## Abstract

Aims We aimed to assess potential associations between different leucocyte components and coronary heart disease (CHD) in a prospective cohort study, and to put these findings in context of other relevant prospective studies in a meta-analysis.

Methods and results We report data on differential leucocyte count and CHD derived from the first National Health and Nutrition Examination Survey (NHANES I) and the NHANES 1 Epidemiologic Follow-up Study (NHEFS) involving 4625 individuals followed, on average, for 18 years. The NHEFS involved 914 incident CHD cases and yielded an adjusted risk ratio of 1.09 (0.93–1.29) comparing individuals with neutrophil counts in the top third versus those in the bottom third of the population. In a meta-analysis involving the NHEFS and four other studies comprising a total of 1764 incident CHD cases, the association of CHD with neutrophil counts was somewhat stronger than those with other specific leucocyte components (combined risk ratio=1.33 [1.17–1.50]) but there was substantial heterogeneity between the separate studies (

$${\chi}^{2}_{4}=18.0$$
,
$$p{<}0.001$$
).

Conclusions Although the present synthesis provides the most comprehensive assessment so far of specific leucocyte components in CHD, additional prospective data will be needed to resolve whether neutrophil counts are much stronger predictors of CHD risk than other components.

## Introduction

The idea that atherosclerosis may, in part, be an inflammatory disease is supported by the presence of mononuclear cells in arterial lesions1 and by the ability of leucocyte counts to predict vascular disease. A meta-analysis of prospective studies, has reported that people in the top third of baseline total leucocyte counts have about a 40% (95% CI: 30–50%) increased risk of coronary heart disease (CHD) over the subsequent decade compared with those in the bottom third of the population.2 The causal relevance of this association, however, remains uncertain because it is not clear to what extent the association of the total leucocyte count with CHD risk merely reflects the impact of established risk factors (such as cigarette smoking), the extent of existing atherosclerosis, or both. As leucocytes have such a wide range of biological effects, some potentially protective against vascular disease and some potentially damaging, identification of more specific (and potentially stronger) associations between particular components of the leucocyte count and CHD should help elucidate the association. For example, a particular role for neutrophils has been suggested by reports of strong associations between risk of Myocardial infarction and genetic determinants of 5-lipoxygenase activating protein, a pathway that includes products such as leukotriene B4, which is produced by stimulated neutrophils.3

Available prospective data on differential leucocyte counts and CHD are, however, limited and unsynthesised. To help provide additional data and to clarify the existing evidence we have (i) reported previously unpublished data from the first National Health and Nutrition Examination Survey Epidemiologic Follow-up Study (NHEFS), (ii) sought summary data from other prospective studies with previously unpublished data on differential leucocyte counts, and (iii) conducted a meta-analysis of these newly available data in the context of the limited published prospective data.

## Methods

### NHANES I Epidemiologic Follow-up Study

There were 9881 persons aged 25–74 years in the total NHANES I sample, of which 6913 attended baseline interviews and examinations in 1971–1975 and were eligible for the NHEFS4 (response rate 70%). The NHEFS involved follow-up in four waves during 1982–1984, 1986, 1987 and 1992.5 Incident CHD cases were identified using routine data sources: death certificate with underlying or non-underlying cause of death coded ICD-9 410–414 or one or more hospital stays during the follow-up period with discharge diagnosis of ICD-9 410-414. Hospital discharge summaries for the follow-up period were obtained from any hospitals for which participants reported a stay. Cases were counted only once, with the incidence date taken as the date of first hospital stay, or the date of death for fatal CHD cases with no prior record of hospitalisation. Blood samples were obtained at baseline for differential leucocyte count in 6913 individuals. Duplicate smears were made by the wedge technique from blood preserved with EDTA. Five qualified medical laboratory technicians counted 100 white blood cells using a microscope and recorded the white blood cell type using published cell-typing criteria.6–10 In addition to the 358 (5.2%) individuals lost to follow-up, the statistical analysis excluded 80 (1.2%) individuals from racial groups other than “white” and “black”, 1448 persons with unknown baseline leucocyte count, systolic blood pressure, serum cholesterol, diabetes history, number of cigarettes smoked or body mass index, four persons with extreme leucocyte counts (>18,000 cells/mm3), and 398 persons with a positive history of heart disease at baseline, leaving 4625 individuals in the present analysis. Incidence rates were calculated as number of CHD cases divided by total person years expressed in thousands. Hazard ratios were derived from Cox proportional hazards regression models using SAS statistical analysis software (PHREG procedure).11 There was some evidence of violation of the proportional hazards assumption in the full data set, but not when the first 5 years of follow-up were excluded. But, as estimates were virtually identical by the two methods, results over the full follow-up period are presented in detail, but any differences between the adjusted estimates have been noted. Multivariable models included age at baseline, gender, ethnicity, recreational activity, educational level, systolic blood pressure, serum cholesterol, body mass index and history of diabetes. Sex-specific regression models and combined models including interaction terms with sex were used in a subsidiary analysis. All p-values quoted are derived from two-sided tests.

### Meta-analysis

Prospective epidemiological studies of leucocyte count and CHD with >1 year follow-up, published before January 2003, were sought by computer-based searches scanning the reference lists of all relevant studies and review articles, hand-searching of relevant journals and correspondence with authors of studies. Computer searches in Medline used keywords relating to leucocytes (e.g., leucocyte, leukocyte, white blood cell, neutrophil, eosinophil, basophil, monocyte, granulocyte, lymphocyte) in combination with coronary heart disease (e.g., coronary heart disease, isch[a]emic heart disease, vascular disease, myocardial infarction and arteriosclerosis) and study design (e.g., prospective or cohort). All available studies that reported on differential leucocyte counts in relation to CHD were included. In addition, we corresponded with authors of all previously published prospective studies of total leucocyte count to obtain, where available, tabular data on differential leucocyte counts and CHD risk. The following information was abstracted from each study: geographical location, study size, number and definition of CHD cases, assay (or counting) method, mean age and percentage male cases, follow-up duration sampling method, and degree of adjustment for potential confounders. Adjustment as shown in the figure is denoted as: +++ for age, sex, smoking plus some other established vascular risk factors; ++++ for these plus markers of socio-economic status; +++++ for these plus evidence of pre-existing vascular disease. In studies with published results on differential leucocyte counts we corresponded with investigators to check our estimates and seek tabular data on the most up-to-date results. Results were expressed as the adjusted log risk ratio between individuals in the top third of the distribution of the leucocyte component count and those in the bottom third of disease-free individuals. Where this log risk ratio (and its standard error) was not available as a comparison of extreme thirds from the published report or from correspondence with investigators, it was estimated from the reported risk ratios (e.g., RR per SD, per unit, or other comparisons, such as top and bottom quarters) using log-linear scaling and assuming normality of the leucocyte count distribution, as previously described.2 Fixed and random effect summary estimates were calculated for each leucocyte component. Heterogeneity between study estimates was assessed by standard

$${\chi}^{2}$$
tests,12 although potentially relevant subgroup analyses (such as by age and sex) could not be reliably investigated since individual data were not available in studies other than the NHEFS. Results from studies with fewer than 50 incident CHD cases were treated separately in the analysis, but combined for presentation in the figure. To make some allowance for multiple comparisons, 99% confidence intervals were used.

## Results

### New evidence from NHANES I Epidemiologic follow-up study

Nine hundred and fourteen incident CHD cases were recorded among the 4625 individuals in the present analysis, who had a mean age at baseline of 48.1 years (46% male) and a median follow-up of 18.3 years. Table 1 provides a summary, by thirds of leucocyte components, of crude CHD incidence per 1000 person-years and rate ratios for CHD with increasing degrees of adjustment for baseline values of some established vascular risk factors. In age and sex only adjusted comparisons of leucocyte values in the top third with those in the bottom third of the control population, the rate ratios for CHD were about 1.2–1.3 in people with higher counts of total leucocytes, granulocytes (i.e., a combination of neutrophils, eosinophils and basophils), neutrophils, or lymphocytes, but each of these estimates were only moderately statistically significant. Further adjustment for baseline levels of smoking and other established risk factors weakened these associations to non-significance (Table 1). Similar results were obtained in subsidiary analyses restricted to men and women separately, and there was no significant interaction between sex and any individual leucocyte component in fully adjusted analyses.

Table 1

Summary of CHD incidence and rate ratios by thirds of total leucocyte baseline counts and of leucocyte components in the NHANES I Epidemiologic Follow-up Study

Leucocyte component

Third

No. CHD cases/No. at risk

CHD incidence/1000 pya

Rate ratios and 95% CI, adjusted for

Age and sex

Age, sex and smoking

Age, sex, smoking and risk factorsb,d

Total leucocytes 293/1583 11.7 1.00 1.00 1.00
333/1526 13.8 1.21 (1.03–1.42) 1.15 (0.98–1.34) 1.07 (0.92–1.26)
288/1516 12.3 1.26 (1.07–1.48)** 1.12 (0.95–1.33) 1.01 (0.85–1.20)

Granulocytesc 282/1540 11.5 1.00 1.00 1.00
311/1544 12.8 1.14 (0.97–1.34) 1.11 (0.95–1.31) 1.06 (0.90–1.25)
321/1541 13.5 1.32 (1.12–1.55)** 1.20 (1.02–1.41)* 1.10 (0.93–1.31)

Neutrophils 278/1539 11.3 1.00 1.00 1.00
319/1544 13.2 1.20 (1.02–1.41) 1.18 (1.00–1.38) 1.12 (0.95–1.32)
317/1542 13.3 1.33 (1.13–1.56)** 1.20 (1.02–1.42)* 1.09 (0.93–1.29)

Lymphocytes 285/1540 11.7 1.00 1.00 1.00
317/1542 13.1 1.22 (1.04–1.44) 1.19 (1.01–1.39) 1.13 (0.96–1.33)
312/1543 13.1 1.22 (1.04–1.43)* 1.14 (0.97–1.34) 1.05 (0.89–1.24)

Monocytes

334 /1540 13.8 1.00 1.00 1.00
292/1542 12.0 0.89 (0.76–1.04) 0.88 (0.76–1.03) 0.92 (0.79–1.08)
3

288/1543

12.1

0.95 (0.81–1.12)

0.94 (0.80–1.10)

0.96 (0.82–1.13)

Leucocyte component

Third

No. CHD cases/No. at risk

CHD incidence/1000 pya

Rate ratios and 95% CI, adjusted for

Age and sex

Age, sex and smoking

Age, sex, smoking and risk factorsb,d

Total leucocytes 293/1583 11.7 1.00 1.00 1.00
333/1526 13.8 1.21 (1.03–1.42) 1.15 (0.98–1.34) 1.07 (0.92–1.26)
288/1516 12.3 1.26 (1.07–1.48)** 1.12 (0.95–1.33) 1.01 (0.85–1.20)

Granulocytesc 282/1540 11.5 1.00 1.00 1.00
311/1544 12.8 1.14 (0.97–1.34) 1.11 (0.95–1.31) 1.06 (0.90–1.25)
321/1541 13.5 1.32 (1.12–1.55)** 1.20 (1.02–1.41)* 1.10 (0.93–1.31)

Neutrophils 278/1539 11.3 1.00 1.00 1.00
319/1544 13.2 1.20 (1.02–1.41) 1.18 (1.00–1.38) 1.12 (0.95–1.32)
317/1542 13.3 1.33 (1.13–1.56)** 1.20 (1.02–1.42)* 1.09 (0.93–1.29)

Lymphocytes 285/1540 11.7 1.00 1.00 1.00
317/1542 13.1 1.22 (1.04–1.44) 1.19 (1.01–1.39) 1.13 (0.96–1.33)
312/1543 13.1 1.22 (1.04–1.43)* 1.14 (0.97–1.34) 1.05 (0.89–1.24)

Monocytes

334 /1540 13.8 1.00 1.00 1.00
292/1542 12.0 0.89 (0.76–1.04) 0.88 (0.76–1.03) 0.92 (0.79–1.08)
3

288/1543

12.1

0.95 (0.81–1.12)

0.94 (0.80–1.10)

0.96 (0.82–1.13)

a

py, person years.

b

Systolic blood pressure, serum cholesterol, body mass index, history of diabetes, recreational activity, educational level and ethnic group.

c

Granulocytes=neutrophils, eosinophils and basophils.

d

In analyses excluding incident cases in the first 5 years of follow-up, rate ratios (95% CI) involving comparisons of those in the middle and highest thirds, respectively, with those in the bottom third, adjusted for age, sex, smoking and risk factors were as follows: total leucocytes 1.12 (0.94–1.33), 0.96 (0.79–1.16); granulocytes 1.07 (0.90–1.29), 1.04 (0.86–1.25); neutrophils 1.15 (0.96–1.37), 1.01 (0.84–1.21); lymphocytes 1.16 (0.97–1.38), 1.01 (0.84–1.21); monocytes 0.95 (0.80–1.13), 0.90 (0.76–1.08).

In analyses of leucocyte components as continuous variables, rate ratios (95% CI) per 1 standard deviation increase, adjusted for age, sex, smoking and risk factors were as follows: total leucocytes 1.02 (0.95–1.09); granulocytes 1.01 (0.94–1.09); neutrophils 1.01 (0.94–1.08); lymphocytes 1.02 (0.95–1.09); monocytes 0.98 (0.91–1.06).

In sex-specific analyses, rate ratios (95% CI) involving comparisons of those in the middle and highest thirds, respectively, with those in the bottom third, adjusted for age, smoking and risk factors were as follows: in men, total leucocytes 0.93 (0.74–1.18), 1.04 (0.82–1.32); granulocytes 0.93 (0.73–1.17), 1.11 (0.88–1.39); neutrophils 1.01 (0.79–1.27), 1.14 (0.90–1.44); lymphocytes 1.10 (0.88–1.38), 1.20 (0.96–1.50); monocytes 0.86 (0.69–1.06), 0.80 (0.64–1.00); in women, total leucocytes 1.25 (1.00–1.57), 0.98 (0.76–1.26); granulocytes 1.24 (0.99–1.56), 1.10 (0.86–1.41); neutrophils 1.30 (1.04–1.63), 1.05 (0.82–1.35); lymphocytes 1.16 (0.92–1.47), 0.89 (0.70–1.14); monocytes 1.01 (0.80–1.27), 1.22 (0.96–1.53).

*

$$p{<}0.05.$$

**

$$p{<}0.01.$$

### Meta analysis

Five published studies of differential leucocyte counts and CHD were identified, plus two further cohorts (NHEFS and the Bruneck Study13) with previously unpublished findings. Overall, these studies involved a total of 1764 CHD cases with a weighted mean age at baseline of 57 years (55% male) and a mean follow-up of approximately 12 years (Table 2). Apart from a few dozen cases of angina reported in two of the smaller studies,14,15 all CHD cases involved either death, certified as due to coronary heart disease (generally using ICD criteria) or confirmed non-fatal myocardial infarction (generally using WHO criteria). In a combined analysis of the top third versus the bottom third of baseline leucocyte counts, the risk ratios for CHD were 1.32 (95% CI: 1.15–1.51) and 1.33 (1.17–1.50) in those with higher granulocyte and neutrophil counts, respectively (Fig. 1 ). It should be noted that these two risk ratios are not independent because, as mentioned above, neutrophils are included in the granulocyte count. There was, however, substantial heterogeneity among the studies contributing to each of these estimates (

$${\chi}^{2}_{4}=18.0$$
,
$$p=0.001$$
;
$${\chi}^{2}_{4}=18.9$$
,
$$p=0.001$$
, respectively), with the majority due to differences between the two largest studies. Such heterogeneity implies a need to factor greater uncertainty into the CI around the overall relative risks than is provided by those above, such as by use of a random effects model that takes additional account of study variation (the risk ratios for CHD were 1.51 [0.99–2.30] and 1.48 [1.02–2.15] under such a model, in those with higher granulocyte and neutrophil counts, respectively). Combined analyses of studies of lymphocyte and monocyte counts yielded risk ratios of 1.11 (0.99–1.25) and 1.10 (0.98–1.24), respectively, which were of marginal statistical significance and did not involve significant heterogeneity among the seven contributing studies (Fig. 1); (
$${\chi}^{2}_{6}=1.8$$
,
$$p=0.94$$
;
$${\chi}^{2}_{6}=8.1$$
,
$$p=0.24$$
, respectively).

Table 2

Characteristics of available long-term prospective studies with data on differential leucocyte counts and CHD risk

Cohort (reference)

Geographical location

Population/sampling methodc

Year of baseline survey

No. participants at baseline

Percent male

Age range at baseline (years)

Mean age of cases (years)

No. incident CHD cases

Mean follow-up (years)

Type of CHD cases

Method of leucocyte counting

NHEFS USA National framework/probability 1971–1975 4625 53 25–74 58 914 18 MI/CHD death Microscopic examination
ARIC17 USA Listing of households/random 1987–1989 11,306 44 45–64 54 527 10 MI/CHD death Automated counter
Caerphilly and Speedwell18 UK Electoral rolls/random 1979–1983 2163 100 45–63 NS 143 MI/CHD death Technicon automated counter
Hiroshima and Nagasaki15 Japan Atomic bomb survivors in two cities/random 1958 7597 38 35–75 approx 60 97 16 MI/CHD death/anginaa Newbauers chamber and Wrights stain
Paris Prospective Study II14 France Civil servants and rail workers/complete aged 30–50 1980–1985 3705 100 30–50 42 46 5.5 MI/CHD death/anginab Coulter counter
Santiago19 Spain Coronary angiography patients/complete 1991 152 100 29–87 63 24 CHD death Coulter counter
Bruneck13

Italy

Population register/random

1990

826

50

40–79

$$>$$
70

13

5

MI/CHD death

Microscopic examination

Cohort (reference)

Geographical location

Population/sampling methodc

Year of baseline survey

No. participants at baseline

Percent male

Age range at baseline (years)

Mean age of cases (years)

No. incident CHD cases

Mean follow-up (years)

Type of CHD cases

Method of leucocyte counting

NHEFS USA National framework/probability 1971–1975 4625 53 25–74 58 914 18 MI/CHD death Microscopic examination
ARIC17 USA Listing of households/random 1987–1989 11,306 44 45–64 54 527 10 MI/CHD death Automated counter
Caerphilly and Speedwell18 UK Electoral rolls/random 1979–1983 2163 100 45–63 NS 143 MI/CHD death Technicon automated counter
Hiroshima and Nagasaki15 Japan Atomic bomb survivors in two cities/random 1958 7597 38 35–75 approx 60 97 16 MI/CHD death/anginaa Newbauers chamber and Wrights stain
Paris Prospective Study II14 France Civil servants and rail workers/complete aged 30–50 1980–1985 3705 100 30–50 42 46 5.5 MI/CHD death/anginab Coulter counter
Santiago19 Spain Coronary angiography patients/complete 1991 152 100 29–87 63 24 CHD death Coulter counter
Bruneck13

Italy

Population register/random

1990

826

50

40–79

$$>$$
70

13

5

MI/CHD death

Microscopic examination

NS, not stated; NHEFS, first National Health and Nutrition Examination Survey (NHANES I) Epidemiologic Follow-up Study; ARIC, Atherosclerosis Risk in Communities.

a

Approximately 20 cases were defined by angina pectoris alone.

b

Approximately 20 cases were defined by angina pectoris alone.

c

Sampling method: random, a randomly selected subset of eligible persons was invited to participate; complete, all eligible persons in the study population were invited to participate; probability, random sampling with over-sampling of specific groups such as women of reproductive age, the elderly.

Fig. 1

In the figure, areas of squares are proportional to the sample size, horizontal lines indicate confidence intervals, and open diamonds indicate totals. Adjustment as shown in the figure is denoted as: +++ for age, sex, smoking plus some other established vascular risk factors; ++++ for these plus markers of socio-economic status; +++++ for these plus evidence of pre-existing vascular disease. (*) “Cases” refer to incident confirmed non-fatal myocardial infarction or death due to coronary heart disease. “Controls” refer to individuals who were not defined as cases at the time of analysis. (†) Although all studies measured similar component cell types (neutrophils, eosinophils, basophils, monocytes and lymphocytes), results are included in the figure only where adjusted estimates were available in the published literature or through correspondence with investigators.

Fig. 1

In the figure, areas of squares are proportional to the sample size, horizontal lines indicate confidence intervals, and open diamonds indicate totals. Adjustment as shown in the figure is denoted as: +++ for age, sex, smoking plus some other established vascular risk factors; ++++ for these plus markers of socio-economic status; +++++ for these plus evidence of pre-existing vascular disease. (*) “Cases” refer to incident confirmed non-fatal myocardial infarction or death due to coronary heart disease. “Controls” refer to individuals who were not defined as cases at the time of analysis. (†) Although all studies measured similar component cell types (neutrophils, eosinophils, basophils, monocytes and lymphocytes), results are included in the figure only where adjusted estimates were available in the published literature or through correspondence with investigators.

## Discussion

As it has been proposed that certain leucocyte components, such as neutrophils, are particularly relevant to atherosclerosis,13–15,17–19 prospective epidemiological data on differential leucocyte counts should help to clarify the nature of the associations previously observed between the total leucocyte count and CHD risk.2 The present meta-analysis of 1764 incident cases of CHD from seven long-term prospective studies, involving a total of 30,374 participants, provides the most comprehensive assessment so far of the potential impact of differential leucocyte counts on CHD.

The interpretation of the present evidence, however, is complicated by potentially important differences in the two largest studies, which together comprise approximately 80% of the available CHD cases. Whereas there was less scope for random error in the NHEFS results (since it comprised the largest single study in the meta-analysis, involving 914 incident CHD cases), likely imprecision in its manual method of counting leucocytes (even with trained laboratory staff) may have resulted in under-estimation of the true associations between leucocyte components and CHD. By contrast, the ARIC study, involving 527 incident CHD cases used an automated leucocyte counter. Moreover, the ARIC study may have defined CHD more precisely by using clinical review against standardised criteria to supplement routine CHD ascertainment methods, whilst the NHEFS relied on routine CHD ascertainment methods alone. So, although confidence intervals were wider in the ARIC study than those in NHEFS, the more extreme associations observed in ARIC between elevated neutrophil counts (and, hence, also granulocyte counts) and CHD risk may have been related to its greater precision. By comparison, associations of CHD with lymphocyte and monocyte counts observed in the present meta-analysis were probably weaker than those for granulocyte and neutrophil counts (Fig. 1).

In summary, although our study indicates that neutrophil count may be a somewhat stronger predictor of CHD risk than other leucocyte components, more reliable quantification of its predictive ability will require fresh evidence from prospective studies involving large numbers of CHD cases, with serial measurements of leucocyte components using optimal assay methods, and measurement of other inflammatory biomarkers.16 As it is likely that the neutrophil count is, at best, only a comparatively modest predictor of CHD in general populations, there is a need for sufficiently powered studies, to help assess the separate and combined impact of inflammatory markers on CHD risk.

1

doi:10.1016/j.ehj.2004.06.009.

This work was supported by a British Heart Foundation programme grant. JD has been supported by the Raymond and Beverly Sackler Award in the Medical Sciences.

The following investigators provided additional information from their studies: Antonio Amaro Cendón, Edith Feskens, Aaron Folsom, Paul Froom, Sandy Irving, Stefan Kiechl, Matti Mänttäri, Emily Marino, Ross Prentice, Goya Wannamethee and John Yarnell.

## References

[1]
Ross
R
. Atherosclerosis: an inflammatory disease.
N. Engl. J
.
1999
;
340
:
115
–126.
[2]
Danesh
J
, Collins R, Appleby P et al. Association of fibrinogen, C-reactive protein, albumin, or leukocyte count with coronary heart disease: meta-analysis of prospective studies.
JAMA
.
1998
;
279
:
1477
–1482.
[3]
A
, Manolescu A, Thorleifsson G et al. The gene encoding 5-lipoxygenase activating protein confers risk of myocardial infarction and stroke.
Nat. Genet
.
2004
;
36
(3):
233
–239.
[4]
Cohen
BB
, Barbano HE, Cox CS et al. Plan and operation of the NHANES I Epidemiologic Follow-up Study: 1982–1984.
Vital Health Stat. 1
.
1987
;
22
:
1
–142.
[5]
Cox
CS
, Mussolino ME, Rothwell ST et al. Plan and operation of the NHANES I Epidemiologic Follow-up Study, 1992. National Center for Health Statistics.
Vital Health Stat 1
.
1997
;
35
:
1
–231.
[6]
Miller
HW
. Plan and operation of the Health and Nutrition Examination Survey, United States, 1971–73. National Center for Health Statistics.
Vital Health Stat 1
.
1973
;
1
(10a):.
[7]
Plan and operation of the Health and Nutrition Examination Survey, United States, 1971–1973. National Center for Health Statistics. Vital Health Stat 1 1973;1(10b).
[8]
Engel A, Murphy RS, Maurer K et al. Plan and operation of the HANES I Augmentation Survey of Adults 25–74 Years, United States, 1974-75. National Center for Health Statistics. Vital Health Stat 1 1978;1(14).
[9]
McGrath
CR
, Hitchcock DC, van Assendelft OW. Total white blood cell counts for persons ages 1–74 years with differential leukocyte counts for adults ages 25–74 years, United States, 1971–75. National Center for Health Statistics.
Vital Health Stat 1
.
1982
;
11
(220):.
[10]
Manual of hematologic and biochemical laboratory procedures of the Health and Nutrition Examination Survey, United States, 1971–1975. National Center for Health Statistics, 2002.
[11]
SAS technical report P-217. SAS/STAT Software: the PHREG procedure. SAS Institute, Inc., Cary, NC, 1991, p. 1–59.
[12]
Egger
M
, Davey Smith G, Altman D. Systematic reviews in health care: meta-analysis in context. 2nd ed. London: BMJ; 2001. .
[13]
Willeit
J
, Kiechl S. Prevalence and risk factors of asymptomatic extracranial carotid artery atherosclerosis: a population-based study.
Arterioscler. Thromb
.
1993
;
13
:
661
–668.
[14]
Olivares
R
, Ducimetiere P, Claude JR. Monocyte count: a risk factor for coronary heart disease?
Am. J. Epidemiol
.
1993
;
137
(1):
49
–53.
[15]
Prentice
RL
, Szatrowski TP, Fujikura T et al. Leukocyte counts and coronary heart disease in a Japanese cohort.
Am. J. Epidemiol
.
1982
;
116
(3):
496
–509.
[16]
Danesh
J
, Wheeler JG, Hirschfield GM et al. C-reactive protein and other circulating markers of inflammation in the prediction of coronary heart disease.
New Engl J Med
.
2004
;
350
:
11387
–11397.
[17]
Lee
CD
, Folsom AR, Nieto FJ. White blood cell count and incidence of coronary heart disease and ischaemic stroke and mortality from cardiovascular disease in African–American and white men and women.
Am. J. Epidemiol
.
2001
;
154
(8):
758
–764.
[18]
Sweetnam
PM
, Thomas HF, Yarnell JWG. Total and differential leukocyte counts as predictors of ischemic heart disease: the Caerphilly and Speedwell studies.
Am. J. Epidemiol
.
1997
;
145
(5):
416
–421.
[19]
Amaro
A
, Gonzalez-Juanatey JR, Inglesias C. Leukocyte count as a predictor of the severity ischaemic heart disease as evaluated by coronary angiography.
Rev. Portuguese Cardiol
.
1993
;
12
(11):
913
–917.