81% of the baby foods in one US city were ultra-processed

Researchers photographed the baby/toddler aisles at 21 grocery stores in Austin and analyzed 2,783 products marketed for children 6–36 months, excluding infant formula, beverages, and purees; 81% were classified as ultraprocessed under the NOVA system.

Note

It's a conference abstract presented at NUTRITION 2026 (American Society for Nutrition, July 25–28, National Harbor MD), not peer-reviewed, so it should be treated as preliminary. Lead author Erin A. Hudson, PhD candidate at UT Austin, with Marissa Burgermaster.
One metro area, shelf audit — it measures what's available, not what toddlers actually eat.
Industry pushback focuses on there being no universally accepted scientific definition of "ultraprocessed", which is a fair methodological criticism of NOVA.

From: UT Austin release (cns.utexas.edu) and the ASN release on EurekAlert (news release 1136466).

Asked Claude AI How many of the increased infant health problems are associated with UPF

Short answer: nobody has published an attributable-fraction estimate for infant or toddler health problems. Every population-attributable-fraction (PAF) model I can find is adults-only — 4% to 14% of premature deaths across eight countries, 21% of obesity cases in Belgium, 23–38% of CVD events in Canada. There's no equivalent for the 0–3 age group, so any "X% of rising infant problems" number would be invented.

What has been measured in early life:

Childhood overweight/obesity — the largest study, 19,958 mother-child pairs from NHS II and GUTS, found a 26% higher risk in offspring of mothers in the highest quintile of UPF intake during the child-rearing period vs the lowest, after adjusting for the child's own UPF intake, activity, and sedentary time. Worth noting: peripregnancy UPF intake was not significantly associated — which argues against a straightforward in-utero programming story and toward the household food environment.

Infant atopic dermatitis — Korean MOCEH cohort, 861 pregnant women, highest vs lowest quartile of maternal UPF energy (≥15.5% vs ≤6.8%) gave an adjusted OR of 2.19 (1.11–4.32) for AD in the first 12 months.

Childhood asthma — SENDO cohort, published May 2026: children getting >30% of energy from UPF had HR 3.76 for incident asthma (95% CI 1.15–11.51). That CI is the story — it's compatible with a 15% increase or a 10-fold one. No association with allergic asthma, atopic dermatitis, aeroallergen sensitization, or food allergy, which the authors read as a non-atopic inflammatory pathway.

Food allergy — the EAACI task force review found early introduction of commercial baby foods, many classified as UPF, linked to higher risk of challenge-proven food allergy in infancy, though across studies the associations held in many but not all cohorts.

Why the "how many" question resists an answer:

All observational. UPF intake correlates hard with income, maternal education, breastfeeding duration, maternal BMI — and in one of these cohorts, lower maternal UPF consumption was associated with longer breastfeeding. Residual confounding could account for most or none of these effect sizes.
No infant RCTs, and there won't be.
Each rising condition has three or four competing explanations already (diagnostic expansion, C-section and antibiotic exposure, microbiome, vitamin D status, hygiene hypothesis). UPF is a candidate, not a settled contributor, and these explanations aren't independent — UPF-heavy households also skew low on sun exposure and supplementation, so vitamin D and UPF confound each other in both directions.
NOVA classification itself is contested, so the exposure variable isn't stable across studies.

Attributable fraction in infants is unquantified.

81% of the baby foods in one US city were ultra-processed

Related in VitaminDWiki