associated with systemic inflammation [47–49], which in turn may influence neural and emotional development. Though evidence in infancy remains limited, inflammatory processes have been implicated in both psychological distress and depression-like behaviors in animal studies [26]. Our finding that early distress predicts RIWG may therefore reflect shared inflammatory pathways influencing both emotional and metabolic regulation.

Alternatively, behavioral mechanisms may explain the association. Feeding practices play a crucial role in shaping infant weight gain, and parental responses to infant cues are central to this process. Infants are born with the ability to sense hunger and satiety cues but depend on caregivers to interpret these signals accurately [50]. A recent study found that formula feeding, early introduction of complementary foods, scheduled feeding, and TV exposure while feeding all increase the risk of RIWG [5]. They argue that mothers who are responsive during feeding, read infant hunger and satiety cues accurately and respond appropriately might promote infants’ awareness of and appropriate response to their internal hunger and satiety cues, preventing overeating and greater weight gain. In line with these results, we find that infants who have been breastfed for at least 4 months have significantly lower odds of RIWG. However, infants are limited in their ability to communicate and thus parents may find it difficult to interpret infant signals. An infant turning its head might for example both be a satiety cue and at the same time a symptom of withdrawal behavior. Thus, it is possible that infants showing more withdrawal are underfed, which then triggers a fear of food scarcity leading to RIWG [51]. Future studies should explore how feeding practices and social withdrawal interact in their association with RIWG.

Our results may also partly reflect underlying neurodevelopmental factors. Social withdrawal has been associated with later diagnoses of autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD) [6, 32, 52]. Both conditions are known to involve atypical feeding patterns [53, 54]. These potential underlying neurodevelopmental conditions may partially contribute to the observed association.

The main strength of this study is that we build our analysis on a large cohort of children for whom we have rich data on their socio-economic background from administrative registers in combination with records of weight measures and systematic ADBB and EPDS screenings conducted independently of the research question. While the Danish home visiting program provides universal and systematic data collection, some children were not screened with the ADBB, and missing covariate data reduced the final analytic sample, which may limit internal validity. The study sample is also restricted to children from Copenhagen—the capital of Denmark—thus and replication in more diverse populations is warranted. We have limited information on fathers, and no information about parental sensitivity and ability to read infant hunger and satiety cues accurately. Additionally, we cannot determine causality or completely rule out underlying reverse associations between growth and social withdrawal. As the literature on early-life emotional and metabolic development advances, future research will be better positioned to formalize causal pathways building on the associations documented in this study. Finally, future research should assess whether the associations identified may reflect early manifestations of neurodevelopmental conditions, such as ASD and ADHD.

Taken together, our findings provide novel evidence that early emotional distress, reflected in infant social withdrawal, is associated with disproportionate weight gain in the first year of life. This suggests that the biological and behavioral processes linking emotional regulation and energy balance may begin far earlier than previously recognized. Future research should examine whether these early distress signs predict longer-term trajectories of weight and investigate the underlying mechanisms. Such knowledge may ultimately inform early intervention and prevention strategies aimed at supporting both emotional wellbeing and healthy growth in infancy and childhood.

Author contributions

SS: conceptualization, literature review, data analysis and interpretation, drafting the original manuscript. MSV: conceptualization, acquiring funding, data acquisition, data interpretation, critical revision of the manuscript. MS: data acquisition and management, critical revision of the manuscript. ACS: conceptualization, data interpretation, critical revision of the manuscript. TIAS and JEWM: data interpretation, critical revision of the manuscript.

Data availability

The data that support the findings of this study are securely stored at Statistics Denmark and restrictions apply to the availability of these data. The data were used under license for the current study and so are not publicly available. The data can, however, be accessed remotely from within Danish universities and research institutions. If researchers want to analyze our data for replication purposes, we will provide guidance regarding getting a project approval at Statistics Denmark and provide all programs and instructions to any researcher who should wish to replicate our paper. We will keep the data and programs for five years after acceptance to facilitate such replication in this period.

Competing interests

SS, ACS, JEWM and MSV are employed at the Centre of Excellence in Early Intervention and Family Studies where training in the use of the ADBB scale is offered as part of the continuing education program at the University of Copenhagen.

Ethics approval and consent to participate

Under Danish law, research projects based solely on existing register data do not require individual informed consent. The DAIC project, of which the current study is a subproject, was approved by the ethics committee at the Department of Psychology, University of Copenhagen (ref number: IP-EC-08042025-1).

Supplementary information

The online version contains supplementary material available at 10.1038/s41366-025-01996-y.