Prednisone, but not TNF inhibitors, is associated with an increased risk of serious infection in older RA patients

Older person with rheumatoid arthritis (RA) make up an increasing proportion of those treated with biologic disease modifying anti-rheumatic agents (DMARDs).  The safety of these agents may differ for this subgroup of older RA patients, yet effects have generally not been selectively studied in this population.  Here, Schneeweis et al (Arthritis Rheum 2007; 56: 1754) explore the association of TNF inhibitor therapy and serious bacterial infection in RA patients 65 years of age or older.


Study subjects were low-income enrollees in the Pennsylvania Pharmaceutical Assistance Contract for the Elderly (PACE) program, a program providing prescription drug coverage to elderly residents with incomes not qualifying for Medicaid.  Enrollees from the years 1995 to 2003 with RA were identified and tracked for exposures to biologic and non-biologic DMARDs and glucocorticoids.  The first occurrence of a serious bacterial or opportunistic infection after the initiation of therapy was the primary outcome.  The association of RA therapy with serious infection was modeled, controlling for confounders such as medical comorbidity and RA severity available in the medical claims dataset (i.e. orthopedic procedures, coded extra-articular manifestations of RA, number of tests for inflammatory markers ordered, etc…)


15,597 new courses of therapy were identified (1,900 with methotrexate, 469 with TNF inhibitors, 654 with “cytotoxic” DMARDs (leflunomide, cyclosporine, azathioprine), 1,957 with “non-cytotoxic” DMARDs, and 10,617 with glucocorticoids) with a total exposure time of 5,676 patient-years.  The mean follow-up time per group ranged from 0.20 to 1.29 years.  Subjects were predominantly female (90%) with a mean age of 76 years.  The groups did not significantly differ in the frequency of comorbid disease or RA severity characteristics.  Importantly, 75% of RA patients prescribed glucocorticoids were not receiving any concurrent DMARDs.
The overall incidence rate for serious infection was 2.2 per 100 patient-years.  Compared to methotrexate users, the risk of serious infection was highest for those receiving glucocorticoids, with the adjusted risk more than 2-fold greater for serious bacterial infection (RR 2.34 (95% CI 1.46 – 3.75) and almost 3-fold greater for septicemia (RR 2.91 (95% CI 1.57 – 5.38).  In contrast, compared to methotrexate users, there was no significant difference in the rate of serious bacterial infection for users of TNF-inhibitors or non-biologic DMARDs (either cytotoxic or non-cytotoxic).

For glucocorticoid users, the highest risk for serious bacterial infection was within the first 90 days of use.  Low dose glucocorticoid use (less than 5 mg per day) was not associated with an increased risk of serious infection, while increasing daily doses were associated with an increased risk in a dose dependent fashion.  Compared to methotrexate use, daily use of 20 mg or more of prednisone was associated with an almost 6-fold higher risk of serious bacterial infection (RR 5.48 (95%CI 3.29 – 9.11), with the highest risk for pneumonia (RR 6.69) and septicemia (RR 6.83).

There were 11 hospitalizations for opportunistic infection (5 aspergillosis, 4 pneumocystis, and 2 pulmonary tuberculosis).  The authors did not report in which groups these infections were observed.


Compared to methotrexate, TNF inhibitor use was not associated with an increased risk for serious bacterial infection in older RA patients.  In contrast, corticosteroid use was associated with an increased risk in a dose-dependent pattern.

Editorial Comment

These results are reassuring for the practicing rheumatologist who may be reticent to prescribe TNF inhibitors to older RA patients with active disease for fear of potentiating a life-threatening infection.  Ironically, it is these patients who are commonly prescribed corticosteroids in an effort to control disease activity, which, at least for the population studied here, presents a far greater risk for infection.  As the study was not randomized, confounding by indication may still play a role in the findings—particularly for patients receiving only high dose glucocorticoids without DMARDs, as these may be patients with medical comorbidities that prevent the use of DMARDs who may prone to infection because of comorbid disease rather then the glucocorticoids they were prescribed.  However, the study has a large sample size and this situation likely represents a very small fraction of the overall sample.

Are these results generalizable to the general RA population?  The sample population is a select group (low-income Medicare beneficiaries) and might be expected to have a higher risk of infection due to socio-economic and access to care issues.  Thus, in a more representative community based population, one might actually observe an even lower infection rate than the one reported here.  Care with prescribing TNF inhibitors should always be taken, and older RA patients should be made aware of the signs and symptoms of active infection and have ready access to a physician or, as has been suggested, a broad-spectrum antibiotic on hand to take at their discretion should they develop overt fever or chills.

Jon Giles, MD

About Jon Giles, MD