Abstract
Prof. Andrew Smith and colleagues (Smith et al. 2020) focus on the adverse reaction (AR) data limitations listed by the Medicines and Health Regulatory Agency (MHRA) but omit to mention the crucial second sentence of item 1: “This is because we have limited information about how many people have taken the medicine without experiencing a reaction.” By combining each antibiotic’s prescribing data with its AR data, we accounted for this (Thornhill et al. 2019). MHRA point 1 was, therefore, addressed by the study methodology and the other points in the Limitations section. We also recognize that an AR report in the Interactive Drug Analysis Profile does not necessarily mean that medication caused the reaction.
The prescribing data were for England and corrected annually for population size to match the AR data, which were for the United Kingdom. As stated in the Methods section, they did not capture private prescribing. Private practice, however, does not correlate with private prescribing. The majority of antibiotic prescribing occurs during National Health Service (NHS)–covered emergency and essential dental treatment rather than cosmetic and nonessential dentistry, which accounts for most private practice. Furthermore, the AR risk with private and NHS prescriptions should not substantially differ.
Prof. Smith and colleagues (2020) also focused on the higher AR rate observed with phenoxymethylpenicillin than amoxicillin. The reasons for this difference are unclear but probably relate to the different minor antigenic determinants of each. The role of these in penicillin allergy has been extensively studied (Blanca 1995; Adkinson et al. 2018) and is thought to explain why some individuals are allergic to phenoxymethylpenicillin but not amoxicillin (Blanca et al. 1990) and vice versa (Blanca 1995). The article’s length constraints prevented us fully addressing this and other interesting topics, including narrow-spectrum antibiotics and antimicrobial stewardship.
Despite their limitations, studies like this provide the best estimates currently available of the AR risk with different antibiotics.
