White blood cell count and mortality in the Baltimore Longitudinal Study of Aging

Abstract

Objectives: We investigated the secular trend in white blood cell (WBC) count and the relationship between WBC count and mortality between 1958 and 2002.

Background: The WBC count is a clinical marker of inflammation and a strong predictor of mortality. Limited data exist on the WBC count secular trend and the relationship between WBC and mortality.

Methods: One thousand eighty-three women and 1,720 men were evaluated longitudinally in the Baltimore Longitudinal Study of Aging. Blood samples and medical information were collected at the study entry and every 2 years during follow-up visits. The WBC count and all-cause, cardiovascular, and cancer mortality were assessed.

Results: A downward trend in WBC count was observed from 1958 to 2002. The secular downward trend was independent of age, gender, race, smoking, body mass index, and physical activity. The WBC count was nonlinearly associated with all-cause mortality and almost linearly associated with cardiovascular mortality. Participants with baseline WBC <3,500 cells/mm3 and WBC >6,000 cells/mm3 had higher mortality than those with 3,500 to 6,000 WBC/mm3. Within each WBC group, age-adjusted mortality rates declined in successive cohorts from the 1960s to the 1990s. Participants who died had higher WBC than those who survived, and the difference was statistically significant within 5 years before death.

Conclusions: Our study provides evidence for a secular downward trend in WBC count over the period from 1958 to 2002. Higher WBC counts are associated with higher mortality in successive cohorts. We found no evidence that the decline of age-specific mortality rates that occurred from 1960 to 2000 was attributable to a secular downward trend in WBC.

Figures

Figure 1. Longitudinal Changes in WBC Count by Years of Initial Evaluation, Separately in Men and Women

The upward trend of white blood cell (WBC) count in the oldest cohorts can be explained by participants enrolled in the first 2 enrollment periods, who developed an increase in WBC count in their very old age. Numbers in the table are for participants enrolled in different time periods and, of these, the number who were still alive at the beginning of each decade.

Figure 2. Relationship Between Excess Mortality and WBC Count in the Entire BLSA Trial Sample

The absolute difference between the observed mortality hazard and the expected mortality hazard over time is expressed as excess mortality and plotted against white blood cell (WBC) count for the entire sample. The dashed lines represent the 95% confidence intervals.

Figure 3. Kaplan-Meier Survival Curves and Proportional Hazard Survival Plots According to WBC Count Groups

(A) Kaplan-Meier curves. (B) Proportional hazard survival plots were evaluated at mean values of the explanatory variables. In participants with white blood cell (WBC) count ≤3,500/mm3, the predicted survival was extrapolated up to 20 years of follow-up, because of the absence of events after 20 years.

Figure 4. Longitudinal Changes in WBC, Neutrophil, and Lymphocyte Counts

Longitudinal changes of age- and date-adjusted (A) white blood cell (WBC), (B) neutrophil, and (C) lymphocyte counts observed in the BLSA participants according to time before death for participants who died during the follow-up, and time before censorship for those who were censored. Note that neutrophils and lymphocytes are limited to participants who had differential WBC count (n = 6,227). 95% confidence interval estimated by a bootstrapping method.