Beyond Early Adversity: The Role of Parenting in Infant Physical Health

Abstract

Objective:

Although ample evidence indicates that child health is compromised by early adversity (e.g., abuse and poverty), less is known about the contribution of parenting in low-stress contexts to child health, especially in infancy. This longitudinal study extends previous research on early adversity to ask the question: Does quality of parental care predict infant health in a low-risk community sample?

Method:

Participants were 187 healthy mothers and their full-term infants (86 girls) from the Netherlands, followed from birth to age 1. Home observations of mothers’ behavior were conducted during a naturalistic task (bathing session) when infants were 5 weeks old. Trained researchers interviewed mothers about the infants’ health and prescribed antibiotic use every month for 12 months. Infant health problems were categorized into 4 domains according to the International Classification of Primary Care to capture a range of outcomes: respiratory, digestive, skin, and general illnesses and symptoms.

Results:

Controlling for health-related covariates (e.g., maternal smoking and breastfeeding), maternal sensitivity predicted reduced rates of infant respiratory symptoms and skin conditions and marginally lower prescribed antibiotic use over the first year. Maternal behavior was unrelated to infant digestive and general illnesses.

Conclusion:

Even in the absence of adversity, quality of maternal care may have implications for the development of physical health, beginning as early as the first year of life. That such findings emerge in a low-risk sample helps rule out potential confounders and underscores the importance of parenting for physical and psychological health outcomes.

Previous research on parenting and health has largely focused on at-risk populations. A large body of work has chronicled the effects of extreme caregiving adversity on child health, including abnormal brain development and physical health.¹ Harsh family climate, parental abuse, and neglect have been linked to child low-grade inflammation and increased risk for cardiovascular illness, type 2 diabetes, and other diseases in adulthood.² These consequences persist across the life course, contributing to adult disease burden.^1,2 Emerging research has identified a variety of mechanisms linking early adversity to poor health, including inflammatory processes and disruptions in the stress response system,³ as well as epigenetic mechanisms.⁴

Additional work highlights the role of parenting in the course of illness among children with chronic diseases or those at high risk for developing disease. Research on childhood asthma links parenting difficulties to increased risk of early onset, ongoing morbidity, and mortality across childhood.⁵ Parenting also seems to play a role in symptom severity among children with severe atopic dermatitis⁶ and diabetes.⁷ Emerging evidence suggests that parenting interventions can improve health outcomes and psychological well-being for children with asthma and/or eczema.⁸ In sum, there is increasing recognition that quality of parental care contributes to children’s physical health in at-risk samples.¹

Less is known about the health implications of parental care among community samples that are not at risk because of disease or extreme adversity. This population is particularly important to study, given that (1) a sizeable proportion of patients seeking pediatric care are not at high risk, (2) it is possible that parenting has different effects on child health in high-risk versus community samples, and (3) examining the role of parenting in low-risk contexts may help rule out potential confounders in previous research related to socioeconomic status, such as nutrition. Some evidence suggests that even in the absence of abject adversity, negativity in the parent-child relationship (i.e., high levels of parental anger and conflict) is associated with higher rates of illness later in development.⁹ Conversely, parental warmth and support have been shown to predict improved health in childhood.¹⁰ Furthermore, positive parental relationships have been shown to buffer risk for adult cardiovascular disease,¹¹ obesity,¹² and other health problems.¹³ In 1 prospective longitudinal study, for example, observed quality of parental care at 3 months predicted reduced cardiometabolic risk at age 19.¹⁴ Preliminary evidence for a causal pathway between parenting and health-related risk factors comes from a randomized controlled trial of a family intervention for low-income mothers and their 11-year-olds: Intervention youth had lower inflammation than control youth at age 19, and this effect was partially mediated by improvements in parenting.¹⁵

The Present Study

Notably, however, very little research has examined the contribution of quality of parental care to physical health conditions and symptoms in infancy. Infancy is a particularly important developmental period for understanding the contributors to childhood health and is often considered a critical or sensitive window for the early programming of the immune system.¹⁶ According to the developmental origins of health and disease model, the first 1000 days of life represent a period of increased vulnerability and plasticity in which environmental factors may have lasting effects on developing biological systems that contribute to later health.¹⁷ For example, in a small sample of mother-infant dyads living in poverty (N = 49), researchers found that infants’ quality of attachment to parents was associated with levels of salivary C-reactive protein (a marker of systemic inflammation) at 17 months of age.¹⁸ Yet, whether similar associations are present in larger, nonimpoverished samples and whether they are apparent in markers of physical health symptoms are unknown.

The present study addresses these gaps in the literature by examining the role of maternal care in predicting infants’ physical health symptoms in the first year of life in a low-risk community sample. We leveraged the data available from a longitudinal study of Dutch infants and mothers to test whether parenting behavior contributes meaningfully to child health even in the absence of adversity; if such links emerge in this context—i.e., in a well-educated, low-risk, homogeneous Dutch sample—it would allow us to rule out confounders such as education, race, access to health care, and nutritional differences associated with poverty and point more directly to the unique role of parenting. If links do not emerge in such a sample, it would suggest that additional factors may be at play in the previous literature in high-risk samples.

We examine whether observed maternal behavior in the home predicts infant health symptoms and illnesses from birth to age 1. In contrast to previous studies using adult retrospective accounts of parenting, we implement a prospective longitudinal design, observing maternal behavior at 5 weeks and tracking infant health monthly (12 total assessments) over the first year. We focus specifically on low-level, frequent, ongoing symptoms (e.g., fever, rashes, and coughing), which are underexamined in the infant health literature yet are consequential for quality of life for both parents and children. To best capture such symptoms, which typically do not warrant doctor visits but are quite evident to caregivers, health data are derived from monthly structured interviews that trained researchers conducted with mothers. Given the dearth of research on quality of maternal care and health in early development, we adopt an exploratory approach, examining outcomes across 4 comprehensive domains of child health: respiratory, digestive, skin, and general illnesses and symptoms. We also examine the rates of prescribed antibiotic use.

METHOD

Participants

Pregnant women were recruited through midwife practices in the cities of Nijmegen and surrounding areas (the Netherlands). Inclusion criteria were as follows: an uncomplicated singleton pregnancy, fluency in Dutch, no drug use during pregnancy, no current major physical or mental health problems, full-term delivery (≥37 weeks), and a normal 5-minute infant Apgar score (≥7). Participants provided written informed consent; Radboud University’s Faculty Ethical Committee approved the study (approval number ECG300107).

Of the 220 women who enrolled in the study, 8 were excluded because they did not meet the study criteria (e.g., premature birth). Of the remaining 212 mothers, 19 ceased participation during the first 3 postpartum months because of lack of time or other personal circumstances. No demographic differences emerged between the participating mother-infant dyads (n = 193) and those who dropped out (n = 19). Table 1 displays participant characteristics. Given the inclusion criteria noted above (e.g., no current health problems and full-term delivery), mothers’ education level (77% attending college), and marital status (98% married), the sample was characterized as low-risk, in contrast to previous studies that focused on early adversity in high-risk samples.

Table 1.

Sociodemographics and Descriptive Statistics for the Main Study Variables

	N	%	Mean	SD	Range
Sociodemographics
Mother age (yrs)			32.4	3.8	21.1–42.9
Maternal education level
Primary education	5	3
Secondary education	38	20
College or university	144	77
Maternal marital status: married (n)	183	98
Child sex: girls (n)	86	46
Child Apgar score			9.66	0.63	7.00–10.00
Additional covariates
Smoking during pregnancy	7	4
Alcohol ingestion during pregnancy	27	14
Attendance at center-based child care	106	57
Birth weight (g)			3615.60	469.16	2645.00–4730.00
Duration of breastfeeding (mo)			5.47	4.29	0.00–12.00
No. of siblings			0.74	0.71	0.00–2.00
Maternal depression			5.03	3.15	0.00–22.33
Infant negative emotionality			2.49	0.49	1.33–4.12
Quality of maternal care
Sensitivity and cooperation			5.50	2.06	1.00–9.00
Child health
Respiratory			26.91	8.77	0.00–53.50
Digestive			5.07	3.95	0.00–19.00
Skin			5.81	4.07	0.00–23.50
General			6.76	3.73	0.00–18.00
Prescribed antibiotic use			0.65	1.03	0.00–6.00

Procedure

During a home visit when infants were approximately 5 weeks of age (M = 33.5 days, SD = 4.9), mothers were asked to bathe their infant as they normally would (i.e., undressing, bathing, and dressing). This task has been used in previous research to elicit mild stress in young infants and corresponding caregiving behavior in parents.¹⁹ The unstructured nature of the task in the context of the home environment allows observations to be as naturalistic and ecologically valid as possible. The bathing routine was unobtrusively video-recorded, and recordings were subsequently rated by trained coders for maternal sensitivity and cooperation, as described below.

Information on infant health was obtained through monthly semistructured maternal interviews conducted by a trained health interviewer across the first 12 months of life (3 in person, 9 by phone). Additional mother/infant information was obtained during these interviews and with maternal questionnaires during pregnancy, after delivery, and at 3, 6, and 12 months.

Measures

Quality of Maternal Care

Two independent coders, who had no additional information about the dyad, rated all bathing sessions. Coders used 2 of Ainsworth’s 9-point scales of maternal care: sensitivity versus insensitivity and cooperation versus interference. Sensitivity versus insensitivity reflects the extent to which the mother promptly and appropriately responds to the infant’s signals and needs (e.g., insensitivity is characterized by large delays in maternal responses to infant signals, including crying). Cooperation versus interference taps the extent to which the mother adjusts her behavior to the infant’s ongoing activity, rather than interfering with the infant’s actions (e.g., interference is characterized by maternal direct physical forcefulness of interruptions or restraints). These widely used scales show high stability across the first year of life,²⁰ and measurement at a single time point shows strong predictive validity regarding later child attachment security and other social-emotional outcomes.²¹

In the present study, interobserver reliability was excellent (intraclass correlations ≥ 0.94). Sensitivity and cooperation scores were highly correlated, r = 0.83, p < 0.05, as they were in the Ainsworth original sample (r = 0.87).²¹ Because of this high intercorrelation, the mean of the 2 scores was used as a composite sensitivity/cooperation score, as has been done in previous research.¹⁹ Higher scores reflect more sensitivity/cooperation.

Infant Health

Trained health interviewers administered the structured interviews with mothers every month for 12 months to track rates of infant health symptoms and prescribed antibiotic use. After an initial open-ended question (“Has your child been sick during the past month?”), mothers were asked to respond with “yes” or “no” to the presence of 24 common infant illnesses and health complaints (e.g., coughing, fever, and runny nose). Providing mothers with the 24-item list was intended to aid mothers’ memory and increase validity of the scoring. Subsequently, the health data were coded with the International Classification of Primary Care, which classifies illnesses and health complaints into nonoverlapping categories according to established criteria.²² Following previous research,²³ ear- and eye-related illnesses (mostly infections) were included with general illnesses because of their infrequent occurrence. Finally, frequencies of illnesses and complaints and of antibiotic treatments were summed over the 12 months. Five dependent variables were examined: respiratory, digestive, skin, general illnesses and complaints (which excludes respiratory, digestive, and skin conditions), and antibiotic use. This method has been used successfully in previous work on infant health,²³ and symptom scores on this measure have been linked with health-related biomarkers, such as infant intestinal microbiota.²⁴ Because it is rare for parents to make use of doctor visits for low-level health symptoms in the Netherlands, parent reports are a more valid index of common child health symptoms within this cultural context.

Analytic Approach

Covariates

To rule out potential confounders, we cast a wide net when considering possible covariates: maternal educational level, smoking during pregnancy, alcohol use during pregnancy, infant sex, infant birth weight, number of siblings, center-based care attendance (yes/no), and breastfeeding duration (in months). To control for maternal postnatal depression, mothers completed the Edinburgh Postnatal Depression Scale²⁵ at 3, 6, and 12 months postpartum. Because of moderate inter-correlations (r values > 0.50, all p values < 0.05), mothers’ mean depression score was used in the analyses. To rule out potential confounders with child temperament, mothers completed the Infant Behavior Questionnaire-Revised²⁶ when infants were 3 and 6 months old. The negative emotionality factor showed moderate stability across the 2 time points, r = 0.52; thus, an average negative emotionality score was used in the analyses.

Missing Data

Infants whose mothers missed 6 or more monthly health interviews were excluded from the study (n = 5); this 50% threshold was used to ensure that there were sufficient data to perform reliable multiple imputation. Of the remaining 188 infants, missing health data were as follows: no missing data (n = 125), missing data for 1 month (n = 50), for 2 months (n = 9), for 3 months (n = 3), and for 4 months (n = 1). Data for the missing months were imputed using the monthly average of the corresponding trimester.²³ Quality of maternal care data were missing for 1 infant (no bathing session), yielding a final sample of N = 187. To preserve power, multiple imputation with N = 40 imputations was used to treat missing data for covariates (<5% missingness for all variables); substantial research has demonstrated that multiple imputation is a valid and flexible tool for treating missing data in clinical research.²⁷ The results were the same when conducted using the imputed data versus the original data; thus, we report the results using the imputed data.

Statistical Analyses

Two multiple hierarchical regression models were computed for each infant health outcome. The first model contained all possible covariates in the first step and quality of maternal care in the second step (see Table S1, Supplemental Digital Content, http://links.lww.com/JDBP/A255). To eliminate irrelevant variables and increase power, the second and final model contained only covariates that explained at least 1% of the variance in the first model (following Beijers et al.²³). To correct for positive skewness in the regression residuals, the digestive, skin, and general health outcome variables, as well as antibiotic use, were square-root transformed, which effectively normalized the residuals. The final models are presented in the results section.

RESULTS

Preliminary Analyses

Descriptive statistics and simple correlations are presented in Tables 1 and 2, respectively. Higher quality of care was correlated with fewer infant respiratory symptoms and skin conditions. Furthermore, higher quality of care was related to lower infant antibiotic use. Effect sizes were small (r values ≤ 0.20). Greater antibiotic use was positively associated with all categories of infant health symptoms. Notably, regarding maternal and child characteristics, neither maternal depressive symptoms nor child negative emotionality were significantly related to maternal quality of care (p values > 0.11).

Table 2.

Bivariate Correlations Between Maternal Quality of Care and Infant Health Outcomes

	Infant Illnesses and Symptoms				Prescribed Antibiotic Use
	Respiratory	Digestive	Skin	General
Maternal quality of care	−0.20**	0.05	−0.18*	0.04	−0.15*
Infant illnesses and symptoms
Respiratory	—	0.24**	0.09	0.34***	0.26***
Digestive		—	0.18*	0.37***	0.14*
Skin			—	0.13	0.20**
General				—	0.48***
Prescribed antibiotic use					—

^*p ≤ 0.05,

^**p < 0.01,

^***p < 0.001,

2-tailed.

Infant Health

The regression results are presented in Table 3. Attendance at center-based child care emerged as a significant covariate in all models, likely reflecting, in part, children’s increased exposure to contagious illness in settings involving multiple children. Additional significant covariates were largely consistent with previous work examining environmental contributors to infant health.²³ Controlling for relevant covariates, higher quality of maternal care at 5 weeks predicted reduced infant respiratory symptoms and skin conditions across the first year of life; effect sizes were small (ΔR² = 0.03 and 0.04, respectively, p values < 0.05), although similar in magnitude to observed effect sizes for the established predictors of infant health, such as breastfeeding. The results were not significant for infant digestive or general illnesses and symptoms (Table 3). Regarding medication, after controlling for children’s number of siblings and attendance at center-based child care, higher quality of maternal care was a marginally significant predictor of lower prescribed antibiotic use, with a small effect size (ΔR² = 0.02, p = 0.088).

Table 3.

Regression Results Predicting Child Health Outcomes

	B (SE)	β	t	p	ΔR2
Respiratory illnesses and symptoms
Step 1					0.20*
Siblings	2.01 (0.84)	0.16	2.41	0.016
Alcohol use during pregnancy	−2.86 (1.68)	−0.12	−1.70	0.089
Duration breastfeeding	−0.34 (0.14)	−0.17	−2.48	0.013
Maternal depression	0.48 (0.19)	0.17	2.55	0.011
Attendance at center-based child care	6.83 (1.19)	0.39	5.73	<0.001
Step 2					0.03*
Quality of maternal care	−0.69 (0.28)	−0.16	−2.44	0.015
Digestive illnesses and symptoms
Step 1					0.11*
Maternal depression	0.06 (0.02)	0.20	2.70	0.007
Child negative emotionality	0.31 (0.14)	0.16	2.13	0.033
Attendance at center-based child care	0.35 (0.13)	0.19	2.65	0.008
Step 2					0.01
Quality of maternal care	0.03 (0.03)	0.07	0.97	0.332
Skin conditions
Step 1					0.09*
Duration breastfeeding	0.05 (0.02)	0.22	3.15	0.002
Maternal depression	0.05 (0.02)	0.18	2.59	0.009
Attendance at center-based child care	0.18 (0.13)	0.10	1.38	0.169
Step 2					0.04*
Quality of maternal care	−0.08 (0.03)	−0.19	−2.70	0.007
General illnesses and symptoms
Step 1					0.22*
Siblings	0.26 (0.07)	0.24	3.72	<0.001
Maternal depression	0.05 (0.02)	0.21	3.67	0.003
Child negative affectivity	0.20 (0.11)	0.13	1.79	0.074
Attendance at center-based child care	0.53 (0.10)	0.34	5.08	<0.001
Step 2					0.01
Quality of maternal care	0.04 (0.03)	0.10	1.52	0.129
Prescribed antibiotic use
Step 1					0.05*
Siblings	0.14 (0.07)	0.16	2.20	0.028
Attendance at center-based child care	0.20 (0.09)	0.16	2.20	0.028
Step 2					0.02†
Quality of maternal care	−0.04 (0.02)	−0.13	−1.73	0.083

Antibiotic use and digestive, skin, and general illnesses were square-root transformed to normalize regression residuals.

^*p < 0.05,

^†p < 0.10.

SE, standard error.

Follow-up Analyses

Given the robust role of center-based child care as a covariate, we conducted exploratory follow-up analyses to see whether maternal care interacted with center-based child care in the prediction of health outcomes. Additional follow-up analyses examined potential interactions with maternal and child characteristics (i.e., maternal depression, child negative emotionality). Finally, we tested the potential nonlinear associations between maternal care and health outcomes by adding a quadratic term (quality of maternal care²) to each model. All interaction terms and quadratic terms were nonsignificant (p values > 0.10) and thus were dropped from the final regression models to preserve model parsimony.

DISCUSSION

In the present study, we found preliminary evidence that quality of maternal care—defined as mothers’ observed sensitivity and cooperation with their 5-week-old infants during a mild stressor in the home—predicted lower rates of infant respiratory symptoms and skin conditions, assessed with structured health interviews administered 12 times over the first year of life. In addition, quality of maternal care was marginally associated with less prescribed antibiotic use. Importantly, even in this low-risk sample, we observed (1) wide variability in maternal sensitivity in the first weeks of life; (2) variability in low-level health symptoms, which have not been adequately examined previously but significantly impact quality of life for infants and families; and (3) significant links between sensitivity and these symptoms across the first year, which is earlier than has previously been examined. We found these links while controlling for important potential confounds such as breastfeeding, prenatal smoking, child temperament, maternal depression, center-based child care attendance, and number of siblings. The inclusion of these covariates, as well as our specific focus on a highly educated, low-risk, homogenous Dutch sample, allows us to rule out alternate explanations for these and similar findings; that is, links between parenting and health are not wholly attributable to poverty, poor nutrition, lack of parent education, race, access to health care, or severe forms of negative parenting.

This study is among the few to demonstrate the contribution of typical variation in parental sensitivity—and not just the presence or absence of adversity—to child health, consistent with emerging evidence that secure child attachment to parents is associated with reduced risk factors relevant to physical health.¹⁸ Moreover, whereas most previous studies have focused on links between parental behaviors and child physical health in middle childhood, adolescence, and adulthood, our findings suggest that similar patterns emerge as early as infancy—adding to the growing literature linking parenting and child health.^9–15

This study is the first to document systematic variation in infants’ respiratory symptoms and skin conditions as a function of observed maternal behavior in a community sample. These findings about respiratory symptoms and illness are consistent with previous research on the connections between family relationships and children’s asthma morbidity.⁵ The results are also consistent with a small body of literature linking parenting and skin conditions.^6,8

There are several possible explanations for these findings. One possibility is that less sensitive caregiving contributes to infant experiences of chronic stress—either directly or through a failure to buffer the infant from other environmental stressors. Chronic stress has been shown to lead to proinflammatory tendencies and the development of an immune system that is less effective at resolving exposure to pathogens.²⁸ As such, less sensitive caregiving may calibrate children’s immune systems in ways that increase risk for developing health problems.² Another possibility is that more sensitive and cooperative mothers are better attuned to their children’s cues of distress or discomfort, which may signal underlying health issues. For example, more sensitive mothers may be quicker than less sensitive mothers to notice infants’ distress caused by minor respiratory problems and to seek remedies before the symptoms become more severe and result in illness. It is also possible that other factors, such as infant genetics, temperament, or parental traits, account for the observed link by influencing both maternal sensitivity and infant health. Regarding temperament, it is important to note that infant negative emotionality was unrelated to maternal sensitivity in the present sample, and moderation analyses revealed that the observed link between maternal sensitivity and infant health did not differ by infant temperament.

It is unclear why maternal sensitivity was predictive of respiratory symptoms and skin conditions, but not digestive problems or general illness. Given the ease with which digestive illnesses are transmitted, the best predictor of these illnesses may be simply having greater exposure to other symptomatic children; high-quality maternal care may not be able to buffer children from direct exposure. In addition, because frequencies of general illnesses and digestive problems were somewhat low, statistical power may not have been sufficient to detect the effects of parenting.

The fact that higher-quality maternal care is even marginally associated with lower antibiotic use is notable, given that this study was carried out in a Dutch sample. In the Netherlands, consumption of antibiotics is, in general, low (<10 defined daily doses per 1000 inhabitants per day²⁹). We therefore encourage future studies to examine the link between maternal quality of care and antibiotic use in countries in which antibiotics consumption is higher. In addition, given our study design, we cannot evaluate whether infants of less sensitive mothers were actually sicker (and needed a prescription) or whether less sensitive mothers were more likely to seek antibiotics for comparable symptoms (perhaps because of a lack of confidence in their ability to manage their infants’ illnesses). Importantly, however, infant health problems were positively correlated with reports of antibiotic use; thus, it seems likely that the link between maternal sensitivity and antibiotic use is indirect, with lower maternal sensitivity predicting greater infant health problems, which then predict greater use of prescribed antibiotics; research with larger samples could test this hypothesized model. Future studies should also investigate whether the link between maternal sensitivity and antibiotic use reflects the differences in infant health or differences in health care utilization, because it is important to understand the factors predicting antibiotic use.

Although our study identified quality of maternal care as a predictor of infant health and antibiotic use, the effect sizes are small, and many additional factors are known to contribute to infant health. For example, future studies should consider the role of the family’s level of social support, exposure to community and domestic violence, and socioeconomic status. Of note, however, is that the observed effect sizes for maternal care are comparable with those of other well-recognized contributors to health, such as those observed in the present study for breastfeeding. Furthermore, the effects of maternal care were significant over and above these key covariates. Importantly, the study’s correlational design precludes causal interpretation of the results; thus, randomized controlled trials of parenting interventions that examine the influences on infant health will be important for examining causality. It may be that caregiving behaviors influence health in part by buffering against risk factors, such as genetic susceptibility to particular health problems. Some risk factors may be difficult to avoid, but interventions have been successful at improving parenting behaviors, and these interventions may lead to improvements in children’s physical health (see Reference 15 for the reductions in systemic inflammation associated with improvements in parenting after a family intervention). It is also possible that the small effects observed in infancy grow over time because of cumulative caregiving experiences across childhood. Future longitudinal studies that follow infants through childhood and adolescence will provide insight into developmental trajectories associated with less sensitive care in infancy.

The reliance on maternal interviews of infant health in the present study merits discussion. Because we aimed to capture common, subtle symptoms that typically do not warrant doctor visits, physician reports were not an ideal assessment of infant physical health. Furthermore, we note that in this sample, doctor visits during the first year were extremely rare, and thus, reliance exclusively on data from doctor visits would miss valuable information about daily health experiences that characterize early development. Instead, to best capture these daily experiences, we believed it more informative to have trained health researchers administer 12 monthly structured maternal interviews; we view this procedure as a methodological strength. Future research using the mothers’ reports may wish to use more frequent sampling to minimize the potential for recall problems (e.g., with online sampling or diary reports).

It is likely that mothers are relatively good reporters of their infants’ everyday health problems, being privy to low-level daily symptoms that may be overlooked in a single doctor visit; indeed, maternal interviews of infant health have been linked to infant intestinal microbiota.²⁴ At the same time, we recognize that personality traits may shape self-report of health symptoms in adults and mothers’ perceptions of their infants,³⁰ and mothers’ reports can be biased. Notably, however, our findings remained significant even when controlling for maternal depressive symptoms, which suggests that results were not simply a reflection of mothers’ negative biases because of low mood. Still, other dimensions of maternal personality (e.g., neuroticism) could bias reporting of infant health. On the other hand, it is unclear whether other approaches would avoid this limitation, because these same personality characteristics could similarly influence mothers’ tendency to seek medical visits and their reporting of symptoms to physicians. Moreover, mothers’ perceptions of their children meaningfully shape child development and thus are important to examine in themselves. Taking into account both the strengths and limitations of our approach, we anticipate that the present study will motivate future work to tease these explanations apart. Among the most important avenues for future research on this topic will be to use multimethod assessments of infant health—including maternal reports alongside reports from additional sources and biological specimens to measure specific aspects of immune function.

An additional consideration is that the present study examined 2 specific parenting dimensions of maternal behavior: sensitivity and cooperation. Extensions to other dimensions of maternal behavior and to multiple time points will be important to evaluate the extent to which the frequency, timing, and duration of specific practices shape child health across development. We also recommend that future studies extend observations to other caregivers, such as fathers and grandparents, who may play important roles in children’s daily lives.

Future research should also work to characterize the mechanisms through which parental care could shape infant physical health.³ Animal studies and emerging research in humans have begun to shed light on the mechanisms by which environmental stressors, including poor parenting, get “under the skin” to affect biological systems; such mechanisms include programming effects that lead to aggressive proinflammatory tendencies, neurobiological responses to stress, and differential methylation of genes involved in the stress and immune systems.^2–4,18 Emerging research suggests that some of these programming effects may be observed as early as infancy.¹⁸ Further examination is needed, however, of how these mechanisms operate in the first months of life and of the role of developmental timing (i.e., how parenting affects physical systems at different ages, whether mechanisms of influence change with development, and whether sensitive periods exist during which specific parenting behaviors have the greatest impact on specific health-related systems).

CONCLUSIONS

An important goal of pediatric research is to understand the parameters of parental care that contribute to childhood health and disease risk, and to do so as early in life as possible. Although the effects of early adversity have been a focus of considerable research, a crucial next step is to understand the potential contribution of typical variation in parenting within community samples to child health. Our preliminary findings suggest that variability in parenting behaviors in low-risk contexts are meaningfully linked to infant respiratory symptoms, skin conditions, and (at the trend level) antibiotic use over the first year. Because the immune system is still developing across infancy and early childhood, early parental programming effects could have long-term health implications. The findings reaffirm calls for clinicians to consider parenting as an opportunity for primary prevention while also taking into account the unique contexts and characteristics of families.³¹ (For practice recommendations in support of this perspective, see the American Academy of Pediatrics guidelines in Bright Futures.³²) We encourage medical professionals and researchers to form interdisciplinary partnerships (with, for instance, psychologists, molecular biologists, immunologists, epidemiologists, and pediatricians) to further explore these important questions about the role of caregiving for children’s physical health (e.g., Healthy Steps³³). Understanding the many environmental contributors to health in infancy—particularly parents—can enhance early prevention efforts to promote child health.