Association of exposure to synthetic phenols and metal(loid)s with early puberty in Spanish girls: a multicentric case–control study

Olivas-Martinez and colleagues measured eight metal(loid)s and synthetic phenols in first morning urine from girls recruited in six Spanish hospitals during 2018–2022. Higher urinary zinc was associated with greater odds of early puberty and precocious puberty in adjusted models, while the continuous models for arsenic, copper, mercury and nickel did not identify statistically significant associations. The combined chemical mixture was associated with early puberty, with bisphenol A contributing the largest positive weight. This is human biomonitoring and observational health evidence; no sunscreen, fish, food or other consumer product was sampled.

Key numbers

There were 310 participants: 182 cases and 128 controls. Cases comprised 101 girls with precocious puberty, 74 with premature thelarche, six with early pubarche and one with isolated vaginal bleeding. Metal measurements were available for 180 cases and 127 controls; synthetic phenol measurements were available for 181 cases and 126 controls. Mean ages were 7.44 years (SD 0.78) and 6.24 years (SD 1.28), respectively. Recruitment occurred during 2018–2022; the paper provides no per-year or per-region metal concentration estimates (PDF pp. 1–4, Table 1; p. 5, Table 2).

The following are urine concentrations in the paper’s native unit, ng/mL, not concentrations in a consumer product. Methods describes measuring specific gravity for dilution adjustment, but Table 2 does not explicitly label its medians as adjusted or unadjusted. Specific-gravity medians were 1.024 in cases and 1.023 in controls. Each metal row uses 180 cases and 127 controls (PDF p. 5, Table 2).

The table footnote specifies a Mann–Whitney test for chemicals detected in more than 85% of samples and a chi-squared comparison of detected versus undetected categories for chemicals detected in fewer than 85%. Therefore, the reported p-values for Cd, Mn and Pb should not be presented as median-comparison tests. The Mn median is censored below 0.07 ng/mL; it is not zero. The main PDF does not report a complete table of metal LODs or LOQs.

Continuous-exposure models included As, Cu, Hg, Ni and Zn because their detection frequencies exceeded 85%. Odds ratios below are per two-fold increase in urinary concentration; adjusted models include age, hospital and maternal education (PDF p. 3, Methods; p. 6, Table 3).

Zinc in the third versus first tertile was associated with an OR of 2.28 (95% CI 1.10–4.73) for all early-puberty diagnoses. The authors also reported inverse associations for categorized Cd and Cu exposures. Figure 1 has no numeric labels for these estimates, so their exact odds ratios and confidence intervals are not reconstructed from plot positions. These observations do not demonstrate a protective effect of cadmium or copper (PDF p. 6, Results; p. 7, Figure 1).

BPA medians were 1.78 ng/mL in cases and 0.60 ng/mL in controls. Its adjusted OR per doubling was 1.44 (95% CI 1.19–1.73) for all diagnoses, 1.29 (1.05–1.58) for premature thelarche and 1.69 (1.26–2.27) for precocious puberty. This phenol context matters when interpreting results for the joint mixture (PDF pp. 5–6, Tables 2–3).

The mixture contained BPA, methylparaben, benzophenone-3, As, Cu, Hg, Ni and Zn. Effects below are per one-quartile increase in joint mixture concentration, adjusted for age, hospital and maternal education (PDF p. 7, Table 4).

BPA’s reported positive-direction mixture weights were 46% for all diagnoses and 52% for precocious puberty. A model weight is not the proportion of cases caused by that chemical, and the joint estimate cannot be attributed to metal exposure alone (PDF p. 6, Results; p. 8, Figure 2).

Methods (brief)

This was a multicentric hospital-based case–control study. Participants collected a single first morning void urine sample in a polypropylene container. Samples remained frozen at −80 °C before analysis at the University of Granada. Metals were measured using an Agilent 8900 triple quadrupole inductively coupled plasma mass spectrometer. Synthetic phenols were measured by dispersive liquid–liquid microextraction and ultra-high performance liquid chromatography with tandem mass spectrometry. Specific gravity was measured with a handheld refractometer to address urine dilution. The main paper refers sample preparation and analytical quality control to supplementary material (PDF pp. 2–3).

Values below the LOD were assigned LOD/√2 for analysis. Biomarkers with detection frequencies above 85% entered log-transformed and tertile models; those detected in 50–85% entered categories based on the LOD and median; those detected in fewer than 50% entered detected/undetected models. Quantile g-computation used quartiles of eight frequently detected chemicals and 1,000 bootstraps. Primary confounder adjustment comprised age, hospital and maternal schooling. Sensitivity analyses added BMI or excluded girls from the hospital without controls (PDF p. 3).

The absence of controls at HUVA, unequal case/control ages, one urine measurement, residual confounding and multiple comparisons limit interpretation. The authors also identify urine as an unsuitable measure of chronic Pb and Hg exposure, potentially causing exposure misclassification. They report similar sensitivity-analysis results, but the detailed supplementary estimates were not available in the main PDF (PDF pp. 6, 8–9).

Implications

The study supports source-native urinary medians, detection frequencies and associations with contemporaneous pubertal diagnoses. It is an exposure and health-effect source (lane a4), with urine as its matrix and Spain as its population jurisdiction. Product and ingredient arrays are empty because a research-pull query mentioning sunscreen does not establish a measured product or an exposure source.

The case–control design and a single urine sample cannot establish exposure before puberty onset or a causal effect. Arsenic and mercury were measured without reported speciation: their results are retained as total/unspeciated elemental measurements, not inorganic arsenic or methylmercury. The values cannot supply product contamination concentrations, ingredient contamination profiles, product-specific exposure attribution or clinical risk thresholds. Detailed analytical quality-control information and several sensitivity analyses are delegated to a supplement absent from the preserved main PDF.

This source contributes urinary exposure and reproductive-health context to the relevant metal pages. It separates individual-metal associations from a mixture signal driven primarily by BPA, and keeps negative or inconclusive findings visible alongside the zinc association. It neither identifies the products responsible for exposure nor supports individual clinical predictions.

The source was located under a sunscreen query, but its actual participants and laboratory matrix govern routing. The prior fish and mineral-sunscreen routes were unsupported and should not contribute to product occurrence evidence. No regional concentration gradient, historical exposure trend or remediation effect was measured.

Wiki pages this source may touch

• Arsenic: unspeciated urinary arsenic medians and observational association estimates.
• Cadmium: urinary cadmium medians, detection frequencies and categorical association context.
• Copper: urinary copper results and contrasting continuous/categorical associations.
• Mercury: unspeciated urinary mercury results and the authors’ chronic-exposure matrix limitation.
• Manganese: low detection frequency and censored urinary medians.
• Nickel: urinary nickel medians and adjusted association estimates.
• Lead: urinary lead results and the authors’ chronic-exposure matrix limitation.
• Zinc: urinary zinc medians and individual/mixture reproductive-health associations.

Verification notes

The complete 13-page PDF was read using pdftotext -layout. Tables 1–4 and Figures 1–2 were visually checked on rendered PDF pages 4–8. DOI, full title, publication, year and all 14 authors were taken from the PDF byline and bibliographic header; the CC BY 4.0 license appears on page 9. Publisher record.

The stable source path is the raw_path above, through the repository’s ingredient-sweep symlink. It resolves to /Volumes/HMI-Library/research-pulls/ingredient-sweep/Baby sunscreen chemical UV filters/Baby sunscreen chemical UV filters Africa heavy metals/s00431-026-06919-1.pdf. The file is 1,031,210 bytes and has SHA-256 3d2a58257d6b952bd3877dbc6982687a831105072d7ece250d8d7fa939f33280. The discovery manifest identifies the same DOI and an Europe PMC retrieval URL, https://europepmc.org/api/getPdf?pmcid=PMC13106254. Its query and folder name are discovery metadata, not study geography or product identity.

This correction rolls up the same DOI and file hash previously recorded under the malformed legacy cite key authors2026-association-of-exposure-to-synthetic-phenols-and-metal and the mismatched raw handle RPMEHG_10-1007-s00431-026-06919-1. The author-based key and RPIS handle identify this same work; it must not remain as a second source record under the old key.

Table 2 is retained over contradictory narrative: arsenic medians are 22.23 and 27.66 ng/mL in the table, versus 22.20 and 27.60 in the preceding prose. The prose also calls total phenols significantly different, whereas Table 2 gives p = 0.24; no significant total-phenol difference is asserted here. Table 1’s unusually large printed BMI z-score values and malformed category inequalities are source-quality issues, not values corrected by this ingest.

Methods calls the phenol-family sums molecular-weight-based molar sums while the Table 2 heading uses ng/mL. Those sums are retained in structured evidence with this unit ambiguity and are not converted or combined with individual mass concentrations. The full supplementary methods, exact categorical estimates shown only graphically, and participant-level data remain explicit extraction gaps. Source-native values were preserved without concentration conversion; arsenic species, mercury species, urinary and product matrices remain separate.

Update history

No substantive edit history is available in this build. The full commit record is available in git.