Original Article
Ann Clin Biochem 1992; 29: 422-425
Within-subject and between-subject biological variation of prothrombin time and activated partial thromboplastin time Dolors Dot, Jaume Mir6 and X Fuentes-Arderiu! From the Laboratori d'Analisis Cliniques, Hospital de Viladecans, Viladecans, and 'Servei de Bioquimica Clinica, Hospital Princeps d'Espanya, L 'Hospitalet de Llobregat, Barcelona, Spain
The within-subject and between-subject biological variation of prothrombin time (PT) and activated partial thromboplastin time (APTT) have been studied in a group of 19 women and 20 men over a period of 9 months. All measurements were performed with a mechanical coagulometer. The between-subject coefficients of variation were 6' 80/0 for PT and 8· 9% for APTT. The medians and the 90th percentiles of within-subject coefficients of variation were 1. 7% and 5· 9%, respectively, for PT, and 0·0% and 6·1% for APTT. These data were used to calculate the critical difference for significant change detection and the desirable between-day imprecision.
SUMMARY.
Additional key phrases: analytical goals; index of individuality; critical difference; quality control
The application in clinical laboratory sciences of data on within-subject biological variation of biological quantities is now well established. I These applications are: (i) the establishment of metrological goals [in this article the adjective metrological (i.e. related to metrology) instead of analytical is used. The rationale is that metrological is a more general term than analytical, and, moreover, PT and APTT are time measurements but not analyses.] ; (ii) the evaluation of the significance of a change between two successive results; (iii) the assessment of the suitability of population reference values; (iv) the establishment of the number of specimens to be collected and analysed for the estimation of the 'homeostatic value' of a quantity; and (v) selection between biological quantities with the same clinical utility. However, to our knowledge, there is only one publication on within-subject biological variation of prothrombin time (PT) and activated partial thromboplastin time (APTT). 2 This article provides data on within-subject and betweensubject biological variation of these quantities, obtained over a 9 month period. Correspondence: Dr X Fuentes-Arderiu, Servei de Bioquimica Clfnica, Hospital Princeps d'Espanya, Feixa Llarga sin, 08907 L'Hospitalet de Llobregat, Spain.
METHODS AND MATERIALS
Subjects and specimens The participants were 39 apparently healthy volunteers, 19 women, aged 22 to 49 years, and 20 men, aged 20 to 45 years. The volunteers maintained their life styles, which involved no strenuous exercise, throughout the study period; none of them were taking drugs affecting the coagulation system. Over 9 months, generally at monthly intervals, venous blood specimens were drawn between 0800 and 1000 h with the volunteers in a sitting position. These collections were usually performed by a single phlebotomist using the Venojectf system (Terumo Europe, Leuven, Belgium), using 0'9x25 mm needles and evacuated blood-collection tubes containing trisodium citrate (final concentration: 0'013 mol/L), Blood specimens were centrifuged at 1880g for 10 min within 4 h of collection, and the plasmas were separated and measured. Using this protocol, the pre-metrological variation was considered negligible. Methods of measurement Both quantities were measured with the Amelung mechanical coagulometer model KC 10A (Heidrich Amelung, Lemgo, Germany). The reagent used for PT measurement was Thromboplastin a
422
Downloaded from acb.sagepub.com by guest on June 4, 2016
Biological variation of PT and APTT (Boehringer Mannheim, Mannheim, Germany, Ref. 244252), and the reagents used for APTT were PIT a (Boehringer Mannheim, Ref. 886050) and CaCl 2 O' 025 mollL. PT measurements are expressed using the international normalized ratio (lNR) and APTT measurements in relative time (ratio) with respect to the average time of all volunteers .. Each working day the quality of the measurements was controlled using one control material with 'physiological' values of both quantities (Ci-Trol I-X, Merz + Dade, Diidingen, Germany, Ref. 291086). Throughout the study the same batch of the control material was used. Statistical treatment For each quantity, the between-day metrological variance (S~b) and between-day metrological coefficient of variation (CV Mb) were estimated using the results observed in the 'physiological' control material included in every run. Between-subject biological variation was estimated as the coefficient of variation of the individual means (referred to as homeostatic values) of all volunteers. For each volunteer, the within-subject biological variance (s~w) was estimated as follows: 2
2
2
S Bw = S Tw - S Mb
where S2 is the overall within-subject variance.! The within-subject biological coefficients of variation (with respect to the homeostatic values; CV BW) were also calculated. After demonstrating that no significant differences existed between men and women with respect to the CV Bw values (using the Mann-Whitney test), the mean, 50th percentile (median) and 90th percentile of all CV Bw values were estimated. For each quantity the index of individuality used to assess the utility of reference values was calculated as i 2 2 -imedian sTjsob
wh~re. s~~ is the between-subject biological variation. The minimal difference between two successive measurements of a quantity in the same patient to be considered significant (P~ O' 05), i.e. the critical difference;' was calculated as
RESULTS The between-day metrological coefficients of variation (and the corresponding mean values
423
of the control material) were 6·40/0 (mean = 1.00) for PT, and 5·6% (mean=0'99) for APTT. The biological variation results are shown in Table 1. The critical differences (P ~ O'05) between two consecutive measurements calculated using the mean, the median and the 90th percentile of s~w are presented in Table 2. The desirable betweenday imprecisions of PT and APIT measurements have been defined as one-half of the withinsubject biological variation.s-? and these are also shown in Table 2. These desirable imprecisions have been estimated by averaging the medians of the CV Bw values obtained in this study and those obtained in the other published study.P Table 2 also shows the indices of individuality for PT and APTT. DISCUSSION Within-subject biological variation As may be seen from Table I, the within-subject coefficients of biological variation for PT and APTT observed in this study are clearly smaller than those observed by Costongs et al.' using frozen specimens assayed in one run for each subject. In the two studies the time span and the age and sex distribution of subjects were similar; only the number of volunteers was very different (n = 274 in the study of Costongs et al., and n = 39 in this study), but this fact is not sufficient to explain the observed differences. As PT and APTT values are determined by the simultaneous measurement of activities of different coagulation factors, the within-subject biological variation of PT and APTT should be determined by the within-subject biological variation of each participating coagulation factor. Because the biological variations of the different coagulation factor concentrations are not necessarily synchronous, their biological fluctuations should lead to a very small PT and APTT within-subject biological variation. The data shown in this study seem consistent with this hypothesis. Between-subject biological variation To our knowledge, no previous data have been published on the between-subject biological variation of PT and APTT. However, in spite of the fact that PT and APIT are measurements of a mixture of catalytic (enzymatic) activities in plasma, the CV Bb observed for these quantities are generally lower than the CV Bb observed in other studies for the catalytic activities of serum enzymes commonly measured in clinical chemistry. 8,9
Downloaded from acb.sagepub.com by guest on June 4, 2016
424
Dot, Mira and Fuentes-Arderiu
1. Mean. median and 90th percentile of the within-subject biological variation (CV Bw)' and betweensubject biological variation (CV B/J' all expressed as percentage coefficient of variation). PT = Prothrombin time; APTT = activated partial thromboplastin time
TABLE
Costongs et al.
This study
PT APTT
CV Bw (mean)
CV Bw (median)
CV Bw (90th percentile)
CV Bb
CV Bw (median)
CV Bw (90th percentile)
2·3 2·1
1'7 0·0
5·9 6·1
6·8 8·9
5·8 6·8
11·6 12·7
TABLE 2. Critical differences (P:S;; O·05), desirable between-day imprecision (CV D' %) and index of individuality. PT=Prothrombin time; APTT=activated partial thromboplastic time; INR = international normalized ratio
Critical difference using: 2
Quantity
SBw
PT (INR) APTT (ratio)
0·19 0·16
'
mean
s~w' median
s ~w' 90th percentile CV 0
0·18 0·15
0·25 0·23
Critical differences Not long ago, to evaluate the difference between two successive measurements of a quantity in a patient only the metrological variation was considered.P'! It is now well established that within-subject biological variation should be also considered.'? However, some controversy remains about the use of the mean, the median or other fractile of the different CV Bw values; it is clear from Table 2 that the use of different fractiles for the quantities studied produces very different critical differences. Moreover, it should be noted that the between-day metrological variance is the basis for the calculation of the critical differences, and should therefore be calculated by each laboratory. Bearing this in mind, it is obvious that the critical differences shown in Table 2 are only indicative. Desirable between-day imprecision Two proposals have been published for the establishment of desirable between-day imprecision for methods of measurement of haematological quantities: (i) a subjective approach using the opinion of physicians;" and (ii) an objective approach using the within-subject biological variation," Although the latter approach has been criticized.Pi" in the present study we have estimated desirable imprecision for each method of measurement using the mean within-subject biological variation, by averaging the data of this study and the previous report(Table 2). Thus, the values obtained this way are the best objective estimation of the desirable imprecision at present.
1·9 1·7
Index of individuality 0·96 0·68
Index of individuality The estimated indices of individuality suggest that PT and APTT are more appropriate for longitudinal comparisons (monitoring or individual reference values) than for transverse comparisons (diagnosis using population reference values)." This means that preoperative screening for coagulation disorders using PT and APTT may not be useful because a within-subject change of these quantities could be 'masked' by the between-subject biological variation. This lack of utility for preoperative screening is consistent with studies about the clinical performance of these quantities.Pi'? REFERENCES
2
3 4
5
6
Fraser CG, Harris EK. Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 1989; 27: 409-37 Costongs GMP1, 1anson PCW, Bas BM, Hermans 1, Brombacher P1, van Wersch 1W1. Short-term and long-term intra-individual variations and critical differences of haematological laboratory parameters. J Clin Chem Clin Biochem 1985; 23: 69-76 Harris EK. Distinguishing physiologic variation from analytic variation. J Chron Dis 1970; 23: 469-80 Harris EK. Effects of intra- and interindividual variation on the appropriate use of normal ranges. Clin Chem 1974; 20: 1535-42 Fuentes-Arderiu X, Cases-Regany E, Frey-Gonzalez E, et al. Interpretaci6n de un cambio entre dos valores consecutivos de una magnitud bioquimica. Qu{m Cltn 1989; 8: 357-61 Harris EK. Statistical principles underlying analytic goal-setting in clinical chemistry. Am J Clin Pathol 1979; 72: 374-82
Downloaded from acb.sagepub.com by guest on June 4, 2016
Biological variation of PT and APTT 7 Fraser CG. Desirable standards of hematological tests: a proposal. Am J Clin Pathol 1987; 88: 667-9 8 Juan-Pereira LJ. Variabilitat biologica intraindividual de les magnituds bioquimiques. Aplicacions cliniques. Doctoral Thesis. Barcelona: Facultat de Farrnacia, Universitat de Barcelona, 1989 9 Queralt6 JM. Analisis de series temporales en Bioquimica Clinica. Doctoral Thesis. Barcelona: Facultat de Medicina, Universitat Autonoma de Barcelona, 1990 10 Raphael SS. Lynch's Medical Laboratory Technology. Philadelphia: Saunders, 1976 II Miale JB. Laboratory Medicine. Hematology. Saint Louis: Mosby, 1977 12 Barnett RN. Clinical Laboratory Statistics. Boston: Little, Brown & Co., 1979 13 Harris EK, Yasaka T. On the calculation of 'reference change' for comparing two consecutive measurements. Clin Chem 1983; 29: 25-30
425
14 Skendzel LP, Barnett RN, Platt R. Medical useful criteria for analytic performance of laboratory tests. Am J Clin Pathol 1985; 83: 200-5 15 Barnett RN. Analytical goals. Am J Clin Pathol 1988; 89: 577 16 Skendzel LP. Response to the proposal of setting goals for imprecision based on average intraindividual biologic variation. Am J Clin Pathol 1988; 89: 578 17 Fraser CG. The application of theoretical goals based on biological variation data in proficiency testing. Arch Pathol Lab Med 1988; 112: 404-15 18 Suchman AL, Griner PF. Diagnostic uses of the activated partial thromboplastin time and prothrombin time. Ann Int Med 1986; 104: 810-16 19 Roizen MF. Routine preoperative evaluation. In: Miller R, editor. Anesthesia. 2nd ed. New York: Churchill Livingstone, 1986: 225-53
Accepted for publication 9 December 1991
Downloaded from acb.sagepub.com by guest on June 4, 2016
Ann Clin Biochem 1992; 29: 422-425
Within-subject and between-subject biological variation of prothrombin time and activated partial thromboplastin time Dolors Dot, Jaume Mir6 and X Fuentes-Arderiu! From the Laboratori d'Analisis Cliniques, Hospital de Viladecans, Viladecans, and 'Servei de Bioquimica Clinica, Hospital Princeps d'Espanya, L 'Hospitalet de Llobregat, Barcelona, Spain
The within-subject and between-subject biological variation of prothrombin time (PT) and activated partial thromboplastin time (APTT) have been studied in a group of 19 women and 20 men over a period of 9 months. All measurements were performed with a mechanical coagulometer. The between-subject coefficients of variation were 6' 80/0 for PT and 8· 9% for APTT. The medians and the 90th percentiles of within-subject coefficients of variation were 1. 7% and 5· 9%, respectively, for PT, and 0·0% and 6·1% for APTT. These data were used to calculate the critical difference for significant change detection and the desirable between-day imprecision.
SUMMARY.
Additional key phrases: analytical goals; index of individuality; critical difference; quality control
The application in clinical laboratory sciences of data on within-subject biological variation of biological quantities is now well established. I These applications are: (i) the establishment of metrological goals [in this article the adjective metrological (i.e. related to metrology) instead of analytical is used. The rationale is that metrological is a more general term than analytical, and, moreover, PT and APTT are time measurements but not analyses.] ; (ii) the evaluation of the significance of a change between two successive results; (iii) the assessment of the suitability of population reference values; (iv) the establishment of the number of specimens to be collected and analysed for the estimation of the 'homeostatic value' of a quantity; and (v) selection between biological quantities with the same clinical utility. However, to our knowledge, there is only one publication on within-subject biological variation of prothrombin time (PT) and activated partial thromboplastin time (APTT). 2 This article provides data on within-subject and betweensubject biological variation of these quantities, obtained over a 9 month period. Correspondence: Dr X Fuentes-Arderiu, Servei de Bioquimica Clfnica, Hospital Princeps d'Espanya, Feixa Llarga sin, 08907 L'Hospitalet de Llobregat, Spain.
METHODS AND MATERIALS
Subjects and specimens The participants were 39 apparently healthy volunteers, 19 women, aged 22 to 49 years, and 20 men, aged 20 to 45 years. The volunteers maintained their life styles, which involved no strenuous exercise, throughout the study period; none of them were taking drugs affecting the coagulation system. Over 9 months, generally at monthly intervals, venous blood specimens were drawn between 0800 and 1000 h with the volunteers in a sitting position. These collections were usually performed by a single phlebotomist using the Venojectf system (Terumo Europe, Leuven, Belgium), using 0'9x25 mm needles and evacuated blood-collection tubes containing trisodium citrate (final concentration: 0'013 mol/L), Blood specimens were centrifuged at 1880g for 10 min within 4 h of collection, and the plasmas were separated and measured. Using this protocol, the pre-metrological variation was considered negligible. Methods of measurement Both quantities were measured with the Amelung mechanical coagulometer model KC 10A (Heidrich Amelung, Lemgo, Germany). The reagent used for PT measurement was Thromboplastin a
422
Downloaded from acb.sagepub.com by guest on June 4, 2016
Biological variation of PT and APTT (Boehringer Mannheim, Mannheim, Germany, Ref. 244252), and the reagents used for APTT were PIT a (Boehringer Mannheim, Ref. 886050) and CaCl 2 O' 025 mollL. PT measurements are expressed using the international normalized ratio (lNR) and APTT measurements in relative time (ratio) with respect to the average time of all volunteers .. Each working day the quality of the measurements was controlled using one control material with 'physiological' values of both quantities (Ci-Trol I-X, Merz + Dade, Diidingen, Germany, Ref. 291086). Throughout the study the same batch of the control material was used. Statistical treatment For each quantity, the between-day metrological variance (S~b) and between-day metrological coefficient of variation (CV Mb) were estimated using the results observed in the 'physiological' control material included in every run. Between-subject biological variation was estimated as the coefficient of variation of the individual means (referred to as homeostatic values) of all volunteers. For each volunteer, the within-subject biological variance (s~w) was estimated as follows: 2
2
2
S Bw = S Tw - S Mb
where S2 is the overall within-subject variance.! The within-subject biological coefficients of variation (with respect to the homeostatic values; CV BW) were also calculated. After demonstrating that no significant differences existed between men and women with respect to the CV Bw values (using the Mann-Whitney test), the mean, 50th percentile (median) and 90th percentile of all CV Bw values were estimated. For each quantity the index of individuality used to assess the utility of reference values was calculated as i 2 2 -imedian sTjsob
wh~re. s~~ is the between-subject biological variation. The minimal difference between two successive measurements of a quantity in the same patient to be considered significant (P~ O' 05), i.e. the critical difference;' was calculated as
RESULTS The between-day metrological coefficients of variation (and the corresponding mean values
423
of the control material) were 6·40/0 (mean = 1.00) for PT, and 5·6% (mean=0'99) for APTT. The biological variation results are shown in Table 1. The critical differences (P ~ O'05) between two consecutive measurements calculated using the mean, the median and the 90th percentile of s~w are presented in Table 2. The desirable betweenday imprecisions of PT and APIT measurements have been defined as one-half of the withinsubject biological variation.s-? and these are also shown in Table 2. These desirable imprecisions have been estimated by averaging the medians of the CV Bw values obtained in this study and those obtained in the other published study.P Table 2 also shows the indices of individuality for PT and APTT. DISCUSSION Within-subject biological variation As may be seen from Table I, the within-subject coefficients of biological variation for PT and APTT observed in this study are clearly smaller than those observed by Costongs et al.' using frozen specimens assayed in one run for each subject. In the two studies the time span and the age and sex distribution of subjects were similar; only the number of volunteers was very different (n = 274 in the study of Costongs et al., and n = 39 in this study), but this fact is not sufficient to explain the observed differences. As PT and APTT values are determined by the simultaneous measurement of activities of different coagulation factors, the within-subject biological variation of PT and APTT should be determined by the within-subject biological variation of each participating coagulation factor. Because the biological variations of the different coagulation factor concentrations are not necessarily synchronous, their biological fluctuations should lead to a very small PT and APTT within-subject biological variation. The data shown in this study seem consistent with this hypothesis. Between-subject biological variation To our knowledge, no previous data have been published on the between-subject biological variation of PT and APTT. However, in spite of the fact that PT and APIT are measurements of a mixture of catalytic (enzymatic) activities in plasma, the CV Bb observed for these quantities are generally lower than the CV Bb observed in other studies for the catalytic activities of serum enzymes commonly measured in clinical chemistry. 8,9
Downloaded from acb.sagepub.com by guest on June 4, 2016
424
Dot, Mira and Fuentes-Arderiu
1. Mean. median and 90th percentile of the within-subject biological variation (CV Bw)' and betweensubject biological variation (CV B/J' all expressed as percentage coefficient of variation). PT = Prothrombin time; APTT = activated partial thromboplastin time
TABLE
Costongs et al.
This study
PT APTT
CV Bw (mean)
CV Bw (median)
CV Bw (90th percentile)
CV Bb
CV Bw (median)
CV Bw (90th percentile)
2·3 2·1
1'7 0·0
5·9 6·1
6·8 8·9
5·8 6·8
11·6 12·7
TABLE 2. Critical differences (P:S;; O·05), desirable between-day imprecision (CV D' %) and index of individuality. PT=Prothrombin time; APTT=activated partial thromboplastic time; INR = international normalized ratio
Critical difference using: 2
Quantity
SBw
PT (INR) APTT (ratio)
0·19 0·16
'
mean
s~w' median
s ~w' 90th percentile CV 0
0·18 0·15
0·25 0·23
Critical differences Not long ago, to evaluate the difference between two successive measurements of a quantity in a patient only the metrological variation was considered.P'! It is now well established that within-subject biological variation should be also considered.'? However, some controversy remains about the use of the mean, the median or other fractile of the different CV Bw values; it is clear from Table 2 that the use of different fractiles for the quantities studied produces very different critical differences. Moreover, it should be noted that the between-day metrological variance is the basis for the calculation of the critical differences, and should therefore be calculated by each laboratory. Bearing this in mind, it is obvious that the critical differences shown in Table 2 are only indicative. Desirable between-day imprecision Two proposals have been published for the establishment of desirable between-day imprecision for methods of measurement of haematological quantities: (i) a subjective approach using the opinion of physicians;" and (ii) an objective approach using the within-subject biological variation," Although the latter approach has been criticized.Pi" in the present study we have estimated desirable imprecision for each method of measurement using the mean within-subject biological variation, by averaging the data of this study and the previous report(Table 2). Thus, the values obtained this way are the best objective estimation of the desirable imprecision at present.
1·9 1·7
Index of individuality 0·96 0·68
Index of individuality The estimated indices of individuality suggest that PT and APTT are more appropriate for longitudinal comparisons (monitoring or individual reference values) than for transverse comparisons (diagnosis using population reference values)." This means that preoperative screening for coagulation disorders using PT and APTT may not be useful because a within-subject change of these quantities could be 'masked' by the between-subject biological variation. This lack of utility for preoperative screening is consistent with studies about the clinical performance of these quantities.Pi'? REFERENCES
2
3 4
5
6
Fraser CG, Harris EK. Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 1989; 27: 409-37 Costongs GMP1, 1anson PCW, Bas BM, Hermans 1, Brombacher P1, van Wersch 1W1. Short-term and long-term intra-individual variations and critical differences of haematological laboratory parameters. J Clin Chem Clin Biochem 1985; 23: 69-76 Harris EK. Distinguishing physiologic variation from analytic variation. J Chron Dis 1970; 23: 469-80 Harris EK. Effects of intra- and interindividual variation on the appropriate use of normal ranges. Clin Chem 1974; 20: 1535-42 Fuentes-Arderiu X, Cases-Regany E, Frey-Gonzalez E, et al. Interpretaci6n de un cambio entre dos valores consecutivos de una magnitud bioquimica. Qu{m Cltn 1989; 8: 357-61 Harris EK. Statistical principles underlying analytic goal-setting in clinical chemistry. Am J Clin Pathol 1979; 72: 374-82
Downloaded from acb.sagepub.com by guest on June 4, 2016
Biological variation of PT and APTT 7 Fraser CG. Desirable standards of hematological tests: a proposal. Am J Clin Pathol 1987; 88: 667-9 8 Juan-Pereira LJ. Variabilitat biologica intraindividual de les magnituds bioquimiques. Aplicacions cliniques. Doctoral Thesis. Barcelona: Facultat de Farrnacia, Universitat de Barcelona, 1989 9 Queralt6 JM. Analisis de series temporales en Bioquimica Clinica. Doctoral Thesis. Barcelona: Facultat de Medicina, Universitat Autonoma de Barcelona, 1990 10 Raphael SS. Lynch's Medical Laboratory Technology. Philadelphia: Saunders, 1976 II Miale JB. Laboratory Medicine. Hematology. Saint Louis: Mosby, 1977 12 Barnett RN. Clinical Laboratory Statistics. Boston: Little, Brown & Co., 1979 13 Harris EK, Yasaka T. On the calculation of 'reference change' for comparing two consecutive measurements. Clin Chem 1983; 29: 25-30
425
14 Skendzel LP, Barnett RN, Platt R. Medical useful criteria for analytic performance of laboratory tests. Am J Clin Pathol 1985; 83: 200-5 15 Barnett RN. Analytical goals. Am J Clin Pathol 1988; 89: 577 16 Skendzel LP. Response to the proposal of setting goals for imprecision based on average intraindividual biologic variation. Am J Clin Pathol 1988; 89: 578 17 Fraser CG. The application of theoretical goals based on biological variation data in proficiency testing. Arch Pathol Lab Med 1988; 112: 404-15 18 Suchman AL, Griner PF. Diagnostic uses of the activated partial thromboplastin time and prothrombin time. Ann Int Med 1986; 104: 810-16 19 Roizen MF. Routine preoperative evaluation. In: Miller R, editor. Anesthesia. 2nd ed. New York: Churchill Livingstone, 1986: 225-53
Accepted for publication 9 December 1991
Downloaded from acb.sagepub.com by guest on June 4, 2016
