SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Objective

To investigate mortality rates, causes of death, time trends in mortality, prognostic factors for mortality, and the relationship between disease activity and mortality over a 23-year period in an inception cohort of rheumatoid arthritis (RA) patients.

Methods

A prospective inception cohort of RA patients diagnosed between January 1985 and October 2007 was followed for up to 23 years after diagnosis. Excess mortality was analyzed by comparing the observed mortality in the RA cohort with the expected mortality based on the general population of The Netherlands, matched for age, sex, and calendar year. Period analysis was used to examine time trends in survival across calendar time. Prognostic factors for mortality and the influence of the time-varying Disease Activity Score in 28 joints (DAS28) on mortality were analyzed using multivariable Cox proportional hazards models. Causes of death were analyzed.

Results

Of the 1,049 patients in the cohort, 207 patients died. Differences in observed and expected mortality emerged after 10 years of followup. No improvement in survival was noted over calendar time. Significant baseline predictors of survival were sex, age, rheumatoid factor, disability, and comorbidity. Higher levels of DAS28 over time, adjusted for age, were associated with lower survival rates, more so in men (hazard ratio [HR] 1.58, 95% confidence interval [95% CI] 1.35–1.85) than in women (HR 1.21, 95% CI 1.04–1.42).

Conclusion

Excess mortality in RA emerged after 10 years of disease duration. Absolute survival rates have not improved in the last 23 years and a trend toward a widening mortality gap between RA patients and the general population was visible. Higher disease activity levels contribute to premature death in RA patients.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Rheumatoid arthritis (RA) is an autoimmune inflammatory articular disease of unknown etiology. Most patients experience a chronic fluctuating course of disease that, despite therapy, may result in progressive joint destruction, deformity, severe disability, and even premature death (1).

Since 1953 (2), numerous studies have investigated mortality among patients with RA. Most of these studies demonstrated reduced life expectancy in RA patients compared with the general population (3, 4). However, the majority of these studies were based on cohorts of established RA (5–11). Although several reports from inception cohorts of RA patients also found excess mortality, the observed differences in death rates between the RA patients and the general populations were not substantial (12–16). In fact, a few inception cohorts did not detect any increase in mortality in early RA (17–19). This might be explained by the fact that inception cohorts are more likely to recruit the earlier and milder type of RA as well (20). However, a recently published inception cohort study suggested that RA already carries an elevated risk of mortality in the first few years of disease (16).

Overall, there seems to be little doubt that patients with RA, at least to some extent, have lower life expectancies than members of the general population of the same age and sex (4). It can be assumed that the mortality among RA patients may decline over time as a result of the considerable improvement in the treatment of RA over the past years. The time to disease-modifying antirheumatic drug (DMARD) initiation has noticeably shortened and more effective therapy regimens have become available (21). Studies examining time trends in mortality have so far not found improved survival over time (6, 12, 13), which would be an important outcome of the treatment of RA.

As in the general population, age and sex are the strongest predictors of death in RA (9, 17, 18, 22, 23). Interestingly, it is unclear whether or not mortality can be predicted by disease-related factors such as rheumatoid factor (RF) (16, 24), functional disability (18, 23), and inflammatory markers (17, 19). If disease-related factors play a role, identifying and targeting these risk factors could potentially reduce mortality in RA.

Chronic inflammation is intrinsic to RA. However, the influence of disease activity over time on mortality in RA has not yet been clarified. Previous studies mostly included measures of disease activity at only one point in time (11, 16, 19, 24–26). One study found that significantly increased mortality was related to mean disease activity expressed by the Stoke Index over 6 and 12 months of followup (25). A causal relationship between mortality and disease activity over a longer period of time, expressed by the Disease Activity Score in 28 joints (DAS28), has not been evaluated.

The objectives of this study were to determine whether patients with RA have excess mortality compared with the general population, to compare causes of death with the general population, to explore whether RA mortality improved over time, to examine disease-specific predictors for mortality, and to investigate the relationship between survival and disease activity over time.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Study design.

The present study is a prospective observational study using the Nijmegen inception cohort of early RA. This is an ongoing inception cohort at the Departments of Rheumatology of the Radboud University Nijmegen Medical Centre and the Sint Maartenskliniek in Nijmegen, The Netherlands. In The Netherlands, all of the patients with RA receive standard treatment by a rheumatologist in hospital-based outpatient clinics. General practitioners refer all of the patients with arthritis that do not respond to nonsteroidal antiinflammatory drugs (NSAIDs) within 6–12 weeks to a rheumatologist. Since January 1985, patients with the diagnosis of early RA (disease duration <1 year), according to the classification criteria of the American College of Rheumatology (formerly the American Rheumatism Association) (27), without prior use of DMARDs were consecutively included in the cohort (28). All of the patients were followed in a standardized way until death or loss to followup.

Patients.

The study sample consisted of patients with early RA who were consecutively included in the Nijmegen RA inception cohort. All of the patients with a new diagnosis of RA were asked to participate in the Nijmegen inception cohort, regardless of age or comorbidities.

For the current study, patients were followed until death or they were evaluated on March 15, 2008, or on the date last seen before the voluntary end of participation. Followup ranged from less than 1 year to up to 23 years.

Treatment.

Patients were not treated according to strict treatment protocols, but treatment decisions were left to the consideration of the rheumatologist. Therapies used to treat the cohort patients were standard of the era they were treated in, and all of the patients started DMARD treatment upon diagnosis. Between 1985 and 1995, the prevailing strategy was sequential monotherapy, commonly using methotrexate or sulfasalazine, with the addition of corticosteroids as bridging therapy and NSAIDs on the basis of need. From 1995 onward, add-on DMARD combination therapy was also used if the first DMARD failed, which was generally based on methotrexate and sulfasalazine. Since January 1, 2000, 215 patients were treated with one of the anti–tumor necrosis factor α (anti-TNFα) agents.

Assessments.

Patients were followed once in 3 months by their rheumatologists and by specially trained research nurses. Quantitative clinical and laboratory data, including disease activity measures, functional capacity measured by the Health Assessment Questionnaire (HAQ) score, medication use, and comorbidities, were collected by the research nurses for each patient during the prospective followup. IgM-RF was analyzed by an enzyme-linked immunosorbent assay, with a value of ≥10 IU/ml indicating positive RF. Data regarding weight and height were collected from the medical records in order to calculate body mass index (BMI). Smoking status at baseline was classified as ever smoking versus never smoking and was retrospectively gathered by means of questionnaires. The smoking history of the deceased patients was obtained through their surviving relatives. Comorbidity was assessed using the medical records and was defined as the presence of at least one of the following conditions: cardiovascular disease (including hypertension, angina pectoris, heart failure, myocardial infarction, and coronary atherosclerosis), cerebrovascular accident, all grades and complications of diabetes mellitus, liver failure, kidney failure, chronic obstructive pulmonary disease, asthma, and cancer.

By March 15, 2008, the death/life status of all of the patients who were included in the study was verified using the Municipal Personal Records Database, which also provided death dates. Death/life status was also verified from the 22 patients who withdrew from followup assessments. These patients were evaluated in the longitudinal analysis of the DAS28 in relation to the risk of mortality. Information on age, sex, calendar year, and cause-specific mortality rates in the general population, as well as primary and secondary causes of death in patients of the inception cohort, were provided by Statistics Netherlands (CBS) (29). Causes of death were coded according to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (30). Primary and secondary causes of death were carefully examined. In the case of comorbidities, a condition is regarded as a primary cause of death if its lethality is known to exceed the lethality of the other causally unrelated condition; if a lethal condition can be regarded as a complication of a lethal underlying disease, the underlying disease was regarded as the primary cause of death. The study was approved by the local Ethics Committee.

Statistical analyses.

In order to investigate excess mortality in patients with RA compared with the general population, Kaplan-Meier survival analyses were used in which the survival of the RA group was compared with the expected survival based on death rates of the general population of The Netherlands, matched for 5-year age groups, sex, and calendar year. According to data of CBS, death rates from the region of Nijmegen (Gelderland) do not differ from the death rates of The Netherlands. The standardized mortality ratio (SMR) was estimated for the cohort and observed and expected death rates were compared for important causes of death.

Time trends in 5-year and 10-year survival were investigated using a novel form of survival analysis called “period analysis” to derive more up-to-date estimates of survival (31). In period analysis, all observation times falling in a calendar period are used; as a consequence, observation time can be evaluated to the right and to the left. In classic cohort analysis, observation time of the patients included in a calendar period is used and observation time is only evaluated to the right. Therefore, in cohort analysis, an improvement in survival (e.g., from 2000 onward) is partitioned over all of the past cohorts and observations of trends in survival lag behind. For 5-year survival, the periods 1990–1994, 1995–1999, 2000–2004, and 2005–2007 were compared regarding absolute and relative survival rates. Period analysis of 10-year survival started with 1995–1999, because by then the first patients had completed 10 years of followup. Period-, age-, and sex-specific mortality rates of the Dutch population were retrieved from the database of CBS.

Prognostic factors for mortality were analyzed by univariate and multivariate Cox proportional hazards models, including sex (1 = male), age, RF, DAS28, HAQ, BMI, smoking status (1 = ever smoked), and comorbidity (≥1 condition). All of the variables were measured at inclusion. Age, DAS28, HAQ, and BMI were kept continuous, whereas sex, RF, smoking, and comorbidity were dichotomous. The final prediction model was achieved using a backward stepwise selection approach with inclusion and exclusion P values of less than or equal to 0.05 and greater than 0.05, respectively.

Cox regression was used to study the relationship between the time-varying DAS28 and survival among RA patients. The DAS28 was put in the model as a segmented time-varying covariate linking the DAS28 to time. Potentially confounding factors suspected to be associated with both mortality and the DAS28 included age, sex, RF, smoking status, baseline BMI and HAQ score, and the presence of ≥1 comorbid condition during followup. Variables were added one by one to the Cox proportional hazards model with the time-varying DAS28 as an independent factor and survival as a dependent variable. If the added factor changed the crude effect of the estimate of the time-averaged DAS28 on mortality with more than 10%, it was considered a confounder and was included in the model. One-way interactions were examined for evidence of important effect modifiers. The hazard ratio (HR) resulting from the model should be interpreted as the proportional change in the risk of death associated with a 1-unit change in DAS28. The proportional hazards assumption of categorical variables was tested by a visual inspection of the baseline hazard function of each category of the variable. The nonproportionality of continuous variables was graphically evaluated using the partial residual method. All of the variables used in the Cox regression satisfied the proportional hazards assumption. Missing data at baseline were imputed using linear regression with an error term, based on the assumption that the data were missing at random.

All of the statistical analyses were carried out using the statistical software packages SPSS, version 14.0 (SPSS, Chicago, IL), and SAS, version 8.2 (SAS Institute, Cary, NC).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

Mortality rate.

Between January 1985 and October 2007, 1,081 patients presenting with early RA had entered the study. Of them, 27 were excluded because they eventually received another diagnosis and 5 were excluded because their death/life status was unknown and as they were lost to followup immediately after inclusion, their followup could not be evaluated. In total, 1,049 patients were included in the analyses. Table 1 shows the baseline characteristics of the study group.

Table 1. Baseline characteristics of the rheumatoid arthritis study sample (n = 1,049)*
VariableValue
  • *

    Values are the median (range) otherwise indicated. RF = rheumatoid factor; DAS28 = Disease Activity Score in 28 joints; ESR = erythrocyte sedimentation rate; BMI = body mass index.

  • Includes hypertension, cardiovascular disease, cerebrovascular accident, diabetes mellitus, liver failure, kidney failure, chronic obstructive pulmonary disease, asthma, and cancer.

Women, no. (%)677 (64.5)
Age at study start, mean ± SD years55 ± 13.9
Seropositive RF, no. (%)809 (77.8)
DAS28, mean ± SD5.1 ± 1.3
 ESR, mm/hour30 (1–135)
 Number of swollen joints9 (0–26)
 Number of tender joints6 (0–28)
 Patient assessment of general health46 (1–100)
Health Assessment Questionnaire score0.63 (0–2.75)
BMI, mean ± SD kg/m226 ± 4.1
Smoking (ever), no. (%)544 (51.9)
Comorbidity (≥1 condition), no. (%)240 (22.9)

Variables with missing values at baseline included RF (n = 9), DAS28 (n = 69), erythrocyte sedimentation rate (n = 25), number of swollen joints (n = 65), number of tender joints (n = 65), patient assessment of general health (n = 77), HAQ score (n = 96), BMI (n = 70), and smoking (n = 238).

The mean ± SD duration of followup was 9.0 ± 5.7 years. Of the total study group of 1,049 patients, 207 patients died at a mean age of 75 years during the period of observation; 95 men at age 73 years and 112 women at age 76 years.

Figure 1 shows survival functions, including 95% confidence intervals (95% CIs), of the RA cohort compared with the general population in The Netherlands, matched for age, sex, and calendar year. The survival curves started to diverge at ∼10 years of followup and reached statistically significant difference from the general population after 12.7 years of followup, where the death rate of the RA patients became higher than that of the general population. After 20 years, the estimated all-cause SMR was 1.40 (95% CI 1.09–1.77), indicating a 40% increased risk of mortality compared with the general population. Excess mortality in female RA patients could be detected 2 years earlier in the disease course than excess mortality in male RA patients (data not shown). An age-adjusted HR of 1.68 (95% CI 1.28–2.21) shows an increased mortality risk in male patients compared with female patients.

thumbnail image

Figure 1. Kaplan-Meier survival curves of 1,049 patients with recent-onset rheumatoid arthritis compared with the general population of The Netherlands, matched for age, sex, and calendar year.

Download figure to PowerPoint

Primary causes of death.

The causes of death are summarized in Table 2. In 6 patients, the cause of death could not be verified. Cardiovascular disease represented the leading cause of death, followed by malignancies, infections, and respiratory diseases. After circulatory diseases, in men, the most important causes of death were malignancies and infections, while in women, infections came in second place, followed by malignancies (data not shown). The observed rate of death did not exceed the expected rate of death caused by coronary heart disease (HR 0.84, 95% CI 0.57–1.19), myocardial infarction (HR 1.21, 95% CI 0.78–1.81), and malignant neoplasm in total (HR 0.83, 95% CI 0.64–1.07). The observed rate of infections (HR 8.8, 95% CI 5.4–13.7) and lymphoid malignancies (HR 2.3, 95% CI 1.12–4.3) as the primary cause of death was significantly elevated in the inception cohort.

Table 2. Primary causes of death in the inception cohort
 No. (n = 207)
Cardiovascular73
 Heart disease48
  Coronary heart disease31
   Acute myocardial infarction24
   Chronic ischemic heart disease7
  Heart failure6
  Other heart disease11
 Cerebrovascular16
 Peripheral vascular7
 Pulmonary embolism2
Malignancies60
 Gastrointestinal18
 Lung cancer17
 Leukemia/lymphoid malignancy9
 Urogenital5
 Breast cancer4
 Brain cancer1
 Unknown6
Infection20
 Pneumonia10
 Empyema2
 Sepsis4
 Infection vascular graft1
 Infection joint prosthesis1
 Urinary tract infection1
 Decubitus wound infection1
Respiratory18
 Chronic obstructive pulmonary  disease/emphysema14
 Interstitial fibrosis1
 Acute respiratory failure3
Gastrointestinal11
 Liver/biliary disease3
 Intestinal disease6
 Gastric ulcer/hemorrhage2
Neurologic10
 Dementia6
 Parkinson's disease2
 Other2
Diabetes mellitus1
Rheumatoid arthritis2
Hematologic1
Injury/accident5
Unknown6

Period analysis.

Results of the period analysis, as shown in Figure 2, demonstrated that there was no clear trend of improvement in absolute or relative survival over time. The relative survival curves (Figure 2) imply that survival rates in RA patients show a declining trend in 2005–2007 as compared with the general population, although the results were not statistically significant.

thumbnail image

Figure 2. Period analysis: absolute and relative (compared with the general population) 5- and 10-year survival rates per followup period. Five-year survival: 1990–1994, proportion of deaths 7% and number of living 364; 1995–1999, proportion of deaths 8% and number of living 626; 2000–2004, proportion of deaths 4.5% and number of living 826; 2005–2007, proportion of deaths 9% and number of living 913. Ten-year survival: 1995–1999, proportion of deaths 19% and number of living 626; 2000–2004, proportion of deaths 19% and number of living 826; 2005–2007, proportion of deaths 18% and number of living 913.

Download figure to PowerPoint

Cox proportional hazards model in ordinary survival analysis also showed that calendar year was not a significant predictor of survival (P = 0.51), corrected for age and sex. Together, these analyses demonstrated no evidence of improvement in survival over time.

Baseline predictors of mortality.

The results of the analysis considering baseline predictors for mortality are shown in Table 3. Using univariate Cox proportional hazards models, significant predictors at baseline were age, sex, BMI, DAS28, and HAQ score. BMI and DAS28 were not statistically significant predictors for death in the multivariate analyses (Table 3), whereas RF and HAQ score appeared to be. The predictive value of BMI and DAS28 in the univariate model could be explained by age, and the actual predictive value of RF was revealed after adjusting for age. Smoking status did not show predictive value, regardless of which analyses were performed. Male patients appeared to have a considerably lower survival probability than female patients. Higher age, positive RF at onset, a higher HAQ score, and comorbidity (≥1 condition) at study entry showed an independent negative influence on life expectancy.

Table 3. Univariate and multivariate reduced Cox regression models of the predictive value of baseline characteristics for mortality*
 Univariate modelMultivariate model
HR (95% CI)PHR (95% CI)P
  • *

    HR = hazard ratio; 95% CI = 95% confidence interval; HAQ = Health Assessment Questionnaire; RF = rheumatoid factor; BMI = body mass index; DAS28 = Disease Activity Score in 28 joints.

  • Includes hypertension, cardiovascular disease, cerebrovascular accident, diabetes mellitus, liver failure, kidney failure, chronic obstructive pulmonary disease, asthma, and cancer.

Age, years1.11 (1.09–1.12)< 0.0011.10 (1.09–1.12)< 0.001
Sex (male)1.72 (1.31–2.26)< 0.0011.90 (1.43–2.52)< 0.001
HAQ score2.10 (1.72–2.56)< 0.0011.46 (1.19–1.79)< 0.001
Comorbidity (≥1 condition)2.84 (2.16–3.74)< 0.0011.83 (1.38–2.42)< 0.001
RF positivity1.00 (0.73–1.39)0.9811.47 (1.05–2.05)0.023
Smoking (ever)1.30 (0.96–1.77)0.090  
BMI, kg/m21.03 (1.00–1.06)0.080  
DAS281.31 (1.17–1.46)< 0.001  

Effect of disease activity over time on survival.

The effect of disease activity over time on survival is shown in Figure 3. DAS28 values were missing to some extent in 56 patients (5–60% of DAS28 values during followup, with 5 patients having 40–60% missing) and were imputed. The mean time-averaged DAS28 was 3.7 (range 0.8–7.3). The relationship between DAS28 and mortality is displayed in the upper half of Figure 3. For the purpose of the figure, time-averaged DAS28 values were used to divide patients into high (above the median) and low (below the median) disease activity. In the crude analysis, the DAS28 was handled as a continuous variable (lower half of Figure 3). Of all of the variables tested for confounding, including age, sex, RF, smoking status, BMI, and HAQ assessed at baseline, and comorbidity (≥1 condition) assessed throughout the followup, only age and baseline HAQ score appeared to be confounders. The analyses were also performed separately for men and women, as sex acted as an effect modifier. Both the unadjusted and the adjusted HRs indicate a greater influence of disease activity on the mortality risk in male patients than in female patients (Figure 3). The average disease duration at death for male patients with a DAS28 above the median of 3.72 was 6.5 years, and with a DAS28 below the median, 11 years. In female patients, this difference was 9 years as opposed to 11 years, respectively.

thumbnail image

Figure 3. Cumulative survival rates in rheumatoid arthritis patients with high versus low disease activity, divided by the median time-averaged Disease Activity Score in 28 joints (DAS28). Results are shown separately for men and women. 95% CI = 95% confidence interval; HAQ = Health Assessment Questionnaire.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

In this prospective study, the mortality in patients with RA was analyzed in an inception cohort. According to the results of this study, excess mortality emerged after 10 years of disease duration, with cardiovascular disease as the most common cause of death. Despite improved therapies in the last decade, there was no improvement in survival, at diagnosis survival was predicted by age and sex, RF, HAQ, and the presence of comorbidities, and higher inflammation over time was associated with worse survival.

During the first 10 years of followup, the observed mortality rate was comparable with the expected mortality, whereas survival rates of RA patients declined in comparison to the general population, pointing to excess mortality in RA. This finding is in agreement with previous studies that also found no substantial increase in mortality early in RA (17–19). In contrast, Young et al (16) found excess mortality only in the first 7 years of the disease in a British early RA cohort. Different calculation of observed and expected survival rates and a different composition of the reference population may account for the discrepancies between our study and the British study.

Altogether, the results are in line with the predominant evidence that RA is associated with shorter life expectancy than the general population (13, 32, 33). Life expectancy in this RA cohort was reduced on average, with 2 years in men and 5 years in women. This implies that although absolute survival is worse in men, compared with the general population, relative survival is more reduced in female than in male RA patients. A generally higher age at onset of RA in men could explain this (34). Similar to the general population, cardiovascular pathologies were the leading causes of death in female RA patients. In recent years, cancer has overtaken cardiovascular diseases as the leading cause of death in men in general (29), whereas male RA patients still primarily die of cardiovascular diseases. This finding emphasizes the impact of RA on the pathogenesis of cardiovascular diseases. In this inception cohort, the observed rate of death exceeded the expected rate when caused by infections or lymphoid malignancies. An elevated infection risk may be associated with a compromised immune system in RA, being more prone to infections (35), in combination with intensive antiinflammatory treatment from early disease on. Higher incidence of lymphoid malignancies in RA has also been found by other studies (36).

Given the remarkable declines in overall mortality rates in the general population and the considerable change in the treatment of RA over the last decades, it could be possible that survival of RA patients had improved over time. For the analysis, we used period analysis instead of cohort analysis to be able to detect early trends in improved survival. Nevertheless, we were unable to demonstrate a statistically significant change in 5- or 10-year survival, which may be explained by the finding that excess mortality appeared after 10 years of disease duration. This finding concurs with previous findings that survival in patients has not improved in recent years, despite significant advances in treatment (6, 13, 14, 37). As the survival of the general population improved over time, the mortality gap between RA patients and the general population may widen (38).

Different baseline variables were analyzed for their prognostic value for mortality. Male sex, higher age, seropositive RF, higher HAQ score, and comorbidity were shown to have independent predictive value with respect to the mortality of RA patients. Similar to findings in the general population, male sex and age were the strongest predictors of survival. Remarkably, Anderson (39) did not show a clear association between sex, age, and mortality in RA after a comprehensive review of 25 articles on age and mortality in RA. In accordance with the present study, the HAQ score has been considered to be a good prognostic factor for RA mortality (40–42). It is quite conceivable that excess disability, being more disabled than explained by age and sex, points to larger frailty and consequently increased mortality (23). RF (16, 25, 43) and comorbidity (13, 16, 24, 44, 45) have also been shown to be associated with reduced survival. The observation that comorbidity is a major contributor to mortality in RA should serve to focus attention on other diseases that may affect RA patients. In agreement with a study by Goodson et al (46), but in contrast to a study by Young et al (16), smoking status did not provide predictive information with respect to RA mortality. Potential misclassification of the smoking status obtained from questionnaires might form a limitation of this study. However, it has been shown that retrospective questionnaires are a valid tool to assess smoking habits and that relatives of deceased smokers can accurately report on overall smoking status (47–49). BMI and DAS28 at baseline were also not shown to be significant predictors of death among patients with RA. An explanation for the latter is that the DAS28 at onset does not adequately reflect the level of disease activity over time.

To our knowledge, our study is the first to investigate a possible causal relationship between the DAS28 over time and the survival of patients with RA. The results showed that increase in disease activity has a significant negative influence on the life expectancy of patients with RA, adjusted for age and baseline functional status. High disease activity was shown to have a greater negative influence on the survival of male compared with female RA patients.

The major strengths of this study include the length of the followup and the inception cohort design, which allows study of a wide spectrum of RA and is more likely to retain patients with definite but milder disease, some of whom achieve prolonged remission. Therefore, the varied severity and prognosis in RA will be reflected more accurately. Studies of patients with established RA may well be biased toward the more severe patients, who would be more likely to be retained in the clinical setting, partly because of comorbidity.

Missing values unfortunately always occur in observational studies. The maximum amount of missing values per variable in this study was 23% (smoking status). Regression with an error term was used to impute missing data, which has been proven to yield valid results (50).

RA patients in our cohort experienced worse survival compared with the general population, with excess mortality emerging after 10 years of disease duration. Despite significant improvement in the treatment of RA and a declining mortality in the general population, the life expectancy of patients with RA has not improved during the last 2 decades, which results in a widening mortality gap. This increased excess mortality in RA patients may be attributable to the presence of mainly cardiovascular comorbidities that were shown to have a worse outcome in patients with RA than in the general population (51). Systemic inflammation and immune dysfunction in RA seem to accelerate the pathogenesis of comorbidity and mortality. Indeed, to our knowledge, our results are the first to show that higher disease activity over time, expressed by the DAS28, significantly reduces life expectancy in RA.

The fact that intensive antiinflammatory treatment can improve cardiovascular status is shown by studies that found a reduction in the incidence of myocardial infarction in patients who responded to anti-TNFα therapy (44, 52, 53). However, the era of the new biologic therapies might just be too short to significantly influence survival in RA patients. Continuous tight disease control in all age groups of RA patients and increased attention to comorbid diseases should be able to improve the survival of patients with RA in the future.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

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 published. Dr. Radovits 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. Radovits, Fransen, van Riel, Laan.

Acquisition of data. Radovits, Al Shamma, Eijsbouts, van Riel, Laan.

Analysis and interpretation of data. Radovits, Fransen, Al Shamma, Eijsbouts, van Riel, Laan.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES

We would like to thank Professor Andre Verbeek for epidemiologic advice and Erik Brummelkamp for statistical and technical support.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. Acknowledgements
  9. REFERENCES