CLIN. CHEM. 20/8, 1062-1070 (1974)

1062 CLINICAL CHEMISTRY. Vol. 20. No.8. 1974

Chemical and Clinical Evaluation of theContinuous-flow Analyzer “SMAC”

Morton K. Schwartz, Victor G. Bethune, Martin Fleisher, Gina Pennacchia,Celia J. Menendez-Botet, and Donald Lehman1

“SMAC” (Sequential Multiple Analyzer plus Computer)is a high-speed computer-controlled multitest analyzer.A 20-channel prototype SMAC (glucose, urea nitrogen,creatinine, carbon dioxide content, total bilirubin, calci-um, phosphorus, cholesterol, iron, uric acid, chloride,sodium, potassium, total protein, albumin, creatine ki-nase, alkaline phosphatase, lactate dehydrogenase, andaspartate and alanine aminotransferases) has beenevaluated for: (a) method precision during within-dayruns and on a day-to-day basis over a period of time; (b)method linearity over a range established on a chemicalbasis and related to clinical requirements, with use ofboth aqueous standards and protein matrix referencematerial; and (c) correlation of SMAC values with thoseobtained by the methods routinely in use in our depart-ment.

AddItIonalKeyphrases: AutoAnalyzer #{149}continuous-flowanalysfs

In 1972 the Technicon Instrument Corporation in-troduced its third generation continuous-flow system,called “SMAC” (Sequential Multiple Analyzer plusComputer) (1). This instrument makes use of mini-aturized continuous-flow cartridges, new optical de-tecting systems, ion-selective electrodes, many newand more specific chemical procedures, and an in-strument-dedicated computer, as well as new andsimplified sample handling and loading techniques.It offers the clinical chemistry laboratory an opportu-nity for total laboratory automation in the sense thatthe operator’s role is merely placing samples in theinstrument. Calibration, analysis, and correction ofmachine problems, as well as calculation of resultsare done by the computer. On our prototype instru-ment it is possible to do as few as one or as many as20 different assays simultaneously at a rate of 150samples per hour. The Department of Biochemistryof Memorial Hospital for Cancer and Allied Diseaseshas been evaluating this instrument during the pasteight months. The evaluation included studies of theprecision of the methods, linearity of the procedures,and correlation of the values generated by SMACwith those obtained in our laboratory by convention-al methods. This report describes that evaluation.

Department of Biochemistry, Memorial Hospital for Cancer andAllied Diseases, New York, N. Y. 10021.

1 Technical consultant on assignment to Memorial Hospitalfrom Technicon Instrument Corp., Tarrytown, N. V. 10591.

Received May 13, 1974;. accepted May 24, 1974.

Materials and Methods

InstrumentA 20-channel SMAC was used in these studies. The

channels included glucose, urea nitrogen, creatinine,total protein, albumin, calcium, sodium, potassium,chloride, carbon dioxide content, inorganic phospho-rus, cholesterol, iron, total bilirubin, uric acid, alka-line phosphatase (EC 3.1.3.1), creatine kinase (EC2.7.3.2), lactate dehydrogenase (EC 1.1.1.27), and as-partate and alanine aminotransferases (EC 2.6.1.1and 2.6.1.2). The analytical methods have all beenpreviously described (Table 1). Because of program-ming difficulties, it was not possible to evaluate thetwo aminotransferase channels at the same time asthe other channels, and data concerning them will bepresented in a subsequent report.

ReagentsTable 1 lists the necessary reagents and the sched-

ule of their preparation. Most are supplied in a pre-packed form. An appropriate weighed portion of re-agent is accompanied by an exact volume of diluent

in a mixing bottle. Preparation of the reagent merelyinvolves adding the chemical to the solvent bottle,shaking to dissolve, and placing the bottle in theproper reagent slot at the back of the instrument.Each day it is necessary to prepare 10 reagents, thecontrol sera, and to change the cholesterol-channelreagent-pump tubes. We also found it necessary tochange the potassium electrode membrane daily.Technicon lyophilized control serum was used to cali-brate the instrument. The values (Table 2) for eachconstituent were determined in our laboratory by ourusual procedures (Table 8) and these values enteredon the magnetic-tape programs used in the operationof the machine.

Operation of MachineThe operation of SMAC is relatively easy and re-

quires two operators. The standard procedure used inour laboratory in starting the machine each morningis as follows: (a) The daily reagents are prepared andif necessary, the stable reagents in reservoir bottlesbehind the machine are replenished; (b) SMACprinter and console power switches are turned to

“on” position and the machine is allowed to “warm-up” for 30 mm; (c) switches on the operator panel for

Table 1. Reagent Preparation SchedulePreparation schedule

Method Reagent Daily Weakly BIweekly Monthly Stable

Calci um() (a) 8.Hydroxyquinoline, cresolphthalein complexone x(b) Dimethylamine, potassium cyanide x

Carbon (a) Diluent, sulfuric aciddioxide(S) (b) Carbonate buffer, phenolphthalein

Creatine (a) Creatine ptosphate, cysteine, xkinase(4) adenosine diphosphate

imidazole buffer(b) N.Ethyimaleimide x(c) Diacetyl, orcinol(d) Sodium hydroxide-EDTA x

Glucose(5) (a) Diluent, sodium chloride(b) Glucose oxidase, phosphate buffer x(c) Peroxidase, acetate buffer x(d) MBTH (3.methyl.2.benzo thiazolinone hyd razone) x(e) DMA (N,N-dimethylaniline) x(f) Working MBTH DMA x

Phosphorus(6) (a) Diluent, sulfuric acid x(b) Ammonium molybdate, sulfuric acid x

Lactate (a) NAD, aminomethylpropanol buffer, pH 9.0 xdehyd.(7) (b) Lactic acid, aminomethyipropanol buffer x

Asp. aminotr.(8) (a) Diluent, phosphate buffer x(b) Malic dehydrogenase, NADH x

L-aspartic acid, a-ketogtutaric acid, phosphate buffer

Alanine (a) Diluent, phosphate buffer xaminotr.(8) (b) Alanine, lactic dehydrogenase,

NADH, a-ketoglutaric acid, phosphate buffer x

Bilirubin(9) (a) Diluent, caffeine, sodium benzoate, sodiumacetate x

(b) Diazo reagent (sodium nitrite, sulfanilic acid) x(c) Sodium hydroxide, sodium potassium tartrate x

Iron(1O,11) (a) Diluent, hydrochloric acid, neocuproine,sodium chloride, ascorbic acid x

(b) Sodium acetate x(c) Iron color, ferrozine, sodium chloride,

hydrochloric acid xTotal protein(s) (a) Biuret reagent with potassium sodium tartrate x

(b) Blank reagent, sodium hydroxide x

AIbumin(1) (a) Bromcresol green x(b) Succinic acid x

Uric acid(13) (a) Diluent, sodium chloride x(b) Sodium tungstate x(c) Hydroxylamine x(d) Phosphoric acid x

Urea (a) Diacetyl monoxime, thiosemicarbazide-EDTA xnitrogen(14) (b) Sulfuric acid, phosphoric acid, ferric chloride x

Creatinine(15) (a) Diluent, sodium chloride X(b) Sodium hydroxide x(c) Saturated picric acid x

Cholesterol(16) (a) Diluent, water X

(b) Acetic anhydride, glacial acetic acid, sulfuric acid x

Chloride(S) (a) Diluent, nitric acid X

(b) Mercuric thiocyanate, ferric nitrate xSodium(17) (a) Tris buffer, pH 8.0, sodium chloride X

Potassium(17) (a) Tris buffer, pH 7.5, potassium chloride X

Alkaline (a) Magnesium chloride, amino methyl propanolphosphatase buffer pH 10.3, p.nitrophenol x(18) phosphate

CLINICAL CHEMISTRY Vol. 20 Nn 9 1974 10R2

Table 2. Concentration of Constituents inCalibration (Reference) Sera

Calibration Sara

Table 4. SMAC Average Within-Day Precision(Mean Value of Daily CV’S)

Concentration

Constituent

Glucose’Urea nitrogen’Creatinine’Uric acid’Albumin’Totalprotein2Sodium’Potassium’Chloride’C023Calcium’Phosphorus’lronBilirubin’Cholesterol’Alk p’ase’Lactate dehydr.’Asp. aminotr.’Creatine kinase’

24368

5.18.6

4.16.8

1475.4

1072910.35.7

1302.1

21090

370

II

1102.8

154650

Constituent

Glucose’Urea Nitrogen’Creatinine’Uric acid’Albumin’Total protein’Sodium’Potassium’Chloride’CO,’Calcium’Phosphorus’Iron4Bilirubin’Cholesterol’Alk p’ase’Lactate dehydr.’Asp. aminotr.’Creatine kinase’

Low mean

(CV, %)

83(0.71)12(0.91)1.2(7.3)

3.6(2.0)2.6(1.8)3.6(1.5)120(0.66)

2.9(2.2)84(0.84)18(1.7)6.7(1.2)2.3(2.6)50(3.9)0.21(18.6)72(2.1)26(4.1)

147(3.1)

a

Mid-value

mean (CV, %)224(0.96)52(0.67)4.4(3.2)

7.0(2.1)3.9(1.0)

5.8(1.3)135(0.77)5.1(3.7)102(0.94)26(1.5)

9.3(1.2)5.5(2.2)142(2.1)

3.08(2.5)143(1.3)141(3. 6)349(2.3)150(2.1)

200(6.00)

High mean

(CV, %)

356(0.24)85(0.86)7.0(2.3)10. 1(1. 1)5.1(0.71)7. 7(0. 63)148(0.92)7.4(3.7)116(0. 65)34(1.0)11. 9(1. 2)

8.3(2.2)

6. 95(0. 99)193(0.84)281(1. 3)501(2.0)

a

mg/dl; 2 g/dl; mmol/liter; g/dl; U/liter.

Table 3. SMAC Within-Day Precisionn = 48

mg/dl; ‘g/dl; ‘ mmol/Iiter; gfdl;’ U/liter.Data not accumulated because of lack of appropriate

material for analysis.

Glucose’Urea nitrogen’Creatinine’Uric acid’Albumin’Totalprotein’Sodium’Potassium’Chloride’CO,’Calcium’Phosphorus’Iron4Bilirubin’Cholesterol’Alk p’ase’Lactate dehydr.5Asp. aminotr.#{176}Creatinekinase5

Concentration Table 6. SMAC CarryoverConstituent

Glucose’Urea nitrogen’Creatinine’Uricacid’Albumin’Totalprotein’Sodium’Potassium’Chloride3CO,’Calcium’Phosphorus’Iron4Bilirubin’Cholesterol’Alkalinephosphatase’Lactate dehydr.’Asp. aminotr.’Creatine kinase’

(Interaction)Carryover(%)

1.34.10.752.64.43.42.53.03.57.55.82.83.03.32.03.76.0

-

6.0

and DriftDrlft(r%)

+0.66+0.86+5.99+0.58+0.41-0.28

+0.08+0.42

+0.36+0.54+1.36+0.71-1.85-5.94

+1.03+1.47+1.26-1.37

+3.00

Low mean(CV, %)84(0.60)12(0.00)1.3(7.2)3.5(0.96)2.6(1.60)3.6(1.37)118(0.67)2.9(1.22)

84(0.67)18(2.68)6.9(1.23)2.3(2.19)50(3.32)0.23(4.01)

74(2.20)

28(3.62)136(2.17)

a

109(5.45)

Mid-valuemean (CV, %)

219(0.60)49(0.42)

4.3(4.00)6.8(0.90)4.0(0.36)5.7(1.03)133(0.33)5.0(0.29)

101(0.38)28(1.63)9.3(1.03)5.7(1.88)127(1.00)3.3(1.04)132(1.47)153(1.41)335(1.49)153(1.10)196(2.70)

High mean(CV, %)

310(0.17)86(0.77)7.1(3.21)10.0(0.64)5.1(0.57)7.7(0.60)146(0.99)7.0(0.88)115(0.40)35(0.41)

11.9(0.95)8.1(1.90)241(0.52)7.02(1.18)190(0.84)297(1.90)469(1.24)

a

308(2.92)mg/dI; 2 g/dl; ‘ mmol/Iiter; ‘ ,g/dI; & U/liter.

mg/dl; ‘g/dl; ‘ mmol/liter; ,g/dl; U/liter.Data not accumulated because of lack

material for analysis.of appropriate

platens, pumps, and the wash cycle are turned to“on”; (d) the SMAC computer is activated with fourswitches on the rear of the machine; (e) two differentreconstituted reference sera (“calibration stan-dards”) (Table 2) are placed in refrigerated wells inthe sampler module; (/) on the telewriter keyboardthe code “1,0,0,0,1” and the calendar date are enteredinto the computer to place SMAC in a standby posi-

tion; and (g) sample “IDEE” (identification) datamay then be entered.

SMAC remains in a standby position until the op-erator requests automatic or semiautomatic opera-tion by depressing the appropriate switch. If the au-tomatic mode is chosen, the instrument proceeds topump first water and then reagents for 20 mm each.After this, SMAC enters a “Test and Calibrate”mode and four samples of calibration material I andthree samples of calibration material II are aspiratedand analyzed. The operator examines the printed re-

1064 CLINICAL CHEMISTRY Vol 20 Nn 5. 1974

Constituent

Concentration

Low Mid-value High

Mean (CV, %)‘ (CV, %) Mean (CV, %)‘ (CV, %)“ Mean (CV, %)“ (CV, %)‘

Glucose’ 83 (5.6) (5.8) 227 (1.9) (2.1) 327 (6.0) (6.1)Urea nitrogen’ 12 (1.3) (1.5) 49 (2.2) (2.6) 85 (0.8) (1.2)Creatinine’ 1.2 (13.3) (15.8) 4.2 (2.4) (4.5) 7.0 (2.6) (3.5)Uricacid’ 3.6 (3.2) (3.8) 6.8 (3.5) (3.4) 10.1 (1.2) (1.7)Albumin’ 2.6 (1.5) (1.8) 3.9 (1.3) (2.0) 5.1 (0.0) (0.7)Totalprotein’ 3.6 (1.1) (1.8) 5.7 (1.8) (1.8) 7.7 (1.1) (1.0)Sodium’ 120 (0.6) (0.8) 134 (1.7) (1.8) 148 (0.9) (1.3)

Potassium’ 2.9 (5.0) (5.3) 4.9 (6.9) (8.2) 6.9 (5.7) (7.7)Chloride’ 84 (1.0) (1.1) 101 (1.1) (1.3) 116 (0.8) (0.9)CO,’ 18 (4.8) (4.8) 27 (5.6) (5.8) 34 (2.6) (2.6)Calcium’ 6.7 (2.0) (2.2) 9.2 (1.5) (2.0) 11.9 (0.6) (1.4)Phosphorus’ 2.3 (5.7) (6.1) 5.4 (3.7) (3.6) 8.3 (3.5) (4.1)Iron4 50 (2.6) (4.7) 99 (11.2) (11.4) C

Bilirubin’ 0.20 (36.5) (41.4) 3.4 (4.9) (5.5) 6.9 (1.4) (1.8)Cholesterol’ 72 (3.5) (4.2) 136 (2.9) (3.2) 192 (1.7) (2.0)Alk p’ase’ 26 (10.3) (11.6) 161 (8.0) (8.8) 281 (5.7) (6.0)Lactate dehydr.’ 147 (4.8) (5.9) 348 (3.8) (5.0) 501 (5.4) (5.8)Asp. aminotr.’ 150 - (2.1) “ - -

Creatine kinase’ ‘ - - 200 - (6.0) ‘ - -

‘ mg/dl; 2 g/dI ‘ mmol/liter; ,,g/dl; ‘ U/liter.a Calculation from mean of means.b Calculated by variance equation (see text).C Data not accumulated because of lack of appropriate material for analysis.

port of these data, which include the photoelectric high-value materials were used in appropriate ratioscell voltage and the absorbancy of each channel for for the linearity studies. Cumulative (day-to-day)the last two aspirated specimens of each of the cali- precision was evaluated by determining the coeffi-bration materials, as well as those of the water and cients of variation of the individual means of thereagent baseline. If he thinks that the data are ac- daily 48 specimen runs and by a statistical methodceptable, he depresses a switch marked “request cali- for determining a variance when sets of determina-bration.” The instrument recalibrates itself and be- tions are considered (25). In this method, the fol-gins the aspiration of the “IDEE” identified samples. lowing equation is used:Every 48 specimens (20 mm) the instrument aspi-rates reference material and recalibrates itself.

Data Output and Statistical Evaluation

SMAC prints the values of each of the constituentsfor each specimen on a chart. During our evaluation,data were also entered on a teletype-generated papertape and transmitted via a dataphone to a time-shared computer (Time Shared Resources, Flushing,N. Y.) where the means, standard deviations, and

cv(%) = M0 tat

J(xi - )2 + (x2 - X2)2 . . (xk - Xk)2

n1 + n . . . + k k

Where “Mtti” is the mean of all the values, repre-sents the mean of each set, x, the individual values ofeach set, n the number of values in each set, and kthe total number of sets included in the statistical

coefficients of variation for the precision data and re- treatment (degrees of freedom).

gression equations by the method of least squares forthe linearity and correlation studies were calculated

Drift was defined as the percent difference be-tween values of the first and 48th sample in the daily

on an IBM 360/50 computer. For the generation ofthe precision data, single preparations of lyophilizedcontrol serum (Versatol Automated, High and Low;

precision experiments. Carryover was determined bysequential analysis of three high- and three low-con-centration specimens and then use of the equation:

General Diagnostics, Morris Plains, N. J. 07950) wereused. The mid-value material was a mixture of equalamounts of the two sera. Each day sufficient materialwas reconstituted according to manufacturer’s in- where L1 and L3 are the first and third low-concen-structions to permit a series of 48 successive tubes of tration specimens and H3 the third high-concentra-the same concentration to be analyzed. The low- and tion specimen (26).

Table 5. SMAC Day-to-Day Precision

MIrAI f’WAIQTDV %!.I ‘30 .I. 0 107.1

%Interaction = L1 - L3 x iooH3 - L3

Table 7. Linearity of SMAC ProceduresDeviations from linearity at two highest

concentrationsRegression Correlation Concentration Concentration

Constituent equation coefficient deviation (%) deviation (%)Total protein’ y=l.O3x-O.20 0.909 8.7(+0.3) 11.3(+2.1)

Creatinine(prot)’ y=O.97x+0.O8 0.999 7.9(-1.2) 10.3(-1.7)Creatinine(aq) y=0.99x-0.10 0.999 7.4(-0.6) 9.6(-2.4)

Uricacid(prot)’ y=0.99x+0.01 0.999 8.4(+0.2) 10.0(+0.1)Uricacid(aq) y=0.88x+0.43 0.987 9.8(-9.6) 11.4(-22.7)

Urea nitrogen (prot)’ y=1.OOx-1.25 0.999 66(+1.6) 85(+2.6)Urea nitrogen (aq) y = 0.99x -4.17 0.999 106(+0.6) 140(+0.9)

Albumin(prot)’ y=0.95x+0.02 0.997 4.2(+1.7) 4.7(1.0)

Sodium(prot)’ y=1.04x-4.03 0.993 143(+0.2) 151(+0.2)Sodium(aq) y=0.96x-2.67 0.999 135(+0.2) 147(+0.1)

Potassium (prot)’ y = 0.94x + 0.35 0.989 6.2(+1.4) 7.4(+4.7)Potassium(aq) y=1.OOx-0.27 0.994 6.1(+3.0) 7.4(+5.1)

Chloride (prot)’ y = 1.02x - 1.88 0.997 108(+0.1) 116(-0.6)Chloride (aq) y = 0.98x + 4.98 0.994 115(-1.1) 129(-1.4)

CO, (prot)’ y = 1.05x -0.74 0.997 31(+1.1) 35(+3.4)CO,(aq) y=0.94x-2.7 0.970 30(-5.6) 36(-12.7)

Calcium (prot)’ y = 1.02x -0.33 0.993 10.3(+0.5) 11.8(+2.3)Calcium(aq) y=0.96x-0.61 0.997 fl.3(+1.4) 13.8(+2.7)

Phosphorus (prot)’ y = 1.03x -0.26 0.997 6.8(+1.1) 8.3(-0.3)Phosphorus(aq) y=O.99x+O.22 0.999 8.2(+0.4) 10.4(-0.7)

Bilirubin(prot)’ y=l.OOx-O.Ol 0.999 8.4(+0.6) 10.7(+0.6)

Glucose (prot)’ y = 0.98x + 6.1 0.999 287(+2.4) 353(+1.8)Glucose(aq) y=l.O3x-3.3 0.999 433(+1.9) 574(+2.2)Cholesterol (prot)’ y = 1.OOx -0.80 0.999 456(+0.7) 587(+0.1)

Lactate dehydr. (prot)’ y = 1.llx - 16.0 0.998 503(+7.0) 653(+7.1)

Creatine kinase (prot)’ y = 0.96x+ 7.0 0.996 544(0.0) 658(-8.O)

Alk p’ase (prot)’ y = 1.OOz-2.0 0.999 336(+1.4) 429(+0.3)

Iron (prot)4 y = 0.98x + 3.0 0.999 337(+1.3) 420(-1.6)

mg/dl; ‘g/dl; mmol/liter; ,g/dl; U/liter.

Table 8. Correlation of SMAC Procedures with Those We Routinely UsedMean values’

RegressionNo. Corral, equation Correlation

Constituent (ref.) Correlation method spec. method SMAC (y = SMAC) coefficientAlbumin(12) Manual BCG 120 3.2 3.1 y = 1.03x - 0.2 0.977Albumin(19) SMA12/6OHABA 149 3.6 4.0 y=0.85x+0.94 0.946

Total protein(s) SMA12/60biuret 179 6.7 6.5 y=l.06x-0.6 0.940Cholesterol(20) AAII Isopropanol extraction 152 233 233 y = 0.95x + 12 0.977Glucose(21) AAI-Ferricyanide 124 123 133 y1.08x+0.16 0.992Phosphorus(22) SMA 12/60-Molybdate SnCI, 188 3.7 3.7 y = 1.03x - 0.10 0.973Creatinine(15) AAI-Picrate 148 2.1 2.2 y=l.08x-O.lO 0.995Uric acid(13) SMA 12/60-Phosphate. 185 5.3 5.1 y = 0.932x + 0.2 0.983

tungstate reductionChloride(S) AAI-Hg(NO,), 261 99 100 y = 0.88x + 12.5 0.923Carbon dioxide(3) AAI-Phenolphthalein 197 28 27 y = 1.Olx - 1.8 0.956

bicarbonate bufferBilirubin(9) SMA12/600iazo 75 0.9 0.7 y=0.99x-0.2 0.930Sodium(3) IL 143 Flame Photometer 144 139 140 y = 0.85x + 21 0.933Potassium(28) IL 143 Flame Photometer 144 4.3 4.3 y = 0.89x + 0.49 0.988Lactate dehydr.(24) SMA 12/60 Colorimetric 185 242 279 y = 1.lOx + 7.5 0.969Alkp’ase(18) SMA12/60p.nitrophenol 113 94 149 y=1.33x+24 0.967

phosphateCreatine kinase(4) AAI Creatine phosphate 234 56 63 y = 1.Olx + 6 0.970Iron(10) AAI Tripyridyl triazine 101 88 80 y = 1.09x- 16 0.976Urea nitrogen(14) SMA 12/60 Diacetyl monoxime 123 21 21 y = 1.OOx+ 0.1 0.998Calcium(2) SMA12/60Cresolphthalein 101 9.9 10.0 y=O.99x+O.6 0.893

complexoneThe units are thosedescribed for each constituent in Table 7.

1066 CLINICAl CHMITRY Vol fl No 9 1974

S..

0E

S

130 220 310Auto Analyzer K Extraction mg /dl

Fig. 2. Correlation of SMAC cholesterolmethod with isopro-panol extraction AutoAnalyzer Il method

Results

Precision

197(P<0.6)

Mean (com-parison toSMAC)

Correlation-regressionequation(xSMAC)

Correlationcoefficient

194(P<0.2)

y=0.87x+24.3 y=0.86x+23.7

0.994 0.998

GLUCOSE

N

E2

:3

I1

Fig. 1. Linearity of glucose

Theoretical value mg/dI

Broken line, aqueous solution: unbroken line,protein SOlUtIOn

CHOLESTEROL

CLINICAL CHEMISTRY, Vol. 20, No.8,1974 1067

Table 9. 25-Specimen Comparison of Results forGlucose by Three Methods

o-Toluldlne Hexokinas. SMAC

mg/dl

87 79 7391 96 8290 95 8391 91 8073 76 64

112 126 105114 111 102110 110 10299 111 101

103 108 101147 135 138162 159 151204 196 196204 194 196157 158 150227 213 225237 223 231255 235 245295 280 293207 197 212363 355 400265 341 376

365 378 418474 472 509313 332 366

The precision of a single day’s run of 48 replicatesamples in each category is shown in Table 3. Themean values and the coefficients of variation are list-

ed at three concentrations for each constituent ex-199 cept aspartate aminotransferase, for which data were

only collected at an activity of 150 U/liter. Table 4lists the averages of the coefficients of variation of

the total number of “48-specimen” precision runs.

- Except for aininotransf erase and creatine kinase,“mid-value concentration” material was analyzed on

- each of 14-28 runs over a period of 15 days and thelow and high concentration material on each of about10 days. The data for aspartate aminotransferaserepresent three days, and for the creatine kinase,three runs on two different days. The difference inthe number of runs included in these studies is due toan operational rule established during the evaluation,that if, during a run, there was a problem (reagent,instrument, manifold, etc.) with a channel only thatchannel would be de-activated and the others wouldcontinue to operate. In addition, there were certaindays when, by design, only the “mid-value concentra-tion” material was analyzed. The average coefficientof variation of the mid-value concentration speci-mens for the “within-day” runs was 2.5% or less forall constituents except creatinine, which was 3.2%(SD, 0.14 mg/dl). In specimens with low concentra-tionsthe precisioncoefficientswere somewhat higher

and were greater than 2.5% for creatinine (SD, 0.08mg/dl); inorganic phosphorus (SD, 0.06 mg/dl); iron(SD, 1.9 g/dl); bilirubin (SD, 0.04 mg/dl); alkalinephosphatase (SD, 1.0 U/liter); and lactate dehydroge-nase (SD, 4.6 U/liter). As isto be expected, the coeffi-cient of variation of the day-to-day precision valuesare higher.

0.960.673.22.11.01.30.773.70.941.51.22.22.12.51.33.62.32.1

2.12.64.53.42.01.81.88.21.35.8

2.03.6

11.45.5

3.28.85.02.1

Table 10. Comparison of SMAC Precision (Coefficients ofThat Reported in Other

Variation-Mid-ValueStudies

Concentration) with

1968CAP 33SMAC Survey

refereeConstituent WithIn-day Day-to-day labs (30)

Universityhospitals

(31)

Medically State ofsignificant the art

(29) (29)

Glucose’Urea nitrogen’Creatinine’Uric acid’Albumin’Totalprotein2Sodium’Potassium’Chloride’C02Calcium’Phosphorus’Iron4BiIiru bin’CholesteroltAlk p’ase’Lactate dehydr.’Asp. aminotr.’Creatine kinase’ - - - -

mg/dl; ‘g/dl; ‘mmol/Iiter; ,g/dI; 6 U/liter. First two columns of data are mid.value concentrations; othersarecoefficientsof variation, in percent.

6.517.8

7.310.55-32.13.63.12.0

19.812.5

10.45.1

3.15.8

7.84.8

2.61.12.51.04.32.34.9

7.65.110.6

11.8

5.07.4

8.37.14.31.58.32.2

2.35.6

20.08.0

5.38.3

5.88.03.91.83.72.1

2.88.4

23.79.1

6.0 6.0

GLUCOSE

320 -

240 -

160

60

N0’6(.)

(‘I

0’6

PHOSPHORUS

2

9

6

3

Co-I I

Table 5 lists the day-to-day (cumulative) precision

data. The coefficients of variation are higher thanthose of the within-day runs and in the case of potas-sium, carbon dioxide content, alkaline phosphatase,and iron, much higher than the generally acceptedlimits for these constituents.However, it must be

kept in mind that these data were generated during a

period when minor changes in reagents,methodolo-

gy, and computer programming for these procedures

1068 CLINICAL CHEMISTRY. Vol. 20. No. 8. 1974

Co - - 240 320

Auto Analyzer terricyonide mq/dI

Fig. 3. Correlation of SMAC glucose method with ferricyanideAutoAnalyzer I method

3 6 9 12

SMA 12iOmg/dt

Fig. 4. Correlation of SMAC inorganic phosphorus methodwith SMA 12/60 method

were being made. A true picture of the overall day-to-day precision is not yet available and can only beobtained when the final production model of the in-strument is permitted to operate over a long periodof time without modification in instrument or tech-nique. These studies are now underway.

Carryover (Interaction) and Drift

Table 6 gives information on carryover (% interac-

96-

72 -

. 46L)

U)

21

400 -

300-

..U) I

100- #{149}

I I0 24 48 72

SMA (Colorimetric) (*10’) U/L

Fig. 5. Correlation of SMAC lactate dehydrogenase methodwith SMA 12/60 colorimetric method

100 200 300 400

Auto Analyzer U/L

Fig. 6. Correlation SMAC creatine phosphokinase methodwith AutoAnalyzer I creatine phosphokinase procedure

LDH CPK

CLINICAL CHEMISTRY. Vol. 20. No. 8. 1974 1069

tion) and linear drift of the individual channels. It is

possible to program the computer to correct for bothinteraction and drift. Because the aim of our study

was to determine these factors, computer correctionwas not made for any data included in the evaluation.

Linearity

Table 7 givesthe regressionequations and correla-

tion coefficientsfor the linearityof five equidistant

concentrations of each of the constituents in a pro-tein matrix and, where appropriate, in an aqueous so-lution. Each of the values used in the calculationsrepresents the average of eight analyses.The devia-

tions from linearity were calculated by preparing aregression equation of the lowest three concentra-tions and then calculating the deviation of the aver-age analytical value of the 4th and 5th concentrationfrom the predicted concentration calculated from thethree-point linear regression curve. The curves foraqueous uric acid and carbon dioxide content are

much less linear than those obtained with these con-stituents in a protein solution. These differences arepresumably attributable to the effects of dialysis onthe linearity. Figure 1 shows typical linearity for glu-cose.

Correlation

Table 8 lists data on the correlation of results bythe SMAC methods with those obtained by proce-dures routinely used in our laboratory. Althoughsome of the methods we used might be consideredreference procedures, comparison of the SMACmethod with “reference methods” will be the subjectof individual future communications. The correlationdata for cholesterol, glucose, phosphorus, lactate de-hydrogenase, and creatine kinase are also shown inFigures 2 to 6. In 25 specimens glucose was deter-

mined on SMAC and by manual o-toluidine (27) andhexokinase (28) procedures. These data, which indi-cate no significant differences attributable to meth-od, are listed in Table 9.

DiscussionThe SMAC system has been introduced as a major

advance in laboratory automation. The attractive-ness of the instrument we evaluated is that it can an-alyze for one or as many as 20 components at a rate of150 specimens per hour, a sample volume of less than0.6 ml, consumption of of the reagent volume re-quired in SMA 12/60 continuous-flow systems, withtwo operators to perform assays now requiring, in ourlaboratory, nine technologists. These advantages canonly be acceptable if the SMAC precision and accura-cy are at least the equal of that available with otherpresently available and acceptable systems.

Our initial evaluation of a prototype instrumenthas indicated that within-day precisionisas good as

that reported for other automated systems and muchbetter than the state-of-the-art variability, medicallysignificantprecision, or intralaboratory variability

reported in surveys of refereeor universityhospitals(Table 10). Although our studies were not designedto evaluate day-to-day precision,itappears that for

all channels except potassium, carbon dioxide con-tent, iron, and alkaline phosphatase this precision isacceptable. Changes (new membranes, reagents, hy-draulics, computer software, etc.) are and have beenmade for these unacceptable channels and their day-to-day precision is expected to improve. The degreeof correlation with methods used in our laboratoryindicate that the system will provide values thatcompare favorably with those achieved with our ac-ceptable routine procedures. Further evaluation is re-quired to see how well results from SMAC correlate

1070 CLINICAL CHEMISTRY. Vol. 20, No. 8. 1974

with those for “reference” methods. SMAC has suffi-cient flexibility to permit evolution and developmentof methods and techniques so that as there is user fa-miliarity with the system, modifications will be madeto achieve all of the promised advantages.

We wish to express our appreciation to Mr. Henry Lee of theTechniconCorporation, who wrote the computer programs used inthe calculation of the data, and to the technologists of the Depart-ment of Biochemistry.

References1. Isreeli, J., and Smythe, W., SMAC: The third generation se-quential multiple analyzer. In Advances in Automated Analysis,Technicon International Congress 1972, 1, Mediad Inc., Tarry-town, N. Y., 1973, pp 13-18.2. Gitebnan, H. J., An improved automated procedure for the de-termination of calcium in biological specimens. Anal. Biochem. 18,521 (1967).3. Skeggs, L. T., and Hochstrasaer, H., Multiple automated se-quential analysis. Clin. Chem. 10,918(1964).4. Siegel, A. L., and Cohen, P. S., An automated determination ofcreatine phosphokinase. In Automation in Analytical Chemistry,Technicon Symposium 1966, 1, N. B. Scova et al., Eds., MediadInc., White Plains, N. Y., 1967, pp 474-476.5. Gochman, N., and Schmitz, J. M., Application of a new perox-ide indicator reaction to the specific, automated determination ofglucose with glucose oxidase. Clin. Chem. 18,943 (1972).

6. Amador, E., and Urban, J., Simplified serum phosphorus analy-ses by continuous-flow spectrophotometry. Clin. Chem. 18, 601(1972).

7. Morgenstern, S., Flor, R., Kessler, G., and Klein, B., Automated

determination of NAD-coupled enzymes, determination of lacticdehydrogenase. Anal. Biochem. 13, 149 (1965).

8. Kessler, G., Rush, R. L., Leon, L., Delea, A., and Cupiola, R.,Automated 340 nm measurement of SGOT, SGPT and LDH. InAdvances in Automated Analysis, Technicon International Con.gress 1970, 1, Thurman Assoc.,Miami, Fla., 1971, pp 67-74.

9. Gambino, S. R., Automatisehe Bestimmung von direktern undGesamt-Bilirubin. In Automation in der Anaiytischen Chemie,1964 Internationales Technicon Symposium. TechniconGmbH,Frankfurt, 1965,p 437. (available as reprint No. 54, TechniconCorp., Tarrytown, N.Y. 10591).

10. Giovanniello, T. J., DiBenedetto, G., Palmer, D. N., and Pet-ters, T., Jr., Fully automated method for the determination ofserum iron and total iron-building capacity. In Automation in An-alytical Chemistry, Technicon Symposium 1967, 1, Mediad Inc.,White Plains,N.Y.,1968, pp 185-188.

11. Stookey, L. L., Ferrozine-A new spectrophotometric reagentfor iron, Anal. Chem. 42, 779 (1970).

12. Doumas, B. T., Watson, W., and Biggs, H. G., Albumin stan-dards and the measurement of serum albumin with bromcresolgreen. Clin. Chim. Acta 31, 87 (1971).

13. Musser, A. W., and Ortigonza, C., Automated determination ofuric acid by the hydroxylamine method.Tech. Bull. Regist. Med.Technol. 36, 21(1966).

14. Marsh, W. H., Fingerhut, B., and Miller, H., Automated andmanualdirectmethodsfor the determination of blood urea.Clin.Chem. 11, 624 (1965).15. Chasson, A. L., Grady, H. T., and Stanley, M. A., Determina.tion of creatinine by means of automatic chemical analysis. Amer.J. Clin. Pat hol. 35,83 (1961).

16. Levine, J., Morgenstern, S., and Vlastelica, D., A direct Lieber-mann-Burchard method for serum cholesterol. In Automation inAnalytical Chemistry, Technicon Symposium 1967, 1, MediadInc., White Plains, N. Y., 1968, pp 25-28.17. Rao, J., Pelavin, M. H., and Morgenstern, S., SMAC: Highspeed, continuous flow, ion-selective electrode for sodium and po-tassium: Theory and design.In Advances in Automated Analysis,Technicon International Congress 1972, 1, Mediad Inc., Tarry.town, N. Y., 1973, pp 33-36.

18. Morgenstern, S., Kessler, G., Auerbach, J., Flor, R., and Klein,B., An automated p-nitrophenylphosphate serum alkaline phos-phatase procedure for the AutoAnalyzer. Clin. Chem. 11, 876(1965).

19. Ness, A. J., Dickerson, H. C., and Pastewka, J. V., The deter-mination of human serum albumin by its specific binding of theanion dye, 2-(4’-hydroxybenzeneazo)-benzoic acid. Clin. C/tim.Acta 12,532 (1965).

20. Fleisher, M., Bethune, V., Pennacchia, G., and Schwartz, M.K., Evaluation of automated cholesterol and triglycerides method-ology. Clin. Chem. 18,693 (1972).21. Hoffman, W. S., Rapid photoelectric method for determina.tion of glucose in blood and urine, J. Biol. Chem. 120,51(1937).22. Hurst, R. 0., The determination of nucleotide phosphoruswith a stannous chloride-hydrazine sulphate reagent. Can. J. Bio-chem. 62, 287 (1964).23. Pennacchia, G., Bethune, V. G., Fleisher, M., and Schwartz, M.K., Automated flame photometry of serum sodium and potassium.Clin. Chem. 17, 339 (1971).24. Hochella, N. J., and Weinhouse, S., Automated assay of lacticdehydrogenase in urine. Anal. Biochem. 13, 322 (1965).

25 Dixon, W. J., and Massey, F. J., Jr., Introduction to StatisticalAnalysis, 3rd ed., McGraw-Hill, New York, N. Y., 1969, pp 112-113.26. Young, D. S., and Gochman, N., Methods for assuring qualitydata from continuous-flow analyzers, In Standard Methods ofClinical Chemistry, 7, Academic Press, New York, N. Y., 1972, pp293-304.27. Hyvarinen, A., and Nikkila, E. A., Specific determination ofblood glucose with o-toluidine. Clin. Chim. Acta 7, 140(1963).

28. Stein, M. W., D-Glucose: Determination with hexokinase andglucose-6-phosphate dehydrogenase. In Methods of EnzymaticAnalysis, H. V. Bergmeyer, Ed., Academic Press, New York, N. Y.,1963,p 117.

29. Barnett, R. N., Medical significance of laboratory results.Amer. J. Clin. Pathol. 50, 671 (1968).

30. Copeland, B. E., and Skendzel, L. P., and Barnett, R. N., In.terlaboratory comparison of university hospital referee laboratoryand community hospital laboratories, using results of the 1968 Col-lege of American Pathologists clinical chemistry survey. Amer. J.Clin. Pathol. 58, 281 (1972).31. Straumfjord, J. V., Jr., and Copeland, B. E., Clinical chemistryquality control values in thirty-three university medical schoolhospitals: Preliminary report. Amer. J. Clin. Pathol. 44, 252(1965).