Research Article
Annals of Clinical Biochemistry
2016, Vol. 53(4) 452–458
! The Author(s) 2015
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0004563215602028
acb.sagepub.com
The influence of storage time and temperature on the
measurement of serum, plasma and urine osmolality
Karla Bezuidenhout1, Megan A Rensburg2, Careen L Hudson3, Younus Essack4 and
M Razeen Davids1
Abstract
Background: Many clinical laboratories require that specimens for serum and urine osmolality determination be
processed within 3 h of sampling or need to arrive at the laboratory on ice. This protocol is based on the World
Health Organization report on sample storage and stability, but the recommendation lacks good supporting data. We
studied the effect of storage temperature and time on osmolality measurements.
Methods: Blood and urine samples were obtained from 16 patients and 25 healthy volunteers. Baseline serum, plasma
and urine osmolality measurements were performed within 30 min. Measurements were then made at 3, 6, 12, 24 and
36 h on samples stored at 4–8 C and room temperature. We compared baseline values with subsequent measurements
and used difference plots to illustrate changes in osmolality.
Results: At 4–8 C, serum and plasma osmolality were stable for up to 36 h. At room temperature, serum and plasma
osmolality were very stable for up to 12 h. At 24 and 36 h, changes from baseline osmolality were statistically significant
and exceeded the total allowable error of 1.5% but not the reference change value of 4.1%. Urine osmolality was
extremely stable at room temperature with a mean change of less than 1 mosmol/kg at 36 h.
Conclusions: Serum and plasma samples can be stored at room temperature for up to 36 h before measuring osmolality. Cooling samples to 4–8 C may be useful when delays in measurement beyond 12 h are anticipated. Urine osmolality
is extremely stable for up to 36 h at room temperature.
Keywords
Analytes, laboratory methods
Accepted: 31st July 2015
Introduction
Serum and urine osmolality are frequently measured in
clinical practice to assist with the diagnosis and management of polyuria, hypernatraemia and hyponatraemia. It is
also applied in the calculation of serum or urine osmolal
gaps and can be used in the estimation of urinary ammonium excretion.1 Specimens for serum and urine osmolality
determination are often required to be processed within 3 h
of sampling or must be transported and delivered to the
laboratory on ice. This protocol has been in place in many
1
Division of Nephrology, Department of Medicine, Stellenbosch
University and Tygerberg Hospital, Cape Town, South Africa
2
Division of Chemical Pathology, Stellenbosch University and National
Health Laboratory Service, Cape Town, South Africa
3
National Health Laboratory Service, Greenpoint Complex, Cape Town,
South Africa
4
PathCare Laboratories, Cape Town, South Africa
Corresponding author:
Razeen Davids, Division of Nephrology at the Faculty of Medicine and
Health Sciences of Stellenbosch University, Francie van Zijl Avenue,
Tygerberg 7505, Cape Town, South Africa.
Email: mrd@sun.ac.za
Bezuidenhout et al.
laboratories since 2009 and is based on the World Health
Organization (WHO) report on sample storage and stability.2 Implementation of this protocol results in the rejection
of many samples with obvious cost implications and a
negative impact on effective clinical decision making.
According to the WHO report, serum/plasma osmolality is stable for 3 h at 20 to 25 C, for 1 day at 4–8 C
and for three months at 20 C. It also states that urine
osmolality is stable for 3 h at 20–25 C, for seven days at
4–8 C and for more than three months at 20 C. The
recommendations are based on a study by Zhang et al.3
on the effects of prolonged serum-clot contact time on
the determination of commonly measured analytes.
They found that potassium and glucose concentrations
had clinically relevant changes at 3 h and that phosphate
concentration was elevated by 6 h. Sodium, calcium,
urea and chloride concentrations were stable over 24 h.
The WHO recommendations for plasma and serum
osmolality stability are therefore based on the instability
of analytes that contribute to osmolality, rather than on
stability studies which measured osmolality directly.3
Boyanton and Blick4 demonstrated a decrease in glucose and HCO3 concentration and an increase in lactate and CO2 concentrations when serum and plasma
separation is delayed. These changes are attributed to
on-going glycolysis by red blood cells in vitro. The formation of lactate during glycolysis yields Hþ ions that
are buffered by HCO3. HCO3 concentration consequently decreases and CO2 concentration increases over
time. Changes in the concentrations of HCO3 and
CO2 were shown to influence plasma osmolality by
Boning and Maassen.5 It is therefore reasonable to
expect osmolality to change over time when separation
from cellular components is delayed.
Osmolality stability studies where serum or plasma is
separated from cellular components almost immediately after sampling and then stored have limited relevance for clinical practice where there are often delays
in getting samples to the laboratory and where plasma
or serum has not yet been separated.6,7 The direct effect
on osmolality of delaying serum separation was examined by Redetzki et al.8 who showed that the osmolality
of serum samples stored for 1–4 h at room temperature
before separation was, on average, 2.8 mosmol/kg
higher than that of baseline samples which were
cooled immediately and measured within 20 min of
sampling. They demonstrated that this difference was
attributable to the formation of lactate. In another set
of samples, they also demonstrated an average increase
in serum osmolality of 3.5 mosmol/kg over 5 h at room
temperature and 1.5 mosmol/kg at 0 C. The changes
from baseline correlated very well with the changes in
lactate concentration. The effect of storage temperature
and delayed serum separation was, however, not studied beyond 5 h.
453
Most laboratories consider plasma and serum osmolality to be identical.6,9 Serum is considered to be more
stable than plasma for the measurement of many analytes, when there is prolonged contact with cells. In
Boyanton and Blick’s stability study,4 the changes
seen over time in lactate, glucose, potassium, pH,
CO2 and HCO3 were more pronounced in plasma
than in serum. This suggests that clotted serum samples
might be the preferred medium for measuring osmolality when separation is delayed.
To the best of our knowledge, the only stability data
available for urine osmolality is a white paper from an
osmometer manufacturing company, which reported
good stability over 24 h.10 The WHO report refers to
Cooper et al.,11 but this study has no data on the stability of urinary analytes or osmolality. It is generally
advised that urine specimens be refrigerated or chemically treated to prevent bacterial growth if samples are
to be stored before analysis. Bacteria may cause conversion of glucose into acids and alcohols, and urea
into ammonia. They also utilize glucose and ketones,
causing a decrease in their concentration. For determining urine osmolality, refrigeration, but not the addition
of preservatives, is advised.12
In summary, guidelines for the handling of samples
for osmolality lack good supporting data. The stability
of chemical analytes contributing to serum and plasma
osmolality has been studied well, but only one study
directly investigated the effect of temperature and
delayed serum separation on the determination of
osmolality, and then only over a 5-h period. No studies
have directly examined the stability of urine and plasma
osmolality. In view of this knowledge gap, we investigated the effect of delayed serum and plasma separation, and cooling, on osmolality measurements. We
also compared serum with plasma osmolality to determine the most stable medium. Last, we investigated the
effect of storage time and temperature on the stability
of urine osmolality.
Materials and methods
Assessment of the sample rejection rate based on
current laboratory protocols
The sample rejection rate for serum and urine osmolality was determined by a search of the Tygerberg
Hospital NHLS database for the period 1 January to
31 December 2012.
Participants and samples for osmolality
measurements
We obtained a total of 30 blood samples and 40 urine
samples over a period of six weeks. These were
454
Annals of Clinical Biochemistry 53(4)
provided by 16 patients and 25 healthy volunteers. The
patients were selected on the basis of having conditions
such as hyperglycaemia, hyponatraemia and renal failure, which are known to affect osmolality. To ensure a
good spread of urine osmolality, some volunteers were
sampled after an overnight fast and others after taking
a water load.
and 0.36% for urine. Each measurement was performed in duplicate. When measurements differed by
more than 5 mosmol/kg, a third measurement was
made. The average of the two closest values was
taken to be the true value.
Specimen collection and handling
Repeated-measures analysis of variance was used to
compare baseline values with subsequent measurements
at each time point, to compare serum and plasma, and
cooled and room temperature specimens. We used
Bland–Altman difference plots to illustrate the deviation from baseline osmolality at different time
points. The total allowable error for serum osmolality
of 1.5% as specified in the Westgard Biological
Variation Database, and the reference change value of
4.1% was used to determine serum and plasma sample
stability at the various time points.13 The reference
change value (RCV) indicates whether a true clinical
change in the value has been observed. The RCV was
calculated using a bidirectional Z-score of 1.96, a
within-subject biological variation of 1.3% and an analytical imprecision of 0.7%.14 The biological variation
in urine osmolality is large, hence the large total allowable error of 39.5% quoted in the Westgard Database13
and the RCV of 88%. This is too large to be clinically
useful and, in the absence of other guidelines, we considered a deviation from baseline values of >5% to be
of clinical significance. A P-value of <0.05 was used to
indicate statistical significance.
After obtaining written informed consent, 45–50 mL of
blood was drawn from a forearm vein. The sample was
immediately divided among 11 serum-separating tubes
and 11 lithium heparin tubes (BD VacutainerÕ , Becton,
Dickinson South Africa, Gauteng, South Africa).
Serum-separating tubes were each filled with 1 mL of
blood and heparin lithium tubes with 3 mL of blood.
One serum and one plasma tube were centrifuged, and
osmolality was determined within 30 min to provide a
baseline value for each medium. Five tubes from each
medium were stored at 4–8 C and the remaining five
tubes at room temperature. Pairs of serum and plasma
samples from the cooled and room temperature sets of
samples were centrifuged and measured at 3, 6, 12, 24
and 36 h.
The stability of urine osmolality was assessed in a
similar manner. A fresh urine sample was divided
amongst 11 non-sterile, plastic 5 mL tubes containing
no additives. A baseline sample was centrifuged and
measured within 30 min. The remaining samples were
split into a group of tubes stored at 4–8 C and a group
stored at room temperature. Samples from each group
then had osmolality determinations at 3, 6, 12, 24 and
36 h.
Osmometry
All osmolality measurements were determined by freezing-point depression. Samples from the 16 patients were
analysed on the Advanced Micro-Osmometer 3320
(Advanced Instruments, Inc., Norwood, MA, USA).
The instrument has a within-day coefficient of variation
(CV) of 0.9% for measurements below 400 mosmol/kg
and a CV of 0.5% for measurements between 400 and
2000 mosmol/kg. Baseline measurements falling outside
the 280–295 mosmol/kg reference intervals for serum/
plasma osmolality were repeated as per the standard
operating procedure in our laboratory. Measurements
at later time points were repeated if they differed by
more than 10 mosmol/kg from the baseline value to
minimize random errors.
A Micro-Osmometer 3300 (Advanced Instruments,
Incorporated Norwood, MA, USA) was used to measure the samples from the healthy volunteers. This
instrument has a within-day CV of 0.87% for serum
Statistical analysis
Results
There were 483 requests for serum osmolality and 628
requests for urine osmolality in 2012 at Tygerberg
Hospital. One-hundred and nine (22%) of the serum
samples and 132 (21%) of the urine samples were
rejected because they were not transported on ice and
had not reached the laboratory within 3 h of sampling.
In the samples obtained from our participants, blood
osmolality measurements ranged from 252 to 311
mosmol/kg (n ¼ 30). No haemolysed specimens were
received. Urine osmolality ranged from 49 to 1157
mosmol/kg (n ¼ 40).
Table 1 shows the changes in serum, plasma and
urine osmolality over time. The average room temperature during the period of data collection was 22 C.
Figure 1 demonstrates the serum and plasma osmolality at 4–8 C and room temperature at each time
point. In serum samples kept at room temperature,
there was no difference between measurements at baseline and those made at 3, 6 or 12 h. The average change
from baseline was 0.4 mosmol/kg at 3 h (P ¼ 0.60),
1.0 mosmol/kg at 6 h (P ¼ 0.20) and 1.1 mosmol/kg
Bezuidenhout et al.
455
Table 1. Changes in serum, plasma and urine osmolality in stored samples over time.
Serum
Plasma
Urine
Time from
sampling (h)
4–8 C
Room temperature
4–8 C
Room temperature
4–8 C
Room temperature
3
6
12
24
36
0.2 3.3
0.5 3.0
0.7 3.1
0.2 3.7
0.9 4.4
0.4 2.9
1.0 4.4
1.1 2.5
3.5 3.6 a
6.4 4.2 a
2.0 4.5
1.8 4.3
2.1 4.3
0.6 5.7
0.9 10.1
1.1 4.9
0.1 4.7
0.7 4.2
3.5 5.7 a
3.3 3.3 a
1.6 4.0
1.0 4.6
0.3 4.5
1.5 3.7
1.2 3.2
0.1 6.1
1.7 4.4
0.1 5.1
0.7 4.3
0.5 5.2
Mean differences from baseline values SD are shown in mosmol/kg.
a
Indicates statistically significant changes from baseline values.
Figure 1. Serum and plasma osmolality at each time point. Mean values and 95% confidence intervals are shown. In serum samples at
room temperature, there was an increase of 3.5 mosmol/kg (P < 0.005) at 24 h and 6.4 mosmol/kg (P < 0.005) by 36 h. In cooled
samples, there were no significant changes from baseline measurements for up to 36 h. In plasma samples at room temperature, the
average change from baseline at 24 h was 3.5 mosmol/kg (P ¼ 0.001), with no further increase at 36 h. In cooled samples, there was a
fall in plasma osmolality at 12 h of 2.1 mosmol/kg (P ¼ 0.04) but no significant changes at the other time points.
at 12 h (P ¼ 0.15). Statistically significant changes were
present at 24 and 36 h. At 24 h, the average change
from baseline was an increase of 3.5 mosmol/kg
(P < 0.05), and by 36 h, the mean increase was 6.4
mosmol/kg (P < 0.05).
In plasma samples kept at room temperature, no
significant change from baseline measurements was
observed up to 12 h. Like serum, the average change
from baseline at 24 h was 3.5 mosmol/kg, but at 36 h,
there was no further increase.
Serum and plasma samples stored at 4–8 C
showed no significant change from baseline osmolality measurements (see Table 1). For serum, the
average change from baseline in the cooled samples
was less than 1 mosmol/kg at all time points. For
plasma, there was a small decease in osmolality from
baseline values of approximately 2 mosmol/kg at 3,
6 and 12 h, and then at 24 and 36 h small increases
of less than 1 mosmol/kg from baseline values were
recorded.
Figures 2 and 3 illustrate the changes in serum osmolality at 24 and 36 h for samples stored at room temperature. Figure 4 illustrates the changes in plasma
osmolality at 36 h at room temperature. Dashed lines
indicate the limits of the total allowable error and
dotted lines indicate the reference change value.
456
Annals of Clinical Biochemistry 53(4)
Figure 2. Serum osmolality of samples at room temperature: deviation from baseline at 24 h. Dashed lines indicate the limits of the
total allowable error and dotted lines indicate the reference change value.
Figure 3. Serum osmolality of samples at room temperature: deviation from baseline at 36 h. Dashed lines indicate the limits of the
total allowable error and dotted lines indicate the reference change value.
Urine osmolality was found to be extremely stable.
Analysis of variance with repeated measures for the
factor time indicated no significant differences between
baseline measurements and measurements made at 3, 6,
12, 24 and 36 h at 4–8 C or at room temperature
(P ¼ 0.21). The average deviations from baseline osmolality measurements did not exceed 2 mosmol/kg at any
time point (see Table 1). Figure 5 illustrates percentage
changes from baseline urine osmolality at 36 h.
Absolute differences are shown in Figure S4 of the supplementary materials. Most measurements do not
deviate by more than 2% (or 10 mosmol/kg) from baseline values.
Discussion
The investigation of salt and water disorders and the
calculation of serum and urine osmolal gaps depend on
the accurate measurement of osmolality. Our study has
demonstrated good stability of osmolality in blood and
urine samples stored at room temperature for up to
36 h. The wide range of blood and urine osmolality
Bezuidenhout et al.
457
Figure 4. Plasma osmolality of samples at room temperature: deviation from baseline at 36 h. Dashed lines indicate the limits of the
total allowable error and dotted lines indicate the reference change value.
Figure 5. Urine osmolality of samples at room temperature: percentage deviation from baseline at 36 h. Most measurements do not
deviate by more than 2% from baseline values.
included in the study makes the results applicable to a
variety of clinical settings, and it therefore appears that
the rejection of so many samples based on the current
recommendations can be avoided.
There were no changes in the osmolality of serum
and plasma samples stored at 4–8 C for up to 36 h. In
the samples stored at room temperature, we found no
changes in the osmolality of samples stored for up to
12 h. At 24 and 36 h, there were statistically significant
increases in the osmolality of the samples kept at room
temperature. At 24 h, the average increase for both
serum and plasma osmolality was 1.2%. At 36 h,
there was no further change for plasma samples but
the osmolality of the serum samples increased to
2.1% above baseline values, exceeding the total allowable error. Importantly, these changes did not exceed
the reference change value and were not regarded as
clinically significant.
In our setting, most samples for osmolality reach the
laboratory within 12 h. Sending samples on ice or
458
storing them at 4–8 C before measurement is therefore
not necessary. In rural clinical settings, the time from
sampling to measurement may well exceed 24 or even
36 h. In such cases, it would be acceptable to do serum
osmolality measurements up to 36 h, but cooling serum
specimens to 4–8 C or sending a heparinized specimen
for measurement of plasma osmolality may be preferable if delays beyond 12 h are anticipated.
Our data revealed excellent stability of urine osmolality over 36 h. There was no difference in osmolality
measurements between urine samples stored at 4–8 C
and those kept at room temperature. The average deviation from baseline at 36 h was 0.46 mosmol/kg in samples at room temperature. The total allowable error for
urine osmolality is 39.5% and the RCV is 88%.12 Most
of our samples measured at 36 h were within 2%
(or within 10 mosmol/kg) of baseline osmolality
values. Sending urine samples on ice or storing them
at 4–8 C before measurement is therefore not required.
The effect on osmolality of storing samples at higher
ambient temperatures was not evaluated in this study.
We therefore cannot comment on the stability of samples when very warm transportation or storage conditions exist.
In conclusion, we suggest that existing recommendations for sample stability and storage for osmolality be
amended. Serum and plasma samples can be stored for
up to 36 h before separation and measurement of osmolality. Cooling blood samples to 4–8 C is advised when
delays in measurement beyond 12 h are anticipated.
Samples for urine osmolality can be stored for up to
36 h before measurement and cooling of these samples
is not required.
Note
Since the acceptance of this article the authors have become aware of the letter
by Abbadi et al.15 reporting good stability of osmolality in heparinised whole
blood samples kept at room temperature for up to 24 h.
Acknowledgements
We thank Wessel Kleinhans for his help in determining the specimen rejection
rate as well as Tarryn Abrahams and the technicians at the Tygerberg Hospital
NHLS and the PathCare laboratories for the osmolality measurements. We
also thank Justin Harvey of the Stellenbosch University Centre for Statistical
Consultation for assistance with data analysis.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the
research, authorship, and/or publication of this article.
Annals of Clinical Biochemistry 53(4)
Funding
The osmolality measurements were provided by the National Health
Laboratory Service and PathCare Laboratories of South Africa.
Ethical approval
Ethical approval was granted by the Health Research Ethics Committee of
Stellenbosch University (reference number S13/03/052).
Guarantor
MRD.
Contributorship
KB and MRD conceived the study, developed the initial protocol and gained
ethical approval. MAR, CLH and YE contributed to refining the protocol. KB,
MAR, CLH and YE were involved with data collection. KB, MRD and MAR
were involved with data analysis. KB wrote the first draft of the manuscript. All
authors reviewed earlier drafts of the manuscript and approved the final version.
References
1. Halperin ML and Goldstein MB. Fluid, electrolyte, and acid-base physiology: a problem-based approach, 3rd ed. Philadelphia, PA: W.B.
Saunders, 1999.
2. World Health Organization. Use of anticoagulants in diagnostic laboratory
investigations and stability of blood, plasma and serum samples. Geneva:
World Health Organization, 2002.
3. Zhang DJ, Elswick RK, Miller WG, et al. Effect of serum-clot contact time
on clinical chemistry laboratory results. Clin Chem 1998; 44: 1325–1333.
4. Boyanton BL Jr and Blick KE. Stability studies of twenty-four analytes in
human plasma and serum. Clin Chem 2002; 48: 2242–2247.
5. Boning D and Maassen N. Blood osmolality in vitro: dependance on pCO2,
lactic acid concentration, and O2 saturation. J Appl Physiol Respir Environ
Exerc Physiol 1983; 54: 118–122.
6. Seifarth CC, Miertschischk J, Hahn EG, et al. Measurement of serum and
plasma osmolality in healthy young humans – influence of time and storage
conditions. Clin Chem Lab Med 2004; 42: 927–932.
7. Bohnen N, Terwel D, Markerink M, et al. Pitfalls in the measurement of
plasma osmolality pertinent to research in vasopressin and water metabolism. Clin Chem 1992; 38: 2278–2280.
8. Redetzki HM, Hughes JR and Redetzki JE. Differences between serum and
plasma osmolalities and their relationship to lactic acid values. Proc Soc
Exp Biol Med 1972; 139: 315–318.
9. Bunting PS, Reid SR and Ngan M. Comparison of serum and plasma
osmolalities in a surgical ICU. Crit Care Med 1986; 14: 650.
10. Curria A. Refrigerated and room temperature storage stability of urine osmolality measurements. Norwood, MA: Advanced Instruments, Inc., 2011.
11. Cooper GR, Sampson EJ and Smith SJ. Preanalytical, including biological,
variation in lipid and apolipoprotein measurements. Curr Opin Lipidol
1992; 3: 365–371.
12. Garcia LS. Clinical laboratory management. Washington, DC: ASM Press,
2004.
13. Ricos C, Alvarez V, Cava F, et al. Desirable specifications for total error,
imprecision, and bias, derived from intra- and inter-individual biologic
variation. http://www.westgard.com/biodatabase1.htm (accessed 23 May
2015).
14. Fraser CG. Biological variation: from principles to practice. Washington,
DC: AACC Press, 2001.
15. Abbadi A, El-Khoury JM and Wang S. Stability of serum and plasma
osmolality in common clinical laboratory storage conditions. Clin
Biochem 2014; 47: 686–687.