JOURNAL OF CLINICAL MICROBIOLOGY, June 1988, p. 1079-10840095-1137/88/061079-06$02.00/0Copyright © 1988, American Society for Microbiology

Rapid and Overnight Microdilution Antibiotic Susceptibility Testingwith the Sensititre Breakpoint Autoreader System

FREDERICK S. NOLTE,* KAREN K. KRISHER, LUISA A. BELTRAN, NANCY P. CHRISTIANSON, ANDGAIL E. SHERIDAN

Department of Microbiology and Immunology, University of Rochester Medical Center, Rochester, New York 14642

Received 12 November 1987/Accepted 22 February 1988

The Sensititre Breakpoint Autoreader system (SBAS) is a broth microdilution method with one to threeconcentrations of each antibiotic and innovative fluorescence technology to define inhibitory endpoints. Wetested 248 gram-negative bacilli and 80 gram-positive cocci using both the rapid (5 h) and overnight (18 h)SBAS procedures. Inhibitory endpoints were also determined by visual inspection of the microdilution traysafter 18 h of incubation. SBAS results were compared with those obtained by a standardized disk diffusion(SDD) procedure. Agreement between the rapid SBAS and SDD results for all antibiotic-organism combina-tions was found in 3,730 of 4,571 (81.6%) tests, with 3.9% very major, 6.5% major, and 7.9% minordiscrepancies noted. Data analysis by organism group revealed 86.8, 57.3, 71.4, and 62.3% agreement formembers of the family Enterobacteriaceae, Pseudomonas spp., staphylococci, and enterococci, respectively.The results of the overnight SBAS and SDD agreed in 4,154 of 4,654 (89.2%) tests, with 2.3% very major,1.3% major, and 7.1% minor discrepancies recorded. Concordance was noted in 90.4, 78.1, 90.6, and 83.3%of the comparisons for the members of the Enterobacteriaceae, Pseudomonas spp., staphylococci, andenterococci, respectively. The inhibitory endpoints determined with the Autoreader were as reliable as thosedetermined by visual inspection after 18 h of incubation.

The endpoints of antibiotic susceptibility tests have beentraditionally determined by direct visual recognition or tur-bidimetric measurements of bacterial growth. The SensititreAutoreader system (Sensititre, Inc., North Andover, Mass.)is unique in that bacterial growth is monitored by measure-ment of fluorescence produced by the action of bacterialenzymes on fluorogenic substrates present in the brothmedium. This principle provides a sensitive means of mea-suring bacterial growth that permits both full-range MIC andbreakpoint susceptibility testing in as little time as 5 h. TheSensititre Breakpoint Autoreader system (SBAS) is a brothdilution method that employs from one to three concentra-tions of each antibiotic and innovative fluorescence technol-ogy to define inhibitory endpoints. The pattern of fluores-cence detected by a multichannel fluorometer is analyzed bya microcomputer after either 5 or 18 h of incubation, and a

category result of susceptible, intermediate (moderatelysusceptible), or resistant is assigned for each antibiotic.

Staneck et al. (8) demonstrated very good correlationbetween full-range MICs determined after overnight incuba-tion with the Autoreader and MICs determined by visualinspection of the Sensititre microdilution plates. In anotherstudy, concordance between the results of visually readSensititre breakpoint panels and the disk diffusion procedurewas observed for 88.1% of the comparisons (2). Recently,Doern et al. (3) compared same-day and overnight SensititreAutoreader breakpoint results with those obtained with theSensititre manual overnight tests. They found good correla-tion (94.4%) between the Autoreader and the visually deter-mined endpoints for the members of the family Enterobac-teriaceae after 5 h of incubation and for both Pseudomonasaeruginosa (92.2%) and the members of the Enterobac-teriaceae (97.2%) after 18 h of incubation.Staneck et al. (9) compared Autoreader-determined full-

range MICs after 5 h of incubation with manually read MICs

* Corresponding author.

after 18 h of incubation and found good correlation betweenthe two methods for Staphylococcus aureus and members ofthe family Enterobacteriaceae.

In this study, the antibiotic susceptibilities of members ofthe family Enterobacteriaceae, Pseudomonas spp., staph-ylococci, and enterococci were assessed by both rapid (5 h)and overnight (18 h) SBAS procedures. Endpoints were alsodetermined by visual inspection of microdilution plates afterovernight incubation. Results obtained with the SBAS werecompared with those obtained by a standardized disk diffu-sion (SDD) procedure.

(This material was presented in part at the 87th AnnualMeeting of the American Society for Microbiology [K.Krisher, L. Beltran, N. Christianson, G. Sheridan, and F.Nolte, Abstr. Annu. Meet. Am. Soc. Microbiol. 1987, C60,p. 333].)

MATERIALS AND METHODS

Organisms. A total of 328 organisms including stockstrains and recent clinical isolates were examined. Many ofthe stock strains were obtained from C. Thornsberry, Cen-ters for Disease Control, Atlanta, Ga., and were usedpreviously to challenge the MS-2 (Abbott Diagnostics, Irv-ing, Tex.) system (11). The study organisms included wereboth susceptible and multiply resistant strains, and theresistance patterns are shown in Table 1. Vancomycin wasthe only antibiotic against which no resistant strains weretested.The 200 enteric bacilli selected for study included 71

strains of Escherichia coli, 37 Proteus spp., 8 Providenciaspp., 3 Morganella morganii, 23 Enterobacter spp., 22Klebsiella spp., 14 Serratia marcescens, 10 Citrobacterspp., 5 Salmonella spp., 3 Shigella spp., and 1 each ofHafnia alvei, CDC enteric group 17, Aeromonas hydrophila,and a lactose fermenter not further identified. We tested 48strains of Pseudomonas: 44 P. aeruginosa, 2 P. cepacia, and2 P. maltophilia. The gram-positive cocci included in the

1079

Vol. 26, No. 6

on August 22, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

1080 NOLTE ET AL.

TABLE 1. Resistance patterns of study organisms as determined by the SDD method

% Resistant organisms"Antibiotic Enterobacteriaceae Pseudomonas spp. Staphylococci Enterococci

(n = 200) (n = 48) (n = 61) (n = 19)

Amikacin 0 (0.5) 16.7 (6.3) NTb NTAmpicillin 46.5 (2.0) NE` 88.5 (0) 0 (47.4)Amoxicillin-clavulanic acid NT NT 13.1 (9.8) NECarbenicillin 27.5 (2.5) 45.8 (8.3) NT NTCefamandole 9.0 (2.5) NE NT NTCefazolin NT NT 11.5 (1.6) NECefotaxmine 0.5 (4.0) 41.7 (35.4) NT NTCefoxitin 13.0 (3.5) NE NT NTCephalothin 23.0 (7.0) NE 23.0 (0) NEChloramphenicol 10.0 (6.0) 75.0 (18.8) 18.0 (1.6) NEClindamycin NT NT 37.7 (0) NEErythromycin NT NT 52.4 (0) 10.5 (21.1)Gentamicin 4.5 (0.5) 45.8 (10.4) 16.4 (0) NEMoxalactam 0.5 (0) 35.4 (43.8) NT NTNalidixic acid 3.0 (2.0) NE NT NTNitrofurantoin 21.0 (8.0) NE 0 (0) 0 (0)Oxacillin NT NT 44.3 (23.3) NEPenicillin NT NT 88.5 (0) 0 (36.8)Piperacillin 7.5 (8.0) 10.4 (6.3) NT NTSulfamethoxazole 18.0 (1.5) NE NT NTSulfisoxazole NT NT 6.6 (0) NETetracycline 29.5 (6.5) NE 23.0 (0) 73.7 (0)Tobramycin 3.5 (1.5) 29.2 (2.1) NT NTTMP-SMZ 3 (2.0) NE 6.6 (0) NEVancomycin NT NT 0 (0) 0 (0)

" Number in parentheses shows percent moderately susceptible or intermediate.b NT, Not tested.C NE, Not evaluated.

study were 39 strains of Staphylococcus aureus (17 oxacillinresistant), 22 coagulase-negative staphylococci (10 oxacillinresistant), and 19 enterococci. Accuracy and reproducibilitystudies were performed with E. coli (ATCC 25922) and S.aureus (ATCC 25923).

Breakpoint microdilution plates. Test panels consisted offrom one to three concentrations of each dehydrated antibi-otic repeated in a pattern three times per microdilution plate.The test panel for gram-negative organisms (APB3A) con-tained the following antibiotics and concentrations (micro-grams per milliliter) when reconstituted with 50 jl of broth:ampicillin (1, 16); carbenicillin (32, 128); piperacillin (16, 64);cephalothin, cefamandole, cefoxitin, and chloramphenicol(8, 16); cefotaxime and moxalactam (8, 32); amikacin (16,32); gentamicin and tobramycin (4, 8); tetracycline (8);nitrofurantoin (64); trimethoprim-sulfamethoxazole (TMP-SMZ) (2/38, 8/152); sulfisoxazole (256); and nalidixic acid(16). The test panel for gram-positive cocci (AGPBP) con-tained (micrograms per milliliter): penicillin (0.03, 0.12, 8);ampicillin (0.25, 2, 8); cephalothin, cefazolin, and chloram-phenicol (8, 16); gentamicin (4, 8); clindamycin and erythro-mycin (0.5, 4); tetracycline (4, 8); vancomycin (4, 16);nitrofurantoin (64); TMP-SMZ (2/38, 8/152); amoxicillin-clavulanic acid (4/2, 16/8); oxacillin (2); and sulfisoxazole(256). The oxacillin well also contained NaCI at a finalconcentration of 2% when reconstituted with 50 tls of broth.SBAS. The SBAS consists of an automated inoculator, an

automated fluorometer (autoreader), a semiautomated platereader (Sensitouch keyboard), and a DEC Pro 350 computersystem (Digital Equipment Corp., Northboro, Mass.). Theautomated inoculator delivers 50 ,ul of inoculated assaybroth to each well of the microdilution plate. The computer-controlled automated fluorometer moves each well of the

microdilution plate over an optical window and measures thefluorescence produced by the bacteria from the fluorogenicsubstrates present in the assay broth after either 5 or 18 h ofincubation. The raw fluorescence values from the antibiotic-free wells are compared with the values from the antibiotic-containing wells by the computer. A mathematical algorithminterprets the values for the different concentrations of eachdrug and assigns a category of susceptible, intermediate(moderately susceptible), or resistant. The Sensitouch key-board is a semiautomated plate reader consisting of a mir-rored viewbox equipped with indicator lights that match acomputer graphic display of the breakpoint panel. It facili-tates the visual interpretation and keyboard entry into thecomputer of the inhibitory endpoints.

Test methods. For the Sensititre procedures, a bacterialsuspension in sterile distilled water matching a 0.5 McFar-land turbidity standard was prepared from isolates grownovernight on 5% sheep blood agar (BBL MicrobiologySystems, Cockeysville, Md.). All frozen stock strains weresubcultured three or more times before testing. A Sensititre5-h or 18-h fluorogenic detection strip was aseptically addedto a tube containing either 10 ml of Sensititre Rapid Broth or10 ml of cation-supplemented Mueller-Hinton broth (18-hassay). The tube was vigorously shaken and incubated at35°C for 30 min before inoculation. For the 5-h test, 200 ,uI ofthe bacterial suspension (10 pI for Proteus spp.) was trans-ferred to the broth to achieve a final inoculum concentrationin the assay of 2 x 106 CFU/ml (1 x 105 CFU/ml for Proteusspp.). For the 18-h test, 10 p.1 (1 pul for Proteus spp.) wasadded to the broth to achieve a final organism density of 105CFU/ml (104 CFU/ml for Proteus spp.).

Using the automated inoculator, 50 ptI of the broth sus-pension containing the fluorogenic substrate was dispensed

J. CLIN. MICROBIOL.

on August 22, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

SENSITITRE AUTOMATED BREAKPOINT TESTING 1081

into wells of the appropriate breakpoint susceptibility plate.After inoculation, the plates were sealed and incubated in an

ambient atmosphere at 35°C for either 5 or 18 h. The patternof fluorescence produced in each well of the plate was

detected with the automated fluorometer. The rapid test wasread with the fluorometer at 5 h (± 10 min). The 18-h test wasread both visually and with the fluorometer after 18 h (± 40min) of incubation. Plates read visually were examined forturbidity or the presence of a button in the bottom of thewell, and those wells with growth were marked and enteredinto the computer with the Sensitouch keyboard. Test re-sults were recorded as susceptible, intermediate, moderatelysusceptible, or resistant.The SDD test was performed by the guidelines of the

National Committee for Clinical Laboratory Standards (5),using the standard method of inoculum preparation. All testprocedures were performed concurrently from the sameblood agar subcultures of the test strains. SBAS reproduc-ibility and accuracy studies were done by repeated dailytesting of E. coli ATCC 25922 and S. aureus ATCC 25923.Data analysis. The results of the SBAS and the SDD were

compared for individual organism-antibiotic combinations,and only those combinations suggested by the NationalCommittee for Clinical Laboratory Standards (5) as appro-priate for routine testing and reporting were evaluated.Discrepancies occurring between test results were assignedto one of three categories: minor, major, and very major.Very major discrepancies were defined as susceptible by theSBAS and resistant by the SDD. Major discrepancies weredefined as resistant by the SBAS and susceptible by theSDD. A minor discrepancy represented an intermediateversus a susceptible or resistant result.

RESULTS

The comparisons of 4,571 test results obtained with therapid SBAS and the SDD methods are shown in Table 2. Theagreement between the two test methods for all organism-antibiotic combinations was 81.6%, with 3.9% very major,6.5% major, and 7.9% minor discrepancies. When the datawere analyzed by organism group, we found 86.8, 59.3, 71.4,and 62.3% agreement for members of the Enterobac-teriaceae, Pseudomonas spp., staphylococci, and entero-cocci, respectively.There were 3.7% very major, 2.5% major, and 7.0% minor

discrepancies that occurred with the 200 members of theEnterobacteriaceae tested. Nitrofurantoin and tetracyclineaccounted for 93 of 180 (51.7%) of the very major discrep-ancies. We found 25% major discrepant results when themembers of the Enterobacteriaceae were tested againstampicillin, carbenicillin, chloramphenicol, and TMP-SMZ.Ampicillin, carbenicillin, and cephalothin together contrib-uted 113 of 237 (47.7%) of the minor discrepancies. No verymajor, 2 (0.3%) major, and 16 (2.7%) minor discrepanciesoccurred with the 600 Enterobacteriaceae-aminoglycosidedatum pairs.Agreement was found between the rapid SBAS and the

SDD for 227 of 383 (59.3%) tests with Pseudomonas spp.There were 1.8% very major, 21.9% major, and 17.0% minordiscrepancies. Low levels of agreement were found for boththe P-lactam and aminoglycoside antibiotic datum pairsevaluated.We found 71.4% concordance between the rapid SBAS

and the SDD for 675 tests performed with staphylococci.Valid breakpoints were obtained for only 8 of 22 (36.4%)

TABLE 2. Comparison of antibiotic susceptibility test results obtained by the 5-h SBAS and the SDD method

No. of tests with"a: Full agreement

Enterobacteriaceae Pseudomonas spp. Staphylococci Enterococci by drug (%)

Amikacin 0, 1, 2, 197 0, 12, 8, 28 NTb NT 90.7Ampicillin 2, 14, 57, 127 NE"' 1, 3, 0, 41 0, 3, 7, 9 66.9Amoxicillin-clavulanic acid NT NT 4, 2, 2, 37 NE 82.2Carbenicillin 2, 10, 27, 161 0, 23, 5, 19 NT NT 72.8Cefamandole 4, 9, 10, 177 NE NT NT 88.5Cefazolin NT NT 6, 5, 2, 32 NE 71.1Cefotaxime 0, 3, 12, 185 2, 12, 15, 19 NT NT 82.3Cefoxitin 2, 7, 10, 181 NE NT NT 90.5Cephalothin 4, 7, 29, 160 NE 13, 5, 0, 27 NE 76.3Chloramphenicol 4, 10, 18, 168 1, 3, 6, 38 3, 16, 1, 26 NE 79.2Clindamycin NT NT 9, 2, 1, 33 NE 73.3Erythromycin NT NT 6, 3, 4, 32 0, 1, 5, 13 70.3Gentamicin 0, 1, 4, 195 4, 3, 4, 37 0, 15, 3, 27 NE 88.4Moxalactam 0, 1, 4, 194 0, 9, 17, 22 NT NT 87.4Nalidixic acid 3, 0, 4, 193 NE NT NT 96.5Nitrofurantoin 57, 2, 14, 127 NE 0, 3, 0, 42 0, 2, 0, 17 70.5Oxacillin NT NT 2, 8, 1, 34 NE 75.5Penicillin NT NT 1, 3, 0, 41 0, 5, 12, 2 67.1Piperacillin 5, 5, 17, 173 0, 19, 9, 20 NT NT 77.8Sulfamethoxazole 3, 0, 4, 193 NE NT NT 96.5Sulfisoxazole NT NT 1, 29, 0, 15 NE 33.3Tetracycline 36, 0, 13, 151 NE 1, 7, 7, 30 0, 2, 1, 16 74.6Tobramycin 0, 0, 10, 190 0, 3, 1, 44 NT NT 94.3TMP-SMZ 4, 16, 2, 178 NE 1, 5, 1, 38 NE 88.2Vancomycin NT NT 0, 7, 0, 38 0, 3, 2, 14 81.3

Full agreement by organism (%) 86.8 59.3 71.4 62.3a Numbers indicate very major discrepancies, major discrepancies, minor discrepancies. and concordant results, respectively.b NT, Not tested.C NE, Not evaluated.

VOL. 26, 1988

on August 22, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

1082 NOLTE ET AL.

coagulase-negative staphylococci and for 37 of 39 (94.9%) S.aureus. Among the 15 antibiotics tested against S. aureus,levels of agreement of >90% were found for penicillin,ampicillin, vancomycin, and nitrofurantoin. The rapid SBAScorrectly categorized 15 of 16 (94.0%) oxacillin-resistant S.aureus (ORSA); however, 4 of 22 (18.0%) susceptible strainswere also categorized as resistant. False-susceptibility tocephalothin occurred for 15 of 16 (93.7%) ORSA strains bythe rapid SBAS and for 4 of 17 (23.5%) ORSA strains by theSDD.When the results of the rapid SBAS and the SDD proce-

dures were compared for 19 strains of enterococci, we found62.3% agreement, with no very major, 14.0% major, and23.7% minor discrepancies. Penicillin and ampicillin resultsshowed the lowest levels of agreement.The comparisons of 4,654 overnight SBAS-SDD datum

pairs are shown in Table 3. The agreement between the twotest procedures for all organism-antibiotic combinations was89.2%, with 2.3% very major, 1.3% major, and 7.1% minordiscrepancies. When the data were analyzed by organismgroup, we found 90.4, 78.1, 90.6, and 83.3% agreement formembers of the Enterobacteriaceae, Pseudomonas spp.,staphylococci, and enterococci, respectively.The results obtained with the members of the Enterobac-

teriaceae at 18 h were similar to those obtained with the 5-hSBAS with the following exceptions. There was substantialimprovement in the level of agreement with the SDD resultsfor tetracycline, with the number of major discrepanciesreduced from 18 to 2%. There were three very majordiscrepancies recorded for moxalactam at 18 h, as opposedto none for the rapid procedure.

Overnight incubation of Pseudomonas spp. improved thelevels of agreement for all antibiotics between the SBAS and

the SDD to 78.1%. Piperacillin was the drug for which thegreatest improvement was demonstrated, with 2.1, 0, and3.0% very major, major, and minor errors, respectively. Thenumber of very major discrepancies that occurred with thePseudomonas spp.-moxalactam combination was greaterwith the overnight SBAS procedure, as noted previously forthe members of the Enterobacteriaceae.An overall improvement was also seen at 18 h for the

gram-positive cocci tested. The results obtained for clinda-mycin and erythromycin with the staphylococci were inclose agreement with the SDD, 100 and 94.2%, respectively.The number of very major discrepancies recorded for cefa-zolin and cephalothin was substantially reduced. Althoughovernight incubation reduced the number of major discrep-ancies that occurred with oxacillin from eight to one, thenumber of very major errors increased from two to four. The18-h SBAS results for augmentin and staphylococci showedonly 78.8% agreement with the SDD results; however, alldiscrepancies were minor. Although there was some im-provement, sulfisoxazole also proved to be a problem for theovernight SBAS procedure, with 3.8% very major, 34.6%major, and no minor discrepancies. Valid breakpoint resultswere obtained after 18 h of incubation for only 15 of 22(68.1%) coagulase-negative staphylococci. We foundmarked improvement in the levels of agreement for all drugstested against the enterococci with the overnight SBASprocedure.We found essentially the same level of agreement between

the overnight SBAS and the SDD results regardless of howthe endpoints were determined. Concordance with the ref-erence method was achieved in 89.3% of the Sensititrebreakpoint tests read visually and in 89.2% of the tests readwith the fluorometer.

TABLE 3. Comparison of antibiotic susceptibility test results obtained by the 18-h SBAS and the SDD method

No. of tests with': Full agreementAntibioticbyduEnterobacteriaceae Pseudoinonas spp. Staphylococci Enterococci by drug (%)

Amikacin 0, 0, 3, 197 1, 0, 6, 41 NTb NT 96.0Ampicillin 1, 10, 62, 127 NE' 3, 0, 0, 49 0, 0, 6, 9 69.3Amoxicillin-clavulanic acid NT NT 0, 0, 11, 41 NE 78.8Carbenicillin 5, 2, 22, 171 2, 1, 13, 32 NT NT 81.9Cefamandole 4, 2, 6, 188 NE NT NT 94.0Cefazolin NT NT 0, 5, 1, 46 NE 88.5Cefotaxime 0, 0, 5, 195 0, 0, 15, 33 NT NT 91.9Cefoxitin 2, 2, 13, 183 NE NT NT 91.5Cephalothin 2, 3, 24, 171 NE 7, 0, 4, 41 NE 84.1Chloramphenicol 4, 2, 11, 183 2, 0, 11, 35 0, 0, 2, 50 NE 89.3Clindamycin NT NT 0, 0, 0, 52 NE 100Erythromycin NT NT 0, 0, 3, 49 0, 0, 0, 15 95.5Gentamicin 0, 2, 5, 193 2, 0, 6, 40 0, 0, 0, 52 NE 95.0Moxalactam 3, 0, 1, 196 4, 0, 16, 28 NT NT 90.3Nalidixic acid 3, 0, 4, 193 NE NT NT 96.5Nitrofurantoin 40, 2, 18, 140 NE 0, 0, 0, 52 0, 1, 0, 14 77.1Oxacillin NT NT 4, 1, 1, 46 NE 88.5Penicillin NT NT 3, 0, 0, 49 0, 0, 8, 7 83.6Piperacillin 0, 0, 19, 181 1, 0, 3, 44 NT NT 90.7Sulfamethoxazole 3, 0, 4, 193 NE NT NT 96.5Sulfisoxazole NT NT 2, 18, 0, 32 NE 61.5Tetracycline 4, 2, 13, 181 NE 0, 0, 0, 52 0, 0, 0, 15 92.9Tobramycin 0, 2, 7, 191 0, 0, 1, 47 NT NT 96.0TMP-SMZ 4, 5, 2, 189 NE 1, 2, 5, 44 NE 92.5Vancomycin NT NT 0, 0, 0, 52 0, 0, 0, 15 100

Full agreement by organism (%) 90.4 78.1 90.6 83.3"Numbersindicateverymajordiscrepancies,major discrepancies,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

" Numbers indicate very major discrepancies, major discrepancies,b NT, Not tested.' NE, Not evaluated.

minor discrepancies, and concordant results, respectively.

J. CLIN. MICROBIOL.

on August 22, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

SENSITITRE AUTOMATED BREAKPOINT TESTING 1083

The reproducibility and accuracy of the SBAS resultswere determined by daily testing of two strains used rou-tinely for quality control of the SDD method. E. coli (ATCC25922) was tested 27 times by the 5-h SBAS procedure.There were 26 (5.6%) discrepant results found for the 459tests. Twenty-three minor errors with ampicillin, one minorerror with carbenicillin, one major error with chloramphen-icol, and one major error with sulfamethoxazole were noted.The same strain was tested 26 times by the 18-h SBASprocedure. There were 23 discrepant results: 22 minor errorswith ampicillin and one minor error with cephalothin.

S. aureus (ATCC 25923) was tested 24 times by the rapidSBAS procedure, and there were 31 (8.6%) discrepanciesrecorded for 360 tests. We found 28 major errors (sulfiso-xazole, 18; TMP-SMZ, 3; and 1 each for penicillin, ampicil-lin, chloramphenicol, gentamicin, nitrofurantoin, cefazolin,and oxacillin) and 3 minor errors (1 each for tetracycline,TMP-SMZ, and amoxicillin-clavulanic acid). With the over-night SBAS procedure, we found 29 (7.2%) discrepancies for405 tests. There were 28 major errors (sulfisoxazole, 25; andTMP-SMZ, 3) and 1 minor error with TMP-SMZ.

DISCUSSION

In this study, we compared the performance of both therapid and overnight SBAS procedures with the SDD forbacterial species commonly subjected to susceptibility test-ing in clinical laboratories. To our knowledge, this study isthe first published comparison of the SBAS with an acceptedreference test method. A previous study (2) demonstratedgood correlation between the results of visually read Sensi-titre breakpoint plates and the SDD method. Subsequentevaluations of the Autoreader system (3, 8, 9) have com-pared the results with visually determined results by usingmicrodilution plates prepared by the same manufacturer.This approach is biased toward the SBAS and provideslimited information regarding the performance of the systemrelative to a reference method.No consensus exists among clinical microbiologists as to

an acceptable level of agreement between the results ofdifferent susceptibility test methods. Thornsberry (10) pro-posed that complete correlation between a new test methodand a reference method should be >90% and that thecombined major and very major discrepancies attributable tothe new system should be <5%. Sherris and Ryan (7)recommended that the overall agreement between the resultsof the two methods should be .95% and that the very majorerrors should not exceed 1.5%.

If we apply the criteria proposed by Thornsberry to ourdata, we find that both the 5- and 18-h SBAS procedures areacceptable alternatives to the SDD for members of theEnterobacteriaceae if nitrofurantoin and tetracycline areexcluded. The overnight SBAS procedure is also found to beacceptable for staphylococci. However, the SBAS resultswith Pseudomonas spp. and enterococci showed unaccept-ably low levels of agreement with the reference method. Weincluded selected strains as well as recent clinical isolates inour evaluation to ensure that the system was challenged withmost of the clinically relevant antibiotic resistance determi-nants. Because of this, our study may represent a moresevere test of the SBAS than the previously publishedevaluation.A disproportionate percentage of very major discrepan-

cies that occurred between the SBAS and the SDD werewith nitrofurantoin and tetracycline when tested againstmembers of the Enterobacteriaceae. The nitrofurantoin very

major discrepancies occurred almost exclusively (90%) withProteus spp., Providencia spp., M. morganii, and Serratiaspp., bacteria generally considered resistant to nitrofuran-toin in vivo (12). The very major discrepancies between therapid SBAS and the SDD results for tetracycline occurredwith 22 Proteus mirabilis, 5 Serratia spp., 4 E. coli, 2Enterobacter aerogenes, and 1 each Klebsiella pneumoniae,Salmonella typhimurium, and Shigella dysenteriae. Theproblem with the rapid tetracycline results most likely has itsorigin in the algorithm used to interpret the fluorescentsignals at 5 h because there was a dramatic reduction in thenumber of very major discrepancies recorded for tetracy-cline with the overnight SBAS procedure.The problem with nitrofurantoin appears to be a funda-

mental one in that poor correlation with the SDD methodwas found for all Sensititre procedures regardless of thelength of incubation or how the endpoints were determined.Doern et al. (2) also noted a large percentage of very majordiscrepancies occurring with members of the Enterobac-teriaceae and nitrofurantoin in their evaluation of manuallyread Sensititre breakpoint plates. They attributed the dis-crepancies to false-resistant results by SDD since the break-point results were confirmed by the Sensititre full-range MICprocedure in 14 of 17 cases.Most of the minor discrepancies for members of the

Enterobacteriaceae occurred with five drugs: ampicillin,cephalothin, carbenicillin, piperacillin, and chlorampheni-col. Minor discrepancies can result from differences betweenthe methods in technical performance or interpretive crite-ria. For ampicillin, carbenicillin, and pipericillin, substantialdifferences exist between the concentrations of drug selectedfor incorporation in the breakpoint panels and the NationalCommittee for Clinical Laboratory Standards equivalentMIC breakpoints for the SDD categories (5). These differ-ences probably help explain the large number of minordiscrepancies that occurred between the two methods, es-pecially for ampicillin.The combination of ampicillin and Enterobacter spp. has

proved problematic for a number of 4- to 8-h antibioticsusceptibility test systems (4, 11). The problems have beenmostly false-susceptible results with the rapid systems. Wefound no false ampicillin susceptibility among the 23 resis-tant Enterobacter spp. we tested with the 5-h SBAS.

Reliable detection of ORSA has been difficult for mostrapid susceptibility testing systems (10). Systems that re-quire incubation for 4 to 8 h do not allow enough time for thegrowth of heteroresistant strains, and the result is oftenfalse-susceptibility. We found excellent correlation (95%)between the oxacillin results of the rapid SBAS and the SDDfor oxacillin-susceptible S. aureus and ORSA. One ORSAwas not recognized in either the rapid or overnight SBASprocedure. The AGPB test panel used in this study con-tained oxacillin as the representative semisynthetic penicillinand NaCI at a final concentration of2% to enhance detectionof heteroresistant strains (6). Staneck et al. (9) found thatonly 1 of 8 (12.5%) of methicillin-resistant S. aureus testedsuccessfully with the rapid Autoreader full-range MIC pro-cedure. However, the microdilution plates contained methi-cillin without the increased concentration of NaCl.

False-susceptibility to the cephalosporins for ORSA oc-curred more frequently with the SBAS system than with theSDD method. However, all ORSA should be consideredresistant to the cephalosporins regardless of the results of invitro susceptibility tests.The coagulase-negative staphylococci generated sufficient

fluorescent signal in the growth control wells for a valid test

VOL. 26, 1988

on August 22, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

1084 NOLTE ET AL.

breakpoint in only 36.4 and 68.1% of the 5- and 18-h SBAStests, respectively. This appeared to result from poor growthin the broth medium rather than failure to liberate fluorogensince visual reading at 18 h often showed scant bacterialpellets in the growth control wells. Staneck et al. (9) foundthat the Autoreader gave MIC results for only 4 of 42 (9.5%)coagulase-negative staphylococci after 5 h of incubation.The SBAS results of sulfisoxazole testing for staphylo-

cocci were unreliable at both 5 and 18 h of incubation.Susceptibility testing of sulfisoxazole is very medium depen-dent; however, testing on Mueller-Hinton agàr works rea-sonably well providing an endpoint of 80% inhibition is used(1). The endpoints in broth microdilution methods may bedifficult to define because even sensitive strains may gothrough several growth generations before endogenous andexogenous folate is depleted.The SBAS is mechanically reliable, and we experienced

no major problems with the automated inoculator, fluoro-meter, or computer during the 5-month evaluation period.The reproducibility and accuracy of the SBAS was assessedby repeated daily testing of E. coli (ATCC 25922) and S.aureus (ATCC 25923). With the exception of ampicillin, theresults with E. coli were highly reproducible and accurate.The quality control range of MICs for this strain is 2.0 to 8.0ug/lml (6). Since the concentrations of ampicillin in theAPB3A plate were 1.0 and 16 ,ug/ml, it follows that theampicillin results were consistently interpreted by the sys-tem as intermediate. Accuracy was also a problem forsulfisoxazole and staphylococci. As discussed above, de-fining inhibitory endpoints in broth dilution tests with sulfi-soxazole is often difficult.This study demonstrates the considerable potential that

fluorogenics holds for antibiotic susceptibility testing. TheSBAS has several features that distinguish it from otherautomated systems. It offers the user the option of perform-ing either same-day or overnight susceptibility testing withthe same instrument. In addition, plates that are incubatedovernight can be read visually. Visual reading serves as abackup in case of instrument malfunction because the fluo-rescence assay does not interfere with bacterial growth.Inhibitory endpoints obtained by breakpoint microdilutiontesting may be more distinct and easier to interpret thanthose obtained by either the SDD method or full-range MICs(2). The SBAS microdilution plates contain dehydrated,stabilized antibiotics, and as a result, the plate may be storedfor long periods at room temperature. Finally, the cost pertest is kept to a minimum by incorporation of three separateantibiotic panels in each microdilution plate.

ACKNOWLEDGMENT

This study was supported in part by Sensititre, Inc., NorthAndover, Mass.

LITERATURE CITED

1. Bauer, A. W., and J. C. Sherris. 1964. The determination ofsulfonamide susceptibility of bacteria. Chemotherapia 9:1-19.

2. Doern, G. V., A. Dascal, and M. Keville. 1985. Susceptibilitytesting with the Sensititre breakpoint broth microdilution sys-tem. Diagn. Microbiol. Infect. Dis. 3:185-191.

3. Doern, G. V., J. L. Staneck, C. Needham, and T. Tubert. 1987.Sensititre autoreader for same-day breakpoint broth microdilu-tion susceptibility testing of members of the family Enterobac-teriaceae. J. Clin. Microbiol. 25:1481-1485.

4. Hansen, S. L., and P. K. Freedy. 1983. Concurrent comparabil-ity of automated systems and commercially prepared microdi-lution trays for susceptibility testing. J. Clin. Microbiol. 17:878-886.

5. National Committee for Clinical Laboratory Standards. 1984.Performance standards for antimicrobial disk susceptibilitytests. Approved standard M2-A3. National Committee for Clin-ical Laboratory Standards, Villanova, Pa.

6. National Committee for Clinical Laboratory Standards. 1985.Methods for dilution antimicrobial susceptibility tests for bac-teria that grow aerobically. Approved standard M7-A. NationalCommittee for Clinical Laboratory Standards, Villanova, Pa.

7. Sherris, J. C., and K. J. Ryan. 1982. Evaluation of automatedand rapid methods, p. 1-5. In R. C. Tilton (ed.), Rapid methodsand automation in microbiology. American Society for Micro-biology, Washington, D.C.

8. Staneck, J. L., S. D. Allen, E. E. Harris, and R. C. Tilton. 1985.Automated reading of MIC microdilution trays containing fluo-rogenic enzyme substrates with the Sensititre autoreader. J.Clin. Microbiol. 22:187-191.

9. Staneck, J. L., S. D. Allen, E. E. Harris, and R. C. Tilton. 1988.Rapid MIC testing with the Sensititre Autoreader. J. Clin.Microbiol. 26:1-7.

10. Thornsberry, C. 1985. Automated procedures for antimicrobialsusceptibility tests, p. 1015-1018. In E. H. Lennette, A. Ba-lows, W. J. Hausler, Jr., and H. J. Shadomy (ed.), Manual ofclinical microbiology, 4th ed. American Society for Microbiol-ogy, Washington, D.C.

11. Thornsberry, C., T. L. Gavan, J. C. Sherris, A. Balows, J. M.Matsen, L. D. Sabath, F. Schoenknecht, L. D. Thrupp, and J. A.Washington' Il. 1975. Laboratory evaluation of a rapid, auto-mated susceptibility testing system: report of a collaborativestudy. Antimicrob. Agents Chemother. 7:466-480.

12. Washington, J. A., Il. 1969. Antimicrobial susceptibility ofEnterobacteriaceae and nonfermenting gram-negative bacilli.Mayo Clin. Proc. 44:811-824.

J. CLIN. MIC'RO'BIOL.

on August 22, 2020 by guest

http://jcm.asm

.org/D

ownloaded from