Is vancomycin ototoxicity a significant risk?

Forouzesh and colleagues recently reported an increase in vancomycin ototoxicity in parallel with the increasing use of higher vancomycin doses (4). The report suggests an overall incidence of vancomycin-induced ototoxicity of 12%, doubling historic estimates (3). This is a surprisingly high estimate for a drug that many experts do not consider to be ototoxic (5, 6). The authors have outlined several limitations to their nonrandomized, retrospective review, yet several other limitations must be taken into consideration. First, the premise of this study was to highlight the importance of toxicity during an era of aggressive dosing for vancomycin. The authors have reported vancomycin ototoxicity to be dose dependent as justification for their review; however, this is contrary to consensus guidelines that have recently been published (5). Serum vancomycin levels do not correlate with ototoxicity, presumably why the authors found no differences in vancomycin trough levels between patients with and without hearing impairment. Similarly, the vancomycin doses for the patients with audiogram changes do not reflect a more aggressive dosing trend at this institution. Only one patient received more than 2 grams of vancomycin per day and that patient had a range of vancomycin trough levels of 8.8 to 9.8 mg/liter. This pattern of vancomycin prescribing should not be considered a reflection of the era of increasing doses. Second, the authors have documented hearing impairment on an ordinal scale, defining categories as normal, mild, moderate, severe, and profound. However, Table 3 in the study of Forouzesh et al. (4) makes reference to categories that have not been defined, such as mild to moderate. This type of scale is also misleading. For instance, a patient may have a change of 2 decibels (dB) from baseline to follow-up but still meet the definition for a worsening audiogram. This is significant, because a change of 5 dB is normal variability in an alert, oriented patient (1). Similarly, there can be variability in the instrumentation used; a confounder may produce evidence of hearing loss by technical failure alone (2). Certainly, tremendous variability is possible given these inconsistencies in combination with background noise and the use of ordinal categories to report results. Categorizing hearing loss is not an accurate method of measuring the ototoxic effects of a drug; instead it is suggested to report a change from baseline to follow-up. This has been reported previously, using a change of 30 dB, at any frequency, to define hearing loss (3). Third, the inclusion of patients on concomitant ototoxic medications does not allow for an unbiased evaluation on the effects of vancomycin. The exclusion of the three patients who received additional ototoxic agents decreases the incidence of vancomycin-induced ototoxicity from 12% to 8.9%. It is agreed that ototoxicity from antibiotics has potential quality-of-life implications, as the authors allude, but we find it difficult to assess these implications given the methodology used. Because baseline audiometry is not routine in most facilities, it is important to provide a clinical evaluation of the patients included in this study to extrapolate the results. Undoubtedly, a control group is necessary to elicit a more accurate assessment of the ototoxic effects of vancomycin. Further studies are clearly needed to control for previously mentioned variables to predict an accurate incidence of vancomycin ototoxicity and associated determinants of risk. While an investigation in the modern era is indeed warranted, the conclusion of a significant risk is not.