Impedimetric screening for bacteriuria

(1)

JOURNALOFCLINICAL MICROBIOLOGY, Mar.1978,p. 273-278 Vol 0095-1137/78/0007-0273$02.00/0

Impedimetric

Screening

for

Bacteriuria

P.CADY,* S. W. DUFOUR, P. LAWLESS, B.NUNKE, ANDS.J. KRAEGER

Bactomatic, Inc., Palo Alto,California94303

Receivedforpublication16May1977

A rapid, automated instrumental procedure for distinguishing urine cultures

containing greater than 105organisms per ml is described. The method is based

upon the measurement ofchanges in impedance that takeplaceasmicroorganisms

alter the chemical composition of the mnedium. The time required to detect

impedancechangeis

inversely

relatedtothe initial concentration of

microorga-nisms in thesample.By defininganimpedance-positivecultureas onethatgives

detectable impedance change within 2.6 h,95.8% of 1,133 urine cultures tested werecorrectlyclassifiedascontainingmorethanorfewer than105 organismsper ml. Selection ofa longer detection time decreases false

negative

results atthe

cost of increased false positive results. Impedance screening is compared with

screening data reported in the literatureusing adenosine-5'-triphosphate

detec-tion, microcalorimetry, electrochemical measurements, andoptical microscopy.

1. 7,No. 3 {in U.S.A.

A screeningtest forbacteriuria, ifrapid,

ac-curate, and inexpensive, could have many

ad-vantages. Itwouldspeed apreliminaryreport to

the clinicianwhocould prescribetreatment on

the basisofpositive

laboratory

evidence of in-fection.Itwould eliminatetheunrewarding plat-ing andreadingofnegative cultures thatin most

laboratoriescomprisethelarge majorityofurine

cultures. Finally, it wouldencourage screening ofasymptomatic populations thatare not

pres-ently screened foreconomic reasons.

A number of rapid methods have been de-scribed. These includemeasurementof bacterial adenosine-5'-triphosphate by luciferase (1,4, 11), production ofradioactivecarbondioxideby

me-tabolism of labeled substrates (J. P. Kilbourn and J. L. Bramhall, Abstr. Annu. Meet. Am.

Soc. Microbiol. 1974, M278, p. 112),

measure-mentof thepotential generated bygrowing

mi-croorganisms using a platinum and a calomel

electrode(7),measurementofheatgenerated by

metabolizing microorganisms using a

microca-lorimeter (2;K. A. Bettelheim, S.M.O'Farrell,

S.Al-Salihi,E. J.Shaw, andA.E.Beezer,Abstr.

2nd Int.Symp. Rapid MethodsAuto.Microbiol.

1976, p.8),measurement of theimpedance ofa

suspensionof organismsfiltered from urine and

eluted into water(12), and measurement of

par-ticle size and distribution by

electronic-pulse

height analysis (L. S. Gall and W. A. Curby,

Abstr. Annu. Meet. Am. Soc. Microbiol. 1976,

C88,p. 40). Inaddition, anumber of

investiga-torshave usedmicroscopicexamination of urine

as ascreening procedure(8-10).

Thispaper reports on the use in 1,133clinical

urine cultures of an

impedance-measuring

in-strument to detect samples containing more

than105microorganismsperml,aconcentration

of microorganisms commonly associated with

the presence ofurinarytractinfection.Itshould

be pointedoutthat Kass (6) showed that over

105 of one single type ofmicroorganism per ml

was indicative of a urinary tract infection,

whereasthe methoddescribedherescreensfor

over105total organismsperml. It isoccasionally

the case that a urineculture will contain over

105organismsbutless than105of anysingletype

perml.

The pnnciple upon which impedance

moni-toring is based (3) can be briefly summarized

here. As microorganisms grow andmetabolize,

the chemicalcomposition of thesupporting

me-dium is alteredas nutrientsare consumedand

metabolic end products are produced.

Associ-ated with thischange inchemical composition

is a corresponding change in the resistance to

the flow of an alternating current (i.e., the

impedance)when apairofelectrodesare placed in the medium. When a small concentration of

microorganismsispresent, the changein

imped-anceis notdetectable.However,iftheorganisms

areallowedto

replicate,

they will,intime, reach numbers sufficienttocause adetectable

imped-ance change. This concentration of

microorga-nismsisdesignatedthethresholdconcentration.

The threshold concentration depends, in part,

onhowdetectable

impedance

changeisdefined.

When detection is defined asthefirst noticeable

change in impedance ratio, the corresponding

threshold concentration is about 105organisms

per ml. When detection is defined as 0.8% change in impedance ratio, a larger and more clearly

273

on February 7, 2020 by guest

http://jcm.asm.org/

(2)

274 CADY ET AL.

obvious change, the corresponding threshold

concentrations for most organisms are in the

range of10'to107organismsperml.

The timerequiredtoreach thethreshold

con-centrationiscalled thedetectiontime.Itisclear

that the smaller the initial concentration, the

longerthetimerequiredtoreachthreshold. The

detection time is, thus, a function oftheinitial

concentration; i.e., as theinitial concentration is

reduced, the detection time increases. Detection

timeisalsoafunction of the generationtimeof

the population of microorganisms being

mea-sured and is prolonged when organisms are

growing with slow generation times and

shortened with more rapid growth. If a

popula-tionof organisms has similar generation times,

the detection time can beused to estimate initial

concentrations. The application of thisprinciple

toestimating theinitialconcentrations of

micro-organismsin clinicalurine samples,especially in

establishing whether a sample contains more

than or less than 105 organisms per ml, is the subject of this paper.

A preliminary report based on part of this

data has already been presented (B. Nunke, S.

W. Dufour, and S. J. Kraeger, Abstr. Annu.

Meet. Amer. Soc. Microbiol. 1975, C25, p. 31).

MATERIALS AND METHODS

Instrumentation. The impedance-measuring in-strumentused in allexperiments was a BACTOME-TER 32 microbial monitoring system (Bactomatic, Inc.,PaloAlto, Calif.), which has been described else-where (5).

Electrodes. All sample chambers used in these

experiments were in clusters of 16, comprised of 8

samplechambers and8referencechambers, knownas

printed-circuit modules (Fig. 1) and have been de-scribed elsewhere (5). In thisstudy,both gold-plated

and stainless-steel electrodes were used. After use,

modulesweredecontaminated withAmphyl (Sterling Drug), rinsedthoroughly,and soakedovernightin 7X

FIG. 1. Printed-circuit board module (cluster of impedance chambers).

detergent (Linbro). Before sterilization, the module was extensively rinsed (1 h minimum) with tap and

distilledwater. Ourexperience showed that this

pro-cedurewas required notonly toreduce any residual

phenolic compounds to nontoxic levels, but also to assure moreuniform electrode response.

Impedance measurements. Ethylene

oxide-ster-ilizedmoduleswereasepticallyfilled withmedium, 1.0 ml in all reference chambers and0.5ml inall sample

chambers. All chambers were capped, and the

elec-trodes were allowed to equilibrate atroom tempera-turefor at least 1 h before obtaining urine samples. Modulesweregenerally preparedtheprevious day.

Afterequilibration,duplicate samplesof fresh urine (0.5 ml) were pipetted into sample wells. When all

sample wells werefilled, they wererecappedandthe

module wasthen inserted into the incubator portion

of the BACTOMETER32 microbial monitoring sys-tem, which had been preheated to 35°C. A 40-mV

sinusoidalsignal of2,000 Hzwas then passed across both reference and sample wells in series, and the

impedance changewasthenautomaticallyand

contin-uously recorded by the instrument's strip chart re-corder.

Impedimetric detection time.Todistinguishthe onsetofimpedance changecausedbymicrobialgrowth

from nonmicrobial changes, such asdrift, noise, and temperature fluctuation, a detection standard was

used. An accelerating change of 0.8% in impedance had to take place before an impedance change was

ascribed tomicrobialactivity.The timeatwhich the

accelerating-impedance changereached 0.8%was

des-ignatedthedetectiontime.Sometimes,especiallywith large concentrations of microorganisms, there was a strong, immediate, nonaccelerating change in

imped-ance. If this change exceeded 10% in 1 h, it was

designatedasdetection.

Cultures. All clinical urine specimens were

ran-domlyselected and obtained from Kaiser Foundation

Hospital (Santa Clara, Calif.) and transportedtoour laboratory. Three plates from eachsamplewere

pre-pared:two dilutionswere platedontoblood agar and one dilution was plated on Levine eosin-methylene

blue agar. All plates were prepared by using a

cali-brated-loop technique. Organisms from positive cul-tures wereidentifiedby conventional microbiological procedures.

Media. All media were made from commercially prepared dehydrated bases obtained from Baltimore

Biological Laboratory (Cockeysville, Md.) according

to manufacturer's directions. These included brain heartinfusion(BHI),BHIwith 0.05%agar,Trypticase

soy broth, Columbia broth, and thioglycolate broth. Othermedia,blood culture medium withand with-out0.05% agar contained BHI base(BBL),0.5%yeast extract, 0.025%sodium polyanetholsulfonate, 0.0001% vitamin K (dissolved in 95% ethanol), and 0.005% hemin.

RESULTS

AND

DISCUSSION

The medium and the sample inoculum size

were determined by preliminary studies

per-formed withurine samplesknowntohavehigh

concentrations of organisms. Six media, BHI,

J. CLIN. MICROBIOI,.

on February 7, 2020 by guest

http://jcm.asm.org/

(3)

IMPEDIMETRIC SCREENING FOR BACTERIURIA

blood culture medium without agar, blood

cul-ture medium with agar,Trypticasesoy broth,

Columbia broth, and thioglycolate broth, were

compared with respect to their promotion of

rapid growth, impedance drift,and response size.

After62urinesampleswereanalyzed,BHIwas

chosen, not because it wasdemonstrated to be

clearly superior, but because the other media

werenobetter with respecttothe above

param-eters.

Various urine-BHI ratios were investigated,

ranging from 1% urine and 99% BHI to 100%

urine. Thepredominantly urine mixtures have

theadvantage ofmaintainingahigh

concentra-tion ofmicroorganisms and, hence, of shortening

detection times.Theyhave thedisadvantage of

introducing a large variable, the urine itself,

which can differ considerably from patient to

patient with respect to pH and ionic content.

Conversely, a small inoculum such as 10

p1

of

urine minimizes thevariabilityof the urine and

dilutes inhibitory substances, but it markedly

cutsdown the initial concentration oforganisms

in theimpedance chamber, thusprolonging

de-tection times. (Detection times with 1% urine

and 99% BHIweretypically3hlater than those

with 100% urine.) A workable compromise was

found with0.5ml of urine and0.5mlof medium.

Detection times for 1,133 urine specimens

S10'

N,

U-U 10'

z

82o

I-U z

0 2

U

z.

101

1ooL

0

* * .I

*;s

'

.

_-plotted againstinitialconcentration are shown

inFig. 2.Each urine specimen isrepresented by

asingle point on the graph, the coordinates of

whichdefine the detection time in hours and the

initial concentration in colony-forming units.

Detection times greater than 12h areshown at

the extreme right hand of the graph along a

common vertical line. The numbers next to

larger black circles indicate the number of

sam-ples having the samedetection time and initial

concentration displayed together as one large

datapoint.

The horizontal line at 105

organisms

per ml

indicates the level chosentodivide

positive

and

negative urine cultures.

Thus,

all

points

above

the linerepresentcultures definedas

positive by

platecount,and allpoints below the line

repre-sentculturesdefinedas

negative. By

this

crite-rion, 188sampleswere

positive

and945

samples

werenegative.

We maychooseatimecalledthecutofftime

thatwill dividetheculturesintothosedetected beforethecutoff(and henceare

impedance

pos-itive) and those detected after the cutoff time (and are thus impedance

negative).

One such cutofftimeis

represented by

thevertical lineat

2.6 h. All points now fall into four

quadrants.

Thepoints in theupper-leftquadrantrepresent

cultures positive both

by

plate

count and

by

,..L.. .5...

* .**A*.

*.!***^: ..*,£.-:.

It

t..

I *9

456

*81 .22

I_I_I_I__ .I I n78

1 2 3 4 5 6 7 8 9 10 11 12 OVER12

__ _-HOURS

DETECTION TIME (HOURS)

FIG. 2. Detection times of 1,133 randomly selected urinesamples plotted against the initial concentration incolony-formingunits permilliliter (by plate count). The horizontal line dividesallcultures into those with greater orfewerthan 10 organisms perml. The verticalline divides all cultures into impedancepositive

(detection time,:2.6h)orimpedancenegative (detection time, >2.6 h). Some points are larger than the others and have a number next to them. The number represents the number of data pointsoccupying the same

positiononthegraph.Allsampleswith detection timesgreater than 12 h are shown at the extreme right-hand

sideofthegraph.

is, mi. .

- t.

I .

VOL. 7,1978 275

107

I

I~~~~~~~~~~

on February 7, 2020 by guest

http://jcm.asm.org/

(4)

276 CADY ET AL.

impedance.Conversely, the points in the

lower-right quadrant represent cultures negative by

both criteria.Culturesdesignated as positive by

impedancebutnegative by platecount are rep-resented by points in the lower-left quadrant

and are false positive results. Cultures positive

by plate count but negative by impedance are

false negative results and are represented by

points intheupper-rightquadrant.

Bychangingthe cutoff time, the numbers of

false positive and false negative results can be

altered. Short cutoff times minimize false

posi-tive but increase false negative results. Long

cutoff times minimize falsenegativebut increase

falsepositiveresults.

Maximum agreement between the two

meth-ods isachievedwithacutoff time of2.6h. With

thiscriterion, 95.8% of allsamplesarecorrectly classified, with22 false positive samplesand 26

falsenegative samples.Thus, 1.9% of allsamples

are false positive and 2.3% of all samples are

false negative. These numbersare summarized

in Table 1, where the upper-left box contains

the number ofsamples found positive by both

systems; upper right, those positive by plate

count only; lowerleft, those positive by

imped-ance only; and lower right, those negative by

both systems. Also summarized in Table 1 are

the percentages of falsenegativeandfalse

posi-tivesamples (aspercentagesof thetotal number ofcultures) and positive samples missed (as a

percentage of thenumber of cultures found

pos-itivebyplate count).

Maximum agreement may not be the most

desirable criterion for selecting a cutoff time.

Current practicein many clinics is to treat all

patientsonthebasis ofhistory,symptoms, and a urinalysis, using a culture only to verify the

diagnosis. Arapidscreen would enable the

cli-nician to withhold treatment until he had

spe-cificlaboratory confirmation of bacteriuria.

In-creasing the cutoff time, thereby minimizing

false negative atthe expense ofincreasing false

positive results, will assurethat fewer cases of

bacteriuriaaremissed. The additional false

pos-TABLE 1. Numberofpositiveandnegativesamples

obtainedbyculture andbyimpedance usinga2.6-h

cutoff

Impedance" (cutoff time) Culture(organisms/ml)

+(c2.(6h) -(>2.6h)

+ (-10") 162 26

- (<10'') 22 923

J. CLIN. MICROBIOL.

itive samples so engendered would mean that

morepatientswould receive unnecessary

treat-ment. A greater number of positive samples

would alsomean morework for the

microbiolo-gist, assuming that all cultures positive by the

screenwould need furtherworkup.

Toreduce false negative results, and thereby

reduce the percentage of positive cultures

missed, onecanincrease the cutofftime; this has

the effect ofincreasingfalsepositivesamples as

well. Using 3.5 h as the cutoff time, to reduce

the falsenegativeresults and the percentage of

positive cultures missed, gives agreement of

93.3% with 14false negative and 55 false positive

samples. Thus, the percentage of all samples

that are false negative falls from 2.3 and 1.2%.

The false positive cultures rise from 1.9 to

4.9%, and the percentage of positive samples

missed drops from 13.8 to7.5% (Table 2).

Anotherway to view these percentages is in

terms of the percentage ofsamples

impedimet-rically classified positive or negative for which

the plate countyielded the sameclassification.

The 2.6-h cutoffyields 184 impedance-positive

cultures, with 162,or88%, also positive by plate

count, and 949 impedance-negative cultures,

with 923,or97.3%, also negative by plate count.

Increasing the cutoff timeto 3.5hincreases the

latter percentage to 98.5% at the expense of

decreasingthe formerto76%.

There are several likelyreasons for the false

negativeresults that occurred with the

imped-ancemethod. One sourceoffalse negative data

is urine samples that contain antimicrobial

agents. In only a few cases was information

availablethat urinesamplescamefrompatients activelybeingtreated with antibiotics. In some

of these specimens, we noted that the

impedi-metric assessment was negative, whereas the

plate count was positive. This mayresult from

the fact that agreaterdilution ofantimicrobial

agents takes place on the surface of a culture

platethan ispossibleinanimpedancechamber. Thesmallvolume ofacalibratedloopisspread

overthe entire surface of theplate,and

presum-TABLE2. Numberofpositiveandnegativesamples

obtainedbyculture andbyimpedance usinga3.5-h

cutoff

Impedance" (cutoff time) Culture(organisms/ml)

+(c3.5h) -(>3.5h)

+ (-10") 174 14

(<

10')

55 890

aNumber of false positive samples, 55/1,133 (4.9%);falsenegative samples,14/1,133(1.2%);positive samplesmissed,14/188(7.4%);agreement,1,064/1,133

(93.9%).

aNumber of false positive samples, 22/1,133

(1.9%); falsenegativesamples,26/1,133(2.3%);positive

samples missed, 26/188 (13.8%); agreement, 1,085/ 1,133(95.8%).

on February 7, 2020 by guest

http://jcm.asm.org/

(5)

VOL. 7, 1978

ably the volume into which the anti

diffuse is, by

contrast, very

large. Alt4

antibiotics may be bound to agar in

countprocedure,whereas this is not ]

the liquid medium of the impedance Some false negative cultures ma

causedby

slower-growing

organismst notreach threshold as rapidly as,fo: Escherichia coli. Itshould be noted that many organisms with generat longer than E. coli, such as staphyloc( tococci, lactobacilli,yeast, etc., wered

im,pedance positives in thisstudy-p

duetoveryhigh initial concentration the organisms encountered inthe cu]

ignated positiveby platecountappeal 3.

One should examine thedata from

of view of

work

saved. If one relies

screen and does nofurther work wit

thatarescreenedasnegative,

conside]

issaved. Thissavings ismostmarked

TABLF, 3. Distributionof organisms in

samples positive by culture(>lt0'

Organism No. of

cultures

Escherichia coli.107

Proteus mirabilis.. 13

Staphylococcus aureus 12

Klebsiella pneumoniae ... 9

Enterococcus 7

Pseudomonasaeruginosa .... 5

Enterobacter sp. 3

Streptococcus sp. 2

Lactobacilli and diphtheroids 2

Serratiamarcescens 1

Candida sp. ... ... 1

Unidentified gram-negative

rods .... 18

Unidentifiedgram-positive

cocci 4

Mixedorganisms ... 4

IMPEDIMETRIC SCREENING FOR BACTERIURIA 277

ibiotic can percentageof positive samples isgenerallysmall,

ernatively, asmight be expected in anoutpatient clinic. If

Lthe plate allimpedance negativesampleswerediscarded

possible in andonly impedance positive cultureswere

quan-chamber. titatively cultured,acutofftimeof 3.5 hwould

Ly also be yield79.8% reduction in work. The work saved

thatwould isevengreaterwiththe shortercutofftime,i.e.,

r example, 83.8%.

L,

however, In Table4theimpedancescreeniscompared

tion times with othertechniques. The dataare all castinto

Dcci,strep- thesameformasTables 1and2, andaretaken

letectedas from the cited references. The results of each

Iresumably techniquearecompared with platecountresults

s.A list of by using thesamecriterionforapositive culture

Itures des- of105organismsorgreaterperml. All techniques

rsinTable are rapid, ranging in time from afew minutes

for microscopy to 4 h for the electrochemical.

kthepoint Also, in general, the agreement obtained was

upon the highfor allmethods(Table 5). Microcalorimetry

;h cultures had the lowest agreementdue tothelarge

num-rable labor ber of false positive results; however, the very

Iwhenthe low percentage ofpositive samples missed is a

very strong feature of thismethod. Impedance

188urine hasthe bestoverallagreement(seeeither Table

/ml) 1or2) butdoesnothaveaslowapercentageof

% positive cultures missed as were seen with

mi-crocalorimetry or microscopy. The luciferase

56.9 method missed over ¼4 ofthe samples positive

6.9 by plate count, which is a severelimitation to

6.4 this procedure. More confidence canbeplaced

4.8 inresults fromthose studies having the largest

3.7 sample sizes. Thus, it is possible that further

2.7 studiesmayleadtoslightly different conclusions

1.6 with those methods where the sample size is

1.1 verysmall.

0.5 The use of

microscopy,

although

reasonably

0.5 accurateand

by

far themost

rapid,

isnot

wide-spread.Thismaybe due tothe burden itplaces

9.6 on the staff performing the work, which is

ex-acting and tedious. Adenosine-5'-triphosphate

2.1 assays require chemical manipulation of the

2.1 sample,which isabarrier formanylaboratories

TABLE4. Comparisonof the number ofpositiveandnegativesamples obtained by culture and various screening procedures

Screening procedure

Culture Electrochemical" Microcalorimetry^ Adenosine-5'-triphos-_phate" Microscopy"

+ -+ -+ _+_

+ 37 3 146 4 46 17 712 31

- 2 26 44 105 20 265 188 1633

aDatataken fromFig. 1of reference7. "See reference2.

'*Seereference 1.

"See reference 8.Onlythe results fromdefiningapositiveresult by culture as samples having>O0'organisms perml andapositiveresultby microscopyasmultiplecellsareshown.

on February 7, 2020 by guest

http://jcm.asm.org/

(6)

TABLE5. Agreementof variousscreening procedures

Screening ed No. of sam- False positive sam- False negative Positive samples Agreement

ples ples(%) samples (%) missed(%)

Electrochemical" 68 2/68(2.9) 3/68 (4.4) 3/40(7.5) 63/68 (92.6)

Microcalorimetryb 299 44/299(14.7) 4/299 (1.3) 4/150(2.7) 251/299 (84.0) Adenosine-5'-tri- 348 20/348(5.7) 17/348 (4.9) 17/63 (27.0) 311/348 (89.4)

phosphate

lucif-erase"

Microscopy" 2,564 188/2,564(7.3) 31/2,564 (1.2) 31/743(4.2) 2,345/2,564(91.5)

aDatataken fromFig.1ofreference7.

bSee reference2.

cSeereference 1.

dSee reference8.

in terms of labor cost, and, in addition, the reagentsareexpensive. The majordifficulty with

microcalorimetry is thatmostmicrocalorimeters

are single-channel instruments.

Multiple-chan-nel instruments have been made, but they are

costly and not readily available. The

electro-chemical method iseasytouseandshowsgreat

promise, but this method also appears to be

difficulttoextend(at low cost)tomonitoring of

large sample numbers. Thus, of the methods

compared above, the impedance approach

ap-pears to offer advantages in ease of use and

abilitytohandlelargenumbers ofsampleswhile

retaining a high level of agreement with the

plate counttechnique.

ACKNOWLEDGMENTS

We thankEdithSteward and themedical technologistsof Kaiser Foundation Hospital, Santa Clara, Calif., for their cheerfulcooperationandassistance inobtaining clinical

spec-imens. Thepatienttechnical assistance ofPatriciaCastillo, Alice D.Cavette,VivianGodinez,and FatimaZaman, and the dedicated secretarial skills of LuYutzy,aregratefully acknowl-edged.

LITERATURE CITED

1. Alexander, D.N.,G. M.Ederer, and J. M. Matsen. 1976. Evaluation ofanadenosine5'-triphosphateassay asascreeningmethodtodetectsignificantbacteriuria. J.Clin.Microbiol.3:42-46.

2. Beezer,A.E.,K. A.Bettelheim,R. D.Newell,and J. Stevens. 1974. The diagnosisof bacteriuria byflow

microcalorimetry, a preliminary investigation. Sci. Tools 21:13-15.

3. Cady,P. 1975.Rapid automated bacterial identification byimpedance measurement, p. 73-99. In C. G. Heden andT.Illeni(ed.), New approaches to the identification ofmicroorganisms. John Wiley & Sons, Inc., New York. 4. Chappelle,E.W.,andG. W. Picciolo.1971.Bacterial adenosinetriphosphate (ATP)as ameasureofurinary tract infection. NASA publication TSP1-10051, Na-tional Aeronautics andSpace Administration, Green-belt, Md.

5.Hadley, W. K., and G. Senyk.1975.Early detection of microbial metabolism and growth by measurement of electricalimpedance, p.12-21.In D.Schlessinger (ed.), Microbiology-1975. American Societyfor Microbiol-ogy,Washington, D. C.

6.Kass, E. H. 1957.Bacteriuria and thediagnosis of infec-tions of the urinary tract. Arch. Intern. Med. 100:709-714.

7.Lamb,V.A.,H. P.Dalton, and J. R. Wilkins.1976. Electrochemical method for theearly detection of uri-narytractinfections.Am. J.Clin. Pathol.66:91-95. 8.Lewis, J. F., and J. Alexander. 1976.Microscopy of

stained urinesmears todetermine the need for quanti-tativeculture. J.Clin. Microbiol.4:372-374.

9. Robins,D.G.,R. H.White,K.B.Rogers, and M. S. Osman.1975.Urinemicroscopyas anaidin the detec-tion of bacteriuria.Lanceti:476-478.

10. Rosenthal,S.1976.Microscopic screeningforbacteriuria. N.Y.StateJ.Med.76:209-211.

11.Thore, A.,S.Ansehn,A.Lundin,andS.Bergman. 1975. Detection of bacteriuria byluciferase assay of adenosinetriphosphate.J.Clin. Microbiol.1:1-8. 12. Wheeler,T.G.,and M. C. Goldschmidt. 1975.

Deter-mination ofbacterial cell concentrations by electrical measurements.J.Clin. Microbiol.1:25-29.

278

CADY ET AL. J. CLIN. MICROBIOL.