PEDIATRICS Vol. 118 No. 6 December 2006, pp. e1650-e1656 (doi:10.1542/peds.2006-0023)
ARTICLE |
Extent of Agreement in Gentamicin Concentration Between Serum That Is Drawn Peripherally and From Central Venous Catheters
a Departments of Pharmacy
b Nursing
c Division of Haematology/Oncology
d Population Health Sciences, Research Institute, Hospital for Sick Children
e Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
OBJECTIVE. At our institution, patients who receive once-daily dosing of gentamicin have serum concentrations determined 3 and 6 hours after dose administration. Patients with single-lumen central venous catheters have the 3-hour samples drawn peripherally. The objective of this study was to evaluate the extent of agreement between peripheral and central venous catheter serum gentamicin concentrations drawn 3 hours after dose administration.
METHODS. In this prospective, observational study, patients provided both a peripheral and a central blood sample for determination of serum gentamicin concentration. The order of sampling (central venous catheter versus peripheral first) was randomized. Agreement was assessed by determination of the intraclass correlation coefficient and Bland-Altman analysis. The clinically acceptable targets for the lower limit of the intraclass correlation coefficient and Bland-Altman limits of agreement were defined a priori as >0.80 and ±6%, respectively. Differences between the theoretical dose adjustments using the central venous catheter versus the peripheral sample result were described.
RESULTS. Forty-five pairs of samples were collected: 42 from single-lumen implantable central venous catheters (ports) and 3 from peripherally inserted central venous catheters. The intraclass correlation coefficient was 0.91. However, the Bland-Altman analysis resulted in a mean percentage difference (central venous catheter versus peripheral) of –0.92% and limits of agreement of –27.9% to 26.0%. The gentamicin dose adjustment based on the central venous catheter sample result would have led to clinically significant dose adjustments in 19 (42%) cases, when compared with the peripheral sample result.
CONCLUSIONS. These results indicate a lack of agreement between peripheral and single-lumen central venous catheter samples. In particular, ports are not appropriate sites for monitoring serum gentamicin concentrations.
Key Words: gentamicin • therapeutic drug monitoring • reliability • central venous catheters • blood sampling
Abbreviations: CVC—central venous catheter • SGC—serum gentamicin concentration • CV—coefficient of variation • ICC—intraclass correlation coefficient • PICC—peripherally inserted central catheter • CI—confidence interval
For the purposes of serum drug concentration monitoring, it generally is recommended that blood samples be drawn from a catheter that is not used for drug administration. This technique, although feasible for patients with multilumen central venous catheters (CVCs), is not possible for patients with single-lumen CVCs. Consequently, patients with single-lumen CVCs must have samples obtained from a finger lancet puncture or venipuncture. Not only does peripheral blood sampling cause pain, but also patients may have to be wakened at night for samples to be obtained.
In our institution, patients who receive once-daily gentamicin doses have serum gentamicin concentrations (SGCs) determined 3 hours and 6 hours after the dose.1 The blood sample for the 3-hour gentamicin concentration determination is obtained peripherally (finger lancet puncture or venipuncture) because of a concern that results that are obtained from a sample that is drawn from the CVC may be falsely elevated by residual gentamicin that is present in the CVC. The 6-hour sample is obtained from the single-lumen CVC used for drug administration because it is thought that sufficient intravenous fluid by then has been administered through the CVC to clear any residual gentamicin. Patient quality of life would be enhanced if it were possible to obtain the 3-hour sample via the CVC. To do so with confidence, it is necessary to ensure agreement between the results that are generated from peripheral and CVC samples. The primary objective of this study was to evaluate the extent of agreement between peripheral and single-lumen CVC SGCs that are measured from blood that is drawn 3 hours after the administration of a gentamicin dose.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This prospective study was approved by our institution’s Research Ethics Board. Informed consent for study participation was obtained from children who were 16 years and older and the parent(s)/guardian(s) of children who were younger than 16 years; informed assent was obtained from children who were aged 7 to 15 years. Children were asked to participate in this study once per hospital admission and no more than twice in total.
Study Population
Patients who had a diagnosis of cancer, were 18 years or younger, had a single-lumen CVC, and were receiving once-daily gentamicin were eligible to participate in this study.
Study Procedures
Participating patients had a peripheral blood sample drawn 3 hours after a gentamicin dose as per the standard of care and also had a blood sample drawn from the CVC that was used to administer the dose within 15 minutes of the peripheral sample. Block randomization (block size of 4) was used to randomize patients to the order of blood sampling site (ie, peripheral followed by CVC or vice versa).
Blood samples from the CVC were obtained by Hematology/Oncology nursing staff with a discard volume of 3 to 5 mL as per hospital policy. Peripheral blood samples were drawn by the phlebotomy team with a needle that was inserted into a peripheral vein or via finger lancet puncture. A blood volume of 0.5 mL was obtained for each sample. The gentamicin dosage, number of gentamicin doses received, type of CVC (implantable subcutaneous port, peripherally inserted, or external), discard blood volume, and volume of intravenous fluid administered through the CVC between the gentamicin dose and blood sampling were recorded for each patient.
With 1 exception, SGCs were analyzed by a microparticle automated immunoassay analyzer (Bayer Technicon Immuno 1; Bayer Corp, Tarrytown, NY). One sample pair was assayed by a random-access integrated immunoassay analyzer (Integrated Modular System; Bayer Diagnostics). Each pair of peripheral and CVC serum samples was assayed for gentamicin content on the same assay run.
Repeat Analysis of Samples
For provision of an independent assessment of the variation of the serum gentamicin assay, 15 patients were randomly selected to provide 2 additional blood samples from the CVC at the time of sampling. The triplicate CVC samples that were provided from each of these patients were used to estimate a coefficient of variation (CV) within the study period. A convenience sample of 15 patients was selected for this analysis. For these patients, 1 of the 3 CVC samples was arbitrarily selected to be used for assessment of agreement with the peripheral sample.
Calculated Gentamicin Dosage Adjustments on the Basis of Both 3-Hour Concentrations
The results that were obtained from the CVC and peripheral 3-hour samples were paired individually with the 6-hour sample results to calculate each patient’s parameters of gentamicin pharmacokinetic disposition using standard first-order equations.2 Theoretical gentamicin dosage adjustments were calculated to achieve a back-extrapolated maximum SGC at the end of the infusion of 20 to 25 mg/L and a drug-free interval (the period during which the gentamicin concentration is <2 mg/L) of at least 4 hours.1 Differences in gentamicin dosage adjustment that was generated by the result pairs were described.
When the second 3-hour blood sample was drawn at a time different from the first, the theoretical simultaneous SGC result for the second sample was calculated by back-extrapolation using the SGC result that was obtained 6 hours after dose administration. Differences in gentamicin dosage adjustment that was generated by including the theoretical simultaneous result pairs were described. A gentamicin dosage adjustment of >20% was deemed to be clinically significant.
Statistical Analysis
A minimum sample size of 45 gentamicin serum concentration pairs was necessary to achieve 80% power (ß = 0.2) to detect an intraclass correlation coefficient (ICC) of 0.90 with a significance level of .05.3,4 Power Analysis and Sample Size (PASS Statistical Software, Kaysville, UT) software was used for sample size calculation.
The agreement between the peripherally and centrally obtained SGC results was assessed using 2 methods: the Bland-Altman method5,6 and ICC analysis.7 In the Bland-Altman analysis, the percentage difference (bias) between CVC and peripheral concentrations was plotted against the mean of the 2 measurements (percentage difference defined as peripheral concentration subtracted from CVC concentration, divided by the mean of the CVC and peripheral concentrations, multiplied by 100%). Clinically acceptable limits of agreement were based on analysis of calculations for once-daily gentamicin dosage adjustment. Thus, ±6% was set as the clinically acceptable limits of agreement between the CVC and the peripheral SGC results because differences within this range would result in values for maximum concentration that remained within the target range. Percentage difference between the sample results rather than the absolute difference was selected for analysis because the absolute size of the differences in SGC increased as the magnitude of the SGCs increased.8,9 A 1-way random-effect model for single measures ICC also was used to evaluate agreement.7 An ICC of 
0.8 represents good agreement, and a value >0.90 is considered to indicate excellent agreement.3,10
The relationships between the difference in SGC results from different sampling sites and potential confounders were explored using parametric (gentamicin dosage) or nonparametric (number of gentamicin doses received, intravenous fluid volume, blood discard volume, and timing between samples) correlation analysis as appropriate. Differences in SGC results that were attributable to the type of CVC and the number of CVC samples drawn were assessed using the Mann-Whitney or Kruskal-Wallis test. Descriptive statistics were reported as means ± SD for normally distributed data and medians with range for data that were not normally distributed. Statistical analysis other than the Bland-Altman analysis was performed using SPSS 13.0.1 for Windows (SPSS, Chicago, IL). Statistical significance was defined a priori as P < .05.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Consent/assent to participate in the study was requested from 47 patients. Parents did not consent on 4 occasions; 5 children did not assent and; in 1 case, study requisitions were not completed correctly. A total of 45 pairs of samples were obtained from 37 patients for this analysis. Forty-2 CVC samples were drawn from implantable subcutaneous CVCs (ports), and 3 were drawn from peripherally inserted CVCs (PICCs).
Patients ranged in age from 5 months to 18 years (mean: 7.9 ± 4.4 years). Twenty-one patients were randomly assigned to have the peripheral sample taken before the CVC sample; 24 were randomly assigned to have the CVC sample taken first. Samples for 4 patients were inadvertently obtained in an order that was different from the preassigned order; the peripheral sample was obtained first rather than the CVC. These sample pairs were included in the analysis. Fourteen patients were documented to have had samples drawn simultaneously.
Estimation of Intrastudy CV of Gentamicin Assay
Fifteen patients provided CVC samples for estimation of CV analysis (10 peripheral samples drawn first; 5 CVC samples drawn first). All of these samples were analyzed using the Immuno 1. On the basis of these samples, the intrastudy CV was 5.4%. The CV for the Immuno 1 reported during the same period by our laboratory ranged from 2.8% to 4.75%. The CV for the Integrated Modular System immunoassay was <4.7% during the study period.
Extent of Agreement
The ICC for the 45 CVC:peripheral result pairs was 0.91 (95% confidence interval [CI]: 0.84 to 0.95). In the Bland-Altman analysis, the mean percentage difference (bias) between the 45 result pairs was –0.92% (95% CI: –5.06 to 3.22%). However, the limits of agreement were –27.9% (95% CI: –35.0 to –20.7%) to 26.0% (95% CI: 18.9% to 33.2%; Fig 1).
|
The ICC for the theoretical simultaneous CVC:peripheral result pairs was 0.92 (95% CI: 0.85 to 0.95). In the Bland-Altman analysis, the mean percentage difference of the theoretical simultaneous CVC:peripheral result pairs was –0.82% (95% CI: –4.44 to 2.80%) with limits of agreement of –24.4% (95% CI: –30.7 to –18.1%) to 22.8% (95% CI: 16.5% to 29.0%).
Children received gentamicin in dosages that ranged from 60 to 520 mg (mean: 257.2 ± 93.6 mg). The median number of gentamicin doses that were administered before sampling was 2 (range: 1–10). The median volume of intravenous fluid that was administered between gentamicin administration and the 3-hour sample was 282 mL (range: 117–722 mL); the mean blood discard volume was 3.78 mL ± 0.71 mL. Correlation analysis did not reveal significant associations between the dosage, number of doses received, intravenous fluid volume administered or discard volume, and percentage difference in gentamicin concentration (P > .05; data not shown).
The median time between samples was 0 minutes (range: –35 to 20 minutes) relative to the peripheral sample. A moderate negative correlation was found between the length of time between samples and the percentage difference in SGC results (Spearman 
= –.507, P < .05). Three sample pairs were drawn >15 minutes apart (20 minutes, –27 minutes, and –35 minutes). These samples were from patients with ports. The SGC results from these CVC samples differed from their matched peripheral samples by –31.6%, 23.3%, and 18.2%. When these pairs were removed from analysis, the strength of the correlation between the length of time between samples and the percentage difference in SGC results was reduced (r = –0.410, P < .05).
Because the numbers of patients with a CVC other than a port were so few, it was not possible to evaluate the effect of the type of CVC on the extent of agreement between serum gentamicin results. However, it is noteworthy that the percentage difference that was observed in the PICC samples relative to their respective peripheral samples was always positive (1.9%, 6.8%, and 17.5%).
Calculated Gentamicin Dosage Adjustments on the Basis of Both 3-Hour Concentrations
Had the CVC result rather than the peripheral result been applied clinically, 19 (42%) patients would have received clinically significantly different gentamicin dosages; that is, the dosages would have differed by 20% or more. On 7 occasions, significant dosage changes (–57%, –46%, –28%, –21%, 25%, 38%, and 50%) would have been implemented on the basis of the CVC concentration, whereas no dosage adjustment would have been recommended using the peripheral. On another 7 occasions, significant dosage changes would not have been recommended on the basis of the CVC concentration, whereas one would have been made using the peripheral (–46%, –30%, –25%, –24%, 22%, 36%, and 54%). In 5 cases, a dosage change would have been made regardless of which result (peripheral or CVC) was used. In 2 of these cases, the use of the CVC result would have led to clinically significant dosage reductions (–51% and –29%). In the remaining 3 cases, clinically significant dosage increases (26%, 32%, and 36%) would have been implemented had the CVC result been used.
Using the theoretical simultaneous SGC pairs, 17 (38%) of 45 recommendations for dosage adjustments would have differed by >20% had the CVC result rather than the peripheral result been used. On 6 occasions, significant dosage changes (–57%, –46%, –21%, 25%, 50%, and 63%) would have been implemented on the basis of the CVC result, whereas no dosage adjustment would have been made using the peripheral. On another 6 occasions, significant dosage changes would not have been recommended on the basis of the CVC concentration, whereas one would have been recommended using the peripheral (–40%, –25%, –24%, 36%, 42%, and 54%). In the remaining 5 cases, a dosage change would have been made regardless of which result (peripheral or CVC) was used, although the magnitude of the dosage change would have differed significantly (–50%, –29%, 26%, 32%, and 37%).
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
We observed a clinically significant lack of agreement between peripheral and CVC SGCs obtained 3 hours after gentamicin dose administration. Dosage adjustments ranging up to 57% would have been made had the CVC result rather than the peripheral result been applied clinically.
SGCs from peripheral sites and CVCs have been compared by other investigators. Shulman et al11 used the paired t test to compare results and Bland-Altman analysis to measure the extent of agreement. These authors suggested that several drugs, including gentamicin, could be monitored accurately via CVC when compared with peripheral blood samples. The type of CVC evaluated in this study is unknown. Details of the Bland-Altman analysis are not provided by these investigators; furthermore, this study’s method does not meet the analytical criteria required for a Bland-Altman analysis.12 Specifically, the limits of agreement were not defined a priori, and CIs of the differences between the 2 sample sites were not reported. Given these limitations, these data do not allow evaluation of the extent of agreement between the 2 sampling sites.
The ability to obtain serum methotrexate concentrations via implantable CVCs has been evaluated, although the statistical analysis that was applied to the data (paired t test analysis) was not sufficient to evaluate agreement.13 No statistical differences were found between serum methotrexate concentrations that were obtained from peripheral and CVC samples. However, in 5 of 33 cases, the difference was believed to be clinically significant. Detailed information regarding the data pairs was not provided by these investigators, so the number of data pairs, if any, for which the CVC sample was lower than the peripheral sample is unknown.
More recently, McBeth et al14 conducted a similar study with patients who were receiving tobramycin and had blood samples drawn 1 and 3 hours after dose administration. In 10 of 86 data pairs, a clinically significant dosage reduction would have been made had the CVC result been applied clinically. Of these 10, 7 were sampled from 2 patients with ports. On the basis of their results, practice changed in their institution to permit sampling from PICCs but not ports. Details regarding the overall proportions of CVC type evaluated and specific information about the pairs of data were not provided by these investigators. It also should be recognized that in addition to a lack of sample size justification, the statistical analyses that were performed by these investigators (comparison of means with t tests and correlation) are not appropriate for evaluating agreement. It is interesting to note, however, that these authors commented on a potential lack of agreement that may exist between results that are obtained from samples that are drawn from PICCs and ports. In our study, the serum gentamicin results that were obtained from PICC samples were always higher, although not always clinically significantly so, than the respective peripheral sample result.
The ICC is a single value that quantifies the reliability of 2 methods. It depends on the range of measurement, assumes that the measurement error of both methods is identical, and is not related to the size of discrepancy between the 2 methods that may be clinically permissible. The ICC for our data indicated excellent agreement. However, the ICC alone cannot describe patterns of discord that may be present among differences in the data. The Bland-Altman method aids in the determination of whether 2 methods agree sufficiently to use either method interchangeably in a clinical setting and is independent of the variability in the measured values.5,6,12 The methods being evaluated may be used interchangeably if the calculated limits of agreement between the 2 methods are not clinically important.5,6 The slight negative bias (–0.92%) that was seen in our comparison suggests that the CVC concentration was lower than the peripheral, on average, with limits of agreement of –27.9 to 26.0%. The size of difference between SGC results that were obtained from the 2 sample sites that was considered clinically acceptable was ±6% a priori. Although these limits, in retrospect, may have been overly stringent, the calculated limits of agreement in our analysis were –27.9% to 26.0%, which clearly exceed the clinically acceptable limits of agreement. The Bland-Altman plot indicates that the degree of agreement is not sufficient to permit sampling interchangeably from either the CVC or a peripheral site. This conclusion is substantiated by the clinically significant differences between gentamicin dosages that were calculated using the peripheral or CVC result.
Unlike a result such as a serum potassium concentration, an SGC result is manipulated mathematically to make gentamicin dosage adjustments. Differences between results are magnified by the calculations that are required for their interpretation. In this situation, the ICC alone is insufficient to evaluate agreement between results. The Bland-Altman method and clinical contextualization of agreement are critical to the appreciation of small, seemingly inconsequential numerical differences that may have a significant impact on patient care.
Of the possible confounding factors evaluated, only the elapsed time between peripheral and CVC samples was found to be significant. That is, the greater the time between samples, the greater the percentage difference between results. Likewise, the sampling order (peripheral first, CVC first, or both samples drawn at the same time) was found to have a significant impact on the percentage difference in sample pairs. These findings are to be expected because drug clearance will continue during the time between sampling. However, the time between samples does not account for the observed lack of agreement because the percentage differences that were observed between the 14 sample pairs that were documented to have been drawn simultaneously varied from –19% to 28%. Furthermore, extrapolation to a theoretical simultaneous SGC result did not change the Bland-Altman analysis. It must be noted that calculation of the extrapolated SGC relies on the accuracy of the 6-hour SGC result that was drawn from the CVC. The results of the current study now call into question the use of the 6-hour result.
We did not expect to observe SGC results from the CVC to be lower than those from the peripheral sample. Direct venipuncture is considered to be the ideal blood sampling technique for the purposes of therapeutic drug monitoring.15 However, it often is not feasible in children. It commonly is thought that sampling from the same lumen as dose administration may result in artificially elevated serum drug concentration because of the presence of residual drug within the catheter. Consequently, peripheral blood samples often routinely are obtained via finger or heel lancet puncture. Aminoglycoside concentrations from capillary samples have been shown to be comparable to concentrations from venous samples.16–18
Potential sources of differences between peripheral and CVC results may relate to the sampling procedures, specifically the discard volume, the force with which the sample is withdrawn into a syringe, the gauge of the needle, the depth at which the needle tip is placed within the port, and perhaps patient position. Implantable subcutaneous or internal CVCs, such as the ports that were used in our patients, now are the preferred type of CVC used in pediatric oncology. Compared with external CVCs, they have been shown to be associated with a lower rate of infection, thrombosis, and need for premature removal.19 The ports that were in use during this study contain a reservoir with a dead-space volume of up to 0.62 mL (Vital-Ports, polysulfone or titanium single chamber with polyurethane or silicone catheter; Cook Venous Access Devices, Leechburg, PA). Our current hospital policy stipulates that 3 to 5 mL of blood be discarded before a clinical sample is obtained from a subcutaneous implantable vascular device and is in accordance with the advice of the manufacturers of this device. It may be possible for a small volume of fluid to be sequestered in the reservoir of a port that is not removed with the discard volume. If the sequestered volume instead is removed with the clinical blood sample (0.5 mL), then the measured SCG result would be falsely elevated or lowered depending on the gentamicin concentration of the sequestered fluid. It is interesting that of the 2 previous studies that have attempted to evaluate the reliability of serum drug concentration results that are obtained from port samples, both have identified problems,11,14 yet ports are recommended for blood sampling by manufacturers (Cook Venous Access Devices) and commonly are used for this purpose.5 Recently, other investigators20,21 performed in vitro tests on procedures to obtain accurate drug concentration measurements from blood samples that are drawn from lines that are used to administer drug doses. These types of studies are critical to the development of sampling techniques that then must be validated clinically before their application to therapeutic drug monitoring or pharmacokinetic studies.
The original objective of this study was to minimize the number of times children experience peripheral blood sampling for the purposes of gentamicin therapeutic drug monitoring. However, our data indicate that this is not possible, at least for the purposes of determining the 3-hour SGC, when the blood sample is obtained according to our current policy. It is important to realize that there were 19 (42%) occasions whereby significantly different gentamicin dosages would have been given to patients had the CVC rather than the peripheral result been applied clinically. Subtherapeutic dosing could place a patient at risk for life-threatening infection, whereas overdosing could place a patient at risk for adverse effects such as nephrotoxicity.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
There is insufficient agreement between SGC results that are obtained from that is blood drawn from a single-lumen subcutaneous CVC port and from a peripheral sample 3 hours after the dose to permit routine sampling via the port. We therefore recommend that ports no longer be used to obtain blood samples for SGC monitoring at any time and that blood sampling via ports for other laboratory tests be reevaluated. Alternative procedures that reliably will clear ports of fluid should be investigated.
|
ACKNOWLEDGMENTS |
|---|
We acknowledge the Department of Pharmacy for generous support and funding of this project.
We thank the Nursing and Pharmacy Staff, Division of Hematology/Oncology, staff from the Department of Pediatric Laboratory Medicine, and the Department of Pharmacy, Hospital for Sick Children, for assistance with this study. In addition, we specifically thank Winnie Seto, PharmD, and Warren Walsh, BScART, for support and assistance.
|
FOOTNOTES |
|---|
Accepted Jun 22, 2006.
Address correspondence to Sabrina Boodhan, RPh, HonBSc, BScPhm, ACPR, Department of Pharmacy, Hospital for Sick Children, 555 University Ave, Toronto, Ontario, Canada M5G 1X8. E-mail: sabrina.boodhan{at}sickkids.ca
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
1. Dupuis LL, Sung L, Taylor T, et al. Tobramycin pharmacokinetics in children with febrile neutropenia undergoing stem cell transplantation: once-daily versus thrice daily administration. Pharmacotherapy. 2004;24 :564 –573[CrossRef][Web of Science][Medline]
2. DiPiro JT, Spruill WJ, Blouin RA, Pruemer JM. Aminoglycosides. In: DiPiro JT, Spruill WJ, Blouin RA, Pruemer JM, eds. Concepts in Clinical Pharmacokinetics: A Self-Instructional Course. 3rd ed. Bethesda, MD: American Society of Health-System Pharmacists; 2002:189–210
3. Donner A, Eliasziw M. Sample size requirements for reliability studies. Stat Med. 1987;6 :441 –448[Web of Science][Medline]
4. Walter S, Eliasziw M, Donner A. Sample size requirements and optimal designs for reliability studies. Stat Med. 1998;17 :101 –110[CrossRef][Web of Science][Medline]
5. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1 (8476):307–310
6. Bland JM, Altman DG. Comparing methods of measurement: why plotting differences against standard method is misleading. Lancet. 1995;346 :1085 –1087[CrossRef][Web of Science][Medline]
7. Muller R, Buttner P. A critical discussion of intraclass correlation coefficients. Stat Med. 1994;13 :2465 –2476[Web of Science][Medline]
8. Dewitte K, Fierens C, Stockl D, Thienpont LM. Application of the Bland-Altman plot for interpretation of method-comparison studies: a critical investigation of its practice.
Clin Chem. 2002;48
:799
–801
9. Altman DG, Bland JM. Commentary on quantifying agreement between two methods of measurement.
Clin Chem. 2002;48
:801
–802
10. Lee J, Ong CN. Statistical evaluation of agreement between two methods for measuring a quantitative variable. Comput Biol Med. 1989;19 :61 –70[CrossRef][Web of Science][Medline]
11. Shulman RJ, Ou C, Reed T, Gardner P. Central venous catheters versus peripheral veins for sampling blood levels of commonly used drugs.
JPEN J Parenter Enteral Nutr. 1998;22
:234
–127
12. Mantha S, Roizen MF, Fleisher La, Thisted R, Foss J. Comparing methods of clinical measurement: reporting standards for Bland and Altman analysis.
Anesth Analg. 2000;90
:593
–602
13. Cash M, Schafhauser B, Byers JF. Venipuncture versus central venous access: a comparison of methotrexate levels in pediatric leukemia patients. J Pediatr Oncol Nurs. 1999;16 :189 –193[CrossRef][Medline]
14. McBeth CL, McDonald RJ, Hodge MB. Antibiotic sampling from central venous catheters versus peripheral veins. Pediatr Nurs. 2004;30 :200 –202[Medline]
15. Jusko WJ. Guidelines for collection and analysis of pharmacokinetic data. In: Burton ME, Shaw LM, Schentag JJ, Evans WE, eds. Applied Pharmacokinetics & Pharmacodynamics. Principles of Therapeutic Drug Monitoring. 4th ed. Baltimore, MD: Lippincott Williams & Wilkins; 2006:9–26
16. Bosso JA, Mead RA, Matsen JM, Chan GM, Gooch WM 3rd. Agreement between capillary and arterial serum gentamicin concentrations in neonates. Pediatr Infect Dis. 1985;4 :142 –144[Web of Science][Medline]
17. Lewis AS, Taylor G, Williams HO, Lewis MH. Comparison of venous and capillary blood sampling for the clinical determination of tobramycin serum concentrations. Br J Clin Pharmacol. 1985;20 :597 –601[Web of Science][Medline]
18. Murphy JE, Peltier T, Anderson D, Ward ES. A comparison of venous versus capillary measurements of drug concentration. Ther Drug Monit. 1990;12 :264 –267[Web of Science][Medline]
19. McClean TW, Fisher CJ, Snively BM, Chauvenet AR. Central venous lines in children with lesser risk acute lymphoblastic leukemia: optimal type and timing of placement.
J Clin Oncol. 2005;23
:3024
–3029
20. Skolnik JM, Paccaly DA, Barrett JS, Adamson PC. Mechanisms to enhance enrollment in pediatric trials: an in vitro procedure to clear residual chemotherapeutics from indwelling catheters [abstract]. J Clin Pharmacol. 2005;45 :1085
21. Paccaly DA, Skolnik JM, Adamson PC, Barrett JS. A model-based approach for evaluating potential contamination of blood samples from a common indwelling catheter and its effects on estimated actinomycin-d pharmacokinetics [abstract]. Clin Pharmacol Ther. 2006;79 :55
PEDIATRICS (ISSN 1098-4275). ©2006 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
