Fg/ml into the gentamicin and tobramycin panels, and 12 and 24 pig/ml into the amikacin. panels. Minimal inhibitory concentration (MIC)

Transcription

1 ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Nov. 1983, p /83/ $02.00/0 Copyright C 1983, American Society for Microbiology Vol. 24, No. 5 Error Rates Associated With the Use of Recently Proposed Breakpoints for Testing Pseudomonas aeruginosa versus Gentamicin, Tobramycin, and Amikacin by the Standardized Disk Agar Diffusion Test BERT F. WOOLFREY*, JOAN M. K. FOX, CHARLES 0. QUALL, AND RICHARD T. LALLY Clinical Microbiology Section, Department ofanatomic and Clinical Pathology, St. Paul-Ramsey Medical Center, St. Paul, Minnesota Received 5 April 1983/Accepted 17 August 1983 Two hundred fifteen Pseudomonas aeruginosa isolates were tested in parallel by the disk agar diffusion test, using a standardized agar preparation, and by a microbroth test, using dilutions differing by small arithmetic increments. For gentamicin, recently proposed breakpoints of resistance (R) c 12 mm and supceptibility (S) - 16 mm produced error rates of 20 and 6.8%, respectively. Limiting the error rate for susceptible interpretations to s 2% produced a widening of the intermediate zone to include 67.4% of the isolates tested. For tobramycin, the recently proposed breakpoints of R s 12 mm and S 2 15 mm were associated with error rates of 66.7 and 1.4%, respectively. Breakpoints of R 5 12 mm and S ; 13 mm were demonstrated to be equally effective when the error rate for susceptible interpretations was limited to 2% by - error rate-bound analysis. For amikacin, proposed breakpoints of R s 14 mm and S a 17 mm were associated with error rates of 27.3 and 3.2%, respectively. Limiting the error rates for susceptible interpretations to s 2% required breakpoints of R s 14 mm and S 2 18 mm. The ability to establish effective susceptibility breakpoints for tobramycin and amikacin appeared not to be related to the disk agar diffusion test process itself but rather to the high degree of susceptibility of the P. aeruginosa population. These findings severely limit the usefulness of the disk agar diffusion procedure for testing P. aeruginosa versus the aminoglycosides. For this purpose, we recommend dilution tests which employ small arithmetic increment schemes. Despite extensive investigative work and many published reports, the usefulness of the standardized disk agar diffusion test for assessing the susceptibility of Pseudomonas aeruginosa to the aminoglycosides remains in question. Over a decade ago, initial recommendations for inhibition zone diameter breakpoints for P. aeruginosa versus gentamicin were 12 mm or less for assigning resistance (R), and 13 mm or greater for susceptibility (S) (13). Since then, a number of changes in breakpoints for gentamicin, as well as for new aminoglycosides, have been proposed by the National Committee on Clinical Laboratory Standards (NCCLS) and others (12, 14, 15). Most recently, Barry and coinvestigators have proposed new breakpoints of R c 12 mm and S :16 mm for gentamicin, R c 12 mm and S - 15 mm for tobramycin, and R c 14 mm and S-17 mm for amikacin (2, 3). These breakpoints were determined by using reference microbroth test panels employing twofold dilutions from 0.5 to 64 Fag/ml with the incorporation of two intermediate concentrations of 6 and Fg/ml into the gentamicin and tobramycin panels, and 12 and 24 pig/ml into the amikacin panels. Minimal inhibitory concentration (MIC) values of s6 and >8 ;ig/ml were, respectively, used to define S and R for gentamicin and tobramycin, with s12 and >16 ;.g/ml, respectively, representing S and R for amikacin. The inhibition zone diameter breakpoints which were determined and proposed on this basis were said to provide acceptable accuracy and precision for the disk agar diffusion test. In contrast, other investigators, working largely with P. aeruginosa versus gentamicin, have reported significant difficulties in establishing useful breakpoints (15, 16, 20, 23, 24), and our own observations have recently been supported (C. W. Stratton, H. B. Hawley, L. S. Patterson, and M. E. Evans, Abstr. Annu. Meet. Am. Soc. Microbiol. 1983, C312, p. 363). Such differing observations and recommendations are no doubt the result of problems encountered in controlling a number of medium-related variables which are known to influence the measurement of MIC

2 VOL. 24, 1983 values of aminoglycosides against P. aeruginosa, especially the concentrations of divalent cations (1, 4, 6-9, 15-17). Such factors appear to be more easily controllable for broth dilution tests than for agar-based tests. This appears to be due to the difficulties encountered in the manufacture and standardization of Mueller- Hinton agar (MHA), and to variations which occur in MHA medium as prepared on a day-today basis in microbiology laboratories. Recently, manufacturers have attempted to produce standardized MHA preparations which have been adjusted by manipulation of calcium and magnesium content, etc., to meet susceptibility test performance standards. The use of such standardized MHA preparations, in conjunction with strict daily quality control of MHA plates carrying P. aeruginosa ATCC 27853, has now been generally promoted as providing acceptable results when testing P. aeruginosa versus aminoglycosides. The present study was designed to investigate the error rates associated with the newly proposed breakpoints for gentamicin, tobramycin, and amikacin when applied to a purportedly performance-standardized commercial MHA preparation. A small arithmetic increment microbroth dilution test was used to minimize error which might have been introduced into our previously reported results (18, 21) as a consequence of using twofold dilution schemes. Results were evaluated by both regression-correlation analysis and by our modification of the error rate-bound method. DISK AGAR DIFFUSION TEST ERROR RATES 765 MATERIALS AND METHODS Experimental design. Two hundred fifteen P. aeruginosa clinical isolates were tested in parallel by the standardized disk agar diffusion test (14), using Mueller-Hinton II agar (BBL Microbiology Systems, Cockeysville, Md.), and by a reference microbroth test (19), using dilutiohs differing by small arithmetic increments. The small increment schemes were used to improve on test precision afforded by conventional twofold dilution steps. Paired MIC and inhibition zone diameter data were obtained for gentamicin, tobramycin, and amikacin. These data were studied by conventional regression and correlation analysis (5), and by our modifications of the Metzler and DeHaan error rate analysis procedure (11, 20, 23), to evaluate the usefulness of the recently introduced breakpoint schemes for the aminoglycosides and the standardized disk agar diffusion test. Microorganisms. Two hundred fifteen P. aeruginosa clinical isolates were studied at the time of isolation in the St. Paul-Ramsey Medical Center Clinical Microbiology Laboratory. Isolates were identified as P. aeruginosa by using the API-20E system (Analytab Products, Plainview, N.Y.) as outlined in a previous report (22). Microbroth dilution tests. MICs were determined for aminoglycosides against each microorganism at the time of isolation, using the MIC-2000 (Dynatech Laboratories, Alexandria, Va.) microbroth dilution system (19). Mueller-Hinton broth, supplemented with calcium and magnesium to achieve concentrations of 5.5 ± 0.2 and 2.5 ± 0.2 mg/dl, respectively, was prepared by the clinical microbiology laboratory and was used in all dilution tests. Antimicrobial agent dilutions were prepared so as to differ by small arithmetic increments rather than by conventional twofold dilution schemes. Gentamicin and tobramycin concentrations were prepared in 1-,ug/ml increments ranging from 1 through 16,ug/ml. Amikacin concentrations were prepared in 2-,ug/ml increments ranging from 2 through 32 ^g/ml. The aminoglycosides were supplied from the manufacturers (Schering Corp., Bloomfield, N.J.; Eli Lilly & Co., Indianapolis, Ind.; and Bristol Laboratories, Syracuse, N.Y.) as desiccated powders. Details relating to MIC tray production, inoculum preparation, and incubation conform to those previously described (19). Quality control was performed daily by using P. aeruginosa ATCC The mean MIC for gentamicin was 2.8,ug/ml, with 80% of MICs at 3.0,ug/ml and 20o at 2.0,ug/ml, which corresponded to the narrow dispersion previously reported for small arithmetic increment MICs (18, 21), i.e., 72.9o of the MIC values being modal, with a 95% confidence limit of ±1 small arithmetic unit from the modal value, and 100% of values falling within ±2 small arithmetic increments. The small arithmetic increment microbroth test has been used in our laboratory for a number of years for testing P. aeruginosa and other problem organisms against the aminoglycosides and other potentially toxic agents. Throughout this period, the performance characteristics, as mentioned above for the smallincrement microbroth test, have been consistently matched or bettered. Standardized disk agar diffusion test. Standardized disk agar diffusion tests were performed in the clinical microbiology laboratory at the time of isolation of each microorganism. All susceptibility plates were prepared from a single lot of Mueller-Hinton II agar lot no. J3DKWG, using 20 ml in each 100-mm-diameter petri dish. This lot of Mueller-Hinton II agar was alleged to be appropriately adjusted and tested by the manufacturer for satisfactory susceptibility testihg of P. aeruginosa. It was felt that the error rate which would be determined by using the single lot of Mueller-Hinton II agar might represent optimum rates achievable by standardized MHA preparations and would not be complicated by errors associated with the use of multiple lots. Calcium and magnesium concentrations in the Mueller-Hinton II agar preparations were measured by atomic absorption spectrophotometry to be, respectively, 4.0 ± 0.4 and 1.3 ± 0.1 mg/dl. Total calcium and magnesium content was measured to obtain a base of reference relative to concentrations which may be expected in standardized agar preparations. No measurements of bound, soluble, or ionized calcium and magnesium were made because of both complexity of measurement and interpretation of values, as well as their being outside the scope and point of the investigation. Inoculum suspensions were prepared in cation-adjusted Mueller-Hinton broth. Except for the use of 100-mm petri dishes, the disk agar diffusion test was performed according to NCCLS standards (13). Gentamicin 10-,ug, tobramycin 10-l±g, and amikacin 10-p,g susceptibility disks (General Diagnostics, Inc., Morris Plains, N.J.) were symmetrically

3 766 WOOLFREY ET AL. >16 H- 16 " TR 14 ' _ c 9 Ct 8L IR ~SR 2K!51 -I 0D RI ZR@ TI )D(9 00D 9)( () (D (!) OSI 04 Zs ANTIMICROB. AGENTS CHEMOTHER. placed on the surface of each susceptibility plate midway between the central portion and the edge. Plates were closely monitored to insure exactness of depth and surface characteristics. Each batch preparation was subjected to the standard quality control procedures of the clinical microbiology laboratory. Daily plate quality control was also performed by using P. aeruginosa ATCC versus gentamicin. During the experimental period, inhibition zone diameters ranged from 16 through 19 mm, with a mean inhibition zone diameter of 17.8 mm. Data analysis. The MIC-inhibition zone diameter data sets for gentamicin, tobramycin, and amikacin were analyzed by conventional regression and correlation methods (5) and by our modifications of the Metzler and DeHaan approach (11, 20, 23). These modifications of the error rate analysis procedure, as used for evaluating the recently proposed aminoglycoside breakpoints, are illustrated in Fig. 1 which is a scatter diagram of MIC-inhibition zone diameter data points. Lines ZR and Zs represent breakpoints for R and S as applied to the disk agar diffusion test. In the present example, the breakpoints are those recommended by Barry (2, 3) for P. aeruginosa versus gentamicin for which R is assigned for inhibition zone diameters which are 512 mm and S is assigned for inhibition zone diameters >16 mm. The line M1 represents an arbitrary but realistic breakpoint for the microbroth dilution test for which isolates requiring MICs -Ml are judged to be susceptible. Similarly, M2 represents an MIC value for which isolates requiring MICs.M2 are judged to be resistant. The lines thus divide the scatter diagram into nine areas which categorize the standardized disk agar diffusion test interpretations as true resistant (TR), intermediate resistant (IR), susceptible and not resistant (SR), resistant intermediate (RI), true intermediate (TI), susceptible intermediate (SI), true susceptible (TS), intermediate susceptible (IS), and resistant and not susceptible (RS). The error rate for false-susceptible interpretations, Es, may thus be calculated as ES = [(RS + IS)/(TS + IS + RS)] x 100, in which TS, IS, and RS represent the number of isolates falling into each of these categories. Likewise, an error rate for false-resistant interpretations, ER, may be calculated as ER = [(IR + SR)/(TR + IR + SR)] x 100. Using the scatter diagrams representing the MIC-inhibition zone diameter data sets, error rates associated with the use of several inhibition zone diameter breakpoints (see Tables 2 to 4) were determined for P. aeruginosa versus gentamicin, tobramycin, and amikacin, using several commonly encountered M1 and M2 definitions of susceptibility and resistance for the reference broth dilution test. RESULTS Table 1 summarizes the results of regression and correlation analysis as applied to the MICinhibition zone diameter data for all isolates, and as applied to a more limited data set representing those isolates for which both the MIC and the inhibition zone diameter values were definite onscale measurements. For all three antimicrobial agents, correlation coefficients did not exceed For the on-scale isolates, correlation coefficients were significantly lower than those found by using all isolates combined. Table 2 summarizes the error rates which were found to be associated with three inhibition zone diameter breakpoint schemes for gentamicin, as judged in reference to three commonly used broth dilution MIC breakpoint sets. The 0D Q0 +@@ 0D 00(j RS is (3 (Qf) (I) (D TS 0 I I 00I I I I I Zone Diameter in mm Scatter diagram of MIC-inhibition zone diameter data points for P. aeruginosa versus gentamicin, FIG. 1. illustrating the modified method of error rate-bound analysis. -M lvi

4 VOL. 24, 1983 TABLE 1. Summary of the results of regression and correlation analysis as applied to MIC-inhibition zone diameter data sets for P. aeruginosa versus gentamicin, tobramycin, and amikacin Antimicrobial Isltse No. of Correlation agent Isolate set isolates coefficient Gentamicin All isolates On-scale isolatesa Tobramycin All isolates On-scale isolates Amikacin All isolates On-scale isolates a On-scale isolates were those with both MIC and inhibition zone diameter values as on scale measurements. lowest error rates were observed when MIC values of <6 and >8 Fxg/ml, respectively, were used as definitions of susceptibility and resistance for the broth dilution reference test. Under these circumstances, use of the Barry breakpoints, in comparison with those proposed by NCCLS, improved ES from 10.2 to 6.8%. Concomitantly, however, the percentage of isolates which were classified as intermediate increased from 15.8 to 24.6%. An ER of 20% was found for both the NCCLS and Barry proposals. When Es was limited by error rate-bound analysis to TABLE 2. DISK AGAR DIFFUSION TEST ERROR RATES 767 <2%, on the basis of rationale to be discussed later, 67.4% of all isolates were given intermediate classifications. Table 3 summarizes the error rates which were observed for tobramycin breakpoints. In contrast to gentamicin, almost all isolates were susceptible to tobramycin for the NCCLS and Barry breakpoints. An Es of 1.4% occurred for 212 susceptible interpretations. Error ratebound analysis demonstrated that Es could be limited to 2% or less by using breakpoints of R < 12 mm and S 2 13 mm. In all instances, the R s 12 mm breakpoint was associated with relatively high ER values. Table 4 summarizes the results of error analysis for amikacin breakpoints. As with tobramycin, the vast majority of isolates were notably susceptible. Use of the NCCLS breakpoints in reference to MIC breakpoints of S < 16,g/ml and R > 18,ug/ml resulted in an Es of 3.2%. Limiting Es to 2% or less necessitated a widening of the intermediate zone by 1 mm, as compared with the NCCLS proposal, so that breakpoints became R 14 - mm and S - 18 mm. In all instances, the breakpoints of R c 14 mm produced relatively high ER values. DISCUSSION Despite the use of a single lot of standardized MHA for this study, correlation coefficients Summary of error rates associated with the use of various inhibition zone diameter breakpoints for assessing susceptibility of P. aeruginosa to gentamicin by the standardized disk agar diffusion test" Disk test breakpoints Broth test ER %I ES (mm) breakpoints (>g/ml) (TR, IR, SR) (RI, TI, SI) (TS, IS, RS) NCCLS (12, 15) -6, (8, 6, 1) (3, 17, 14) (149, 13, 4) <4, > (12, 2, 1) (12, 18, 4) (115, 42, 9) s6, > (12, 2, 1) (12, 8, 14) (149, 8, 9) Barry (12, 16) -6, (8, 6, 1) (5, 22, 26) (137, 8, 2) -4, > (12, 2, 1) (15, 32, 6) (113, 28, 6) <6, > (12, 2, 1) (15, 12, 26) (137, 4, 6) ERBA-2% (12, 19) <6, (8, 6, 1) (12, 24, 109) (54, 1, 0) ERBA-2% (12, _-b) c4, >8 ERBA-2% (12, 19) s6, > (12, 2, 1) (21, 15, 109) (54, 1, 0) a ER and ES are defined in the text. I, Percentage of all isolates judged to be intermediate by the disk agar diffusion test. (TR, IR, SR) = number of isolates judged to be TR, IR, and SR; (RI, TI, SI) = number of isolates judged to be RI, TI, and SI; (RS, IS, TS) = number of isolates judged to be RS, IS, and TS. (See the text for definitions.) ERBA-2% = Error rate-bound analysis with Es limited to <2%. b Es not limitable to -<2%.

5 768 WOOLFREY ET AL. ANTIMICROB: AGENTS CHEMOTHER. TABLE 3. Summary of error rates associated with the use of various inhibition zone diameter breakpoints for assessing susceptibility of P. aeruginosa to tobramycin by the standardized disk agar diffusion testa Disk test breakpoints Broth test ER %I ES (mm) breakpoints (,ug/ml) (TR, IR, SR) (RI, TI, SI) (TS, IS, RS) NCCLS, Barry (12, 15) s6, <4, > (1, 1, 1) (0, 0, 0) (207, 4, 1) 6, > ERBA-2% (12, 13) c6, ERBA-2% (12, 17) <4, > (1, 1, 1) (0, 2, 2) (205, 2, 1) ERBA-2% (12, 13) <6, > a Abbreviations are the same as those in Table 2. were found to be unacceptably low and were were used, correlation coefficients became, resimilar to those reported by Krasemann and spectively, 0.67 and These observations, in Hildenbrand (10). Of particular interest is the conjunction with our findings, indicate a poor finding of lower correlation coefficients for iso- correlation of MICs and inhibition zone diamelates having on-scale MICs and inhibition zone ters for ranges extending from well below to well diameters. This is significant in that regression above the inhibition zone diameter breakpoints and correlation studies have sometimes errone- recently recommended for testing P. aeruginosa ously included off-scale values in their computa- versus the aminoglycosides by the disk agar tions, thereby producing artificially high correla- diffusion test. tion values. On this basis, we examined the data Tables 2 to 4 summarize the results of error of Barry et al. (3) for their 470 gram-negative analysis relating to the three aminoglycosides. isolates, which included 130 P. aeruginosa iso- For gentamicin, there was an improvement in Es lates that were tested in parallel by microbroth from 10.2 to 6.8% when the recently proposed dilution tests and the disk agar diffusion test. breakpoints of R c 12 mm and S - 16 mm were When data for all isolates were analyzed, we substituted for the previous NCCLS proposal of found correlation coefficients of 0.93 and 0.93, R < 12 mm and S 2 15 mm. For toxic antimicrorespectively, for gentamicin and tobramycin. In bial agents, such as gentamicin, which are used contrast, when data for only on-scale isolates for the treatment of critically ill patients, it is TABLE 4. Summary of error rates associated with the use of various inhibition zone diameter breakpoints for assessing susceptibility of P. aeruginosa to amikacin by the standardized disk agar diffusion test' Disk test breakpoints Broth test ER %I ES (mm) breakpoints (jig/ml) (TR, IR, SR) (RI, TI, SI) (TS, IS, RS) NCCLS, Barry (14, 17) <12, (3, 7, 1) (1, 5, 8) (178, 11, 1) <16, > (5, 3, 3) (1, 3, 10) (184, 4, 2) <16, > (8, 0, 3) (2, 2, 10) (184, 2, 4) ERBA-2% (14, 20) -12, (3, 7, 1) (1, 15, 51) (135, 1, 1) ERBA-2% (14, 18) <16, > (5, 3, 3) (1, 6, 18) (176, 1, 2) ERBA-2% (14, 18) <16, > (8, 0, 3) (3, 4, 18) (176, 0, 3) a Abbreviations are the same as those in Table 2.

6 VOL. 24, 1983 desirable that Es be less than 1%. In this way, no more than 1 patient in 100 might receive inappropriate treatment on the basis of the laboratory test. In consideration of the 6.8% Es which was found for the recently proposed breakpoints, an important question must be addressed. What is the magnitude of error contributed by the reference test itself? We have previously found (18, 21) that dispersion characteristics for the small-increment microbroth test are approximately the same as those for replicated inhibition zone diameters. Because of this, in the search for appropriate breakpoints by error rate-bound analysis, it was deemed appropriate to limit ES to <2% rather than <1%. Breakpoints determined on this basis produced a marked widening of the intermediate zone to incorporate 67.4% of the isolates. R breakpoints were not manipulated, and ER was found to be unacceptably high in all instances. For tobramycin, the use of the proposed breakpoints of Rs 12 mm and S 2 15 mm resulted in an ES of <2% but an unacceptably high ER. Error rate-bound analysis demonstrated that an ES of <2% could actually be achieved by S - 13 mm. However, the apparent usefulness of such breakpoints appears to be largely determined by the fact that the P. aeruginosa population was highly susceptible to tobramycin rather than being related to the reliability of the disk agar diffusion test process itself. This is evidenced by the poor correlation of MIC and inhibition zone diameter values which were found for the on-scale isolates. Had the population been more resistant to tobramycin, such breakpoints would have been less effective, and error rates would have approached those which were found with gentamicin. For amikacin, the proposed breakpoints of R. 14 mm and S 2 17 mm were associated with an Es of 3.2%. Limiting the error rate to.2% resulted in a somewhat wider intermediate zone, with R 2 14 mm and S < 18 mm. ER was unacceptably high in all instances. As with tobramycin, the P. aeruginosa population was highly susceptible to amikacin. The findings described above, which were based on the use of a single lot of standardized MHA for the disk agar diffusion test and an improved small-increment microbroth reference test, support our previous observations (20, 23, 24), which were based on the use of nonstandardized MHA preparations and a reference microbroth test using twofold dilution schemes. Data for this study were accumulated by performing parallel tests on a day-to-day basis, using batches of MHA susceptibility plates as prepared for day-to-day use in the clinical microbiology laboratory. Relationships of total, soluble, and ionized calcium and magnesium, as well DISK AGAR DIFFUSION TEST ERROR RATES 769 as other media characteristics, are known to be influenced by such factors as temperature, boiling time, storage conditions, etc., and may unpredictably vary from batch to batch (1, 4, 6-9, 15-17). Although a single lot of manufacturer's standardized MHA was used throughout the study, the use of multiple batches of susceptibility plates made from the same lot of agar may have resulted in media variations of sufficient magnitude to affect MIC measurements in the clinical laboratory setting. It is interesting that daily quality control tests and tests on individual batch preparations with P. aeruginosa ATCC conformed to NCCLS standards. This inability of the quality control system to predict performance problems is unexplained but might be related to differences in- degrees of media dependency for various P. aeruginosa strains. In light of these findings and observations, it is plausible to assume, until demonstrated to the contrary, that other lots and preparations of standardized MHA might also perform poorly when introduced into a clinical laboratory setting, and that manufacturers will face a difficult task in producing a standard MHA preparation. The resolution of this problem goes beyond the scope of the present study and must await larger and more comprehensive investigations such as those provided by large collaborative studies. On the basis of our findings, we make the following conclusions and recommendations. Unacceptably high error rates are associated with the recently proposed disk agar diffusion test breakpoints for gentamicin and amikacin. For gentamicin, limiting Es to <2% makes the test impractical because approximately twothirds of P. aeruginosa isolates will be classified as intermediate. For amikacin, limiting ES to.2% necessitates the use of a breakpoint of S- 18 mm rather than 2 17 mm as recently proposed. For tobramycin, the recently proposed breakpoints of R < 12 mm and S - 15 mm provide a satisfactory ES but an unacceptable ER. Actually, as demonstrated by error ratebound analysis, breakpoints of R s 12 mm and S - 13 mm are equally effective. As P. aeruginosa populations develop increasing resistance to tobramycin and amikacin, it is to be expected that breakpoints will need to be modified and error rates for susceptible interpretations will increase. With progressive widening of intermediate zones, the disk agar diffusion test may also become impractical for tobramycin and amikacin. Because of the uncertainties and high error rates associated with the disk agar diffusion test, we suggest that small-increment dilution tests should be used for testing P. aeruginosa versus the aminoglycosides. Even then, limitations related to test precision and the question of test accuracy must be kept in mind in the final

7 770 WOOLFREY ET AL. application of test results to clinical decision making. ACKNOWLEDGMENT Funds for this study were provided by the St. Paul-Ramsey Hospital Medical Education and Research Foundation, grant no LITERATURE CITED 1. Barry, A. L., and L. J. Effinger Performances of Mueller-Hinton agars prepared by three manufacturers. Am. J. Clin. Pathol. 62: Barry, A. L., C. Thornsberry, and R. N. Jones Gentamicin and amikacin disk susceptibility tests with Pseudomonas aeruginosa: definition of minimal inhibitory concentration correlates for susceptible and resistant categories. J. Clin. Microbiol. 13: Barry, A. L., C. Thornsberry, R. N. Jones, and H. Gerlach Gentamicin, tobramycin, and sisomicin disc susceptibility tests. Revised zone standards for interpretation. Am. J. Clin. Pathol. 75: Casillas, E., M. A. Kenny, B. H. Minshew, and F. D. Schoenknecht Effect of ionized calcium and soluble magnesium on the predictability of the performance of Mueller-Hinton agar susceptibility testing of Pseudomonas aeruginosa with gentamicin. Antimicrob. Agents Chemother. 19: Colton, T Regression and correlation, p In T. Colton (ed.), Statistics in medicine. Little, Brown & Co., Boston. 6. Garrod, L. P., and P. M. Waterworth Effect of medium composition on the apparent sensitivity of Pseudomonas aeruginosa to gentamicin. J. Clin. Pathol. 22: Gilbert, D. N., E. Kutscher, P. Ireland, J. A. Barnett, and J. P. Sanford Effect of the concentrations of magnesium and calcium on the in-vitro susceptibility of Pseudomonas aeruginosa to gentamicin. J. Infect. Dis. 124 (Suppl.): Hameister, V. W., and H. Wahlig Der einflus von Mg++-und Ca++-Ionen auf die wirksamkeit von gentamycin gegen Pseudomonas aeruginosa im agardiffusionstest. Arzneim. Forsch. 21: Kenny, M. A., H. M. Poilock, B. H. Minshew, E. Casillas, and F. D. Schoenknecht Cation components of Mueller-Hinton agar affecting testing of Pseudomonas aeruginosa susceptibility to gentamicin. Antimicrob. Agents Chemother. 17: Krasemann, C., and G. Hildenbrand Interpretation of agar diffusion tests. J. Antimicrob. Chemother. 6: Metzler, C. M., and R. M. DeHaan Susceptibility tests of anaerobic bacteria: statistical and clinical considerations. J. Infect. Dis. 130: Minshew, B. H., H. M. Pollock, F. D. Schoenknecht, and ANTIMICROB. AGENTS CHEMOTHER. J. C. Sherris Emergence in a burn center of populations of bacteria resistant to gentamicin, tobramycin, and amikacin: evidence for the need for changes in zone diameter interpretative standards. Antimicrob. Agents Chemother. 12: National Committee for Clinical Laboratory Standards Proposed performance standards for antimicrobial disc susceptibility tests, ASM-2. National Committee for Clinical Laboratory Standards, Villanova, Pa. 14. National Comndttee for Clinical Laboratory Standards Performance standards for antimicrobic disc susceptibility tests, 2nd ed. National Committee for Clinical Laboratory Standards, Villanova, Pa. 15. Poilock, H. M., B. H. Minshew, M. A. Kenny, and F. D. Schoenknecht Effect of different lots of Mueller- Hinton agar on the interpretation of the gentamicin susceptibility of Pseudomonas aeruginosa. Antimicrob. Agents Chemother. 14: Reller, L. B., F. D. Schoenknecht, M. A. Kenny, and J. C. Sherris Antibiotic susceptibility testing of Pseudomonas aeruginosa: selection of a control strain and criteria for magnesium and calcium content in media. J. Infect. Dis. 130: Washington, J. A. II, R. J. Snyder, P. C. Kohner, C. G. Wiltse, D. M. listrup, and J. T. McCall Effect of cation content of agar on the activity of gentamicin, tobramycin, and amikacin against Pseudomonas aeruginosa. J. Infect. Dis. 137: Woolfrey, B. F., J. M. K. Fox, R. T. Laily, and C. 0. Quall Broth microdilution testing of Pseudomonas aeruginosa and aminoglycosides: need for employing dilutions differing by small arithmetic increments. J. Clin. Microbiol. 16: Woolfrey, B. F., J. M. Fox, and C. 0. Quail A comparison of minimum inhibitory concentration values determined by three antimicrobic dilution methods for Pseudomonas aeruginosa. Am. J. Clin. Pathol. 75: Woolfrey, B. F., J. M. Fox, and C. 0. Quall An analysis of error rates for disc agar-diffusion testing of Pseudomonas aeruginosa versus aminoglycosides. Am. J. Clin. Pathol. 75: Woolfrey, B. F., R. T. Lally, and C. 0. Quail Comparative evaluation of the Micro-Media System, Sceptor, and MIC-2000 microdilution methods for testing Pseudomonas aeruginosa against gentamicin, tobramycin, and amikacin. J. Clin. Microbiol. 17: Woolfrey, B. F., C. 0. Quail, and J. M. Fox Inability of the API-20E system to speciate the fluorescent group of pseudomonads. Am. J. Clin. Pathol. 75: Woolfrey, B. F., W. A. Ranadel, and C. 0. Quall Inability of the standardized disk agar-diffusion test to measure susceptibility of the fluorescent group of pseudomonads to gentamicin. Am. J. Clin. Pathol. 70: Woolfrey, B. F., W. M. Ramadei, and C. 0. Quall Evaluation of the moving intermediate zone concept for determining susceptibility of pseudomonads to gentamicin by the standardized disk agar-diffusion test. Am. J. Clin. Pathol. 72: