Mortality in established rheumatoid arthritis

Rheumatoid arthritis (RA) is associated with reduced life expectancy. Whether the development of RA initiates this process of premature ageing or is part of it is not clear. The excess mortality is apparent within the first few years of disease and increases with RA disease duration. Most of the excess deaths are attributable to infection, cardiovascular disease (in particular coronary heart disease) and respiratory disease. Deaths due to lung cancer and non-Hodgkin's lymphoma, but not other cancers, are also increased. There is some evidence that effective disease-modifying therapy can improve survival but, overall, survival in RA patients has not improved to the same degree as in the general population over recent decades.

Introduction

It has been clear since the 1950s that patients with rheumatoid arthritis (RA) experience premature mortality. The earliest mortality study¹ was published before the introduction of the 1958 American Rheumatism Association (ARA) diagnostic criteria² and included some males with ankylosing spondylitis. Most studies published since 1958 have used the ARA criteria but have varied as to whether they included patients with probable RA. Studies also varied in the method of selection of patients, the method of establishing the cause of death, the length and completeness of follow-up, the choice of controls and in the method of statistical analysis. In particular, a number of early studies used proportional mortality rather than standardised mortality ratios (SMR). Nevertheless, the great majority of these studies (Table 1) suggested that patients with RA, from whichever setting they were recruited, died earlier than members of the general population. The pattern of causes of death led Eric Bywaters, in 1961, to state, ‘Patients with rheumatoid arthritis died from the same causes as the general population but at an earlier age.’³ This observation was accepted for many years and it is only in the last 5–10 years that rheumatologists have begun to ask why RA patients experience what might be called premature ageing. Is it that the burden of chronic inflammatory activity leads to a faster ageing process, or is RA a manifestation of a premature ageing process that was already taking place within the individual? If disease activity itself contributes to premature mortality, why has improved treatment not led to improved survival? This chapter addresses these issues.

There have been many excellent reviews of the early mortality papers published in RA (e.g. Grimstadt-Kvalvik 1996; Ref. 4). This chapter, therefore, focuses on literature published since 1990. It examines the available literature on trends in mortality in RA patients and attempts to summarise the growing literature on cardiovascular mortality in this disease. It then looks at predictors of mortality within rheumatoid cohorts and finally explores the influence of treatment, in particular methotrexate and the anti-tumour necrosis factor (TNF) agents, on survival.

It is important to emphasise some of the pitfalls when comparing studies that are based on SMRs. The SMR is calculated by dividing the observed number of deaths by the number of deaths that would have been expected in a cohort of the same age and sex distribution recruited from the general population. In each case, the general population used should be that from which the RA subjects have been drawn. Background mortality rates in the general population vary by geographical site, calendar time and ethnic group. It is, therefore, not appropriate to make direct comparisons between, for example, an SMR calculated for a cohort of RA patients in London with one in the African Bush. The age- and sex-specific mortality rates in the two RA populations might be identical but the SMRs will be very different. Similarly, the SMR is influenced by the length of follow-up. The longer the length of follow-up, the more likely the SMR is to approach 1. This is because all individuals eventually die and if the length of follow-up is extended, say for 100 years, then the SMR will be 1. SMRs are, therefore, of great value and interest when trying to understand the mortality experience of an individual cohort but of less value when trying to compare the experience between cohorts.