- Email: [email protected]
Urine-creatinine concentration as a marker of urine dilution: Reflections using a cohort of 45,000 samples
Forensic Science International 186 (2009) 48–51
Contents lists available at ScienceDirect
Forensic Science International journal homepage: www.elsevier.com/locate/forsciint
Urine-creatinine concentration as a marker of urine dilution: Reflections using a cohort of 45,000 samples T. Arndt * Bioscientia Institut fuer Medizinische Diagnostik GmbH, Konrad-Adenauer-Straße 17, D - 55218 Ingelheim, Germany
A R T I C L E I N F O
A B S T R A C T
Article history: Received 12 August 2008 Received in revised form 7 January 2009 Accepted 14 January 2009 Available online 11 February 2009
Background: Statistical data from a wide cohort of subjects should provide arguements for a more valid interpretation of urine-creatinine concentrations as laboratory marker of urine dilution. Methods: Unselected, consecutive urine-creatinine concentrations from 11,811 women and 13,009 men in a clinical chemistry laboratory (mainly from clinical trials and employment medicine departments) and from 7300 women and 12,456 men in a toxicological chemistry laboratory (mainly from drug screenings for re-issuing drivers licenses or from employment medicine departments) were evaluated by descriptive and comparative statistics. Results: Women (clinical chemistry lab, toxicological chemistry lab): mean 723 mg/L, 921 mg/L; median 568 mg/L, 728 mg/L; 2.5–97.5% percentile range 189–2198 mg/ L, 129–2690 mg/L. Men (clinical chemistry lab, toxicological chemistry lab): mean 975 mg/L, 1395 mg/L; median 802 mg/L, 1241 mg/L; 2.5–97.5% percentile range 256–2660 mg/L, 204–3520 mg/L. The rate of urine-creatinine concentrations of >3000 mg/L (up to 3-fold of the upper limit of the reference range) was higher for men in both laboratories and for both genders in the toxicological chemistry lab compared with the clinical chemistry lab (toxicological chemistry lab: 697 for men (5.6%) and 111 for women (1.5%), clinical chemistry lab: 200 for men (1.5%) and 93 for women (0.8%)). Conclusions: Utmost caution should be taken when interpreting urinary creatinine concentrations that fall below so-called cut-offs. Cut-offs greater than the gender-specific 2.5% percentiles bear a high risk of misinterpretation regarding urine adulteration. Such cut-offs are no longer acceptable. At present, the borderline range of >50 mg/L to <200 mg/L given by the Australian Standard AS/NZS4308:2008 and indicating dilute urines but are not implicated in adulteration seems to fit best with the clinical and forensic requirements. Nevertheless, using gender-independent urine-creatinine concentration cut-offs can discriminate women since women have in general lower muscle mass and thus lower urinary creatinine concentrations compared with men. Future concepts of drug screen in urine should use gender-specific and creatinine-adjusted decision limits. ß 2009 Elsevier Ireland Ltd. All rights reserved.
Keywords: Creatinine Cut-off Drug screen Gender Urine
1. Introduction Urine-creatinine concentrations have been used for a long time for detecting diluted (adulterated) urine samples. However, the definition of a valid cut-off value for the differentiation of a diluted from a physiological urine is still a challenge because of the broad and gender-specific reference ranges of urine-creatinine concentrations (e.g. 1st morning urine women: 280–2170 mg/L, men: 390–2590 mg/L [1]). Cut-off values of e.g. 300 mg/L [2], 250 mg/L [3], 200 mg/L [4,5] and 50 mg/L [6] have been suggested. Current German regulations for re-issuing drivers licences rate urinecreatinine concentrations of <100 mg/L as ‘‘water’’, of <200 mg/L
* Tel.: +49 6132 781 349. E-mail address: [email protected]. 0379-0738/$ – see front matter ß 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.forsciint.2009.01.010
as ‘‘not useable’’ and already concentrations of <900–1000 mg/L as ‘‘suspicious of urine adulteration’’ [7]. The latter has been clearly in contrast to the experience from urine-creatinine measurements in a clinical chemistry laboratory with approx. 70% of the urinecreatinine concentrations below that range of 900–1000 mg/L. This observation prompted a statistical evaluation of approx. 25,000 urine-creatinine concentrations from the clinical chemistry laboratory and of additionally approx. 20,000 urine-creatinine concentrations from the toxicological chemistry laboratory of Bioscientia Ingelheim. The aim of the study was to provide statistical data from large analysis series of unselected probands and patients, postulating that the results are representative for the whole population. It was also aimed as an impulse for a genderspecific evaluation of the urine-creatinine concentrations as a marker of urine dilution (adulteration) and for the establishment of analyte/creatinine ratio cut-offs in urinary drug analysis.
T. Arndt / Forensic Science International 186 (2009) 48–51 2. Materials and methods 2.1. Patients Approx. 45,000 unselected, consecutive urine-creatinine concentrations measured in the clinical chemistry laboratory (11,811 women, 13,009 men) and the toxicological chemistry laboratory (7300 women, 12,456 men) of Bioscientia (Ingelheim, Germany) were evaluated. Samples from the clinical chemistry laboratory were sent to the laboratory mainly from pharmaceutical companies (regular check-up as part of clinical trials), from centers for employment medicine, hospitals and to a smaller part from general practitioners. Samples from the toxicological chemistry laboratory were sent for drug screen analysis mainly from centers for traffic medicine (re-issuing drivers licenses), from employment medicine departments (for regular check-ups of workers in jobs with increased safety demands), from addiction therapy centers and to a minor part from general practitioners (family checks for drug use of the children). Urine-creatinine is a regular part of such analyses in this laboratory. 2.2. Determination of urine-creatinine Creatinine was determined using a modification of the Jaffe method: formation of a yellow-orange creatinine-picric acid complex after addition of picric acid to the alkaline urine sample, kinetic measurement of the absorption (with compensation of pseudocreatinines and unspecific background absorption) and quantification of creatinine by a linear calibration function. Urine-creatinine was determined with a Hitachi 917 system (Roche, Germany) in the clinical chemistry laboratory and with a Hitachi 912 system (Roche, Germany) in the toxicological chemistry laboratory using original reagents in accordance with the test instructions. Both methods are Conformite´ Europe´enne (EU) certified analysis methods (being in accordance with the regulations of the European Union). The Hitachi 917 creatinine analysis in short: HiCo Creatinine Jaffe method, rate blanked and compensated reagent (standardized against isotope dilution mass spectrometry) delivered by Roche (Germany). The method uses a 1-point calibrator (Precinorm PUC 967 mg/L calibrator, Roche) and linear regression forcing the calibration curve through the Zero point. Linearity of the method is checked semiannually by dilution of a native urine sample with a high creatinine concentration with water. The last linearity check was done with an urine sample with 4800 mg creatinine/L and 6 dilutions of up to 1000-fold. Linear regression yielded y = 0.99x 14 (y = measured, x = target). The 95% confidence intervals for the slope (0.9688 to 1.0217) and for the intercept ( 74 to 44) were not statistically different from the ideal values 1 and 0, proving the method to be with 95% probability linear between 5 and 4800 mg creatinine/L (correlation coefficient r = 1). The Hitachi 912 creatinine analysis in short: DRI1 Creatinine-Detect1 Test reagent (Thermo Fisher Scientific Microgenics, Germany); 2-point calibration (CreatinineDetect Calibrator Kit 20 mg/L, 200 mg/L calibrators, Thermo Fisher Scientific Microgenics, Germany) and linear regression with extrapolation from the lower calibration point to the Zero point. The linearity of the method is regularly checked as described above for the Hitachi 917 system. The last linearity check yielded y = 0.99x 40 (y = measured, x = target). The 95% confidence intervals for the slope (0.9399 to 1.0442) and for the intercept ( 156 to 77) were not statistically different from the ideal values 1 and 0, proving the method to be with 95% probability linear between 5 and 4800 mg creatinine/L (correlation coefficient r = 1). Intra-assay CVs (n = 10) for creatinine concentrations <10 mg/L were 30–80% (minimum values in Table 1), for 50 mg/L <10% and for >50 mg/L <3%. Inter-assayCVs calculated monthly were regularly <5% for the two analytical systems. Annotation 1: Creatinine concentrations higher than the upper level of the tested linearity ranges (e.g. the Maximum concentrations in Table 1) have been determined by re-analysis after urine dilution using a standardized (and accredited) dilution protocol. Annotation 2: The two analysis systems (Hitachi 917 with Roche reagents vs. Hitachi 912 with Thermo Fisher Scientific Microgenics reagents) were tested for comparability (agreement) by the method of Passing–Bablok [8], using the data from 6 ring trials (Instand eV, Germany) with each 2 samples: the regression function was Hitachi 912 = 0.923 Hitachi 917 + 1.604 with 95% confidence
49
intervals of 0.842–1.011 for the slope and 4.763 to 7.418 for the intercept. Because these intervals include the ideal values of 1 for the slope and 0 for the intercept, slope and intercept of the regression function are with 95% probability not different from 1 and 0. Thus, the regression function was with 95% probability Hitachi 912 = Hitachi 917 and the urine-creatinine concentrations with 95% probability identical (comparable). 2.3. Quality control Internal and external quality control included measurement of commercially available QC material with physiological and pathological urine-creatinine concentrations (LiquicheckTM Urine Chemistry Control Level 1 and Level 2, BioRad, Germany) in each analysis series and documentation of the analysis results in QC protocols. The laboratory participated regularly (successful) in national and international ring trials (INSTAND e. V., Germany; CAP – College of American Pathologists, USA; DGKL – United German Society of Clinical Chemistry and Laboratory Medicine). Bioscientia GmbH is a CAP- and DIN EN ISO 17025 accredited laboratory. 2.4. Statistics Statistics have been calculated with the Analyse-it for Microsoft Excel software (Analyse-it Software Ltd., Leeds, United Kingdom).
3. Results Mean, median, 2.5% and 97.5% percentile of the urine-creatinine concentrations were in both laboratories (as expected) lower for women when compared with men (Table 1; t-test for the means between women and men in the clinical chemistry lab p < 0.0001 and between women and men in the toxicological chemistry lab p < 0.0001). 3.1. Urine-creatinine in the clinical chemistry laboratory Altogether, 77% percent of the urine samples from women and 57% of those from men had urine-creatinine concentrations of <900 mg/L. In detail, 35 samples from men (0.3% of 13,009) and 28 from women (0.2% of 11,811) had to be qualified as ‘‘water’’ (urinecreatinine <100 mg/L), 343 samples from men (2.6%) and 177 from women (1.5%) as ‘‘not useable’’ (urine-creatinine <200 mg/L) and 7462 samples from men (57%) and 9156 from women (77%) as ‘‘suspicious of urine adulteration’’ when applying the criteria given in [7]. 3.2. Urine-creatinine concentrations in the toxicological chemistry laboratory Altogether, 61% of the urine samples from women and 34% of the urine samples from men had urine-creatinine concentrations <900 mg/L. In detail, 83 samples from men (0.7% of 12,456) and 32 from women (0.4% of 7300) had to be qualified as ‘‘water’’ (urinecreatinine <100 mg/L), 279 samples from men (2.2%) and 328 from women (4.5%) as ‘‘not useable’’ (urine-creatinine <200 mg/L) and 4306 samples from men (34%) and 4427 from women (61%) as ‘‘suspicious of urine adulteration’’ when applying the criteria given in [7].
Table 1 Descriptive statistics of approx. 45,000 consecutively measured, unselected urine-creatinine concentrations [mg/L] from a clinical chemistry and a toxicological chemistry laboratory. 2.5% = 2.5% percentile, 97.5% = 97.5% percentile of the urine-creatinine concentration (per definition the lower and upper limits of a reference range). Min/ Max = minimum/maximum concentration; CI = confidence interval. Annotation: Min values are outside the calibration range and are strongly effected by high CVs (up to 80% in 10-fold analysis). These values are given only for documentation purposes. Max values have been obtained after sample dilution according to a standardized dilution protocol.
Women (toxic.-chem. lab) Women (clin.-chem. lab) Men (toxic.-chem. lab) Men (clin.-chem. lab)
n
Min
Max
Mean (95% CI)
Median (95% CI)
2.5%
97.5%
Ref.-range [1]
7300 11,811 12,456 13,009
9 12 1 1
7698 6800 8886 5727
921 723 1395 975
728 568 1241 802
129 189 204 256
2690 2198 3520 2660
280–2170 280–2170 390–2590 390–2590
(905–937) (713–732) (1379–1410) (964–986)
(710–743) (562–576) (1218–1261) (792–813)
50
T. Arndt / Forensic Science International 186 (2009) 48–51
3.3. Comparison between the clinical chemistry and the toxicological chemistry laboratories Urine-creatinine concentrations from the toxicological chemistry laboratory were in average higher for both genders (Table 1; e.g. mean for men clinical chemistry lab vs. toxicological chemistry lab: 975 mg/L vs. 1395 mg/L, 2-side t-test p < 0.0001; mean for women clinical chemistry lab vs. toxicological chemistry lab: 723 mg/L vs. 921 mg/L, 2-side t-test p < 0.0001). The 2.5–97.5% percentile concentration range was broader for each gender in the toxicological chemistry lab (Table 1). Unphysiologically high urine-creatinine concentrations of up to 3-fold of the upper limit of the reference range (row Max in Table 1) were found for both genders and in both laboratories. However, the rate of urine-creatinine concentrations of >3000 mg/L was higher in the toxicological chemistry lab with 697 men (5.6%) and 111 women (1.5%) compared with the clinical chemistry lab with 200 men (1.5%) and 93 women (0.8%).
4. Discussion Special care was taken to provide statistical data from large analysis series of unselected probands and patients, postulating that the results are representative for the whole population. Therefore, urine-creatinine concentrations were neither grouped e.g. with regard to age, body mass (index) or race of the patient nor according to such criteria like absence or presence of kidney and/or muscle disease. The rational behind this concept was that even basic information like age and body mass is most often missing in the analysis request formulars based on patient codes. Kidney, muscle or liver diseases can cause diminished or increased creatinine concentrations but can be unknown (or still undiagnosed) at the time of urine sampling and thus will not be considered in the interpretation of urine-creatinine concentrations as a marker of urine dilution (adulteration). It becomes clear from this study, urine-creatinine concentrations of <900–1000 mg/L cannot be rated as ‘‘suspicious of urine adulteration’’ like recommended in [7]. Applying such a cut-off would cause an unplausible high rate of ‘‘suspicious urines’’ (57%
Fig. 1. Frequency distribution of the urine-creatinine concentrations from 11,811 women and 13,009 men from a clinical chemistry laboratory and from 7300 women and 12,456 men from a toxicological chemistry laboratory. For reasons of a better resolution in the relevant concentration range, urine-creatinine results of >3000 mg creatinin/L (clinical chemistry laboratory: 93 women and 200 men, toxicological chemistry laboratory: 111 women and 697 men) are not presented here (but are considered in the statistics of Table 1).
T. Arndt / Forensic Science International 186 (2009) 48–51
for men, 77% for women) even in a clinical chemistry laboratory with mainly healthy patients from clinical trials and no need for a urine adulteration for the persons tested. A revision of this cut-off is needed. The 2.5–97.5% percentile ranges of the urine-creatinine concentrations obtained for women (189–2198 mg/L) and men (256–2660 mg/L) in the clinical chemistry lab are in good agreement with urine-creatinine concentration reference ranges from the literature, being 280–2170 mg/L for women and 390– 2590 mg/L for men in the 1st morning urine [1] (Table 1). However, the 2.5% percentile of the urine-creatinine concentration for women (189 mg/L) was clearly below suggested cut-off levels of <300 mg/L [2], of <250 mg/L [3] and of <200 mg/L [4,5] and the same is true for men (256 mg/L 2.5% percentile) when using cutoffs of >250 mg/L. This situation is not acceptable since it bears the risk of a high rate of false classifications between physiological and adulterated urine samples. Furthermore, it is important that employees or others undergoing drug testing should not be punished for maintaining a good level of hydration. Consequently, the 2008 revision of the Australian Standard AS/NZS4308 [9] has changed the lower urinary creatinine level indicating dilute urine from >100 mg/L to >50 mg/L. With the current version [9], urinary creatinine >50 mg/L but <200 mg/L are dilute but are not implicated in adulteration. It should be pointed out that any gender-independent cut-off or even a borderline range for the urine creatinine concentration bears the risk of discriminating women: the fact that women have in general a lower muscle mass and thus lower urine-creatinine excretion compared with men is well known. It is reflected in the lower reference range for women [1] and is once more confirmed by the data from the ca. 45,000 subjects of this study. Because of this, gender-independent urine-creatinine cut-offs are in any case closer to the lower limit of the reference range for women when compared with that for men. This can cause more ‘‘dilute urine’’ or even ‘‘adulterated urine’’ results for women compared with men. Data from an (unpublished) in-house longitudinal study on the excretion of ethyl glucuronide as a short time marker of alcohol intake clearly confirm this risk: even before the ingestion of 200 mL of water or red wine, urine-creatinine concentrations of <300 mg/L were found in 18 out of 70 measurements and out of these 18 observations of <200 mg/L in 13 cases and of <100 mg/L in 2 cases. Analytical errors, metabolic or organ disorders (esp. kidney malfunction) and urine adulteration were clearly ruled out as potential reasons for these very low urine-creatinine analysis results. The latter could clearly be attributed to a very slender habitus of the women tested. The consequences would have been: 18 false-positives (26% out of 70 measurements) when applying a cut-off of <300 mg/L according to [2], 13 false-positives (19%) when using a cut-off of <200 mg/L like given in [4,5] and still 2 false-positives (3%) with a cut-off of <100 mg/L. The latter (3%) are still a relatively high rate of false-positives from the forensic point of view. Following from this, the cut-off should have been further reduced to e.g. <50 mg/L like suggested in [6] and like established in the current Australian Standard AS/NZS4308:2008 [9]. The findings from the clinical chemistry laboratory are qualitatively also valid for the toxicological chemistry lab. The main important difference is the trend to higher and broader distributed urine-creatinine concentrations in the toxicological lab (Table 1, Fig. 1). The reason for this finding remains unclear. It was, however, beyond the scope of this statistical evaluation to assess potential causes for this finding.
51
Finally, modern concepts of urine analysis in clinical chemistry are based on the determination of analyte concentrations and analyte/creatinine ratios. Analyte/creatinine ratios and creatinineadjusted reference ranges in urine from children are considered as the Gold Standard in Paediatrics. Current diagnosis and follow-up of e.g. proteinuria or catecholamine-producing malignomas are unthinkable without the determination of analyte concentrations and analyte/creatinine ratios in spot urine [10]. Based on this experience and taking into account the broad and gender-specific reference ranges of urinary creatinine, everything should be undertaken to adapt these concepts even in toxicological (forensic) analysis of urine. 5. Conclusion Utmost caution should be taken when interpreting urinary creatinine concentrations that fall below so-called cut-offs. At present, the borderline range of >50 mg/L to <200 mg/L given by the Australian Standard As/NZS4308:2008 and indicating dilute urines but are not implicated in adulteration seems to fit best with the clinical and forensic requirements. Nevertheless, using genderindependent urine-creatinine concentration cut-offs bears the risk of discriminating women since women have in general lower urinary creatinine concentrations compared with men. Future concepts of drug screen in urine should use gender-specific and creatinine-adjusted decision limits. Acknowledgements Parts of this study have been published in German in the members journal of the German Society of Toxicological and Forensic Chemistry [11,12]. I thank Prof. F. Pragst (Berlin, Germany) for kindly lifting the copyright. References [1] B.C. Mazzachi, M.J. Peake, V. Ehrhardt, Reference range and method comparison studies for enzymatic and Jaffe´ creatinine assays in plasma and serum and early morning urine, Clin. Lab. 46 (2000) 53–55. [2] S.D. Needleman, M. Porvaznik, D. Ander, Creatinine analysis in single collection urine specimens, J. Forensic Sci. 37 (1992) 1125–1133. [3] M. Goll, G. Schmitt, B. Ganssmann, R.E. Aderjan, Excretion profiles of ethyl glucuronide in human urine after internal dilution, J. Anal. Toxicol. 26 (2002) 262–266. [4] A.D. Fraser, J. Zamecnik, Substance abuse monitoring by the correctional service of Canada, Ther. Drug Monit. 24 (2002) 187–191. [5] SAMHSA guidelines, substance abuse and mental health service administration. www.samhsa.gov/library/searchreal.aspx. [6] C.S. Barbanel, J.W. Winkelmann, G.A. Fischer, A.J. King, Confirmation of the department of transportation criteria for a substituted urine specimen, J. Occup. Environ. Med. 44 (2002) 407–416. [7] W. Schubert, R. Mattern (Eds.), Beurteilungskriterien. Urteilsbildung in der medizinisch-psychologischen Fahreignungsdiagnostik., Kriterienx Verlag, Bonn, 2005. [8] H. Passing, W. Bablok, A new biometrical procedure for testing the equality of measurements from two different analytical methods. Application of linear regression procedures for method comparison studies in clinical chemistry, Part I, J. Clin. Chem. Clin. Biochem. 21 (1983) 709–720. [9] Australian/New Zealand StandardTM. Procedures for specimen collection and detection and quantitation of drugs of abuse in urine. AS/NZS4308:2008. Standards Australia/Standards New Zealand, 2008. [10] A.M. Gressner, T Arndt (Eds.), Lexikon der Medizinischen Laboratoriumsdiagnostik, Springer, Heidelberg—New York, 2007. [11] T. Arndt, Urin-Kreatininkonzentration: Kenngro¨ße zur Pru¨fung auf Probenver¨ berlegungen aus ca. 25000 Urin-Kreatininbestimmunwertbarkeit? Kritische U gen in einem klinisch-chemischen Labor, Toxichem+Krimtech 74 (2007) 94–99. [12] T. Arndt, Urin-Kreatininkonzentration: Kenngro¨ße zur Pru¨fung auf Probenverwertbarkeit? Teil 2.Auswertung von ca. 20. 000 Kreatinin-Analysen im Rahmen des Drogenscreenings, Toxichem+Krimtech 74 (2007) 155–158.
Contents lists available at ScienceDirect
Forensic Science International journal homepage: www.elsevier.com/locate/forsciint
Urine-creatinine concentration as a marker of urine dilution: Reflections using a cohort of 45,000 samples T. Arndt * Bioscientia Institut fuer Medizinische Diagnostik GmbH, Konrad-Adenauer-Straße 17, D - 55218 Ingelheim, Germany
A R T I C L E I N F O
A B S T R A C T
Article history: Received 12 August 2008 Received in revised form 7 January 2009 Accepted 14 January 2009 Available online 11 February 2009
Background: Statistical data from a wide cohort of subjects should provide arguements for a more valid interpretation of urine-creatinine concentrations as laboratory marker of urine dilution. Methods: Unselected, consecutive urine-creatinine concentrations from 11,811 women and 13,009 men in a clinical chemistry laboratory (mainly from clinical trials and employment medicine departments) and from 7300 women and 12,456 men in a toxicological chemistry laboratory (mainly from drug screenings for re-issuing drivers licenses or from employment medicine departments) were evaluated by descriptive and comparative statistics. Results: Women (clinical chemistry lab, toxicological chemistry lab): mean 723 mg/L, 921 mg/L; median 568 mg/L, 728 mg/L; 2.5–97.5% percentile range 189–2198 mg/ L, 129–2690 mg/L. Men (clinical chemistry lab, toxicological chemistry lab): mean 975 mg/L, 1395 mg/L; median 802 mg/L, 1241 mg/L; 2.5–97.5% percentile range 256–2660 mg/L, 204–3520 mg/L. The rate of urine-creatinine concentrations of >3000 mg/L (up to 3-fold of the upper limit of the reference range) was higher for men in both laboratories and for both genders in the toxicological chemistry lab compared with the clinical chemistry lab (toxicological chemistry lab: 697 for men (5.6%) and 111 for women (1.5%), clinical chemistry lab: 200 for men (1.5%) and 93 for women (0.8%)). Conclusions: Utmost caution should be taken when interpreting urinary creatinine concentrations that fall below so-called cut-offs. Cut-offs greater than the gender-specific 2.5% percentiles bear a high risk of misinterpretation regarding urine adulteration. Such cut-offs are no longer acceptable. At present, the borderline range of >50 mg/L to <200 mg/L given by the Australian Standard AS/NZS4308:2008 and indicating dilute urines but are not implicated in adulteration seems to fit best with the clinical and forensic requirements. Nevertheless, using gender-independent urine-creatinine concentration cut-offs can discriminate women since women have in general lower muscle mass and thus lower urinary creatinine concentrations compared with men. Future concepts of drug screen in urine should use gender-specific and creatinine-adjusted decision limits. ß 2009 Elsevier Ireland Ltd. All rights reserved.
Keywords: Creatinine Cut-off Drug screen Gender Urine
1. Introduction Urine-creatinine concentrations have been used for a long time for detecting diluted (adulterated) urine samples. However, the definition of a valid cut-off value for the differentiation of a diluted from a physiological urine is still a challenge because of the broad and gender-specific reference ranges of urine-creatinine concentrations (e.g. 1st morning urine women: 280–2170 mg/L, men: 390–2590 mg/L [1]). Cut-off values of e.g. 300 mg/L [2], 250 mg/L [3], 200 mg/L [4,5] and 50 mg/L [6] have been suggested. Current German regulations for re-issuing drivers licences rate urinecreatinine concentrations of <100 mg/L as ‘‘water’’, of <200 mg/L
* Tel.: +49 6132 781 349. E-mail address: [email protected]. 0379-0738/$ – see front matter ß 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.forsciint.2009.01.010
as ‘‘not useable’’ and already concentrations of <900–1000 mg/L as ‘‘suspicious of urine adulteration’’ [7]. The latter has been clearly in contrast to the experience from urine-creatinine measurements in a clinical chemistry laboratory with approx. 70% of the urinecreatinine concentrations below that range of 900–1000 mg/L. This observation prompted a statistical evaluation of approx. 25,000 urine-creatinine concentrations from the clinical chemistry laboratory and of additionally approx. 20,000 urine-creatinine concentrations from the toxicological chemistry laboratory of Bioscientia Ingelheim. The aim of the study was to provide statistical data from large analysis series of unselected probands and patients, postulating that the results are representative for the whole population. It was also aimed as an impulse for a genderspecific evaluation of the urine-creatinine concentrations as a marker of urine dilution (adulteration) and for the establishment of analyte/creatinine ratio cut-offs in urinary drug analysis.
T. Arndt / Forensic Science International 186 (2009) 48–51 2. Materials and methods 2.1. Patients Approx. 45,000 unselected, consecutive urine-creatinine concentrations measured in the clinical chemistry laboratory (11,811 women, 13,009 men) and the toxicological chemistry laboratory (7300 women, 12,456 men) of Bioscientia (Ingelheim, Germany) were evaluated. Samples from the clinical chemistry laboratory were sent to the laboratory mainly from pharmaceutical companies (regular check-up as part of clinical trials), from centers for employment medicine, hospitals and to a smaller part from general practitioners. Samples from the toxicological chemistry laboratory were sent for drug screen analysis mainly from centers for traffic medicine (re-issuing drivers licenses), from employment medicine departments (for regular check-ups of workers in jobs with increased safety demands), from addiction therapy centers and to a minor part from general practitioners (family checks for drug use of the children). Urine-creatinine is a regular part of such analyses in this laboratory. 2.2. Determination of urine-creatinine Creatinine was determined using a modification of the Jaffe method: formation of a yellow-orange creatinine-picric acid complex after addition of picric acid to the alkaline urine sample, kinetic measurement of the absorption (with compensation of pseudocreatinines and unspecific background absorption) and quantification of creatinine by a linear calibration function. Urine-creatinine was determined with a Hitachi 917 system (Roche, Germany) in the clinical chemistry laboratory and with a Hitachi 912 system (Roche, Germany) in the toxicological chemistry laboratory using original reagents in accordance with the test instructions. Both methods are Conformite´ Europe´enne (EU) certified analysis methods (being in accordance with the regulations of the European Union). The Hitachi 917 creatinine analysis in short: HiCo Creatinine Jaffe method, rate blanked and compensated reagent (standardized against isotope dilution mass spectrometry) delivered by Roche (Germany). The method uses a 1-point calibrator (Precinorm PUC 967 mg/L calibrator, Roche) and linear regression forcing the calibration curve through the Zero point. Linearity of the method is checked semiannually by dilution of a native urine sample with a high creatinine concentration with water. The last linearity check was done with an urine sample with 4800 mg creatinine/L and 6 dilutions of up to 1000-fold. Linear regression yielded y = 0.99x 14 (y = measured, x = target). The 95% confidence intervals for the slope (0.9688 to 1.0217) and for the intercept ( 74 to 44) were not statistically different from the ideal values 1 and 0, proving the method to be with 95% probability linear between 5 and 4800 mg creatinine/L (correlation coefficient r = 1). The Hitachi 912 creatinine analysis in short: DRI1 Creatinine-Detect1 Test reagent (Thermo Fisher Scientific Microgenics, Germany); 2-point calibration (CreatinineDetect Calibrator Kit 20 mg/L, 200 mg/L calibrators, Thermo Fisher Scientific Microgenics, Germany) and linear regression with extrapolation from the lower calibration point to the Zero point. The linearity of the method is regularly checked as described above for the Hitachi 917 system. The last linearity check yielded y = 0.99x 40 (y = measured, x = target). The 95% confidence intervals for the slope (0.9399 to 1.0442) and for the intercept ( 156 to 77) were not statistically different from the ideal values 1 and 0, proving the method to be with 95% probability linear between 5 and 4800 mg creatinine/L (correlation coefficient r = 1). Intra-assay CVs (n = 10) for creatinine concentrations <10 mg/L were 30–80% (minimum values in Table 1), for 50 mg/L <10% and for >50 mg/L <3%. Inter-assayCVs calculated monthly were regularly <5% for the two analytical systems. Annotation 1: Creatinine concentrations higher than the upper level of the tested linearity ranges (e.g. the Maximum concentrations in Table 1) have been determined by re-analysis after urine dilution using a standardized (and accredited) dilution protocol. Annotation 2: The two analysis systems (Hitachi 917 with Roche reagents vs. Hitachi 912 with Thermo Fisher Scientific Microgenics reagents) were tested for comparability (agreement) by the method of Passing–Bablok [8], using the data from 6 ring trials (Instand eV, Germany) with each 2 samples: the regression function was Hitachi 912 = 0.923 Hitachi 917 + 1.604 with 95% confidence
49
intervals of 0.842–1.011 for the slope and 4.763 to 7.418 for the intercept. Because these intervals include the ideal values of 1 for the slope and 0 for the intercept, slope and intercept of the regression function are with 95% probability not different from 1 and 0. Thus, the regression function was with 95% probability Hitachi 912 = Hitachi 917 and the urine-creatinine concentrations with 95% probability identical (comparable). 2.3. Quality control Internal and external quality control included measurement of commercially available QC material with physiological and pathological urine-creatinine concentrations (LiquicheckTM Urine Chemistry Control Level 1 and Level 2, BioRad, Germany) in each analysis series and documentation of the analysis results in QC protocols. The laboratory participated regularly (successful) in national and international ring trials (INSTAND e. V., Germany; CAP – College of American Pathologists, USA; DGKL – United German Society of Clinical Chemistry and Laboratory Medicine). Bioscientia GmbH is a CAP- and DIN EN ISO 17025 accredited laboratory. 2.4. Statistics Statistics have been calculated with the Analyse-it for Microsoft Excel software (Analyse-it Software Ltd., Leeds, United Kingdom).
3. Results Mean, median, 2.5% and 97.5% percentile of the urine-creatinine concentrations were in both laboratories (as expected) lower for women when compared with men (Table 1; t-test for the means between women and men in the clinical chemistry lab p < 0.0001 and between women and men in the toxicological chemistry lab p < 0.0001). 3.1. Urine-creatinine in the clinical chemistry laboratory Altogether, 77% percent of the urine samples from women and 57% of those from men had urine-creatinine concentrations of <900 mg/L. In detail, 35 samples from men (0.3% of 13,009) and 28 from women (0.2% of 11,811) had to be qualified as ‘‘water’’ (urinecreatinine <100 mg/L), 343 samples from men (2.6%) and 177 from women (1.5%) as ‘‘not useable’’ (urine-creatinine <200 mg/L) and 7462 samples from men (57%) and 9156 from women (77%) as ‘‘suspicious of urine adulteration’’ when applying the criteria given in [7]. 3.2. Urine-creatinine concentrations in the toxicological chemistry laboratory Altogether, 61% of the urine samples from women and 34% of the urine samples from men had urine-creatinine concentrations <900 mg/L. In detail, 83 samples from men (0.7% of 12,456) and 32 from women (0.4% of 7300) had to be qualified as ‘‘water’’ (urinecreatinine <100 mg/L), 279 samples from men (2.2%) and 328 from women (4.5%) as ‘‘not useable’’ (urine-creatinine <200 mg/L) and 4306 samples from men (34%) and 4427 from women (61%) as ‘‘suspicious of urine adulteration’’ when applying the criteria given in [7].
Table 1 Descriptive statistics of approx. 45,000 consecutively measured, unselected urine-creatinine concentrations [mg/L] from a clinical chemistry and a toxicological chemistry laboratory. 2.5% = 2.5% percentile, 97.5% = 97.5% percentile of the urine-creatinine concentration (per definition the lower and upper limits of a reference range). Min/ Max = minimum/maximum concentration; CI = confidence interval. Annotation: Min values are outside the calibration range and are strongly effected by high CVs (up to 80% in 10-fold analysis). These values are given only for documentation purposes. Max values have been obtained after sample dilution according to a standardized dilution protocol.
Women (toxic.-chem. lab) Women (clin.-chem. lab) Men (toxic.-chem. lab) Men (clin.-chem. lab)
n
Min
Max
Mean (95% CI)
Median (95% CI)
2.5%
97.5%
Ref.-range [1]
7300 11,811 12,456 13,009
9 12 1 1
7698 6800 8886 5727
921 723 1395 975
728 568 1241 802
129 189 204 256
2690 2198 3520 2660
280–2170 280–2170 390–2590 390–2590
(905–937) (713–732) (1379–1410) (964–986)
(710–743) (562–576) (1218–1261) (792–813)
50
T. Arndt / Forensic Science International 186 (2009) 48–51
3.3. Comparison between the clinical chemistry and the toxicological chemistry laboratories Urine-creatinine concentrations from the toxicological chemistry laboratory were in average higher for both genders (Table 1; e.g. mean for men clinical chemistry lab vs. toxicological chemistry lab: 975 mg/L vs. 1395 mg/L, 2-side t-test p < 0.0001; mean for women clinical chemistry lab vs. toxicological chemistry lab: 723 mg/L vs. 921 mg/L, 2-side t-test p < 0.0001). The 2.5–97.5% percentile concentration range was broader for each gender in the toxicological chemistry lab (Table 1). Unphysiologically high urine-creatinine concentrations of up to 3-fold of the upper limit of the reference range (row Max in Table 1) were found for both genders and in both laboratories. However, the rate of urine-creatinine concentrations of >3000 mg/L was higher in the toxicological chemistry lab with 697 men (5.6%) and 111 women (1.5%) compared with the clinical chemistry lab with 200 men (1.5%) and 93 women (0.8%).
4. Discussion Special care was taken to provide statistical data from large analysis series of unselected probands and patients, postulating that the results are representative for the whole population. Therefore, urine-creatinine concentrations were neither grouped e.g. with regard to age, body mass (index) or race of the patient nor according to such criteria like absence or presence of kidney and/or muscle disease. The rational behind this concept was that even basic information like age and body mass is most often missing in the analysis request formulars based on patient codes. Kidney, muscle or liver diseases can cause diminished or increased creatinine concentrations but can be unknown (or still undiagnosed) at the time of urine sampling and thus will not be considered in the interpretation of urine-creatinine concentrations as a marker of urine dilution (adulteration). It becomes clear from this study, urine-creatinine concentrations of <900–1000 mg/L cannot be rated as ‘‘suspicious of urine adulteration’’ like recommended in [7]. Applying such a cut-off would cause an unplausible high rate of ‘‘suspicious urines’’ (57%
Fig. 1. Frequency distribution of the urine-creatinine concentrations from 11,811 women and 13,009 men from a clinical chemistry laboratory and from 7300 women and 12,456 men from a toxicological chemistry laboratory. For reasons of a better resolution in the relevant concentration range, urine-creatinine results of >3000 mg creatinin/L (clinical chemistry laboratory: 93 women and 200 men, toxicological chemistry laboratory: 111 women and 697 men) are not presented here (but are considered in the statistics of Table 1).
T. Arndt / Forensic Science International 186 (2009) 48–51
for men, 77% for women) even in a clinical chemistry laboratory with mainly healthy patients from clinical trials and no need for a urine adulteration for the persons tested. A revision of this cut-off is needed. The 2.5–97.5% percentile ranges of the urine-creatinine concentrations obtained for women (189–2198 mg/L) and men (256–2660 mg/L) in the clinical chemistry lab are in good agreement with urine-creatinine concentration reference ranges from the literature, being 280–2170 mg/L for women and 390– 2590 mg/L for men in the 1st morning urine [1] (Table 1). However, the 2.5% percentile of the urine-creatinine concentration for women (189 mg/L) was clearly below suggested cut-off levels of <300 mg/L [2], of <250 mg/L [3] and of <200 mg/L [4,5] and the same is true for men (256 mg/L 2.5% percentile) when using cutoffs of >250 mg/L. This situation is not acceptable since it bears the risk of a high rate of false classifications between physiological and adulterated urine samples. Furthermore, it is important that employees or others undergoing drug testing should not be punished for maintaining a good level of hydration. Consequently, the 2008 revision of the Australian Standard AS/NZS4308 [9] has changed the lower urinary creatinine level indicating dilute urine from >100 mg/L to >50 mg/L. With the current version [9], urinary creatinine >50 mg/L but <200 mg/L are dilute but are not implicated in adulteration. It should be pointed out that any gender-independent cut-off or even a borderline range for the urine creatinine concentration bears the risk of discriminating women: the fact that women have in general a lower muscle mass and thus lower urine-creatinine excretion compared with men is well known. It is reflected in the lower reference range for women [1] and is once more confirmed by the data from the ca. 45,000 subjects of this study. Because of this, gender-independent urine-creatinine cut-offs are in any case closer to the lower limit of the reference range for women when compared with that for men. This can cause more ‘‘dilute urine’’ or even ‘‘adulterated urine’’ results for women compared with men. Data from an (unpublished) in-house longitudinal study on the excretion of ethyl glucuronide as a short time marker of alcohol intake clearly confirm this risk: even before the ingestion of 200 mL of water or red wine, urine-creatinine concentrations of <300 mg/L were found in 18 out of 70 measurements and out of these 18 observations of <200 mg/L in 13 cases and of <100 mg/L in 2 cases. Analytical errors, metabolic or organ disorders (esp. kidney malfunction) and urine adulteration were clearly ruled out as potential reasons for these very low urine-creatinine analysis results. The latter could clearly be attributed to a very slender habitus of the women tested. The consequences would have been: 18 false-positives (26% out of 70 measurements) when applying a cut-off of <300 mg/L according to [2], 13 false-positives (19%) when using a cut-off of <200 mg/L like given in [4,5] and still 2 false-positives (3%) with a cut-off of <100 mg/L. The latter (3%) are still a relatively high rate of false-positives from the forensic point of view. Following from this, the cut-off should have been further reduced to e.g. <50 mg/L like suggested in [6] and like established in the current Australian Standard AS/NZS4308:2008 [9]. The findings from the clinical chemistry laboratory are qualitatively also valid for the toxicological chemistry lab. The main important difference is the trend to higher and broader distributed urine-creatinine concentrations in the toxicological lab (Table 1, Fig. 1). The reason for this finding remains unclear. It was, however, beyond the scope of this statistical evaluation to assess potential causes for this finding.
51
Finally, modern concepts of urine analysis in clinical chemistry are based on the determination of analyte concentrations and analyte/creatinine ratios. Analyte/creatinine ratios and creatinineadjusted reference ranges in urine from children are considered as the Gold Standard in Paediatrics. Current diagnosis and follow-up of e.g. proteinuria or catecholamine-producing malignomas are unthinkable without the determination of analyte concentrations and analyte/creatinine ratios in spot urine [10]. Based on this experience and taking into account the broad and gender-specific reference ranges of urinary creatinine, everything should be undertaken to adapt these concepts even in toxicological (forensic) analysis of urine. 5. Conclusion Utmost caution should be taken when interpreting urinary creatinine concentrations that fall below so-called cut-offs. At present, the borderline range of >50 mg/L to <200 mg/L given by the Australian Standard As/NZS4308:2008 and indicating dilute urines but are not implicated in adulteration seems to fit best with the clinical and forensic requirements. Nevertheless, using genderindependent urine-creatinine concentration cut-offs bears the risk of discriminating women since women have in general lower urinary creatinine concentrations compared with men. Future concepts of drug screen in urine should use gender-specific and creatinine-adjusted decision limits. Acknowledgements Parts of this study have been published in German in the members journal of the German Society of Toxicological and Forensic Chemistry [11,12]. I thank Prof. F. Pragst (Berlin, Germany) for kindly lifting the copyright. References [1] B.C. Mazzachi, M.J. Peake, V. Ehrhardt, Reference range and method comparison studies for enzymatic and Jaffe´ creatinine assays in plasma and serum and early morning urine, Clin. Lab. 46 (2000) 53–55. [2] S.D. Needleman, M. Porvaznik, D. Ander, Creatinine analysis in single collection urine specimens, J. Forensic Sci. 37 (1992) 1125–1133. [3] M. Goll, G. Schmitt, B. Ganssmann, R.E. Aderjan, Excretion profiles of ethyl glucuronide in human urine after internal dilution, J. Anal. Toxicol. 26 (2002) 262–266. [4] A.D. Fraser, J. Zamecnik, Substance abuse monitoring by the correctional service of Canada, Ther. Drug Monit. 24 (2002) 187–191. [5] SAMHSA guidelines, substance abuse and mental health service administration. www.samhsa.gov/library/searchreal.aspx. [6] C.S. Barbanel, J.W. Winkelmann, G.A. Fischer, A.J. King, Confirmation of the department of transportation criteria for a substituted urine specimen, J. Occup. Environ. Med. 44 (2002) 407–416. [7] W. Schubert, R. Mattern (Eds.), Beurteilungskriterien. Urteilsbildung in der medizinisch-psychologischen Fahreignungsdiagnostik., Kriterienx Verlag, Bonn, 2005. [8] H. Passing, W. Bablok, A new biometrical procedure for testing the equality of measurements from two different analytical methods. Application of linear regression procedures for method comparison studies in clinical chemistry, Part I, J. Clin. Chem. Clin. Biochem. 21 (1983) 709–720. [9] Australian/New Zealand StandardTM. Procedures for specimen collection and detection and quantitation of drugs of abuse in urine. AS/NZS4308:2008. Standards Australia/Standards New Zealand, 2008. [10] A.M. Gressner, T Arndt (Eds.), Lexikon der Medizinischen Laboratoriumsdiagnostik, Springer, Heidelberg—New York, 2007. [11] T. Arndt, Urin-Kreatininkonzentration: Kenngro¨ße zur Pru¨fung auf Probenver¨ berlegungen aus ca. 25000 Urin-Kreatininbestimmunwertbarkeit? Kritische U gen in einem klinisch-chemischen Labor, Toxichem+Krimtech 74 (2007) 94–99. [12] T. Arndt, Urin-Kreatininkonzentration: Kenngro¨ße zur Pru¨fung auf Probenverwertbarkeit? Teil 2.Auswertung von ca. 20. 000 Kreatinin-Analysen im Rahmen des Drogenscreenings, Toxichem+Krimtech 74 (2007) 155–158.