Download tematic constant difference (intercept different from zero) and a

Clinical Chemistry 45, No. 2, 1999
315
Fig. 1. Difference plot with (y 2 x) as a
function of (x 1 y)/2 (left) and x 2 y
plot (right) with estimated Deming regression line (——) and diagonal y 5 x
(zzzzzz).
tematic constant difference (intercept
different from zero) and a systematic
proportional difference (slope different from 1). Therefore, the paired
t-test should not be applied uncritically to method comparison data.
Only when the graphical display
suggests that a systematic constant
difference, but not a systematic proportional difference, is involved
should this test be applied. With this
background, it appears surprising
that a clinical chemistry journal has
directly prohibited the use of regression analysis in method comparison
studies, a point of view also expressed in another journal (7, 8 ). Opposition against this practice has previously been put forward (9 ).
References
1. Bland JM, Altman DG. Statistical methods for
assessing agreement between two methods of
clinical measurement. Lancet 1986;i:307–10.
2. Pollock MA, Jefferson SG, Kane JW, Lomax K,
MacKinnon G, Winnard CB. Method comparison—a different approach. Ann Clin Biochem
1992;29:556 – 60.
3. Petersen PH, Stöckl D, Blaabjerg O, Pedersen
B, Birkemose E, Thienpont L, et al. Graphical
interpretation of analytical data from comparison of a field method with a reference method
by use of difference plots. Clin Chem 1997;43:
2039 – 46.
4. Westgard JO, Hunt MR. Use and interpretation
of common statistical tests in method comparison studies. Clin Chem 1973;19:49 –57.
5. Westgard JO, deVos D, Hunt MR, Quam EF,
Carey RN, Garber CC. Concepts and practices
in the evaluation of clinical chemistry methods.
Part III. Statistics. Am J Med Technol 1978;44:
552–70.
6. Linnet K. Performance of Deming regression
analysis in case of misspecified analytical error
ratio in method comparison studies. Clin Chem
1998;44:1024 –31.
7. Hollis S. Analysis of method comparison studies [Editorial]. Ann Clin Biochem 1996;33:1– 4.
8. Hollis S. Analysis of method comparison studies. JIFCC 1997;9:8 –12.
9. Stöckl D. Beyond the myths of difference plots
[Letter]. Ann Clin Biochem 1996;36:575–7.
Kristian Linnet
Laboratory of Clinical Biochemistry
Psychiatric University Hospital
DK-8240 Risskov, Denmark
E-mail [email protected]
Interlaboratory Variability for Total
Homocysteine Analysis in Plasma
To the Editor:
Total plasma homocysteine consists
of free homocysteine and homocysteine that is complexed with itself or
with other amino acids or proteins.
Free homocysteine has been measured previously as part of a biochemical screen for inherited metabolic disorders. More recently,
increased total plasma homocysteine
has been suggested as an independent risk factor for atherosclerotic
coronary artery disease [reviewed in
Ref. (1 )]. In addition, increased total
homocysteine is associated with a
poor prognosis in patients with angiographically demonstrated coronary artery disease (2 ). These studies
have prompted clinicians to include
total homocysteine analysis as part of
the risk assessment profile of patients with premature coronary artery disease. However, it has yet to
be shown that reducing total plasma
homocysteine concentration leads to
a decrease in cardiovascular risk, although vitamin supplementation
may effectively lower or normalize
circulating homocysteine concentrations (3 ).
From a laboratory standpoint,
problems exist in homocysteine analysis. External quality-assurance programs for total homocysteine (proficiency testing) are not available at
present, and interlaboratory correlations of total homocysteine measurements have not been evaluated formally. We have contacted many of
the reference laboratories in the US
that offer this assay and have found
that the reference ranges vary considerably between laboratories. Although most laboratories offer reference intervals based on in-house
studies, a few base reference intervals on a review of the literature.
Other laboratories offer a “target
range” based on prospective studies,
which correlate total homocysteine
concentrations with risk for cardiovascular disease or with mortality (2).
To evaluate laboratory variability of
total homocysteine analysis, we sent
five samples of frozen plasma in
EDTA tubes, all drawn and pooled
from the same fasting subject, to five
different reference laboratories. The
analytical variation of the results produced a wide range of possible risk
estimates for that subject (Table 1).
We evaluated the extent of variability attributable to inherent test
imprecision as opposed to bias between laboratories. The variance of
the reported homocysteine test results is composed of the sum of variance within the laboratories and
variance between laboratories. To estimate variance within the laboratories, we obtained the coefficients of
316
Letters
Table 1. Test results and risk assessment for total homocysteine in a specimen that has been aliquoted and sent to five
reference laboratories.
Reference
laboratory
Test result,
mmol/L
Reference or target
range, mmol/L
Test result/median
of reference interval
Study A
mortality ratioa
Study B
relative riskb
A
B
C
D
E
13.5
11.8
11.0
10.1
8.7
4.9–14.6c
8.0–12.0c
4–17c
6.4–13.7c
,7.2d
1.4
1.2
1.0
1.0
1.1
0.9
0.8
0.75
0.7
3.0
1.5
1.0
1.9
0.8
a
Relative risk of mortality in patients with angiographically confirmed coronary artery disease during a follow-up period of ;4 years. Mortality ratios were estimated
from Fig. 2 in Nygard et al. (2).
b
Relative risk of vascular disease at the time of homocysteine testing. The relative risk was derived from Fig. 4 in Graham et al. (4).
c
Reference range.
d
Target range described by Nygard et al. (2).
variation (CVs) for homocysteine
testing from the five reference laboratories. At homocysteine concentrations ranging from 8 to 15 mmol/L,
the participating laboratories reported CVs of 4.9 –11%. The variances of homocysteine test results
within each reference laboratory
were estimated to be between 0.44
and 1.46 (mmol/L)2, with an average
variance of 0.96 (mmol/L)2. By subtracting the average variance within
laboratories from the total variance
of 3.25 (mmol/L)2, we found the variance between laboratories to be 2.29
(mmol/L)2. Thus, bias between the
laboratories contributed more than
random error to the observed variability.
Multiple factors could account for
the interlaboratory variability, including different methodologies (most common methods are based on fluorescence polarization immunoassays and
HPLC), different approaches to test
calibration, lack of availability of reference materials for the various forms of
homocysteine, or varying efficiencies
for the dissociation of homocysteinecontaining complexes. The observed
variability makes it difficult to interpret the laboratory result in terms of
patient risk for development or progression of cardiovascular disease.
References
1. Welch GN, Loscalzo J. Homocysteine and
atherothrombosis. N Engl J Med 1998;338:
1042–50.
2. Nygard O, Nordrehaug JE, Refsum H, Ueland
PM, Farstad M, Vollset SE. Plasma homocysteine levels and mortality in patients with coronary artery disease. N Engl J Med 1997;337:
230 – 6.
3. Malinow MR, Duell PB, Hess DL, Anderson PH,
Kruger WD, Phillipson BE, et al. Reduction of
plasma homocysteine levels by breakfast cereal fortified with folic acid in patients with
coronary artery disease. N Engl J Med 1998;
338:1009 –15.
4. Graham IM, Daly LE, Refsum HM, Robinson K,
Brattstrom LE, Ueland PM, et al. Plasma homocysteine as a risk factor for vascular disease.
The European Concerted Action project. JAMA
1997;277:1775– 81.
Stephen C. Eliason1
Detlef Ritter1,2*
Hyung D. Chung1,2
Michael Creer1
1
Department of Pathology
Saint Louis University
School of Medicine
St. Louis, MO 63104
2
Pathology and
Laboratory Medical Service
John Cochran Veterans Affairs
Medical Center
St. Louis, MO 63106
*Address correspondence to this author at: Clinical Laboratories, 3635 Vista
at Grand, St. Louis, MO 63110-0250. Fax
314-289-7073; e-mail [email protected].
Urinary Free Cortisol Is Not Affected
by Short-term Water Diuresis
To the Editor:
Urinary free cortisol (UFC) has been
shown to be a reliable measure of
adrenocortical secretion, and it is
generally accepted as being an index
of the free fraction in the plasma.
However, the interpretation of UFC
results reportedly can be compromised when the urine volume in
patients is considerably increased because a high fluid intake (5 L/day)
increases UFC in healthy subjects (1 )
and UFC was reported to be closely
related with the changes in urine
volume in women (2 ). The aim of
this work was to determine whether
short-term changes in urine volume
influence UFC.
I studied 15 volunteers (6 women
and 9 men) with normal body mass
indexes (23.9 6 1.9 kg/m2) and ages
from 23 to 52 years. Informed consent was obtained from all volunteers and our institution’s responsible committee. Subjects were asked
to empty their bladders at 0900 and
to collect urine samples at 30-min
intervals. At 1000, 10 of the volunteers (6 women and 4 men) were
asked to drink 1 L of water within 5
min. The volume of each urine sample was measured and adjusted to a
final volume of 300 mL. Urine samples with a volume .300 mL were
not diluted. Creatinine and cortisol
were measured by a colorimetric
method (3 ) or by RIA (4 ). The results
are presented as mean 6 SE. Statistical comparisons were made using
the Mann–Whitney U-test for unpaired data.
As expected, the urine volume was
significantly greater during water diuresis (P ,0.002, 60 –180 min after
water ingestion) than during the control period (0 –180 min, no water ingested). In contrast, neither urinary
excretion of creatinine nor of UFC
changed significantly (Fig. 1). A similar lack of change of UFC was found
when cortisol concentrations in 40 of
the samples were measured again by