Abstract

Objective: A recently proposed guideline from the NCCLS recommends that results only from the first isolate of a species per patient be used in calculation of percentage susceptibilities to antimicrobial agents. Because this is apparently based on the comparison of various calculation methods for results for oxacillin against a fairly small number of isolates of Staphylococcus aureus, we have applied these methods to a wider range of antibiotic/organism combinations.

Methods: Antibiotic susceptibility results from our hospital laboratory database were analysed. Rates of antimicrobial susceptibility were calculated using the various criteria proposed by the NCCLS, including exclusion of results from duplicate isolates and surveillance specimens from the calculations.

Results and conclusion: Analysis of results for methicillin against S.aureus, gentamicin against Klebsiella spp., vancomycin against enterococci (all in-patient specimens), and amoxicillin and cefuroxime against Escherichia coli (general practice specimens) confirm that, if duplicates and surveillance specimens are excluded, results obtained with the various patient- and episode-based methods for the calculation of percentage susceptibility are very similar. Because of its simplicity and unambiguity, we agree with the suggestion of the NCCLS group that results for the ‘first isolate of a given species per patient per analysis period, irrespective of body site, antimicrobial susceptibility profile or other phenotypic characteristics’ should be used in the calculation of susceptibility frequencies.

Received 30 April 2002; returned 16 July 2002; revised 9 August 2002; accepted 28 August 2002

Introduction

We have reported previously the effects on apparent resistance frequencies of different criteria for the classification of isolates as duplicates.1 A recently proposed guideline2 from the NCCLS on the analysis and presentation of cumulative antimicrobial susceptibility test data recommends that results from the first isolate only of a species per patient be used in the calculation of percentage susceptibilities. This is apparently based on the comparison of various calculation methods for results for oxacillin against a fairly small number of isolates of Staphylococcus aureus and information that is not given in the NCCLS document. We have, therefore, used these methods on our results for the same antibiotic/organism combinations as in our earlier paper, to see whether the recommendation is more generally applicable.

Materials and methods

Antibiotic susceptibility results from the Microbiology Computer System in the laboratory of the Guy’s and St Thomas’ Hospitals NHS Trust were analysed. Duplicate isolates were detected by a computerized system on the basis of patients’ names, identification numbers and organism identification. In some cases the interval between specimens was also taken into account (see below). Except where explicitly indicated otherwise, surveillance specimens (e.g. those in which methicillin-resistant isolates of S. aureus were sought, but methicillin-susceptible isolates were ignored) were excluded from the calculations. The comparative method3 was used for the determination of susceptibility of antimicrobial agents until the autumn of 2000, after which the British Society for Antimicrobial Chemotherapy method4 was introduced. Intermediate categories of susceptibility or resistance were not used in either method.

Various calculation methods were compared in Appendix B of the NCCLS document.2 These are: (i) all isolates, which includes results for all isolates of a given species, even those from patients with multiple isolates; (ii) first isolate per patient, which takes only the result for the first isolate of a given species recovered from each patient during the study period, regardless of susceptibility profile, body source or specimen type; (iii and iv) most resistant, or most susceptible, interpretation per patient, which calculates rates based on only the most resistant or susceptible result for each antimicrobial agent tested among all isolates of a given species from an individual patient; (v) average result, for which the average susceptibility result for each drug is calculated for each patient and these are then used to calculate the overall average percentage susceptibility; (vi) first isolate per episode (7 day interval or 30 day interval), which considers only the first isolate of a given species recovered from each episode of infection, with an episode defined as the set of all isolates from a patient in which the interval between consecutive isolates is less than or equal to 7 days or 30 days, respectively.

Results and discussion

The results for methicillin susceptibility in isolates of S. aureus from inpatients from 1995–2001 are summarized in Figure 1, and results for 2000 are given in Table 1. Inclusion of all isolates gives the lowest susceptibility frequencies. Results calculated on the basis of ‘first isolate per patient’, ‘average result’ and ‘first isolate per episode’ with a 30 day interval were always within the range bounded by the most resistant and most susceptible interpretations per patient. Inclusion of surveillance specimens resulted in a reduction in the calculated susceptibility frequencies.

Results for gentamicin susceptibility in isolates of Klebsiella spp. from in-patients from 1995–2001 are summarized in Figure 2, and results for 2000 are given in Table 1. The effects of different calculation methods were generally similar to those found for methicillin and S. aureus. However, the susceptibility frequency calculated for first isolate per episode with a 30 day interval were slightly lower than those for most resistant interpretation per patient for 1998 and 1999 when results for surveillance specimens were excluded from the calculations.

There were few, if any, surveillance specimens for vancomycin resistance in enterococci from in-patients or for amoxicillin and cefuroxime resistance in Escherichia coli from general practice patients. Consequently, surveillance specimens were excluded for these organism/antibiotic combinations. For vancomycin against enterococci, most of the results, except those for all isolates, were within the range bounded by the most resistant and most susceptible interpretations per patient (Figure 3, Table 1). All the results were close for amoxicillin or cefuroxime against E.coli (Figure 4, Table 1); indeed for cefuroxime they overlap so closely that four of the seven lines are obscured.

Our results confirm that it is appropriate for the NCCLS to recommend the exclusion of results from surveillance specimens and duplicate isolates from the same patient in the calculation of resistance rates. With one exception, we also agree that results obtained with the various patient- and episode-based methods for calculation of percentage susceptibility are very similar and that ‘differences between these analysis options should generally not have a significant impact on empiric therapy decisions or for following underlying trends in resistance’.2

Our one reservation is that, for our results, the use of a 7 day interval to define an episode often gives percentage susceptibilities lower than those based on the most resistant interpretation per patient. It also produces fewer duplicates than other methods, suggesting that 7 days is too short a cut-off period for a single episode of infection or colonization.

The effect of exclusion of results for surveillance specimens is smaller if results for the first isolate are compared than if results for all isolates are compared (Figures 1 and 2, Table 1). Nevertheless there was an 8% difference for methicillin against S. aureus and a 6% difference for gentamicin against Klebsiella spp. in 2000 (Table 1), so the recommendation of the NCCLS is justified.

Because of its simplicity and unambiguity, we agree with the suggestion of the NCCLS group that results for the ‘first isolate of a given species per patient per analysis period, irrespective of body site, antimicrobial susceptibility profile or other phenotypic characteristics’ should be used in the calculation of susceptibility frequencies. The method applied should be stated along with the results.

*

Corresponding author. Tel. +44-207-922-8383; Fax: +44-207-928-0739; E-mail kevin.shannon@kcl.ac.uk

Figure 1. Comparison of different calculation methods for susceptibility of S. aureus from in-patients to methicillin. Key: filled circles, first isolate per patient; filled squares, all isolates; filled inverted triangles, most resistant interpretation per patient; filled triangles, most susceptible interpretation per patient; open squares, average result; open triangles, first isolate per episode (7 day interval); open circles, first isolate per episode (30 day interval).

Figure 1. Comparison of different calculation methods for susceptibility of S. aureus from in-patients to methicillin. Key: filled circles, first isolate per patient; filled squares, all isolates; filled inverted triangles, most resistant interpretation per patient; filled triangles, most susceptible interpretation per patient; open squares, average result; open triangles, first isolate per episode (7 day interval); open circles, first isolate per episode (30 day interval).

Figure 2. Comparison of different calculation methods for susceptibility of Klebsiella spp. from in-patients to gentamicin. Key: solid circles, first isolate per patient; solid squares, all isolates; solid inverted triangles, most resistant interpretation per patient; solid triangles, most susceptible interpretation per patient; open squares, average result; open triangles, first isolate per episode (7 day interval); open circles, first isolate per episode (30 day interval).

Figure 2. Comparison of different calculation methods for susceptibility of Klebsiella spp. from in-patients to gentamicin. Key: solid circles, first isolate per patient; solid squares, all isolates; solid inverted triangles, most resistant interpretation per patient; solid triangles, most susceptible interpretation per patient; open squares, average result; open triangles, first isolate per episode (7 day interval); open circles, first isolate per episode (30 day interval).

Figure 3. Comparison of different calculation methods for susceptibility of enterococci from inpatients to vancomycin. Key: filled circles, first isolate per patient; filled squares, all isolates; filled inverted triangles, most resistant interpretation per patient; filled triangles, most susceptible interpretation per patient; open squares, average result; open triangles, first isolate per episode (7 day interval); open circles, first isolate per episode (30 day interval).

Figure 3. Comparison of different calculation methods for susceptibility of enterococci from inpatients to vancomycin. Key: filled circles, first isolate per patient; filled squares, all isolates; filled inverted triangles, most resistant interpretation per patient; filled triangles, most susceptible interpretation per patient; open squares, average result; open triangles, first isolate per episode (7 day interval); open circles, first isolate per episode (30 day interval).

Figure 4. Comparison of different calculation methods for susceptibility of E. coli from general practice patients to amoxicillin and cefuroxime. Key: filled circles, first isolate per patient; filled squares, all isolates; filled inverted triangles, most resistant interpretation per patient; filled triangles, most susceptible interpretation per patient; open squares, average result; open triangles, first isolate per episode (7 day interval); open circles, first isolate per episode (30 day interval).

Figure 4. Comparison of different calculation methods for susceptibility of E. coli from general practice patients to amoxicillin and cefuroxime. Key: filled circles, first isolate per patient; filled squares, all isolates; filled inverted triangles, most resistant interpretation per patient; filled triangles, most susceptible interpretation per patient; open squares, average result; open triangles, first isolate per episode (7 day interval); open circles, first isolate per episode (30 day interval).

Table 1.

 Comparison of the effect of different calculation methods on susceptibility frequencies

Antibiotic, organism and calculation method Number of isolates Susceptible (%) 
Methicillin susceptibility in S. aureus from in-patients in 2000 (excluding surveillance specimens) 
all isolates 6382 42.2 
first isolate per patient 2355 59.1 
average result 2355 57.6 
most resistant interpretation per patient  2355 55.1 
most susceptible interpretation per patient  2355 60.9 
first isolate per episode (7 day interval) 3259 52.1 
first isolate per episode (30 day interval) 2591 57.6 
Methicillin susceptibility in S. aureus from in-patients in 2000 (including surveillance specimens) 
all isolates 9912 27.2 
first isolate per patient 2706 51.0 
average result 2706 49.1 
most resistant interpretation per patient  2706 53.1 
most susceptible interpretation per patient  2706 49.1 
first isolate per episode (7 day interval) 4048 41.4 
first isolate per episode (30 day interval) 3015 49.0 
Gentamicin susceptibility in Klebsiella from in-patients in 2000 (excluding surveillance specimens) 
all isolates 1220 58.6 
first isolate per patient 557 76.7 
average result 557 75.1 
most resistant interpretation per patient  557 72.5 
most susceptible interpretation per patient  557 78.8 
first isolate per episode (7 day interval) 763 67.6 
first isolate per episode (30 day interval) 629 73.8 
Gentamicin susceptibility in Klebsiella from in-patients in 2000 (including surveillance specimens) 
all isolates 1503 47.6 
first isolate per patient 599 70.3 
average result 599 68.4 
most resistant interpretation per patient  599 65.9 
most susceptible interpretation per patient  599 73.0 
first isolate per episode (7 day interval) 867 58.9 
first isolate per episode (30 day interval) 685 66.7 
Vancomycin susceptibility in Enterococcus from in-patients in 2000 (excluding surveillance specimens) 
all isolates 1802 82.9 
first isolate per patient 1003 90.9 
average result 1003 90.2 
most resistant interpretation per patient  1003 88.1 
most susceptible interpretation per patient  1003 92.5 
first isolate per episode (7 day interval) 1260 87.8 
first isolate per episode (30 day interval) 1093 89.8 
Amoxicillin susceptibility in E. coli from GP patients in 2000  
all isolates 4813 54.4 
first isolate per patient 4236 55.7 
average result 4236 55.5 
most resistant interpretation per patient  4236 54.5 
most susceptible interpretation per patient  4236 56.6 
first isolate per episode (7 day interval) 4769 54.6 
first isolate per episode (30 day interval) 4609 55.1 
Cefuroxime susceptibility in E. coli from GP patients in 2000 
all isolates 4813 98.98 
first isolate per patient 4236 99.03 
average result 4236 99.04 
most resistant interpretation per patient  4236 98.98 
most susceptible interpretation per patient  4236 99.08 
first isolate per episode (7 day interval) 4769 98.99 
first isolate per episode (30 day interval) 4609 99.02 
Antibiotic, organism and calculation method Number of isolates Susceptible (%) 
Methicillin susceptibility in S. aureus from in-patients in 2000 (excluding surveillance specimens) 
all isolates 6382 42.2 
first isolate per patient 2355 59.1 
average result 2355 57.6 
most resistant interpretation per patient  2355 55.1 
most susceptible interpretation per patient  2355 60.9 
first isolate per episode (7 day interval) 3259 52.1 
first isolate per episode (30 day interval) 2591 57.6 
Methicillin susceptibility in S. aureus from in-patients in 2000 (including surveillance specimens) 
all isolates 9912 27.2 
first isolate per patient 2706 51.0 
average result 2706 49.1 
most resistant interpretation per patient  2706 53.1 
most susceptible interpretation per patient  2706 49.1 
first isolate per episode (7 day interval) 4048 41.4 
first isolate per episode (30 day interval) 3015 49.0 
Gentamicin susceptibility in Klebsiella from in-patients in 2000 (excluding surveillance specimens) 
all isolates 1220 58.6 
first isolate per patient 557 76.7 
average result 557 75.1 
most resistant interpretation per patient  557 72.5 
most susceptible interpretation per patient  557 78.8 
first isolate per episode (7 day interval) 763 67.6 
first isolate per episode (30 day interval) 629 73.8 
Gentamicin susceptibility in Klebsiella from in-patients in 2000 (including surveillance specimens) 
all isolates 1503 47.6 
first isolate per patient 599 70.3 
average result 599 68.4 
most resistant interpretation per patient  599 65.9 
most susceptible interpretation per patient  599 73.0 
first isolate per episode (7 day interval) 867 58.9 
first isolate per episode (30 day interval) 685 66.7 
Vancomycin susceptibility in Enterococcus from in-patients in 2000 (excluding surveillance specimens) 
all isolates 1802 82.9 
first isolate per patient 1003 90.9 
average result 1003 90.2 
most resistant interpretation per patient  1003 88.1 
most susceptible interpretation per patient  1003 92.5 
first isolate per episode (7 day interval) 1260 87.8 
first isolate per episode (30 day interval) 1093 89.8 
Amoxicillin susceptibility in E. coli from GP patients in 2000  
all isolates 4813 54.4 
first isolate per patient 4236 55.7 
average result 4236 55.5 
most resistant interpretation per patient  4236 54.5 
most susceptible interpretation per patient  4236 56.6 
first isolate per episode (7 day interval) 4769 54.6 
first isolate per episode (30 day interval) 4609 55.1 
Cefuroxime susceptibility in E. coli from GP patients in 2000 
all isolates 4813 98.98 
first isolate per patient 4236 99.03 
average result 4236 99.04 
most resistant interpretation per patient  4236 98.98 
most susceptible interpretation per patient  4236 99.08 
first isolate per episode (7 day interval) 4769 98.99 
first isolate per episode (30 day interval) 4609 99.02 

References

1.
Shannon, K. P. & French, G. L. (
2002
). Antibiotic resistance: effect of different criteria for classifying isolates as duplicates on apparent resistance frequencies.
Journal of Antimicrobial Chemotherapy
 
49
,
201
–4.
2.
National Committee for Clinical Laboratory Standards. (
2000
). Analysis and Presentation of Cumulative Antimicrobial Susceptibility Test Data—Proposed Guideline M39-P. NCCLS, Wayne, PA, USA.
3.
Report of the Working Party on Sensitivity Testing of the British Society for Antimicrobial Chemotherapy. (
1991
). A guide to sensitivity testing.
Journal of Antimicrobial Chemotherapy
 
27
, Suppl. D,
1
–50.
4.
Andrews, J. M. for the BSAC Working Party on Susceptibility Testing. (
2001
). BSAC standardized disc susceptibility testing method.
Journal of Antimicrobial Chemotherapy
 
48
, Suppl. 1,
43
–57.