Beyond the First Feeling: What 73,518 Families Reveal About Support and Child Development

At a municipal counter where a new Maternal and Child Health Handbook is issued, or in an obstetric waiting room where a first questionnaire is unfolded, there may be a short question: “How did you feel when you found out you were pregnant?” A single answer cannot contain a life. Yet plans, work, housing, money, relationships, previous loss and fears about health can all fold into that one line.

Some people answer “very happy” without hesitation. Others were surprised but happy, surprised and confused, troubled, or felt nothing in particular. Still others cannot find themselves in any offered category and select “other.” None of those words is a verdict on what kind of parent a person will become. Feelings arise inside circumstances, and circumstances can change.

A team led by Professor Mitsue Nagamine of Science Tokyo’s Institute for Liberal Arts, with Professor Kenta Matsumura of Aomori University of Health and Welfare, Junko Sakai, Akiko Tsuchida and Professor Emeritus Hidekuni Inadera of the University of Toyama, and the JECS Group, analyzed 73,518 mother–child pairs in the Japan Environment and Children’s Study. They placed three observations on one timeline: feelings recorded in early pregnancy, social support reported 2½ years after birth and developmental screening at age 3.

The paper appeared online in the Journal of Health Psychology on June 28, 2026, and Science Tokyo announced it on July 22. Compared with “very happy,” every other response was associated with more suspected areas of developmental delay at age 3. Later social support statistically mediated part of each association. The direction for public health is therefore not toward blaming a first feeling. It is toward widening the circle around the family afterward.

A study connecting three moments

A major strength is that this was not a cross-sectional survey asking about feeling, support and development all at once. The pregnancy reaction was reported in a questionnaire distributed around the first trimester. When the child was 2 years 6 months old, the mother answered three questions about affection, help with consultation and decisions, and contact with someone she could trust. At age 3, a parent completed the Japanese ASQ-3.

That order supplies the temporal sequence required by mediation analysis: an earlier exposure, an intermediate candidate and a later outcome. Sequence alone, however, does not establish causation. Finances, partner relationships, maternal mental and physical health, a child’s health and temperament, and changes across the intervening years can influence both perceived support and developmental reporting.

Time	What was measured	Role in the analysis
Around the first trimester	Feeling when pregnancy was discovered, six response categories	Initial exposure or predictor
Child age 2 years 6 months	Affection, decisional support and trusted contact, three items	Candidate social-support mediator
Child age 3	ASQ-3 communication, gross motor, fine motor, problem-solving and personal–social domains	Parent-reported developmental screen

Six answers, and the lives inside them

Of the 73,518 mothers, 49,395—67.2%—answered “very happy.” Another 17,899, or 24.3%, said the pregnancy was unexpected but they were happy. A further 4,704, or 6.4%, were surprised and confused. Those three groups together comprised 97.9% of the sample.

The smaller categories were “troubled,” selected by 358 mothers, or 0.5%; “felt nothing in particular,” selected by 344, also 0.5%; and “other,” selected by 818, or 1.1%. Science Tokyo reported that mothers outside the “very happy” reference group tended to have lower social support.

The distribution shows why pregnancy “intention” and pregnancy “feeling” are not interchangeable. Roughly one quarter of the sample had not expected the pregnancy but was happy. A planned pregnancy can still generate fear after a medical result, a job loss or a financial shock. An administrative planned–unplanned binary cannot capture welcome, ambivalence or acceptance. The study’s single item asked about the subjective moment of discovery.

A first reaction is not an entrance examination for parenthood. It can be the opening of a conversation about whether support is reaching the family.

Why 1.68 is not destiny

The strongest published association was for mothers who said they “felt nothing in particular.” Compared with the “very happy” group, their adjusted odds ratio for having a greater number of ASQ-3 domains below cutoff at age 3 was 1.68, with a 95% confidence interval of 1.41 to 2.00.

It would be wrong to translate 1.68 as “a 68% probability of delay,” or automatically as “68% greater risk.” Odds divide the probability that an event occurs by the probability that it does not. Odds ratios approximate risk ratios when an outcome is rare, but diverge as it becomes more common. The public university release does not provide group-specific absolute rates, so the estimate cannot responsibly be turned into a number of additional children.

The category also contained only 344 mothers, 0.5% of the analyzed sample. The entire cohort is vast; the group producing the most prominent comparison is not. A confidence interval that excludes 1 supports a statistical association. It does not make one response a precise individual diagnostic test.

“Nothing in particular” can carry many meanings: emotion that has not yet become language, shock, exhaustion, a cultural reluctance to display feeling, poor fit with the options, or genuinely neutral affect. The study did not separate those possibilities. It should not be collapsed into indifference, weak attachment or poor parenting.

Mediation is not a magic shield

The operative phrase in the paper’s title is “mediating effect.” Mediation analysis asks through which path an association may travel. Here, a complicated initial reaction was associated with lower later support, which in turn was associated with developmental screening. The model divided this indirect path from the remaining direct and unmeasured paths.

According to Science Tokyo, the proportion mediated by support was 45.0% in the “other” group, 43.0% among those who were troubled, 27.8% among those who felt nothing in particular, 22.8% among those surprised but happy, and 22.2% among those surprised and confused. Support did not explain everything. Even the largest estimate was under half.

Mediation is not identical to testing an interaction—whether the pregnancy–development association itself becomes statistically smaller at high versus low support. Nor was this a randomized trial in which researchers delivered a support program and observed whether development improved. One cannot conclude that adding support would prevent exactly 22% to 45% of suspected delays.

Still, mediation has practical value. A first reaction is already in the past; present support may be alterable. Even without causal certainty, the analysis identifies a plausible point for services and a concrete hypothesis for intervention trials.

Measuring a circle of support in three questions

The age-2½ social-support measure was not a count of relatives, friends or phone contacts. It asked whether someone showed the mother affection or goodwill, whether someone helped with consultation and decision-making, and whether she maintained contact with someone she could trust. Scores ran from 3 to 15, with 9 or lower designated low support.

This mainly captures emotional and informational support. It does not fully count instrumental help such as cooking, child care, transport, money, housing or a night of uninterrupted sleep. A large network can still feel unusable. One dependable person can sometimes produce a stronger sense of support than many nominal connections.

Timing matters. The mediator in this study was not support during pregnancy. It was support when the child was 2½. A headline claiming that “support the pregnant woman and the problem is solved” changes the design. The data instead suggest that the first experience of pregnancy and the social relationships surrounding a family nearly three years later may belong to one trajectory.

A separate JECS analysis published in 2023 found the largest adverse estimate for maternal mental health at 2½ years among mothers who had lost close friends or neighbors only by that later time, followed by loss of emotional support. Support should not be treated as a short service that expires after birth. It changes through toddlerhood.

The ASQ-3’s five windows

The Ages and Stages Questionnaires, Third Edition, is a parent-completed developmental screener covering ages 1 month through 5½ years. At age 3, it observes five domains: communication, gross motor, fine motor, problem-solving and personal–social development.

Each domain has six questions scored “yes” at 10, “sometimes” at 5 and “not yet” at 0, creating a score from 0 to 60. The activities are visible in daily life: understanding and using words, jumping or kicking, manipulating objects with fingers, working out sequences, dressing and interacting in play. The Japanese translation has undergone psychometric evaluation and uses age-specific cutoffs.

ASQ-3 is not a diagnosis. A score below cutoff is a signal to consider monitoring or further assessment. Ordinary variation, temporary illness, the respondent’s interpretation, opportunity to observe a behavior, prematurity, culture and experience can all affect a score. In this study, the same broad reporting source supplied the mother’s perceived support and the child’s developmental screen, leaving room for common-method bias.

The researchers counted how many of the five domains were below cutoff. They did not turn any one domain into a named neurodevelopmental condition. The result therefore means “more areas of concern on a broad screen at age 3,” not “this reaction caused a specific disorder.”

JECS: following 100,000 families

The infrastructure behind the analysis is the Japan Environment and Children’s Study, launched by the Ministry of the Environment in fiscal 2010. Pregnant women were recruited from January 2011 through March 2014 in 15 regions stretching from Hokkaido to Okinawa. The National Institute for Environmental Studies serves as the Core Centre and the National Center for Child Health and Development as the Medical Support Centre, working with university-based Regional Centres.

The original target was 100,000 mother–child pairs. Biological samples including maternal blood and urine, cord blood and breast milk were combined with medical records, household measurements and questionnaires, with children followed to age 13. The central purpose was to investigate chemical and other environmental influences on health. Yet “environment” is not limited to molecules. Family relationships, income, education, work, stress and social connection are also contexts in which development unfolds.

The baseline cohort included 100,778 pregnancies resulting in delivery, 95,248 unique mothers and 100,148 live births, making JECS one of the world’s largest birth cohorts. The 73,518 pairs in the new analysis are not the entire cohort. Miscarriage or stillbirth, loss to follow-up, unanswered questionnaires and missing items reduce the available sample by the third birthday.

That attrition is not clerical trivia. Families facing isolation, poverty, language barriers, migration or illness may be less able to sustain years of questionnaires. If those needing the greatest support are more likely to disappear from the analysis, estimates can shift. Size is a major strength, but no sample becomes immune to selection bias merely by becoming enormous.

From a 1942 handbook to a continuing offer of help

Japan’s system for connecting a pregnancy to health and municipal services has an 84-year history. The wartime Maternal Health Record Book of 1942 recorded health and also helped establish priority for scarce rice, sugar, cotton and other necessities. In 1948, maternal and infant records were integrated into the Mother and Child Handbook. After the 1965 Maternal and Child Health Act, the modern Maternal and Child Health Handbook took shape in 1966.

The book places prenatal care, delivery, health examinations, vaccination and growth on one timeline and is issued by municipalities after notification of pregnancy. As infectious disease and malnutrition receded, revisions expanded toward development, mental health, partner participation, family diversity and digital use.

The question examined in the new study is practical precisely because it is often asked during the interview for handbook issuance or prenatal care. It requires no laboratory or specialist instrument. That simplicity creates both power and danger. Used as an invitation, it can open support. Used as a test of the “correct” emotion, it encourages concealment and may drive the people most in need away.

The handbook’s origin in a system that connected a record to tangible rations is instructive. Documentation alone does not protect a family. A question becomes useful when food, health care, time, listening, child care and financial resources can follow the answer.

Fifty years of the stress-buffering idea

A landmark year for social-support science was 1976. American physician Sidney Cobb described support as information leading people to believe that they are cared for, valued and part of a network of mutual obligation. He argued that this social information could moderate the health effects of major life stress.

In 1985, Sheldon Cohen and Thomas Wills organized a large literature around two models. The “main effect” model proposes that social integration can benefit health broadly, whether stress is high or low. The “buffering” model proposes that appropriately matched support is especially protective when stress is severe.

The distinction matters in pregnancy. The number of relatives or neighbors is not the same as being heard when advice is needed, having someone share a decision or being able to sleep while another person takes over. Well-intended advice that conflicts with a mother’s needs may not function as support; criticism or control may deepen stress. Network size, support received and support perceived as available are related but different.

The new study’s three items translate that theoretical history into ordinary language: affection, consultation and trusted contact. A concept debated for half a century becomes three questions posed to a mother raising a 2½-year-old.

Several possible roads from support to development

How might support for a mother be associated with a child’s development? One route is maternal mental health. A relationship that listens, helps solve problems and reduces isolation can change the appraisal of stress and may lessen depression or anxiety. A 2022 JECS study of 88,711 mothers also modeled associations between support during pregnancy and perinatal or postpartum depressive states.

A second route is time and physical capacity. When transport, meals, housework, night care or care of siblings can be shared, a parent can rest, attend appointments and spend more time talking, reading, playing and moving with a child. Those daily interactions are practice grounds for the language, movement, problem-solving and social behaviors observed by ASQ-3.

A third route is connection to services. A trusted person may notice a small concern, then help a family contact a public-health nurse, pediatrician, developmental service or nursery. Early consultation does not guarantee a higher score, but it prevents the family from carrying uncertainty alone.

A fourth possibility is a common social cause. Stable income, safe housing, flexible work, nearby child care and relationships free of violence can make support easier to obtain while also influencing health directly. What appears to be “the effect of support” may partly reflect social and economic resources that were measured imperfectly.

Evidence accumulated within the same cohort

The new finding did not appear in isolation. A 2021 JECS analysis followed 24,324 women who had psychological distress in the first trimester. Negative feelings about pregnancy and later mother–infant bonding were associated with persistent psychological distress through 12 months postpartum. The first reaction was linked to the mother’s own longer psychological trajectory.

In 2024, an Osaka University-led team analyzed 68,442 JECS pairs and reported a dose-response association: higher social support during pregnancy was linked to lower odds of falling below cutoff in all five ASQ-3 domains at age 3. Adjusted odds ratios comparing the highest and lowest support levels ranged from 0.49 to 0.58 across domains. That study treated prenatal support as the main exposure; the 2026 study treats support at age 2½ as a mediator. The questions complement but do not duplicate one another.

Another 2026 JECS paper examined bonding difficulties among mothers without prior postnatal depression. Negative pregnancy feelings, low support and difficulty holding a distressed infant all showed strong associations. None of these observational studies proves causality, but together they thicken a picture in which pregnancy reaction, maternal mental health, social support, bonding and child development are connected parts of a process rather than isolated boxes.

The power of a cohort lies less in one dramatic number than in repeated analyses illuminating the same life course. But studies using the same participants and related questionnaires are not independent replications. Other countries, different Japanese populations, clinician-assessed outcomes and intervention trials are still necessary.

What even a vast study cannot prove

First, the feeling about pregnancy was measured with one item. The study cannot recover its intensity, duration, change over time, cause or the partner’s response. Six boxes make analysis possible; they do not exhaust human experience.

Second, ASQ-3 was parent-reported, not a clinician’s diagnostic examination. Depression, anxiety, expectations and cultural norms can shape how behavior is rated. If a mother’s psychological state affects both perceived support and developmental reporting, the association may be strengthened by shared reporting.

Third, support was measured at 2½ and development only six months later. If a child already had language, movement or behavioral differences, the parent might have found it harder to socialize—or received more assistance from relatives and professionals. Reverse causation remains possible.

Fourth, the data are observational. Statistical adjustment cannot fully remove unmeasured relationship quality, violence, chronic disease, local service availability or work flexibility. A mediated proportion is an estimate conditional on the model’s assumptions.

Fifth, JECS participants were recruited mainly through prenatal visits and municipal handbook offices, and had to be able to complete Japanese-language self-administered questionnaires. The findings cannot automatically be generalized to people who miss prenatal care, require language support or were lost during follow-up. Nothing in this research justifies using a feeling to judge parental fitness, trigger punishment or intensify surveillance.

Make the question a door, not a scorecard

The first rule for applying this research is not to pathologize surprise, confusion, trouble or emotional neutrality. Pregnancy can transform work, education, income, housing and relationships at once. Ambivalence is not rare, and joy and fear can coexist. An interview that teaches respondents the “right” answer destroys the very information it seeks.

The second rule is to ask again. Giving a phone number to someone who felt troubled in pregnancy does not ensure support at age 2½. Prenatal appointments, birth, the one-month examination, infant health checks, entry into child care and return to work are each points to review whether the family has someone to call, time to rest, safety, food, housing, income and care.

The third is not to make the mother both the sole recipient and sole producer of support. Partners, families, employers, municipalities, medicine and child care must share responsibility. Telling an isolated person simply to “ask for help” misses the central fact that a usable helper may not exist. Support is not only a personality trait; it is social infrastructure.

The small handbook of 1942 began as a device that made pregnancy visible and connected records to material resources. The 100,000-family cohort launched in 2011 used some of those same municipal entry points to turn the years from pregnancy through childhood into longitudinal evidence. In 2026, the evidence returns a question to the counter: if we record a person’s first feeling, can that record become a promise not to leave the family alone?

The work of supporting child development is not handed exclusively to a mother when a line appears on a pregnancy test. A broader circle cannot erase the first shock, but it can prevent that shock from being carried in isolation. That is the most humane reading of this study—and the one most faithful to what the data can actually say.

Sources and references

This article is based primarily on Science Tokyo’s July 22, 2026 release and the bibliographic record for the Journal of Health Psychology paper, supplemented by official JECS materials and peer-reviewed work on cohort design, Japanese ASQ-3 validation, pregnancy feelings, social support and the history of Japan’s Maternal and Child Health Handbook. It does not turn observational mediation into a proven intervention effect or individual diagnosis.

• Science Tokyo, feelings when pregnancy was discovered, social support and child development (July 22, 2026)
• Nagamine et al., The mediating effect of social support on negative feelings about pregnancy and child developmental delays, Journal of Health Psychology (2026)
• Ministry of the Environment, Japan Environment and Children’s Study—Study Overview
• National Institute for Environmental Studies, JECS Programme Office
• Kawamoto et al., Rationale and study design of JECS, BMC Public Health (2014)
• Michikawa et al., Baseline profile of JECS participants, Journal of Epidemiology (2018)
• Mezawa et al., Psychometric profile of the Japanese ASQ-3, Pediatrics International (2019)
• Imanishi et al., Maternal social support during pregnancy and development at age 3, Environmental Health and Preventive Medicine (2024)
• Matsumura et al., Social support and perinatal/postpartum depressive state in JECS, Journal of Affective Disorders (2022)
• Matsumura et al., Time-varying support and maternal health at 2.5 years postpartum, Journal of Epidemiology (2023)
• Tokuda et al., Feelings about pregnancy, bonding and persistent distress, Journal of Psychiatric Research (2021)
• Delgado-Ron & Janus, Pregnancy planning or intention and early child development, PLOS Global Public Health (2023)
• Cobb, Social support as a moderator of life stress, Psychosomatic Medicine (1976)
• Cohen & Wills, Stress, social support, and the buffering hypothesis, Psychological Bulletin (1985)
• Suzuki, History and revision of Japan’s Maternal and Child Health Handbook (2024)