1382 CLINICAL CHEMISTRY, Vol. 30, No. 8, 1984

of laboratory results in the elderly. Postgrad Med 72, 147-152(1982).4. Cohen T, Gitman L, Lipschutz E. Liver function studies in theaged. Geriatrics 15, 824-836 (1960).

5. Gambino SR, Schrieber H. The measurement and fractionaliza-tion of bilirubin on the AutoAnalyzer by the method of Jendrassikand Grof. In Automation in Analytical Chemistry, Technicon Sym-posia 1964, Mediad Inc., New York, NY, 1965.6. Jendrassik L, Grof P. Vereinfachte photometrische Methoden

zur Bestimmung des blut Bilirubins. Bicchem Z 297, 81-89 (1938).

CLIN. CHEM. 30/8, 1382-1386 (1984)

7. Reed All, Henry RJ, Mason B. Influence of statistical methodused on the resulting estimate of normal range. Clin Chem 17,275-284 (1971).

8. Werner M, Tolls RE, Hultin JV, Mellecker J. Influence of sexand age on the normal range of eleven serum constituents. Z KimChem Kim Biochem 8, 105-115 (1970).

9. Muraca M, Degroote J, Fevery J. Sex differences of hepaticconjugation of bilirubin determine its maximal biliary excretion innon-anesthetized male and female rats. Clin Sci 64, 85-90 (1983).

Carbon Monoxide in Blood: An Improved Microliter Blood-Sample CollectionSystem, with Rapid Analysis by Gas ChromatographyHendrik J. Vreman, Linda K. Kwong, and David K. Stevenson1

We examined the sensitive assay for carboxyhemoglobinbased on reaction with K3Fe(CN)6 and gas chromatographyof the liberated CO. Our improvements included increasedbaseline stability, shorter analysis time, and simpler stan-dardization. EDTA-containing Vacutainer Tubes (lavender-stoppered) increase the carboxyhemoglobin content of bloodstored in them. The carboxyhemoglobin content of bloodstored in capillary tubes containing solid heparin and saponinremained stable for two weeks. Using our improved proce-dures, we measured the carboxyhemoglobin content of bloodfrom adults and neonates collected via venipuncture or heelor fingersticks. We observed no significant difference incarboxyhemoglobin content of blood obtained by venipunc-ture or heel stick for premature infants, 0.19 ± 0.04 vs 0.18 ±

0.03 mL of CO per 100 mL of blood, respectively (mean ±

SD). Nonsmoking adults (n = 19) had CO values (mean ±

SD) of 0.19 ± 0.03 and 0.17 ± 0.04 mL per 100 mLof blood,and smoking adults (n = 7) gave CO values of 0.96 ± 0.49and 0.91 ± 0.49 mL/dL, for venipuncture and fingerstick,respectively.

AddItIonal Keyphrases: neonates . pediatric chemistry ref-erence interval blood gases variation, source of head-space analysis

The usefulness of measuring blood carboxyhemoglobin

(COHb) in detecting increased hemolysis in newborns hasbeen documented extensively (1, 2). Several methods fordetermining CO bound to hemoglobin in blood have beendescribed (3-5). Gas chromatography (GC) is selective andsensitive for this (3, 4). The present study is based on themethod initially reported by. Collison et al. (3) with modifi-cations as described by Ostrander et al. (2).The method

depends on oxidation of hemoglobin to methemoglobin byK3Fe(CN)6, the released CO going to the reactor head space,and the headspace gas being then analyzed by GC. Themethod is sufficiently accurate but time consuming, andbaseline instability limits the number of samples that canbe analyzed to about 15 per day. Recently, a rather abrupt

Department of Pediatrics S222, Stanford University School ofMedicine, Stanford, CA 94305.

1 Direct correspondence to this author.Received April 16, 1984; accepted May 23, 1984.

increase of COHb values in blood from normal and high COproducers prompted us to re-examine the sampling proce-dures, sample-storage conditions, and the analytical methoditself; for sources of error. We found a serious potentialsource of error when Vacutainer Tubes were used for sample

storage, and we have significantly improved the analyticalprocedure and sample handling.

Materials and Methods

Subjects and Samples

This investigation was approved by the Committee for theProtection of Human Subjects in Research at StanfordUniversity, and signed informed consent was obtained be-

fore each study.Blood was sampled from infants in the Intensive Care

Nursery and from smoking and nonsmoking adult volun-teers from our staff.

Original method: Collect blood specimens, 150-500 �L, bydirect venipuncture or, via a heparin-lock, with a 1-mLtuberculin syringe. Transfer the specimens immediately tolavender-stoppered Vacutainer Tubes (47 x 10 mm, 2.4 mL;Becton-Dickinson and Co., Rutherford, NJ 07070) contain-

ing 40 �L of aqueous EDTA (7.5 g/L) as anticoagulant.Analyze the samples immediately or store at 4#{176}Cfor nolonger than two weeks.

Modified method: Obtain blood by heel stick (infant), orfingerstick (adult) and venipuncture. Collect sample directlyinto microhematocrit tubes (75 x 1 mm i.d., 70 4,; Scien-

tific Products, McGraw Park, IL 60085) that have beencoated with dried heparmn and saponin (see below). Fill thetubes to near capacity. Insert a 15 x 0.75 mm stainless steelbar, also coated with dried heparin and saponin, into eachtube, and seal with “Critocap” tube closures (Fisher Scien-tific Co., Pittsburgh, PA 15219). Mix blood and reagentsthoroughly by inverting the tubes about 10 times or untilthe liquid has become nearly clear. Analyze the sampleimmediately or store at 4 #{176}Cfor no longer than two weeks.

Reagents

Coat the interior of each microhematocrit tube with about2 4, of a freshly made 100-mg solution of saponin (SigmaChemical Co., St. Louis, MO 63187) in 1 mL (1000 USPunits) of sodium heparin. Keeping the tubes horizontal,

\�NJECTION

I GAS FLOW

101

SIDE PORT

A

Fig. 1. Needle assemblies used for purging the reaction vials (A) and sampling the vial headspace (�ga = gauge

B

CLINICAL CHEMISTRY, Vol. 30, No.8, 1984 1383

allow the water to evaporate at ambient temperature or inan oven at reduced pressui’e at about 50#{176}C.Wet thestainless steel bars with the above heparin/saponin solution(200 4, per 50 bars) in a 1.5-mL polypropylene centrifugetube, and allow the water to evaporate.

Prepare the reaction mixture for the improved method byadding the following to about 5 mL of distilled water: 50 mgof saponin, 1 g of K3Fe(CN)6, and 1 mL of potassiumphosphate buffer, 1 molIL, pH 6.0. Adjust the volume to 10mL. Prepare the solution freshly each day and keep atambient temperature, out of direct sunlight. For purging the

vials, use CO-free gas obtained by passing air through aHopcalite catalytic converter at 100 #{176}C(Trace Analytical,Menlo Park, CA 94025). This converter reduces the COconcentration in air to a negligible 20 nL/L.

Standardization

Collison et al. (3) prepared the COHb standard byexposing a hemoglobin solution to 100% CO. The COcontent of this solution was determined spectrophotometri-cally and was stable for at least three months at 4 #{176}C(2).

Instead, we use a precisely determined gas mixture of COto standardize the gas chromatograph. Obtain this standard

gas from a commercial source (we used CO in nitrogen, 23.441L, from Airco, Inc., Santa Clara, CA 95051), or mix 204, of pure CO gas (99.9%, Matheson Gas Products, EastRutherford, NJ 07073) with 1000 mL of CO-free gas in a1000-mi, acrylic syringe (Hamilton Co., Reno, NV 89510)sealed with a septum fitted into an 18-gauge needle hubwithout needle. The CO concentration in this syringe re-mains stable for at least 8 h. Using a gas-tight, low needle-volume syringe (Precision Sampling Corp., Baton Rouge,LA 70895), inject 0,25, 50,75, or 100 4, of the standard gasinto the purged reaction vials containing reagent. Analyzethe vial head space after letting it stand 10-30 mm atambient tempera ..ire.

Analytical Procedure

Our previous method for liberating hemoglobin-bound COfrom blood was based on a modification (2) of the GC methodof Collison et al. (3) in which a reaction tube (29 x 7.5 mm,1.3 mL) with fitting was substituted for a gas sample loop ofthe GC injection valve. The tube contained a magneticstirring bar (7 x 2 mm), 5 4, of 100 milL Triton X-100

(Rohm and Haas Co., Philadelphia, PA) solution to lyseerythrocytes, 60 4, of 100 g/L K3Fe(CN)6 solution for

liberating the hemoglobin-bound CO, and 5 4, of capryl

alcohol, an antifoaming agent. We then attached the tube tothe 10-port GC injection valve via a compression fitting withseptum and carrier gas inlet and outlet. The reactionmixture was stirred and purged with CO-free carrier gas for3 mm. Then, after switching the carrier gas to bypass, weadded 2 4, of blood through the septum with a 5-4, gas-tight syringe (Hamilton Co.). After letting the mixture react

for 4 mm, we let the carrier gas sweep the head space of thereaction tube assembly for 3 miii, for determination of theliberated CO.

The improved method for liberating hemoglobin-boundCO is based on the same reaction, but the reaction mixtureis different to eliminate problems with GC baseline stabil-ity. As reactor, use a vial (12 x 32 mm, 1.5-2.0 mL; PierceChemical Co., Rockford, IL 61105) with a screw-cap and ahigh-temperature septum (8-mm diam., blue silicone sheet;Alltech Associates, Los Altos, CA 94022). Place the vials inan 80-place polypropylene rack (Cole Parmer Instr. Co.,Chicago, IL 60648) with a shortened peg length (about 20mm). Dispense 204, of reagent onto the bottom of each vialwith a 1-mL gas-tight Hamilton syringe in a repeatingdispenser; seal the vial, and purge the head space with CO-free air for 2 s at 200 miimin. Use the needle assembly

shown in Figure l#{192}to avoid excess pressure in the vial.Repurge the vials for 2 5 just before adding the blood sample.We typically prepare a series of 16 vials per analytical runand analyze the blood samples in duplicate. After adding theblood or standard gas to all vials kept at ambient tempera-ture at t = 0 miii, attach the first vial 10 miii later to the GCinjection valve by means of the headspace access needleassembly shown in Figure lB.

Gas Analysis

We determined the amount of CO liberated into thereaction vial head space with a gas chromatograph equippedwith a reduction gas detector (Trace Analytical). A 10-portpneumatic sample-injection valve controlled by a digitalvalve-sequence controller (both from Valco Instruments Co.,Houston, TX 77024) injects the gas sample onto the 90 x0.53 cm (i.d.) stainless steel column, which is packed with60-80 mesh molecular sieve 5A (Alltech Associates). Col-uinn temperature is maintained at 110 #{176}C.Air carrier gas,freed from reducing gases by passage through the catalyticconverter and a mole-sieve moisture and CO2 trap (TraceAnalytical), flows through the column at a rate of 50mL/mun. Under these operatingconditions, the elution timefor CO is about 70 s and as little as 0.1 nL of CO is

1 AIR OUTLET) (2Oga * 31,2” ,pmM nisdIe)

AIR INLET��TL..STAINLESS STEEL TUBING

� (1.6n,mODn1.Omm ID)

SILICON SLEEVE(6.3mm 00* 1.6mm ID)

3.2

� 2.8

02.4

_ 2.0

N

1.60LU

1.2

LU� 0.8-J

� 0.4

00 10 20 30 40 50 60

MINUTES

Fig. 2. Reaction time course at different temperatures for a typical bloodsample from a nonsmoker

Day 06

1384 CLINICAL CHEMISTRY, Vol. 30, No. 8, 1984

detectable. A 10-mV recorder (Linear Instrument Corp.,Irvine, CA 92705) charts the detector responses at a speed of20 cm/h.

Inject vial headspace gas onto the column by switchingthe injection valve from bypass to sampling for 6 s, thenback to bypass. A moisture trap (45 x 3.5 mm i.d.) filledwith methylene-blue-dusted Mg(Cl04)2 is positioned be-tween the injection valve and the column to keep watervapor from entering the column and interfering with theanalysis. Use peak heights to correlate detector responsesbetween samples and standards. Vials and septa can bereused after washing in dilute NH4OH (50 mLfL) andrinsing with distilled H20.

Calculations

For clinical purposes, concentrations of COHb are ex-pressed as percentage of saturation, calculated from bloodhemoglobin concentrations as follows:

COHb (% sat) = Volco X 100%/([Hb] X 1.34)where Volco is the amount of CO (mL) bound to 100 mLwhole blood, [Hb] is the concentration of blood hemoglobin(g/100 mL; we used the cyanmethemoglobin method ofSigma Chemical Co.), and 1.34 is a factor expressing themaximum CO-binding capacity of 1 g of hemoglobin. Most ofthe data in this study are expressed as milliliters of CO per

100 mL of blood, and are not corrected for ambient COcontent (2, 6).

Results and Discussion

Available assays for COHb are time-consuming and haveother disadvantages. To reduce the sample analysis timeand handling, we modified the original reaction vesselconfiguration so that it allowed preparation and reaction ofa number of samples simultaneously rather than sequen-tially. We use septum-sealed vials that can be purged andreadied for reaction before attachment to the CC sampleinjection valve. Preparation and analysis time for a typicalset of 16 analyses was 1 h, compared with about 4 h by theolder method, for a similar set of analyses.

We found that problems of CC baseline deterioration were

related to three components of the previous reaction mix-ture: water vapor, capryl alcohol, and volatile contaminantsin the Triton X-100. These volatiles elute slowly from theCC column and cause baseline deterioration after repeatedanalyses. Water vapor was excluded from the column bytrapping it with Mg(Cl04)2. The reduced volume of reactionmedium (204, vs 704,) of the new method made continu-

ous mixing unnecessary, causing no foaming, and thereforeno capryl alcohol was required. Besides unremovable vola-

tileimpurities, Triton X-100 also contained compounds thatreacted with K3Fe(CN)6; we therefore replaced Triton withsaponin, an excellent lysing agent. An observation by Ce-burn et al. (4) that alkaline pH promotes liberation of COfrom hemoglobin breakdown led us to change the composi-tion of the reaction mixture further (6). Rodkey and Collison(5) reported that oxygenated blood released more CO than

reduced blood when they used neutral K3Fe(CN)6. Becausethe release of CO from COHb was complete and constant atpH 1.5-6.0, regardless of the degree of oxygenation, weadjusted the pH of our reaction medium to 6.0 by addingpotassium phosphate buffer. However, this did not complete-ly eliminate background CO generation at 25#{176}C.

COHb dissociation is complete in 6 mm at 25 or 37 #{176}C,buttakes 25 mm at 0#{176}C(Figure 2). At 25 or 37 #{176}C,a secondaryreaction occurs that continues to liberate CO, most likelybecause of nonspecific hemoglobin oxidation (4) or becauseof CO diffusing from the septum. For this study we carriedout the reaction for 10-30 mm at ambient temperature; all

sample values were corrected for a reagent blank becausewe could not find septum material that does not bleed CO orgenerate CO upon contact with the reaction mixture. How-

ever, this problem can be overcome by performing thereaction for at least 25 mm at 0#{176}C.At this temperaturethere is no upper reaction time limit and no correction fornon-COHb CO is necessary. Therefore, our present proce-

dure includes a reaction time of at least 30 miii at 0#{176}C.COfrom photo-oxidation of organic compounds (9) does notappear to interfere in our system. However, amber-coloredreaction vials are available to obviate this.

We also studied the capacity of the reagent to releasecovalently bound CO from blood. With K3Fe(CN)6 in largeexcess, the amount of blood that can be analyzed dependsonly on the saponin content of the reagent mixture. We cananalyze a maximum of 6 � of blood with 10 j�g of saponinper 20 pL of reagent. The standard reaction mixture con-tains 100 j�g of saponin per 20 p�L and allows analysis of upto 30 4, of blood, which may be necessary to generateenough CO for less-sensitive CO detectors.

Our modified assay made the standardization procedure

reproducible and more direct than the old method. Prepar-ing a COHb standard is time-consuming compared withpurchasing or preparing a dilute CO gas standard. Thereproducibility of preparing 10 such gas mixtures with useof the 200-4, sample loop was within 2% (10.00 ± 0.11pL/L, mean ± SD). The reproducibility of headspace sam-pling of 5.85 nL of CO injected into reaction vials withreagent, in terms of detector response, was 139 ± 1 mV(mean ± SD, n = 10), and for 9.36 nL ofCO, 208 ± 2 mV(n= 10). The standard curve was linear to about 200 mV. A 2-4, blood sample from a normal premature infant or non-

Table 1. WIthin-Day and Day-to-DayReproducibility of COHb Determinations In Blood

Samples from Seven Premature Infants____________ CO, mL per 100 mL blood

a b C Day3 Day5

0.25 0.29 0.28 0.30 0.270.20 - 0.22

0.200.240.21

0.220.21

- - 0.14 0.15 0.160.27 0.28 - 0.36 0.300.17 0.20 0.21 0.210.13 0.14 0.15 0.16

C Capillary (heel-stick) blood, collected in capillary tubes containing saponin

and heparin.b Duplicate (in most cases) analyses of the same samples on the same day.

CUNICAL CHEMISTRY, Vol. 30. No.8, 1984 1385

Table 2. COHb Content of Blood Samples Collected and Stored under DIfferent CoCO, mL/100 mL of blood, mean ± SD

nditlons

V.nlpuncture

Lavender Vacutalner TubeCapIllary Heel or tlngerstick

Subjects Intact Opened tubs capIllary tubeaLiterature values

(and ret)b

Healthy premature infants (n = 10) 0.57 ± 0.07 0.32 ± 0.10 0.19 ± 0.04 0.18 ± 0.03r4onsmokingadults(n = 11) 0.50 ± 0.08 0.19 ± 0.03 0.19 ± 0.03 0.17 ± 0.04Smoking adults (n = 7) 1.46 ± 0.39 1.00 ± 0.49 0.94 ± 0.49 0.91 ± 0.49

0.12(3), 0.23(4)0.18(5)0.19(6)0.85 (5)

Heel sticks for premature infants, fingersticks for adults.bReponed values (% COHb saturation) were converted to mL/100 mL of blood, with an assumed Hb concentration of 16 gIlOO mL.

smoking adult yielded about 4 nL of CO, but the sameamount of blood from smokers contained about 20 nL of CO.

To estimate the precision of the assay, we made 10repeated analyses of two normal blood samples. The meanCOHb concentrations were 0.180 (SD 0.004) and 0.208 (SD0.006) mL of CO per 100 mL of blood. Repeat analyses ofblood samples from seven normal premature infants withina single day (Table 1) shows a 10% maximum variation from

the mean, with a mean variation of 5%. We then analyzedthe same seven samples on three consecutive days and founda maximum variation from the mean of 13% (mean varia-

tion, 6%). The results of our analyses of blood samples frompremature infants and adults are listed in Table 2, which

also lists data reported by others.The sensitivity of the CC and detector system determines

the minimum amount of detectable COHb. We could detect0.1 nL of CO in 24, of blood with our system, correspondingto a saturation of about 0.005% COHb in blood (assuming 16g of Hb in 100 mL of blood).

Previously, blood samples from neonates were obtainedfrom arterial lines or by venipuncture, with the sample(100-500 4,) being transferred to a pediatric lavender-toptube (Vacutainer Tube no. 6496) containing 40 ML of liquidEDTA. The error caused by dilution can be considerablewhen only small samples of blood are obtained, but can bereduced by selecting an appropriate sample size.

In an earlier study (2) we unstoppered the tubes beforeadding the blood sample. Recently, when the tubes were leftstoppered, the COHb content of the samples (which wereabout 0.25 mL) nearly doubled. Controlled experimentssubsequently showed that the intact Vacutainer Tube con-tributes CO to the blood sample, but this problem can bealmost eliminated if the tubes are opened before the sampleis introduced (Table 2). We further found that the intactVacutainer Tube head space and liquid EDTA contained COand substantial amounts of other nonidentified compoundsthat eluted before CO (22 and 40 vs 75 s). The head space(2.46 mL of the tubes) contained 0.4-1.0 /LL of CO (two lotnos.) after being filled with CO-free gas, and the EDTAsolution (40 /LL) contained 4 nL of CO. Other workers alsohave reported on various contaminants for head-space anal-ysis in blood-sample containers (10-14).

We determined the CO content of blood stored either inintact Vacutainer Tubes or capillary tubes and found that250 /LL of blood from smoking and nonsmoking adults hadadsorbed 0.26 and 0.52 /LL, respectively, of CO from theVacutainer Tube head space, which still contained substan-tial amounts of a compound that eluted at the same time asCO, as well as earlier eluting contaminants. The COHbcontent in the blood samples does not appear to increasefurther after several hours of storage in the VacutainerTubes (Table 3). Clotting, which is a problem with thesyringe/Vacutainer sample-collection method, cannot al-ways be completely prevented; this is probably related to

Table 3. Effect of Blood Sample Storage at 4#{176}Con COHb Content

CO mLIlOO mL of blood, mean ± SD

Storsge contaIner DayO Day 7 Day 14

Intact Vacutainera 0.50 ± 0.08 0.40 ± 0.11 0.41 ± 0.03(n = 10)

Opened Vacutainera 0.19 ± 0.03 0.17 ± 0.02 0.17 ± 0.02(n = 10)

Capillary tube 0.17 ± 0.04 0.15 ± 0.02 0.15 ± 0.02(n = 10)

C Values corrected for anticoagulant dilution. No differences from day 0 were

significant by Students Mest.

difficulties involved in quickly obtaining blood from prema-ture infants.

To prevent contamination of our blood samples with CO,we use microhematocrit tubes coated with heparin andsaponin with easily removable silicone rubber closures thatdo not leach CO. Sample-tube volumes can be reducedfurther by cutting off a section of the tube. As Table 2indicates, blood stored in capillary tubes has lower COHbconcentrations than blood kept in Vacutainer Tubes. Fur-ther experiments (unpublished) indicate a stability of capil-lary tube samples of up to six months. Moreover, becausethere is no significant difference between the two methods ofdrawing blood, use of heel or fingersticks, which are lessinvasive than venipuncture, is preferable, particularly fornewborns.

The use of the capillary system, although not availablecommercially, has several advantages: ease of blood collec-tion; small sample size; no variable sample dilution byliquid anticoagulant; permanently homogeneous sample;absence of sample contamination; no coagulation; and stableCOHb content during storage.

In addition, improvements in the analytical techniquemake this assay more reliable and less time-consuming toperform.

This study was supported by an NIH grant (HD-14426), MeadJohnson Nutritional Division, and the Harrison Grant.

References1. Necheles TF, Rai US, Valaes T. The role of haemotysis inneonatal hyperbilirubinaemia as reflected in carboxyhaemoglobinlevels.Acta Paedio.tr Scand 65, 361-367 (1976).2. Ostrander CR, Cohen RS, Hopper AO, et al. Paired determina-tions of blood carboxyhemoglobin concentrations and carbon mon-oxide excretion rate in term and preterm infants. J Lab Clin Med100, 745-755 (1982).3. Collison HA, Rodkey FL, O’Neal JD. Determination of carbonmonoxide in blood by gas chromatography. Clin Chem 14, 162-171(1968).4. Coburn RF, Danielson GK, Blakemore WS, Forrester RE. Car-bon monoxide in blood: Analytical method and sources of error. JAppi Physiol 19, 510-515 (1964).

1386 CLINICAL CHEMISTRY, Vol. 30, No.8, 1984

5. Rodkey FL, Collison HA. An artifact in the analysis of oxygenat-ed blood for its low carbon monoxide content. Clin Chem 16, 896-899 (1970).

6. Coburn RF, Forster RE, Kane PB: Considerations of the physio-logical variables that determine blood carboxyhemoglobin in man.

J Clin Invest 44, 1899-1910 (1965).

7. Bullister JL, Guinasso NL, Schink DR. Dissolved hydrogen,carbon monoxide, and methane at the CEPEX site. J Geophys Res87, 2022-2034 (1982).

8. Jensen WE, O’Donnell RT, Rosenberg IH. Gaseous contaminantsin sterilized evacuated blood-collectiontubes. Clin Chem 18, 1406(1982). Letter.

9. Shang-Qiang J, Evenson MA. Effects of contaminants in blood-

CLIN. CHEM. 30/8, 1386-1388 (1984)

collection devices on measurements of therapeutic drugs. ClinChem 29, 456-461 (1983).

10. Ettre IS, Kolb B, Hurt SG. Techniques of headspace gaschromatography. Am Lab 15,76-84(1983).

11. Karlin DA, Jones RD, ODonnell RT. Comparison of siliconeand nonsilicone coated glass tubes for storage and transport ofsamples used in breath analysis. Gastroenterology 80, 1187 (1981).Abstract.12. Rodkey FL, Collison HA, Engel RR. Release of carbon monox-ide from acrylic and polycarbonate plastics. JAppiPhysiol 27,554-555 (1969).

13. Kandall S. Landaw SA, Thaler MM. Corrected carboxyhemo-globin (COHb). A sensitive index of hemolysis in jaundiced new-borns (NB). Pediatr Res 7, 356 (1973). Abstract.

Effect of Dialysis on Interference by Cefoxitin with Determination ofCreatinineMartin H. Kroll, Claire Hagengruber, and Ronald J. Elm

Interference by cefoxitin with determination of creatinine is

less with the Technicon SMAC than with other commercialanalytical systems. The SMAC assay involves a single-pointkinetic meihod with dialysis, whereas most other commercialmethods are multipoint kineticwithout dialysis.The apparent

creatinine concentration measured for aqueous solutions ofcefoxitin was 73 mmol of creatinine per mole of cefoxitin withthe SMAC, 135 mmol/mol with a manual method. Furthermore,we determined for the SMAC that the average fraction ofcreatinine dialyzed was 0.128 and for cefoxitin, 0.064. Thus,the concentration of and interference by cefoxitin in thereaction mixture for SMAC are reduced by half (i.e., theapparent creatinine concentration for cefoxitin with the man-ual system multiplied by 0.5 is essentially that noted with theSMAc: 68 vs 73). Thus we conclude that the diminished

interference be cefoxitin with determination of creatinine by

SMAC is primarily ascribable to the dialysis step.

Additional Keyphrases: antibiotics . analytical error fri.uous-iow analysis

Some of the naturally occurring compounds that cause apositive interference with the classic Jaffe technique forcreatinine are acetoacetate, oxaloacetate, acetone, and glu-cose (1,2). The cephamycin antibiotic, cefoxitin, also causesa documented positive interference with the Jaffe alkaline

picrate reaction for creatinine (3-6), cefoxitin and picrateforming a product for which the absorption spectrum issuperimposable on that for the creatinine-picrate product

(7). The degree of interference varies among instruments. Ofall the commercial methods, the s�c continuous-flowmethod shows the least interference from cefoxitin (5, 6).This method differs from most other methods for creatininein that it is a single-point kinetic method (8) and involves adialysis step. We determined the effect of the s�c dialysisstep on the magnitude of cefoxitin interference with the

Clinical Chemistry Service, Clinical Pathology Department,Clinical Center, National Institutes of Health, Building 10, Room2C407, Bethesda, MD 20205.

Received March 23, 1984; accepted May 9, 1984.

determination of creatinine and compared the results withthose of other methods, which do not involve dialysis.

Materials and Methods

Equipment

We used the following instruments: a s�c (Technioon

Instruments Corp., Tarrytown, NY 10591), a spectropho-tometer (Model 25, Beckman Instruments, Inc., Brea, CA92621), and a Cobas Bio centrifugal analyzer (Roche Analyt-

ical Instruments, Inc., Nutley, NJ 07110).

Reagents

Creatinine (Standard Reference Material No. 914, Mr 113;National Bureau of Standards, Gaithersburg, MD 20234).Dissolve in doubly distilled, de-ionized water and use within8 h.

Cefoxitin sodium, sterile (Mr 449, lot no. 0673H; Merck,Sharp and Dohme, West Point, PA 19486). Dissolve in 10mmolJL phosphate buffer and use within 8 h.

Picric acid (2,4,6-trinitrophenol, 99+%, Gold Label; Al-drich Chemical Co., Milwaukee, WI 53201). Dilute so final

concentration is 10 mmol of picrate per liter.Phosphate buffer, 10 mmol/L, pH 7.1(25 #{176}(J).Prepare from

reagent-grade NaH2PO4, N2HPO4, and doubly distilled de-ionized water.

The following three Technicon reagents were used withthe SMAC and for absorbance studies with the Beckmanspectrophotometer: the creatinine color reagent, picric acid(product no. Tol-0724), 13 g/L; the creatinine recipient

solution consisting of 1.0 mL of Brij-35 surfactant, 300 mL/Lsolution (product no. T21-0110), plus 1 L of de-ionized water;and sodium hydroxide solution, 750 mmoIfL (product no.T01-0858).

Methods

We prepared aqueous solutions of creatinine standards(0.15, 0.50, 1.0, and 1.5 mmol/L) and cefoxitin standards inphosphate buffer (1.0, 2.0, 5.0, and 10.0 mmolJL). Cefoxitinwas added to pooled serum to give final concentrations of1.0, 2.0, 4.0, and 5.57 mmol/L.