• platelet function;
  • PFA-100;
  • smoking;
  • oral contraceptives


  1. Top of page
  2. Abstract
  6. Acknowledgements
  7. References

The PFA-100® (PFA) diagnostic system for the detection of platelet dysfunction was evaluated to determine reference ranges in a normal population. The PFA determines the primary haemostasis capacity (PHC) of anticoagulated whole blood, expressed by the system's closure time (CT). In this study the CT reference ranges were determined for blood samples collected in 105 mmol/l (3.2%) buffered citrate and the effect of gender, smoking, and use of oral contraceptives on reference ranges was assessed. Each of the 309 healthy blood donors from five blood centres was confirmed to have normal platelet function before inclusion in the study. Blood samples were tested in duplicate with both the collagen/epinephrine (Col/Epi) and collagen/ADP (Col/ADP) test cartridges.

PFA reference ranges (90% central intervals of measured closure times) for both cartridge types were similar for all groups. Subgroup analysis showed that neither gender nor oral contraceptive usage had any effect on PHC. The 95% cut-off value for the Col/Epi CT was slightly higher for smokers than for non-smokers, an effect more pronounced in female than in male donors. However, the small difference did not justify establishment of specific reference ranges for smokers. Data from all included subjects were pooled to calculate the CT reference ranges for blood samples collected in 105 mmol/l buffered citrate (Col/Epi 82–150 s; Col/ADP 62–100 s). Normal levels of fibrinogen, as well as normal platelet counts and normal haematocrit levels, appeared not to influence the PHC. Because slight but significant differences of the reference ranges were observed between some of the participating sites, in-house confirmation of these reference range guidelines is recommended.

Evaluation of platelet function in a patient undergoing surgery is important for haemostatic management, particularly if the patient has a history of clinically significant bleeding ( George & Shattil, 1991). Traditionally, the most widely used screening test for platelet disorders is the in-vivo bleeding time test, which suffers from inevitable variance, a lack of predictive value for postoperative bleeding ( Rodgers et al, 1990 ; Lind, 1991) and a certain amount of discomfort for both patient and technician. Furthermore, the test set-up of the bleeding time and associated scarring disqualifies the method for multi-testing and follow-up of the patient to assess the effect of haemostatic therapy. As a reference method, platelet aggregometry is generally accepted as the preferred technology to diagnose platelet dysfunction. Drawbacks of aggregometry include the need for a relatively large sample size, suboptimal standardization, and the fact that reaching a final diagnosis takes considerable time, skill and technical expertise.

Recently, the PFA-100® test system was introduced as an aid in the detection of platelet dysfunction ( Kundu et al, 1995 ). The PFA demonstrated good clinical performance as indicated by both high specificity for normal platelet function and high sensitivity for platelet dysfunction ( Mammen et al, 1998 ). In particular, with respect to acetyl salicylic acid (ASA)-induced platelet dysfunction, the PFA showed superior sensitivity when compared to the traditional bleeding time test ( Marshall et al, 1997 ). Recent detailed characterization of the PFA system demonstrated that its mode of action correlates well with known mechanisms of in-vivo primary haemostasis in that the PFA test was dependent on shear, platelet receptors GPIb and GPIIb–IIIa, platelet count, haematocrit, and von Willebrand factor (VWF) (Kundu et al, 1996). The importance of VWF in the PFA system was further illustrated in several recent reports demonstrating the system's high sensitivity for detection of the dysfunctional primary haemostasis in von Willebrand disease ( Fressinaud et al, 1998 ; Carcao et al, 1998 ).

Most of the aforementioned studies were conducted with blood samples collected in 129 mmol/l (3.8%) buffered sodium citrate. Since the potential influence of the collection system was demonstrated earlier ( Heilmann et al, 1997 ), and a global trend toward use of low-citrate collection systems is anticipated ( Wiseman, 1996, 1998), this study aimed to determine the reference ranges of samples collected in 105 mmol/l (3.2%) buffered sodium citrate. Because smoking and oral contraceptives usage are suspected to influence platelet function, this study also assessed the effects of these behavioural parameters on PFA results. Furthermore, the effect of gender, normal levels of platelets, haematocrit, and fibrinogen on PFA results was evaluated. Finally, the multi-centre set-up of the study allowed assessment of within-site variance, to estimate overall reproducibility of PFA testing, and inter-site variance, to determine any local population influences on PFA testing.


  1. Top of page
  2. Abstract
  6. Acknowledgements
  7. References

Subject recruitment and enrolment

From five blood centres in Germany, 359 committed (regular donations in one blood centre) blood donors were recruited and tested according to the study protocol, in accordance with guidelines specified in the Declaration of Helsinki, and after signing a letter of informed consent. Intended enrolment comprised 60 donors/site, divided into five groups: non-smoking males (M), non-smoking females (F), smoking males (MS), smoking, non-contraceptive-using females (FS), non-smoking, oral-contraceptive-using females (OC). Subjects were accepted for the smoking group when their usage exceeded five cigarettes per day (a lower usage resulted in non-selection). All oestrogen-based contraceptives were acceptable for inclusion in the OC group.

Inclusion of a subject in the study was dependent upon a history of no haemostatic disorders and normal results of laboratory tests for primary and secondary haemostasis as well as complete blood count. Based on these data, the responsible investigator at each site confirmed the platelet function in the collected blood samples to be normal.

Clinical history

Each subject's clinical history was taken and checked for accordance with inclusion criteria: age  >18 years, ostensibly healthy; and exclusion criteria: history of platelet and/or bleeding disorder, liver disease or renal insufficiency; ingestion of ASA, ASA containing compounds, ibuprofen, antibiotics, antihistamines, non-steroid anti-inflammatory drugs, oral anticoagulants, or any other drugs affecting platelet function within 12 d of sample collection; vascular defects or collagen impairment; use of DDAVP (1-deamino-8- D-arginine vasopressin), blood products or antifibrinolytics within 8 d of sample collection, current pregnancy

Tests for primary and secondary haemostasis

All collected samples were tested for: activated partial thromboplastin time (aPTT) in seconds, prothrombin time (PT) in % compared to control plasma, fibrinogen (g/l), factor VIII:C in % compared to control plasma, complete blood count including platelet count (109/l) and haematocrit (%), platelet aggregation in platelet-rich plasma with ADP, collagen, epinephrine, arachidonic acid and ristocetin as agonists. All tests (including agonist concentrations, activities and sources) were performed according to the respective guidelines and operation procedures at each site. Aggregometry studies were carried out within 3 h of collection. Acceptance levels to fulfil the normality criteria were in alignment with the appropriate in-house normal ranges. Platelet counts > 150 × 109/l and haematocrit levels > 35% were also required.

Collection of blood samples

Prior to the regularly scheduled blood donation at each site, blood samples were collected in EDTA for the complete blood count, in the standard in-house citrate collection system for aggregometry, and in 105 mmol/l buffered citrate for PFA-testing (S-Monovette®). Exchange of collection tubes was facilitated by using a 19G Multifly® canula (all from Sarstedt, Nümbrecht, Germany).

PFA-100 Test System

The in-vitro platelet function test system, PFA-100® (Dade Behring Inc., Deerfield, Ill., U.S.A.), is an aid for the detection of platelet dysfunction in an anticoagulated blood sample. The test system has been described in detail by Kundu et al (1995 ). Briefly, the system consists of a disposable test cartridge where a platelet plug occludes a microscopic aperture cut into a membrane coated with collagen and either epinephrine (Col/Epi) or ADP (Col/ADP). The plug formation occurs under high-shear flow produced and controlled by a constant vacuum and a capillary. The time required for occlusion, the closure time (CT), indicates the primary haemostatic capacity (PHC) of the blood sample. Since the Col/ADP test cartridge has no significant sensitivity for acetyl salicylic acid (ASA), and the Col/Epi test cartridge has, the combination of both test cartridges enables discrimination of ASA effects from other causes of platelet dysfunction ( Marshall et al, 1997 ; Mammen et al, 1998 ).

Each of the PFA systems in use had on-board software, version 1.20. All PFA testing was performed in duplicate, and was completed within 4 h of blood collection. Retesting was performed when the coefficient of variance between the duplicates exceeded 20% or in case of a test being interrupted due to flow obstruction (automatically denoted by the system).

Test cartridges

In order to minimize any effect of lot-related variance during the evaluation, for both the Col/Epi and Col/ADP test cartridges, four lots were used in a random order throughout the study.

Statistical rationale and methods

Reference ranges for collagen/epinephrine (Col/Epi) and collagen/ADP (Col/ADP) test cartridges were estimated with pooled data from all sites. 309 individuals confirmed to have normal platelet function were included in calculation of reference ranges. This number of subjects exceeds the recommended number (>120) of the NCCLS guideline C 28-A ( Sasse, 1995). Reference ranges for the PFA test system were determined by calculating the 5th and 95th percentiles and the corresponding 90% central intervals.

Statistical analysis of differences between the means of the five study groups, the five sites or any subgroups was based on Student two-tail, two-sample t-testing if the skewness-value of the data allowed parametric analysis (skewness < 2.0). To analyse differences in cut-off values, we compared 90% confidence intervals around these 95th percentiles ( Reed et al, 1971 ). Within-site variance was expressed as the mean coefficient of variance (CV) value of all duplicate tests for the Col/Epi and Col/ADP test cartridges. Inter-site variance was calculated by comparing the site-specific Col/Epi and Col/ADP mean values with Student t-testing, in order to assess possible local population effects on PHC.


  1. Top of page
  2. Abstract
  6. Acknowledgements
  7. References


Based on their clinical background, a total of 359 subjects was found eligible for inclusion into the study. Of these subjects 50 (14%) were excluded from further evaluation, because normal platelet function in the collected samples could not be confirmed. In most cases exclusion was due to subnormal platelet aggregation results; in a few cases the platelet count was  <150 × 109/l, or haematocrit was < 35%. Distribution (skewness) of CT results of the included 309 subjects allowed non-parametric testing for further statistical analysis (Fig 1). Demographics, coagulation and haemostasis parameters of all 309 subjects and the respective subgroups are shown in 1Table I.


Figure 1. 00 s for Col/ADP.

Download figure to PowerPoint

Table 1. Table I. Demographics and laboratory parameters of 309 subjects (mean ±SD). * 5–95% central interval represents the time-window of 90% of the measured closure time (s).Thumbnail image of

Influence of gender

Although gender-related differences could be demonstrated in one demographic parameter (weight) and in several haemostasis parameters such as platelet count, haematocrit, fibrinogen, haemoglobin and prothrombin time, no significant effect on CT was observed ( 1 Table I). This result permitted pooling of CT results from the two non-smoking groups (M and F) and from the two smoking groups (MS and FS).

Effect of smoking

Comparing 5th and 95th percentiles of the smoking group (MS+FS) with the non-smokers (M+F) demonstrated similar 5th percentiles for both test cartridge types and revealed a higher Col/Epi 95th percentile in the smoking group than the non-smoking group (161 s. v 143 s) ( Table II). Comparing the 90% confidence intervals around the Col/Epi 95th percentile ( Reed et al, 1971 ) the difference was significant for females (FS v F). In contrast, the 95th percentiles for Col/ADP were very similar. Compared to the non-smoking group, the smoking group also demonstrated significantly higher fibrinogen levels (2.74 v 2.50 g/l).

Table 2. Table II. Demographic and laboratory parameters of the smoking and non-smoking group (mean ±SD). * Values < 0.05 are indicated and considered significant.† 5–95% central interval represents the time-window of 90% of the measured closure times (s).‡ Analysis of the 90% confidence intervals around the 95th percentiles showed a significant difference for the Col/Epi but not for the Col/ADP test cartridge.Thumbnail image of

Effect of oral contraceptive usage

The OC group was compared exclusively to the female control group (F). CTs for Col/Epi and Col/ADP were very similar in both groups ( Table I). Significant differences were observed only for prothrombin time (P < 0.001) and aPTT (P = 0.02).

Inter-site and within-site variance

Whereas inter-site comparison of the calculated site-specific ranges showed occasional, small but significant differences in the 90% central intervals ( 3 Table III), no significant differences could be detected when site-specific ranges were compared with the overall reference ranges. The within-site variance of the duplicate measurements showed similar average CV values in the five sites for the Col/Epi Test cartridges from 4.7% to 5.8% (range 0–21%), and for Col/ADP Test cartridges from 3.7% to 4.9% (range 0–33%).

Table 3. Table III. Inter-site comparison by Student t testing of the means. Indicated are P values or non-significance (NS). * Mean ±SD and 90% central interval.Thumbnail image of

Reference ranges for the Col/Epi and Col/ADP test cartridges

Because of the similar central interval results for the five sites and five subgroups, all obtained data were pooled to calculate the Col/Epi and Col/ADP reference ranges for blood samples collected in buffered 105 mmol/l sodium citrate (Fig 1).

For the Col/Epi test cartridge a 90% central interval of 82–150 s was calculated, and for the Col/ADP the central interval was 62–100 s. Further details are given in 1Table I.

Effect of normal levels of fibrinogen, haematocrit and platelets on PHC

Normal levels of fibrinogen (1.00–4.5 g/l), haematocrit (35–50%) and platelet count (150–400 109/l) did not show any correlation with Col/Epi CT based on linear regression analysis. Similar results were obtained for Col/ADP (r2<0.01 for all three parameters with both types of test cartridges). In contrast, a clear correlation was observed between Col/Epi and Col/ADP CTs (r2 = 0.49).

Effect of FVIII concentrations on PHC

Linear regression analysis demonstrated a weak but significant inverse correlation between plasma levels of FVIII (60–240%) and Col/Epi CTs (r2 = 0.04) or Col/ADP CTs (r2 = 0.07) in normals.


  1. Top of page
  2. Abstract
  6. Acknowledgements
  7. References

As with the introduction of any new diagnostic test system, the clinical acceptance of the PFA-100® relies for the major part on clinical performance and standardization. To allow a correct interpretation of the results generated by the test system, establishment of appropriate reference ranges and cut-off values is a prerequisite. Similar to platelet aggregometry and other test systems dependent on functional cellular blood components, the PFA system cannot be calibrated by the use of control blood, since such controls are not available. Instead, the testing of an adequate number of blood samples, in which normal platelet function can be anticipated or has been confirmed, provides a reasonable alternative for calibration. In this study we evaluated blood samples from ostensibly healthy blood donors with confirmed normal platelet function.

A similar study conducted in the U.S.A. ( Mammen et al, 1998 ) characterized the performance of PFA with samples collected in 3.8% (129 mmol/l) buffered sodium citrate. Because more and more haematology laboratories might convert to using 3.2% (105 mmol/l) citrate based collection systems, consistent with NCCLS guideline H3-A4, the aim of the present study was to evaluate PFA performance in normal healthy subjects from the general population and to establish reference ranges in a standard 3.2% collection system.

From the 359 blood donors recruited 50 were excluded for further evaluation because a normal platelet function in the blood sample could not be confirmed, mostly due to subnormal aggregometry results. Retrospectively, it was determined that most of these drop-outs had normal CTs with the PFA, suggesting a higher specificity for PFA compared to platelet aggregometry, when results were compared to the normal clinical histories obtained from these blood donors.

Both smoking and use of oral contraceptives are considered parameters for increased risk of thrombosis ( Frederiksen et al, 1970 ). Uncertain still is their mode of action which might be directed primarily towards the vessel wall endothelium and/or cellular blood components like platelets ( Levine, 1973) and leucocytes ( Lehr et al, 1994 ). Smoking has been demonstrated to enhance TxA2 formation and raise the cellular level of platelet activation, suggesting that PHC might also be increased in people who smoke. Smoking also stimulates platelets in patients with increased blood pressure ( Gleerup & Winther, 1996), and has a somewhat obscure antagonistic behaviour to aspirin in coronary artery patients ( Davis et al, 1985 ). In the present study no effect of smoking on PHC was observed at the lower CT cut-off value for both Col/Epi and Col/ADP. However, the 95% cut-off for Col/Epi was slightly higher with borderline significance for the combined male and female smokers, and was significantly higher in the female smokers when compared to the non-smoking controls. Although an immediate and reversible platelet activating effect during smoking was not investigated, the results in the blood donors suggest that a certain level of desensitization in platelet function occurs in regular smokers. Further evaluations are needed to clarify whether the apparent inhibitory effects of smoking on platelet function in some donors warrants having such donors withdraw from smoking for a certain time prior to donation.

Oral contraceptive use has also been associated with increased risk of thromboembolic events ( Frederiksen et al, 1970 ). Even newer birth control medications are associated with increased risk for venous thrombosis ( Helmerhorst et al, 1997 ; Comp, 1997). The only significant differences observed between the OC and female control group were higher PT levels and shorter aPTT values in the OC group, although a trend for higher FVII:C levels was also seen. No effect on platelet function as measured in the PFA system was seen. This observation suggests that the prothrombotic character of OC medication is more likely to be related to changes in the balance of the plasma coagulation system rather than to an altered platelet function ( Kluft & Lansink, 1997). Moreover, the apparent changes in plasma concentrations of coagulation factors indicate that such factors do not play a major role in the PFA-100 test mechanism, an observation consistent with the findings that patients with haemophilia A or B have normal closure times ( Fressinaud et al, 1998 ; Carcao et al, 1998 ). Whether the PFA-100 test mechanism is only independent of these factors under anticoagulated conditions is still unclear.

Comparison of the closure time mean values and distributions of the participating sites demonstrated in a few cases a small difference, mostly of borderline significance. An explanation for these differences is unclear. Aside from possible ethnic and dietary differences from region to region, the mere fact that in-house logistics and pre-analytical routing of the blood samples differ could attribute to these findings. Therefore the reference ranges determined in this study should be considered general guidelines. Establishing in-house reference ranges is recommended, though expected to agree with these general guidelines.

In a seven-site field trial conducted in U.S.A., the corresponding reference ranges for normal samples collected in 129 mmol/l buffered citrate were 94–193 s for the Col/Epi and 71–118 s for the Col/ADP test cartridge ( Mammen et al, 1998 ). Compared to that study we found a significant shift to lower closure times for the 105 mmol/l citrate, an observation described earlier by Mammen et al (1995 ) in a small number of subjects. This finding is also in agreement with a more recent publication ( Heilmann et al, 1997 ), suggesting again that platelet function is influenced by different citrate concentrations in the collection systems. Therefore the citrate concentration used in the blood collection system has to be considered for correct interpretation of PFA test results.

The overall within-site variation ranging from 3.7% to 5.8% in the duplicate tests was acceptable, considering the complex test matrix in which the test process is taking place.

Moreover, the low number of samples which needed re-testing (1.45%), due to an unacceptable variation of >20% between the duplicates (0.8%) or to test interruptions resulting from flow obstructions in one of the two replicates (0.65%), strongly suggests that in routine usage there is no need to perform duplicate testing. This study does not exclude that the variance in abnormal samples could be higher.

The weak correlation between factor VIII levels and PHC is probably due to underlying VWF levels (not evaluated in this study), which are known to influence PHC ( Fressinaud et al, 1998 ) and, because of the VWF–VIII protein organization, correlate with FVIII levels in normals.

In an earlier report ( Kundu et al, 1995 ) the inverse correlation of platelet count and haematocrit level with PFA-100 closure time was described. These observations could not be confirmed by our data, possibly because in this study only subjects with a normal blood profile were included. Apparently, platelets and haematocrit effect PHC only at pathological levels. A possible role of fibrinogen in the test process could not be found, albeit that again only normal levels were evaluated.

In summary, centre specific reference ranges for Col/ADP and Col/Epi closure times show slight, mostly non-significant, local differences indicating the need for establishment of in-house reference ranges for the PFA-100 test system. OC usage and gender do not significantly influence PHC in normal subjects. The small effect of smoking on an increased Col/Epi upper cut-off value needs further explanation. Both smoking and oral contraceptive usage do not effect the closure time lower cut-off values suggesting that these behavioural variables do not cause a significant change in platelet function. Normal levels of haematocrit and platelet count in normal subjects do not significantly influence PHC. Reference ranges for 3.2% buffered citrate show a significant shift to shorter CTs when compared to the 3.8% reference ranges described in literature.


  1. Top of page
  2. Abstract
  6. Acknowledgements
  7. References

The excellent technical assistance of G. Franz, J. Klink, D. Lehmann, K. Lüdemann, S. Rahrig and Ralf Müller during this clinical evaluation is gratefully acknowledged.


  1. Top of page
  2. Abstract
  6. Acknowledgements
  7. References
  • 1
    Carcao, M.D., Blanchette, V.S., Dean, J.A., He, L.., Kern, M.A., Stain, A.M., Sparling, C.R., Stephens, D., Ryan, G., Freedman, J., Rand, M.L. (1998) The Platelet Function Analyzer (PFA-100®): a novel in-vitro system for evaluation of primary haemostasis in children. British Journal of Haematology, 101, 70 73.
  • 2
    Comp, P.C. (1997) Thrombophilic mechanisms of OCs. International Journal of Fertility and British Journal of Womens' Medicine, Suppl. 1, 170 176.
  • 3
    Davis, J., Hartman, C., Lewis, H., Shelton, L., Eigenberg, D., Hassanein, K., Hignite, C., Ruttinger, H. (1985) Cigarette smoking-induced enhancement of platelet function: lack of prevention by aspirin in men with coronary artery disease. Journal of Laboratory and Clinical Medicine, 105, 479 483.
  • 4
    Frederiksen, H. & Ravenholt, R. (1970) Thromboembolism, oral contraceptives and cigarettes. Public Health Reports, 85, 197 205.
  • 5
    Fressinaud, E., Veyradier, A., Truchaud, F., Martin, I., Boyer-Neumann, C., Trossaert, M., Meyer, D. (1998) Screening for von Willebrand disease with a new analyzer using high shear stress: a study of 60 cases. Blood, 91, 1325 1331.
  • 6
    George, J.N. & Shattil, S.J. (1991) The clinical importance of acquired abnormalities of platelet function. New England Journal of Medicine, 324, 27 39.
  • 7
    Gleerup, G. & Winther, K. (1996) Smoking further increases platelet activity in patients with mild hypertension. European Journal of Clinical Investigation, 26, 49 52.
  • 8
    Heilmann, E.J., Kundu, S.K., Sio, R., Garcia, C., Gomez, R., Christie, D. (1997) Comparison of four commercial citrate blood collection systems for platelet function analysis by the PFA-100TM system. Thrombosis Research, 87, 159 164.
  • 9
    Helmerhorst, F.M., Bloemenkamp, K.W., Rosendaal, F.R., Vandenbroucke, J.P. (1997) Oral contaceptives and thrombotic disease: risk of venous thromboembolism. Thrombosis and Haemostasis, 78, 327 333.
  • 10
    Kluft, C. & Lansink, M. (1997) Effect of oral contraceptives on haemostasis variables. Thrombosis and Haemostasis, 78, 315 326.
  • 11
    Kundu, S.K., Heilmann, E.J., Sio, R., Garcia, C., Ostgaard, R. (1995) Characterization of an in vitro Platelet Function Analyzer, PFA-100TM. Clinical and Applied Thrombosis/Hemostasis, 2, 241 249.
  • 12
    Lehr, H.-A., Frei, B., Arfors, K.-E. (1994) Vitamin C prevents cigarette smoke-induced leukocyte aggregation and adhesion to endothelium in vivo. Proceedings of the National Academy of Sciences of the United States of America, 91, 7688 7692.
  • 13
    Levine, P. (1973) An acute effect of cigarette smoking on platelet function: a possible link between smoking and arterial thrombosis. Circulation, 68, 619 623.
  • 14
    Lind, S.E. (1991) The bleeding time does not predict surgical bleeding. Blood, 77, 2547 2552.
  • 15
    Mammen, E.F., Alshameeri, R.S., Comp, P.C. (1995) Preliminary data from a field trial of the PFA-100TM System. Seminars in Thrombosis and Haemostasis, 21, 113 121.
  • 16
    Mammen, E.F., Comp, Ph.C., Gosselin, R., Greenberg, Ch., Hoots, W.K., Kessler, C.M., Larkin, E.C., Liles, D., Nugent, D. (1998) PFA-100TM System: a new method for assessment of platelet dysfunction. Seminars in Thrombosis and Haemostasis 24, 195 202.
  • 17
    Marshall, P.W., Williams, A.J., Dixonm, R.M., Growcott, J.W., Warburton, S., Armstrong, J., Moores, J. (1997) A comparison of the effects of aspirin on bleeding time measured using the Simplate method and closure time measured using the PFA-100®, in healthy volunteers . British Journal of Clinical Pharmacology, 44, 151 155.
  • 18
    Reed, A., Henry, R., Mason, W. (1971) Influence of a statistical method used on the resulting estimate of normal ranges. Clinical Chemistry 17, 275 284.
  • 19
    Rodgers, R.P.C. & Levin, J. (1990) A critical reappraisal of the bleeding time. Seminars in Thrombosis and Haemostasis, 16, 1 20.
  • 20
    Sasse, E.A. (1995) How to define and determine reference intervals in the clinical laboratory: approved guideline. NCCLS C28-A.
  • 21
    Wiseman, J.D. (1996) Evaluated tubes and additives for blood specimen collection–fourth edition; approved standard. NCCLS H1-A4.
  • 22
    Wiseman, J.D. (1998) Procedure for the collection of diagnostic blood specimens by venipuncture: approved standard–fourth edition. NCCLS H3-A4.