The curve that a child’s body mass index traces between the first and seventh birthdays may be one of the most revealing vital signs that pediatric medicine is not yet systematically watching. That is the central message of an eight-year study of 1,072 mother–child pairs in Tianjin, China, published on 28 August 2026 in the World Journal of Pediatrics. A team led by Ming Gao and Jing Li of Tianjin Medical University, with colleagues at the Tianjin Women and Children’s Health Center, Dalhousie University and Pennington Biomedical Research Center, followed annual height and weight measurements through early childhood and then, when the children turned eight, assessed their blood pressure, fasting glucose and blood lipids. Two patterns of BMI development emerged as distinctly dangerous. Children whose BMI entered the obese range and stayed there from toddlerhood onward — a trajectory the researchers call the persistent obesity growth pattern — faced nearly four times the odds of clustering multiple cardiometabolic risk traits by age eight. Children who veered into obesity later, on a so-called late obesity growth pattern, still carried 70 percent higher odds than peers on a normal course.

The findings arrive against the backdrop of an accelerating global childhood obesity epidemic. The World Obesity Federation’s 2024 atlas documents the scale of the problem, and forecasting analyses conducted for the Global Burden of Disease Study 2021 project that the prevalence of child and adolescent overweight and obesity will continue climbing through 2050. Pediatric researchers have long observed that cardiovascular risk factors — elevated blood pressure, high fasting glucose, unfavorable lipid profiles — tend not to appear one by one in adulthood but begin clustering together early in life. That clustering is consequential because childhood cardiometabolic risk is not a benign, temporary state: decades-long international cohorts have shown that risk factors measured in childhood predict actual cardiovascular events, including heart attacks and strokes, in middle age. The clinical difficulty has always been timing. Blood pressure, glucose and cholesterol are usually unremarkable in healthy young children, and a single weight measurement at any one visit reveals little about where a child is heading. The Tianjin study’s premise was that the trajectory — the shape, timing and steepness of BMI change — carries predictive information that no single snapshot can provide.

To capture that shape, the investigators drew on the Tianjin Antenatal Network, a municipal system of maternal and child health services through which the 1,072 mother–child pairs were recruited and followed for eight years. Study personnel measured each child’s height and weight annually from age one to age seven, allowing the team to calculate body mass index — weight in kilograms divided by the square of height in meters — at seven consecutive ages. At age eight the children underwent a cardiometabolic assessment comprising blood pressure measurement, fasting glucose and lipid concentrations. The researchers summarized these results in two complementary ways. The first was a modified continuous cardiometabolic risk score, or modified cMetS, in which each measured component is converted into a standardized Z-score and aggregated so that higher values indicate a worse overall risk profile. The second was a classification of cardiometabolic risk traits, flagging children with clinically elevated values across these measures. The team then used generalized linear and logistic regression models, adjusted for potential confounding factors, to estimate how each adverse trajectory related to the age-eight outcomes relative to a control trajectory.

The analytical backbone was a technique known as group-based trajectory modeling, a form of finite-mixture modeling that does not merely average BMI change across the cohort but searches for hidden subgroups of children whose BMI followed similar developmental courses. Fed seven years of repeated measurements, the algorithm sorts the population into a small number of latent trajectories, each with its own estimated curve and share of members. In this cohort the procedure isolated two adverse patterns alongside the reference group. The persistent obesity growth pattern captured children whose BMI was already elevated early and remained high across the entire window — an unbroken stretch of excess adiposity. The late obesity growth pattern told a different story: children who tracked a comparatively normal course through the toddler and preschool years and then rose into the obese range in the later portion of the observation period. Distinguishing the two was the point. Persistent and late-onset excess weight may impose different biological pressures on a developing cardiovascular system, and collapsing all children with childhood obesity into a single category would erase that clinically meaningful difference.

The results were graded by severity. In fully adjusted models, children on the late obesity growth pattern scored 0.88 points higher on the modified cMetS than controls, with a 95 percent confidence interval of 0.53 to 1.22, and had 1.70 times the odds — 95 percent confidence interval 1.20 to 2.41 — of exhibiting cardiometabolic risk traits. Children on the persistent obesity growth pattern fared considerably worse: a 1.59-point elevation in the continuous risk score, with a 95 percent confidence interval of 0.87 to 2.31, and an odds ratio of 3.91, spanning 1.92 to 7.99. Because the confidence intervals exclude zero for the coefficients and one for the odds ratios, the associations hold at conventional thresholds of statistical significance. When the team dissected the individual components of the composite outcome, one signal dominated: hypertension. The trajectory effects concentrated most clearly on elevated blood pressure, an observation consistent with pediatric guidelines that increasingly urge clinicians to look for high blood pressure in childhood rather than treating it as an adult-only condition. Glucose and lipid measures contributed to the composite score, but blood pressure emerged as the trait most consistently linked to early BMI patterns.

Beyond the shape of the whole curve, the researchers asked a second, more quantitative question: does the sheer accumulated dose of excess weight matter, independent of when it arrives? To answer it they computed excess BMI-years, a metric borrowed from adult cardiovascular epidemiology that integrates both the magnitude and the duration of excess adiposity. Conceptually it is the area between a child’s observed BMI and the level expected for a healthy child of the same age and sex, summed across the years of observation. A child who sits moderately above the healthy curve for five years and a child who sits far above it for two years can accumulate comparable doses, and the metric deliberately treats them as equivalent exposures. In the Tianjin cohort, greater excess BMI-years were associated with higher modified cMetS scores and greater odds of cardiometabolic risk traits at age eight. The result reframes childhood obesity less as a categorical diagnosis and more as a cumulative exposure — analogous to pack-years in tobacco research — whose accumulating insult to the cardiovascular system can be tallied in unit-years rather than cigarettes.

The timing analysis centered on a developmental landmark most parents have never heard of: the adiposity rebound. In typical development, BMI rises steeply in infancy, declines through the toddler and preschool years as children lean out, bottoms out around age five or six, and then turns upward again. That inflection point has fascinated obesity researchers for decades because children whose BMI bottoms out early and begins climbing ahead of schedule tend to carry more body fat later in childhood and into adulthood. The Tianjin data reinforce the signal: an earlier adiposity rebound was associated with higher cardiometabolic risk scores and greater odds of risk traits at age eight. The underlying biology is still debated — an early rebound may reflect early adiposity itself, altered growth signaling, or early-life nutritional programming — but its practical value is that it is observable in ordinary clinical measurements. A clinic that plots BMI at every well-child visit can, in principle, watch the rebound occur and recognize an unusually early one, converting an abstract epidemiological concept into a visible warning at precisely the visit where it appears.

The team then decomposed the seven-year window into three intervals — ages one to three, three to five, and five to seven — and asked whether rapid BMI gain in any particular interval carried disproportionate weight. The methodological instrument was the conditional BMI Z-score gain: for each interval, a child’s current Z-score is regressed on the prior Z-score, and the residual — the gain that cannot be predicted from where the child already stood — serves as a measure of interval-specific acceleration. This statistical maneuver strips out the strong persistence of BMI over time that would otherwise swamp any period analysis. The verdict was pointed. Greater conditional BMI gains during ages three to five and ages five to seven were each associated with higher modified cMetS scores and higher odds of cardiometabolic risk traits, whereas gains between ages one and three did not show the same associations. The pattern aligns with evidence from the Dutch Terneuzen Birth Cohort, which found that BMI change between ages two and six was the childhood measure most predictive of adult cardiometabolic risk. Together, such studies point to the preschool-to-early-school transition as a sensitive period for cardiometabolic programming.

The authors conclude that the late and persistent obesity growth patterns in early childhood are associated with increased cardiometabolic risk by age eight, and that greater cumulative excess BMI burden, an earlier adiposity rebound and accelerated BMI gain within specific windows may help identify children at higher risk. The practical appeal is that none of these signals requires new technology: annual height and weight measurements are already routine in pediatric care, and the study relied on precisely such ordinary anthropometry plus a single age-eight blood draw and blood pressure check. The research, supported by the National Natural Science Foundation of China and the International Diabetes Federation’s BRIDGES program, also carries caveats. It is observational, so unmeasured differences among children — diet, physical activity, sleep, genetic predisposition — could partly explain the associations despite statistical adjustment. The cohort came from a single Chinese city, and a continuous risk score, while standard in pediatric research, is not equivalent to a clinical diagnosis of metabolic syndrome. Even so, the message lands with unusual clarity: by the time a child turns eight, seven years of growth data have already been written, and the body appears to keep the record.