Article inclusion and trends

After removing duplicates, 5,164 studies were identified; 234 (4.5%) met our criteria for full-text review, and 4,930 (95.5%) were excluded for reasons summarized in Fig. 1. Overall, 61 studies had pertinent findings and an appropriate control group to allow for data abstraction (Supplementary Material online). Of these 61 studies, 39 were published in peer-reviewed journals, and the remaining 22 were published as conference abstracts (Fig. 1). After contacting abstract authors, 1 abstract had been advanced to peer-reviewed publication and was included in the final analysis.^{Reference Lam, Staicu, Conn and Ramsey16} There was not sufficient additional information for inclusion of the remaining 21 abstracts. Also, 1 study was published shortly after our search and was included in the final analysis.^{Reference Imlay, Krantz and Stohs17} Ultimately, 41 articles met eligibility criteria and were included. Only 1 of these studies was published prior to 2000; 6 studies were published between 2000 and 2010; 7 studies were published between 2011 and 2015; and 27 of these studies were published between 2016 and 2019 (Fig. 2).

Fig. 1.

Flow diagram for study selection on antibiotic allergy. Overall, 5,164 studies were eligible for screening after duplicates were removed. After irrelevant studies were removed, 234 full-text studies were assessed for eligibility; 173 studies were excluded for indicated reasons. Of 61 eligible studies, 39 were published in peer-reviewed journals and ultimately included. Also, 1 study was identified and included after the original screen and another study was advanced from conference abstract to peer-reviewed publication after our screen and was included.

Fig. 2.

Number of studies per year included in the final analysis.

Population, setting, exposure, and study design

Of the 41 studies included for data abstraction, 35 (85%) focused exclusively on adult patients age >18 years; 4 (9.8%) focused solely on pediatrics; and 2 studies (5%) analyzed data from patients of all ages (Table 1). Most studies (n = 34, 83%) investigated inpatient cases, and 7 (17%) analyzed data from the outpatient setting (Table 1). Of the 41 studies, 20 (49%) were conducted in the United States, 8 (20%) were conducted in Australia, 5 (12%) were conducted in countries within the European Union, 3 (7%) were conducted in the United Kingdom, 2 (5%) were conducted in Canada, and 1 study (2%) was included from each of the following: China, Israel, or multinational. The number of study participants ranged from 72 to 301,339. Only 5 studies (12%) were prospective, and the remaining 36 (88%) were retrospective by design. More than half (n = 21, 51%) of studies focused exclusively on penicillin allergy, while 9 (22%) used β-lactam allergy labels as a primary exposure; the remaining 11 studies included either all “antibiotic” or “antimicrobial” allergy labels (Table 1). Broadly, antibiotic prescribing patterns and exposures were the most common domain reported, with 34 studies (83%) analyzing antibiotic exposure by drug class, number of doses administered, and/or total duration of treatment (Table 1 and Table 2). Clinical outcomes, such as LOS, ICU admission, readmission rates, opportunistic infection or MDR, and mortality were examined in 31 studies (76%) (Tables 1 and 3). Healthcare-related costs were the least examined outcome domain; 9 studies (~22%) reported these data (Tables 1 and 4). Also, 27 studies (66%) reported results from >1 domain (Table 1).

Table 1.

Characteristics of included studies. The study population and setting, number of patients, study design, primary exposure (allergy label type) and domains analyzed are indicated for each included study

Table 2.

Antibiotic Exposure and Use in Allergy-Labeled and Control Patients^a

Note. CI, confidence interval; IQR, interquartile range; IRR, incidence rate ratio; OR, odd ratio; RR, relative risk; SD, standard deviation.

^a For each study analyzing antibiotic exposure, effect of allergy labels on antimicrobial administration (column 2) and effect on number of antimicrobials or duration of treatment (column 3) is indicated.

Table 3.

Health Outcomes in Allergy-Labeled and Control Patients^a

Note. CAP, community-acquired pneumonia; ICU, intensive care unit; LOS, length of hospital stay; MDR, multidrug resistant; MRSA, methicillin-resistant Staphylococcus aureus; SSI, surgical site infection; UTI, urinary tract infection; VRE, vancomycin resistant Enterococcus.

^a For each study analyzing clinical outcomes, results on hospitalization characteristics including length of stay (LOS), ICU admission, and hospital readmission are indicated in columns 2–4. Multi-drug resistant or opportunistic infection, if analyzed, is indicated in column 5 and patient mortality in column 6. Other less commonly reported outcomes, such as surgical site infections, are reported in column 7.

Table 4.

Healthcare-Related Costs in Allergy-Labeled and Control Patients^a

Note. AU, Australian dollars; CAP, community-acquired pneumonia; EUR, Euros; USD, United States dollars.

^a Results from studies that reported cost analyses, including inpatient and outpatient antibiotic cost, as well as total cost of hospitalization (cost of admission) are reported in columns 2–4.

Antibiotic allergy labels and antibiotic exposure

Most studies reported at least 1 result suggesting that antibiotic allergy labels increased exposure to broad-spectrum therapy, the number of antibiotics administered, or duration of treatment (Fig. 3, first column). Of the studies examining antibiotic exposure, 32 of 34 (94%) found that allergy-labeled patients received more broad-spectrum antibiotics (eg, vancomycin, linezolid, clindamycin, metronidazole, macrolides, fluoroquinolones and carbapenems) than non–allergy-labeled control patients (Table 2). Several studies that used penicillin or β-lactam allergy as the primary exposure confirmed that allergy-labeled patients received fewer, if any, drugs in these classes (Table 2). Notably, 1 study, which examined guideline adherence in penicillin allergic patients, found that there was increased adherence to guideline-recommended therapy when an allergy label was present.^{Reference Phillips, Gilchrist and Cooke18}

Fig. 3.

Studies grouped by broad results in each of 3 domains: antibiotic exposure and prescribing patters (34 studies), clinical outcomes (31 studies) and healthcare related costs (9 studies).

Additionally, 7 of 10 studies examining inpatients (70%) found that antibiotic allergy labels significantly increased the number of antibiotics administered.^{Reference Macy and Contreras2,Reference Charneski, Deshpande and Smith19–Reference Sade, Holtzer, Levo and Kivity24} The other 3 studies reported equivocal results.^{Reference Imlay, Krantz and Stohs17,Reference Lucas, Arnold and Sommerfield25,Reference Trubiano, Cairns and Evans26} The only 2 outpatient studies to examine number of antibiotics per patient found that allergy-labeled patients received significantly more antibiotic prescriptions over a defined time period.^{Reference Su, Broekhuizen, Verheij and Rockmann27,Reference Kraemer, Caprye-Boos and Berman28} Also, 2 studies examined duration of treatment and both found that antibiotic allergy-labeled patients received longer treatment courses in the hospital and after hospital discharge.^{Reference Huang, Cluzet, Hamilton and Fadugba20,Reference Sade, Holtzer, Levo and Kivity24} Of 3 studies examining adverse drug reactions (and had enough observations to draw meaningful comparisons), 2 found that allergy-labeled patients were more likely to have adverse reactions to alternative therapy.^{Reference Desai, Kaplan, Chen and Macy29,Reference MacFadden, LaDelfa and Leen30} Furthermore, 2 studies demonstrated that patients with a penicillin allergy label were more likely to have other antibiotic allergies.^{Reference Blumenthal, Ryan, Li, Lee, Kuhlen and Shenoy31,Reference Blumenthal, Shenoy and Huang32} Additionally, one study found that allergy-labeled patients were less likely to have antibiotics narrowed while in the hospital, and another study found that more allergy-labeled patients received parenteral therapy than controls.^{Reference van Dijk, Gardarsdottir, Wassenberg, Oosterheert, de Groot and Rockmann23,Reference Conway, Lin and Sellick33}

Antibiotic allergy labels and clinical outcomes

Overall, 31 studies examined the effects of antibiotic allergy labels on clinical outcomes (Table 3). Moreover, 23 (74%) of these studies found that allergy labels negatively influenced at least 1 outcome while having no positive effects. Also, 4 studies (13%) reported equivocal results that did not reach statistical significance. Another 4 studies (13%) had at least 1 result suggesting that allergy labels negatively influenced an outcome, but another suggesting it positively influenced another (Fig. 3, column 2).^{Reference Imlay, Krantz and Stohs17,Reference Charneski, Deshpande and Smith19,Reference Sade, Holtzer, Levo and Kivity24,Reference Blumenthal, Ryan, Li, Lee, Kuhlen and Shenoy31,Reference Blumenthal, Shenoy and Huang32,Reference Ponce, Raines, Reed, Vick, Richman and Hawn34–Reference Sousa-Pinto, Cardoso-Fernandes, Araujo, Fonseca, Freitas and Delgado36}

Of the 31 (71%) studies reporting clinical outcomes, 22 examined some aspect of hospitalization: LOS, ICU admission, or hospital readmission. Of the 20 studies examining LOS, 11 (55%) found that allergy-labeled patients had significantly longer LOS.^{Reference Macy and Contreras2,Reference Charneski, Deshpande and Smith19,Reference Huang, Cluzet, Hamilton and Fadugba20,Reference Lucas, Arnold and Sommerfield25,Reference Desai, Kaplan, Chen and Macy29,Reference MacFadden, LaDelfa and Leen30,Reference Sousa-Pinto, Cardoso-Fernandes, Araujo, Fonseca, Freitas and Delgado36–Reference Turner, Moehring and Sarubbi40} Also, 8 studies found equivocal results and 1 found that allergy-labeled patients were hospitalized for shorter duration.^{Reference Imlay, Krantz and Stohs17,Reference Trubiano, Leung, Chu, Worth, Slavin and Thursky21,Reference van Dijk, Gardarsdottir, Wassenberg, Oosterheert, de Groot and Rockmann23,Reference Sade, Holtzer, Levo and Kivity24,Reference Blumenthal, Ryan, Li, Lee, Kuhlen and Shenoy31–Reference Conway, Lin and Sellick33,Reference Khumra, Chan and Urbancic41,Reference Knezevic, Sprigg and Seet42} Of 6 studies, 1 (17%) found that allergy-labeled patients were more likely to be admitted to the ICU, but all other studies reported no difference (Table 3).^{Reference Charneski, Deshpande and Smith19} Furthermore, 13 studies examined healthcare contact, and 10 of these studies (77%) found that allergy-labeled patients were more likely to be readmitted, to have infection recurrence, and to have more primary-care visits (Table 3).^{Reference Macy and Contreras2,Reference Huang, Cluzet, Hamilton and Fadugba20,Reference Trubiano, Leung, Chu, Worth, Slavin and Thursky21,Reference van Dijk, Gardarsdottir, Wassenberg, Oosterheert, de Groot and Rockmann23,Reference Lucas, Arnold and Sommerfield25,Reference Su, Broekhuizen, Verheij and Rockmann27,Reference Kraemer, Caprye-Boos and Berman28,Reference MacFadden, LaDelfa and Leen30,Reference Turner, Moehring and Sarubbi40,Reference Knezevic, Sprigg and Seet42} Also, 1 study found that allergy-labeled patients were readmitted less than control patients, and the remaining 2 studies had equivocal results (Table 3).^{Reference Charneski, Deshpande and Smith19,Reference Conway, Lin and Sellick33,Reference Khumra, Chan and Urbancic41}

Of the studies examining MDR or opportunistic infections, 10 of 15 (67%) found that allergy-labeled patients were more likely to have an MDR infection or to be colonized with an MDR organism. Of the 9 studies that addressed C. difficile infection specifically, 4 (44%) found that allergy-labeled patients were more likely to experience infection.^{Reference Macy and Contreras2,Reference Huang, Cluzet, Hamilton and Fadugba20,Reference Khumra, Chan and Urbancic41,Reference Blumenthal, Lu, Zhang, Li, Walensky and Choi43} Most studies that analyzed methicillin-resistant Staphylococcus aureus (MRSA) (4 of 4, 100%) and vancomycin-resistant Enterococcus (VRE) infections (2 of 3, 67%), respectively, found that allergy-labeled patients were more likely to develop infection or to be colonized with a resistant organism than control patients (Table 3).^{Reference Macy and Contreras2,Reference Lam, Staicu, Conn and Ramsey16,Reference Blumenthal, Ryan, Li, Lee, Kuhlen and Shenoy31,Reference Sousa-Pinto, Cardoso-Fernandes, Araujo, Fonseca, Freitas and Delgado36,Reference Blumenthal, Lu, Zhang, Li, Walensky and Choi43}

Overall, 13 studies analyzed hospital mortality. Furthermore, 3 studies (23%) found that allergy-labeled patients were more likely to die in the hospital, and 1 study (8%) found that allergy-labels conferred a mortality benefit.^{Reference Charneski, Deshpande and Smith19,Reference Huang, Cluzet, Hamilton and Fadugba20,Reference MacFadden, LaDelfa and Leen30,Reference Sousa-Pinto, Cardoso-Fernandes, Araujo, Fonseca, Freitas and Delgado36} All other studies (n = 8, 69%) reported equivocal results that did not reach statistical significance (Table 3).^{Reference Macy and Contreras2,Reference Trubiano, Leung, Chu, Worth, Slavin and Thursky21,Reference van Dijk, Gardarsdottir, Wassenberg, Oosterheert, de Groot and Rockmann23,Reference Sade, Holtzer, Levo and Kivity24,Reference Blumenthal, Shenoy and Huang32,Reference Turner, Moehring and Sarubbi40–Reference Knezevic, Sprigg and Seet42}

Furthermore, 2 studies found that antibiotic allergy labels were associated with an increased risk of surgical site infection.^{Reference Blumenthal, Ryan, Li, Lee, Kuhlen and Shenoy31,Reference Sun, Wang and Tang44} Also, 1 study that focused on obstetric patients found that there were more cesarean deliveries in a large cohort of mothers with a penicillin allergy label than among control patients (Table 3).^{Reference Desai, Kaplan, Chen and Macy29} Other less frequently reported outcomes are included in the “other” column of Table 3.

Antibiotic allergy labels and healthcare costs

Only 9 studies examined the impact of antibiotic allergy labels on healthcare-related costs, and 7 (77.7%) found that allergy labels increased at least 1 cost component (Fig. 3, column 3).^{Reference Macy and Contreras2,Reference Huang, Cluzet, Hamilton and Fadugba20,Reference Sade, Holtzer, Levo and Kivity24,Reference Kraemer, Caprye-Boos and Berman28,Reference Sousa-Pinto, Cardoso-Fernandes, Araujo, Fonseca, Freitas and Delgado36,Reference Irawati37,Reference MacLaughlin, Saseen and Malone45} Cost data included multiple currencies from various countries. No studies found that allergy labels were associated with cost savings (Fig. 3, column 3). For example, 2 of 3 studies examining the cost of inpatient antibiotics found that allergy-labeled patients had a significantly higher cost of treatment than nonallergic control patients (Table 4).^{Reference Sade, Holtzer, Levo and Kivity24,Reference Irawati37} Similarly, 3 of 3 studies that examined the costs of outpatient antibiotic treatment found that patients with an allergy label were prescribed more expensive drugs (Table 4).^{Reference Sade, Holtzer, Levo and Kivity24,Reference Kraemer, Caprye-Boos and Berman28,Reference MacLaughlin, Saseen and Malone45} Of 5 studies examining the total cost of hospitalization, 4 found that allergy-labeled patients incurred higher costs than nonallergic control patients, and the fifth study showed a trend toward significance (Table 4).^{Reference Macy and Contreras2,Reference Huang, Cluzet, Hamilton and Fadugba20,Reference Sousa-Pinto, Cardoso-Fernandes, Araujo, Fonseca, Freitas and Delgado36,Reference Irawati37,Reference Sousa-Pinto, Araujo, Freitas and Delgado39}

Evidence grading

The quality of studies ranged from 3 to 9 stars. Only 9 of 41 studies (22%) scored a maximum quality score of 9 stars, but 20 studies (49%) received scores of 7 or 8 stars (Supplementary Material online). Also, 5 studies (12%) had scores of 6, and 7 studies (17%) had scores ≤5. The mean value of stars achieved by all studies was 7.05, and the quality of studies did not differ by domain: 7.24 for antibiotic exposure, 7.28 for clinical outcomes, and 7.55 for healthcare cost. Studies that failed to achieve a high-quality score typically had concerns over comparability of allergy-labeled and control cohorts, as well as poorly defined follow-up for outcomes of interest.