| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
General Clinical Chemistry |
Departments of
1
Pathology and
2
Biostatistics, Virginia Commonwealth University, Richmond, VA 23298.
3
American Medical Laboratories, Inc., P.O. Box 10841,
Chantilly, VA 20153-0841.
a Address correspondence to this author at: Medical College of Virginia Hospitals of Virginia Commonwealth University, Richmond, VA 23298-0286. Fax (804) 828-0353; e-mail millerg{at}hsc.vcu.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Prolonged contact of serum with the clot can cause preanalytical variation. The optimum time interval between sample collection and separation of serum from the clot should be long enough to allow complete clot formation but be shorter than the time in which a significant change in test result is induced by serum-clot contact. The minimum clotting time suggested by the Tietz Textbook of Clinical Chemistry (1) is 20–30 min. During a prolonged contact time between serum and clot, both biological activity of the cells and trans-membrane diffusion can change the concentrations of certain analytes in the serum. NCCLS Procedures for the Handling and Processing of Blood Specimens (2) recommends that serum or plasma should be physically separated from contact with cells as soon as possible, unless conclusive evidence indicates that longer contact times do not contribute to result inaccuracy. A maximum limit of 2 h from the time of collection to the time of separation was also recommended.
Each individual analyte has a different tolerance to a delay in separating serum from clot. Many analytes are stable for much longer than 2 h. In hospitals and outpatient clinics, transportation of specimens from a phlebotomy site to a laboratory sometimes takes longer than 2 h. If overly stringent transportation requirements are set for all tests, many acceptable specimens would be rejected unnecessarily. Ideally, a specific allowable transportation time should be applied for each analyte in a specimen. In practice, analytes are usually grouped into time blocks in which serum-clot contact cause no changes in analyte concentrations. Generally, specimens arrive in the laboratory at the Medical College of Virginia Hospitals <6 h after collection. Therefore, knowing the test stability within 6 h was critical to determine the cutoff times for acceptable specimens.
Information about test stability after prolonged contact of serum with
clot is available in the literature for 41 chemistry tests
(3)(4)(5)(6)(7). All studies except Chu et al. (5)
reported 24-h stability at room temperature (summarized in Table 1
). Some reports also included data at different times and
temperatures.
|
Laessig et al. (3) reported the serum-clot contact effect on 25 tests. Whole-blood samples were incubated at room temperature for 1, 2, 4, 8, 24, and 48 h before serum-clot separation. The sera were assayed along with quality-control (QC)1 material immediately after separation from clot. The results of the 1-h sera and QC material were considered as target results and the percentage change in results at 2–48 h was calculated. The significance of the change was determined by comparing the percentage of change in results of the sera to that of the QC material. They reported that significant changes occurred for glucose, lactate dehydrogenase (LD), and potassium at 2 hand for iron and chloride at 8 h. The other tests were stable up to 48 h.
Ono et al. (4) studied the effect of serum-clot contact time at 4, 23, and 30 °C for 25 analytes. The contact time effect was tested at 2, 4, 6, 8, 24, and 48 h. All the sera were stored at -20 °C after separation from clots and thawed and assayed in the same run. The results of the 2–48 h samples were compared with that of the zero hour samples. Statistically significant changes (by Student's t-test) at 30 °C were observed for alanine aminotransferase (ALT, 6 h), aspartate aminotransferase (AST, 6 h), LD (8 h), glucose (2 h), sodium (6 h), potassium (4 h), calcium (48 h), and inorganic phosphorus (8 h). The stability of these analytes was sensitive to temperature.
Chu et al. (5) reported the changes in results after 3 days of serum-clot contact. The results from the stored samples were compared with those from sera separated from cells within 3 h of collection. The changes were classified as significant if the percentage of change was greater than the analytical variability of the method.
In the study by Rehak and Chiang (6), whole blood specimens were stored at 3, 10, 15, 22, 25, 30, and 38 °C for 24 h before serum-clot separation. The results of these specimens were compared with the results from the fresh sera. They concluded that significant changes occurred for creatinine, glucose, inorganic phosphorus, potassium, AST, and ALT and that all of these changes were sensitive to temperature.
Heins et al. (7) reported the effect of storage temperature and time on 22 analytes. The whole-blood specimens were stored at 9 °C and room temperature for 1–7 days. The results from the stored specimens were compared with those from the corresponding fresh sera. The changes were classified as significant if they exceeded the maximum allowable inaccuracy according to the recommendation of the German Federal Medical Council.
Because of a variety of experimental designs, it is difficult to interpret information on specimen stability from the literature. First, only two of the previous reports (3)(4) presented the test results with contact times <24 h. Second, specimens from healthy donors were used, which had concentrations of some analytes so low that precise measurements were hard to achieve. Third, the control samples and the test samples were assayed in different analytical runs in all of these studies except Ono et al. (4). Thus, run-to-run imprecision was introduced into the test results. Fourth, statistical significance of the change in test results was evaluated by analytical variations of QC results or the Student's t-test. However, a statistically significant change may or may not have any impact on the interpretation of the test result. The clinical relevance of the change was either not addressed or determined by the opinions of the authors.
In this study, we investigated the effect of serum-clot contact time on the laboratory results of 63 analytes. The specimen donors were selected to obtain analyte concentrations high enough to achieve precise measurement. The clinical impact of the change was evaluated with consideration of both analytical and biological variation of each test. The results of this study provided justification to make optimum decisions regarding specimen stability before reaching the laboratory.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. Each blood specimen was collected into four red-topped Vacutainer
Tubes (Becton Dickinson), which were randomly labeled to eliminate the
effects of the drawing sequence. The first tube was allowed to clot for
30 min at room temperature before the serum was separated from the
clot. The rest of the tubes were incubated in a circulating water bath
at 32 °C, to mimic a warm transportation condition, for 3, 6, and
24 h before serum-clot separation. To study the behavior of
potassium at different incubation temperatures, two groups of specimens
were incubated at room temperature or at 32 °C for 3, 6, and 24
h before serum-clot separation. All specimens were centrifuged for 10
min at 1500g to separate the serum from the clot. The sera
from all serum-clot contact times were stored at 4 °C until assay.
To eliminate the run-to-run analytical imprecision, the four specimens
from each donor were assayed in the same run. Specimens from different
donors for an analyte may have been run at different times. The
specimens used for light-sensitive analytes, carotene, vitamin A, and
bilirubin, were protected from light during the process.
|
The effect of specimen transportation conditions on the concentration of LD was evaluated using blood specimens collected from 12 donors. Three serum separation tubes were drawn on each individual. Two tubes were filled completely and one was half-filled. One full tube remained in the laboratory as a control sample. The other full tube and the half-filled tube were hand-carried to various pneumatic tube stations in our hospital, from which they were sent to our laboratory. The pneumatic tube system (PEVCO Systems International) used 6-inch carriers, and transport distances varied between 500 and 1750 feet. The average transport speed was ~5.2 m/s. When the samples arrived in the laboratory, the control sample and the two samples transported through the tube system were centrifuged and analyzed for LD in the same run. The percentage difference was calculated as (LDt - LDc) x 100/LDc, where LDt was the LD result for the sample transported through the tube system, and LDc was the LD result for the control sample that remained in the laboratory.
methods for specimen analysis
Albumin, alkaline phosphatase, ALT, amylase, AST, bicarbonate,
calcium, creatine kinase (CK), chloride, creatinine,
-glutamyltransferase, glucose, iron, LD, lipase, magnesium,
inorganic phosphorus, potassium, sodium, total protein, urea, and uric
acid were measured with a Vitros 700 (Johnson & Johnson).
Apolipoprotein A (apo A), apolipoprotein B (apo B), haptoglobin, IgA,
IgG, IgM, prealbumin, and transferrin were measured with the Array 360
system (Beckman). Cholesterol, triglycerides, and the cholesterol
component in HDL and LDL were measured on Hitachi 911. A
heparin-manganese method was used for HDL pretreatment and a direct-LDL
kit from Sigma was used for LDL separation. 
-Fetoprotein, ferritin,
ß-human chorionic gonadotropin, IgE, prolactin, and
thyroid-stimulating hormone were measured with IMX (Abbott
Laboratories). Cortisol, thyroxine uptake, and thyroxine were measured
with TDX (Abbott Laboratories). Folate and vitamin B12 were measured
with RIA (Diagnostic Products). CK-MB was measured with the Opus Plus
(Behring Diagnostics). All of the above tests were performed in the
Department of Pathology clinical chemistry laboratory at the Medical
College of Virginia Hospitals.
Estriol, estradiol, follicle-stimulating hormone (FSH), and luteinizing hormone were measured with Immulite (Diagnostic Products) in the Department of Obstetrics and Gynecology fertility laboratory at the Medical College of Virginia Hospitals.
The following tests were performed at the American Medical Laboratories (Chantilly, VA). Aldosterone (Diagnostic Products), C-peptide (INCSTAR), and vitamin D (Nichols Diagnostics) were measured by RIA. Thyroxin-binding globulin was measured with Immulite (Diagnostic Products). Testosterone was measured with ACS:180 (Chiron). Fructosamine was measured with a colorimetric method (Roche Diagnostics). Ceruloplasmin was measured with the Behring Nephelometer II (Behring Diagnostics). Vitamin C was measured with a colorimetric method. Carotene, vitamin A, and vitamin E were measured with HPLC methods.
statistical analysis
To estimate the average effect of prolonged serum-clot contact,
the mean of the results from all specimens for each analyte at each
serum-clot contact time was used to evaluate the change. The approach
to the analysis was derived from a quality-control paradigm. The
X chart, originally proposed by Shewhart (8),
tracks the mean of a process and, considering the noise in the system,
determines when the process is out of control. In our case, the process
was the measurement of analyte content after different contact times
between serum and cells. X is the mean result of the 30-min
samples. A variety of methods have been proposed for determining
process limits, such as 2 or 3 SD from the mean or probability limits
using the gaussian distribution (9). Both approaches require
an estimate of the SD, which can be estimated from the process or
preferably, known from some external source. In this experimental
design, each measurement under the same treatment condition was made
once; thus an estimate of analytical variance could not be made from
the data. Therefore, we used the well-established analytical SD from
quality-control data for each method. Thus, the X chart
with probability limits is:
 | (1) |
is the SD; and N is the sample size. The analytical CV of
each test was estimated by the annual cumulative CV of quality-control
results at an analyte value close to the mean of the 30-min samples.
The goal of this statistical analysis was to determine which analytes
were altered enough by serum-clot contact to exceed the total
measurement error for a method. Thus, the within-run SD was not used as
an estimate of analytical imprecision.
It was necessary to assume that the analytical CV was constant over the
concentration ranges tested. To ensure validity of this assumption, we
selected donors to give increased concentrations of those analytes that
would have had changes in analytical CV at low reference range
concentrations. Therefore, the imprecision contributed by the
analytical variation at the 30-min value can be expressed as
CVa x X, which can be placed in Eq. 1
.
The 95% confidence interval based on the measurement variation defines
the analytically acceptable process limits as:
 | (2) |
Because of the existence of intraindividual biological variation, a
mean result at a contact time that exceeds the analytical
imprecision-based process limits defined by Eq. 2
may not be clinically
relevant. We combined analytical and intraindividual biological
variations of each test as:
 | (3) |
provided clinically acceptable
process limits as:
 | (4) |
In this study, the experimental design did not allow estimation of biological variation from the data because the specimens from each patient were collected via a single phlebotomy and then divided for different incubation times. The intraindividual biological CV of each analyte (CVb) was obtained from Fraser (10)(11), in which he summarized the intraindividual biological variation of several tests reported in the literature. In those reports, the data were obtained by collecting specimens from cohorts of subjects for certain time spans and analyzing them in duplicate. The analytical, within-subject, and between-subject components of variation were derived by nested analysis of variance after exclusion of outliers. We took the median intraindividual biological variations of healthy individuals from the literature as CVb after excluding the data collected in a time span of <5 days. For FSH and prolactin, where the biological variation is much higher in females than in males, the higher female intraindividual biological variations were used. When no data about intraindividual biological variation could be found in the literature, one-quarter of the reference range was used as an estimation of the intraindividual biological SD (12). Ricos et al. (13) reported a similar biological variation of total and conjugated bilirubin. Therefore, the median biological variation of total bilirubin was adopted as the CVb for total bilirubin, conjugated bilirubin, and unconjugated bilirubin.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. The tests that were suitable for analysis after
24 h of serum-clot contact, based on clinically acceptable limits,
were -fetoprotein, aldosterone, ALT, alkaline phosphatase, amylase,
apo A, apo B, AST, bilirubin, calcium, carotene, ceruloplasmin,
chloride, cholesterol, creatinine, CK, CK-MB, cortisol, estradiol,
estriol, ferritin, folate, fructosamine, FSH, -glutamyltransferase,
human chorionic gonadotropin, haptoglobin, IgA, IgE, IgG, IgM, LD,
luteinizing hormone, lipase, magnesium, pre-albumin, prolactin, sodium,
thyroxine, thyroxin-binding globulin, testosterone, transferrin,
thyroid-stimulating hormone, thyroxine uptake, urea, uric acid, vitamin
A, vitamin B12, vitamin C, and vitamin E. Tests that were
suitable for analysis at a 6-h incubation but not suitable for analysis
at a 24-h incubation were bicarbonate, chloride, C-peptide, iron, HDL,
LDL, and total protein. Albumin and inorganic phosphorus were suitable
for analysis at a 3-h incubation but not suitable for analysis at
6 h. Glucose and potassium had clinically relevant changes at
3 h.
Time-course plots for selected analytes that illustrate various
stability situations are presented in Fig. 1
. Glucose, potassium, and inorganic phosphorus were the most
unstable tests. Note that potassium changes were temperature-dependent.
Calcium was a stable analyte. Albumin exceeded acceptability criteria
over time. Changes in LD exceeded the process limit determined by
analytical imprecision but were within the clinically acceptable limits
determined by both analytical imprecision and intraindividual
biological variations.
|
The differences in LD results of the samples before and after pneumatic
tube transportation are presented as percentages in Fig. 2
. LD value changes ranged from -1.0% to 13.9% for the samples
in full Vacutainer Tubes and from 8.6% to 30.7% for the samples in
half-filled Vacutainer Tubes.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The statistical evaluation developed here was based on a QC paradigm. The critical question was if the result from a specimen was altered by serum-clot contact enough to exceed the variability expected for physiological processes. A QC model performs exactly that evaluation for each sequential data point in a process by evaluating that point vs fixed limits based on expected variability in the process. In this case, each serum-clot contact time represented a unique preanalytical condition that could change the analyte stability independently of any physiologic condition in the patient. To establish clinically meaningful evaluation limits within the technical limitations of the measurement process, we combined literature estimates of biological variation, which do not include an analytical component, with the known imprecision for the measurement process.
Glucose concentrations decreased with increasing serum-clot contact time. It is well-documented that glycolysis of cells consumes glucose and causes a decrease in glucose concentration in blood during transportation and that the rate of decrease is sensitive to temperature. Serum glucose decreases 100–200 mg · L · h at 37 °C (14) and 50–100 mg · L · h at room temperature (15). The rate of glycolysis is markedly increased in the presence of leukocytosis (15). Glycolysis can be inhibited and glucose concentration can be stabilized by using sample collection tubes with sodium fluoride as an additive (16).
Inorganic phosphorus increased after three hours serum-clot contact. The concentration of organic phosphates is about seven times higher in erythrocytes than that in serum (17). These organic phosphates are susceptible to hydrolysis to produce inorganic phosphorus, which leaks from the cells and consequently increases the concentration in serum (18).
The effect of serum-clot contact on serum potassium depended on
incubation temperature (Fig. 1
). At room temperature, the serum
potassium values increased with contact time; the change became
clinically significant by 3 h. At 32 °C, the potassium values
fluctuated with an initial decrease followed by a large increase after
6 h. The change of potassium is the net effect of glycolysis,
which moves potassium into cells, and passive diffusion, which allows
potassium to diffuse out of the cells (19)(20).
At 32 °C, glycolysis was dominant initially; thus serum potassium
concentrations decreased. As time went on, the glucose in serum was
depleted, and passive diffusion of potassium from cells became
dominant, producing an increase in potassium after longer serum-clot
contact. At room temperature, glycolysis was slower; thus passive
diffusion was dominant, allowing potassium to increase.
A small increase in concentration was observed for albumin at 6 h
(Fig. 1
) and total protein at 24 h, which was outside the
clinically acceptable limits. These limits are very small because of
the small analytical and biological variations. Although these analytes
exceeded the clinically acceptable limits defined by Eq. 4
, such a
small increase generally will not cause changes in clinical treatment
or diagnosis. Because the small increases occurred with both albumin
and total protein, they were possibly due to water shifting into the
cells.
Previous studies are contradictory regarding the stability of LD during
serum clot contact. Rehak and Chiang (6) found no
significant change of LD up to 24 h, whereas Laessig et al.
reported that LD concentration increased after 2 h of contact
between serum and cells (3). Fig. 1
shows an increasing
trend in LD concentration over time, with all values remaining within
the clinically acceptable limits. We observed substantial increases for
samples transported through the pneumatic tube (Fig. 2
). The effect of
transportation on half-filled Vacutainer Tubes was more pronounced than
the full Vacutainer Tubes, presumably because of greater agitation of
cells. We concluded that prolonged contact between serum and cells
without mechanical agitation did not cause a substantial increase in LD
results. Agitation of the sample before serum/cell separation was the
major contributor to the preanalytical variation of LD results. In this
study, except for the samples in the transportation study, blood
specimens were handled with care to avoid any agitation.
In conclusion, only a few routine analytes required stringent control of delivery times before serum-clot separation. Most routine tests can tolerate fairly long delays in transportation without changes in analyte content. Samples for glucose, potassium, and inorganic phosphorus should be spun and separated from the clot within 3 h after collection. Samples for albumin, bicarbonate, chloride, C-protein, HDL, iron, LDL, and total protein should be processed within 6 h. The remaining analytes evaluated were stable for 24 h.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  B. S. Drammeh, R. L. Schleicher, C. M. Pfeiffer, R. B. Jain, M. Zhang, and P. H. Nguyen Effects of Delayed Sample Processing and Freezing on Serum Concentrations of Selected Nutritional Indicators Clin. Chem., November 1, 2008; 54(11): 1883 - 1891. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. Kumar and B. S. Karon Comparison of Measured and Calculated Bicarbonate Values Clin. Chem., September 1, 2008; 54(9): 1586 - 1587. [Full Text] [PDF]
|
|
|
|
 |
|
 M. Tanner, N. Kent, B. Smith, S. Fletcher, and M. Lewer Stability of common biochemical analytes in serum gel tubes subjected to various storage temperatures and times pre-centrifugation Ann Clin Biochem, July 1, 2008; 45(4): 375 - 379. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W S Waring, L E Evans, and C T Kirkpatrick Glycolysis inhibitors negatively bias blood glucose measurements: potential impact on the reported prevalence of diabetes mellitus J. Clin. Pathol., July 1, 2007; 60(7): 820 - 923. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. L. Myers, W. G. Miller, J. Coresh, J. Fleming, N. Greenberg, T. Greene, T. Hostetter, A. S. Levey, M. Panteghini, M. Welch, et al. Recommendations for Improving Serum Creatinine Measurement: A Report from the Laboratory Working Group of the National Kidney Disease Education Program Clin. Chem., January 1, 2006; 52(1): 5 - 18. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. van Eijsden, M. F. van der Wal, G. Hornstra, and G. J. Bonsel Can Whole-Blood Samples Be Stored over 24 Hours without Compromising Stability of C-Reactive Protein, Retinol, Ferritin, Folic Acid, and Fatty Acids in Epidemiologic Research? Clin. Chem., January 1, 2005; 51(1): 230 - 232. [Full Text] [PDF]
|
|
|
|
|
|
E. J. Giltay, J. M. Geleijnse, E. G. Schouten, M. B. Katan, and D. Kromhout High Stability of Markers of Cardiovascular Risk in Blood Samples Clin. Chem., April 1, 2003; 49(4): 652 - 655. [Full Text] [PDF]
|
|
|
|
|
|
B. L. Boyanton Jr and K. E. Blick Stability Studies of Twenty-Four Analytes in Human Plasma and Serum Clin. Chem., December 1, 2002; 48(12): 2242 - 2247. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
24 h serum-clot
contact.1
) indicates incubation at 32 °C; (
) indicates incubation at
24 °C; ([- - -]); indicates acceptable limits based on analytical
variation; ([—–]) indicates acceptable limits based on combined
analytical imprecision and clinical variation.
) indicates blood specimens transported in full collection tubes,
and (
) indicates blood specimens transported in half-filled
collection tubes.