Disk Diffusion Method for Susceptibility Testing of

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, Aug. 1991, p /91/ $02.00/0 Copyright 1991, American Society for Microbiology Vol. 29, No. 8 Disk Diffusion Method for Susceptibility Testing of Neisseria gonorrhoeae SIGNE RINGERTZ,* MARGARETA RYLANDER, AND GORAN KRONVALL Department of Clinical Microbiology, Karolinska Hospital & Karolinska Institute, Stockholm S , Sweden Received 27 February 1991/Accepted 30 April 1991 The standard medium for disk diffusion and MIC testing of Neisseria gonorrhoeae (that of the National Committee for Clinical Laboratory Standards) was tested to establish zone correlations for the MIC breakpoints currently used in Sweden. Eight gonococcal control strains representing both susceptible and resistant strains and 50 clinical isolates were tested. The standard medium did not support the growth of two control strains and three clinical isolates when the standardized inoculum was used in the disk diffusion test. The same medium with the addition of hemoglobin was introduced. This medium supported the growth of all strains. The correlations between the MICs and the zones of inhibition were calculated for penicillin, ampicillin, cefuroxime, erythromycin, tetracycline, doxycycline, ciprofloxacin, and spectinomycin. The range of MICs for the clinical isolates were broad, without bimodal distribution, for all antibiotics, except ciprofloxacin and spectinomycin. With the susceptibility distribution of MICs and zones near the current susceptible and intermediate or intermediate and resistant limits, a low reproducibility of tests and a high frequency of minor interpretive errors can be expected. A revision of MIC breakpoints seems warranted but can only be done after renewed clinical evaluation of different treatment regimens. A standardized method is the basis for accurate and reproducible results for both dilution and diffusion antibiotic susceptibility testing of bacteria. Standard methods for nonfastidious bacteria are available from the National Committee for Clinical Laboratory Standards (NCCLS) and from other national reference authorities, such as the Swedish Reference Group for Antibiotics (SRGA) (1, 6, 10). Fastidious bacteria requiring other media to support their growth also require other standards for susceptibility testing. Recently, a report on the standardization of disk diffusion and agar dilution susceptibility tests for Neisseria gonorrhoeae was published (4). For the disk diffusion and agar dilution tests, a standard medium and a standardized methodology were presented, as were interpretive zone breakpoints and quality control guidelines. The disk diffusion test is a convenient antibiotic susceptibility test with an acceptable accuracy when species-specific interpretive breakpoints are applied. We therefore tested the standardized disk diffusion method for N. gonorrhoeae as described by Jones et al. (4) in our laboratory. Current methods and criteria used for antimicrobial susceptibility testing of gonococci in Sweden have been based on dilution methods only. The reference method is MIC determination on GC agar base supplemented with horse serum and defibrinated horse blood and, for routine testing, the agar dilution screen method with two plates is recommended. Only a few laboratories use the disk diffusion method for susceptibility testing of gonococci, and no standardized method or interpretive breakpoints are available from the SRGA reference authority. In preliminary experiments, it became clear that the standard medium recommended by the NCCLS, GC agar with 1% defined GC supplement (4), did not support the growth of all clinical isolates. In addition, the reading of inhibition zones was more difficult on this medium than on other media and supplements. The present investigations * Corresponding author were therefore performed to compare the NCCLS GC medium with another combination of medium and supplement selected as an alternative standard for disk diffusion testing of gonococci. Tentative interpretive breakpoints for susceptibility categories were calculated. MATERIALS AND METHODS Bacterial strains. Eight gonococcal control strains (ATCC 49226, WHO A, WHO C, and five clinical isolates) were selected to represent both susceptible and resistant strains for each antibiotic tested. No strains resistant to ciprofloxacin were available. All strains showed either the same level of susceptibility to spectinomycin or were highly resistant. The Staphylococcus aureus ATCC quality control strain represented nonfastidious bacteria. Enterococcus faecalis ATCC was used to test medium antagonism of trimethoprim-sulfamethoxazole (6). Fifty consecutive N. gonorrhoeae clinical isolates from cultures isolated in 1990 were also used in comparisons between the media and used to test the selected zone diameter breakpoints for susceptibility category interpretations. MIC limits. The MIC limits for susceptibility categories for N. gonorrhoeae were obtained from Jones et al. (4). For other antibiotics, the MIC limits issued by the SRGA and the Swedish Reference Group for Sexually Transmitted Diseases were applied. For ciprofloxacin, tentative MIC breakpoints were based on reports of strains with reduced susceptibility (2) (Table 1). The definition of susceptibility categories by the SRGA has been based on clinical outcome, in that susceptible means >95% cure and resistant means <90% cure for uncomplicated genital gonorrhea. MIC breakpoint interpretive categories have generally been based on clinical results following single-dose treatment (penicillin, ampicillin, and cefuroxime) and are accordingly lower than those for other bacteria. For other antibiotics (erythromycin, doxycycline, and trimethoprim-sulfamethoxazole), the MIC breakpoints have been the same as those for other bacteria. Antimicrobial agents and disks. The antimicrobial agents

2 VOL. 29, 1991 SUSCEPTIBILITY TESTING OF N. GONORRHOEAE 1605 TABLE 1. Antimicrobial susceptibility testing of N. gonorrhoeaea Antibiotic Disk (,u.g) MIC limit (,ug/ml) Zone diam breakpoint (mm) Criterion Susceptible (s) Resistant (2) Susceptible (2) Resistant (s) source Penicillin NCCLS and Sweden Ampicillin Sweden Cefuroxime Sweden Erythromycin NCCLS and Sweden Tetracycline NCCLS Doxycycline Sweden Ciprofloxacin Spectinomycin NCCLS a Interpretive MIC limits and antimicrobial disk contents for disk diffusion tests are shown. The MIC limits are those of Jones et al. and NCCLS (4, 6) and from Sweden. Tentative ciprofloxacin interpretive MIC limits were defived from the data of Jephcott and Turner (2). Interpretive zone breakpoints for Hb+ GC medium were calculated by regular regressions analysis or by SCA for spectinomycin. tested and their corresponding disks are shown in Table 1. Disks for routine use were purchased from AB Biodisk (Solna, Sweden). Disks containing 100,ug of spectinomycin and disks containing antimicrobial agents with contents other than those in routine disks were prepared by the investigators. All antimicrobial agents were obtained as dry experimental substances with known potencies. The sources of these reagents were as follows: ampicillin, Hoechst AG, Frankfurt, Germany; penicillin and erythromycin, Astra Research Centre, Sodertalje, Sweden; cefuroxime, Glaxo Group Research, Greenford, Middlesex, United Kingdom; tetracycline, Kabi Pharma, Solna, Sweden; doxycycline, Pfizer Inc., New York, N.Y.; ciprofloxacin, Bayer, Leverhusen, Germany; and spectinomycin, The Upjohn Co., Kalamazoo, Mich. Test media. Two different growth media were tested: GC agar base II (batch K2DWPE; BBL) supplemented with 1% IsoVitaleX (BBL) (Hb- GC medium) and the same medium with the addition of 1% hemoglobin (BBL) (Hb+ GC medium). With the exception that IsoVitaleX contains cysteine, Hb- GC medium was equal to the NCCLS medium (4). The cysteine component was shown to cause the inactivation of carbapenems but had no adverse effects on penicillin, ceftriaxone, tetracycline, or spectinomycin (4). MIC determinations. MIC determinations were performed by the agar dilution method on both test media. The inoculum was prepared by suspending growth from an overnight hematin agar plate in phosphate buffer solution containing 0.1% Tween 80 to a concentration of 107 to 108 CFU/ml, corresponding to a 1.0 McFarland standard. The plates were inoculated with a multipoint applicator, delivering 104 to 105 CFU per spot. The control strains were each tested on three different days with log2 dilutions of antimicrobial agents and with one intermediate step, and the geometric means were calculated. The plates were incubated in 5% CO2 at 37 C for 20 to 22 h. The MIC was determined as the lowest concentration that permitted the growth of no more than one colony or a faint haze within the area of inoculation. MIC determinations for the 50 clinical isolates were performed as a single test for each strain and with log2 dilution steps. Disk diffusion testing. Disk diffusion testing was performed on the two GC media as described above with an inoculum of 106 to 107 CFU/ml, except for two control strains (898 and 8563), for which a 10-fold-higher inoculum was required for adequate growth. The control strains were each tested on three different days, and arithmetric means were calculated. Also, 25 separate tests were performed on four control strains, two with normal growth characteristics (N. gonorrhoeae ATCC and WHO C) and two with higher nutritional requirements (WHO B and 108), as demonstrated by weaker growth. These tests were performed by different technologists, each preparing a new inoculum for each separate strain and test day. The 50 clinical isolates were tested once, also by different technologists, each preparing the inoculum. The three clinical isolates that did not grow on Hb- GC medium with the standard inoculum were not retested. Both media were tested for antagonism against trimethoprim-sulfamethoxazole by disk diffusion (6). The plates were incubated in 5% CO2 at 37 C for 20 to 22 h, except for some clinical isolates requiring 30 h before adequate growth was established on Hb- GC medium. The inhibition zones were read with calipers on Hb+ GC medium from the front of the open plates by reflected light and on the optically clear Hb- GC medium against a black background. MIC-inhibition zone correlations. MIC-zone size correlations and correlation coefficients were calculated by the method of least squares with the MIC as the independent variable x and the zone diameter as the dependent variable y (reverse of U.S. analyses). For each antimicrobial agent, six control strains for which MICs represented a range were included on both test media: for ampicillin, the MIC range was 0.03 to 2,ug/ml (no P-lactamase-producing strains were included); for cefuroxime, it was 0.01 to 1,ug/ml; for cefotaxime, it was to 0.1,ug/ml; for erythromycin, it was 0.05 to 6,ugIml; for tetracycline, it was 0.25 to 8,ug/ml; for doxycycline, it was 0.1 to 3,ugIml; and for ciprofloxacin, it was to 0.1,ug/ml. For spectinomycin, a regular regression line was not calculated, since only strains for which MICs were 8 to 32,ug/ml or >256,ug/ml, with no measurable corresponding zones, were available. Regression lines were also calculated for 50 clinical isolates. The standard curve regression analysis (SCA) method for establishing the relationship between MIC and zone size is based on regression analysis of the correlation between the logarithm of the disk content and zone size (squared for better linearity in the calculations), generally referred to as the standard curve (5, 7). The slope and position of the standard curve express the same relationship between MIC and zone size as the regression line calculated by log2 MIC-zone size comparisons with many strains for which there are different MICs. With the SCA method, only two control strains for which there are different MICs are required, but one strain will suffice in some cases (5, 7). Strains WHO A and SR 39 were used, except for ciprofloxacin, for which strains 1408 and SR 39 were used, and for spectinomycin, for which only strain SR 39 was used, since two suitable control strains for which there are different MICs were not available (Table 2). Four different disk contents

3 1606 RINGERTZ ET AL. J. CLIN. MICROBIOL. TABLE 2. MICs for eight control strains of N. gonorrhoeae representing both susceptible and resistant strains for each antimicrobial agenta Geometric mean MIC (,ug/ml) of: Strain Strmmedium GCmedium Ampicillin Cefuroxime Erythromycin Tetracycline Doxycycline Ciprofloxacin Spectinomycin ATCC Hb Hb WHO A Hb >256 Hb >256 WHO C Hb Hb SR 39 Hb Hb Hb+ > Hb- > Hb Hb Hb Hb Hb Hb s < S. aureus ATCC Hb Hb a Results from tests on two different media, Hb+ GC and Hb- GC, were compared. Strain 1408 is a,b-lactamase producer, and strains 898 and 8563 have weaker growth characteristics. S. aureus represents nonfastidious bacteria. Three determinations were made. were tested for each antimicrobial agent on both media: ampicillin disks, 0.3, 1, 3, and 10 jig; cefuroxime disks, 0.1, 0.3, 1, and 10,ug; erythromycin disks, 5, 15, 50, and 150,ug; tetracycline and doxycycline disks, 3, 10, 30, and 100,ug; ciprofloxacin disks, 0.1, 0.3, 1, and 10 jig; and spectinomycin disks, 30, 100, 300, and 1,000 jig. Zone diameter breakpoints corresponding to MIC limits were calculated as described previously (5, 7). Statistical analysis. The means and standard deviations of inhibition zone diameter values obtained from 25 separate disk diffusion tests were calculated for each control strain and antimicrobial agent. MICs for 50% of strains (MIC50s) and MIC90s were calculated by probit analysis (9). RESULTS MIC determinations. MICs for the eight gonococcal control strains and S. aureus ATCC on the two media are shown in Table 2. The MICs obtained on the two media were not significantly different, except for tetracycline and doxycycline, the MICs of which were one-half to one log2 dilution lower on Hb- GC medium. MIC50s and MIC90s determined by log probit analysis for 50 consecutive clinical isolates are shown in Table 3 (9). The differences between the two media were less than one-half log2 dilution step for all antibiotics tested. For the clinical isolates, there was a wide range of MICs with all antibiotics, except for ciprofloxacin and spectinomycin. Disk diffusion tests. The mean zone diameters for four control strains tested (25 times each) are shown in Table 4. The inhibition zones were in general larger on Hb- GC medium. The differences between the two media were smaller for the two strains with normal growth characteris- TABLE 3. MIC50s and MIC90s of 50 consecutive clinical isolates of N. gonorrhoeae (including 7 P-lactamase-producing strains) from Stockholm in 1990a Antibiotic GC medium mdu 50% 90% Range MIC (,g/ml) Penicillin Hb _4b Hb b Ampicillin Hb b Hb b Cefuroxime Hb Hb Erythromycin Hb Hb Tetracycline Hb Hb Doxycycline Hb Hb Ciprofloxacin Hb Hb Spectinomycin Hb Hb a MICs were calculated by log probit analysis. Results from tests on two different media, Hb+ GC and Hb- GC, were compared. b For,-lactamase-producing strains, >4,ug/ml.

4 VOL. 29, 1991 SUSCEPTIBILITY TESTING OF N. GONORRHOEAE 1607 Strain TABLE 4. Stran Mean zone diameters of growth inhibition of four control strains of N. gonorrhoeae tested 25 separate times by the disk diffusion testa GC Zone diam (mm) for: medium mdu Ampicillin Cefuroxime Erythromycin Tetracycline Doxycycline Ciprofloxacin Spectinomycin ATCC Hb Hb Diff WHO C Hb Hb Diff WHO B Hb Hb Diff Hb Hb Diff a The difference in zone diameter (diff) between the results on two media, Hb- GC and Hb+ GC, is given for comparison. Strains WHO B and 108 have weaker growth characteristics. tics (from 0.2 mm for spectinomycin to 5 to 6 mm for tetracycline, doxycycline, and ciprofloxacin). For the two strains with compromised growth, the differences between the two media were larger with all antibiotics (e.g., up to 10 to 15 mm for tetracycline and doxycycline and 12 mm for ciprofloxacin). The clinical isolates showed a wide range of zone diameters with all antibiotics, except for ciprofloxacin and spectinomycin. A direct comparison between inhibition zones for the two growth media was made by the calculation of regression lines as shown in Fig. 1A and B. The zone diameters with all antibiotics were slightly larger on Hb- GC medium than on Hb+ GC medium. The largest differences were noted for tetracycline, doxycycline, and ciprofloxacin. An inhibition zone of 35 mm with ampicillin on Hb+ GC medium corresponded to 37.3 mm on Hb- GC medium (Fig. 1A). The corresponding values with tetracycline on the two media were 35 and 43.5 mm (Fig. 1B). The precision of the tests measured as standard deviations for the combined 25 measurements of each control strain did not differ significantly between the two media, being approximately 2 to 3 mm for both media and all antibiotics. Ability of the media to support the growth of gonococci. The two test media did not differ significantly in MIC results or precision of the tests. However, two of the control strains did not grow adequately during disk diffusion testing on Hb- GC medium with an inoculum of 106 to 107 CFU/ml, an inoculum which was ideal for the other strains. The 10-foldhigher inoculum required by these strains yielded too dense a growth of the other strains, according to the international collaborative study standard (1). Of the 50 consecutive clinical isolates, 3 did not grow on Hb- GC medium with the standardized inoculum. The higher inoculum required by the NCCLS may account for the better performance reported for Hb- GC medium (4). Hb- GC medium is suboptimal for use in susceptibility testing of gonococci in Sweden, and the alternative medium with 1% hemoglobin is therefore preferred for the determination of interpretive zone diameter breakpoints for routine use. Tests with control strain E. faecalis ATCC and the trimethoprim-sulfamethoxazole disk showed growth of the strain near the disk on both media, suggesting that there are inhibitory substances in the Zone mm Hb+ GC med. / // // / CP Inhibition zone diam., mm Inhibition zone diam., mm FIG. 1. Comparison between inhibition zone diameters for clinical isolates of N. gonorrhoeae on two different growth media for gonococci, Hb+ GC and Hb- GC media, for ampicillin (A) (plactamase positive strains excluded) and for tetracycline (B). The regression lines for zones on Hb- GC medium as the independent x variable are shown (solid lines). For comparison, the regression lines for zones on Hb+ GC medium as the independent variable x were also calculated (broken lines). 0 Hb- GC med.

5 1608 RINGERTZ ET AL. Zone mm MIC ug/mi MIC ug/mi FIG. 2. Ciprofloxacin (A) and tetracycline (B) MICs plotted versus inhibition zones for clinical isolates of N. gonorrhoeae. The regression lines obtained by regular regression analysis for welldefined control strains are shown (solid lines), as are SCA-derived regression lines (dotted lines). MIC limits and zone diameter breakpoints for susceptibility categories are also shown (broken lines). The regression lines for the plots of clinical isolates were similar (data not shown). GC agar base supplemented with IsoVitaleX, irrespective of the hemoglobin content (6). Determination of interpretive zone diameter breakpoints. Two different methods were used for the calculation of interpretive breakpoints on Hb+ GC medium. Regression analysis of the correlation between MIC and inhibition zone diameters with a small number of well-defined gonococcal strains for which there were different, well-defined MICs was performed first (Fig. 2). The regression lines obtained for the clinical isolates tested only once corresponded to these control strain regression lines. SCA was then used for calculating interpretive breakpoints, and the two sets were compared. For ampicillin, the interpretive breakpoints for susceptible and resistant (greater than or equal to for susceptible and less than or equal to for resistant) (in millimeters), in accordance with the MIC limits in Table 1, were determined to be 40 and 21 by regular regression analysis or 40 and 19 by SCA. For cefuroxime, the interpretive breakpoints were 45 and 31 or 45 and 29; for erythromycin, they were 32 and 21 or 32 and 21; for tetracycline, they were 39 and 31 or 38 and 30; for doxycycline, they were 33 and 22 or 31 and 19; and for ciprofloxacin, they were 39 and 32 or 40 and 31. Regular regression analysis could not be performed for spectinomycin, since the range of MICs was too narrow. SCA, however, permitted calculations of spectinomycin interpretive breakpoints to 20 and 13. For penicillin, interpretive breakpoints calculated by linear regression analysis were 43 and 29. The resulting breakpoints determined by the two calculation methods for the various antibiotics were very similar, but the SCA method required much less laboratory effort. Currently, MIC limits and disk contents for spectinomycin in Sweden (susceptible, -16,ug/ml; resistant,.32,ug/ml; 30-pLg disk) are different from those that we have selected. Breakpoints calculated in accordance with the Swedish MIC limits indicated that the 30-,ug disk was not sufficient to distinguish between strains for which the MIC was 32,ug/ml, because the zone of inhibition would be near 6 mm. Also, the MIC breakpoint bisected the homogeneous population of susceptible strains. The parameters that we have adopted from Jones et al. (4) for our studies (susceptible, s32,ug/ml; resistant,.128 pg/ml; 100-,ug disk) are therefore preferable. We suggest that these be adopted in Sweden. Breakpoints corresponding to the 100-,ug spectinomycin disk were calculated by the SCA method. Application of interpretive MIC limits and zone diameter breakpoints. Examples of the interpretation of MICs or zone diameters into susceptibility categories are shown in Fig. 2A and B. For ciprofloxacin, the zone diameters of the strains correctly interpreted these as belonging to the susceptible or intermediate category corresponding to the MIC interpretive breakpoints (Fig. 2A). All clinical strains were correctly identified as susceptible to spectinomycin when the MIC interpretive breakpoint of resistant to.128 txg/ml was applied. For tetracycline, however (Fig. 2B), as well as for the other antibiotics tested, the use of neither MIC limits nor zone diameter breakpoints for susceptibility categories could reproducibly identify strains as belonging to one single category, since the range of the MICs for the clinical isolates was broad and without a bimodal distribution. With the susceptibility distribution of MICs and zones for many strains being near the susceptible and intermediate limits or the intermediate and resistant limits, a low reproducibility of tests and a high frequency of minor interpretive errors can be expected. DISCUSSION J. CLIN. MICROBIOL. There is a need for a simple and reliable routine susceptibility test for gonococci, both for epidemiological surveillance and for the management of patients. A standardized disk diffusion test gives reproducible results and is easy to use even by laboratories with only an occasional need to test isolates. There are several crucial points regarding the standardization of in vitro antimicrobial susceptibility tests: first, the definition of susceptibility categories and the consequent setting of MIC limits; second, the choice of a medium which not only influences the ability to support the growth of all strains and the formation of inhibition zones but also affects the actual MICs for the strains; and third, the determination of interpretive zone diameter breakpoints for disk diffusion testing. Interpretive MIC breakpoints for gonococcal susceptibility categories have been based on single-dose treatment for most drugs, and these limits are accordingly lower than those for other bacteria. For the penicillins and cefuroxime, the majority of the clinical isolates of N. gonorrhoeae are designated intermediately susceptible by these MIC limits used in Sweden. The question of whether these isolates are

6 VOL. 29, 1991 clinically susceptible to the single-dose treatment regimens used today has not been answered, nor has the question of whether a multiple-dose treatment regimen would cure a higher percentage of patients. Given normal treatment regimens and, consequently, MIC limits like those for other bacteria, these isolates would be designated susceptible. Interpretive MIC breakpoints are generally set to fit bimodal distributions of fully susceptible strains and strains with a reduced susceptibility. However, there is no clear bimodal distribution of the current clinical isolates with respect to MICs of several antibiotics tested, but rather a continuum of increased levels of resistance with a wide range of MICs. This lack of bimodality and the overlapping of MICs for strains for which the MICs are close to the breakpoints give rise to problems concerning the reproducibility of in vitro susceptibility testing of gonococci. A high frequency of minor errors of interpretation will occur when there are many strains around the susceptible and intermediate limits or the resistant and intermediate limits, irrespective of whether the method is MIC determination or disk diffusion (4). This problem with every in vitro susceptibility testing method for gonococci has to be accepted as long as MIC breakpoints cut through major populations of clinical strains. A change in MIC limits can only be made after renewed clinical evaluations of different treatment regimens have been performed. For instance, in a recent paper by Jones et al. (3), the interpretive criterion for cefuroxime was susceptible to <1,ug/ml, in contrast to susceptible to 0.06,ug/ml in Sweden. However, the single-dose treatment used against gonococci with cefuroxime in the United States is twice the dosage recommended in Sweden. Inhibition zone diameter breakpoints for susceptibility interpretations are most accurate if the regression line used for the calculation is species specific, as is the case for the present standardization of an N. gonorrhoeae disk diffusion test. The regression lines were calculated in two different ways, and both correlated well with the regression analysis of the less-well-defined clinical isolates. It is therefore of interest to note that the simple method of SCA is capable of setting breakpoints for disk diffusion testing. This method can even be performed in an individual laboratory as a quality control measure to ensure adherence to established standards. In view of the facts considered above regarding MIC susceptibility breakpoints, the disk diffusion method done with the proposed interpretive breakpoints for susceptibility categories is considered to give results as correct as those given by the MIC determination method. The final point to consider and the main issue of the present study is the choice of growth medium. The medium proposed by Jones et al. (4) was tested with the intention that we should adhere to an internationally accepted standard medium for the testing of gonococci. However, the use of this medium, not containing hemoglobin, may not be acceptable because of the auxotropic requirements of some strains and the inability of this medium to support the growth of SUSCEPTIBILITY TESTING OF N. GONORRHOEAE 1609 such strains. The medium proposed here includes hemoglobin, which has also been found suitable as a supplement for the susceptibility testing of Haemophilus influenzae (8). The addition of hemoglobin to the medium for the susceptibility testing of N. gonorrhoeae does not significantly affect the MICs, except for the tetracyclines. The medium which we propose, containing hemoglobin, produces slightly higher MICs than does the NCCLS medium and therefore gives a more conservative susceptibility interpretation with the same MIC breakpoints. ACKNOWLEDGMENT Medical laboratory technician Kerstin Jacobson is acknowledged for skilled technical assistance. REFERENCES 1. Ericsson, H. M., and J. C. Sherris Antibiotic sensitivity testing. Report of an international collaborative study. Acta Pathol. Microbiol. Scand. Sect. B Suppl Jephcott, A. E., and A. Turner Ciprofloxacin resistance in gonococci. Lancet 335: Jones, R. N., P. C. Fuchs, J. A. Washington II, T. L. Gavan, P. R. Murray, E. H. Gerlach, and C. Thornsberry Interpretive criteria, quality control guidelines, and drug stability studies for susceptibility testing of cefotaxime, cefoxitin, ceftazidime, and cefuroxime against Neisseria gonorrhoeae. Diagn. Microbiol. Infect. Dis. 13: Jones, R. N., T. L. Gavan, C. Thornsberry, P. C. Fuchs, E. H. Gerlach, J. S. Knapp, P. Murray, and J. A. Washington II Standardization of disk diffusion and agar dilution susceptibility tests for Neisseria gonorrhoeae: interpretive criteria and quality control guidelines for ceftriaxone, penicillin, spectinomycin, and tetracycline. J. Clin. Microbiol. 27: Kronvall, G Analysis of a single reference strain for determination of gentamicin regression line constants and inhibition zone diameter breakpoints in quality control of disk diffusion antibiotic susceptibility testing. J. Clin. Microbiol. 16: National Committee for Clinical Laboratory Standards Tentative standard M2-T4. Performance standards for antimicrobial disk susceptibility tests. National Committee for Clinical Laboratory Standards, Villanova, Pa. 7. Ringertz, S., and G. Kronvall On the theory of the disk diffusion test. Evidence for a non-linear relationship between critical concentration and MIC, and its practical implications for susceptibility testing of Haemophilus influenzae. APMIS 96: Ringertz, S., B. Olsson-Liljequist, and G. Kronvall Antimicrobial susceptibility testing of Haemophilus influenzae. Improvement of accuracy of the disc diffusion test. J. Antimicrob. Chemother. 26: Schmidt, L. H The MIC-50/MIC-90 assessments of in vitro activities of antimicrobial agents that facilitate comparative agent-agent and agent-species susceptibility comparisons. Antimicrob. Newsl. 4: Swedish Reference Group for Antibiotics Antimicrobial susceptibility testing of bacteria. National Bacteriological Laboratory, Stockholm.