In depth
Rheumatoid arthritis and environment
With one exposure here, risk climbs as exposure climbs, and cutting it costs nothing. That exposure is smoking, and the evidence behind it is stronger than anything else on this page by a considerable distance. The rest range from a real workplace risk to a question the studies still disagree about.
What the research found.
One meta-analysis pooled 10 studies and 4552 people with rheumatoid arthritis. Against never smoking, risk rose 26 percent at 1 to 10 pack-years, at a relative risk of 1.26 from 1.14 to 1.39. It roughly doubled at 21 to 30 pack-years, at 1.94 from 1.65 to 2.27, so the gradient runs across the whole range.
Di Giuseppe and colleagues, Arthritis Research and Therapy, 2014
One Swedish population study covered 11285 incident cases and 115249 matched controls. Silica-exposed men had an odds ratio of 1.4 for antibody-positive disease, from 1.2 to 1.6. Risk climbed with the years a man was exposed, reaching 2.3 from 1.4 to 3.8, at a trend P under 0.0001.
One review pooled 31 studies on inhaled exposures at work, and seven of the exposure types came out linked to rheumatoid arthritis. They were silica, asbestos, solvents, pesticides, fertilizers, animal dust, and engine exhaust. The relative risks across those seven ranged from 1.25 to 1.49, which is a modest range and a consistent one.
One genetic study tested whether air pollution causes rheumatoid arthritis and found nothing for any of four pollutants. Fine particulate matter came in at an odds ratio of 0.71, from 0.27 to 1.91. Nitrogen dioxide came in at 0.482, from 0.173 to 1.343, and both intervals are wide enough to contain a doubling of risk.
In a large US sample, PFAS were linked to rheumatoid arthritis. PFOA came in at an odds ratio of 1.63, PFOS came in at 1.41, and PFNA came in at 1.40. The mixture effect, which estimates what real combined exposure does, came out at 1.06, from 1.02 to 1.10.
Lv and colleagues, International Journal of Molecular Sciences, 2025
Start with the one that has a dose-response
Smoking has the strongest evidence of any exposure in this disease, and the shape of that evidence is what makes it strong rather than the size of it. It isn't one number on its own, which is the usual situation on an environment page. It's a gradient, and gradients are harder to produce by accident than single associations are.
One review pooled 3 forward-looking cohorts and 7 case-control studies, covering 4552 people with rheumatoid arthritis between them. It asked how risk changed with lifetime smoking, counted in pack-years. That question is the one worth asking, because a link that grows with the dose is much harder to explain away.
Against never smoking, risk rose 26 percent at 1 to 10 pack-years, at a relative risk of 1.26 with the true value from 1.14 to 1.39. At 21 to 30 pack-years it roughly doubled, at 1.94 from 1.65 to 2.27. So the gradient runs across the whole range the review examined.
Two details make that more convincing than a single number would have been on its own. The studies didn't disagree with each other, at a heterogeneity P of 0.32, which is unusual in a literature where disagreement is the norm. Publication bias didn't affect the result either, so the finding isn't an artifact of which studies got published.
Two things that figure does and doesn't say
It says the lightest lifetime exposure studied was already associated with raised risk, which is the part people miss. The authors put it plainly in their own conclusion. Lifelong smoking was linked to risk even among smokers with a low lifelong exposure.
It also says risk stopped climbing past a point, and above 40 pack-years the relative risk was 2.07, from 1.15 to 3.73. That's no higher than the 21 to 30 band, and the plateau could be real. It could also reflect how few people are in that band and how roughly they recall their own history, and the interval ranging from 1.15 to 3.73 hints at the second.
Read that plateau carefully, because it describes who develops this disease and nothing else. It says nothing about heart disease, lung disease, or cancer. In lupus, current smokers had 1.42 times the risk of cardiovascular disease, from 1.21 to 1.66, which is a different illness and a different question.
One more split in that review is worth knowing about. For rheumatoid factor positive disease the link was 2.47, from 2.02 to 3.02, and for rheumatoid factor negative disease it was 1.58, from 1.15 to 2.18. That's a large difference between the two, and it comes up again when you get to silica.
Other people's smoke
One review pooled six studies on passive smoking, of which three were case-control studies and three were cohort studies. They agreed with each other fairly well, at an I2 of 34.9 percent. That agreement is the strongest thing about the review, and the pooled risk estimate it produced is a good deal weaker than the active smoking evidence.
Passive smokers had 12 percent higher risk than people with no exposure at all, and adult exposure came out at 12 percent too. Exposure during childhood came out higher, at 34 percent. That childhood figure is the one most often repeated, and it deserves the caveats in the next paragraph.
Hold all of that more loosely than the active smoking figures, for two reasons. The review reports no confidence intervals at all, so the precision of those percentages is simply unknown. It also found no real rise with more years of exposure, which is the very thing that made the active smoking evidence convincing.
Work
This is the part of the environment question most often left out, and it has real numbers behind it. Two of those numbers are worth keeping in mind. The first comes from a pooled review of inhaled exposures, and the second from a Swedish population study of silica.
One review pooled 31 studies on inhaled exposures at work, of which 25 entered meta-analyses across 10 exposure types. Seven of those types came out linked to rheumatoid arthritis. They were silica, asbestos, solvents, pesticides, fertilizers, animal dust, and engine exhaust, and the relative risks ranged from 1.25 to 1.49.
Silica is the best studied of the seven, and the clearest evidence for it comes from Sweden. A population study covered 11285 incident cases and 115249 matched controls, adjusted for age, sex, county, year, alcohol, and smoking. Silica-exposed men had an odds ratio of 1.4 for antibody-positive disease, from 1.2 to 1.6, and risk climbed with years exposed to reach 2.3 from 1.4 to 3.8, at a trend P under 0.0001.
A second Swedish study, of 31139 cases, found something worth noticing about the antibodies. Silica-exposed men had 1.22 for antibody-positive disease and 1.23 for antibody-negative disease. Those two are equal, and that counts, because smoking raises antibody-positive risk far more than antibody-negative risk, so silica isn't simply tracking smoking here.
A French survey came at the question from the other side, by measuring lifetime silica exposure in 97 rheumatoid arthritis patients. It compared them against controls drawn from a general population panel of 2911 people, with the matching covering age, sex, and tobacco. Patients had higher exposure, at P under 0.0001. The difference came from work rather than from home or hobbies, which is the point of doing the survey that way.
Two limits apply to all of that, and the first is that exposure in the register studies is assigned by job title rather than measured. That blurs it, because two people whose job titles match can have very different exposures. The second is that these findings are reported for men, and one study notes that women had lower odds ratios along with lower duration and intensity of exposure. So a woman reading this shouldn't conclude the exposure is irrelevant to her, and she shouldn't conclude it's been shown to apply either.
Air pollution, where the evidence argues with itself
Eight studies here found raised risk and one study built to test cause found nothing at all. That disagreement is the finding, and this page won't resolve it for you. What it can do is lay both sides out with their intervals attached, which is the only honest way to present an unsettled question.
One Californian study followed 42152 adults over 65, and fine particulate matter came in at a hazard ratio of 1.20 per 2 micrograms per cubic meter, from 1.16 to 1.23. Nitrogen dioxide came in at 1.44 per 10 micrograms, from 1.35 to 1.52. A UK Biobank study of 342973 people found fine particulate matter at 1.07, from 1.01 to 1.13, with three of its five pollutants showing nothing at all. A Taiwanese study of 322301 people found nitrogen dioxide at 1.54, from 1.45 to 1.64, and its confidence intervals are tight enough to be worth questioning.
One genetic study tested whether air pollution causes the disease rather than merely tracking it. Fine particulate matter came in at an odds ratio of 0.71, from 0.27 to 1.91, and PM10 came in at 0.657, from 0.374 to 1.154. Nitrogen dioxide came in at 0.482, from 0.173 to 1.343. Nitrogen oxides came in at 0.868, from 0.317 to 2.373, and not one of the four reached significance, while sensitivity checks found no sign of the two problems that most often undo this method.
Read those intervals before concluding anything, because every one of them is wide enough to contain both a halving and a doubling of risk. That's an absence of evidence rather than evidence of absence, and genetic instruments for where somebody lives are weak instruments. So the honest position is that this is open. The cohorts consistently find something, the design built to test cause doesn't, and anybody telling you it's settled hasn't read both halves.
PFAS, and the number that circulates
One 2025 study looked at a large US sample and at a group of chemicals called PFAS. Single chemicals showed clear links, with PFOA at an odds ratio of 1.63, PFOS at 1.41, and PFNA at 1.40. Those are the numbers that get repeated everywhere PFAS come up.
That study produced another number, and it's the one that hardly ever gets quoted with the first three. The mixture effect, which estimates what combined exposure does, came out at 1.06, with the true value between 1.02 and 1.10. Combined exposure is what people really have, because these chemicals turn up together.
Look at the distance between 1.63 and 1.06, because that's the difference between one chemical on its own and the mixture people really live with. Both figures are real and both come out of one study. Quoting the first as the risk from PFAS describes a situation real people aren't in, which is why this page gives you both.
The organic food study, read with care
One study gets quoted as proof that organic food lowers inflammation, and what it found is weaker than the retelling. It's worth walking through, because the way it weakens is instructive. The effect shrinks each time the analysis accounts for something else about how people eat and live.
Eating organic was associated with lower C-reactive protein, at a beta of negative 0.096, with the true value from negative 0.159 to negative 0.033. Then the study accounted for how well people ate overall, using a Mediterranean diet score. The link shrank to negative 0.080 once that adjustment went in.
A second marker did worse than that once the adjustment went in. Cystatin C went from a significant negative 0.033 to a non-significant negative 0.019, which means it lost its link entirely. A finding that disappears when you account for diet quality was probably about diet quality.
Two things about that study count for more than either beta does. Its authors say they can't rule out that healthy eating and income explain the rest of the association. It also measured no rheumatic outcome at all, so it's a study about one inflammation marker in the general population rather than a study about rheumatoid arthritis.
How to weigh all this
One umbrella review helps set the scale, because it graded 42 links between environmental exposures and rheumatic diseases. Only nine of the 42 reached convincing or highly suggestive evidence. Thirty-two had studies that disagreed sharply with each other, and twelve showed small-study effects or excess significance bias.
Smoking and rheumatoid arthritis was among the nine that survived, which is the company that exposure keeps. Very little else on this page is in it. That ranking is worth holding onto when you decide where to spend your attention and your money.
So the order on this page reflects the strength of the evidence rather than how much attention each exposure gets. Smoking comes first, because risk climbs with exposure there and an umbrella review backs it. Work comes next with pooled estimates and risk climbing with years exposed in men, then air pollution while it stays unsettled, and PFAS last because the effect is real and small at the exposures people really live with.
Not one of these studies changed an exposure and watched what happened, because every one measured what people already had. So not one of them can tell you what cutting something would do, which is a different statement from saying it wouldn't help. The comparison worth holding is with treatment, where the trials that lifted remission from 16 to 65 percent tested drug strategies against each other in people who already had the disease. Both kinds of evidence are worth having, and they answer different questions.
Common misconceptions.
Myth. Only heavy smoking counts.
Reality. The lowest band studied was already associated with raised risk, which is the part that surprises people. One meta-analysis pooled 10 studies and 4552 cases, and at 1 to 10 pack-years risk rose 26 percent, at a relative risk of 1.26 from 1.14 to 1.39. That's a light lifetime exposure, and the authors' own conclusion says lifelong smoking was linked to risk even among smokers with a low lifelong exposure.
Myth. If I've smoked for decades there's no point stopping.
Reality. That isn't what these figures say, and this page can't advise you either way. Risk stopped climbing above 20 pack-years in that review, and more than 40 pack-years came in at 2.07, which is no higher than the 21 to 30 band. That's a finding about who develops the disease rather than a statement about anything else. Smoking also raises cardiovascular risk, and in lupus current smokers had 1.42 times the risk of heart disease.
Myth. Air pollution is a proven cause here.
Reality. Eight cohort studies here found raised risk, with hazard ratios mostly falling between 1.07 and 1.54. One genetic analysis was built to test cause, and it found nothing for any of four pollutants, with every interval crossing 1. Those intervals are wide enough to be an absence of evidence rather than evidence of absence. So the question is open, and this page says so rather than picking a side.
Myth. PFAS raise the risk by more than 60 percent.
Reality. That's the single-chemical figure for PFOA, at 1.63, and the mixture effect came out at 1.06, from 1.02 to 1.10. The mixture effect estimates what combined exposure does, and combined is what people really have. So quoting 1.63 as the risk from PFAS describes a situation real people aren't in.
Myth. Workplace exposure is an old problem.
Reality. The evidence is recent and it's specific, from a review that pooled 31 studies on inhaled exposures at work. Seven of those exposures came out linked to rheumatoid arthritis. One Swedish study of 11285 cases found silica-exposed men at 1.4 for antibody-positive disease, rising to 2.3 with years exposed, and the pooled relative risks across the seven ranged from 1.25 to 1.49.
Cautions specific to this condition.
- Smoking has the strongest evidence of any exposure here, and it's the only one that costs nothing to cut. Raise it with your team if it applies to you.
- Read a single-chemical odds ratio and a mixture effect as two different things. The first is one chemical alone. The second estimates the combined exposure you really have.
- Tell your team what you do or did for work. Silica, asbestos, solvents, pesticides, fertilizers, animal dust, and engine exhaust all appear in the pooled evidence.
- Treat a link as a reason to look further, not a reason to spend money. Every cohort study here measured exposure. Not one of them assigned it.
- Ask what a study really measured. The organic food study measured C-reactive protein and cystatin C. It measured no rheumatic outcome at all.
- Don't let worry about chemicals crowd out your treatment. The trials that lifted remission from 16 to 65 percent tested drug strategies, not exposures.
Discuss any change with the rheumatologist who manages your care. Nothing here replaces that conversation.
Questions patients ask.
Which exposure has the best evidence?
Smoking has the best evidence of any exposure here, by a distance. One review pooled 10 studies covering 4552 cases, and risk rose 26 percent at 1 to 10 pack-years and roughly doubled at 21 to 30. The studies agreed with each other, and no publication bias affected the result either. A dose-response like that is one of the better signs that a link is real rather than incidental.
Does the amount smoked matter?
Up to a point, and then it stops climbing. That review found 1.26 at 1 to 10 pack-years, 1.94 at 21 to 30, and 2.07 above 40, which is no higher than the band below it. The link was stronger for antibody-positive disease, at 2.47 from 2.02 to 3.02. For antibody-negative disease it was 1.58, from 1.15 to 2.18, which is a real difference between the two.
What about other people's smoke?
One review pooled six studies on passive smoking, and passive smokers had 12 percent higher risk than people with no exposure. Childhood exposure came in higher still, at 34 percent. That review found no real rise with more years of exposure, so risk never climbed with the dose the way it does with active smoking. It also reports no confidence intervals at all, so hold those percentages loosely.
Does my job matter?
It may, and one review pooled 31 studies on inhaled workplace exposures. Seven of them came out linked to rheumatoid arthritis, and the relative risks across those seven ranged from 1.25 to 1.49. Those seven were silica, asbestos, solvents, pesticides, fertilizers, animal dust, and engine exhaust. Telling your team what you do or did for work is a fair thing to raise at a visit.
How strong is the silica evidence?
Strongest of the workplace exposures, and it's mostly about men. One Swedish study covered 11285 cases, and silica-exposed men came in at 1.4 for antibody-positive disease, from 1.2 to 1.6, climbing to 2.3 with years exposed. A second Swedish study found 1.22 for antibody-positive and 1.23 for antibody-negative disease. Those two being equal counts, because smoking affects antibody-positive and antibody-negative disease very differently.
Does air pollution cause rheumatoid arthritis?
The studies disagree and this page won't pick a side for you. Eight cohort studies here found raised risk, with most of them falling between 1.07 and 1.54. One genetic study was built to test cause, and it found nothing for any of four pollutants, with every interval crossing 1. Those intervals are wide, so it's an absence of evidence rather than evidence of absence.
Should I be avoiding PFAS?
Wanting less of a chemical that never breaks down is reasonable, and that holds for reasons nothing to do with this disease. No study has tested whether cutting PFAS changes any rheumatic outcome. So this page can tell you that the link is real, that it's small at everyday combined exposure, and that it comes from watching people rather than from testing anything.
Is there anything I can change that has been tested?
Not in the sense of a trial, because every study on this page measured an exposure rather than changing one. So not one of them can tell you what happens if you cut something. That's different from saying the environment doesn't count, and it means the studies that would answer the question haven't been done.
References.
- Di Giuseppe D; Discacciati A; Orsini N et al. Cigarette smoking and risk of rheumatoid arthritis: a dose-response meta-analysis. Arthritis Res Ther. 2014;16:R61. 10.1186/ar4498Dose-response random-effects meta-regression of 3 prospective cohorts and 7 case-control studies
- Ilar A; Klareskog L; Saevarsdottir S et al. Occupational exposure to asbestos and silica and risk of developing rheumatoid arthritis: findings from a Swedish population-based case-control study. RMD Open. 2019;5:e000978. 10.1136/rmdopen-2019-000978Swedish population-based case-control study
- Liu Q; Song X; Mauro E et al. Exposure to Occupational Inhalants and the Risk of Developing Rheumatoid Arthritis: A Systematic Review and Meta-Analysis. Arthritis Rheumatol. 2026;78:830-847. 10.1002/art.43446Systematic review and meta-analysis following PRISMA
- Yang C; Du Z; Ma J et al. Causal relationship between air pollution and rheumatoid arthritis: A two-sample Mendelian randomization study. Medicine (Baltimore). 2025;104:e42901. 10.1097/MD.0000000000042901Two-sample Mendelian randomisation using genome-wide association data for four air pollution metrics as exposures and European rheumatoid arthritis cohorts as outcomes
- 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
- Zhang X; Zhang X; Yang Y et al. Association between passive smoking and the risk of rheumatoid arthritis: a systematic review and meta-analysis. Clin Rheumatol. 2023;42:663-672. 10.1007/s10067-022-06433-3Systematic review and meta-analysis of three case-control and three cohort studies on passive smoking and rheumatoid arthritis
- Wrangel O; Graff P; Bryngelsson IL et al. Silica Dust Exposure Increases Risk for Rheumatoid Arthritis: A Swedish National Registry Case-Control Study. J Occup Environ Med. 2021;63:951-955. 10.1097/JOM.0000000000002281Nationwide registry case-control study of all rheumatoid arthritis cases in Sweden between 2005 and 2016: 31
- Cavalin C; Lescoat A; Sigaux J et al. Crystalline silica exposure in patients with rheumatoid arthritis and systemic sclerosis: a nationwide cross-sectional survey. Rheumatology (Oxford). 2023;62:2707-2715. 10.1093/rheumatology/keac675Nationwide cross-sectional survey using a purpose-built and validated lifetime exposure questionnaire covering occupational and non-occupational crystalline silica exposure
- Wu AH; Tseng C; Park SY et al. Late-Onset Rheumatoid Arthritis and Air Pollution in a Multiethnic Cohort. Arthritis Rheumatol. 2026;78:1425-1436. 10.1002/art.70054Prospective analysis of 42
- Zhang J; Fang XY; Wu J et al. Association of Combined Exposure to Ambient Air Pollutants, Genetic Risk, and Incident Rheumatoid Arthritis: A Prospective Cohort Study in the UK Biobank. Environ Health Perspect. 2023;131:37008. 10.1289/EHP10710Prospective cohort study of 342
- Jung CR; Hsieh HY; Hwang BF. Air Pollution as a Potential Determinant of Rheumatoid Arthritis: A Population-based Cohort Study in Taiwan. Epidemiology. 2017;28 Suppl 1:S54-S59. 10.1097/EDE.0000000000000732Retrospective cohort study of 322
- Belbasis L; Dosis V; Evangelou E. Elucidating the environmental risk factors for rheumatic diseases: An umbrella review of meta-analyses. Int J Rheum Dis. 2018;21:1514-1524. 10.1111/1756-185X.13356Umbrella review of meta-analyses of observational studies on environmental risk factors for the most common rheumatic diseases
- Ludwig-Borycz E; Guyer HM; Aljahdali AA et al. Organic food consumption is associated with inflammatory biomarkers among older adults. Public health nutrition. 2021;24:4603-4613. 10.1017/S1368980020005236Observational
- Hock E; Martyn-St James M; Wailoo A et al. Treat-to-Target Strategies in Rheumatoid Arthritis: a Systematic Review and Cost-Effectiveness Analysis. SN Comprehensive Clinical Medicine. 2021;3:838-854. 10.1007/s42399-021-00727-4Systematic review of randomised controlled trials
- Zhang WT; Liu Z; Zhu BC et al. Effects of tobacco smoking on cardiovascular disease in patients with systemic lupus erythematosus: A systematic review and meta-analysis. Front Immunol. 2022;13:967506. 10.3389/fimmu.2022.967506Systematic review and meta-analysis of 10 studies comprising 6
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.
Across the site