Infant Regulation and Child Mental Health Concerns: A Longitudinal Study

Maternal and child health nurses across 6 local government areas of Melbourne, Australia, consecutively approached all mothers who were attending the 8-month, free well-child appointment between September 2003 and April 2004 and invited them to take part in the Early Language in Victoria Study.¹⁰ The mothers of 1910 infants (82% of those approached) consented to take part. Data in the current study were drawn from questionnaires that were administered to mothers at child ages 8 to 10 months (wave 1: n = 1759), 12 months (wave 2: n = 1759), 5 years (wave 6: n = 1002), and 11 years (wave 10: n = 871). Infants with serious disabilities or developmental delays (eg, Down syndrome or cerebral palsy) were excluded. Ethical approval was obtained from the Royal Children’s Hospital Human Research Ethics Committee (27078 and 33195) and the La Trobe University Human Ethics Committee (03–32). All participants provided written, informed consent.

Data on child age, sex, birth weight, weeks’ gestation, whether the child was admitted to the special care nursery (SCN) or NICU, and birth order were gathered from the participating mothers. Participating mothers also reported their own ages; whether they were partnered, born in Australia, and of non–English-speaking background (NESB); and whether they completed secondary school. At child age 12 months, participating mothers also completed the Kessler 6 Psychological Distress Scale (K6), which is a measure of psychological distress with strong psychometric properties.¹¹ In keeping with previous studies, scores ≥8 indicated significant psychological distress.¹² The Socio-Economic Indexes for Areas (SEIFA) index of relative socioeconomic disadvantage was assigned on the basis of 106 unique home postal codes, which gave a broad indication of the relative disadvantage within the geographic area of each participant (higher scores indicated less disadvantage).¹³ 

Several measures relating to infant regulation that have been used in previous research¹⁴ were completed by participating mothers. Participating mothers indicated the presence and severity of their children’s sleep problems, excessive crying, temper tantrums, and mood swings by scoring each as either none (0), mild (1), moderate (2), or severe (3).^15,16 Global infant temperament was measured by using the following item: “Compared with other children, I think my child is: 1 = much easier than average, 2 = easier than average, 3 = average, 4 = more difficult than average, 5 = much more difficult than average.”^15,16 Participating mothers were asked whether their infants coughed and/or choked or gagged (yes or no) on pureed or smooth foods (eg, yogurt), mashed foods or solids (eg, potatoes), or lumpy foods (eg, peas). These items were scored with 1 point for each problem that was indicated, and this score was collapsed to form 1 variable that we used to indicate the number of feeding problems (0 for no feeding problem, 1 for 1 feeding problem, 2 for 2 feeding problems, and 3 for 3 feeding problems).

At child ages 5 and 11 years, participating mothers completed the parent-report version of the Strengths and Difficulties Questionnaire (SDQ), which is a well-validated, 25-item behavioral screen for children.¹⁷ A total difficulties score was generated (0–40), and scores were categorized within normal (<14), borderline (14–16), or clinical (17–40) ranges. The SDQ was shown to be a reliable and valid measure for identifying children who met the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria for requiring psychological treatment.¹⁸ A score in the clinical range indicates a child would likely receive a diagnosis of a mental disorder.

Latent class analysis was conducted to identify profiles of regulation at 12 months of age (Mplus version 8; Mplus Development Team, Los Angeles, CA). All 6 items in the Early Language in Victoria Study 12-month questionnaire (wave 2) pertaining to infant regulation (infant sleep problems, excessive crying, feeding problems, mood swings, temper tantrums, and global temperament) were used as indicators in the model. We aimed to identify the smallest number of profiles by starting with a parsimonious 1-class model and fitting successive models with increasing numbers of profiles. The likelihood-ratio statistic (L²), Bayesian information criterion (BIC), and Akaike information criterion (AIC) were examined across each model, with lower L², BIC, and AIC values indicating a better model fit. Entropy values were also examined (higher probability values indicated greater certainty in allocating participants to the profiles that were identified), and the Vuong-Lo-Mendell-Rubin likelihood ratio (VL) indicated the significance of differences in model fit between each model compared with the model with 1 less profile. Each infant was classified according to his or her most likely profile membership for subsequent analyses.

Linear and logistic regression models were used to compare the sociodemographic characteristics of infants in each identified profile with those of the infants in the most settled profile (the profile with few or no regulatory problems). Linear regression was conducted by comparing each of the identified profiles with the most settled (reference) profile on the SDQ total difficulties score at 5 and 11 years of age. Multinomial logistic regression was conducted to compare the proportion of children who met borderline and clinical categories on the SDQ total difficulties score between each profile and the settled profile, and estimates reflecting relative odds were calculated. Estimated associations between regulatory profiles and behavior scores and/or clinical categories were controlled for potential confounders (identified a priori through an examination of previous research), including maternal characteristics (age, marital status, country of birth, NESB, school completion, and socioeconomic disadvantage), child characteristics (age, sex, birth weight, weeks’ gestation, NICU or SCN admittance, and birth order), and maternal mental health (K6 score ≥8).

As is indicated in Table 1, our study sample comprised 1759 12-month-old infants who were of average birth weight overall (mean 3444.38 g; SD 530.49 g), with 50% being boys and 50% being firstborn infants. Participating mothers were on average ∼32 years old (SD 4.46), ∼97% were partnered, and 6% were NESB. Overall, the mothers in the sample lived in areas of slightly less socioeconomic disadvantage (mean 1037.80; SD 59.72) than the state average (mean 1000; SD = 100).

Latent class models for 1 to 6 classes were estimated (Table 2). The 5-class model had the indexes with the best fit (lower L², BIC, and AIC values), and the VL indicated a significant difference between the 4- and 5-class models, which suggests the latter was a more appropriate fit for the data. The 5-class model had an acceptable entropy value (0.74), which indicates high certainty in allocating participants to their most likely profile.

Profiles of infant regulatory behavior for each of the 5 profiles are demonstrated in Table 3. Names given to each profile are necessarily brief and indicate the predominant regulatory feature that was reported. The largest group of infants (36.8%) indicated few problems and as such was termed the settled profile. Infants in another group reported mostly mild, isolated temper tantrums (referred to as the tantrums profile; 21.3%), and a group characterized by mild-to-moderate sleep problems was termed as the sleep problem profile (25.4%). Infants in a moderately unsettled profile (13.2%) predominantly reported mild-to-moderate sleep problems, mild crying problems, mild-to-moderate tantrums, and mild-to-moderate mood swings. Infants in the smallest profile (3.4%) were identified as being severely unsettled because of greater rates of moderate-to-severe sleep problems, moderate crying problems, moderate-to-severe temper tantrums, more difficult temperament, and feeding problems. Of note, infants with severe tantrums tended to fall into the severely unsettled profile rather than the tantrums profile, and similarly, those with moderate-to-severe sleep problems tended to fall into the severely unsettled profile rather than the sleep problems profile. Those with more severe sleep or tantrum problems tended to have many other regulatory problems, and, if mild, tantrums and sleep problems only tended to occur in isolation.

As is indicated in Table 4, mothers with infants in the severely unsettled profile compared with those who had infants in the settled profile were slightly younger (mean age 31.27 years [SD 4.98] versus 32.39 years [SD 4.02]), less likely to be born in Australia (71.2% vs 84.8%), more likely to be NESB (13.6% vs 5.4%), less likely to have completed secondary school (63.2% vs 78.8%), more likely to live in an area of comparative socioeconomic disadvantage (although still greater than the state average; mean 1014.95 [SD 81.47] versus 1042.42 [SD 54.20]), and more likely to have significant mental distress (27.1% vs 5.0%). Infants who were moderately unsettled were more likely to be firstborn (57.4%) than the infants who were settled (44.4%). Mothers of infants in the sleep problems profile were more likely to be experiencing mental distress than mothers with infants in the settled profile (9.0% vs 5.0%). Mothers of infants in the tantrums profile were younger (mean maternal age 31.15 years [SD 4.55]) than those in the settled profile (mean maternal age 32.39 years [SD 4.02]), and infants in the tantrums profile were more likely to be firstborn (59.6%) than those in the settled profile (44.4%).

Adjusted regression analyses (Table 5) revealed that the 4 profiles of regulatory problems in infancy were associated with greater total behavioral difficulties at 5 years of age than the settled profile (tantrums: adjusted mean difference [AMD] 1.14 [95% confidence interval (CI): 0.35 to 1.93; P = .01]; sleep problems: AMD 0.87 [95% CI: 0.15 to 1.59; P = .02]; moderately unsettled: AMD 2.17 [95% CI: 1.20 to 3.14; P < .01]; severely unsettled: AMD 3.65 [95% CI: 1.87 to 5.42; P < .01]). At 11 years of age (Table 5), infants in the sleep problems profile, moderately unsettled profile, and severely unsettled profile were associated with greater total behavioral difficulties than those in the settled profile (sleep problems: AMD 1.27 [95% CI: 0.25 to 2.29; P = .02]; moderately unsettled: AMD 1.96 [95% CI: 0.57 to 3.35; P = .01]; severely unsettled: AMD 5.42 [95% CI: 3.10 to 7.74; P < .01]), but no differences were evident for infants in the tantrums profile.

In adjusted analyses (Table 6, Fig 1), children in the tantrums profile had higher odds of scoring in the clinical range (odds ratio [OR] 3.79; CI: 1.53 to 9.41; P < .01) for total difficulties at 5 years of age. Children in the severely unsettled profile had higher odds of scoring in the borderline range (OR 7.44; CI: 2.11 to 26.22; P < .01) and clinical range (OR 9.35; CI: 2.49 to 35.11; P < .01) for total difficulties at 5 years of age. At 11 years of age, children in the moderately unsettled profile had higher odds of scoring in the clinical range for total difficulties (OR 2.85; CI: 1.28 to 6.36; P < .01). Children in the severely unsettled profile had higher odds of scoring in the borderline range (OR 5.56; CI: 1.61 to 19.25; P < .01) and clinical range (OR 10.37; CI: 3.74 to 28.70; P < .01) for total difficulties at 11 years of age.

In a community cohort of 12-month-old infants and their mothers, an analysis of 6 brief items pertaining to multiple facets of infant regulation revealed 5 distinct profiles. In comparison with infants in the settled profile, those in each profile reported greater total behavioral difficulties at 5 and 11 years of age (except for the tantrums profile at 11 years of age). Broadly, as the complexity and severity of reported infant regulatory problems increased, so did the tendency to report mental health difficulties during childhood. At 5 years of age, infants from the severely unsettled profile in comparison with those in the settled profile staggeringly had 7.44 and 9.35 times higher odds of scoring in the borderline and clinical ranges for mental health concerns, respectively, even after accounting for potential confounding variables. Previous research suggests that those scoring in the clinical range would be likely to receive a formal diagnosis of a mental disorder.^17,18 The odds of scoring in the clinical range further increased for this profile to 10.37 at 11 years of age, which indicated that on average, symptoms for the severe profile worsened over time. Notably, infants in the moderately unsettled profile had 2.85 times greater odds of scoring in the clinical range for total difficulties at 11 years of age than those in the settled profile. Although infants in the tantrums profile had 3.79 times greater odds of scoring in the clinical range at 5 years of age, this was no longer apparent at 11 years of age, which indicates a shift toward recovery during the early school years. For infants in this profile, early tantrums and the associated poor outcomes in early childhood seem to be outgrown by middle childhood.

To our knowledge, the 5 profiles of infant regulation that we have identified and described here have not been identified in the scientific literature. Our results confirm that sleep and tantrum problems do seem to occur in isolation for some infants (25% and 21%, respectively), but these tend to be only mild in intensity and have few repercussions for mental health during early-to-middle childhood. However, for ∼16% of infants, regulation difficulties tend to be complex and involve multiple, overlapping symptoms of moderate-to-severe intensity. Examining a single component of regulation in isolation (such as sleep) may not provide enough detail to identify those at risk for later mental health difficulties. Although infants in each profile had increased total difficulties in comparison with those in the settled profile, only infants with multiple moderate-to-severe problems had greater odds of scoring in the clinical range for mental health problems at 11 years of age. That is, ∼30% of the infants in the severely unsettled profile had mental health concerns in the borderline or clinical ranges at 5 years of age, and this increased to >46% by 11 years of age.

Infants from the moderately unsettled profile and tantrums profile were more likely to be firstborns, and mothers of infants in the severely unsettled profile, moderately unsettled profile, and tantrums profile were slightly younger than those who had infants in the settled profile. Mothers of firstborn infants and younger mothers may report more problematic infant behavior because of unfamiliarity with normal infant behavior. Previous research indicates young mothers may engage in fewer positive parenting practices and report more difficult child behavior.¹⁹ Mothers of infants in the severely unsettled profile compared with those who had infants in the settled profile were less likely to be born in Australia, more likely to be NESB, less likely to have completed secondary school, and more likely to live in areas of comparative socioeconomic disadvantage (although these areas were less disadvantaged than the state average). Mothers of infants in the moderately unsettled profile were also more likely to report being NESB and to live in areas of comparatively greater socioeconomic disadvantage than those with infants in the settled profile. These mothers may have less social and familial support and face language barriers when seeking help for their infants who are unsettled.²⁰ 

Approximately 5% of mothers of infants in the settled profile reported clinically significant psychological distress; however, this number was significantly greater in the sleep problems profile (∼9%), moderately unsettled profile (12.5%), and severe profile (∼27%). Previous research has repeatedly revealed an association between infant sleep problems and maternal depression and distress.^21,–25 Our findings indicate that multiple moderate-to-severe infant regulatory problems are associated with greatly increased rates of maternal distress, with approximately one-third of mothers with infants who are severely unsettled reporting distress in the clinical range. Interventions that are targeted at the mother-infant dyad rather than the infant regulatory problem in isolation may be the most appropriate for these families.

The infant regulation items that we administered in the current study are not exhaustive and may not accurately encompass all aspects of infant regulation. Children did not complete diagnostic interviews for mental disorders. Replication of this work in other longitudinal cohorts is required. This study has several strengths, including a prospective design and large community sample. We used a data-driven approach that allowed for a more comprehensive examination of the overlap and severity of regulation problems than has been conducted previously.

Infants with multiple moderate-to-severe regulatory problems are at greatly increased risk of later mental health concerns and as such may benefit from targeted intervention and referral to support services. Infants with 1 or 2 mild regulatory problems should be closely monitored. Providers inquiring about the presence and severity of multiple regulatory problems may identify infants with a high burden of comorbidity that extends into childhood. The presence of multiple moderate-to-severe infant regulatory problems may indicate the beginning of a trajectory toward poor child mental health. Subsequent research should investigate whether early intervention can improve outcomes for these children and potentially curtail the rising trend of emergency department attendance for child mental health problems.^26,27 

Infants with multiple moderate-to-severe regulatory problems experience >10 times the odds of clinically significant mental health concerns during childhood, and these symptoms appear to worsen over time. Six brief items pertaining to the presence and severity of regulatory problems may be used to identify infants who require more in-depth assessment. When parents report concern about a regulatory problem in their 12-month-old infant, clinicians must inquire about the extent, complexity, and severity of other regulatory problems to identify those in most urgent need of intervention and support.

AIC

Akaike information criterion

AMD

adjusted mean difference

BIC

Bayesian information criterion

CI

confidence interval

K6

Kessler 6 Psychological Distress Scale

L²

likelihood-ratio statistic

NESB

non–English-speaking background

OR

odds ratio

SCN

special care nursery

SDQ

Strengths and Difficulties Questionnaire

SEIFA

Socio-Economic Indexes for Areas

VL

Vuong-Lo-Mendell-Rubin likelihood ratio