Routine developmental assessments accurately identified infants whose early milestone failures continued through age 2 and were indicative of persistent—rather than transient—delays, according to a national retrospective cohort study conducted in Israel.
The cohort included infants who were born between 2014 and 2022 and analyzed from July 2024 through April 2025. Among 529,797 infants evaluated at 9 to 12 months, 37,760 (7%) failed to meet at least one milestone during the first visit. Of those who were reassessed (35,163 participants) from 12 to 24 months, 1 in 4 continued to fail at least one milestone, which indicated persistent developmental delays that may indicate underlying neurodevelopmental conditions requiring early intervention.
Transient delays, by comparison, typically result from environmental factors or minor maturational variations.
Persistent delay was further defined as either specific, meaning failure in the same developmental domain, or general, meaning failure in any domain at follow-up. Infants who failed milestones across multiple developmental domains were more likely to experience ongoing delays than those who failed in only one domain.
Rates of specific persistent delay ranged from 2% in fine-motor skills to 22% in gross-motor milestones. General persistent delay ranged from 23% to approximately 31%, depending on the domain. Counting the number of developmental domains in which unmet milestones occurred served as a simple indicator of persistence. Sensitivity analyses that extended follow-up to 36 months and adjusted milestone thresholds produced consistent findings, which supported the robustness of the results.
Machine-learning models using milestone and perinatal data predicted which delays persisted with moderate accuracy and achieved areas under the receiver operating characteristic curve between 0.71 and 0.77.
In their analysis, the researchers used data from Israel’s national maternal and child health clinics, which routinely track 59 developmental milestones across 4 analytic cohorts: language-social (combined), fine motor, gross motor, and any milestone. Infants born at or after 37 weeks’ gestation who completed both assessments were included. Severe failure was defined as not achieving a milestone that 95% of peers reached.
Participants represented a demographically diverse cohort across all Israeli districts, with reassessment rates ranging from 74% to 80% in fine motor milestones and 91% to 95% for at least one language-social or gross motor milestone. The investigators trained gradient-boosting algorithms to predict persistence of developmental delay and compared these algorithms with a rule-based method that counted failed domains. The incorporated models routinely collected clinical and perinatal information from electronic health records that covered about 65% of Israeli children.
“To our knowledge,” wrote lead author Yonatan Bilu, PhD, of the KI Research Institute in Kfar Malal, Israel, with colleagues, “our analysis offers the first quantitative evidence that an intuitive rule already used informally by many clinicians across a wide range of milestone-based monitoring tools does indeed perform reliably, thereby lending an evidence base to everyday practice.”
However, they cautioned that milestone failure does not equal a formal diagnosis of developmental delay, and that their findings may not be generalizable beyond Israel’s health care system. The data lacked detailed information on socioeconomic context, comorbidities, and interventions, all of which may influence child development outcomes.
The authors concluded that the Tipat Halav Israeli Surveillance developmental scale could serve as an effective bridge between routine developmental surveillance and targeted early intervention. Developmental surveillance, they noted, can be implemented in multiple settings: in daycare centers, by parents and caregivers at home, and by allied health professionals such as physiotherapists and speech-language pathologists. By integrating structured milestone assessments with predictive models, they wrote, health systems may improve early identification of persistent delays, optimize pediatric and rehabilitation resources, and enhance equity in child-development monitoring.
Full disclosures can be found in the published study.
Source: JAMA Network Open