Tolerance and effectiveness of anti–tumor necrosis factor α therapies in elderly patients with rheumatoid arthritis: A population-based cohort study


  • Stéphane Genevay,

    1. University Hospitals of Geneva, Geneva, Switzerland
    Search for more papers by this author
    • Drs. Genevay and Finckh contributed equally to this study.

    • Dr. Genevay has received speaking fees (less than $10,000) from MSD and Abbott.

  • Axel Finckh,

    1. University Hospitals of Geneva, Geneva, Switzerland
    Search for more papers by this author
    • Drs. Genevay and Finckh contributed equally to this study.

    • Dr. Finckh has received an unrestricted grant from Wyeth and a research grant from the Swiss Society of Rheumatology sponsored by Abbott.

  • Adrian Ciurea,

    1. Rheumaklinik und Institut für Physikalische Medizin, Universitätsspital Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Anne-Marie Chamot,

    1. Rheumatology Private Practice, Morges, Switzerland
    Search for more papers by this author
  • Diego Kyburz,

    1. Rheumaklinik und Institut für Physikalische Medizin, Universitätsspital Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Cem Gabay,

    Corresponding author
    1. University Hospitals of Geneva, Geneva, Switzerland
    • Division of Rheumatology, University Hospital of Geneva, 26 Avenue Beau-Séjour, 1211 Geneva 14, Switzerland
    Search for more papers by this author
    • Dr. Gabay has received consulting fees and/or honoraria (less than $10,000) from Abbott, Wyeth, Essex, and Roche.

  • Physicians of the Swiss Clinical Quality Management Program for Rheumatoid Arthritis

    Search for more papers by this author
    • The Swiss Clinical Quality Management program for Rheumatoid Arthritis has received grants from the Swiss Health authorities, the Swiss Academy for Medical Sciences, and private companies (Wyeth, Abbott, Essex, Aventis, Bristol-Myers, Mepha, Merck, Novartis, and Roche).



Limited data have been published on tolerance to and efficacy of classic or biologic disease-modifying antirheumatic drugs in elderly patients with rheumatoid arthritis (RA). The goal of the present study was to evaluate the tolerance to and effectiveness of anti–tumor necrosis factor (anti-TNF) agents in elderly patients (≥65 years old) with RA (ERA) in comparison with younger patients (YRA).


The Swiss Clinical Quality Management program for RA is a longitudinal population-based cohort. All patients who had received at least 1 dose of anti-TNF agents between January 1997 and November 2005 were included and categorized according to their age. Tolerance was assessed by analyzing discontinuation rates of anti-TNF agents. Effectiveness of these agents was assessed by analyzing RA disease activity (Disease Activity Score in 28 joints [DAS28]) and functional disability (Health Assessment Questionnaire [HAQ]) after anti-TNF initiation.


Among 1,571 patients with RA treated with anti-TNF agents, 344 were ≥65 years of age at treatment initiation. Drug discontinuation rates (median time 3 years) and mean change in DAS28 scores at 2 years (–0.65 versus –0.58) were identical in ERA and YRA. However, HAQ score improved significantly less in ERA (–0.02) than in YRA (–0.1) and a subsequent analysis revealed that this finding was essentially due to patients >75 years of age.


Age in itself should not interfere with the decision to treat elderly patients with RA with anti-TNF agents. In a subset of patients ages >75 years, no functional improvement according to HAQ should be expected despite improvements in disease activity.


Rheumatoid arthritis (RA) is a chronic inflammatory disease of joints affecting ∼1% of the population and causing joint destruction and disability. Overall, approximately 20–30% of patients with RA will be diagnosed after the age of 60, resulting in an increasing use of disease-modifying antirheumatic drugs (DMARDs) in elderly individuals (1).

Several differences may be expected between young and elderly patients with RA. In elderly patients with RA (ERA), disability resulting from other causes of structural joint damage, such as osteoarthritis, may be increased. The likelihood of adverse drug reaction may also be increased due to age-related changes in drug metabolism. A higher rate of comorbidities may also increase the number of medications and, thus, the risk of drug interaction. The limit of 60 or 65 years of age has classically been used to categorize young patients with RA (YRA) from ERA. Unfortunately, patients >65 years of age are often underrepresented in RA clinical trials, resulting in a paucity of information on DMARDs in this age group.

Due to the efficacy of biologic agents and the presence of 20–50% of patients with RA refractory to conventional DMARDs (2), tumor necrosis factor α (TNFα) inhibitors have become widely used in ERA. However, there are only limited data available on the effectiveness of and tolerance to anti-TNF agents in elderly patients. Furthermore, this information derives primarily from either secondary analyses of randomized controlled trials (3, 4) or cross-sectional analyses of tertiary care centers (5). The present study was undertaken to explore the safety and effectiveness of TNF inhibitors in a large, prospective, population-based cohort of patients with RA. We hypothesized that effectiveness of anti-TNF agents should not change with age, whereas high discontinuation rates could be expected in elderly patients due to increased incidence of adverse events (6).


Study population.

The Swiss Clinical Quality Management program for RA (SCQM-RA) is a longitudinal population-based cohort of patients with RA. All patients have been diagnosed as having RA according to the clinical judgment of their rheumatologist. The SCQM-RA is a multicenter program designed to improve the quality of care for patients with RA by providing feedback on outcomes for individual patients to the physician (provider feedback) (7). Regulatory authorities in Switzerland have recommended continuous monitoring within the SCQM-RA system for all patients receiving new biologic agents (8, 9). Based on a comparison with sales data from the industry, between 70% and 80% of all Swiss patients with RA receiving anti-TNF agents are included in the SCQM-RA. Patients are generally enrolled prior to the initiation of anti-TNF therapy (8). Patients come from a wide range of settings: ∼40% come from private practices, one-third come from 5 academic centers, and one-third come from 11 nonacademic centers. The protocol of the SCQM-RA includes data on physical examination (i.e., tender joint count, swollen joint count, Disease Activity Score in 28 joints [DAS28]), laboratory tests (i.e., erythrocyte sedimentation rate, rheumatoid factor), and several patient questionnaires. These data are obtained routinely once a year or any time DMARD therapy is modified. All patients who received at least 1 anti-TNF agent between January 1997 and the end of November 2005 were included in the analysis.

Assessment of tolerance.

Tolerance was analyzed by examining drug retention rates of current TNFα inhibitors using the time until drug discontinuation, independent of the reason that led to drug interruption. Drug discontinuation rates, or drug survival, reflect the patients' and the doctors' satisfaction with a given treatment and therefore represent a useful summary measure of the overall treatment effectiveness and tolerability (10).

Because information on the reasons of permanent discontinuation of anti-TNF therapy was not consistently available for all patients in the SCQM-RA database, we conducted a detailed survey of the reasons leading to interruption of anti-TNF agents in 3 representative areas of the country (Geneva, Zurich, and Morges). Files from patients with RA who had received at least 1 injection of anti-TNF agents were reviewed and episodes of permanent discontinuation were analyzed. Basic demographic data were recorded along with specific reasons for discontinuing anti-TNF agents, which were categorized into lack of efficacy, allergic reaction, severe infection, cancer, severe cardiovascular disease, and other reasons.

Assessment of effectiveness.

Clinical effectiveness was assessed based on the longitudinal evolution of the DAS28 and the Health Assessment Questionnaire (HAQ) in response to therapy. In addition, we used the European League Against Rheumatism (EULAR) response criteria to analyze the percentage of patients who achieved a good, moderate, or insufficient response 1 year after initiation of anti-TNF therapy (11). EULAR criteria are dependent on both the change in DAS28 score and the absolute disease activity level reached. A secondary outcome of the study was the Rapid Assessment of Disease Activity in Rheumatology questionnaire (12), also called the Rheumatoid Arthritis Disease Activity Index (RADAI), which measures patient-perceived disease activity (13). This self-administered questionnaire includes 5 items and has been found to have good psychometric properties (13).

Statistical analysis.

To compare the effectiveness and survival rates of anti-TNF agents in ERA and YRA, patients were dichotomized into those ages ≥65 years (ERA) and those ages <65 years (YRA) at the time of treatment initiation. In an exploratory analysis, we further categorized geriatric age into 5-year subgroups (<65, ≥65 to <70, ≥70 to <75, ≥75 years) to investigate potential trends in higher age groups. Baseline disease characteristics were compared between YRA and ERA. The significance of differences in mean values of continuous variables was assessed with Student's t-test for normally distributed variables and with the Kruskal-Wallis test for non-normally distributed variables. For dichotomous variables, Pearson's chi-square test was used to evaluate the significance of differences in proportions. All statistical tests were 2-sided and evaluated at the 0.05 significance level. The statistical analysis was performed with Stata software, version 9.2 for Windows (StataCorp, College Station, TX).

The time to discontinuation of anti-TNF agents (drug survival) was analyzed with a Cox proportional hazards model. We examined if the proportional hazards assumption was satisfied using the Schoenfeld residuals. Survival curves of the time to discontinuation of anti-TNF agents (drug survival) were produced using the Kaplan-Meier product-limit method. The evolution of HAQ score and DAS28 were analyzed using generalized mixed models for longitudinal data. We examined whether time as a linear trend or as a polynomial function best fit the data. We selected the best fitting mean response model of the outcome, controlling for predictors of severity and risk factors of treatment resistance. Rheumatoid factor positivity, baseline disease activity (DAS28), baseline functional disability (HAQ), disease duration, and inappropriate response on a previous anti-TNF agent were considered confounders a priori and were forced into the models. We tested other covariates using a backward stepwise selection approach. Additional covariates were included in the model only if they were found to be significant predictors or substantial confounders, using the 10% change in estimate criteria. All patients receiving TNF inhibitors were included in the analysis, although patients without followup data only contributed information to baseline. Because incomplete followup was generally due to recent initiation of anti-TNF therapy, we assumed absent followup data to be missing at random. Moreover, <5% of covariates were sporadically missing; to minimize potential bias, we used the population average of the respective covariates as a substitute.


Baseline characteristics.

At the end of November 2005, 1,571 patients with RA with at least 1 anti-TNFα injection and at least 1 set of followup data were included in the SCQM-RA database. Of these, 344 were ≥65 years of age (ERA) and 1,227 were <65 years of age (YRA), with age ranging from 16 to 64 years in YRA and 65 to 88 years in ERA. Mean ± SD followup time was 22 ± 15 months.

As shown in Table 1, the percentage of female patients and the mean number of DMARDs prior to the introduction of anti-TNF therapy were similar in the 2 groups. The percentage of inappropriate responses on a previous anti-TNF agent was also similar in both groups. However, disease duration and rheumatoid factor positivity were significantly higher in ERA than in YRA. Furthermore, some differences surfaced in the type of concomitant DMARD treatments administered with anti-TNF agents. ERA patients received more glucocorticoids (60% versus 49%) and sulfasalazine (7% versus 3%) than YRA patients, whereas YRA received more methotrexate (42% versus 35%) than ERA. Some differences were also noted in disease activity and functional status. Indeed, mean ± SD DAS28 scores (4.5 ± 1.5 versus 4.2 ± 1.5; P < 0.001) and HAQ scores (1.4 ± 0.6 versus 1.2 ± 0.7; P < 0.001) were slightly higher in ERA than in YRA; however, from a clinical perspective both groups were quite similar. Interestingly, patient-perceived disease activity (RADAI) was slightly higher in YRA than in ERA (mean ± SD 4.3 ± 2.2 versus 4.1 ± 2.0; P = 0.054), which probably reflects different health norm expectations between the 2 groups.

Table 1. Baseline characteristics*
CharacteristicAge groupP
YRA (n = 1,227)ERA (n = 344)
  • *

    YRA = young rheumatoid arthritis (<65 years old); ERA = elderly rheumatoid arthritis (≥65 years old); IQR = interquartile range; RA = rheumatoid arthritis; RF = rheumatoid factor; DMARDs = disease-modifying antirheumatic drugs; TNF = tumor necrosis factor; DAS28 = Disease Activity Score based on 28 joints (higher scores indicate more disease activity); HAQ = Health Assessment Questionnaire (higher scores indicate more functional disability); RADAI = Rheumatoid Arthritis Disease Activity Index (higher scores indicate more disease activity).

Age, median (IQR) years51 (41–58)71 (67–74)< 0.001
Female sex, %
Duration of RA, mean ± SD years11.5 ± 8.814.3 ± 10.0< 0.001
Positive RF, %79.989.0< 0.001
Traditional DMARDs prior to anti-TNF agents, no. median330.096
Type of anti-TNF agents, no. (%)   
 Infliximab339 (27.6)80 (23.2)0.11
 Etanercept465 (38.3)144 (42.1)0.21
 Adalimumab423 (34.4)120 (34.8)0.89
Concomitant medication, no. (%)   
 Methotrexate515 (42.0)121 (35.2)0.02
 Leflunomide146 (11.9)39 (11.6)0.89
 Sulfasalazine40 (3.3)26 (7.6)< 0.001
 Hydroxychloroquine42 (3.4)7 (2.0)0.19
 Other immunosuppressive drugs15 (1.2)5 (1.5)0.53
 Glucocorticoids599 (48.8)206 (59.9)< 0.001
Failed previous anti-TNF agents, no. (%)207 (16.9)62 (18.0)0.61
DAS28, mean ± SD 4.2 ± 1.54.5 ± 1.5< 0.001
HAQ, mean ± SD 1.23 ± 0.711.40 ± 0.68< 0.001
RADAI, mean ± SD 4.3 ± 2.24.1 ± 2.00.054

Drug discontinuation.

Drug discontinuation rates were similar in both populations of patients with RA (Figure 1). The median half life of anti-TNF therapy in these cohorts was ∼3 years (3.08 years for ERA and 3.04 years for YRA). This result was consistent in all geriatric age subgroups. To further analyze the specific reasons of drug discontinuation, we reviewed the charts of 330 patients with RA in 3 representative geographic regions (Geneva, Zurich, and Morges). We found 128 incidences of permanent anti-TNF therapy discontinuation (Table 2). Although the number of events was limited, cancer was significantly more frequent in ERA than in YRA (7.1% versus 0%; P < 0.05). There was also a trend toward a lower rate of discontinuation due to allergic reactions in ERA compared with YRA (10.7% versus 20%). When causes of transient interruptions of anti-TNF agents were included in the analysis, allergic reactions were significantly more frequent in YRA than in ERA (P = 0.04). We did not observe any difference in the rate of serious infections between both groups of patients. The reasons for drug discontinuation (21 events) listed under the term any other cause included patient choice (3 occurrences), nausea (2 occurrences), and various other rare causes (e.g., hallucination, epigastric pain, cirrhosis, pregnancy, lupus-like reaction). Of note, remission was mentioned as a reason for drug discontinuation in only 1 ERA patient treated with adalimumab.

Figure 1.

Discontinuation rates of anti–tumor necrosis factor agents between elderly patients and young patients with rheumatoid arthritis. No difference was seen between the 2 groups.

Table 2. Reasons for permanent discontinuation of anti–tumor necrosis factor agents*
 Age groupP
YRA (n = 100)ERA (n = 28)
  • *

    Values are the number (percentage) unless otherwise indicated. YRA = young rheumatoid arthritis (<65 years old); ERA = elderly rheumatoid arthritis (≥65 years old).

Lack of efficacy60 (60)14 (50)0.39
Allergic reactions20 (20)3 (10.7)0.40
Severe infection12 (12)4 (14.3)0.75
Cancer0 (0)2 (7.1)0.05
Cardiovascular disease0 (0)1 (3.6)0.21
Other reasons14 (14)7 (25)0.25

Evolution of disease activity and function.

DAS28 values decreased significantly 6 months after the initiation of anti-TNFα therapy and the magnitude of this effect was similar in both groups of patients (Figure 2). Similar improvements were also observed in both groups at 1 year (−0.63 in ERA versus −0.59 in YRA) and 2 years (−0.65 in ERA versus −0.58 in YRA). EULAR response criteria at 1 year were somewhat different between ERA and YRA, with less ERA patients classified as good responders (7.2% versus 11.2%; P < 0.05) and more ERA patients classified as poor responders (60.2% versus 51.5%; P < 0.01). The rate of moderate responders (37% in YRA versus 32% in ERA; P = 0.11) was similar in both groups. These results were consistent in all geriatric age subgroups. To further examine the clinical response according to age, we repeated the analysis using the RADAI score as an outcome measure. Consistent with the results of the DAS28, we did not find any significant difference in the evolution of RADAI between ERA and YRA (P = 0.28).

Figure 2.

Change in rheumatoid arthritis (RA) disease activity (Disease Activity Score in 28 joints [DAS28]) after initiation of anti–tumor necrosis factor agents. Similar improvement was observed in elderly patients and young patients with RA.

HAQ scores decreased in both groups after the initiation of anti-TNF therapy. However, at 6 months, the effect size was significantly smaller in ERA compared with YRA (mean ± SD −0.07 ± 0.02 versus −0.09 ± 0.01), which was further confirmed at 1 year (−0.08 ± 0.02 versus −0.12 ± 0.02) and 2 years (−0.02 ± 0.03 versus −0.10 ± 0.02) (Figure 3). The exploratory analysis of HAQ values in geriatric age subgroups revealed the absence of effect of anti-TNF agents on functional score in the 74 patients ages >75 years (ΔHAQ −0.01 at 6 months and 0.03 and 0.2 at 1 and 2 years, respectively), whereas subgroups of patients ages 65–70 years and 71–75 years were similar to YRA (Figure 4).

Figure 3.

Change in functional disability (Health Assessment Questionnaire [HAQ]) after initiation of anti–tumor necrosis factor agents between elderly patients and younger patients with rheumatoid arthritis. The change was significantly different over time (P < 0.001).

Figure 4.

Change in functional disability (Health Assessment Questionnaire [HAQ]) after initiation of anti–tumor necrosis factor agents for 5-year age categories. Patients ages >75 years were significantly different from the 3 other groups.


Despite the fact that elderly individuals represent a substantial percentage of patients with RA, specific information regarding the efficacy and tolerability of classic and biologic DMARDs is scarce and derives mainly from highly selected patients included in clinical trials. In this article, we report the results obtained from a large population-based cohort of patients with RA treated with anti-TNF inhibitors. The most important finding was that drug discontinuation rates, an overall measure of effectiveness and tolerance, and the evolution of RA disease activity following the introduction of TNF inhibitors were similar in ERA and YRA. However, we observed that evolution of functional disability (HAQ) differed between ERA and YRA with more favorable outcomes in YRA. Subsequent analyses of age subgroups revealed that this discrepancy was largely explained by the absence of substantial functional improvement in patients older than 75 years, whereas their DAS28 response was comparable with that of younger patients. A higher number of comorbidities in elderly individuals, in particular the presence of osteoarthritis, may have a negative influence on function in this population. This suggests that changes occurring with age may influence the effectiveness of RA treatment and that conclusions obtained from younger patients may not be readily transferable to older patients. Furthermore, these findings also suggest that geriatric age in RA may truly start only after 75 years of age.

We found only minor differences regarding specific causes of permanent treatment discontinuation. A higher rate of cancer among elderly patients was expected, because this finding may be a consequence of increasing age. However, although the difference was statistically significant, the number of events was very low. Of note, both groups of patients had solid tumor cancer (breast with bone metastasis and pancreas), which has not been specifically associated with the use of anti-TNF treatment so far (14). Cardiovascular events resulting in discontinuation were not significantly higher in older patients. In contrast, although not significantly different, the percentage of discontinuation for allergic reaction was lower in ERA compared with YRA (10% versus 20%). In an exploratory analysis, this difference became significant when transient anti-TNF interruptions were also included (4). Only 3 studies have been published on the tolerance to and the effectiveness of anti-TNF therapies in ERA. Bathon et al (3) and Fleischmann et al (4) reported post hoc analysis encompassing randomized and open-label trials on etanercept and found no difference between ERA and YRA. Nevertheless, because efficacy was analyzed only in patients achieving 1 year of therapy (4), direct comparison with our results is difficult. The overall rate of side effect was similar in both populations; however, consistent with our results, Fleischmann et al reported a significantly lower rate of allergic reaction in ERA (4). This finding could be related to modifications of the immune system with aging (15). A third, small cross sectional study compared DMARD therapies (including anti-TNF agents) in 49 patients older than 70 years matched to 49 patients younger than 60 years (5). They found no difference in disease activity on global assessment, morning stiffness, or rates of discontinuation of anti-TNF agents.

In our population, baseline DAS28 and HAQ scores were significantly higher in ERA than in YRA. Although the difference in DAS28 scores may not be clinically relevant, it has induced some differences in the percentage of EULAR good responders. Because EULAR response criteria are dependent on the absolute level of activity reached and the extent of change, a similar effect size between 2 populations with different baseline levels of activity may have induced some differences in the percentage of patients that could be classified as good responders. The results obtained with the DAS28 were confirmed by the use of the RADAI, a self-administered questionnaire on disease activity. The differences in baseline DAS28 score also suggest that physicians are more reluctant to introduce TNF inhibitors in ERA than in YRA due to potentially higher risk of adverse reaction and cost-effectiveness consideration. A previous study reported a similar difference with traditional DMARDs (16) and is in line with many published recommendations on the use of DMARDs in elderly patients (17–19). In contrast, a recent study did not find any bias regarding the use of aggressive therapy in elderly patients (5). However, this study was conducted in a tertiary center in patients with remarkably low HAQ scores at treatment initiation (mean 0.82 for ERA and 0.73 for YRA), suggesting that these results may not be generalized to other populations.

Some inherent limits to observational studies should be underlined. Although the mean duration of RA in our patients was >10 years, diagnostic criteria were not recorded and we can only assume that diagnoses were correctly established by the rheumatologist in charge. In a similar cohort, it was found that 98% of all patients diagnosed with RA by their rheumatologist fulfilled the American College of Rheumatology (formerly the American Rheumatism Association) 1987 classification criteria (20,21), thus the probability of a correct diagnosis is very high. The mean duration of followup is relatively short (2 years), which limited our ability to analyze the effect of treatment for longer duration. In contrast, drug survival is presented on a longer time frame because retention rates represent the percentage of the population still receiving therapy at each time point. The retention rates could potentially be biased by patients stopping their medication because of sustained remission, if this phenomenon was differential between YRA and ERA. In a representative subset of this study, interruption of TNF-inhibitors was only exceptionally due to permanent remission (1 case in 128 discontinuations), which makes significant bias very unlikely. Missing data is another concern with observational studies. To address this, we previously checked the accuracy of medication data provided by the physicians, including start and stop dates, against records from the pharmaceutical industry and participants' self report (8). We included all patients with RA receiving anti-TNF agents, including some patients missing a followup visit. By far, the most common reason for incomplete followup was the recent initiation of anti-TNF agents, with insufficient time for followup assessment. Baseline disease characteristics of patients without complete followup were similar to those included in the analysis (data not shown), suggesting that patients in the analysis were a representative sample of the entire population.

In conclusion, the results from this large population-based cohort study demonstrate that anti-TNF agents can be administrated to elderly patients with RA with similar levels of effectiveness and tolerability as in younger patients. Therefore, age by itself should not interfere with therapeutic decisions concerning the introduction of anti-TNF agents. However, one should not infer from this study that anti-TNF agents should be used without discernment in ERA. Besides obvious life-threatening comorbidities, the coexistence of osteoarthritis and its role in functional handicap should also be precisely evaluated because it may limit the benefit patients can expect from these drugs.


Dr. Gabay 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 design. Genevay, Finckh, Gabay.

Acquisition of data. Genevay, Ciurea, Chamot, Kyburz.

Analysis and interpretation of data. Genevay, Finckh, Gabay.

Manuscript preparation. Genevay, Finckh, Ciurea, Kyburz, Gabay.

Statistical analysis. Finckh.


We are grateful to the SCQM staff for data management and support and to participating physicians and patients who made this study possible. We thank specifically those rheumatologists who enrolled large numbers of these patients (10 or more): P. Wiedersheim, St. Gallen; T. Lehmann, Bern; H. Tinner, Weinfelden; C. Zenklusen, Neuchatel; B. Müller, Sarnen; T. Cunningham, Geneva; J. L. Meier, Delémont; Th. Langenegger, Unterägeri; G. Thiebaud, Geneva; J. B. Suter, Bern; F. Wicht, Solothurn; M. Saxer, Basel; and P. Pancaldi, Muralto. We also especially thank participating rheumatology clinics who registered large numbers of these cases (20 or more): Divisions of Rheumatology of the University Hospital of Zurich; the University Hospital of Vaud; the University Hospital of Bern; Kantonsspital Aarau; Kantonsspital Luzern; Felix-Platter-Spital, Basel.