Oxybenzone, an organic ultraviolet (UV) filter, is absorbed in small amounts and concern has been raised about its weak estrogenicity. It should be noted that studies have shown that oxybenzone is between 1/1000 and 1/1,000,000 as estrogenic as naturally-occurring estradiol and is far less estrogenic than compounds found in foods such as soy or chick peas. It is for this reason that major scientific organizations that have reviewed this issue have concluded that this compound does not pose a health risk at current exposure levels.
A few studies have looked for correlations between oxybenzone levels and a number of health outcomes. Of these studies, two have reported positive associations; one linking oxybenzone to birth weight, and another linking it to a diagnosis of endometriosis. Neither study is powered to show that oxybenzone causes either outcome, and significant flaws in the design and interpretation of both studies should cause one to question their conclusions. My reasons for this assessment are numerous, and so I have chosen to outline them in a separate post, so that those who are interested in this more detailed, nuanced explanation can access it.
Both studies are primarily limited by the fact that neither study was specifically designed to examine oxybenzone exposure, and as a result they did not control for important confounding variables. Both studies were designed to examine broader risk factors, either for birth outcomes or endometriosis, and the decision to measure oxybenzone levels in the urine came later.
In both cases, oxybenzone levels were measured on only one occasion. Oxybenzone (also known as benzophenone-3) is absorbed through the skin in small amounts, and is excreted in the urine. Its half-life is approximately 16.5 hours, meaning that oxybenzone is completely eliminated in 3-4 days, and a measurement of this ultraviolet filter in the urine reflects exposure to this compound that is very recent. It would be hard to imagine that short-term exposures would be relevant to the health outcomes described, and yet the authors of the study did not take multiple measurements, nor did they take a history of sunscreen use, which would have been crucial in establishing exposure patterns.
The study by Wolff et al. measured 19 metabolites in the urine of pregnant women and attempted to link them to a number of birth outcomes. Such study design is rife with room for error; if enough variables and outcomes are examined in this fashion, some will correlate purely by chance. Oxybenzone levels were, in fact, not linked to any of the study's outcomes, and it was only when the authors looked at sex-specific interactions that they found a link to lower birth weight in girls and higher birth weight in boys. The results themselves are conflicting and inconclusive, and the authors themselves note that they "can not be explained". Additionally, if the relationship were truly significant, one would expect that in addition to birth weight, oxybenzone levels would correlate with head circumference, as the two are often closely correlated. In fact, this was not the case; no relationship was found between oxybenzone levels and head circumference.
The study linking oxybenzone to endometriosis was also not designed to evaluate oxybenzone exposure and as such did not account for important confounding variables. Oxyenzone levels fluctuate with the seasons (people use more sunscreen in the summers), and, while the authors noted that levels were indeed higher in the summer, they failed to control for this in their study. While they matched their study and control groups based on demographic features, they did not control for the time of year at which their urine was sampled. In other words, we have no information on whether the control group was more likely to be sampled in the fall/winter, which would have significantly skewed the results. The authors also did not control for vitamin D levels, which several studies have shown are higher in those with endometriosis. One could postulate that, in this case, oxybenzone exposure is a surrogate marker for more sun exposure and higher vitamin D levels (a couple of studies have shown a trend toward higher vitamin D levels in sunscreen users). While this is purely speculation, the point is that the authors of the study did not control for important variables that one could reasonably argue may have influenced the results.
A link between two variables does not prove cause and effect. Both of these studies are limited by weak design and flawed analysis, and neither can be used to conclude that oxybenzone causes human harm. While they may stimulate further study, a thorough analysis of these papers yields shortcomings that make their findings questionable. In order to truly establish whether these relationships exist, well-designed studies that control for known confounding variables would be needed.