The impact of penicillin allergy labels on antibiotic and health care use in primary care: a retrospective cohort study

Penicillins are the most prescribed antibiotics worldwide [1‐5] due to their high effectiveness, low cost and minimal side effects [6]. Unfortunately, penicillins are also a frequent cause of suspicion of antibiotic allergies [1, 2]. Prevalences of physician or patient reported suspected penicillin allergy (Pen-A) varying from 2 to 16% were reported, however the true prevalence in the general population is unknown, and remains difficult to determine due to varying study populations and study designs [7, 8].

Diagnostic workup for Pen-A includes detailed patient history, skin testing, in vitro testing and oral drug challenge. In Dutch primary care guidelines drug allergy testing is not addressed. Diagnostic workup takes place in specialized academic centers following European Academy of Allergy and Clinical Immunology (EAACI) guidelines [9]. Studies on outpatient adults in secondary and tertiary care showed that an alleged Pen-A can only be confirmed in the minority (16.5–29.0%) of patients [10]. Possible explanations for over reporting of Pen-A might be that non immunological side effects or disease symptoms such as viral exanthema are alleged to drug allergy. Studies in tertiary care showed that Pen-A labeling in patient charts is associated with more use of broad-spectrum antibiotics, particularly macrolides, quinolones, tetracyclin [11] and vancomycin [12, 13]. Use of these reserve antibiotics have been associated with higher risk of treatment failure, adverse drug reactions, complications as Clostridium difficile, MRSA and vancomycin-resistant Enterococcus infections, higher costs and it reinforces the development of antibiotic resistance [11]. Furthermore, Pen-A patients showed longer duration of hospitalization and more readmissions than non-Pen-A patients [11‐13]. Another study in the United States also reported a higher mortality during admission in patients with a Pen-A label [13]. One single recent study conducted in a primary care population in the United States [n = 232,616 patients; 199 general practioners (GPs)] also revealed, that patients with Pen-A documentation received more fluoroquinolones, clindamycin and vancomycin [14].

The impact of Pen-A labels in primary care is however still poorly investigated, while the majority (80%) of antibiotics are prescribed in primary health care [15] and most allergy labels are probably assigned in primary care. The aim of this study was to assess the impact of Pen-A registration in primary care on health care related factors such as the number and type of antibiotics prescribed, second choice antibiotic prescriptions and GP contacts.

Out of the database of 196,440 enlisted patients [52% female; mean (SD) age 38 (22) years] in 2014, 1254 (0.6%) individual patients [66.9% females; mean (SD) age 47 (26) years] with a Pen-A registration were identified with a median duration of follow-up of 4.5 years (IQR 5 years). All patients could be matched with three patients without Pen-A registration resulting in a total study population of 5010 patients. Date of Pen-A registration ranged from 24 November 1989 to 17 February 2014. Atopic comorbidities were significantly more often registered in Pen-A patients compared to non-Pen-A patients (P < 0.05). The most common indications for antibiotic prescription were respiratory (27.5%) and urinary (28.4%) disease followed by skin infections (8.0%) and ear infections (3.6%) without significant difference between both groups (Additional file 1: Table S1). Data concerning indication for antibiotic prescription were missing in 13.9% of the prescriptions (14.2% in the Pen-A group; 13.8% in non-Pen-A group). All other data were complete. For both Pen-A and non-Pen-A patients tetracyclins were the most prescribed second choice antibiotic for respiratory infections. For skin infections it were ‘macrolides, lincosamides and streptogamins’. For ear infections ‘macrolides, lincosamides and streptogamins’ were the most prescribed second choice antibiotic for Pen-A patients, and sulfonamides for non-Pen-A patients (Additional file 2: Table S2a, Additional file 3: Table S2b, Additional file 4: Table S2c).

This is the first study in Europe investigating the impact of a Pen-A registration in primary care. It shows that a Pen-A label is related to a higher use of antibiotics (19% of Pen-A patients received an antibiotic prescription per year compared to 6.9% of non-Pen-A patients), in particular regarding reserve and second choice antibiotics. Furthermore, patients with registration of Pen-A use more primary health care. Remarkably, in contrast to previous studies [11, 14, 23] Pen-A patients did not receive less prescriptions of penicillins compared to non-Pen-A patients in our study. Possible explanation could be that prescribing GPs in our study were aware of the overestimation of Pen-A labeling and consciously prescribed penicillins to Pen-A patients based on their own experience.

Before drawing conclusions from our results, potential limitations should be considered. Although patients with and without Pen-A registration were matched on age, sex, GP practice and follow-up period, results in our study may be explained by residual confounding mainly related to unmeasured comorbidity. We did not have any information concerning comorbidities, antibiotic and health care use before the moment of Pen-A registration at our disposal. Pen-A patients may have more comorbidities, resulting in more antibiotic use and health care use compared to non-Pen-A patients to begin with. This certainly has to be kept in mind when interpreting the results of our study. Another limitation is that our data did not include information about the severity of the allergy reactions or differentiation between immediate and non-immediate allergy.

Although our study did not aim to assess the prevalence of Pen-A labels in primary care, the prevalence of 0.6% is considerably lower than previously reported in the literature (2.0–15.6%) [7, 8, 14, 24]. A recent mono center study in Dutch primary care also showed a very low prevalence of 2.0%, possibly implicating country specific differences [7]. A recent study in an adult primary care population in the United States by Shah et al. [14] observed a high prevalence of 15.6%. This study, included 232,616 patients seen by 199 primary care providers, also showed a high variability in Pen-A documentation between the different primary care providers varying from 7.9 to 24.8% [14]. Therefore differences in reported prevalences might be explained by differences in study designs but also other factors such as the variation in Pen-A documentation between GPs. Stricter antibiotic allergy labeling behavior of the Dutch GPs and the low antibiotic prescribing rate in The Netherlands and in particular in the studied population might contribute to the low prevalence in our study, since a higher antibiotic prescription rate is known to be related to a higher frequency of (suspicion of) antibiotic allergy [12, 13, 25]. In our study only 9.9% of the total study population received at least one antibiotic prescription per patient year compared to 18.2% of the total Dutch population [3]. This might be caused by a superior motivation and attention on following the guidelines more strictly in GPs actively participating in our research network [26]. However, we have no reason to think that this lower prevalence of Pen-A labeling did influence the relation between labeling, antibiotic use and consultation rates.

Our data confirm the results of a recent study conducted in primary care population in the United States (n = 232,616; 199 GPs) reporting that patients with Pen-A registration were of higher risk to receive fluoroquinolones, clindamycin and vancomycin [14]. These reserve antibiotics are associated with higher costs, more adverse reactions and complications and a high risk of developing antimicrobial resistance [27, 28]. Increased usage of macrolides and clindamycin, quinolones and tetracyclin in patients with Pen-A registration was also shown in a recent study in Dutch tertiary care and tertiary care studies from Israel, the United States and the United Kingdom [4, 5, 11, 29]. These studies on hospitalized patients also showed an increased risk of longer duration of hospital admission and re-admission.