Measuring state-level infant and toddler well-being in the United States: Gaps in data lead to gaps in understanding

Children who are nurtured, protected, and supported in the first years of life tend to have better individual outcomes and are more likely to grow to become healthy, productive adults. Child well-being varies across states, yet the field lacks a comprehensive review of infant and toddler indicators measured at the state-level. This paper reviews indicators of well-being from the prenatal period to three years that meet certain a priori criteria. Most of the child-level indicators identified were in the physical health domain; relatively fewer indicators were found in the early cognition and language or social-emotional-behavioral domains. While some states are making progress toward developing integrated early childhood data systems, more work is needed to provide robust data on infant and toddler development. These results highlight the need to develop a broader range of indicators of infant and toddler well-being and improve measurement sources to better inform policies and programs advancing population health.

. Yet not all children have an equitable start in life.Racial and ethnic disparities in well-being emerge even before birth (Wilkinson et al., 2021), and inequities by additional sociodemographic factors (e.g., family income, home language, and maternal education) across domains of development become evident at nine months and grow larger as infants become toddlers (Halle et al., 2009).These early years are crucial to healthy development due to the rapid changes in brain connectivity and skill acquisition that occur during this time (Masten & Cicchetti, 2010;Shonkoff & Phillips, 2000).Accordingly, the first three years have become a central focus of policies and programs designed to improve overall child well-being (Prenatal-to-3 Policy Impact Center, 2021).
Child well-being is globally defined as "the multi-dimensional nature of health that is enhanced when physical, cognitive, and social-emotional-spiritual development is nurtured in developmentally appropriate ways" (The Alliance for Child Protection in Humanitarian Action, 2021, p. 11).The well-being of children from the prenatal period to three years, henceforth referred to as "infant and toddler well-being," varies across states (Keating et al., 2020).For instance, some states have higher infant mortality rates than others (CDC, 2021), which may lead to variations in policy and practice interventions.It is therefore important to develop indicators available for all states and the District of Columbia (DC) to help monitor how well communities fare compared to others on an outcome of interest at a single point in time (Szilagyi & Schor, 1998;Moore & Brown, 2003).Additionally, state-level indicators can be used to monitor trends in infant and toddler outcomes and enable researchers to assess associations between individual factors, policies and programs, and overall health and development.
Although associations with indicators cannot establish causality, they can inform whether policy interventions, such as expanding economic supports for disadvantaged families, are associated with better outcomes.Building on this understanding, indicators can help states set goals or targets for policies and programs, invest in advantageous policies and programs, and respond to the needs of families, educators, policymakers, and public health officials.Moreover, indicators can help states identify sub-populations of children who may be at risk for adverse outcomes, such as children and families who have faced racial discrimination and children of families with lower income (Keating et al., 2020;Wilkinson et al., 2021).
Efforts to identify indicators of child well-being have been undertaken in the United States since the 1970s (Lippman, 2007;Moore, 2020).Most indicator reports have tended to focus on a small set of negatively oriented indicators for preschool and school-aged children measured by well-established data collection systems (Moore, 2020;Moore et al., 2004).A major gap in the literature is a comprehensive review of both positive and negative indicators of infant and toddler well-being available for every state and DC that can inform policies and programs.The goal of this paper is, therefore, to review indicators of infant and toddler well-being that are publicly available, measured across states, representative of state populations, and measured over time.This review builds upon extant work by providing a conceptual framework, identifying a priori inclusion criteria for indicators and measurement sources, identifying gaps in measurement, and informing future efforts related to childhood indicator research.

Methods
We took a four-step approach to review the state of measurement of infant and toddler well-being in the United States.We (1) developed criteria for identifying relevant indicators, (2) compiled a list of potential indicators using those criteria, (3) reviewed measurement sources for each potential indicator, and (4) solicited expert input on the list of indicators and their measurement sources.Each of these steps is described in more detail below.

Criteria for identifying potential indicators
Before identifying indicators of infant and toddler well-being, we first developed criteria for indicators to include in the review.Substantively, we were interested in measures of infant and toddler well-being that are relevant for children prenatally to age three, have a demonstrated connection to long-term outcomes, and focus on the child or their family or caregivers.We did not include more distal contextual indicators, such as those about the neighborhood or policy environments.

List of potential indicators
We next compiled a list of potential indicators of infant and toddler well-being that met the substantive criteria outlined above.Specifically, we conducted a focused review of academic literature in early child development as well as existing indicator tracking work from non-profit research organizations that have published data-driven advocacy or policy work, including the Zero To Three State of Babies Year-book (Keating et al., 2020), the University of Texas at Austin Prenatal-to-3 Policy Impact Center's Roadmap (Prenatal-to-3 Policy Impact Center, 2020), the National Center for Children in Poverty's Improving the Odds for Young Children State Early Childhood Profiles (National Center for Children in Poverty, 2018), and the Child Trends DataBank (Child Trends, 2020).The compiled list of all potential indicators based on this review is available in Appendix A.

Measurement sources for potential indicators
Then, we looked for sources of measurement for these potential indicators.Specifically, we looked for sources that are publicly available, comparable for all 50 states and DC, representative of state populations, and measured at regular time intervals (e.g., annually).The sources that met these criteria and were included in this review are outlined in Table 1.

Expert input on indicators and measurement sources
We convened two meetings of experts in August 2020 to solicit feedback on our process and preliminary findings.Experts were identified by the authors and invited to attend one of two meetings.The first group was comprised of experts on early childhood administrative data, and the second group was comprised of experts on early childhood development. 1 Both groups provided input on relevant child and family/caregiver indicators as well as data quality and coverage.The experts agreed that our review had not missed any substantive indicators or major data sources, and their feedback is incorporated throughout this manuscript.

Results
Following the framework outlined in Fig. 1, we present the identified indicators of infant and toddler well-being that have measurement sources meeting our criteria categorized into three broad domains: physical health, early cognition and language, and social-emotionalbehavioral development.Within each domain, we describe two types of indicators: those at the child-level and those at the family/caregiver level.Some of these indicators apply to multiple domains and have multiple measurement sources.See Appendices B-D for a complete listing of all indicators by developmental domain.This information is current as of Spring 2021.

Early physical health indicators
Many child health indicators are measured at birth, including birthweight (used to identify births with low birthweight); gestational age (used to identify preterm births); some health conditions of the newborn; some treatment received at birth (e.g., neonatal intensive care unit [NICU] admission, antibiotics); and infant mortality (Table 2).Data on these indicators are collected from birth and death certificates and are available through the CDC's National Vital Statistics System (NVSS).From birth to 3 years, child indicators can be constructed using data on immunizations, physical health, oral health, sleep habits, and special health care needs using data from CDC's National Immunization Survey (NIS) and the National Survey of Children's Health (NSCH) (see Appendix B).
Family/caregiver indicators of infant and toddler physical health include whether the child is covered by health insurance (CPS), the adequacy and consistency of that insurance (NSCH), the food security of their household (CPS), whether the child received preventative care visits (NSCH), and whether the child has access to a medical home (NSCH).Information is also available on a series of indicators related to the mother's health, including receipt of timely prenatal care and maternal health during pregnancy (NVSS) and data on breastfeeding initiation and duration (NIS).Finally, rates of child maltreatment, which may influence physical health other child outcomes, are available through the National Child Abuse and Neglect Data System (NCANDS).

Early cognition and language indicators
In the early cognition and language domain, data are available from the NSCH about diagnoses of developmental delays (Table 3).Data are also available from the NSCH for certain cognitive/language developmental milestones (e.g., is the child able to use two words together).Parents of children of all ages are asked about developmental delays whereas the latter set of items are asked about children who are at least 1 year old.
Family/caregiver indicators of infant and toddler well-being related to the early cognition and language domain include NSCH data on caregiver interactions with the child (e.g., reading to children singing and telling stories to children) and whether a health care provider asked about concerns related to the child's learning or development.

Early social-emotional-behavioral indicators
The only nationally available child-level indicator of infant and toddler social-emotionalbehavioral development is a battery of four items about flourishing designed to capture "curiosity and discovery about learning, resilience, attachment with parent(s), and contentment with life" (Table 4).These items are from the NSCH and asked of parents of children 0-5 years (Child and Adolescent Health Measurement Initiative, 2018, p. 55).Information on family/caregiver indicators that may influence social-emotional-behavioral development-such as exposure to adverse childhood experiences (ACEs), parental aggravation, maternal mental health, family resilience, and eating meals together-are also available through the NSCH.Additionally, Vital Statistics indicates whether a father is listed on birth certificates.

Discussion
In this paper, we sought to identify indicators of infant and toddler well-being from the prenatal period to three years that are publicly available, comparable for all 50 states and DC, representative of state populations, and measured at regular time intervals.Most of the child-level indicators we identified were in the physical health domain.Relatively fewer child-level indicators were found in the early cognition and language and socialemotional-behavioral domains.These results highlight the need to develop a broader range of indicators of infant and toddler well-being and improve measurement sources to better inform policies and programs advancing population health.

Strengths in existing infant and toddler indicators and measurement sources
The indicators of infant and toddler well-being included in this review focus primarily on infant and toddler physical health.Indicators of physical health are crucial, as poor infant health can have lifelong consequences (Black et al., 2017).For example, research from Canada has found that health at birth is associated with infant mortality in the short term and educational success and reliance on social programs in the long term (Oreopoulos et al., 2008).
NVSS is the federal compilation of birth and death certificates from the 50 states, five U.S. territories, New York City, and the District of Columbia.A wealth of information is collected from birth certificates, including birthweight and length of gestation, parental characteristics, and health of the newborn.The NVSS is the only source of national population-level birth data on newborns, and it is large enough to facilitate subgroup comparisons (e.g., by race/ethnicity) within states.However, the reliability and validity of birth certificate data vary widely by individual item and those on maternal tobacco and alcohol use, prenatal care, pregnancy complications, delivery events, and congenital malformations have been identified as items that need quality improvement (Northam & Knapp, 2006;Salemi et al., 2017).Additionally, analysts need to take into consideration changes in birth certificate designs over time when making historical comparisons.
The NIS and NSCH are surveys focused specifically on the health of children.The NIS-Child survey component is focused on immunization and breastfeeding histories among toddlers aged 19-35 months.The NSCH is more general-taking a whole child perspective -and captures health settings, behaviors, and outcomes for children ages 0-17 years.It is the primary source of information on the well-being of infants and toddlers beyond immunizations, breastfeeding, and birth experiences.Both the NSCH and NIS are annual data collections with data that can be compared across states.Together, they cover many well-being indicators relevant to infants and toddlers.

Challenges with existing infant and toddler indicators and measurements sources
Indicators of infant and toddler early cognition and language and social-emotionalbehavioral development are more limited than those of physical health.Information needed to measure early cognition and language and social-emotional-behavioral development is frequently collected in clinical settings using child observations or parent questionnaires that capture rapid and interconnected changes in the early years.These clinical assessments are not easily scalable to population-level measurement due to the time and resource commitments needed for such data collection methods (Brito et al., 2019;Paschall et al., 2020).Nonetheless, NSCH has included questions on a child's understanding and use of language and measures of flourishing for children 0-5 years.Ideally, additional measures of early cognition and language and social-emotional-behavioral skills appropriate to children from birth to three years will become more readily available in future years.
Many of the available indicators reflect the absence of positive development (e.g., infant mortality rates, reported cases of child maltreatment) rather than the presence of positive development (e.g., normative language development, demonstrating curiosity) or the presence of supportive contexts for young children's development (e.g., positive parent-child or caregiver-child relationships).This dearth of "positive" indicators of child well-being has been an ongoing gap in the field for decades (Halle & Moore, 1998;Lippman, 2007;Lippman et al., 2014;Moore & Halle, 2001).
Even promising data sources have limitations.Sample sizes severely limit analysts' ability to disaggregate data to examine disparities by race/ethnicity and income at the state level with the NIS and NSCH in a timely manner.For example, the CDC combines multiple years of NIS data to produce its state-level cohort estimates of breastfeeding rates, without looking at subgroups (Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Division of Nutrition, Physical Activity, and Obesity, 2021).Additionally, the public-use version of the NIS contains only four racial/ethnic group identifiers: Hispanic, non-Hispanic Black, non-Hispanic other, and non-Hispanic White, limiting analysts' ability to examine smaller racial/ethnic subgroups (Centers for Disease Control and Prevention, National Center for Immunization and Respiratory Diseases, 2020).The last time the Current Population Survey produced estimates of food insecurity in households with children at the state level, it combined data from 2003 to 2011 to do so (Coleman-Jensen et al., 2013).With the NSCH, subgroup analyses are limited by sample size that is compounded by a survey redesign.The NSCH survey administration methodology was redesigned in 2016, and data before and after the redesign cannot be harmonized, limiting the number of years of data that can be combined (U.S. Census Bureau, 2021a).
Other data sources did not meet our inclusion criteria.For instance, family/caregiver indicators in the Behavioral Risk Factor Surveillance System (BRFSS) cannot be disaggregated for children from the prenatal period to age three.The Pregnancy Risk Assessment Monitoring System (PRAMS), National Health and Nutrition Examination Survey (NHANES), and National Health Interview Survey (NHIS) have indicators for infant and toddler well-being, but data are either not publicly available or not representative of state populations.Newborn screenings are not systemically measured so comparability across states is unclear.Further, administrative data can supply information on the number of children or families receiving varied services and benefits; however, these systems only include those receiving services and do not measure indicators for children who may be at-risk for adverse outcomes.

Ways to address challenges with child indicators and measurement sources
There are promising examples of ways to address the challenges with existing indicators and measurement sources highlighted in this report.In recent years, the World Health Organization developed the Global Scale for Early Development (GSED), a population-level measure of child development for children at birth to age three.The GSED is currently being validated in several countries, including the United States, with the goal of accelerating early childhood development monitoring (Black et al., 2019).Many European countries have population registries that provide basic data about all individuals, including infants and toddlers, that can be aggregated across communities.Data collected from these sources are shared with governments, schools, and community leaders to help make informed policy and program decisions that address the needs of all children.Some states in the United States have also developed ways to address the challenges with existing child indicators and measurement sources.For example, several states have started a process to include home visiting data within early childhood integrated data systems (Lin, 2019).The long-term goal of these efforts is to gain a comprehensive understanding of child and family access to the programs and services infants and toddlers have received.Minnesota's Early Childhood Longitudinal Data System (ECLDS) links data from the state departments of Education, Human Services, and Health.Using a public online platform, data users can access deidentified aggregate-level reports on topics such as birth records, program participation, financial assistance programs, and child and family demographics (Jordan et al., 2018).These efforts hold promise for linking administrative data with data about young children and their families, yet they are not at the stage where they include indicators of young children's development.
Other communities across the United States are exploring ways to use developmental screeners such as the Ages and Stages Questionnaire (ASQ) to create a proxy for understanding whether infants and toddlers are developmentally "on track."Developmental screeners cannot be used for diagnosis, but they can be effective in identifying the need for further assessment.Tracking assessment after developmental screening is complex and requires collaboration and coordination between practitioners, providers, and stakeholders.Efforts in Salt Lake City, Utah; Tarrant County, Texas; and Norwalk, Connecticut are being made to build partnerships, address data coordination challenges, and explore the viability of using screeners as population measures of infant and toddler development (K.Paschall, personal communication, November 4, 2021).
There may also be an opportunity to aggregate data in the United States in a similar fashion to European registries.Most infants and toddlers in the United States receive well-child check-ups (Keating et al., 2020).Information collected at these visits, such as timing of meeting developmental milestones, could theoretically be aggregated.This type of data linkage would be complex and would require integrating medical records across multiple data systems, using a shared data management platform, ensuring that data are de-identified in compliance with HIPAA, and following other relevant regulations.The de-identified, aggregate data source would then be a rich source of information on infant and toddler health and development.

Ways to prioritize future child indicator development and measurement efforts
We recognize that data collection efforts are extremely complex and expensive.There are ways to prioritize indicators collected through existing and upcoming data collection efforts when resources are limited.Here we focus on three considerations relevant to children's outcomes: (i) prevalence, (ii) severity, and (iii) disparate impacts.
Children's health has traditionally been assessed by evaluating indices that include the prevalence of adverse health conditions (e.g., proportion of newborns born small or too early), with a focus on discovering associations and finding effective prevention methods (NRC & IOM, 2004).The prevalence of typical child development is more prevalent than atypical child development, yet measures of appropriate developmental achievements appropriate for children from birth to three years are limited.Balancing indicators that measure adverse health outcomes, typical child development, and supportive contexts may help improve state-based monitoring for all children, including those who are at-risk versus presently demonstrating atypical development.
Another way of prioritizing the development of new indicators is to focus on factors that can severely impact subsequent development.Some measurement sources in this review collect data about conditions that can severely impact development, such as low birthweight or maltreatment.However, we lack comprehensive state-level measures on other potentially damaging conditions, such as malnutrition or psychosocial deprivation (Black et al., 2017).An important consideration when thinking about these types of indicators is whether social desirability bias prevents data from being accurately collected, as disclosing these conditions may be difficult for respondents.
Finally, some outcomes may be more impactful in vulnerable populations, raising equity concerns.A core principle of various indicators frameworks is to promote health equity (Raikes et al., 2017;Black et al., 2019).One example is the UNICEF "human rights-based approach" of selecting indicators for the child-related Sustainable Development Goals (SDG) that realize the rights of every child, especially the most disadvantaged.Measuring differences in well-being indicators across demographic subgroups can inform policy and program decisions to address health disparities.For instance, it is well known that maternal and child health inequities emerge even before birth in the United States (Wilkinson et al., 2021).Nationally representative studies find that measures of child development such as self-regulation or early learning skills vary by race and ethnicity by kindergarten (Piña et al., 2020), with differences even emerging by nine months (Halle et al., 2009).Yet, state-level data on these outcomes are limited for population subgroups.
Investing in resources to increase sample sizes and provide representative and comparable data across states could expanded opportunities for health disparity research.The NSCH has been able to increase its sample sizes through optional state oversamples, both of entire states and of more specific geographic areas.The NSCH costs approximately $16.82 per sampled address, but that is not the total cost per response, since an average of approximately six addresses must be sampled to achieve one completed response.Using a hypothetical example provided by the NSCH, it would cost a state about $58,870 to double their sample size (U.S. Census Bureau, 2021b).Among the 12 states that have or are currently sponsoring an oversample, actual costs for state oversamples have ranged from approximately $20,000 to $500,000 (A. Hirai, personal communication, October 22, 2021).Additionally, the CDC co-sponsored the 2021 NSCH in order to support a national increase in the sample size of children ages 1-5 years.The NSCH is also exploring options to improve state-level sample sizes for underrepresented racial/ethnic populations (A. Hirai, personal communication, October 22, 2021).
In other cases, current sampling procedures limit data sources' ability to be easily expanded.The NIS, for example, is a phone-based survey targeting a very small proportion of the population (as respondents need to have a toddler ages 19-35 months for the child survey).The CPS Food Security Supplement is an add-on to the CPS and is dependent on the CPS sampling frame.Expanding the sample of either the NIS or the CPS Food Security Supplement would be very resource intensive.The NSCH is also exploring options to adapt its sampling to be able to target specific groups of children (A. Hirai, personal communication, October 22, 2021).
Recent national efforts have been established to prioritize equitable data collection and reporting.In January 2021, President Joe Biden signed an executive order to establish an Interagency Working Group on Equitable Data.This group will identify weaknesses in federal data collection efforts and help agencies expand and improve their data collection efforts (Exec. Order No. 13985, 2021).Additionally, the Robert Wood Johnson Foundation has established a National Commission to Transform Public Health Data Systems, aiming to rethink and improve the public health data system to promote health equity.We recognize the overlaps between developmental delay and special health care needs (SCHN) as well as the complexities of identifying and serving these needs for the 0-3 age group c Housing stability and housing instability are catch-all terms that capture the impacts of issues such as crowding, frequency of moves, homelessness, and cost burden of home expenses d Many of the child-level and family/caregiver indicators are related to multiple developmental domains.Select infant and toddler cognition and language indicators available at the child or family/caregiver levels

Fig. 1 .
Fig. 1.Framework for the identification of infant and toddler well-being indicators Note: This model was adapted from Bronfenbrenner (1979)'s social-ecological model for this paper.
development: NSCH asks a series of items for children ages 1-5 about their ability to speak, form sentences, and ask questions, and follow directions.NSCH Developmental Delays: Percent of infants and toddlers whose parents have been told by a health care provider that they have a developmental delay NSCH Family/caregiver indicators Developmental Surveillance: Percent of infants and toddlers whose parents were asked by a health care provider about the child's learning, development, or behavior NSCH Developmental screening: Percent of infants and toddlers whose parents filled out a developmental screener from a health care provider in the past year NSCH Reading to children: Percent of infants and toddlers who are read to every day NSCH Singing and telling stories: Percent of infants and toddlers who are told stories or sung songs to every day NSCH This is a select set of the most well-measured early cognition and learning indicators.See Appendix C for a full set of indicators and details about their measurement properties Child Indic Res.Author manuscript; available in PMC 2022 August 02.

Appendix B.: Physical health indicators Indicator Sub-domain Developmental stage Data source Definition by data source and notes Years available and periodicity
They are listed in the most relevant domain to save space Many of the family/caregiver indicators are related to multiple domains, but to streamline this manuscript, we placed the indicators in the most aligned domain

Appendix C.: Early cognition and language indicators
Many of the family/caregiver indicators are related to multiple domains.To streamline this manuscript, we placed the indicators in the most aligned domain b Vital statistics data were accessed through CDC Wonder (wonder.cdc.gov).Additional data, including more historical data, may be available through the National Center for Health Statistics' Vital Statistics Online Data Portal (https:// www.cdc.gov/nchs/data_access/vitalstatsonline.htm)