Article Text


THU0184 Polymorphonuclear cell counts (pmn) effectively predict mortality in rheumatoid arthritis (ra) patients followed for up to 20 years
  1. F Wolfe1,
  2. HK Choi2
  1. 1National Data Bank for Rheumatic Diseases, Arthritis Research Center Foundation, Wichita, USA
  2. 2Rheumatology Unit, Harvard Medical School, Boston, USA


Background Mortality is increased in rheumatoid arthritis (RA), and this increase is related to disease severity markers, disease activity and psychosocial factors. It has been previously reported that white cell counts (WBC) predict total joint replacement in RA,1 and WBC levels are associated with cardiovascular mortality in non-RA populations.2 The effect of WBC, especially PMN, on mortality has not been studied in RA, but might be clinically important.

Objectives To investigate the relation between PMN and mortality in RA.

Methods During a 20-year period ending in 2000, 1500 consecutive RA patients had 21,581 clinic visits at which 18,850 WBC tests were performed. 544 patients died. The predictability for mortality of PMN and other variables obtained 1) during the first 2 years of follow-up and 2) over the entire course of RA was examined using Cox regression models (Cox). To assess the impact of covariates on the predictability of PMN, we controlled for RF, ESR, age, disease duration, sex, HAQ, and prednisone use in multivariate Cox models. Generalised estimating equations (GEE) were used to describe the relation of PMN to other variables.

Results The mean (SD) of WBC and PMN was 8.0 (2.8) and 6.0 (2.5), respectively. Total PMN but not lymphocyte counts predicted mortality (p < 0.05 and p = 0.522, respectively). The mean PMN values in its quartiles were 3.5, 4.9, 6.2 and 9.0. Over the entire course of RA the hazard ratios (HR) for mortality compared with Q1 were Q2 1.5 (1.1,1.9), Q3 1.9 (1.4,2.5) and Q4 2.8 (2.2,3.7). By contrast, the HR for rheumatoid factor positivity (RF) was 1.6 (1.3, 2.1). HR for PMN during the 1st 2 years of follow-up were Q2 1.5 (1.1,1.9), Q3 2.0 (1.6, 2.6) and Q4 2.4 (1.8, 3.1). For those in Q1-Q4 during the 1st 2 years, the predicted 25% time to death was 11.7, 8.8, 7.2 and 4.6 years, respectively. To assess the effect of covariates on the predictive ability of PMN, covariates that included RF, ESR, age, disease duration, sex, HAQ, and prednisone use were added to the 2-year and lifetime PH models. PMN remained significant in the multivariate Cox models, and was a stronger predictor of mortality than RF or ESR. In GEE analyses, the strongest predictor of PMN was, by far, the use of prednisone, with users having 1.8 (1.6, 2.0) thousand cells increase compared to non-users. Weaker predictors of PMN included sex, HAQ, ESR, RF, disease duration, but not age.

Conclusion Total PMN predicts mortality in RA as effectively as RF, and the predictability is robust to duration of disease as well as to fixed and time-dependent disease severity covariates. This simple test appears to have been overlooked, but adds significantly to our ability to predict mortality in RA. Corticosteroids are the strongest influence on PMN among RA patients, and might be an important factor in mortality increase in RA. Treatment related PMN reduction that occurs in clinical trials of biologic agents may be a marker for increased survival, and is a candidate variable for measurement of pharmaco-economic benefit.


  1. Arthritis Rheum. 1998;41(6):1072–82

  2. N Engl J Med. 2000;343(16):1148–55

Statistics from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.