Don’t take our word for it. Here’s every fact on this site, with its evidence grade, its caveats, and a link to the source, so you can check us.
We start you on the strong ones: findings that have been replicated or come from one large, well-controlled study. Moderate and weak facts are here too, labeled for what they are. Use “Link to this fact” to share a single card.
Health effects / Other evidence
The Cragoe 2026 abstract's headline single-paraben findings run past its own intervals: propylparaben with lower shift rate in girls is printed with a confidence interval of -0.032 to 0.027, which includes zero.
Moderate evidence: Reporting inconsistency within the paper (per-IQR vs per-unit scaling), not a hidden effect; read intervals from the Results and figures.
Source: Cragoe 2026, Environmental Epidemiology Link to this fact
Health effects / Other evidence
The rates of parent-reported autism (13% of families, 32% in India) and ADHD (18%) in this survey are far above real-world prevalence, and the authors warn they must not be compared across countries.
Moderate evidence: Authors state the rates reflect survey bias, not prevalence.
Source: Palmer & Kattari 2026, J Xenobiotics, Journal of Xenobiotics Link to this fact
Health effects / Review of other studies
A 2026 narrative review says a first-time adult ADHD diagnosis should not rest on a screening score alone, because mood, anxiety, substance-use, sleep, stress and personality conditions can produce similar attention complaints.
Moderate evidence: Expert narrative review; it does not estimate how often each condition is mistaken for ADHD.
Health effects / Other evidence
The review's list of adult ADHD mimics is psychiatric and sleep-related; it does not name chemical exposure, chemical intolerance, thyroid disease or long COVID.
Moderate evidence: Silence in one narrative review is not evidence that such conditions never mimic ADHD.
Health effects / Review of other studies
About 6% of US adults reported a current ADHD diagnosis in 2023, and roughly half of those were first diagnosed at age 18 or older; the survey cannot say whether any diagnosis is wrong.
Moderate evidence: Cross-sectional self-report; reflects help-seeking, access and recall as well as prevalence.
Health effects / Other evidence
A rise in adult ADHD diagnoses is compatible with DSM-5 criteria broadening, correction of past underrecognition and expanded access; the review says current evidence does not quantify these and does not show diagnostic inflation or any environmental cause.
Moderate evidence: One field trial; narrative synthesis.
Health effects / Other evidence
Global Burden of Disease 2021 model estimates that about 47 million people under 20 (1.78% of that age group) had ADHD in 2021, with a 95% uncertainty range of 1.2% to 2.6%.
Moderate evidence.
Caveats
Model-based, using studies to about 2017 with imputation for countries without data; the authors say the burden is probably underestimated, and earlier meta-analyses gave about 5-7%.
Source: Cortese 2026 (Mol Psychiatry), Molecular Psychiatry Link to this fact
Health effects / Other evidence
Between 1990 and 2021 the modeled global ADHD prevalence rate in under-20s fell about 6% (from 1,892 to 1,779 per 100,000) while the number of cases rose about 10% with population growth, so the model shows no rising underlying rate.
Moderate evidence: A modeled series with wide uncertainty and input data ending about 2017; it does not capture recent rises in diagnoses in services, and no formal trend test is reported.
Source: Cortese 2026 (Mol Psychiatry), Molecular Psychiatry Link to this fact
Health effects / Other evidence
In the GBD 2021 model, ADHD prevalence in under-20s is about 2.5 times higher in males (2.52%) than females (0.99%) in every world region.
Moderate evidence: The authors note the ratio may partly reflect rating-scale studies, because a meta-analysis found no sex difference in severity in clinical-interview studies.
Source: Cortese 2026 (Mol Psychiatry), Molecular Psychiatry Link to this fact
Health effects / Other evidence
The author states that in 2006 nobody knew how toxic ozone-terpene particles are compared with outdoor particles, and that the effects of inhaling their peroxide and hydroperoxide products remained to be evaluated; one 2-hour human exposure to a VOC mixture with 40 ppb ozone showed no significant symptoms.
Moderate evidence.
Caveats
A 2006 status statement from one review author, partly superseded by later cell and chamber work (e.g., a 2024 bronchial-cell null); the human null is one short exposure in healthy women.
Source: Weschler 2006, Environmental Health Perspectives Link to this fact
Health effects / Other evidence
A 2007 systematic review pooled 102 surveys of 171,756 children and adolescents (1978-2005) and found about 5.3% (95% CI 5.0-5.6%) met ADHD criteria, with large differences between studies.
Moderate evidence.
Caveats
Pooled survey point prevalence of diagnostic criteria; significant heterogeneity, no publication-bias test or quality scoring; 1978-2005 only; four of five authors disclose ADHD-drug industry ties, funders reported no role.
Source: Polanczyk 2007, American Journal of Psychiatry Industry-funded Link to this fact
Health effects / Other evidence
How ADHD was defined and measured was significantly associated with the spread in prevalence between studies: no impairment requirement, DSM-IV (versus ICD-10 or DSM-III-R) and parent or teacher reports (versus a best-estimate procedure) each went with higher estimates, while the 'and rule' went with lower ones.
Moderate evidence.
Caveats
Study-level associations from a metaregression of observational surveys (p-values given, no effect sizes in text); parent-report p is 0.02 in text and 0.03 in Table 2; no I-squared reported.
Source: Polanczyk 2007, American Journal of Psychiatry Industry-funded Link to this fact
Health effects / Review of other studies
The review found sensory problems in ADHD to be measured almost entirely by parent report, and noted that parents of these children often have depression or anxiety and that whether this colors their ratings has not been studied.
Moderate evidence: Author's own caveat; no data on the size of any reporting bias.
Source: Ghanizadeh 2011, Psychiatry Investigation Link to this fact
Health effects / Other evidence
A 2012 meta-analysis of 20 double-blind challenge studies (794 children) found synthetic food colors went with a small increase in parent-rated ADHD-type symptoms (standardized effect 0.18, 95% CI 0.08 to 0.29), shrinking to 0.12 (0.01 to 0.23) after adjusting for possible publication bias.
Moderate evidence.
Caveats
Mostly small, old trials on children preselected as diet responders; heterogeneity moderate (I2 52%); funded by an industry-supported institute and the National Confectioners Association; food colors are not fragrance ingredients.
Source: Nigg 2012 Industry-funded Link to this fact
Health effects / Other evidence
In the same meta-analysis, teacher and observer ratings (10 studies, 323 children) showed no reliable overall effect of food colors (0.07, 95% CI -0.03 to 0.18), and the parent-rated effect was not significant in studies using only FDA-approved colors (13 studies, 0.13, -0.04 to 0.30).
Moderate evidence: Small pooled samples; only one small US-color study used valid measures in non-preselected children; industry-supported funding.
Source: Nigg 2012 Industry-funded Link to this fact
Health effects / Other evidence
A 2013 meta-analysis of 33 studies (10,232 children and adolescents) found a small association between lead exposure and ADHD symptoms: inattention r about .16 (95% CI .12-.20, 27 studies) and hyperactivity/impulsivity r about .13 (.09-.16, 23 studies).
Moderate evidence: Observational studies only; heterogeneity I2 71% (inattention) and 50% (hyperactivity/impulsivity); causation not established.
Source: Goodlad 2013, Clinical Psychology Review Link to this fact
Health effects / Other evidence
In that review, omega-3/omega-6 fatty acid supplements gave a small benefit for ADHD symptoms that survived blinded ratings (standardized difference 0.16, 95% CI 0.01 to 0.31, 11 trials); the authors say its clinical importance is undetermined.
Moderate evidence: Lower bound of the CI is near zero; small effect; the review was supported in part by a fatty-acid supplement maker among other funders (stated no input).
Source: Sonuga-Barke 2013, American Journal of Psychiatry Link to this fact
Health effects / Other evidence
The review shows blinding matters: with the rater closest to treatment all six ADHD treatments tested looked helpful, but with probably blinded raters the behavioral (0.02), neurofeedback (0.29) and cognitive training (0.24) results were not significant, which the authors read as probable rater bias.
Moderate evidence: Blinded ratings may be less reliable or less ecologically valid (the authors raise this too); heterogeneity is high in several domains.
Source: Sonuga-Barke 2013, American Journal of Psychiatry Link to this fact
Health effects / Review of other studies
A 2013 narrative review by Cardiff University psychiatric geneticists concluded that no clear-cut causal environmental risk factor for ADHD had been identified, apart from rare extreme early deprivation; prenatal smoking, alcohol, maternal stress, lead, pesticides and PCBs were graded as risks that are not proven causal.
Moderate evidence: Narrative review, no systematic search or risk-of-bias grading; literature to about 2012; the causal grades are the authors' judgment.
Source: Thapar 2013, Journal of Child Psychology and Psychiatry Link to this fact
Health effects / Review of other studies
ADHD twin heritability is about 71-90%, but the authors stress that this figure already includes gene-environment correlation and interaction, applies to populations and does not mean 90% of an individual's ADHD is caused by genes.
Moderate evidence: Population statistic from twin studies as cited by the review; later work reports a mean of 74%.
Source: Thapar 2013, Journal of Child Psychology and Psychiatry Link to this fact
Health effects / Review of other studies
The review says lead is the one ADHD-linked chemical exposure whose associations are unlikely to be fully explained by residual confounding, but still grades it as risk factor not proven causal, and advises reducing toxin exposure for children.
Moderate evidence: Lead is not a fragrance ingredient; the review is narrative; the 'unlikely confounded' view is the authors' reasoning, not a pooled adjusted estimate.
Source: Thapar 2013, Journal of Child Psychology and Psychiatry Link to this fact
Health effects / Other evidence
A 2014 meta-regression of 135 community studies (1985-2012) found that ADHD prevalence estimates did not rise with year of study once the diagnostic criteria, the informant, and the impairment requirement were held constant (+0.02 percentage points per year, 95% CI -0.16 to 0.20).
Moderate evidence.
Caveats
Compares different cross-sectional studies, not repeated surveys of one population; year of publication was used as a proxy for assessment year in about 30% of studies; data end in 2012; two authors have drug-maker ties (disclosed).
Source: Polanczyk 2014, International Journal of Epidemiology Link to this fact
Health effects / Other evidence
Most of the wide variation in ADHD prevalence between studies comes from how the studies were done: informant, diagnostic criteria, and whether impairment was required together explained 44% of the variance; teacher reports gave about 5.5 percentage points higher rates than best-estimate procedures, and requiring no impairment added about 2.3 points.
Moderate evidence: 56% of the variance remained unexplained; no heterogeneity I-squared or publication-bias test reported; 66% of studies overlap with the authors' 2007 review.
Source: Polanczyk 2014, International Journal of Epidemiology Link to this fact
Health effects / Other evidence
The rise in diagnosed and treated ADHD that US, UK and Canadian administrative data show is, in the authors' reading of this meta-regression, probably awareness and access to services rather than more children meeting criteria; they say a change in clinical practice cannot be ruled out and that this inference is indirect.
Moderate evidence: Inference from flat community estimates, not a direct test of why diagnoses rose; flat prevalence says nothing about causes in individual children.
Source: Polanczyk 2014, International Journal of Epidemiology Link to this fact
Health effects / Other evidence
The authors conclude that the smoking-ADHD association reflects unmeasured familial (genetic or household) confounding rather than a direct effect of smoke exposure, while stating this is no argument for relaxing advice not to smoke in pregnancy because of other proven harms such as low birth weight.
Moderate evidence.
Caveats
Authors' interpretation; cannot rule out a small effect or effects on finer neurocognitive traits; the sibling confidence intervals (0.73-1.06; 0.65-1.06) exclude large but not small effects.
Source: Skoglund 2014, Journal of Child Psychology and Psychiatry Link to this fact
Health effects / Other evidence
Cousin and sibling comparisons of a maternal exposure and child ADHD are a stronger test than adjusting for measured covariates: here adjustment cut the hazard ratio only from 2.50 to 2.04, while sibling comparison removed it, so many other prenatal-exposure and ADHD cohort associations remain untested for this kind of confounding.
Moderate evidence.
Caveats
Methods lesson drawn from one exposure; whether chemical exposures behave like heritable smoking behavior is unknown; not evidence that any specific chemical association is spurious.
Source: Skoglund 2014, Journal of Child Psychology and Psychiatry Link to this fact
Health effects / Other evidence
Traffic air pollution at school was associated with slower cognitive growth in children, but the paper is observational and does not show that traffic pollution causes lower cognitive ability.
Moderate evidence: Authors write 'demonstrated' and 'strong support', while conceding residual confounding could not be discarded completely.
Source: Sunyer 2015 (BREATHE, Barcelona), PLOS Medicine Link to this fact
Health effects / Other evidence
In 116 newly diagnosed Dutch adults with ADHD, self-rated sensory sensitivity scores were higher than US adult norms (40.9 vs 33.7, P<.001) and so were low-registration (under-responsiveness) scores (36.2 vs 30.3, P<.001); about 19-21% scored in the top band that only 2% of norms occupy.
Moderate evidence.
Caveats
Single-clinic cross-sectional self-report (AASP-NL), 68% response, compared with published US norms rather than a matched control group; no autism-free or psychiatric control group.
Source: Bijlenga 2017, European Psychiatry Link to this fact
Health effects / Other evidence
The sensory differences in these adults with ADHD were concentrated in hearing and activity-level items; taste/smell, vision and touch averaged below 3.0 on every quadrant, so the study does not show broad sensory sensitivity, and gives no evidence of smell or chemical sensitivity.
Moderate evidence: Per-modality means have no norm data; activity-level items overlap ADHD symptoms; 9.5% had impaired hearing; taste and smell are combined in one modality.
Source: Bijlenga 2017, European Psychiatry Link to this fact
Health effects / Other evidence
In this adult ADHD sample, each extra point on the ADHD symptom rating scale went with about 8% higher odds of sensory under-responsiveness (OR 1.08, 95% CI 1.01-1.16) and 13% higher odds of over-responsiveness (OR 1.13, 1.05-1.22).
Moderate evidence.
Caveats
Cross-sectional association within an ADHD clinic sample; self-rated symptoms and self-rated sensory traits from the same questionnaire session; models explain 13-28% of variance (Nagelkerke); correlation does not show that ADHD causes sensory differences.
Source: Bijlenga 2017, European Psychiatry Link to this fact
Health effects / Review of other studies
A US EPA meta-analysis of child ADHD studies found that children with higher blood lead had about 3 to 4 times the odds of ADHD (odds ratio 3.08 to 4.06 depending on whether text or table is read), graded moderate certainty.
Moderate evidence: Crude observational pool with I2 96 to 99%; several comparisons per study; probable publication bias; text and Table 1 numbers differ; lead is not a fragrance ingredient.
Source: Nilsen and Tulve 2020, Environmental Research Link to this fact
Health effects / Review of other studies
Despite its title, this review found no study that tested how chemical exposure, a child's MAOA gene variant and childhood trauma combine to affect ADHD; the authors call that literature sparse.
Moderate evidence: The statement is the authors' own; MAOA candidate-gene studies and the three trauma studies point in conflicting directions.
Source: Nilsen and Tulve 2020, Environmental Research Link to this fact
Health effects / Other evidence
A 2021 meta-analysis of 40 population studies in 30 countries estimated that about 2.6% of adults worldwide have ADHD that began in childhood (persistent ADHD) in 2020, roughly 140 million people, after adjusting to the 2020 age structure.
Moderate evidence: Extreme heterogeneity (I2 98.8%), small-study effects (Egger P = 0.01), mixed DSM versions and tools; a modeled projection, not a count of diagnosed people.
Source: Song 2021, Journal of Global Health Link to this fact
Health effects / Other evidence
The same meta-analysis found that about 6.8% of adults (about 366 million) report current ADHD symptoms regardless of childhood onset (2020, age-adjusted); counting only people with childhood onset gives about 2.6%, so the definition changes the answer by roughly 2.6 times.
Moderate evidence: Symptom-scale screens (ASRS) are not clinical diagnoses; I2 97.6% and publication bias flagged by Egger and Begg tests; the authors did not correct for it.
Source: Song 2021, Journal of Global Health Link to this fact
Health effects / Other evidence
Adult ADHD becomes less common with age in this meta-analysis: persistent ADHD falls from about 5% at ages 18-24 to under 1% at 60 and over, and symptomatic ADHD from about 9% to about 4.5%.
Moderate evidence: Cross-sectional data points from different studies, not people followed over time, so age and birth cohort cannot be separated; older ages are sparsely sampled.
Source: Song 2021, Journal of Global Health Link to this fact
Health effects / Other evidence
A 2022 review that searched PubMed (2000-2022) for studies of blue food dyes (Blue No. 1 and No. 2) and ADHD in children included only four papers, none testing a blue dye alone; the human evidence it relies on is about dye mixtures and restriction diets, and the authors say blue-dye research in children is very limited.
Moderate evidence: Description of what the review found, not evidence on blue dyes; the search was narrow (4 of 20 hits kept) and the review is a low-rigor Cureus paper. Food dyes are not fragrance.
Health effects / Other evidence
In the OEHHA review the dye effect showed up mainly in parent ratings: parent-reported results were positive in 7 of 14 group comparisons, teacher-reported in 1 of 7.
Moderate evidence.
Caveats
Parents were blinded in most crossover trials but rater-dependence limits conclusions. The Fig. 1 parent bar (about 12 of 20) combines group and individual results; the 7 of 14 group figure here matches Table 2 and supplement Table A.3 (6 of the 7 significant). 5 of the review's 16 positive trials rest only on individual-child results.
Health effects / Other evidence
Among US children with current ADHD in 2022, about 3 in 4 (77.9%) had another diagnosed condition, most often behavioral or conduct problems (44.1%), anxiety (39.1%) or learning disability (36.5%); 14.4% had autism.
Moderate evidence: Parent report; co-occurrence only, no direction or cause implied.
Source: Danielson 2024 (NSCH 2022), Journal of Clinical Child & Adolescent Psychology Link to this fact
Health effects / Other evidence
A 2025 systematic review of European children and adolescents with ADHD on methylphenidate found 7 eligible studies (2007-2016) and none of them examined genetic variants linked to sensory differences.
Moderate evidence: Absence of studies within a narrow filter (European only, methylphenidate, under 18, English, 2000-2024); not evidence that no genetic link exists.
Source: Recarey-Rama 2025, Journal of Psychiatric Research Link to this fact
Health effects / Other evidence
The meta-analysis did not test whether the pollution-ADHD associations survive sibling or genetic-confounding designs: none of the included studies used one, and adjustment for parental ADHD and socioeconomic status was inconsistent.
Moderate evidence: Not fragrance-related: noise and outdoor air pollutants. Statement about what the pooled studies lack, from the authors' own limitations section.
Source: Zhang 2026, Frontiers in Psychiatry Link to this fact
Health effects / Other evidence
The Fuertes 2016 paper reports no NO2 association with hyperactivity/inattention (odds ratios 1.02-1.04 per interquartile range, all intervals including 1); a meta-analysis plotting a Fuertes NO2 OR of 1.11 appears to have used the PM2.5 mass 15-year estimate (1.11, 1.01-1.22) instead.
Moderate evidence: Match is by identical point estimate and nearly identical interval (upper bound 1.21 vs 1.22), not confirmed by the review authors. Outdoor air pollution is not fragrance-related.
Source: Fuertes 2016, Environment International Industry-funded Link to this fact
Hidden ingredients / Measured in people or real products
In a test of 6 best-selling US air fresheners and laundry products, researchers found 98 chemical detections (58 different chemicals) released into the air; none appeared on any product label, and only 1 (ethanol) appeared on a safety data sheet.
Moderate evidence: N=6 products, headspace only, semi-quantitative.
Source: Steinemann 2009, Environmental Impact Assessment Review Link to this fact
Hidden ingredients / Measured in people or real products
4 of the 6 products released at least one chemical the Clean Air Act classifies as a hazardous air pollutant (acetaldehyde, chloromethane, or 1,4-dioxane), and none was listed.
Moderate evidence: Paper text says 5 of 6; tables support 4. Two of three HAPs come from the base formulation, not fragrance.
Source: Steinemann 2009, Environmental Impact Assessment Review Link to this fact
Hidden ingredients / Self-reported survey
68.6% of Australians were unaware that fragrance ingredients don't have to be fully listed on labels.
Moderate evidence: Awareness survey item.
Source: Steinemann 2017, Preventive Medicine Reports Link to this fact
Hidden ingredients / Self-reported survey
73.4% of Swedes didn't know fragrance ingredients don't have to be fully disclosed, and 70.9% didn't know that fragranced products sold as 'green' or 'organic' can also release potentially hazardous air pollutants.
Moderate evidence: Awareness items; the HAP premise itself rests on other studies.
Source: Steinemann 2018 (Sweden), Air Quality, Atmosphere & Health Link to this fact
Hidden ingredients / Measured in people or real products
When researchers tested 25 best-selling US laundry, cleaning, personal-care, and air-freshener products, they found 133 different chemicals released into the air, but only 1 of them was listed on any product label.
Moderate evidence: Single headspace sample per product; VOC counts depend on detection threshold.
Source: Steinemann 2011, Environmental Impact Assessment Review Link to this fact
Hidden ingredients / Measured in people or real products
11 of 25 best-selling fragranced products (44%) released at least one chemical the EPA classes as a probable cancer-causing air pollutant, such as acetaldehyde or 1,4-dioxane, though the study couldn't tell whether these came from the fragrance or the rest of the formula.
Moderate evidence: Headspace, not exposure; mostly base-formula compounds.
Source: Steinemann 2011, Environmental Impact Assessment Review Link to this fact
Hidden ingredients / Self-reported survey
About two-thirds (64.6%) of US adults did not know fragrance ingredients don't have to be listed on the label, and more than half would prefer fragrance-free workplaces (53.2% vs 19.7% against) and health care (54.8% vs 22.4%).
Moderate evidence: Opinion survey.
Source: Steinemann 2016, Air Quality, Atmosphere & Health Link to this fact
Hidden ingredients / Measured in people or real products
Counting the word "parfum" undercounts scented children's products, because essential oils and individual fragrance chemicals are listed under other names.
Moderate evidence: Undercount size not measured for these products.
Source: Uber 2024, Jornal de Pediatria (Rio de Janeiro) Link to this fact
Hidden ingredients / Measured in people or real products
In a test of 42 fragranced baby products (shampoos, lotions, oils, sprays), 95% of the chemicals they released into the air were not listed on the label, safety data sheet, or company website.
Moderate evidence: Headspace presence only; Australian products.
Source: Nematollahi 2018, Air Quality, Atmosphere & Health Link to this fact
Hidden ingredients / Measured in people or real products
The best-selling US laundry detergent's label and safety sheet disclosed almost none of what it released; only ethanol was listed.
Moderate evidence: Same detergent measured in 2009, 2011, and 2013 papers; count once.
Source: Steinemann 2013, Air Quality, Atmosphere & Health Link to this fact
Hidden ingredients / Review of other studies
One product's safety data sheet listed twelve ingredients, every one of them only as 'trade secret' at '1-100%'.
Moderate evidence: Single example cited by EPA authors.
Source: Goldsmith 2014, Food and Chemical Toxicology Link to this fact
Hidden ingredients / Review of other studies
In a German test of six air fresheners, more than 90% of the chemicals they released were not declared on the product.
Moderate evidence: Secondary citation of Uhde & Schulz 2015 (not in our evidence base); by number of substances (>75% by concentration).
Source: Steinemann 2017, Building and Environment Link to this fact
Hidden ingredients / Review of other studies
Fragrance turns up in products people don't think of as scented: household cleaners and detergents, tampons and panty liners, medicines, medical devices, toys, and candles.
Moderate evidence: Qualitative list of exposure sources.
Source: Sukakul 2024, Acta Dermato-Venereologica Industry-funded Link to this fact
Hidden ingredients / Self-reported survey
When companies replace a chemical of concern, they often swap in one with far less safety data; one firm found a supplier's 'safer' substitute only looked safer because more of its ingredients were kept confidential.
Moderate evidence: Interview quotes and one anecdote.
Source: Scruggs 2011, Environmental Science & Policy Link to this fact
Hidden ingredients / Review of other studies
Testing cited by a 2010 review claimed the fragrance solvent DEP was found in all fragrances tested and in 71% of cosmetics, but the largest independent survey since (Koniecki 2011, n=252) found DEP in 70% of fragrances and 41% of cosmetics overall — lower than either earlier figure.
Moderate evidence.
Caveats
Secondary citation of EWG's 2002 testing (n=72), relayed via Crinnion 2010. Two larger, later, independent surveys (Hubinger & Havery 2006, n=48; Koniecki 2011, n=252) confirm DEP is common and fragrance-concentrated but do not reproduce the specific 100%/71% figures; detection frequency falls as sample size rises across all three surveys. Treat 'all'/'71%' as an upper bound from small early samples, not a population estimate.
Source: Crinnion 2010, Alternative Medicine Review Link to this fact
Hidden ingredients / Review of other studies
FDA's voluntary cosmetic registration listed benzyl alcohol in only 322 formulations in 1998 and 511 products in 2006, far below its real use in moisturizers, shampoos, baby products, and deodorants, which the authors call widespread under-reporting.
Moderate evidence.
Caveats
The 322-formulation count for 1998 is now confirmed against the primary source: CIR's 2001 safety report (Nair 2001) reproduces the identical FDA VCRP count in its own Table 2, with a full 34-category breakdown (largest use hair dyes and colors, 130/1572; baby shampoos 5/21; baby lotions 3/53) — this is the exact table Jacob & Barron's figure traces to, not a secondhand citation. The 511-products-in-2006 figure remains an unpublished personal communication (postdates the 2001 report, so it cannot confirm or deny it). The same 2001 report separately documents that FDA stopped collecting concentration-of-use data after 1992, an independently CIR-acknowledged registry gap that supports treating VCRP counts as an incomplete picture of real use, distinct from the authors' own 'under-reporting' framing.
Hidden ingredients / Review of other studies
The glycol-ether solvent propylene glycol methyl ether (PGME), mostly used in industrial coatings and cleaners, is officially listed as a possible 'fragrance ingredient' in cosmetics, and a Swiss survey found it making up 30–50% of one perfume.
Moderate evidence: INCI function listing plus one Swiss product survey (Dentan 2000, cited secondhand); US registry shows only nail-product uses.
Source: CIR 2008, International Journal of Toxicology Industry-funded Link to this fact
Hidden ingredients / Journalism
Laboratory testing commissioned by US campaigners found 38 chemicals not listed on the label across 17 popular fragrances, 14 per product on average, including hormone-disrupting synthetic musks.
Moderate evidence: NGO report, not peer reviewed; numbers now verified against the primary (Sarantis 2010, in our evidence base); one sample per product, tentative library IDs, no concentrations.
Source: Sealey 2016 (citing Sarantis 2010, Not So Sexy), Environment International Link to this fact
Hidden ingredients / Review of other studies
The single word 'fragrance' on a label can stand for dozens to hundreds of chemicals. The fragrance industry's own inventory (RIFM) lists about 2,600 chemicals in current active use, plus roughly 3,500 discontinued materials it keeps only for internal read-across comparison.
Moderate evidence.
Caveats
Primary now read directly (Date 2020, RIFM): ~2,600 active-use + ~3,500 discontinued = the 'over 6,000' figure in its abstract. Rádis-Baptista 2023 paraphrased this as '6,000 known, 2,000 used,' which is not what Date 2020 reports. This is the industry's own inventory count, corroborated only within RIFM's own publications, not independently, so it stays at moderate rather than strong.
Source: Rádis-Baptista 2023, Journal of Xenobiotics Link to this fact
Information, not medical advice. See also: myths we won’t tell you.