Abstract
Objective
People with rheumatoid arthritis (RA) are at increased risk of serious infection, but less is known about nonserious infections. Our prospective cohort study evaluated associations between medications for RA and the risk of nonserious infections.
Methods
We remotely recruited adults with RA in a community rheumatology practice‐based research network. Participants joined the ArthritisPower Registry (now PatientSpot) and completed a baseline and up to six monthly follow‐up surveys. Using data from consecutive monthly surveys, we assessed associations between medication use at the prior survey and infection report at the subsequent survey, adjusting for confounders.
Results
We recruited 351 people with RA (mean age 60, 84% female) who reported 439 infections (330 infections per 100 patient‐years). Associations between medication use and infection were assessed among 1,075 qualifying observations with 289 (27%) total infections and 146 (14%) infections with health care encounters or antibiotic use. Compared to those receiving conventional synthetic disease‐modifying antirheumatic drugs who were biologic or JAK inhibitor (JAKi) naïve, current biologic or JAKi use was not associated with either infection outcome. Infections were numerically more common with glucocorticoids ≥10 mg/day (odds ratio 1.94, 95% confidence interval 0.89–4.24). Season, previous infection, poorer function, and rural residence were significantly associated with one or both infection outcomes.
Conclusion
Biologics and JAKi were not associated with greater risk for nonserious infections compared to conventional therapies. Given that nonserious infection risk may be due more to exposures and general health status, future studies should assess whether the common practice of interrupting medications in people with infection improves or worsens outcomes.
INTRODUCTION
Infections are a major contributor to morbidity and mortality in people with rheumatoid arthritis (RA). Serious infections are one of the three leading causes of death in people with RA. 1 , 2 , 3 , 4 , 5 , 6 Comorbidities and RA severity contribute to the risk of infection, but immunosuppressive medications are the primary modifiable risk factor. 1
Relatively little, however, is known about the risk of more common, nonserious infections. These infections may frequently not lead to a formal health care encounter or antibiotic prescription and therefore may not be captured completely in administrative data. Additionally, the frequency of data collection and lack of questions specific to nonserious infections in most patient registries and trials limit the ability to fully capture these outcomes and understand their importance to patients. Nonserious infections are much more common than serious infections in people with RA 7 , 8 , 9 and may have a substantial impact on quality of life. 10 These infections may also affect RA treatment decisions and cause treatment interruptions. 11
We conducted a prospective, observational study of people with RA to comprehensively capture nonserious infections based on patient self‐report from monthly surveys. We aimed to assess the associations between different RA treatments (including disease‐modifying antirheumatic drugs [DMARDs] and glucocorticoids) and nonserious infection risk, hypothesizing that higher dose glucocorticoids would be the most strongly associated with nonserious infection risk.
PATIENTS AND METHODS
We remotely recruited adults from July 2022 to July 2023 who were seen in the Excellence Network in Rheumatology to Innovate Care and High‐Impact Research (ENRICH), a national community rheumatology practice‐based research network. Among patients of rheumatology providers who agreed to have their patients contacted for the study, we used the electronic health record (EHR) to identify those who were ≥19, had a diagnosis of RA, no diagnosis of lupus, inflammatory bowel disease, psoriatic arthritis, or ankylosing spondylitis, and who were listed in the EHR as receiving treatment with a tumor necrosis factor inhibitor (TNFi), JAK inhibitor JAKi, abatacept, or methotrexate (without use of a JAKi or biologic). Although people with RA receiving other DMARDs were eligible to join the study, we aimed recruitment efforts at a subset of medications to ensure large enough subgroups for analysis.
First, potentially eligible patients were sent invitations by email to join the study. Later potentially eligible patients were contacted by a remote recruiter who first texted patients, then called to describe the study, and finally either emailed or texted a link to join the study. A second attempt was made in one week if the patient could not be contacted.
Potential participants were directed to a landing page where they agreed to participate and provided informed consent and then answered screening questions to confirm eligibility. Eligible patients were directed to join ArthritisPower (now PatientSpot) using a smartphone or computer. After registering, they were directed to join the study and complete initial surveys in the PatientSpot app. Some additional baseline information was also collected through a web‐based survey. Patients were then instructed to complete follow‐up surveys in the PatientSpot app every 30 days for 6 surveys. When surveys were due, daily reminders were sent by email and lockscreen notification through the app. Those who did not complete surveys were reminded by text message or phone call by the study coordinator. Participants received $15 per completed follow‐up survey.
Data collection
Data collection at baseline included demographics, RA duration, prior RA treatments, comorbidities, and other patient characteristics. We also extracted data from the EHR; diabetes (diagnosis or use of a diabetes medication in the year before enrollment) and chronic obstructive pulmonary disease (COPD) or asthma (diagnoses in the year before enrollment), and number of previous biologics or JAKi (all available data before index). We extracted tender and swollen joint count, physician global, and C‐reactive protein results using data closest to but preceding enrollment. At the baseline survey and all follow‐up surveys in the PatientSpot app, participants reported current medications including glucocorticoid dose, patient‐reported outcomes, and RA flares in the past 30 days. Zip code was used to determine urban versus rural status using county classification from the National Center for Health Statistics. 12
Participants reported infections in the past 30 days (primary study outcome). Those with infections reported infection type, duration, whether infection was ongoing, severity (1–7 Likert scale), the modified Jackson symptom survey measuring presence and severity of eight symptoms of upper respiratory infection on a 1 to 7 Likert scale (range 0–32, ≥5 meeting subjective criteria for clinical illness), 13 and whether infection led to missed work, antibiotics, temporarily stopping RA medications, doctor's office or urgent care visit, emergency department visit, or hospitalization.
Statistical analysis
Cohort characteristics at baseline were assessed descriptively. Incidence of infection was calculated in person‐years based on the number of monthly surveys with infection divided by total surveys multiplied by 12, excluding surveys in which ongoing infection was reported at the previous month's survey to avoid double‐counting infections that spanned two monthly periods. Those with >1 infection in a month were counted as having a single infection.
To assess associations between medication use and infections, we restricted analyses to surveys in which the current survey was preceded by a completed survey the previous month, allowing capture of exposures (eg, medications) at the preceding survey and the outcome (infection in the past 30 days) from the current survey. Again, we excluded those with ongoing infection at the preceding survey.
Key exposures of interest included DMARD use and glucocorticoid dose. DMARDs were categorized into (1) conventional synthetic disease‐modifying antirheumatic drugs (csDMARDs) without a biologic or JAKi, (2) TNFi, (3) non‐TNFi biologic, and (4) JAKi (each of which could be with or without csDMARDs). The csDMARD‐only group was further subdivided into those with no prior biologic or JAKi use (reference group) and those with prior biologic or JAKi use given prior literature indicating significant differences in infection risk in these people (driven by avoidance of biologics or JAKi in those with prior infections or other adverse outcomes while on therapy) 14 as well as our own observation of differential risk from this study. Glucocorticoid dose was categorized as none, ≤5 mg/day, >5 and ≤10 mg/day, or >10 mg/day, based on patient's self‐reported use of these doses.
The primary outcome was any infection reported at the current survey (with the overwhelming majority of these infections being nonserious infections). A prespecified secondary outcome was infection leading to antibiotic use or health care utilization (doctor's visit, urgent care, emergency department visit, or hospitalization), excluding infections in which these details were missing (11 of 523 infections).
Associations between medication use at the preceding survey and infection report from the current survey were evaluated using logistic regression with generalized estimating equations and robust variance estimators of SEs to account for multiple observations per participant, adjusting for prespecified confounders. In the primary analysis, covariates included the following: (1) age in categories, sex, race and ethnicity (self‐report), region, rural, diabetes, asthma or COPD, and smoking status from the baseline survey; (2) opioids, nonsteroidal anti‐inflammatory drugs, RA flare, Patient‐Reported Outcomes Measurement Information System (PROMIS) function, patient global, and infection from the preceding survey; and (3) season from the current survey.
Several sensitivity analyses were also conducted, including (a) additional adjustment for baseline variables that were missing in greater numbers (ie, COVID and influenza vaccination, infections and hospitalizations in the past year, and living with young children), excluding those with missing data, (b) additional adjustment for EHR variables (ie, tender and swollen joint count, physician global, body mass index (BMI), and log‐transformed C‐reactive protein) using multiple imputation with linear regression and chained equations with 10 iterations for those with missing data, (c) limiting to those taking stable DMARD treatment for three months, (d) excluding those with any infection at the preceding survey (not just ongoing infection).
To assess whether the characteristics or severity of infections differed among participants on certain therapies, we compared the types of infection descriptively across DMARD categories and glucocorticoid dose categories and compared measures of infection severity and health care utilization across these categories using linear regression for continuous variables and the chi‐square test or Fisher's exact test for categorical variables.
The study was a substudy of an approved institutional review board (IRB) protocol (Advarra #201607783). The substudy protocol was considered exempt by the IRB. All subjects provided informed consent for the parent study and additional addendum to consent.
RESULTS
Recruitment and cohort characteristics
Among 727 people who went to the landing page and agreed to participate, 660 (90.8%) people completed consent, 641 (88.2%) people passed the screener questions, 418 (57.5%) people registered in PatientSpot, and 351 (48.3%) people began participation and completed the baseline surveys. Of 5,248 patients emailed, 258 (4.9%) patients successfully registered in PatientSpot, and among 1,316 people contacted by the remote recruiter, 155 (11.8%) people registered in PatientSpot.
Among the 351 participants, mean age was 60; 84% were female, 87% non‐Hispanic White (Table 1, Supplemental Table 1). At baseline, 78% of participants were receiving a biologic or JAKi, most commonly a TNFi, and 22% of participants were taking csDMARDs alone (83% methotrexate). Glucocorticoids were used by 24% of participants, with most (71%) participants on ≤5 mg/day. Participants tended to have long‐standing RA, low joint counts, and low physician global scores.
Table 1.
Baseline participant characteristics*
| Characteristics | Subjects with at least one follow‐up survey, n = 293 | Subjects with only baseline survey, n = 58 | All subjects, n = 351 |
|---|---|---|---|
| Number of follow‐up surveys, median (IQR) | 5 (3–6) | 0 (0–0) | 5 (1–6) |
| Age, mean (SD), y | 59.8 (13.6) | 61.4 (12.9) | 60.1 (13.5) |
| Female, n (%) | 251 (85.7) | 44 (75.9) | 295 (84.0) |
| White non‐Hispanic, n (%) | 256 (87.4) | 50 (86.2) | 306 (87.2) |
| Comorbidities, n (%) | |||
| Diabetes a | 37 (12.6) | 4 (6.9) | 41 (11.7) |
| Asthma or COPD a | 54 (18.4) | 16 (27.6) | 70 (19.9) |
| Fibromyalgia or chronic pain b | 67 (25.5) | 16 (30.2) | 83 (26.3) |
| Current smoking a | 17 (5.8) | 4 (6.9) | 21 (6.0) |
| Baseline medications c | |||
| Current DMARD use, n (%) | |||
| csDMARD only, biologic or JAKi naïve | 41 (14.1) | 8 (13.8) | 49 (14.0) |
| csDMARD only, biologic or JAKi experienced | 26 (8.9) | 2 (3.4) | 28 (8.0) |
| TNFi | 122 (41.9) | 34 (58.6) | 156 (44.7) |
| Non‐TNFi biologic | 69 (23.7) | 8 (13.8) | 77 (22.1) |
| JAKi | 28 (9.6) | 5 (8.6) | 33 (9.5) |
| No current DMARD | 5 (1.7) | 1 (1.7) | 6 (1.7) |
| Glucocorticoids (past 30 days), n (%) | |||
| None | 227 (78.0) | 37 (63.8) | 264 (75.6) |
| ≤5 mg | 46 (15.8) | 14 (24.1) | 60 (17.2) |
| >5 to ≤10 mg | 9 (3.1) | 3 (5.2) | 12 (3.4) |
| >10 mg | 9 (3.1) | 4 (6.9) | 13 (3.7) |
| Prior biologics or JAKi a | 248 (84.6) | 49 (84.5) | 297 (84.6) |
| Disease activity, median (IQR) d | |||
| Physician global (0–10) | 1.0 (0–3.5) | 1.5 (0.5–4.2) | 1.0 (0–3.5) |
| Tender joint count (0–28) | 2 (1–6) | 2 (1–8) | 2 (1–6) |
| Swollen joint count (0–28) | 1 (0–4) | 1 (0–4) | 1 (0–4) |
| CRP (mg/dL) | 0.3 (0.1–0.6) | 0.3 (0.1–0.5) | 0.3 (0.1–0.6) |
| Other patient characteristics b , n (%) | |||
| Hospitalized infection past year | 15 (5.7) | 2 (3.8) | 17 (5.4) |
| Influenza vaccine past year | 170 (64.6) | 39 (73.6) | 209 (66.1) |
| COVID vaccine ever | 229 (87.1) | 46 (86.8) | 275 (87.0) |
COPD, chronic obstructive pulmonary disease; CRP, C‐reactive protein; csDMARD, conventional synthetic disease‐modifying antirheumatic drug; DMARD, disease‐modifying antirheumatic drug; EHR, electronic health record; IQR, interquartile range; JAKi, JAK inhibitor; TNFi, tumor necrosis factor inhibitor.
Based on patient report, except EHR data used for 30 patients with follow‐up surveys completed and 5 patients with no follow‐up surveys who had missing baseline survey data. EHR data on smoking were from patient report on tablets during clinic visits when available and otherwise from the EHR record. EHR definition of diabetes based on a diabetes diagnosis or use of a diabetes medication.
Based on patient report, with data missing for 30 patients with follow‐up surveys available and 5 patients with no follow‐up surveys.
Medication missing for two patients.
Using EHR data – most recent before the index date. Physician global missing in 2 with follow‐up surveys completed and 2 with no follow‐up surveys completed, joint counts missing in 34 with baseline surveys completed and 5 with no baseline surveys completed. CRP missing in 52 with follow‐up surveys completed and 10 without baseline surveys completed.
Of the 351 participants, 293 (83%) participants completed ≥1 follow‐up survey (median of 5 follow‐up surveys). Those who completed a follow‐up survey were more frequently female, taking methotrexate, and receiving a non‐TNF biologic, but were less frequently receiving a TNFi or glucocorticoids (Table 1).
Incidence of infection
A total of 1,681 surveys were completed (351 baseline and 1,330 follow‐up) with 523 infections reported. Infection data were missing in 7 surveys, and 84 participants reported ongoing infection at the previous month's survey. Excluding these surveys (to avoid double‐counting) left 439 infections from 1,590 monthly surveys (27.6%). As previously reported, this corresponds to an incidence of infection of 3.3 per person‐year (330 per 100 person‐years). 10
Among these 439 infections, the most common were upper respiratory infections (55.6%), followed by COVID (11.4%), urinary tract infection (8.9%), and gastrointestinal infection (7.1%). Of 429 infections with additional details available, 198 (46.2%) led to a doctor's office or urgent care visit, 174 (40.6%) led to antibiotic use, 24 (5.6%) led to an emergency department visit, and 6 (1.4%) led to hospitalization (ie, 98.6% of reported infections were “nonserious”). In total 231 of 429 (53.8%) led to any health care encounter or antibiotic use. One hundred eighty‐seven of 199 (94%) respiratory infections had a Jackson symptom score ≥5, meeting subjective criteria for clinical illness.
Associations between medication use and infection
Of the 1,330 follow‐up surveys completed, 1,242 (93.4%) participants had completed the survey the previous month. Of these, 35 participants had missing data or were not taking a DMARD at the previous survey, leaving 1,207 (97.2%) follow‐up surveys. Among these, 132 (10.9%) participants reported ongoing infection at the previous month's survey, leaving 1,075 observations among 273 participants for the primary analysis. There were 289 (27%) infections among these 1,075 observations, and 146 (14%) infections with health care encounters or antibiotic use.
The association between medications and infections are shown in Table 2 with adjusted results plotted in Figure 1. Examining categories of DMARD use, the highest frequency of infection was observed in people taking csDMARDs without a biologic or JAKi who had received biologics or JAKi in the past (36.9%), with similar frequency of infection among those receiving a biologic, JAKi, or a csDMARD alone with no prior biologic or JAKi use (range 17.4%–29.8%). There was no statistically significant difference in infection risk across DMARD categories in unadjusted or adjusted analyses. A similar pattern was seen for infections leading to health care encounters or antibiotic use, although people taking csDMARDs alone with previous biologic or JAKi use had a significant greater risk of these infections than people taking a csDMARD alone without previous biologic or JAKi use (22.9% vs 12.1%, adjusted odds ratio 2.65 [confidence interval 1.09–6.45]).
Table 2.
Association between medication use and report of infection at the next monthly survey*
| Observations | Infection, n (%) | Unadjusted OR (95% CI) | Adjusted OR (95% CI) | |
|---|---|---|---|---|
| All infections | ||||
| DMARD | ||||
| csDMARD only, no prior biologic or JAKi | 142 | 38 (26.8) | Reference | Reference |
| csDMARD only, with prior biologic or JAKi | 84 | 31 (36.9) | 1.55 (0.79–3.05) | 1.50 (0.73–3.10) |
| TNFi | 463 | 138 (29.8) | 1.07 (0.65–1.76) | 1.01 (0.61–1.66) |
| Non‐TNFi | 254 | 59 (23.2) | 0.82 (0.47–1.43) | 0.74 (0.41–1.33) |
| JAKi | 132 | 23 (17.4) | 0.54 (0.26–1.11) | 0.59 (0.28–1.23) |
| Glucocorticoids | ||||
| None | 856 | 218 (25.5) | Reference | Reference |
| ≤5 mg/day | 153 | 47 (30.7) | 1.19 (0.78–1.83) | 0.93 (0.60–1.44) |
| >5 to ≤10 mg/day | 33 | 8 (24.2) | 1.04 (0.51–2.13) | 0.72 (0.30–1.72) |
| >10 mg/day | 33 | 16 (48.5) | 2.39 (1.21–4.71) | 1.94 (0.89–4.24) |
| Infection leading to physician visit or antibiotic use a | ||||
| DMARD | ||||
| csDMARD only, no prior biologic or JAKi | 141 | 17 (12.1) | Reference | Reference |
| csDMARD only, with prior biologic or JAKi | 83 | 19 (22.9) | 2.29 (1.05–4.97) | 2.94 (1.20–7.21) |
| TNFi | 461 | 74 (16.0) | 1.30 (0.71–2.41) | 1.25 (0.65–2.39) |
| Non‐TNFi | 251 | 28 (11.2) | 0.92 (0.45–1.86) | 0.92 (0.43–1.96) |
| JAKi | 131 | 8 (6.1) | 0.46 (0.17–1.22) | 0.55 (0.19–1.59) |
| Glucocorticoids | ||||
| None | 849 | 114 (13.4) | Reference | Reference |
| ≤5 mg/day | 152 | 20 (13.2) | 0.81 (0.45–1.48) | 0.54 (0.30–0.98) |
| >5 to ≤10 mg/day | 33 | 5 (15.2) | 1.25 (0.53–2.94) | 0.90 (0.33–2.46) |
| >10 mg/day | 33 | 7 (21.2) | 1.36 (0.54–3.39) | 1.03 (0.43–2.46) |
Instances in which there were two consecutive monthly surveys were included. Medication exposures were determined based on the preceding survey, with the outcome of infection assessed from the subsequent survey, excluding patients with ongoing infection at the time of the preceding survey to avoid capturing prevalent infections. Patients on conventional synthetic DMARDs were divided into those with no prior biologic or JAK inhibitor use (biologic or JAKi naïve) and those with prior biologic or JAKi use. Unadjusted and adjusted results from logistic regression models with generalized estimating equations to account for within‐patient correlations, with DMARDs at the previous survey and glucocorticoids at the previous survey in separate models for unadjusted results. Adjusted results are from models with both DMARDs and glucocorticoids at the previous survey as well as covariates: baseline age, sex, race and ethnicity, region, rural, smoking, diabetes, asthma or COPD, current season, as well as measures from the preceding survey including opioid use, nonsteroidal anti‐inflammatory medication use, disease flare, PROMIS function T score, patient global, and infection. Bold values indicate p < 0.05. CI, confidence interval; COPD, chronic obstructive pulmonary disease; csDMARD, conventional synthetic disease‐modifying antirheumatic drug; DMARD, disease‐modifying antirheumatic drug; JAKi, JAK inhibitor; OR, odds ratio; PROMIS, Patient‐Reported Outcomes Measurement Information System; TNFi, tumor necrosis factor inhibitor.
Data on whether infection led to health care encounters of antibiotic use were missing for 8 observations and were excluded here.
Figure 1.
Association between medication use and report of infection at the next monthly survey. Instances in which there were two consecutive monthly surveys were included. Medication exposures were determined based on the preceding survey, with the outcome of infection assessed from the subsequent survey, excluding patients with ongoing infection at the time of the preceding survey to avoid capturing prevalent infections. Patients on conventional synthetic DMARDs were divided into those with no prior biologic or JAK inhibitor use (biologic/JAKi naïve) and those with prior biologic or JAKi use. aOR from multivariable logistic regression models with generalized estimating equations to account for within‐patient correlations, adjusted for age, sex, race and ethnicity, season, region, rural, smoking, diabetes, and asthma or chronic obstructive pulmonary disease. Also adjusted for measures from the preceding survey including opioid use, nonsteroidal anti‐inflammatory medication use, disease flare, PROMIS Function T score, patient global, infection. aOR, adjusted odds ratio; CI, confidence interval; csDMARD, conventional synthetic disease‐modifying antirheumatic drug; DMARD, disease‐modifying antirheumatic drug; OR, odds ratio; TNFi, tumor necrosis factor inhibitor.
Examining glucocorticoid use, the frequency of any infection or infection leading to health care encounter or antibiotic use was highest among people receiving glucocorticoids >10 mg/day (48.5% and 21.2%, respectively). Although the risk of any infection was significantly different in those receiving >10 mg/day versus no glucocorticoids in unadjusted analyses, there were no significant differences in either outcome in adjusted analyses. Lower dose glucocorticoids were not associated with either outcome in unadjusted or adjusted analyses.
Results of the four sensitivity analyses were similar to the primary adjusted analyses, although in some sensitivity analyses associations between prednisone >10 mg/day and any infection were statistically significant, and in some sensitivity analyses the greater risk of infection leading to health care encounters or antibiotic use in those taking csDMARDs alone but with previous biologic or JAKi use were not statistically significant (Supplemental Tables 2 and 3).
Associations between covariates at the infection outcomes from the full multivariable model are shown in Supplemental Table 4. Winter season and infection at the previous survey were associated with significantly greater risk of infection. Winter and fall season, rural residence, and poorer PROMIS function were associated with significantly greater risk of infection with health care encounters or antibiotics. In addition, those aged ≤40, current smokers, residents in the Midwest or South, and those with diabetes or asthma or COPD reported more of both outcomes although these associations were not statistically significant.
Difference in infection characteristics based on medications used
Specific infection types and characteristics of infections across DMARD treatment groups are shown in Table 3, and across glucocorticoid dose categories are shown in Table 4. Upper respiratory infections were the most common infections across all treatment groups. There were no statistically significant differences in infection duration, health care utilization, or infection severity across DMARD treatment groups, although frequency of infections with severity of six or seven (of seven) was numerically higher among those receiving a biologic or JAKi. The same was true across glucocorticoid dose groups except that those with infection receiving >5 to ≤10 mg/day of glucocorticoids had higher mean disease severity and were more likely to report a severity of six or seven than those in other glucocorticoid dose groups, although these numbers were based on a small number of outcomes.
Table 3.
Specific infection types and characteristics of infections across DMARD treatment groups*
| csDMARD, no prior biologic or JAKi | csDMARD, with prior biologic or JAKi | TNFi | Non‐TNFi biologic | JAKi | |
|---|---|---|---|---|---|
| Number of observations | 142 | 84 | 463 | 254 | 132 |
| Number of infections, n (%) | 38 (26.8) | 31 (36.9) | 138 (29.8) | 59 (23.2) | 23 (17.4) |
| Specific infections, n (%) | |||||
| Upper respiratory infection | 21 (14.8) | 8 (9.5) | 84 (18.1) | 42 (16.5) | 11 (8.3) |
| COVID | 6 (4.2) | 4 (4.8) | 11 (2.4) | 7 (2.8) | 1 (0.8) |
| Influenza/“the flu” | 1 (0.7) | 0 (0.0) | 2 (0.4) | 2 (0.8) | 1 (0.8) |
| Pneumonia | 0 (0.0) | 1 (1.2) | 1 (0.2) | 1 (0.4) | 0 (0.0) |
| Urinary tract infection | 3 (2.1) | 5 (6.0) | 10 (2.2) | 4 (1.6) | 2 (1.5) |
| Skin or soft tissue | 2 (1.4) | 1 (1.2) | 9 (1.9) | 1 (0.4) | 3 (2.3) |
| Shingles | 0 (0.0) | 0 (0.0) | 1 (0.2) | 0 (0.0) | 1 (0.8) |
| Gastrointestinal | 3 (2.1) | 6 (7.1) | 10 (2.2) | 0 (0.0) | 2 (1.5) |
| Other | 2 (1.4) | 6 (7.1) | 10 (2.2) | 2 (0.8) | 2 (1.5) |
| Characteristics of infections with full data available, no. a | 37 | 30 | 136 | 56 | 22 |
| Duration, n (%) | |||||
| ≤1 wk | 13 (35.1) | 12 (40.0) | 54 (39.7) | 20 (35.7) | 10 (45.5) |
| 1–2 wk | 18 (48.6) | 15 (50.0) | 46 (33.8) | 22 (39.3) | 9 (40.9) |
| 2–3 wk | 3 (8.1) | 1 (3.3) | 16 (11.8) | 12 (21.4) | 1 (4.5) |
| 3–4 wk | 3 (8.1) | 2 (6.7) | 9 (6.6) | 1 (1.8) | 1 (4.5) |
| >4 wk | 0 (0.0) | 0 (0.0) | 11 (8.1) | 1 (1.8) | 1 (4.5) |
| Health care utilization, n (%) | |||||
| Antibiotics | 12 (32.4) | 16 (47.1) | 68 (43.6) | 23 (31.1) | 9 (31.0) |
| Doctor or urgent care | 18 (45.0) | 18 (52.9) | 69 (44.2) | 33 (44.6) | 10 (34.5) |
| Emergency department visit | 4 (10.0) | 0 (0.0) | 6 (3.8) | 3 (4.1) | 0 (0.0) |
| Hospitalized | 0 (0.0) | 0 (0.0) | 2 (1.3) | 0 (0.0) | 0 (0.0) |
| Severity or impact | |||||
| Miss work, n (%) | 20 (54.1) | 10 (33.3) | 49 (36.0) | 16 (28.6) | 8 (36.4) |
| Stop RA medications, n (%) | 8 (21.6) | 5 (16.7) | 31 (22.8) | 10 (17.9) | 4 (18.2) |
| Severity (1–7), mean (SD) | 3.92 (1.59) | 3.77 (1.28) | 3.77 (1.63) | 4.11 (1.53) | 3.50 (1.79) |
| Severity 6 or 7 of 7, n (%) | 3 (8.1) | 2 (6.7) | 26 (19.1) | 12 (21.4) | 3 (13.6) |
csDMARD, conventional synthetic disease‐modifying antirheumatic drug; DMARD, disease‐modifying antirheumatic drug; JAKi, JAK inhibitor; RA, rheumatoid arthritis; TNFi, tumor necrosis factor inhibitor.
Instances in which there were two consecutive surveys were included. Medication exposures were determined based on the preceding survey, with the outcome of infection assessed from the subsequent survey. Patients with ongoing infection at the time of the preceding survey were excluded. No statistically significant difference in illness duration, health care utilization, or severity or impact was observed across treatment groups based on linear regression analyses for continuous variables and the chi‐square test or Fisher's exact test for categorical variables.
Table 4.
Specific infection types and characteristics of infections across glucocorticoid dose categories*
| No glucocorticoids | ≤5 mg/day | >5 to ≤10 mg/day | 10 mg/day | |
|---|---|---|---|---|
| Number of observations | 856 | 153 | 33 | 33 |
| Number of infections, n (%) | 218 (25.5) | 47 (30.7) | 8 (24.2) | 16 (48.5) |
| Specific infections, n (%) | ||||
| Upper respiratory infection | 122 (14.3) | 31 (20.3) | 4 (12.1) | 9 (27.3) |
| COVID | 24 (2.8) | 2 (1.3) | 1 (3.0) | 2 (6.1) |
| Influenza/“the flu” | 5 (0.6) | 1 (0.7) | 0 (0.0) | 0 (0.0) |
| Pneumonia | 3 (0.4) | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| Urinary tract infection | 19 (2.2) | 1 (0.7) | 2 (6.1) | 2 (6.1) |
| Skin or soft tissue | 10 (1.2) | 5 (3.3) | 0 (0.0) | 1 (3.0) |
| Shingles | 2 (0.2) | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| Gastrointestinal | 18 (2.1) | 1 (0.7) | 0 (0.0) | 2 (6.1) |
| Other | 15 (1.8) | 6 (3.9) | 1 (3.0) | 0 (0.0) |
| Characteristics of infections with full data available, no. a | 211 | 46 | 8 | 16 |
| Duration, n (%) | ||||
| ≤1 wk | 87 (41.2) | 17 (37.0) | 2 (25.0) | 3 (18.8) |
| 1–2 wk | 84 (39.8) | 14 (30.4) | 4 (50.0) | 8 (50.0) |
| 2–3 wk | 21 (10.0) | 9 (19.6) | 0 (0.0) | 3 (18.8%) |
| 3–4 wk | 12 (5.7) | 3 (6.5) | 0 (0.0) | 1 (6.3) |
| >4 wk | 7 (3.3) | 3 (6.5) | 2 (25.0) | 1 (6.3) |
| Health care utilization, n (%) | ||||
| Antibiotics | 82 (38.9) | 18 (39.1) | 3 (37.5) | 5 (31.3) |
| Doctor or urgent care | 99 (46.9) | 16 (34.8) | 4 (50.0) | 7 (43.8) |
| Emergency department visit | 11 (5.2) | 0 (0.0) | 0 (0.0) | 2 (12.5) |
| Hospitalized | 2 (0.9) | 0 (0.0) | 0 (0.0) | 1 (6.3) |
| Severity or impact | ||||
| Miss work, n (%) | 73 (34.6) | 18 (39.1) | 4 (50.0) | 8 (50.0) |
| Stop RA medications, n (%) | 40 (19.0) | 10 (21.7) | 3 (37.5) | 5 (31.3) |
| Severity (1–7), mean (SD) | 3.77 (1.57) | 3.96 (1.41) | 5.38 (1.77) b | 3.45 (1.82) |
| Severity 6 or 7, n (%) | 32 (15.2) | 6 (13.0) | 5 (62.5) b | 3 (18.8) |
RA, rheumatoid arthritis.
Instances in which there were two consecutive surveys were included. Medication exposures were determined based on the preceding survey, with the outcome of infection assessed from the subsequent survey. Patients with ongoing infection at the time of the preceding survey were excluded.
No statistically significant difference in illness duration, health care utilization, or severity or impact was observed across treatment groups, except that mean severity was higher in those receiving >5 to ≤ 10 mg/day of glucocorticoids (P = 0.005 vs no glucocorticoids by linear regression), and illness severity was more likely to be 6 to 7 (rather than 1–5) in those receiving >5 to ≤10 mg/day glucocorticoids (P = 0.01 by Fisher's exact testing).
DISCUSSION
In this prospective study of people with RA specifically designed to capture nonserious infections, we found that participants using biologic therapies or JAKi were not more likely to report an infection than those receiving conventional DMARD therapies alone. As expected, we found the highest rate of infections in those who had previously used a biologic or JAKi but were currently only taking a csDMARD; previously studies have demonstrated higher infection risk in this population, likely because prior infections or other adverse outcomes led to discontinuation of biologics or JAKi for many of these people 14 ; in other words, this population tends to be sicker than those who have never received a biologic or JAKi. Surprisingly, in our study, glucocorticoids were not associated with a significant increase in the risk of infection, although there were numerically more infections in the few participants receiving >10 mg/day of prednisone. Additionally, we did not find evidence that biologics, JAKi, or glucocorticoids were associated with longer or greater severity of infections.
Biologics and JAKi are known to be associated with a small but measurable increase in the risk of serious infections, 15 , 16 and glucocorticoids with a dose‐dependent increase in the risk of these infections. 17 , 18 Few studies, however, have assessed the association of immunosuppressive medications with nonserious infections (the predominant infections in this study). An analysis of the British Society for Rheumatology Biologics Register (BSRBR) for RA found that the risk of nonserious infections was lower in those receiving csDMARDs than in those receiving TNFi, and higher in those receiving rituximab or interleukin‐6 inhibitors. 19 In contrast, a study of the CorEvitas registry found a small but significant association for both methotrexate and TNFi with nonserious infections. 7 Interestingly, a clinical trial examining methotrexate use for cardiovascular disease also found a small but statistically significant increased risk of any infection adverse event. 20 The lack of association of biologics or JAKi versus csDMARDs with infection seen in our study could be because both are associated with small magnitude increases in nonserious infection risk (as suggested in the CorEvitas registry study and the methotrexate clinical trial). It is also possible that our results reflect a lack of association of any of these medications with nonserious infection risk, with differences compared to prior studies due to differences in patient populations, study design (ie, lack of a new user design), or outcome capture. It is notable that our monthly surveys, directed specifically at infection capture, likely more comprehensively captured nonserious infections than previous studies; indeed we found a much higher frequency in our study (330 per 100 person‐years) than in the BSRBR study (27 per 100 patient‐years) which utilized every six month surveys. One possible explanation for differences between our results and the BSRBR study is that biologics and JAKi are not associated with nonserious infections when the full spectrum of infections are captured (including more mild infections that we captured in our study), but that there is some increase in the most severe of these nonserious infections which might make up a disproportionate share of infections captured in more infrequent surveys. Such results would not be unexpected given known associations with serious infections, and supporting this hypothesis, we did find numerically more infections that patients rated as severe (six or seven of seven) among those receiving a biologic or JAKi, although differences were not statistically significant.
Previous studies of glucocorticoids include a case‐control study using administrative data found a dose‐dependent risk of nonserious infection associated with glucocorticoid use, although the risk in those receiving <10 mg/day of prednisone was small (relative risk 1.1). 21 Similar magnitude results were found in another administrative database study 8 as well as in the two registry studies noted earlier, 7 , 19 and there was also a small increase in infections in a randomized trial of 5 mg of prednisolone. 22 In many of these studies, the associations between glucocorticoids and nonserious infections were notably weaker than between glucocorticoids and serious infections. These weaker associations may have been missed in our study. Residual confounding is also possible. Although we adjusted for a wide range of covariates and also found similar results in sensitivity analyses including other variables such as physician disease activity measures, it is possible that glucocorticoid users had less exposure to infection or less infection reporting; such confounding could explain why ≤5 mg of prednisone was associated with lower rates of infection leading to antibiotics or health care utilization. Few patients in our study received higher doses of glucocorticoids >10 mg, limiting our ability to assess the risk with higher dose glucocorticoids. It seems likely based on prior studies that higher dose glucocorticoids are associated with an increase in the risk of nonserious infection, whereas the risk with low‐dose glucocorticoids, if present, is likely small.
Although we did not find associations between medications and infection, we did find that fall or winter season (with greater infection exposure), previous infection, and poorer PROMIS function were associated with greater risk of infection, while rates of infection were also higher but not statistically significantly associated with diabetes, asthma or COPD, and smoking. These observations suggest that infection exposures, comorbidities, and health status may have a stronger impact on nonserious infection risk than therapies for RA. Previous studies have also demonstrated associations with disease activity and disability, 7 and we found higher rates of infection among those with worse PROMIS function scores. Both high and low BMI have been found to be associated with infections, although BMI was not associated with nonserious infection in our cohort (data not shown). 23 It may seem counterintuitive that younger patients had higher reports of infection in our study, but these patients may have greater exposure to infectious organisms or differential reporting.
We were not able to assess associations between medications and the risk of specific infections. Herpes zoster risk has been shown to be higher in those receiving JAKi and glucocorticoids, 24 and prior studies have also suggested worsened COVID outcomes in those receiving glucocorticoids >10 mg, rituximab, or JAKi. 25 , 26 We also did not have sufficient rituximab use in our study to separately evaluate the effect or rituximab on outcomes, and could not separately evaluate associations with specific csDMARDs which might carry less infection risk (eg, hydroxychloroquine and sulfasalazine).
Although nonserious infections are less severe than serious infections by their nature, they are much more common. In this study, the incidence of infection was approximately 330 per 100 patient‐years (3–4 infections per patient per year). In contrast, the incidence of serious, hospitalized infections in RA is approximately 2 to 5 per 100 patient‐years depending on the definition and the population studied. 16 , 17 , 27 Additionally, nonserious infections frequently led to health care utilization, antibiotic use, or missed work, and approximately 20% of patients temporarily stopped RA medications because of their infection in our study. We have previously reported impacts of these infections on medication interruptions and quality of life. 10 Although physicians may be less aware or less concerned about these infections, they may be of significant concern for patients and may affect adherence and willingness to start new medications. 28 , 29 Understanding to what degree medications contribute to these infections is important to allow physicians to inform patients of the risks and benefits of therapy.
Several limitations should be considered. Infection was based on self‐report rather than physician assessment and misclassification could bias toward the null, but Jackson symptom scores were ≥5 in 94% of respiratory infections, suggesting typical symptoms of a respiratory infection and accurate reporting by patients. Although we used contiguous surveys to ensure that medications were assessed before outcomes, channeling of patients with prior infections or at high infection risk to certain therapies could affect results. However, we specifically separated out patients previously but not currently receiving a biologic or JAKi, and although a new user design was not possible, we found similar results when requiring patients to have been receiving stable therapy for ≥3 months. We did not have physician disease activity data on all patients or information on the presence of interstitial lung disease, but we did adjust for patient global and PROMIS function scores, and we found similar results when adjusting for disease activity measures from the EHR. A minority of invited patients participated in the study, participants were predominantly White, and access to digital technology and English proficiency were needed for participation; these factors could affect generalizability or impact the associations seen. Selection bias could also lead to inclusion of patients who are more likely to report infections than the general population.
In conclusion, we found that nonserious infections were very common in patients with RA, but that biologics, JAKi, and low‐dose prednisone were not associated with greater risk or greater severity of these infections. For patients who experience or who are concerned about the risk of these infections, providers can explain that reducing exposure to infections, avoiding high doses of glucocorticoids, and maintaining RA disease control may have bigger effects on the risk of nonserious infections than commonly used RA therapies. The observation that these very common infections not infrequently lead to treatment interruptions suggests a need to determine whether treatment interruptions are beneficial or whether these interruptions unnecessarily lead to disease flares.
AUTHOR CONTRIBUTIONS
All authors contributed to at least one of the following manuscript preparation roles: conceptualization AND/OR methodology, software, investigation, formal analysis, data curation, visualization, and validation AND drafting or reviewing/editing the final draft. As corresponding author, Dr George confirms that all authors have provided the final approval of the version to be published and takes responsibility for the affirmations regarding article submission (eg, not under consideration by another journal), the integrity of the data presented, and the statements regarding compliance with institutional review board/Declaration of Helsinki requirements.
Supporting information
Disclosure Form:
Data S1 Supporting Information
Kelly Gavigan contributed to this article as an employee of Global Healthy Living Foundation, and the views expressed herein do not necessarily represent the views of Modus Outcomes. W. Benjamin Nowell contributed to this article as an employee of Global Healthy Living Foundation, and the views expressed herein do not necessarily represent the views of Regeneron Pharmaceuticals, Inc.
Supported by the NIH (grant K23‐AR‐073931, with infrastructure for the study supported by grant P30‐AR‐072583).
1Perelman School of Medicine, University of Pennsylvania, Philadelphia; 2Global Healthy Living Foundation, New York, New York; 3Foundation for Advancing Science, Technology, Education, and Research, Birmingham, Alabama; 4University of Alabama at Birmingham. Authors Gavigan and Venkatachalam's current address is Modus Outcomes, a THREAD Company, Waban, Massachusetts. Dr Nowell's current address is Regeneron Pharmaceuticals Inc, Tarrytown, New York.
Additional supplementary information cited in this article can be found online in the Supporting Information section (https://acrjournals.onlinelibrary.wiley.com/doi/10.1002/acr2.90050).
Author disclosures are available at https://onlinelibrary.wiley.com/doi/10.1002/acr2.90050.
Data availability
Deidentified data from this study will be made availability upon reasonable request after author approval of a formal written proposal.
References
- 1. Listing J, Strangfeld A, Kary S, et al. Infections in patients with rheumatoid arthritis treated with biologic agents. Arthritis Rheum 2005;52:3403–3412. [DOI] [PubMed] [Google Scholar]
- 2. Radovits BJ, Fransen J, Al Shamma S, et al. Excess mortality emerges after 10 years in an inception cohort of early rheumatoid arthritis. Arthritis Care Res (Hoboken) 2010;62:362–370. [DOI] [PubMed] [Google Scholar]
- 3. Sokka T, Abelson B, Pincus T. Mortality in rheumatoid arthritis: 2008 update. Clin Exp Rheumatol 2008;26:S35–S61. [Google Scholar]
- 4. van den Hoek J, Boshuizen HC, Roorda LD, et al. Mortality in patients with rheumatoid arthritis: a 15‐year prospective cohort study. Rheumatol Int 2017;37:487–493. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Ogdie A, Maliha S, Shin D, et al. Cause‐specific mortality in patients with psoriatic arthritis and rheumatoid arthritis. Rheumatology (Oxford) 2017;56:907–911. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Johnson TM, Yang Y, Roul P, et al. A narrowing mortality gap: temporal trends of cause‐specific mortality in a national matched cohort study in US veterans with rheumatoid arthritis. Arthritis Care Res (Hoboken) 2023;75:1648–1658. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Au K, Reed G, Curtis JR, et al; CORRONA Investigators . High disease activity is associated with an increased risk of infection in patients with rheumatoid arthritis. Ann Rheum Dis 2011;70:785–791. [DOI] [PubMed] [Google Scholar]
- 8. Lacaille D, Guh DP, Abrahamowicz M, et al. Use of nonbiologic disease‐modifying antirheumatic drugs and risk of infection in patients with rheumatoid arthritis. Arthritis Rheum 2008;59:1074–1081. [DOI] [PubMed] [Google Scholar]
- 9. Bergmans B, Jessurun N, van Lint J, et al. Burden of non‐serious infections during biological use for rheumatoid arthritis. PLoS One 2024;19:e0296821. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Patel H, Gavigan K, Venkatachalam S, et al. Impact of non‐serious infections on medication interruptions, quality of life, and disease flares in patients with rheumatoid arthritis. Rheumatology (Oxford) 2025;65:keaf503. [Google Scholar]
- 11. George MD, Venkatachalam S, Banerjee S, et al. Concerns, healthcare use, and treatment interruptions in patients with common autoimmune rheumatic diseases during the COVID‐19 pandemic. J Rheumatol 2021;48:603–607. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. National Center for Health Statistics . Urban rural classification scheme for counties. December 2, 2019. Accessed May 26, 2020. https://www.cdc.gov/nchs/data_access/urban_rural.htm
- 13. Gwaltney JM, Moskalski PB, Hendley JO. Interruption of experimental rhinovirus transmission. J Infect Dis 1980;142:811–815. [DOI] [PubMed] [Google Scholar]
- 14. Yun H, Xie F, Delzell E, et al. Comparative risk of hospitalized infection associated with biologic agents in rheumatoid arthritis patients enrolled in medicare. Arthritis Rheumat. 2016;68:56–66. [Google Scholar]
- 15. Singh JA, Cameron C, Noorbaloochi S, et al. Risk of serious infection in biological treatment of patients with rheumatoid arthritis: a systematic review and meta‐analysis. Lancet 2015;386:258–265. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Pawar A, Desai RJ, Gautam N, et al. Risk of admission to hospital for serious infection after initiating tofacitinib versus biologic DMARDs in patients with rheumatoid arthritis: a multidatabase cohort study. Lancet Rheumatol 2020;2:e84–e98. [DOI] [PubMed] [Google Scholar]
- 17. George MD, Baker JF, Winthrop K, et al. Risk for serious infection with low‐dose glucocorticoids in patients with rheumatoid arthritis: a cohort study. Ann Intern Med 2020;173:870–878. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Dixon WG, Abrahamowicz M, Beauchamp M‐E, et al. Immediate and delayed impact of oral glucocorticoid therapy on risk of serious infection in older patients with rheumatoid arthritis: a nested case‐control analysis. Ann Rheum Dis 2012;71:1128–1133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Bechman K, Halai K, Yates M, et al; British Society for Rheumatology Biologics Register for Rheumatoid Arthritis Contributors Group . Nonserious infections in patients with rheumatoid arthritis: results from the British Society for Rheumatology Biologics Register for Rheumatoid Arthritis. Arthritis Rheumatol 2021;73:1800–1809. [DOI] [PubMed] [Google Scholar]
- 20. Ridker PM, Everett BM, Pradhan A, et al; CIRT Investigators . Low‐dose methotrexate for the prevention of atherosclerotic events. N Engl J Med 2019;380:752–762. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Dixon WG, Kezouh A, Bernatsky S, et al. The influence of systemic glucocorticoid therapy upon the risk of non‐serious infection in older patients with rheumatoid arthritis: a nested case‐control study. Ann Rheum Dis 2011;70:956–960. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Boers M, Hartman L, Opris‐Belinski D, et al; GLORIA Trial consortium . Low dose, add‐on prednisolone in patients with rheumatoid arthritis aged 65+: the pragmatic randomised, double‐blind placebo‐controlled GLORIA trial. Ann. Rheum. Dis. 2022;81:925–936. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Carey IM, Harris T, Chaudhry UAR, et al. Body mass index and infection risks in people with and without type 2 diabetes: a cohort study using electronic health records. Int J Obes (Lond) 2025;49:1800–1809. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Winthrop KL, Curtis JR, Lindsey S, et al. Herpes zoster and tofacitinib: clinical outcomes and the risk of concomitant therapy. Arthritis Rheumatol 2017;69:1960–1968. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Gianfrancesco M, Hyrich KL, Al‐Adely S, et al; COVID‐19 Global Rheumatology Alliance . Characteristics associated with hospitalisation for COVID‐19 in people with rheumatic disease: data from the COVID‐19 Global Rheumatology Alliance physician‐reported registry. Ann Rheum Dis 2020;79:859–866. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Sparks JA, Wallace ZS, Seet AM, et al; COVID‐19 Global Rheumatology Alliance . Associations of baseline use of biologic or targeted synthetic DMARDs with COVID‐19 severity in rheumatoid arthritis: results from the COVID‐19 Global Rheumatology Alliance physician registry. Ann Rheum Dis 2021;80:1137–1146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Ozen G, Pedro S, England BR, et al. Risk of serious infection in patients with rheumatoid arthritis treated with biologic versus nonbiologic disease‐modifying antirheumatic drugs. ACR Open Rheumatol 2019;1:424–432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Neame R, Hammond A. Beliefs about medications: a questionnaire survey of people with rheumatoid arthritis. Rheumatology (Oxford) 2005;44:762–767. [DOI] [PubMed] [Google Scholar]
- 29. Fraenkel L, Bogardus ST, Concato J, et al. Patient preferences for treatment of rheumatoid arthritis. Ann Rheum Dis 2004;63:1372–1378. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Disclosure Form:
Data S1 Supporting Information
Data Availability Statement
Deidentified data from this study will be made availability upon reasonable request after author approval of a formal written proposal.