Although 8-20% of patients have penicillin allergy labels (PAL), less than 1% of the population are truly allergic. The extent to which a PAL persists in the EMR despite documented penicillin tolerance is currently unknown.
The synthetic derivative (de-identified version) of the electronic medical record (EMR) was mined for patients >18 years or older with >3 visits linked to Vanderbilt ambulatory care from January 2000 to August 2014. Key outcomes including antibiotic utilization and presence of C. difficile infection were compared between cases with (n=11,504) and controls (n=31,084) without PAL. Cases were examined for the persistence of the PAL despite documented tolerance. Categorical variables were analyzed by Pearson Chi-squared test and continuous data by Wilcoxon signed rank test.
Most PAL (67%) were already labeled upon entry into the EMR, and 96% remained persistently labeled. Cases were more likely to develop C. difficile infection (1.2% vs 0.9%, p=0.001). The proportion of prescription encounters for levofloxacin (15% vs 12%), vancomycin (5% vs 4%), clindamycin (8% vs 4%), and aztreonam (1% vs <0.1%) were overrepresented in PAL cases versus unlabeled controls (all p<0.001). Of 11216 PAL, 4321 (39%) had EMR documentation of having received and tolerated a penicillin, however despite this 4045/4321 (94%) retained the PAL.
In this largely ambulatory population, PALs persist within the EMR despite proven tolerance and are associated with higher risk antibiotic treatments and C. difficile infection. Major reform of the EMR to utilize systematic approaches of documenting, reconciling, and removing the PAL is urgently needed.