Prevalence of Neutropenia in the U.S. Population: Age, Sex, Smoking Status, and Ethnic Differences FREE

—The Editors

Neutrophils normally comprise most circulating leukocytes and are critical in providing antimicrobial activity against bacteria and fungi. An inverse quantitative relationship between neutrophil count and infection was established 4 decades ago in patients undergoing chemotherapy for acute leukemia (1). The risk for infection was highest when the neutrophil count was less than 0.5 × 10⁹ cells/L for a prolonged period. Risk is substantially reduced when the duration of neutropenia is shortened or when the neutrophil count is greater than 0.5 × 10⁹ cells/L. In healthy persons, reserves of neutrophils in bone marrow greatly outnumber circulating neutrophils, which implies that isolated minor reductions in peripheral neutrophils should not lead to increased risk for opportunistic infection (2).

Asymptomatic or benign reductions in neutrophils are observed in individuals of all ethnic backgrounds but may be more common in those of African descent. Benign ethnic neutropenia has been described in Africans (3 - 4), African-Caribbean persons (5), West Indians (6), Ethiopians, Yemenite Jews (7 - 8), and certain Arabs (7). Prevalence estimates range from 10% to more than 30%. Asymptomatic leukopenia in African-American workmen was described in 1941 (9) and has subsequently been described in larger cohorts of persons in the United States (10 - 13). Extensive evaluations of these “neutropenic” persons showed low neutrophil counts with normal subpopulations of lymphocytes and other leukocytes, normal bone marrow morphologic features and cellularity, and no increased risk for local or systemic infections (14 - 17). Because previous reports describing neutropenia in African-American persons included only military personnel (12 - 13) or patients having elective surgeries (10), or were limited to a single geographic location ((10), (18)), these findings have limited general applicability. In addition, there is no clear consensus regarding the definition of “neutropenia”; therefore, its prevalence in the United States is unknown.

The National Cancer Institute has established a neutrophil count of 1.5 × 10⁹ cells/L as the threshold for neutrophil toxicity (19). As a result, this threshold has been adopted as the minimum count required to initiate or continue treatment in many therapeutic clinical trials, and a neutrophil count less than 1.5 × 10⁹ cells/L has become the commonly accepted definition of neutropenia. Although reports on neutrophil and granulocyte counts in the U.S. population exist (20 - 21), we are unaware of any detailed analysis regarding the population-based prevalence of neutropenia. We used data from the 1999 to 2004 National Health and Nutritional Examination Survey (NHANES) to describe age-, sex-, ethnicity-, and smoking-related differences in blood counts in the United States, focusing on the neutrophil counts in black persons.

The NHANES consisted of an interview, an examination, and laboratory data that were collected from a complex, multistage, stratified, and clustered probability sample of civilian, noninstitutionalized persons. During the 1999 to 2004 examinations, black persons, Mexican-American persons, and all persons 15 to 19 years of age and older than 60 years of age were oversampled to ensure accurate estimates for these groups (22). Of the 29 608 persons 1 year of age or older who were asked to come to the mobile examination centers, 1668 did not report to the centers and 2715 had missing hematologic laboratory values. After 3 additional persons were excluded because of unrealistic values, there were 25 222 participants with complete blood counts and differentials. Smoking data were available for participants who were age 20 years or older for the first 4 of the 6 years of the study. Current smokers were defined as participants who reported smoking occasionally or every day during the past 7 days.

We collected 3 mL or 5 mL K3 EDTA anticoagulated whole blood from all persons who were 1 year of age or older by using established venipuncture protocols and procedures. Complete blood counts were performed in duplicate on a Coulter MAXM counter (Beckman Coulter, Miami, Florida) (22).

All analyses used weighted samples and considered the stratification and clustering of the design to derive estimates that were applicable to the U.S. population (22). To provide estimates for the entire 6 years, a 6-year, weight-variable sample was created by taking two thirds for the 4-year weight for each person who was sampled in 1999 to 2002 and one third for the 2-year weight for each person who was sampled in 2003 to 2004. All analyses were conducted in SUDAAN, version 9.0.1 (Research Triangle Institute, Research Triangle Park, North Carolina (23)) and SAS, version 9.1.3 (SAS Institute, Inc., Cary, North Carolina). Nonadjusted frequency distribution of leukocyte and neutrophil counts and nonadjusted prevalence of neutropenia regarding age groups, sex, and ethnicity were obtained. These age groups were commonly used in highly stratified NHANES analyses (21 - 22). We obtained means and comparisons of age-adjusted and sex-adjusted blood counts by using linear regression, in which age is modeled as a continuous variable. Adjusted means in leukocyte and neutrophil counts in smokers and nonsmokers were compared. Logistic regression adjusting for age group and sex was then used to generate predictive marginals for having neutropenia. More complex models, which involved interaction between ethnicity and sex as well as between ethnicity and age (as a continuous variable), were also tested and different categorizations of age were considered. These models did not substantially add to the explanatory power of the model.

The 25 222 NHANES participants with valid hematologic indices represented 253.2 million noninstitutionalized residents of the United States. Among 2715 participants with missing data, 35% were white, 30% were black, and 26% were Mexican American; children 5 years of age or younger accounted for 35%, those 6 to 18 years of age accounted for 43%, and adults 18 years of age or older accounted for 32%. We compared age-adjusted and sex-adjusted mean blood counts across the major ethnic groups by using linear regression. Relative to white participants, black participants had lower mean leukocyte counts (mean difference, 0.89 × 10⁹/L; P < 0.001), lower neutrophil counts (0.83 × 10⁹ cells/L; P < 0.001), and similar lymphocyte counts (0.022 × 10⁹ cells/L; P = 0.36). Mexican-American participants had slightly higher mean leukocyte counts (0.16 × 10⁹ cells/L; P = 0.014), higher neutrophil counts (0.11 × 10⁹ cells/L; P = 0.026), and higher lymphocyte counts (0.095 × 10⁹ cells/L; P < 0.001). In addition, leukocyte, neutrophil, and lymphocyte counts were lower in males than in females (Table 1). Appendix Tables 1 and 2 summarize other basic hematologic variables in the 3 major ethnic groups for sex and for the 13 age groups.

For participants who were age 18 years or older, the distribution of leukocyte and neutrophil counts suggested a downward shift of approximately 1.0 × 10⁹ cells/L among black persons compared with white persons and Mexican-American persons (Figure 1; Appendix Table 3). There was a slight upward shift for leukocyte counts among Mexican-American participants, but little difference for neutrophil counts. Similar shifts were seen in the distribution of leukocyte and neutrophil counts in participants who were younger than age 18 years (data not shown).

Mean age-adjusted and sex-adjusted leukocyte and neutrophil counts were compared for smokers and nonsmokers (Table 2). The overall mean leukocyte and neutrophil counts were higher in smokers than in nonsmokers (1.38 × 10⁹ cells/L and 0.87 × 10⁹ cells/L, respectively). Smoking had the greatest effect on increments of neutrophil counts in white participants (0.87 × 10⁹ cells/L), less of an effect in black participants (0.50 × 10⁹ cells/L), and the least effect in Mexican-American participants (0.41 × 10⁹ cells/L).

The prevalence of neutropenia (neutrophil count <1.5 × 10⁹ cells/L) differed by age, sex, and ethnicity (Figure 2; Appendix Table 4). A total of 583 participants had neutrophil counts less than 1.5 × 10⁹ cells/L. The weighted prevalence was 1.2% (95% CI, 1.1% to 1.4%), which represented an estimated 3.1 million persons in the United States. Neutropenia was present in 4.5% (CI, 3.9% to 5.0%) of black participants, 0.79% (CI, 0.57% to 1.0%) of white participants, and 0.38% (CI, 0.24% to 0.52%) of Mexican-American participants. Across ethnic groups, males were more likely to have neutropenia: 6.65% for black males versus 3.57% for black females, 0.90% for white males versus 0.59% for white females, and 0.57% for Mexican-American males versus 0.39% for Mexican-American females. The prevalence of neutropenia was highest in children younger than age 5 years, was generally lower during childhood, and reached the adult level by age 15 to 17 years in females and by age 18 to 24 years in males. We collapsed the 25- to 64-year-old age groups because the prevalence of neutropenia showed no trend in this broad age range (data not shown). For every age and sex category, black participants were more likely than white participants to have neutropenia; in most age and sex categories, neutropenia was less common among Mexican-American participants than among white participants.

Multivariate logistic regression analysis was done to estimate the prevalence of neutropenia adjusted for age, sex, and ethnicity (Table 3). Compared with white participants, black participants were more likely (P < 0.001) and Mexican-American participants were less likely (P = 0.0038) to have neutropenia. The prevalence of neutropenia was lower in childhood and adolescence and was less than 1% from age 18 to 24 years onward. Males were also more likely than females to have neutropenia (P < 0.001). There was no significant interaction between ethnicity and sex (P = 0.73) or between ethnicity and age as a continuous variable (P = 0.125). In a separate logistic regression analysis that was adjusted for age group, sex, and ethnicity, smokers had a nonstatistically significant lower percentage of neutropenia (CI, 0.22 to 1.19; P = 0.113) relative to nonsmokers.

Overall, most participants with neutropenia had a neutrophil count of 1.0 to 1.5 × 10⁹ cells/L. Eighty-nine percent of black participants, 85% of white participants, and 84% of Mexican-American participants with neutropenia had neutrophil counts of at least 1.0 × 10⁹ cells/L. The prevalence of neutrophil counts less than 1.0 × 10⁹ cells/L was 0.57% among black participants, 0.11% among white participants, and 0.08% among Mexican-American participants. The low prevalence of neutrophil counts less than 1.0 × 10⁹ cells/L was also seen across most age ranges in black participants (Table 4). These findings are consistent with the first percentile neutrophil counts in black participants (0.79 × 10⁹ cells/L and 1.17 × 10⁹ cells/L), white participants (1.18 × 10⁹ cells/L and 1.75 × 10⁹ cells/L), and Mexican-American participants (1.38 × 10⁹ cells/L and 1.87 × 10⁹ cells/L) who were younger than 20 years of age and 20 years of age or older, respectively.

Since the description of leukopenia in black workmen in 1941, several reports have confirmed that persons of African descent have substantially lower mean leukocyte counts than do white persons. This relative leukopenia is a result of lower neutrophil counts in black persons (21). The report from the 1999 to 2004 NHANES and from previous NHANES ((20 - 21), (24)) show that the lower mean neutrophil counts in black persons have persisted for more than 3 decades in the U.S. population. This difference is chiefly explained by the downward shift in neutrophil count distribution by 0.8 to 1.0 × 10⁹ cells/L. The difference in leukocyte and neutrophil counts has also been observed in other large surveys from the United States ((12 - 13), (18)) and the United Kingdom ((5 - 6), (25)). Although the relative leukopenia was initially thought to be related to nutritional deficiency ((4), (9), (26)), the persistence of this finding over time and in several geographic locations suggests that other factors explain the observed neutropenia in black persons.

Our analysis extends previous non–population-based observations that leukocyte and neutrophil counts also differ by sex and age. Males had lower leukocyte and neutrophil counts than did their female counterparts, which confirms the findings from a smaller study of 200 volunteers in the United Kingdom (27). In addition, neutrophil counts and prevalence of neutropenia were lower in younger persons. These sex- and age-related differences in blood count were consistently seen in all 3 major ethnic groups.

Ethnicity is a major contributor to variation in leukocyte and neutrophil counts. Although neutropenia has most commonly been described in black adults, the current analysis shows that neutropenia is most common among male black children. A few other reports have described a similar finding ((18), (28 - 29)), but no study has compared neutrophil counts among black, white, and Mexican-American children. The higher percentage of black children with neutropenia was greatly reduced by adulthood, such that approximately 4% of black men and 2% to 3% of black women had neutropenia. The overall prevalence of neutropenia in black participants in the current study (4.5%) was slightly lower than that for U.S. black active-duty soldiers (7.6%) (12) and considerably lower than that for Yemenite Jews (11.7%) (8), Bedouin Arabs and Ethiopian Jews (20%) (7), and Africans in Uganda (>30%) (25).

Most participants with neutropenia had neutrophil counts between 1.0 × 10⁹ cells/L and 1.5 × 10⁹ cells/L. Because of the ambulatory nature and the lengthy examination process, it is reasonable to speculate that the neutropenia was clinically benign and that these participants could be classified as having benign ethnic neutropenia. Although it is tempting to propose a change from 1 universal to ethnicity-based reference ranges for blood counts, such a proposal may be premature. Detailed clinical data about infections or symptoms related to neutropenia in the NHANES participants were not part of these analyses; thus, the benign nature of the neutropenia in these participants cannot be ascertained. Previous reports in which clinical and laboratory information were available included few participants with benign ethnic neutropenia; therefore, whether these persons truly reflect the lower end of the normal range remains to be confirmed.

As expected, the leukocyte and neutrophil counts of current smokers were higher than those of participants who never smoked. Of interest, black smokers had a smaller increase in leukocyte and neutrophil counts than did white smokers. This difference was also seen in the first NHANES, published more than 30 years ago (20). This finding is consistent with an earlier report in which the leukocyte counts of Nigerian smokers were not statistically higher than those of nonsmokers (30). The lack of a robust increase in leukocytes or neutrophils is also consistent with several reports showing that leukocytes and neutrophils were mobilized to a lesser extent in black participants than in white participants after corticosteroid administration ((14), (31 - 32)) or after participation in vigorous professional sports (33 - 34). Although the mechanisms of neutrophil mobilization differed in these aforementioned reports, the lower neutrophil increment suggests a lower bone marrow reserve, an intrinsic marrow difference, an abnormal cytokine response, or any combination of these. Excessive leukocyte and neutrophil adherence to vascular endothelium has been studied and is unlikely to explain the lower neutrophil increment (35). Of interest, although overall leukocyte and neutrophil counts in Mexican-American participants were similar to those in white participants, the lower leukocyte and neutrophil count increment in Mexican-American smokers was unexpected. To our knowledge, this is the first report of such an observation, and its significance is unknown.

The major limitation of our study is that the data from the blood counts were derived at 1 time per participant. Thus, it was not possible to follow the natural course of individuals with neutrophil counts less than 1.5 × 10⁹ cells/L by using serial blood counts. Although blood counts can fluctuate with concurrent viral infections, medical illnesses, or medications, the large sample size and the voluntary nature of the examinations should minimize these effects. In addition, hematologic data were missing for many participants who were 18 years of age or younger. Thus, the prevalence of neutropenia in younger participants may be exaggerated. However, the different prevalence of neutropenia among ethnic groups is probably not affected, because participants with missing data were evenly distributed among all 3 groups.

Although many infectious and inflammatory factors modulate changes in blood counts, this population-based report suggests broad genetic influences. The findings in the current analysis were consistent and robust for the observed differences in age, sex, and ethnicity. Black participants had substantially lower mean leukocyte and neutrophil counts and similar lymphocyte counts than did white participants, whereas Mexican-American participants had slightly higher mean leukocyte, lymphocyte, and neutrophil counts than did white participants. In addition, the leukocyte and neutrophil counts for current smokers were higher than for participants who never smoked. The lower blood counts that were observed in black participants beginning at a young age may have few symptomatic consequences. The overall prevalence of neutropenia in black participants was estimated to be 4.5%, which is lower than that reported in previous non–population-based studies. Most participants with neutrophil counts less than 1.5 × 10⁹ cells/L had counts of at least 1.0 × 10⁹ cells/L. On the basis of these data, it can be concluded that neutropenia is more common in the general population than may be clinically recognized. Decisions about the need for a thorough diagnostic evaluation of neutropenia should take into account not only clinical symptoms but also age, sex, ethnicity, and smoking status.