Glossary
Endocrine-disrupting chemicals
Nineteen studies pooled together point in one direction, and the signal in them is real enough to be worth having. Every one of those studies watched people rather than testing anything on them. That single fact is the limit on what any of it can say, and it's most of what this page is about.
What the research found.
One review pooled 19 observational studies on endocrine-disrupting chemicals and autoimmune disease. Bisphenols showed the strongest link, at an odds ratio of 2.38 whose true value sits from 1.27 to 4.45. Three other classes were linked too, and those were organochlorine pesticides, phthalates, and the compounds in soot known as PAHs.
That review also found no link at all for one whole class of chemical. The industrial oils known as PCBs showed nothing, while the other four classes did. That negative result sits inside the same pooled analysis as the positive ones, which is what makes it worth quoting.
One study measured these chemicals directly in 106 people with autoimmune encephalitis and in 119 people without it. Two preservatives found in blood were linked to higher risk, at odds ratios of 1.69 from 1.25 to 2.27 and 1.51 from 1.14 to 2.00. One measured in spinal fluid was linked at 1.54, from 1.30 to 1.84.
What these chemicals are
Endocrine-disrupting chemicals are man-made substances that get in the way of hormone signals in the body. The name describes what they do rather than what they are. Hormones work by fitting a receptor, rather as a key fits a lock, and these chemicals look enough like a hormone to interfere with that.
Some of them block the lock and some turn it when they shouldn't. They turn up in plastics, pesticides, preservatives, and industrial goods, which is a wide enough spread that nobody avoids all of them. They're measured in blood and in urine, and one study on this page measured them in spinal fluid as well.
What the pooled evidence shows
One review gathered 19 studies on these chemicals and autoimmune disease, all of which watched people rather than testing anything, and pooled what they found. Four classes came out linked to autoimmune disease risk. Bisphenols showed the strongest link, at an odds ratio of 2.38 with the true value from 1.27 to 4.45.
Three others were linked too, and those were organochlorine pesticides, phthalates, and the compounds in soot and smoke known as PAHs. One class showed nothing whatsoever, and those are the industrial oils known as PCBs. They had no link at all in the same pooled analysis.
That negative result is worth noticing rather than skipping over. A review where every exposure comes out positive is usually telling you something about its own methods. A real negative sitting inside the same analysis is a point in the review's favor.
What observational means here
Every one of those 19 studies watched what happened rather than testing anything, and that single fact sets the limit on this whole page. The difference counts for more here than it does in most subjects. Think about how a study like this runs in practice.
People who already have an autoimmune disease have their chemical levels measured, and those get compared against the levels in people who don't have it. So a link between the two has at least two explanations sitting behind it. The chemical may have contributed to the illness, or having the illness may have changed what somebody was exposed to and how much of it their body held onto.
A study that only watches has no way to separate those two explanations. That isn't a criticism of those studies, which were doing the job they set out to do and doing it well. It's a description of what any study built that way can and cannot settle.
What the authors themselves say
This is the part worth quoting directly, because it's more careful than the way these findings usually get repeated in public. The review calls exposure a possible risk factor and no more than that. It says the pathways still need working out, which means nobody has yet shown how these chemicals would produce an autoimmune disease.
It also asks for two kinds of study that nobody has done for this question. One would follow people forward from before they became ill, watching over the years to see which of them develops a disease and which doesn't. The other would test rather than watch, and neither of those exists yet.
When a review's own authors write that, the question is open rather than answered. This page isn't being more careful than the evidence supports, whatever it may look like from the amount of hedging. It's passing on the care that came with the evidence in the first place.
The brain study
One other study on file measured these chemicals directly, where most studies estimate them instead, and that's what earns it a section here. It compared 106 people with autoimmune encephalitis against 119 people without it. Three groups of chemical came up in the largest amounts, which were antioxidants, phthalate breakdown products, and organophosphate esters.
Two preservatives found in blood were linked to higher risk. Ethyl paraben came in at an odds ratio of 1.69, from 1.25 to 2.27, and butyl paraben came in at 1.51, from 1.14 to 2.00. One measured in spinal fluid was linked as well, and methyl paraben came in at 1.54, from 1.30 to 1.84.
That study also measured how much of each chemical crossed from blood into spinal fluid. The share ran from 0.040 percent for one preservative and reached 3.808 percent for another, so most of what sits in the blood stays there. Hold all of it loosely, because it's one study of one uncommon illness that isn't even a rheumatic disease, and the warning about watching rather than testing applies to it too.
The three limits
Three limits hold across everything above, and they happen to sit where most people want answers. Saying them plainly beats leaving them to be inferred from the hedging. Each one is a description of what this kind of evidence is, rather than a complaint about what it isn't.
The first is that every study here was built to find associations rather than causes, and the review's authors say so themselves. The second is that an odds ratio describes a difference between two groups in a population, so it isn't a number that belongs to any one person. Yours isn't a group, and no figure on this page was ever meant to be read as yours.
The third is the one people ask about most. Not one of these studies gave anybody an intervention, and none of them watched what followed. The study that would answer whether cutting exposure changes anything hasn't been done, so the answer isn't no. There simply isn't one yet.
What it does support
There's a signal here worth studying properly, and it differs by chemical class rather than running evenly across all of them. Those two things together are the finding. Four classes showed a link and one didn't, which is a more interesting result than five out of five would have been.
The strongest was bisphenols, at 2.38, with a range running from 1.27 to 4.45. That's broad enough that the size is loosely pinned even where the direction isn't. So hold the direction and let go of the size, which is the honest way to read a pooled odds ratio of that shape.
If somebody tells you this question is settled, they've read further into it than the people who did the review did. That's worth remembering wherever you meet this subject, because you will meet it. The signal is real and the conclusion isn't drawn yet, and both halves of that sentence are doing work.
Common misconceptions.
Myth. These chemicals are proven to cause autoimmune disease.
Reality. Nothing gathered here shows cause, and the review that pools this evidence covers 19 studies that all watched rather than tested. Its authors call exposure a possible risk factor, they say nobody has worked out how it would happen, and they ask for studies that haven't been done. That's careful wording on their part, and it's the right wording for what they found.
Myth. All of them are linked.
Reality. One class in that review showed nothing whatsoever. The industrial oils known as PCBs had no link at all, while four other classes did. That negative result counts for something, because a review where every exposure comes out positive is usually telling you about its own methods rather than about the world.
Myth. The odds ratios tell me my personal risk.
Reality. They don't, because an odds ratio of 2.38 describes a difference between two groups rather than a chance that applies to one person. Its range runs from 1.27 to 4.45, which is broad enough to notice. Most of these studies also compared people who already had the disease against people who didn't, so the exposure was usually measured afterward.
Myth. Avoiding these chemicals will prevent autoimmune disease.
Reality. No study gathered here tested that question at all. All 19 pooled studies looked at exposure that had already happened, none of them gave anybody an intervention, and none followed what happened next. So there's a link worth studying properly, and there's no evidence in either direction about whether changing anything changes anything.
Myth. This is settled science.
Reality. The review's own conclusion says otherwise, in plainer words than most people use about it. It calls exposure a possible risk factor, it says nobody has worked out how these chemicals would do it, and it asks for two kinds of study that haven't been done. When a review's own authors write that, the question is open rather than answered.
Related terms.
- Environment and autoimmune disease, for the wider set of exposures studied.
- Rheumatoid arthritis and the environment, where the exposure evidence is strongest.
- Lupus and the environment, for the silica and sunlight evidence.
Questions patients ask.
What are endocrine-disrupting chemicals?
Man-made substances that get in the way of hormone signals inside the body. Hormones work by fitting a receptor, rather like a key in a lock, and these chemicals look enough like a hormone to get in the way of that. They turn up in plastics, pesticides, preservatives, and industrial goods, and they're measured in blood and in urine.
Which ones are linked to autoimmune disease?
One review pooled 19 studies and found four classes linked, which were bisphenols, organochlorine pesticides, phthalates, and the compounds in soot known as PAHs. Bisphenols showed the strongest link of the four, at an odds ratio of 2.38 from 1.27 to 4.45. One class showed nothing at all, and that was the industrial oils known as PCBs.
How strong is that evidence?
Weaker than the numbers make it look, and the review's own authors are the ones who say so. All 19 pooled studies watched rather than tested, which is the limit on every figure here. The authors call exposure a possible risk factor, they say nobody has worked out how it would happen, and they ask for two kinds of study that haven't been done.
Why does observational count for so much here?
Because the exposure is usually measured after somebody already has the disease, which leaves two ways to read any link. The chemical may have helped cause the illness, or having the illness may have changed what somebody met and how much of it stayed in them. A study that only watches has no way to tell those two apart.
Should I change what I buy?
Nothing on this page speaks to that, and no study gathered here speaks to it either. None of the 19 pooled studies gave anybody an intervention, and none of them watched what followed afterward. So there's no evidence in either direction about whether cutting exposure changes anything, and what the evidence does support is that this is worth studying properly.
What about PFAS?
One study on file links these to rheumatoid arthritis risk, using a sample meant to stand for the whole country. Our record of that study holds no effect size, no range, and no sample size. So this site can say the study exists and it can't tell you what the study found in numbers, which is a fair description of the whole record.
What was found in the brain study?
One study measured these chemicals in 106 people with autoimmune encephalitis and in 119 people without it. Two preservatives found in blood were linked to higher risk, at odds ratios of 1.69 and 1.51, and one measured in spinal fluid was linked at 1.54. That's one study of one uncommon illness, so it's worth holding loosely.
What should I take from all this?
Three things, and they fit together. There's a real signal in the pooled data and it's worth studying properly, though it comes from studies that only watched people, so it can't show cause, and the review's own authors say as much. Nobody has tested whether changing your exposure changes your risk, which means there's no answer to give on that one yet.
References.
- Xu Y; Xu J; Li S et al. Association between environmental endocrine disruptors and autoimmune diseases: A systematic review and meta-analysis. iScience. 2025;28:113822. 10.1016/j.isci.2025.113822Systematic review and meta-analysis of observational studies published from January 2008 onward examining the association between exposure to environmental endocrine disruptors and autoimmune disease outcomes
- Huang S; Xiao L; Luo M et al. Screening for endocrine-disrupting chemicals in the serum and cerebrospinal fluid of patients with autoimmune encephalitis. Ecotoxicol Environ Saf. 2025;299:118399. 10.1016/j.ecoenv.2025.118399Case-control study of 106 patients with autoimmune encephalitis and 119 without
- Lv Y; Zhao C; Xiang Y et al. Perfluoroalkyl Substance (PFAS) Mixtures Drive Rheumatoid Arthritis Risk Through Immunosuppression: Integrating Epidemiology and Mechanistic Evidence. International journal of molecular sciences. 2025;26. 10.3390/ijms26157518Cross-sectional
This page gathers the published research on this subject into one place. The studies behind it were published between 1989 and 2026, and every figure links to the paper it came from. Those studies were peer reviewed. This summary of them was not. Dr. Sarah Luebker is reviewing these pages one at a time and has not reached this one yet, so it carries no medical review date and nothing here is her opinion or her advice to you. Each page gets updated as she reaches it. It is here in the meantime because the science is worth having in one organized place that is easy to find and easy to read. Talk to your own clinician before acting on any of it.