Materials and methods

Starting January 1985, all consecutive patients attending the Nijmegen or Groningen University Hospital with recent onset RA (symptoms for <1 year at study entry) according to the American Rheumatism Association criteria (ARA, 1987), were asked to participate in a long term prospective study. All patients were followed up in a standardised way and new patients were included25 up to April 1997. The patients were treated by different rheumatologists of both departments, and were seen every three months by specially trained research nurses. Quantitative clinical and laboratory data were collected, comprising the Ritchie Articular Index (RAI), the number of tender joints, number of swollen joints, ESR (according to Westergren, mm/1st h), IgM RF (enzyme linked immunosorbent assay (ELISA), normal <10 IU/ml), HLA status, and level of education. Furthermore, the patients completed a Health Assessment Questionnaire (HAQ) every six months and at each visit the disease activity score (DAS) was calculated.26 ,27 The starting date, dose, stopping date, reason for discontinuation, and side effects of the DMARDs and NSAIDs used were recorded.28 DMARDs included antimalarial drugs, methotrexate, sulfasalazine, gold, d-penicillamine, azathioprine, cyclophosphamide, and cyclosporin.

Standard methods of survival analysis (Kaplan-Meier, 1957) were used, in which the mortality of this RA group (n=622) was compared with the expected mortality based on the general population of the Netherlands, matched for age and sex.29 Patients lost to follow up were contacted by phone or mail with the help of family or general physicians to determine the death/life status of the relevant patient at April 1997. Causes and dates of death were obtained from the general practitioner (or rheumatologist who obtain dates and causes of death from the general practitioner in the Netherlands) and were coded according to the International Classification of Diseases, ninth revision, clinical modification (ICD-9-CM) medical diagnoses.

The functional capacity of patients attending the Nijmegen University Hospital (n=322), for whom Health Assessment Questionnaires were available at each year of follow up (n=249), are presented for each year of the follow up (table 2).

Prognostic factors for mortality were analysed multivariately by the Cox proportional hazards model, multivariate with age (start of RA), sex (male or female), IgM rheumatoid factor (IgM RF positive or negative), HLA status, RAI, number of swollen joints (range 0–44), mean DAS of first year of follow up (range 0–10), and ESR fixed in the model. All variables were continuous, except sex, IgM RF, and HLA status. Patients for whom all baseline characteristics were available were included in the model (n=566). The dependent variable in the model was duration of follow up. In a separate analysis the mean HAQ of the first year was included in the Cox proportional hazards model as well.

Results

MORTALITY AND FUNCTIONAL CAPACITY

By April 1997 622 patients had entered the study and were included in the various analyses. Table 1 presents the baseline characteristics of the clinical variables of the study group. Figure 1 shows the number of patients at each time point during 10 years of follow up. Mean duration of follow up was 5.8 years (range 1–10), and 109 patients were followed up for 10 years. Of the total study group of 622 patients with recently diagnosed RA, 55 patients (16% of the patients up to 10 years of follow up) died during the period of observation with a mean age at death of 73 years. The patients included had a similar prevalence of chronic comorbidities as patients with other chronic diseases, also having other comorbidities, living in the same area as the patients with RA.30 ,31 The life/death status could not be traced in two of the 35 patients who were lost to follow up at April 1997.

View this table:

• View inline
• View popup

Table 1

Baseline characteristics of the study group

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Length of follow up of patients with recent onset rheumatoid arthritis.

The causes of death ascribed by the rheumatologists or general practitioners could be classified as cardiovascular (n=26) and respiratory cause (n=6), cancer (n=17), sepsis (n=1), amyloidosis (n=2), renal insufficiency of unknown cause (n=1), perforation of the oesophagus, probably related to NSAID treatment (n=1), and pancytopenia during aurothioglucose treatment (n=1).

Figure 2 presents survival functions (Kaplan-Meyer) of the patients with RA (n=622) compared with the general population of the Netherlands, matched for age and sex. Confidence intervals (95%) are shown for the study group, as patients had to be censored for duration of follow up, because of different inclusion dates of the patients. The death rate was not significantly different from that of the general population of the Netherlands up to 10 years of RA. The survival curves started to diverge at about seven years of follow up, though this was not statistically significant.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Kaplan-Meier survival curves of patients with recent onset rheumatoid arthritis compared with the general population of the Netherlands, matched for age and sex. Confidence intervals (95%) of the study group are shown.

Table 2 shows the course of the functional capacity according to the HAQ during the complete follow up period of the patients attending the Nijmegen University Hospital. The mean value of the HAQ of these patients at entry was comparable with the mean value of the patients not included in the analysis for functional disability. The functional capacity was relatively constant during the first six years after an initial improvement from baseline. Thereafter a gradual decline in HAQ scores was seen, with the functional capacity being at the same level after 10 years of follow up as at the start.

Discussion

In this prospective study, started in 1985, the death rate of an inception cohort of patients with recent onset RA was analysed. Fifty five patients of the study group of 622 patients died during the period of observation. The observed mortality of this study group up to 10 years of follow up was comparable with the expected mortality, calculated with the aid of life tables, of the general population of the Netherlands (matched for age and sex).29 This finding is in contrast with the general belief that RA is associated with an increased death rate.33

Only three studies report the mortality in patients followed up prospectively from onset.9-12 In the study of Corbettet al a small increase in mortality, compared with population expected rates, was found (these data were not analysed statistically).9 This excess mortality increased with increasing duration of disease, but was already apparent after three years of follow up.9 The study of Rasker and Cosh, also including an inception cohort of patients with RA who satisfied the ARA criteria of definite or classical RA within one year of onset, did not compare their results with the general population (that is, the population of the UK). However, an increase in mortality, between the third and tenth year after baseline with a loss of life expectancy of at least five years was suggested.11 A striking difference between the study of Rasker and Cosh and our data is the mean age at death (73 years in this study v 68 years in the study of Rasker et al). In contrast, only the study of Lindquist and Eberhardt found no increased mortality.12

The causes of death observed in our study were comparable with the results seen in the three inception cohort studies, except for two deaths related to treatment which occurred in this study.9 ,11 ,12 In the three previous studies second line drugs were not responsible for any deaths, probably owing to the relatively small numbers of patients in those studies.9 ,11 ,12 In the study of Lindquist and Eberhardt only 62% of the patients were treated with DMARDs and 16% with oral corticosteroids at some time during follow up.12 In our study 91% of the patients were already being treated with DMARDs during the first year of follow up and 95% of the patients were treated with DMARDs at some time during follow up. In addition, 20% of the patients used oral corticosteroids at some time during the first year of observation, and 35% of the patients used oral corticosteroids at some time during the complete follow up period. In the Netherlands there is a general policy of referring patients with RA early in the disease course to a rheumatologist. Therefore none of the patients had started taking DMARDs or corticosteroids before inclusion in the study. Although it seems logical to suggest that the early use of second line treatment might lead to treatment related mortality,34this was not confirmed by our study as no increased mortality was found. In other mortality studies increased death rates in patients with RA were reported as a result of infection, especially respiratory infection, gastrointestinal and lymphoproliferative diseases.1 ,2 ,23 ,35-40 However, most of these studies had different definitions of RA, disease duration before study inclusion, and length of follow up, and most patients received steroid treatment for several years.1 ,2 ,37 ,38

The functional capacity was relatively constant during the first six years after an initial improvement from baseline. After 10 years of follow up the functional capacity had returned to the level at the start of the study. This is in contrast with the study of Corbettet al, in which disability was reported to develop most rapidly during the first year after disease onset, with a slow, nearly linear rate of increase during the period of observation.9 In the study of Eberhardt and Fex, in which no increased mortality was found in an inception cohort of patients with RA, it was concluded that functional impairment of different joints had progressed, but most patients were mildly disabled up to six years of follow up.13 In that study fewer patients were treated with second line drugs or corticosteroids during follow up, and therefore we suggest that the early use of second line treatment in our study, in almost all patients, might have led to improved functional capacity but not to increased mortality.

Different variables at onset were also analysed for their prognostic value for mortality. No univariate analysis was performed, as mortality and several disease activity variables are highly age and sex related. Therefore age would act as a confounder in a univariate analysis. In the regression model only age at start and male sex were statistically significant prognostic factors. These results are in line with the results of the study of Lindquist and Eberhardt.12 This is unsurprising, as age and sex are the strongest predictors of mortality in the general population as well. However, Anderson, who reviewed 25 articles on age and mortality in RA, failed to show a clear association between sex, age, and mortality in RA.41 Corbettet al and Rasker and Cosh did not analyse prognostic factors for mortality. Corbett et al showed that, compared with survivors, those who died had more severe disease, were more often treated with steroids, and became more disabled during the course of their illness.9 Rasker and Cosh suggested a longer life expectancy for women, a greater chance of early death for patients with a bad functional status and for patients with persistently high ESR and with strong seropositivity for RF.11 However, these suggestions were not analysed statistically.

It has long been suggested that patients with RA, especially the more severe cases, have a shortened life expectancy compared with the general population.11 In this study of an inception cohort of patients with recent onset RA we conclude that the death rate up to 10 years of follow up of patients with RA was not significantly different from the expected mortality of the general population. Therefore we suggest that the aggressive therapeutic strategies introduced since the 1980s have not only resulted in an improved disease course but also in a better long term outcome. But we are aware that other possibilities cannot be excluded—for example, some evidence suggesting that RA is getting milder and less common and the fact that patients attending hospitals are more likely to receive preventive treatment for comorbidities. However, as some studies have suggested that the mortality of patients with RA compared with the general population increases with the duration of the disease, further research (longer follow up) may be necessary.9 In this study only age at the start and male sex were statistically significant prognostic factors for mortality.