The influence of factors affecting the `critical population' density of inocula on the determination of bacterial susceptibility to antibiotics by disc diffusion methods

Sir, The recent article by Gosden et al. prompted me to reflect on the widespread use of disc diffusion methods for determining antibiotic susceptibility that ignore an important finding of the definitive studies of Cooper and Linton on the formation of inhibition zones. These investigators showed that, although the intrinsic susceptibility of a bacterial isolate, under the in-vitro conditions of the test, is one parameter that influences zone diameter, of even greater importance are factors that determine the time taken for the bacterial inoculum to reach a ‘critical population’ density. The disc diffusion techniques most commonly used (the Kirby–Bauer and Stokes’ tests) take account of important factors, including the inoculum density, the composition of the medium, the delay between application of the disc and incubation, the temperature of incubation, etc., but the resulting zone of inhibition is then compared with that produced by a standard strain and/or predetermined templates, thereby disregarding the growth rate of the isolate being tested. In other words, these methods assume that all clinical isolates exhibit the same rate of growth in vitro as that of the control or ‘average’ strain. Many years ago, I carried out studies relating to the determination of the susceptibilities of Staphylococcus aureus isolates to erythromycin by the Kirby–Bauer method. I observed that, if a fully susceptible strain with a lag phase and a generation time of 30 min each mutated to one with a lag phase of the same duration, but a generation time of 21 min, the latter would appear to be resistant to erythromycin, even though its MIC was the same as that of the parent strain; a change in the reverse direction would result in a resistant organism’s appearing susceptible. Methods which overcome this difficulty by using discs containing different concentrations have been proposed, but have not become popular, despite requiring only slightly more effort than the techniques in common use. Although I would invert the calculations in the second method referred to above, in practice, calculations are seldom needed, a simple, hand-drawn graph yielding a sufficiently accurate result. Moreover, such methods make the need to control many of the other factors associated with susceptibility testing less critical. The preceding discussion poses an obvious question: how variable are the growth kinetics of strains isolated in clinical practice? The answer may be ‘not very variable’, but it is clearly not in a patient’s best interests to be infected with a bacterium that is amenable to eradication by the antibiotic of choice, yet to be denied treatment with that agent because the organism appears to be resistant when susceptibility is determined by the disc diffusion method, owing to its rapid growth rate; the reverse situation is equally undesirable. I am hopeful that my colleagues in Bristol might turn their attentions to this interesting and potentially important issue.