Clinically Relevant Differences in Prothrombin Time and INR Values Related to Blood Sample Collection in Plastic vs Glass Tubes

Transcription

1 Coagulation and Transfusion Medicine / PT/INR DIFFERENCES IN PLASTIC VS GLASS TUBES Clinically Relevant Differences in Prothrombin Time and INR Values Related to Blood Sample Collection in Plastic vs Glass Tubes Eberhard W. Fiebig, MD, 1,2 Joan E. Etzell, MD, 1,3 and Valerie L. Ng, PhD, MD 1,2 Key Words: Blood collection tubes; Glass; Plastic; Prothrombin time; International normalized ratio; INR; Warfarin therapy Abstract We compared prothrombin times (PTs) and international normalized ratios (INRs) for blood samples drawn into plastic vs glass collection tubes. We collected 60 venous blood samples into 4.5-mL glass and 2 plastic tubes (2.7 and 3.5 ml). An additional 153 samples, including 63 from warfarin-anticoagulated patients, were collected only in glass and 2.7-mL plastic tubes. The PTs and INRs were determined following routine laboratory procedures. A subset of 35 frozen aliquot samples was analyzed with a different instrument-reagent combination. The PTs and INRs for samples in plastic tubes were significantly lower than for samples in glass tubes. The mean INR differences increased with INR magnitude from approximately 0.1 (INR, 1.5) to 0.7 (INR, 4.5). Of the plastic tube INRs, 50% were more than 10% lower than INRs from samples collected in glass tubes. Therapeutic monitoring based on plastic-tube INRs could result in higher doses of warfarin. Clinical laboratories have gradually replaced glass blood collection tubes with plastic tubes in an effort to eliminate potential sharps injuries and biohazardous exposures associated with breakage of glass tubes. In recent years, the transition to plastic tubes has accelerated as regulatory agencies have enforced their use and put laboratories on notice to discontinue use of glass tubes as soon as feasible. Coagulation testing is known to be sensitive to a variety of preanalytic variables related to properties of the sample collection tube, including anticoagulant concentration, 1-4 filling level with blood sample, 5-7 and the composition of the tube itself Previous studies reported significant differences in thrombin time 8 and prothrombin time (PT) test results 9,11 for blood samples collected in commercial plastic vs glass tubes, although with the exception of thrombin time results, 8 the discrepancies were deemed too small to be clinically relevant. 9,11 However, these investigations compared glass and plastic tubes from the same manufacturer (when available) and did not systematically study plastic vs glass tubes from different manufacturers. We report the findings of a verification study that was designed to assess equivalency in PT and international normalized ratio (INR) values obtained on blood samples collected in plastic vs glass tubes. We compared samples drawn into plastic tubes from 2 manufacturers with samples collected in our standard glass tube, which had been in use in our clinical laboratory for decades. We focused on samples obtained from patients receiving long-term warfarin therapy, because therapeutic dosage adjustments are based on INR results and both overdosing and underdosing may be associated with bleeding and thromboembolic complications, respectively. 12 Accurate and consistent determination of these laboratory values in this particular patient population is of special importance. 902 Am J Clin Pathol 2005;124: Downloaded 902 from

2 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE Materials and Methods Study Design The study was designed as a prospective investigation that compared results obtained from blood samples collected into glass blood collection tubes ( reference samples) with those collected in parallel into plastic blood collection tubes ( study samples). The study was conducted under routine working conditions and following established clinical laboratory practices at San Francisco General Hospital (SFGH) and the University of California San Francisco Medical Center (UCSFMC). The study consisted of 3 parts conducted during an approximately 6-month period from May to November The study was approved by and performed in accordance with the UCSFMC Committee on Human Research guidelines. For the first part of the study, reference and study samples were obtained randomly from 60 SFGH outpatients for whom coagulation tests had been requested and who were referred for phlebotomy. Because the analysis of this initial study showed a significant bias for shorter PT results for samples collected in either plastic tube compared with the glass tube, we considered the possibility that the midpoint of PT reference ranges might be different for samples collected into plastic vs glass tubes. If this possibility were true, INR values calculated with the respective geometric means of PT reference ranges determined from specimens collected into glass or plastic tubes might have better agreement than the corresponding PTs. In the second part of the study, we obtained paired samples from 90 SFGH outpatients for whom no coagulation tests had been requested to determine PT reference ranges and corresponding INR values for glass and plastic tube samples. In addition, we compared INRs derived from 28 paired samples from outpatients receiving long-term warfarin therapy (samples from 8 of the 28 patients already had been analyzed as part of the original 60 samples for which only PT results were compared). Oral anticoagulation therapy and duration of warfarin treatment were confirmed by medical record review. In this comparison only 1 plastic tube (tube B, see Collection Tubes and Sample Collection ) was included because the first part of our study showed no relevant PT differences for specimens collected into either of the 2 plastic tubes. In the third part of the study, we used a routine interlaboratory comparison of INR values for quality assurance purposes to assess whether the observed PT and INR differences between samples collected in glass vs plastic tubes were specific to our coagulation instrument-reagent combination. Thirty-five outpatient samples were collected in glass and plastic tube B only. After separation of platelet-poor plasma from blood cells (see Sample Preparation, Coagulation Instrument, and Reagents ), the plasma was divided into 3 aliquots in polystyrene tubes. One part was run fresh at SFGH in parallel with the reference sample collected in glass. The other 2 parts were frozen within 1 hour of plasma preparation and stored at 70 C for fewer than 20 days. One part then was hand-delivered on dry ice within 1 hour to the Clinical Laboratory, UCSFMC, where it was stored for fewer than 48 hours at 20 C and then thawed and assayed under routine clinical laboratory conditions. The remaining aliquot was kept frozen at SFGH and then was thawed and tested on the same day as the aliquot assayed at UCSFMC. Collection Tubes and Sample Collection Reference samples were collected in non silicone-coated glass tubes with a 4.5-mL draw volume containing 0.5 ml (0.109 mol/l, 3.2%) buffered sodium citrate (Monoject Blue Stopper, Tyco Kendall Medical, Mansfield, MA). 13 This tube has been in use in the SFGH Clinical Laboratory for more than 20 years. Study samples were collected in 2 types of plastic tubes, each containing mol/l (3.2%) buffered sodium citrate (3.5-mL Vacuette, Greiner Bio-One North America, Monroe, NC, referred to as tube A, and 2.7-mL Vacutainer Plus, Becton Dickinson, Franklin Lakes, NJ, referred to as tube B). All samples were obtained from peripheral arm veins by certified phlebotomists and collected into evacuated tubes using cannulas or butterfly needles at the discretion of the phlebotomy staff. When samples for different laboratory studies were obtained, the tubes for coagulation testing were filled first, as recommended. 14 All study samples were obtained from the same venipuncture. Sample Preparation, Coagulation Instrument, and Reagents Glass and plastic tubes were received in the laboratory, checked for adequate tube filling, and centrifuged for 10 minutes at 3,200g to prepare platelet-poor plasma (<10,000 platelets per microliter) for 1-stage PT assays. In keeping with Clinical and Laboratory Standards Institute (formerly the National Committee for Clinical Laboratory Standards) guidelines, 15 collection tubes were kept unopened at 18 C to 24 C before separation of cells from plasma, and testing was completed within 6 hours of collection. Reference and study samples were tested within 1 hour of each other. At SFGH, PTs were determined by an automated coagulation analyzer (AMAX 190, Trinity Biotech, Bray, Ireland) using the instrument s mechanical clot detection mode and Thrombomax HS reagent with calcium (lots 082K6033 and M06535), both from Trinity Biotech. The international sensitivity indices (ISIs) assigned by the manufacturer for the instrument-reagent combinations were 1.32 (lot 082K6033) and 1.21 (lot M06535). The ISI 1.32 reagent lot was used for Downloaded from Am J Clin Pathol 2005;124:

3 Fiebig et al / PT/INR DIFFERENCES IN PLASTIC VS GLASS TUBES the initial 60 sample pairs in which we assessed only PT differences and for the first 28 samples from warfarin-anticoagulated patients. The ISI 1.21 reagent lot was in use when the second set of 35 samples from patients receiving warfarin therapy was compared. Both reagent lots were used in determining reagent-specific PT reference ranges and their geometric means from 90 outpatient samples. Two-level control samples (Accuclot I, II, Trinity Biotech) were assayed every 8 hours. PTs at UCSFMC were obtained with the MLA Electra 1400C analyzer (Beckman Coulter, Fullerton, CA) using optical clot detection in combination with Thromboplastin-DS reagent (lot ; Pacific Hemostasis, Huntersville, NC). The ISI assigned by the manufacturer for the optical detection method-reagent combination was At least 2 different level controls were run every 8 hours. Both laboratories participate in quarterly proficiency testing challenges for coagulation testing offered by the College of American Pathologists. PT and INR proficiency testing results during the study period were all within 2 SD from the peer group mean, and all challenges were completed successfully. Calculation of INR and Localized ISI Values INR values were calculated using the conventional formula (see below). For samples analyzed at both SFGH and UCSFMC, each site used ISI values and geometric means of PT reference ranges established for the specific thromboplastin reagent used for PT analysis. To explore mathematical correction as a means to harmonize INR values derived from glass and plastic tube samples at SFGH, we back-calculated a localized ISI value for samples collected in plastic tubes by solving for individual ISIs using the INR formula: INR Glass Tube = PT Plastic Tube Log conversion results in: ISI PT Geometric Mean of PT Reference Range for Plastic Tube Log INR Glass Tube =Log PT Plastic Tube ISI = Log INR Glass Tube Log PT Plastic Tube PT Geometric Mean of PT Reference for Plastic Tube PT Geometric Mean of PT Reference Range for Plastic Tube ISI ISI values thus were calculated for each plastic-glass sample pair in the study. The localized ISI value was the mean of the back-calculated ISI values. Use of the localized ISI minimizes INR differences between sample pairs in the comparison. 16 Statistical Analysis Agreement between PT and INR values for glass vs plastic tube samples was assessed by using Bland-Altman plots of differences between paired samples against the average of the values. 17,18 When appropriate (ie, no significant correlation between differences and average values), mean differences and 2 SD limits of the means (limits of agreement) were calculated for the entire sample population; otherwise mean differences and 95% confidence intervals (CIs) of the means were calculated for subpopulations. Differences between plastic and glass tube PT and INR values were compared by using a paired t test. Differences in proportions of sample pairs for which INR differences exceeded 10% were assessed by using the Fisher exact test. The 10% threshold was chosen because it has been established as a benchmark for clinically relevant INR differences 19 and was used previously to assess acceptable agreement between paired INRs. 11,19,20 Correlation between variables was studied by calculating the Pearson correlation coefficient (r) and Fisher r to z transformation (P). In all statistical comparisons, P values less than.05 were considered significant. Simple linear regression was used to estimate bias size at variable INR values. Percentage differences between glass and plastic tube results were calculated as ratio of plastic/glass tube results 1, ie, assuming the glass tube result to be 100%. Descriptive statistical values (arithmetic and geometric means, SD) were calculated in the conventional manner. Calculations were performed on a personal computer using a statistical software package (Statview 5.0.1, SAS Institute, Cary, NC). Results PT Comparisons As shown in Figure 1A and Figure 1B, PTs obtained on samples collected in either of the 2 plastic tubes were systematically shorter than PTs for samples collected in glass tubes. The differences between glass and plastic tube results strongly correlated with the magnitude of the average PTs (r = 0.87; P <.0001 for both plastic tubes), and average mean differences changed with the degree of PT prolongation. For 37 samples with PTs within the laboratory s reference range (PT 14.4 seconds, previously established from samples collected in glass tubes), the differences between PTs for glass and plastic tubes were negligible (mean, 0.1 second or 1% for tube A; 0.2 second or 2% for tube B; 95% CI, 5% to +3% and 6% to +3%, respectively). For higher PTs, differences were large enough to cause concern. Mean percentage differences (95% CI) for the 23 PTs that exceeded the reference range were 7% ( 16% to +3%) for tube A and 8% ( 18% to +4%) for tube B. The same values for the subgroup of the 5 highest PTs ( 30 seconds) were 12% ( 23% to 1%) and 14% ( 26% to 2%), respectively. 904 Am J Clin Pathol 2005;124: Downloaded 904 from

4 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE A B PT Plastic Tube A Glass (s) PT Plastic Tube B Glass (s) Average PT Plastic Tube A and Glass (s) Average PT Plastic Tube B and Glass (s) C PT Plastic Tube A Plastic Tube B (s) Average PT Plastic Tube A and B (s) Figure 1 Prothrombin time (PT) difference plots for glass and plastic tube samples. A, PT differences for 60 sample pairs obtained in plastic tube A (Vacuette) and a standard glass tube (Monoject Blue Stopper) plotted against the PT average from each sample pair. Added are the linear regression line of PT differences against the averages (dotted line), the zero difference (solid line), and ± 10% limits (dashed lines). The magnitude of PT differences increased disproportionately with the average length of the PTs (regression line crosses lower 10% limit of differences). B, Results were nearly identical to those in A when the differences between PTs from plastic tube B (Vacutainer Plus) and the glass tube were plotted against their averages. C, In contrast, PT differences between the 2 plastic tubes were not significantly correlated with the average PTs from these tubes and were well within ± 10% limits throughout the PT range in this study. For proprietary information, see the text. In contrast, all PTs collected in the 2 plastic tubes agreed well Figure 1C. The mean difference was small (0.1 second or 1%), and the limits of agreement (mean ± 2 SD) were well within ± 10% ( 0.7 to +0.9 second; 3% to +5%). INR Value Comparisons Reference range verification for samples collected in glass tubes and plastic tube B yielded slightly different geometric means (glass, 13.1 seconds; plastic, 12.9 seconds; n = 87; P <.0001). As shown in Figure 2A, INR values for 28 glass and plastic tube samples obtained from patients receiving warfarin therapy, calculated with their respective geometric means and the assigned ISI value (1.32), showed a similar pattern of disagreement as PTs obtained from a different set of specimens described in the preceding PT Comparisons section (samples from 8 patients were included in both comparisons). The differences between INR values for glass and plastic tubes and their averages were strongly correlated (r = 0.76; P <.0001).The bias was lower in the subtherapeutic INR range, ie, less than 2.0, and higher for INR values from 2.0 to 4.6. Based on the observed relationship, the mean INR bias for plastic tubes was projected to be 0.1 at an average INR of 1.5 ( 5%) with an additional bias of approximately 0.2 per 1 point INR increase to 0.7 at an INR of 4.5 ( 16%). As another measurement of agreement, we compared the proportion of sample pairs for which INR differences exceeded 10% in samples from the 28 patients receiving oral anticoagulant therapy: 2 (25%) of 8 sample pairs with average INRs less than 2.0 fell into this category; the differences were 13.0% and 14.2%, respectively. In contrast, 12 (60%) of 20 sample pairs with average INRs from 2.0 to 4.6 INR exceeded the 10% benchmark value with observed differences ranging from 10.01% to 16.9%. Overall, INRs in 14 (50%) of 28 plastic-glass sample pairs differed by more than 10%, ie, were at least 10% lower for samples collected in Downloaded from Am J Clin Pathol 2005;124:

5 Fiebig et al / PT/INR DIFFERENCES IN PLASTIC VS GLASS TUBES A INR Plastic Tube B Glass Average INR Plastic Tube B and Glass B INR Plastic Tube B Glass Average INR Plastic Tube B and Glass Figure 2 International normalized ratio (INR) difference plots for glass and plastic tube samples with and without INR correction of results from plastic tube samples. Added are the linear regression line of INR differences against the averages (dotted line), the zero difference (solid line), and ± 10% limits (dashed lines). A, Differences between INRs from samples obtained in plastic tube B and standard glass tube (n = 28 sample pairs) plotted against the average INRs from each sample pair. INR differences were inversely related to average INRs and increased disproportionately with the magnitude of INRs. Results for samples from patients who had been receiving warfarin for less then 6 weeks are plotted as triangles; warfarin therapy for more than 6 weeks is indicated by open circles. INR values were calculated from the respective geometric means for glass and plastic tube reference ranges and the manufacturer s assigned international sensitivity index (ISI) value (1.32). B, Results for the same variables are shown after INR values for plastic tubes were recalculated with a localized ISI value (1.5) that minimizes differences with glass tube results. Note correction of bias but not result variability. plastic vs glass tubes. In all comparisons, the duration of warfarin therapy (<6 weeks vs >6 weeks) did not seem to influence agreement (Figure 2A). Agreement between samples in glass and plastic tubes could be improved by replacement of the manufacturer s assigned ISI with a back-calculated localized ISI value (1.5) in the INR calculation for samples in plastic tubes Figure 2B. With this mathematical correction, the mean INR bias was reduced to 0.01 or 0.4% (P =.6, glass vs plastic tube INRs) and the proportion of paired samples with more than 10% INR differences to 3 (11%) of 28 samples (P =.003) compared with INRs calculated with manufacturer-assigned ISIs. However, the limits of agreement remained relatively wide ( 0.26 to 0.29 or 12% to +13%) as some of the corrected INR values for plastic tube samples were notably higher than the corresponding values determined from samples in glass tubes (Figure 2B). INR Value Comparisons Using a Second Instrument- Reagent Combination INR comparison was performed with another set of 35 samples in glass and plastic tubes from outpatients receiving longterm warfarin therapy. These samples were analyzed in parallel in 2 clinical laboratories with different reagent-instrument combinations and showed similar, statistically significant INR differences between sample aliquots in glass and plastic tubes Figure 3 and Table 1. Furthermore, the number of sample pairs at each site for which INR differences between samples in glass and plastic tubes exceeded 10% was virtually identical for the 30 of 35 sample pairs with average INRs between 2.0 and 4.0 (SFGH, fresh samples, 15/30 [50%]; frozenthawed samples, 16/30 [53%]; UCSFMC, frozen-thawed samples, 15/30 [50%]; all comparisons were statistically nonsignificant). INRs in these comparisons were calculated with geometric means of PT reference ranges specifically determined for blood samples collected in plastic (13.9 seconds) vs glass tubes (14.1 seconds) for samples analyzed at SFGH or with a single geometric mean value for glass and plastic tube aliquots that had been established for routine clinical use at that site (samples analyzed at UCSFMC). In contrast with glass-plastic tube comparisons, glassto-glass and plastic-to-plastic tube comparisons between the 2 laboratories showed no statistically significant INR differences for aliquots collected in the same tube type (Table 1). 906 Am J Clin Pathol 2005;124: Downloaded 906 from

6 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE A B INR Plastic Glass (SFGH) Average INR Plastic, Glass (SFGH) INR Plastic Glass (UCSFMC) Average INR Plastic, Glass (UCSFMC) Figure 3 International normalized ratio (INR) difference plots for glass and plastic tube samples analyzed at 2 laboratories. Added are the linear regression line of INR differences against the averages (dotted line), the zero difference (solid line), and ± 10% limits (dashed lines). A, INR differences between paired glass and plastic samples against their averages from a second group of samples from 35 patients receiving warfarin therapy (triangles, warfarin <6 wk; open circles, >6 wk). Sample aliquots had been stored frozen at 70 C before testing. Results were nearly identical to those for aliquots that had been tested fresh (not shown). B, The same plot for frozen-thawed sample aliquots that were tested at a second clinical laboratory on the same day as the samples described in A using a different reagent-instrument combination. Note similar result bias at both laboratories. SFGH, San Francisco General Hospital; UCSFMC, University of California San Francisco Medical Center. Discussion The principal finding of this study was a significant bias between PT and INR results derived from blood samples collected in 2 common types of plastic collection tubes (2.7-mL Vacutainer Plus and 3.5-mL Vacuette) compared with our standard glass tube (4.5-mL Monoject Blue Stopper). PTs obtained for samples collected in plastic tubes generally were shorter and INRs lower compared with paired samples collected in glass tubes. The differences increased disproportionately for higher PTs and INRs, thus disproportionately affecting samples from patients with elevated PTs and INRs or patients receiving long-term warfarin therapy in this study. The bias for PT values in the reference range (PT glass tube, <14.4 seconds) was statistically significant but negligible from a practical standpoint, ie, was less than 5% for most paired samples. On the other hand, 50% of 63 evaluable samples in plastic tubes from patients receiving warfarin therapy in this study (54% for those with INRs ) yielded INR values that were more than 10% lower than paired samples in glass tubes. This level of disagreement exceeds what is considered acceptable for clinical purposes because a bias of this magnitude could affect management of oral anticoagulation therapy. 11,19,20 Absolute differences generally were less than 1.0 INR and, thus, were smaller than so-called critical INR differences that establish with 95% Table 1 Comparison of INR Values From Samples Collected in Glass and Plastic Tubes * Sample Glass Plastic Difference P SFGH fresh 2.6 (0.8) 2.3 (0.7) 0.2 (0.3) <.0001 SFGH frozen-thawed 2.6 (0.8) 2.3 (0.7) 0.3 (0.3) <.0001 UCSFMC frozen-thawed 2.4 (1.2) 2.1 (0.9) 0.3 (0.4) <.0001 INR, international normalized ratio; SFGH, San Francisco General Hospital; UCSFMC, University of California San Francisco Medical Center. * Summarized are INR values and their differences (mean, SD) for 35 glass (Monoject) and plastic tube (Vacutainer Plus, tube B) sample aliquots obtained from outpatients receiving warfarin therapy. These were assayed fresh (SFGH only) and after frozen storage (SFGH and UCSFMC) using different coagulation instrument-reagent combinations (see the Materials and Methods section). Although INR differences between glass and plastic tube samples were statistically different at both locations, INR values from samples collected in glass or plastic tubes did not differ statistically when glass-to-glass and plastic-to-plastic tube results were compared. For proprietary information, see the text. P =.2 (SFGH glass vs UCSFMC glass). P =.07 (SFGH plastic vs UCSFMC plastic). Downloaded from Am J Clin Pathol 2005;124:

7 Fiebig et al / PT/INR DIFFERENCES IN PLASTIC VS GLASS TUBES probability a true difference from previous INR measurements 21,22 but in some cases were sufficiently large to raise concern that they could influence warfarin prescription. Because samples collected in plastic tubes systematically gave lower results than those collected in glass tubes, decisions based on the former likely would result in an increase in warfarin dosage (or failure to reduce current dosing) than if they were based on samples collected in glass tubes. Previous studies of the Vacutainer Plus plastic tube also documented statistically significant shorter PTs for samples collected in plastic compared with standard glass tubes. 9,11 The differences, however, generally were less than 10% and, thus, were deemed too small to be clinically relevant. It is important to note that in these studies, Becton Dickinson glass tubes served as standard against which plastic tubes from the same manufacturer or a competitor were tested. In the present study, we compared plastic tubes from 2 manufacturers with a glass tube from a third maker. The results between the 2 plastic tubes were in good agreement but showed the same relatively large bias vis-à-vis the glass tube results. This suggests a systematic difference in glass and plastic tube properties that leads to a PT and INR bias that might be tolerable for some glass and plastic tube combinations but unacceptably large for others, as shown here. Because of the limited scope of this study, which focused on evaluating whether PT and INR results for samples in plastic tubes would be different from those obtained for samples in our current glass tubes, we did not include other brands of glass tubes in the comparison. The reason that coagulation times are shorter for samples collected in plastic compared with glass tubes is unknown. Given that at least in the tubes studied here the concentration of citrate anticoagulant in the tubes was the same, the differences presumably relate to variability in blood draw volumes (Monoject glass tube, 4.5 ml; Vacutainer Plus plastic tube, 2.7 ml; Vacuette plastic tube, 3.5 ml) or, more likely, differences in surface properties of the tube materials that come into contact with the blood sample (glass vs polypropylene). Although it is well documented that PT and INR values might differ depending on the composition of the tubes in which blood samples are obtained, these values are expected to be lower with samples collected in glass compared with plastic tubes, 10,23 not higher, as we observed. Shortening of the PT in blood samples in contact with glass is thought to be due to coagulation factor activation, 10,23 a phenomenon that is more pronounced when samples are stored in the collection tube for 4 hours or longer and when the tubes are stored at 4 C. 10 We explored mathematical correction as a means to harmonize disparate results for samples collected in glass and plastic tubes. INR values for samples in plastic tubes were adjusted by replacing the manufacturer s ISI with a back-calculated localized ISI that minimizes the average difference between samples in glass and plastic tubes. Back-calculated localized ISIs were approximately 0.2 points higher than ISIs assigned by the manufacturer to compensate for the lower PTs obtained on the samples in plastic tubes. As expected, this approach worked well to eliminate result bias between the samples but did not reduce variability to the same degree so that limits of agreement still exceeded the usually acceptable ceiling of ± 10%. Our findings demonstrate again that tube type, ie, glass vs plastic, can affect PT and INR results at a clinically relevant level and needs to be considered among the preanalytic variables, such as citrate anticoagulant concentration and filling level, that can influence these coagulation tests. This emphasizes the need for individual clinical laboratories to perform their own verification studies, including comparison of samples from patients receiving warfarin before switching from glass to plastic tubes, especially when the switch involves a change in manufacturer as well. For the laboratory community in general, systematic peer-reviewed studies of various tube types that are marketed for clinical coagulation testing would be desirable to assess the overall variability contributed by tube type and to devise strategies on how to minimize this concern. Although plastic blood collection tubes have the potential to reduce the sharps exposure from broken glass tubes to laboratory professionals, it is interesting to note that glass tube breakage is not a common event in clinical laboratories. 24 From the 1 Department of Laboratory Medicine, University of California San Francisco; and the Clinical Laboratory, 2 San Francisco General Hospital and 3 University of California San Francisco Medical Center. Address reprint requests to Dr Fiebig: Clinical Laboratory NH 2M9, San Francisco General Hospital, 1001 Potrero Ave, San Francisco, CA Acknowledgments: We gratefully acknowledge the support of the Division of Specimen Collection and Management at SFGH Clinical Laboratory and expert technical assistance by Ambee Miranda, SFGH, and Kathy Kushner and Jackie Barnick, UCSFMC. We also thank Jack Levin for valuable critical comments and suggestions. References 1. Adcock DM, Kressin DC, Marlar RA. Effect of 3.2% vs 3.8% sodium citrate concentration on routine coagulation testing. Am J Clin Pathol. 1997;107: Chantarangkul V, Tripodi A, Clerici M, et al. Assessment of the influence of citrate concentration on the international normalized ratio (INR) determined with twelve reagentinstrument combinations. Thromb Haemost. 1998;80: Duncan EM, Casey CR, Duncan BM, et al. Effect of concentration of trisodium citrate anticoagulant on calculation of the international normalised ratio and the international sensitivity index of thromboplastin. Thromb Haemost. 1994;72: Am J Clin Pathol 2005;124: Downloaded 908 from

8 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE 4. Ingram GI, Hills M. The prothrombin time test: effect of varying citrate concentration. Thromb Haemost. 1976;36: Adcock DM, Kressin DC, Marlar RA. Minimum specimen volume requirements for routine coagulation testing: dependence on citrate concentration. Am J Clin Pathol. 1998;109: Peterson P, Gottfried EL. The effects of inaccurate blood sample volume on prothrombin time (PT) and activated partial thromboplastin time (aptt). Thromb Haemost. 1982;47: Reneke J, Etzell J, Leslie S, et al. Prolonged prothrombin time and activated partial thromboplastin time due to underfilled specimen tubes with 109 mmol/l (3.2%) citrate anticoagulant. Am J Clin Pathol. 1998;109: Flanders M, Crist R, Rodgers G. A comparison of blood collection in glass versus plastic Vacutainers on results of esoteric coagulation assays. Lab Med. 2003;34: Gosselin RC, Janatpour K, Larkin EC, et al. Comparison of samples obtained from 3.2% sodium citrate glass and two 3.2% sodium citrate plastic blood collection tubes used in coagulation testing. Am J Clin Pathol. 2004;122: Palmer RN, Gralnick HR. Cold-induced contact surface activation of the prothrombin time in whole blood. Blood. 1982;59: Tripodi A, Chantarangkul V, Bressi C, et al. How to evaluate the influence of blood collection systems on the international sensitivity index: protocol applied to two new evacuated tubes and eight coagulometer/thromboplastin combinations. Thromb Res. 2002;108: Schulman S. Clinical practice; care of patients receiving longterm anticoagulant therapy. N Engl J Med. 2003;349: Tyco Healthcare Group LP. Monoject Blue Stopper. Available at: Accessed June 21, National Committee for Clinical Laboratory Standards. Procedures for the collection of diagnostic blood specimens by venipuncture; approved standard. Wayne, PA: NCCLS; National Committee for Clinical Laboratory Standards. Collection, transport, and processing of blood specimens for testing plasma-based coagulation assays; approved guideline. H21-A4. 4th ed. Wayne, PA: NCCLS; Ng VL, Levin J, Corash L, et al. Failure of the international normalized ratio to generate consistent results within a local medical community. Am J Clin Pathol. 1993;99: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1: Bland JM, Altman DG. Comparing methods of measurement: why plotting difference against standard method is misleading. Lancet. 1995;346: WHO Expert Committee on Biological Standardization. Annex 3: guidelines for thromboplastins and plasma used to control oral anticoagulant therapy. In: World Health Organization Technical Report Series. 48th report. No Geneva, Switzerland: World Health Organization; 1999: Poller L, Van Den Besselaar AM, Jespersen J, et al. The effect of sample size on fresh plasma thromboplastin ISI determination. Br J Haematol. 1999;105: Lassen JF, Brandslund I, Antonsen S. International normalized ratio for prothrombin times in patients taking oral anticoagulants: critical difference and probability of significant change in consecutive measurements. Clin Chem. 1995;41: Lassen JF, Kjeldsen J, Antonsen S, et al. Interpretation of serial measurements of international normalized ratio for prothrombin times in monitoring oral anticoagulant therapy. Clin Chem. 1995;41: Rapaport SI, Aas K, Owren PA. The effect of glass upon the activity of the various plasma clotting factors. J Clin Invest. 1955;34: Dunn S. Compliance perspectives: the safe sharps race: a steady advance to the finish line. G-2 Compliance Report. 2004;VI:5-8. Downloaded from Am J Clin Pathol 2005;124: