Low-dose aspirin in the primary prevention of rheumatoid arthritis: The women's health study




Low-dose aspirin may reduce the risk of developing rheumatoid arthritis (RA) through its effect on cyclooxygenase activity and its antioxidant pathways. Previous randomized trial data have demonstrated a beneficial effect of low-dose aspirin in reducing other inflammatory diseases, such as asthma and colorectal adenomas, but no trial has evaluated the role of aspirin in RA prevention.


The Women's Health Study is a randomized, double-blind, placebo-controlled trial conducted between 1992 and 2004 designed to evaluate the risks and benefits of low-dose aspirin (100 mg every other day) and vitamin E in the primary prevention of cardiovascular disease and cancer among 39,876 female health care professionals age ≥45 years throughout the US. After excluding women with RA at baseline, 39,144 women were evaluated for the present study. A definite diagnosis of RA was assessed during followup by self-report and confirmed using a connective tissue disease screening questionnaire, followed by a medical record review by a rheumatologist for American College of Rheumatology criteria.


During an average followup of 10 years, 106 women developed definite RA (48 women in the aspirin group and 58 in the placebo group). There was a nonsignificant risk for RA (relative risk [RR] 0.83, 95% confidence interval [95% CI] 0.56–1.21; P = 0.33) associated with aspirin. There were 64 seropositive RA cases (60%) and 42 seronegative RA cases (40%). Aspirin also had no significant effect on either seropositive RA (RR 1.0, 95% CI 0.61–1.63) or seronegative RA (RR 0.62, 95% CI 0.33–1.15).


One hundred milligrams of aspirin taken every other day was not associated with a significant reduction in the risk of developing RA among women in a randomized, double-blind, placebo-controlled trial.


Rheumatoid arthritis (RA) is a chronic autoimmune disease that affects approximately 1% of the population (1). Although the etiology of RA is unclear, abnormal levels of serologic markers and asymptomatic synovitis can precede the disease (2–4), potentially allowing for preventive therapy by antiinflammatory agents such as aspirin.

It is estimated that between 19% and 25% of the general population take low-dose aspirin prophylactically for the prevention of heart disease (5). Aspirin and other nonsteroidal antiinflammatory drugs (NSAIDs) are used to treat inflammatory symptoms in RA. Aspirin exerts its effect on cyclooxygenase (COX) activity, which is linked to inflammation (6). Aspirin also inhibits interleukin (IL-4) and NF-κB gene expression in non–COX-dependent pathways (7). Its role as an antioxidant and modulator of estrogen biosynthesis may also affect RA clinical onset (8). Previous studies have shown that aspirin has beneficial effects on numerous diseases, including cardiovascular disease (9), stroke (10), and colon cancer (11), but to our knowledge, no prior study has evaluated its potential to prevent RA. Evaluating whether a commonly used prophylactic medication such as aspirin can reduce the incidence of RA is an important public health question.

The Women's Health Study (WHS), a large-scale randomized, double-blind, placebo-controlled trial of aspirin and vitamin E, provides a unique opportunity to evaluate whether low-dose aspirin decreases the risk of incident RA in apparently healthy women.


Study participants.

Participants were drawn from the WHS, a randomized clinical trial of the efficacy of low-dose aspirin (100 mg every other day) and vitamin E (600 IU every other day) in the prevention of cardiovascular disease and cancer (12, 13). Details of this trial are provided elsewhere (10). Briefly, beginning in 1992, letters and baseline questionnaires were mailed to 1.7 million female health care professionals, including physicians, nurses, and dentists. A total of 453,787 women completed the initial questionnaire, and 65,169 women qualified and were interested in participating. Eligibility criteria included age ≥45 years; no prior history of coronary heart disease, cerebrovascular disease, cancer (except non-melanoma skin cancer), or other major chronic illnesses; no reported adverse effects from the study medications; no use of anticoagulants or corticosteroids; and no use of individual supplements of vitamin A, vitamin E, or β-carotene, aspirin, or any NSAIDs more than once a week (or willingness to forego their use during the trial). Eligible women were then enrolled in a 3-month run-in period with placebo medications. Following the run-in period, 39,876 women were enrolled in the study and were randomly assigned to low-dose aspirin or placebo and vitamin E or placebo, using a 2 × 2 factorial design (Figure 1). Written informed consent was obtained from all of the participants and the trial was approved by the Brigham and Women's Hospital Institutional Review Board and monitored by an external data safety monitoring board.

Figure 1.

Flow diagram of the aspirin component of the Women's Health Study. RA = rheumatoid arthritis.

For the present study, 732 women who reported having RA on their baseline questionnaires, or during followup with a diagnosis date that preceded randomization were excluded, leaving 39,144 women.

Study treatment and followup.

Women were followed up about their pill compliance by questionnaire every 6 months during the first year and yearly thereafter. They were also queried about potential side effects to aspirin, diagnoses of outcomes of interest including RA, and risk factors, such as cigarette smoking. At the end of the trial (March 31, 2004), morbidity and mortality followup were 97.2% and 99.4% complete, respectively. Participants were considered compliant if they took at least two-thirds of the study pills (reported on followup questionnaires). Non-trial use of aspirin on ≥4 days/month averaged 12% during followup, with no significant difference between active and placebo groups.

Validating self-reports of RA.

The primary end point of this study was definite RA. In the WHS, women self-reported RA on followup questionnaires. We then confirmed the diagnosis in a 2-stage process. Women were mailed a connective tissue disease screening questionnaire (CSQ) (14, 15). Women who screened positive for possible RA on the CSQ (at least 3 RA symptoms: radiographic changes of erosive disease or periarticular osteopenia, morning stiffness, arthritis of ≥3 joints, symmetric arthritis, and arthritis of the hand joints or rheumatoid nodules or a positive rheumatoid factor) had their medical records reviewed. They were independently reviewed by 2 blinded, board-certified rheumatologists (EWK, NAS) according to the American College of Rheumatology (ACR; formerly the American Rheumatism Association) classification criteria (16), evidence of RA-specific treatment, and the treating physician's diagnostic impression. Any discrepancies were resolved in person by the 2 reviewers to determine diagnostic consensus. Definite RA was defined as occurring in those who met ≥4 ACR criteria for RA (95%) or had confirmed RA based on clinical symptoms, laboratory tests, medication treatment, and the expert reviewers' consensus (5%). We also noted seropositive and seronegative RA (defined as definite RA with or without a positive rheumatoid factor test documented in the medical record) and inflammatory polyarthritis (defined as ≥2 ACR criteria for RA on medical record review). Additionally, we also examined possible RA defined using the CSQ (≥3 RA symptoms on the CSQ) and a patient's self-report of RA (that was not later denied during the validation process).

Statistical analysis.

The SAS statistical software package, release 9.1 (SAS Institute, Cary, NC), was used to perform all analyses. We used chi-square tests for proportions and Student's t-tests for continuous variables to determine any demographic differences between the aspirin and placebo groups. Cox proportional hazards regression models were used to calculate relative risks (RRs) and 95% confidence intervals (95% CIs) comparing the event rates of RA between the aspirin and placebo groups. All models were adjusted for age and randomized treatment assignment.


Table 1 outlines the baseline characteristics of women from each of the groups. The mean age of women at study entry was 54.6 years. There were no significant demographic differences between the aspirin and placebo groups with regard to age, smoking status, body mass index, parity, age at menarche, and menopausal status.

Table 1. Baseline characteristics of women from the Women's Health Study by group*
CharacteristicAspirin (n = 19,562)Placebo (n = 19,582)Total (n = 39,144)P
  • *

    Values are the number (percentage) unless otherwise indicated. Numbers do not always sum to group totals due to missing information for some variables. BMI = body mass index; HT = hormone therapy.

Age, mean ± SD years54.6 ± 754.5 ± 754.6 ± 70.59
 45–5410,945 (56)10,998 (56)21,943 (56)0.81
 55–646,297 (32)6,244 (32)12,541 (32) 
 ≥652,320 (12)2,340 (12)4,660 (12) 
Smoking status    
 Current2,512 (13)2,572 (13)5,084 (13)0.61
 Past7,048 (36)6,986 (36)14,034 (36) 
 Never9,986 (51)10,004 (51)19,990 (51) 
BMI, mean ± SD kg/m226.1 ± 526.0 ± 526.0 ± 5.10.19
 <25.09,743 (51)9,806 (51)19,549 (51)0.88
 25–29.95,920 (31)5,923 (31)11,843 (31) 
 ≥303,487 (18)3,462 (18)6,949 (18) 
Age at menarche, years    
 ≤101,675 (9)1,549 (8)3,224 (8)0.18
 113,142 (16)3,187 (16)6,329 (16) 
 125,491 (28)5,587 (29)11,078 (28) 
 135,619 (29)5,647 (29)11,266 (29) 
 ≥143,613 (18)3,587 (18)7,200 (18) 
 Nulliparous2,463 (13)2,498 (13)4,961 (13)0.61
 Parous17,029 (87)17,002 (87)34,031 (87) 
Menopausal status and HT    
 Premenopausal5,416 (28)5,450 (28)10,866 (28)0.94
 Uncertain3,455 (18)3,556 (18)7,011 (18) 
 Postmenopausal, no HT3,214 (16)3,211 (16)6,425 (16) 
 Postmenopausal, past HT use1,422 (7)1,440 (7)2,862 (7) 
 Postmenopausal, current HT use6,008 (31)5,872 (30)11,880 (30) 

Case validation.

At the end of the trial, with an average of 9.99 years of followup, 1,110 women reported having RA. These women were sent a CSQ and were asked to provide more details regarding their self-reported diagnosis of RA. Of the 803 women (72%) who responded to this questionnaire, 456 (41%) subsequently denied having a diagnosis of RA. Nonresponders to the CSQ were similar to the responders with regard to smoking status and age, and were as likely as the responders to have received aspirin or placebo (P = 0.35). A total of 177 (51%) of the remaining 347 responders who did not deny the diagnosis of RA screened positive for RA on the CSQ. Medical record review confirmed RA in 106 women. This confirmation calculates out to an annual incidence rate of 27.1 cases per 100,000 person-years. There were 64 seropositive RA cases (60%) and 42 seronegative RA cases (40%).

Primary end point.

A confirmation of definite RA on the medical record review was the primary end point of the study. Of the 106 subjects with the primary end point, 48 were randomized to the aspirin group and 58 were randomized to the placebo group. The mean ± SD duration from the time of randomization to the diagnosis of RA was 5.8 ± 2.4 years. There was a nonsignificant RR of 0.83 for RA (95% CI 0.56–1.21, P = 0.33) associated with aspirin.

Secondary end points.

We also examined the incidence of RA defined as seropositive RA, seronegative RA, inflammatory polyarthritis, RA defined using the CSQ, or self-reported RA. When we examined seropositive RA (RR 1.0, 95% CI 0.61–1.63) and seronegative RA (RR 0.62, 95% CI 0.33–1.15) separately, aspirin had no effect on either end point. There was no significant risk reduction of RA by aspirin for any of the remaining 3 secondary end points: inflammatory arthritis (RR 0.91, 95% CI 0.65–1.28), RA defined as screening positive on the CSQ (RR 1.16, 95% CI 0.86–1.56), or self-reported RA (RR 1.01, 95% CI 0.89–1.13) (Table 2).

Table 2. Relative risks (RRs) of rheumatoid arthritis (RA) for women from the Women's Health Study by group*
No. of events
OutcomeNo. of casesAspirinPlaceboRR (95% CI)P
  • *

    95% CI = 95% confidence interval; CSQ = connective tissue disease screening questionnaire.

  • Cox proportional hazards models adjusted for age at randomization and randomized treatment assignment.

  • Confirmed on medical record review. Seropositive RA was defined as cases with a positive rheumatoid factor test documented in the medical record.

  • §

    Defined as ≥2 American College of Rheumatology (formerly the American Rheumatism Association) criteria for RA (16) on medical record review.

  • Defined as ≥4 RA symptoms on the CSQ (14).

  • #

    Analysis considers subjects who reported RA on initial questionnaire, but who subsequently refuted the diagnosis of RA as noncases.

Definite RA10648580.83 (0.56–1.21)0.33
Seropositive RA6432321.00 (0.61–1.63)0.99
Seronegative RA4216260.62 (0.33–1.15)0.13
Inflammatory polyarthritis§13464700.91 (0.65–1.28)0.60
RA defined using CSQ17795821.16 (0.86–1.56)0.33
All self-reported RA1,1105575531.01 (0.89–1.13)0.93
Unrefuted self-reported RA#6543323221.03 (0.88–1.20)0.71


In this large randomized, double-blind, placebo-controlled study of 100 mg of aspirin or placebo on alternate days, we found no significant effect of low-dose aspirin on the incidence of definite RA in women. We also did not find an increased incidence of other types of RA, defined using several different case definitions, including seropositive and seronegative RA, inflammatory polyarthritis, or self-reported RA.

We were interested in examining the role of aspirin in preventing the development of RA because several plausible biologic mechanisms exist. Previous studies have demonstrated that low-dose aspirin is effective in preventing other diseases where prostaglandin pathways are involved, such as colorectal adenomas and cancers. In colorectal neoplasia, COX-2 is responsible for the production of prostaglandins that impact proliferation of tumor tissues (11). Inhibition of COX-2 restores apoptosis and inhibits angiogenesis, which also may delay or ameliorate synovitis in early RA (17). Similarly, prostaglandin E2 increases estrogen production in cultured cells (8), and the use of aspirin may inhibit prostaglandin-driven estrogen and progesterone production. This may explain the potential ability of aspirin to reduce hormonally-affected cancers such as breast cancer and prostate cancer (18, 19). Aspirin may influence RA incidence by modulating estrogen biosynthesis (20), as well as by its ability to restore apoptosis and inhibit angiogenesis (21).

It is also possible that aspirin may reduce the risk of RA via its role as an antioxidant. Salicylate inhibits cytokine-dependent induction of NOS2 gene expression through NF-κB activation (22), which reduces the nitrosative stress of cytokine production. Aspirin also has an antioxidant effect on proteins and scavenges hydroxyl radicals (6). An antioxidant effect could prevent or delay RA onset, although a recent report from this trial demonstrated that vitamin E, a dietary antioxidant, did not significantly reduce the risk of RA (23).

A recent analysis from the WHS trial demonstrated lower incidence of asthma among women randomized to aspirin, possibly explained by an antiinflammatory effect of aspirin (24). Aspirin inhibits IL-4 gene expression in T cells, as well as promotes IL-4– and IL-13–induced activation of STAT-6 via non–COX-dependent pathways, which may also explain its role in reducing asthmatic symptoms (7). In this report, administration of 100 mg of aspirin reduced the risk of newly reported asthma by 10%. This study had 872 new cases of asthma in the aspirin group, therefore providing sufficient power to detect small reductions in risk.

We had limited power to detect moderate reductions in risk of RA and in seropositive or seronegative disease subtypes because of the relatively small number of incident RA cases. Previous observational studies of low-dose aspirin supplement use in the prevention of cancer and asthma have reported a between 10% and 30% reduction in disease risk. In the present study, with 106 cases of definite RA, we would have only 40% power to detect a 30% reduction in risk, but we would have 86% power to detect a 50% reduction in risk. For the self-reported and CSQ-positive end points, however, we would have adequate power to detect moderate-sized risk reductions. In addition, while we ascertained RA cases based on self-reports followed by CSQ screening and the medical record review, our use of strict ACR criteria when reviewing medical records may have resulted in labeling some RA cases as controls. However, analyses of other less stringent diagnostic groups demonstrated no associations as well. Furthermore, our cohort consisted of female health care professionals who differ from the general population in some characteristics. For example, study subjects were less likely to smoke than the general population, and more likely to take postmenopausal hormone therapy, but showed similar proportions of elevated blood pressure and obesity. Although the differences may affect the generalizability of our findings, they do not by themselves compromise the validity of the results. Finally, the annual incidence rate of definite RA in this study was 27.1 per 100,000 person-years, which is similar to one population-based study (25), but lower than rates reported by other studies (1, 26). Incidence rates in the present study may be lower because individuals who choose to participate in research studies tend to be healthier than the general population.

We were not able to evaluate every participant's medical record to confirm RA. Twenty-eight percent of women who self-reported RA did not respond to requests for further information, so how many of these participants may have had RA is unknown. However, we did analyze all self-reported RA as a secondary end point, where we also observed no association with aspirin. Although there was a large number of women who initially reported a diagnosis of RA but then denied this diagnosis upon further followup, this is comparable with 2 similar studies, the Iowa Women's Health Cohort Study (27) and the Nurse's Health Study, where only 7% of original self-reports were confirmed as RA (28). An initial diagnosis of RA later refuted by a rheumatologist is a common experience in community health care settings.

We cannot comment on whether higher doses of aspirin than in the present trial might have a protective effect on the development of RA. Higher doses may mask symptoms of joint pain and delay the diagnosis of RA, or higher doses may exert more antioxidant, COX-2 inhibition, and/or antiinflammatory effect. Randomized trials using higher doses of aspirin can provide additional data. Observational studies of higher doses of aspirin are limited, in that there is potential for confounding by indication. Individuals with early RA might take NSAIDs more frequently to alleviate their symptoms, potentially masking an inverse association.

In conclusion, low-dose aspirin through antiinflammatory, antiapoptotic, or antioxidant effects may potentially reduce the risk of developing RA. Despite these plausible biologic mechanisms, as well as data from this trial and others of a reduction in inflammatory diseases such as colorectal neoplasia and asthma, the present study demonstrated no reduction in risk of RA with aspirin. It is possible that higher doses of aspirin may prevent the diagnosis of RA, or that studies with higher numbers of RA cases may demonstrate a significant result with the modest risk reduction observed in the present study. Given the frequency of aspirin use in the general population, this question deserves further study.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Shadick had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Maher, Lee.

Acquisition of data. Shadick, Karlson, Buring, Lee.

Analysis and interpretation of data. Shadick, Cook, Maher, Buring, Lee.