TECHNICAL NOTE

Comparison of blood counts in venous,®ngertip and arterial blood and theirmeasurement variation

Z.-W. YANG*,

S.-H. YANG*,L. CHEN*,

J. QU*,

J. ZHU*,Z. TANG 

*Morphometric Research Laboratory, North Sichuan Medical College and  Department of Clinical

Laboratory Sciences, Af®liated Hospital of North Sichuan Medical College, Nanchong, Sichuan,

China

Summary We compared routine blood pictures in venous, ®ngertip and arterial blood and their

measurement variation. A pair of venous and ®ngertip blood samples were simulta-

neously collected from 24 volunteers, and another pair of venous and arterial samples

from another 12 volunteers. The volunteers were healthy adults, aged 20±22 years,

with males and females in equal numbers. Three aliquots of blood were taken from each

person for three measurements of blood counts, including haemoglobin concentration

and haemocrit, using an automatic haematology analyser. Compared to arterial blood,

there was a signi®cant increase of erythrocyte count (2.7%) and haematocrit (3.1%) in

the venous blood. The total and large leucocyte counts were signi®cantly higher (9.2%

and 12.6%, respectively) in the ®ngertip blood than in the venous blood. There were a

decreasing number of leucocytes in the ®ngertip blood with repeated sampling. The

intermeasurement variation for total leucocyte or haemoglobin measurement was

signi®cantly larger as measured with the ®ngertip blood or arterial blood than with the

venous blood. The intermeasurement coef®cients of variation of haemoglobin (1.3%)

and small leucocyte ratio or platelet (approximately 4.0%) measurements with venous

and arterial blood were the smallest and largest, respectively. Volume of arterial plasma

was diminished by 2±3% after microcirculation. Elevation of total leucocyte count in

the ®ngertip blood was mainly due to the increase of granulocytes. Measurement with

venous blood was of better precision than with ®ngertip or arterial blood.

Keywords Arterial, blood count, ®ngertip, haematocrit, haematology analyser, haemoglobin,

routine blood picture, venous

Introduction

Difference of blood counts in venous and ®ngertip blood,

including haematocrit and haemoglobin concentration,

has long been recognized; however, controversy remains

about the exact aspects or mechanism of the difference.

For example, using automatic haematology analysers, an

increased (Dacie & Lewis, 1984; Daae et al., 1988, 1991;

Li et al., 1997; Yuan et al., 1999) or unchanged (Leppa-

nen 1988; Liu et al., 1999) leucocyte count in the

®ngertip blood was found compared to the venous blood.

Only one measurement was usually obtained from each

blood sample in previous studies, thus variation of

measurement might have partly contributed to the

controversy. This study was undertaken to clarify the

disagreement by measuring each blood sample three times.

Furthermore, comparison between venous and arterial

Accepted for publication 1 May 2001

Correspondence: Zheng-Wei Yang, Morphometric Research Laborat-

ory, North Sichuan Medical College, 234 Fujiang Road, Nanchong,

Sichuan 637007, China. Fax: +86 817 224 2600; E-mail:

zwyang@nc-public.sc.cninfo.net

Clin. Lab. Haem.

2001, 23, 155±159

Ó 2001 Blackwell Science Limited

155

blood was also made to seek an explanation for the

difference between venous and ®ngertip blood.

Materials and methods

Volunteers

Thirty-six healthy adults, year 2±4 medical students from

North Sichuan Medical College, aged 20±22 years, vol-

unteered the blood samples. All samples were collected

within approximately 2 h in the afternoon between

15.30 h and 18.30 h. Eleven (group 1), 13 (group 2)

and 12 (group 3) volunteers were included in the study on

three afternoons, respectively, with males and females in

approximately equal numbers in each group.

Blood sampling and measurement

Fingertip blood from the tip of left ring ®nger after skin-

puncture and venous blood from the left median cubital

vein were simultaneously collected from each person in

groups 1 and 2; venous blood from the left median cubital

vein and arterial blood from the left radial artery simul-

taneously collected from each person in group 3. All

volunteers had been in a state of rest for more than 30 min

and sat with the arm resting on a worktable when blood

was sampled. For the ®ngerprick sampling, one skin

puncture was made using a blood sampling needle (Ruiqi

Technology Ltd, Chengdu, Sichuan, China). The ®rst drop

of blood after gentle squeezing was discarded, and then an

aliquot (20 ll) of blood was taken from the second, third

and fourth drops of blood, respectively. Each aliquot of blood

was immediately put in 10 ml of standard dilution liquid

(Sysmex, Japan). For venous or arterial blood, 4 ml of blood

was taken from each person using a vacuum tube with

EDTA-K2 (Ruiqi Technology Ltd) for anticoagulation, and

then three aliquots (20 ll each) of blood were directly taken

from the tube and immediately put in the standard dilution

liquid. All measurements were performed on the same

automatic haematology analyser (Sysmex F-820, Japan)

within 1 h after blood sampling, one set of measurements

being obtained from each aliquot of blood. The same two

experienced haematology technicians carried out all blood

sampling and measurement at the Department of Clinical

Laboratory Sciences, Af®liated Hospital of North Sichuan

Medical College, where quality control was checked daily as

required by the Ministry of Health, China. Quality control

was performed using control blood samples provided by the

Chinese Ministry of Health, and the relative difference

between the measured values and the control values for

all parameters were within 1±2%. All sampling and

measurement were carried out at room temperature and

local heating of skin was not done for blood sampling.

Statistical analysis

Data are presented as mean � SD. Comparison between

venous and ®ngertip blood or between venous and arterial

blood was made using paired t-test; multiple comparisons

between different parameters were performed using one-

way ANOVA (analysis of variance) in conjunction with

Student-Newman-Keuls method; and multiple compari-

sons between the ®rst, second and third aliquots of blood

were performed using one-way repeated measures ANOVA

in conjunction with Student-Newman-Keuls method.

P £ 0.05 was considered statistically signi®cant.

Results

Comparison between venous and ®ngertip blood

In the 24 volunteers from groups 1 and 2, the haema-

tological parameters, which were signi®cantly higher as

measured with the ®ngertip blood than with the venous

blood, were WBC (total leucocyte count) and large

leucocyte ratio (LLR, i.e. large leucocyte number as a

percentage of total leucocytes), being 9.2% and 3.6%

higher, respectively (Table 1). The large leucocyte count

(LLC) in the ®ngertip blood (5.23 � 1.68 ´ 109/l) was

also signi®cantly higher (12.6%) than in the venous blood

(4.65 � 1.40 ´ 109/l). There were no signi®cant differ-

ences in other parameters such as RBC (erythrocyte count;

P � 0.73), Hb (haemoglobin concentration; P � 0.75),

Hct (haematocrit; P � 0.32), Plt (platelet count; P � 0.58)

or SLC (small leucocyte count; 2.25 � 0.52 versus

2.24 � 0.56 ´ 109/l; P � 0.87).

Comparison between venous and arterial blood

In group 3, the haematological parameters that were

signi®cantly lower as measured with the arterial blood

than with the venous blood were RBC (by 2.7%) and Hct

(by 3.1%); Hb was also smaller (by 1.8%) but statistical

signi®cance was not found (P � 0.08) (Table 2). There

were no signi®cant differences (P > 0.10) in other param-

eters of Table 2, LLC (4.70 � 1.00 versus 4.81 � 0.98 ´109/l) or SLC (2.23 � 0.49 versus 2.27 � 0.51 ´ 109/l).

Variation of measurement

The mean coef®cient of variation (CV) between three

measurements from each blood sample was less than 6.5%

Comparison of venous, ®ngertip and arterial blood counts156

Ó 2001 Blackwell Science Ltd., Clin. Lab. Haem., 23, 155±159

for all parameters of all the blood samples. The CV was

signi®cantly larger as measured with the ®ngertip or

arterial blood than with the venous blood for WBC and Hb

(Tables 1 and 2).

To determine whether a different statistical conclusion

would be reached if only one random aliquot of blood was

obtained from each blood sample, the three measurements

from each blood sample were independently considered,

thus nine statistical comparisons between three venous

and three ®ngertip or between three venous and three

arterial measurements were performed for each param-

eter. Compared to the statistical inferences reached by

using the mean of the three measurements, all nine

comparisons for such parameters as WBC, RBC, Hb, Hct

and Plt showed the same statistical inferences (with or

without signi®cant differences), with only one of the nine

comparisons for the parameters of SLR and LLR showing

a different statistical inference.

To analyse which parameter has a signi®cantly smaller

variation between measurements in ®ngertip blood or in

venous and arterial blood, the CVs were compared

between parameters based on one-way ANOVA; thus 21

multiple comparisons between the seven parameters

shown in Table 1 or 2 were obtained. The CVs of the

®ngertip values (n � 24) are shown in Tables 1 and 20

pairs of the CVs, not including the pair between RBC and

Hct, were signi®cantly different. When venous and arterial

blood samples (n � 48, from groups 1±3) were considered

all together, the CVs were 4.23 � 5.29% (SLR),

3.99 � 2.44% (Plt), 2.97 � 2.28% (WBC), 2.58 � 1.68%

(Hct), 2.31 � 1.55% (RBC), 1.98 � 2.32% (LLR) and

1.26 � 0.78% (Hb), respectively; and 20 pairs of the

parameters, not including the pair between WBC and Hct,

were signi®cantly different.

To evaluate whether there is a signi®cant difference

between the ®rst, second and third aliquots of blood, one-

way repeated measures ANOVA was performed. All the

seven venous parameters (n � 36, from groups 1±3) had

no signi®cant difference (P > 0.46) between different

aliquots, and all the seven arterial parameters (n � 12)

had no such signi®cant difference (P > 0.40) either.

However, while ®ngertip parameters of RBC, Hb, Hct

and Plt were not signi®cantly different (P > 0.09) between

aliquots, there was a signi®cant difference in WBC-related

Table 2. Blood counts in arterial and venous blood, mean � SD (n = 12)

Arterial Venous CV ± A (%) CV ± V (%) Ratio-A/V No. > V

WBC 6.93 � 1.21 7.08 � 1.25 3.05 � 1.84* 1.55 � 1.18 0.980 � 0.044 4

LLR 67.57 � 5.79 67.81 � 5.41 1.36 � 0.58 1.06 � 0.63 0.996 � 0.014 5

SLR 32.43 � 5.79 32.19 � 5.41 2.81 � 1.11 2.14 � 0.89 1.006 � 0.027 7

RBC 4.58 � 0.44  4.70 � 0.39 2.52 � 0.92* 1.54 � 1.08 0.973 � 0.028 1

Hb 14.1 � 1.4 14.3 � 1.2 1.30 � 0.46* 0.80 � 0.53 0.982 � 0.031 5

Hct 40.17 � 3.59* 41.47 � 3.48 2.60 � 1.41 2.19 � 1.34 0.969 � 0.035 2

Plt 212 � 55 210 � 56 3.39 � 1.69 3.59 � 1.51 1.009 � 0.034 9

A, Arterial blood; Ratio-A/V, ratio between the arterial and the corresponding venous values. No. > V, the number of arterial values

bigger than the corresponding venous ones. *P < 0.05;  P < 0.01 (paired t-test: Arterial versus Venous, or CV ± A versus CV ± V). See

Table 1 for other abbreviations.

Table 1. Blood counts in ®ngertip and venous blood, mean � SD (n = 24)

Fingertip Venous CV ± F (%) CV ± V (%) Ratio-F/V No. > V

WBC 7.49 � 1.99  6.89 � 1.74 6.14 � 3.49* 3.65 � 2.60 1.092 � 0.091 22

LLR 69.44 � 5.41  67.12 � 5.56 1.86 � 1.08 2.75 � 3.07 1.036 � 0.037 21

SLR 30.56 � 5.41  32.88 � 5.56 4.59 � 3.43 5.99 � 7.05 0.932 � 0.079 3

RBC 4.76 � 0.52 4.75 � 0.48 3.22 � 1.64 2.60 � 1.87 1.003 � 0.036 13

Hb 14.2 � 1.4 14.3 � 1.4 2.45 � 1.32  1.46 � 0.92 0.998 � 0.034 12

Hct 40.49 � 3.79 40.09 � 3.93 3.20 � 1.63 2.76 � 1.96 1.012 � 0.048 14

Plt 220 � 38 222 � 35 6.47 � 6.57 4.50 � 3.02 0.991 � 0.109 7

F, Fingertip blood; V, venous blood. CV, coef®cient of variation between three measurements from each blood sample. Ratio-F/V, ratio

between the ®ngertip and the corresponding venous values. No. > V, the number of ®ngertip values bigger than the corresponding

venous values. *P < 0.05;  P < 0.001 (paired t-test: Fingertip versus Venous, or CV ± F versus CV ± V). WBC, white blood cell

(leucocyte) count (109/l); LLR, large leucocyte (with sizes > 90 ¯, mostly granulocytes) ratio (percentage of total leucocyte count); SLR,

small leucocyte (with sizes £ 90 ¯, mostly lymphocytes) ratio. RBC, red blood cell (erythrocyte) count (1012/l); Hb, haemoglobin

concentration (g/dl); Hct, haematocrit (%); Plt, platelet count (109/l).

Z.-W. Yang et al. 157

Ó 2001 Blackwell Science Ltd., Clin. Lab. Haem., 23, 155±159

parameters (n � 24). The multiple comparisons with a

signi®cant difference were as followings: WBC between

the ®rst and third aliquots (7.69 � 2.16 versus

7.33 � 1.90 ´ 109/l; the result from the second aliquot:

7.47 � 2.00); LLR between the ®rst (68.72 � 5.55%) and

third (70.23 � 5.63%) and between the second

(69.38 � 5.38%) and the third; SLR between the ®rst

(31.28 � 5.55%) and third (29.78 � 5.63%) and

between the second (30.62 � 5.38%) and the third; SLC

between the ®rst (2.38 � 0.62 ´ 109/l) and second

(2.25 � 0.48 ´ 109/l), between the second and third

(2.13 � 0.51 ´ 109/l) and between the third and ®rst.

The trend of a decreasing LCC from the ®rst, second and

third aliquots of ®ngertip blood (5.31 � 1.72 versus

5.22 � 1.71 versus 5.15 � 1.68 ´ 109/l) was also

noticed although statistical signi®cance was not detected

(P � 0.29).

Discussion

This study demonstrated a signi®cantly higher erythro-

cyte count (2.7%) and haematocrit (3.1%) in venous

blood than in arterial blood (P < 0.05); haemoglobin

concentration also appeared to be larger (1.8%) in

venous blood (P � 0.08). This suggested that 2±3% of

plasma in the arterial blood exudes out of capillary in

microcirculation and remains to form tissue ¯uid in the

surrounding tissues. In contrast, there was no signi®-

cant difference in erythrocyte count, haemoglobin

concentration or haematocrit between ®ngertip and

venous blood. Erythrocytes do not normally migrate

out of capillary as leucocytes do, and ®ngertip blood and

venous blood looked similar in colour. These might

suggest that ®ngertip blood after skin-puncture probably

mainly came from capillary and small veins rather than

arterioles.

The major difference between ®ngertip and venous

blood was elevated total and large leucocyte counts in the

®ngertip blood. This may be the result of immediate local

accumulation of granulocytes on the stimulation of skin-

puncture because no signi®cant differences in small

leucocyte and erythrocyte counts or haematocrit between

the two blood samples were found.

Previous studies demonstrated a decreased (Dacie

& Lewis, 1984; Daae et al., 1988; Leppanen, 1988),

unchanged (Daae et al., 1991) or increased (Yuan et al.,

1999) platelet count in the ®ngertip blood compared to

venous blood. In the present study, no signi®cant

difference was detected between the two blood samples.

As also shown, platelet count was one of the most

unstable haematology parameters, to which the above

discrepancies may be ascribed. The relatively large CV for

platelet count in all blood samples may indicate a relative

unstable status of anticoagulation of blood in vitro, to

which platelets is most sensitive.

In contrast to ®ndings of the current study and other

colleagues (Dacie & Lewis, 1984; Leppanen, 1988), some

previous investigators observed an elevated erythrocyte

count in the ®ngertip blood compared to venous blood

(Daae et al., 1988, 1991; Li et al., 1997; Yuan et al.,

1999). This disagreement, if not being contributed by

different experimental conditions and/or sampling/sys-

tematic errors, is harder to explain.

Most parameters, such as WBC, RBC, Hb, Hct and Plt

as measured with the ®ngertip blood, appeared to have

a larger variation between measurements than that

with the venous blood (Table 1). A larger intermeasure-

ment variation for arterial parameters such as WBC, RBC

and Hb was also observed compared to venous param-

eters (Table 2). Statistically speaking, a blood count in

venous blood is more ef®cient than ®ngertip or arterial

blood.

With repeated measurements of each blood sample, the

present study showed a small intermeasurement vari-

ation, with CV < 6.5% in all cases (Tables 1 and 2). This

variation is contributed by the intramachine variation

together with the interaliquot variation. As measured with

venous and arterial blood, the mean intermeasurement CV

of all seven parameters was 2.76% and that of RBC-related

parameters (RBC and Hb concentrations and Hct) was

2.05%. That is, the total error of the mean of the three

measurements would be less than 1.6% [ �2.76/(31/2)]

for all parameters. Moreover, when each set of the three

measurements from each blood sample were independ-

ently considered, statistical conclusions were essentially

unaffected. These strongly suggest that the result of the

present study is reliable.

In contrast, the intermeasurement CV for total leuco-

cyte count in ®ngertip blood (aliquots of which were

obtained following ®ngertip squeezing several times) was

as high as 6.14% and signi®cantly larger than that in

venous blood. This indicates a considerable interaliquot

variation, i.e. a considerable effect of squeezing action in

the ®ngerprick sampling. This was partly due to the

decreasing leucocyte count with repeated sampling.

Conway et al. (1998) demonstrated different haemo-

globin values using different drops of ®ngerprick blood by

biomedical scientists and laymen. Haemoglobin values

obtained by the two experienced technicians in the present

study were not signi®cantly different between the second,

third and fourth drops of ®ngertip blood. In contrast,

signi®cant between-drop difference in WBC-related

Comparison of venous, ®ngertip and arterial blood counts158

Ó 2001 Blackwell Science Ltd., Clin. Lab. Haem., 23, 155±159

parameters was detected. Small leucocyte count measured

with the third drop of blood is decreased by 5.61%

(P < 0.05) compared to the value with the second drop of

blood, and the count with the fourth drop was further

decreased by 5.01% compared to the third drop. This

suggested that, with repeated ®ngerprick sampling, fewer

and fewer leucocytes, especially small leucocytes (mostly

lymphocytes), accumulated in the blood. As described

above, relatively more large leucocytes accumulated in the

®ngertip blood. These may indicate a different pathological

response to ®ngerprick by the granulocytes and lympho-

cytes.

As expected, there was no signi®cant between-aliquot

difference in all venous values and arterial values because

each set of three aliquots of venous or arterial blood were

taken from the same pool (population) of blood.

The relatively large CV for the small leucocyte ratio

might be partly due to its smaller number in blood:

approximately 32% of total leucocytes and approximately

0.046% of all formed elements. The highest ef®ciency for

haemoglobin measurement in all blood samples may be

attributed to the fact that erythrocytes accounted for

approximately 95.5% of formed elements in blood

(Tables 1 and 2) and haemoglobin measurement was

not affected by the anticoagulation status of blood using

the haematology analyser.

In summary, examination of routine blood picture with

venous blood is more accurate than with ®ngertip or

arterial blood, and measurement of RBC-related param-

eters is more stable than WBC-related parameters or

platelet counts. A decreasing amount of leucocytes

accumulate in the ®ngerprick blood with repeated samp-

ling. The relative difference between ®ngertip and venous

parameters were 6.8±9.2% for WBC-related parameters

and 0.2±1.2% for other parameters (Table 1). Therefore, if

the above variations and differences are acceptable, blood

counts with ®ngertip blood (the sampling of which is

simple and more acceptable to most patients) would still be

acceptable in clinical practice.

Acknowledgements

The authors wish to thank Messrs Jun-An Li and Yong-

Sheng Feng (Department of Clinical Laboratory Sciences)

for their excellent technical work and Dr Jing-Guo Zhou

(Department of Haematology) for his helpful suggestion.

This study was ®nancially supported by an `Excellent

Young Researcher Grant' from the Chinese Ministry of

Education.

References

Conway A.M., Hinchliffe R.F., Earland J. & Anderson L.M. (1998)

Measurement of haemoglobin using single drops of skin

puncture blood: is precision acceptable? Journal of Clinical

Pathology 51, 248±250.

Daae L.N., Halvorsen S., Mathisen P.M. & Mironska K. (1988) A

comparison between haematological parameters in `capillary'

and venous blood from healthy adults. Scandinavian Journal of

Clinical and Laboratory Investigation 48, 723±726.

Daae L.N., Hallerud M. & Halvorsen S. (1991) A comparison

between haematological parameters in `capillary' and venous

blood samples from hospitalized children aged 3 months to 14

years. Scandinavian Journal of Clinical and Laboratory Investiga-

tion 51, 651±654.

Dacie J.W. & Lewis S.M. (1984) Practical Haematology, 6th edn.

Churchill Livingstone, London, 1±6.

Leppanen E.A. (1988) Experimental basis of standardized speci-

men collection: the effect of the site of venipuncture on the

blood picture, the white blood cell differential count, and the

serum albumin concentration. European Journal of Haematology

41, 445±448.

Li C.H., Zhu J.D. & Yin S.J. (1997) Comparison of multi-param-

eter haematological analysis between ®nger and venous blood

from 200 healthy adults. Chinese Journal of Medical and Labor-

atory Science 20, 46±47.

Liu N.D., Yuan H.R., Huang Y., Wang H. & Qin X.G. (1999)

Comparison of 18 haematological parameters between ®nger

and venous blood from 124 cases. Journal of Clinical and

Laboratory Science (China) 17, 88.

Yuan Q., Xu Y.R. & Wang F. (1999) Comparison of multi-

parameters between ®nger and venous blood with Coulter JT

haematology analyser. Information of Henan Medicine (China)

7, 42.

Z.-W. Yang et al. 159

Ó 2001 Blackwell Science Ltd., Clin. Lab. Haem., 23, 155±159