The Prevalence of Suspected and Challenge-Verified Penicillin Allergy in a University Hospital Population


  • Jakob E. Borch,

    Corresponding author
    1. Allergy Centre and Department of Dermatology, Odense University Hospital, University of Southern Denmark, DK-5000 Odense, Denmark
    Search for more papers by this author
  • Klaus E. Andersen,

    1. Allergy Centre and Department of Dermatology, Odense University Hospital, University of Southern Denmark, DK-5000 Odense, Denmark
    Search for more papers by this author
  • C. Bindslev-Jensen

    1. Allergy Centre and Department of Dermatology, Odense University Hospital, University of Southern Denmark, DK-5000 Odense, Denmark
    Search for more papers by this author

Author for correspondence: Jakob E. Borch, Allergy Centre and Department of Dermatology, Odense University Hospital, University of Southern Denmark, DK-5000 Odense, Denmark (fax +45 6312 1507, e-mail


Abstract: Suspected penicillin allergy is common among hospitalised patients, but the quality of the information given by the patient is often doubtful. Alleged penicillin allergics are likely to be treated with more toxic, broad-spectered, and more expensive antibiotics, with effects on microbial resistance patterns and public economy as a consequence. We performed a cross-sectional case-control study with two visits to all clinical departments of a large university hospital in order to find in-patients with medical files labelled “penicillin allergy” or who reported penicillin allergy upon admission. Patient histories were obtained via a questionnaire, and they were offered investigation for penicillin allergy with specific IgE, basophil histamine release, skin prick tests, intradermal tests and drug challenge tests. Finally, the pharmaco-economical consequences of the penicillin allergy were estimated. In a cohort of 3642 patients, 96 fulfilled the inclusion criteria giving a point-prevalence of alleged penicillin allergy of 5% in a hospital in-patient population. Mean time elapsed since the alleged first reaction to penicillin was 20 years. The skin was the most frequently affected organ (82.2%), maculo-papular exanthema (35.4%) and urticaria (10.4%) being the most frequently reported reactions. 25% did not recall the time of their reaction. 82.2% did not remember the name of the penicillin they reacted to. 34.8% had been treated with penicillins after suspicion of penicillin allergy had been raised. None of these reacted to penicillins. 33.3% of the patients receiving antibiotics during their current hospitalisation were prescribed penicillins. 2% developed non-severe exanthema. The average acquisition costs for antibiotics to penicillin allergic patients were € 278, compared to € 119 had they been non-allergics. The prevalence of suspected penicillin allergy was lower than reported elsewhere. A substantial number of patients failed to recall basic information about their alleged allergy. Patients reporting penicillin allergy upon admission and labels stating penicillin allergy on medical files are ignored in almost a third of patients receiving antibiotics. The acquisition costs for antibiotics to penicillin allergic patients were higher, compared to the cost had the patients been non-allergics.

Up to 17% of the population report a history of penicillin allergy (Kerr 1994; Wyatt 1996; Macy et al. 1997; MacLaughlin et al. 2000; Irawati 2003). The quality of such information is difficult to evaluate, and the alleged reaction may have taken place several years before re-evaluation thus reducing the likelihood of documenting penicillin allergy (Chandra et al. 1980; Romano et al. 1995; Blanca et al. 1999). Further, more than 50% of allergies reported by patients may not be of immunological origin (Kerr 1994). Additionally, once an allergy is recorded in the medical record, it will most likely remain there for the rest of the patient's life. These factors may explain why penicillin allergy is not confirmed in the majority of alleged penicillin allergics tested with skin tests and drug challenge tests (Borch & Bindslev-Jensen 2003).

Suspected penicillin allergy has implications for patient health, microbial resistance patterns, and patient and public economy. Patients reporting penicillin allergy are more likely to have cephalosporins, macrolides, or other antibiotics (e.g., quinolones, tetracyclines, nitrofurantoin, or vancomycin) than those not having an allergy documented in their medical record (Solensky et al. 2000). Alternative antibiotics may be broader in spectrum, have a higher potential for adverse reactions, be more expensive, and have the potential for inducing antibiotic resistance (MacLaughlin et al. 2000).

In Denmark the penicillins used in patients are: Penicillin G & V, dicloxacillin, flucloxacillin, ampicillin, amoxicillin, piperacillin, and mecillinam. Piperacillin is only used together with tazobactam.

In this study we investigated the prevalence of alleged penicillin allergy in a Danish University Hospital and the rate of true penicillin allergy. Furthermore, we estimated the consequences of alleged penicillin allergy on acquisition costs of antibiotics.

Materials and Methods

The study was performed as a combined cross-sectional and case-control study with two unannounced visits of one day's duration to all clinical departments of Odense University Hospital, Denmark, in order to find patients with files marked “CAVE! Penicillin” or who reported penicillin allergy upon admission. All departments were visited in the summer of 2003 (April 29th–June 27th), and again in autumn (September 4th–Nov. 11th) yielding a study cohort of 1925+1717=3642 in-patients. A thorough history was taken by a trained nurse using a structured questionnaire for recording demographic details, histories of the allergic reactions (urticaria, maculo-papular rash, angio-oedema, gastrointestinal symptoms as vomit/diarrhoea/stomach ache, dyspnoea, and anaphylaxis), suspected penicillin, other allergies, and cause of current hospitalisation. Data were entered into a Microsoft Excel spreadsheet.

Skin Prick Tests (SPT) for Benzyl Penicilloyl Polylysine (BPO-PL) and Minor Determinant Mixture (MDM) were performed using Allergopen® (Allergopharma, Reinbek, Germany). SPT and intradermal test (IDT) with penicillin G, ampicillin and SPT with amoxicillin were performed with commercially available formulations (Panpharma SA, Fougéres, France; Bristol-Myers Squibb, NY, USA; Yamanouchi, Tokyo, Japan, respectively). SPT was performed in concentrations of 1:1. IDT were performed in concentrations of 1:1 for penicillin G, PPL, MDM, and in concentrations of 1:1000–1:1 of ampicillin. Dilutions were done in commercial vials, and SPT and IDT was done on the volar forearm. Histamine HCl 10 mg/ml (ALK-Abelló, Hørsholm, Denmark) was used as positive control and isotonic saline was used as negative control. Drug Challenge Tests (DCT) with penicillins G and V (Panpharma SA, Fougéres, France & Rosco, Taastrup, Denmark) was performed when all aforementioned tests were negative. IDT with other β-lactams was performed upon suspicion to these or when cross-reactions were suspected. Intravenous challenge with penicillin G was done in dilutions of 1:1000–1:1 of 1,000,000 IU, and oral challenge with 250,000 IU penicillin V, unless contraindicated (Aberer et al. 2003). DCT with other β-lactams was performed when relevant as described for IDT.

Specific IgE was measured using CAP Pharmacia (Stockholm, Sweden). Basophil histamine release (Skov et al. 1985) (RefLab, Copenhagen, Denmark) was measured for penicillins G & V, MDM, BPO-PL, ampicillin, and amoxicillin.

Patch tests were performed in selected patients with penicillin G, ampicillin, amoxicillin, dicloxacillin, and cefotaxime in 10%, pet./aq./eth. (Chemotechnique, Malmö, Sweden), and cefuroxime (AstraZeneca, London, England) in 30% pet. (Barbaud et al. 2001) using standard technique with Finn Chambers on Scanpor (Epitest Ltd Oy, Tuusula, Finland; Alpharma AS, Oslo, Norway). Patch tests were read on D3 and D5–7 according to ICDRG recommendations (Wahlberg 2001).

The study was approved by the regional ethical board (VF20020165) and informed written consent for all the diagnostic procedures was obtained from all patients.


The study cohort consisted of 3642 hospitalised patients (table 1) and 96 fulfilled the inclusion criteria (fig. 1). During the 5–12 months interval between inclusion and planned investigations 24 (25%) died, 59 (61.4%) refrained from participation, and only 13 (13.5%) patients completed the investigations. A comparison between surviving and deceased patients with regard to mean age and number of hospitalisations is shown in table 2. 17 of 96 (17.7%) patients were unable to give reliable information about previous their reactions to penicillin.

Table 1.  Prevalence of alleged penicillin allergy in different departments.
DepartmentPrevalence of
alleged penicillin allergy
(No. of patients)
Cardiology*6.9% (118)
Dental surgery**0.0% (0)  
Dermatology9.0% (33) 
Endocrinology*0.0% (58) 
Gastroenterology, med.*3.9% (51) 
Gastroenterology, surg.**5.0% (138)
Geriatrics4.1% (96) 
Gynaecology-Obstetrics6.4% (139)
Haematology*1.6% (61) 
Intensive care unit0.0% (21) 
Internal medicine*5.8% (239)
Nephrology*4.8% (41) 
Neurology*6.4% (140)
Neurosurgery**1.6% (60) 
Oncology5.0% (80) 
Ophtalmology0.0% (12) 
Orthopaedics**5.6% (143)
Oto-/Rhino-/Laryngology9.0% (33) 
Paediatrics1.8% (166)
Plastic surgery**0.0% (46) 
Psychiatry, adult2.5% (239)
Psychiatry, children0.0% (12) 
Thoracic & vascular surgery**5.7% (123)
Urology**6.1% (49) 
*Medical, total5.4% (657)
**Surgical, total4.6% (559)
Total 5.0% (1216)
Figure 1.

Patients in the study.

Table 2.  Comparison of surviving and deceased alleged penicillin allergics with all hospitalised patients in 2003.
 Mean age
Number of hospitalisations
All hospitalised patients442.5
Surviving alleged penicillin allergics588.0
Deceased alleged penicillin allergics709.7

Mean age was 61 years, range 2–99 years. Sex ratio (f/m) was 1.7. The overall prevalence of alleged penicillin allergy was 5.0% (table 1).

During the hospitalisation 43 of 96 patients (44.7%) were treated with an antibiotic. A total of 36 patients had infections or traumatic/surgical conditions indicating antibiotic treatment. 15/43 (34.8%) patients received penicillins during the current hospitalisation despite of their history of penicillin allergy. No one experienced adverse reactions to the administration of penicillin. In 28/43 patients antibiotic treatment was done without the use of penicillins (fig. 1). Prior to the current hospitalisation 33 of the 96 patients had received penicillins after the suspicion of penicillin allergy had been raised; none of these had reacted. This results in a total percentage of 34.3 of the 96 patients having received penicillins despite the alleged penicillin allergy. 2/43 (4.6%) patients had their first adverse reaction to penicillins during the current hospitalisation. One patient had a rash during treatment with cefuroxime.

79 patients (82.2%) did not remember to which penicillin they had had an allergic reaction. Five (5.2%) reported allergy to more than one penicillin. Six reported reactions to penicillins G/V, six to ampicillin, five to dicloxacillin, and three to amoxicillin. 2/3 reporting reactions to amoxicillin also reported reactions to ampicillin. No reactions were reported to flucloxacillin, piperacillin, and mecillinam, which are the other penicillins in use in Denmark.

Allergies to a total of 25 drugs other than penicillins were reported by 21 (21.8%) patients. Seven reported allergy to opioids and six to sulfa-antibiotics. All other drugs were reported only once.

Other allergies than drug allergy were reported by 21 (21.8%) patients: 11 reported inhalant allergies, eight contact allergies, and two unspecified reactions.

A family history of penicillin allergy was reported by seven (7.2%) patients.

The time of reaction was recalled by 64 patients. The mean time elapsed since the reaction in these patients was 20 years, ranging from 0 to 59 years (1944–2003). 4 patients now aged 17, 22, 50, and 70 reported reaction in childhood. 24 (25%) did not remember the year of reaction. 13 (13.5%) had a reaction to penicillin within the last five years.

Immediate type reactions were experienced by 14 (14.5%) patients. 10/14 described anaphylaxis, angio-oedema, urticaria or vomiting. Seven (7.2%) reported reaction on day two of penicillin treatment, 15 (15.6%) from day two to day five, and three (3.1%) after one week, including one five days after treatment cessation. 57 (59.3%) did not remember the timing of their alleged reaction.

Indication for penicillin treatment was remembered by 42 (43.7%). 14 were treated for airway infections, six for urinary tract infection, six for dermatological infections, four for dental infections, and 12 for miscellaneous infections and fever.

A maculo-papular rash was described by 34 (35.4%), urticarial lesions defined as hives by 29 (30.2%), angio-oedema defined as swellings of face and/or hands by 10 (10.4%). 8 (8.3%) had gastro-intestinal reactions, six (6.2%) remembered a rash, but were unable to describe it in detail. Anaphylactoid symptoms were described by three, generalised pruritus by three, respiratory symptoms by two, Acute Generalized Exanthematous Pustulosis was diagnosed in one patient, who had experienced angioedema to penicillin in the past. Blisters was described by one patient; activation of pre-existing psoriasis by another. Unspecific and non-allergic symptoms were described by four patients. 18 (18.7%) were unable to describe their skin reaction in detail (“rash”) (fig. 1).

Since their reaction 33 (34.3%) had been treated with penicillins, 37 (38.5%) were certain they had not received penicillins since their reaction, and 26 (27.0%) had no recollection. 43 (44.7%) reported to have had occasional or frequent need of antibiotic treatment in the time elapsed since their allergic reaction.

Penicillin allergy could be demonstrated in 1/13. This patient was however not challenged systemically since the intradermal test (IDT) was positive (Torres et al. 2003). Two challenge negative patients were IgE positive (0.45 kU/l, 0.38 kU/l), and one patient was patch test positive (++ for ampicillin and amoxicillin).

In 32 (33.3%) a penicillin was the drug of choice for the infection according to the hospital guidelines for antibiotic treatment. The cost of antibiotic treatment in alleged penicillin allergics are shown in table 3.

Table 3.  Acquisition costs of antibiotic treatment in alleged penicillin allergic patients.
Patient categoryConditions/contextAcquisition costs
€ (average)
Antibiotic courses in cases where any antibiotic were either given or indicated according to guidelinesPatients receiving any antibiotic treatment including those wrongfully receiving penicillins11,939 (278)
 Same – had the patients not been penicillin allergics 5,111 (119)
 Patients rightfully not treated with penicillins 4,345 (155)
Antibiotic courses in cases where penicillins were either given or indicated according to guidelinesPatients receiving penicillins including those wrongfully receiving penicillins 5,000 (147)
 Hospital antibiotic guidelines3,256 (96)
 Danish Medical Association guidelines2,960 (87)


The prevalence of alleged penicillin allergy in a hospital population was 5.0%. This is considerably lower than reported by others, and may be due to variations in antibiotics sales, as illustrated for the European Union (Cars et al. 2001). Antibiotic use including the use of broad spectrum penicillins is lower in Denmark than in most other member states of the European Union, whereas penicillins G & V are used more. Broad spectrum penicillins and other antibiotics have a higher potential for eliciting adverse drug reactions, and thus the low prevalence of alleged penicillin allergy may reflect a low prevalence of confirmed penicillin allergy. However, further studies are required to prove this. Differences in prevalence may also rely on differences in the composition of the study cohort. Some departments have higher incidences (table 1) and in hospitals with either large or none of such departments the prevalence will be affected accordingly. Differences may also be depending on how information is obtained: by direct interview of all patients upon admission or by manual or computerized search of labelled medical files with the risk of information being outdated or not collected. In the present study information was retrieved by a combination of both.

Mean age was higher than we have observed in patients referred for penicillin allergy investigations: 66 years versus 41–43. More women than men reported penicillin allergy, which is also in agreement with our previous findings. Other allergies were reported by 21.8%, which is also in agreement with our previous findings (Borch & Bindslev-Jensen 2003), indicating that the case group reflects the general population in this respect. Surprisingly 25% of alleged penicillin allergic patients died prior to elective investigations. It appeared that alleged penicillin allergics were older and had been hospitalised more often than the average patient. A causal relationship between the high mortality and the alleged penicillin allergy thus seems implausible.

Immediate type reactions were described by 14.5% patients. Mean time elapsed since the initial reaction was 20 years in accordance with others (Irawati 2003) but in contrast to our previous findings (Borch & Bindslev-Jensen 2003). In patients with immediate hypersensitivity to amino-penicillins the number of positive results to skin testing decrease as the reaction – testing interval increase (Romano et al. 1995). It has been demonstrated that reactivity to penicillin in both specific IgE-determination, skin prick test (SPT), intradermal test (IDT), and drug challenge test (DCT) decrease over time (Chandra et al. 1980). Of 42 children with a previous reaction to penicillin and a positive test, 14 had negative re-tests after one year and were treated with penicillins without adverse reactions. In another study two groups of patients with positive skin tests to benzylpenicilloyl or minor determinant mixture (MDM) and selective reactions to amoxicillin with good tolerance of penicillin G respectively were skin tested (SPT, IDT) (Blanca et al. 1999). The patients were followed for five years, and patients with a selective response to amoxicillin tended to loose sensitivity faster than those who responded to several penicillin determinants. In delayed type hypersensitivity early testing seems less important: 60.6% of 33 patients were diagnosed more than two years after their most recent reaction (Romano et al. 1995). A significant number of patients seem to loose reactivity to penicillins in the time elapsed between clinical reaction and challenge test. Since only 13 patients experienced their reaction within the last two years, only three of 15 (20%) of the patients investigated for penicillin allergy in the present study, and 15 of 109 (13.7%) patients with a history of penicillin allergy and negative specific IgE in a previous study (Borch & Bindslev-Jensen 2003) were found to react to skin tests or DCT, this may certainly be the case. Alternatively, the patients may fail to recollect details about their drug allergy, as was the case in 15.7% of the patients in the study by Wyatt (1996). Infectious exanthema is a differential diagnosis in patients with suspected cutaneous drug reactions and explains why suspected drug allergies are not confirmed in some patients with exanthema: The association between mononucleosis and maculo-papular rash due to ampicillin or amoxicillin is well-known (Pullen et al. 1967; van der Linden PD et al. 1998), and an association between drug-induced maculo-papular rash and upper airway and urinary tract infection has been demonstrated (Cohen et al. 2001). In a case group of patients with maculo-papular rash 28.3% had upper airway infection, while only 1.9% of controls had a rash. 11.6% in the case group had an urinary tract infection, where only 0.6% in the control group did.

In our study 25% did not recall even the approximate year of their reaction. 82.2% did not remember the name of the penicillin they reacted to. Irawati (2003) found that only 3.6% remembered the date of their reaction, and 33.3% of patient records contained information about the nature of the reaction and the name of the drug. From the interviews it appeared that a substantial part of the patients did not distinguish penicillins from other groups of antibiotics.

In the majority of patients penicillin allergy was diagnosed by a physician, and the most frequent consequence was discontinuation of penicillin therapy.

As expected the vast majority of reported previous reactions were located to the skin. A previous study demonstrated that urticaria and systemic symptoms were reported with similar rates in challenge-positive and -negative patients (Borch & Bindslev-Jensen 2003). A possible explanation why the skin so frequently is affected by adverse drug reactions compared to other organs may be that the skin delimits and protects the body from external influences. The immune system is always active in the skin. Drugs involved in CADR may be a co-factor necessary for reaction to occur. Alternatively, drugs with a direct toxic effect to a specific tissue, i.e. the skin, cause tissue necrosis and hereby supply co-stimulation (Pichler 2001). In the present study only 4% did not remember the nature of their reaction. This is in contrast to others (Wyatt 1996; MacLaughlin et al. 2000; Irawati 2003), where respectively 66%, 38.6%, and 23.5% of patients with an alleged drug allergy did not remember the nature of their allergic reaction.

Several facts indicate a lack of awareness of penicillin allergy as a potential health hazard in patients as well as medical staff: A high number of patients failed to remember details about their alleged penicillin allergy. In agreement with a recent study who found that 48.5% of patients with a previous allergy-like event to penicillin had penicillin re-prescribed (Apter et al. 2004), at least 34.3% of our patients had penicillins at least once again since their reaction. 28.8% of the patients were prescribed penicillins by a physician during hospitalisation, as also reported recently (Macy et al. 2004). Finally, only 15.6% agreed to the investigation for penicillin allergy offered in the present study. On the other hand, neither of the patients accidentally treated with penicillins experienced adverse reactions, which is in agreement with Apter et al. (2004) who found that only 1.89% of patients describing allergy-like events following penicillin treatment experience another event following later penicillin treatments. This indicates that the label penicillin allergy is often erroneous and the diagnosis should be confirmed by allergy testing, which is also consistent with the conclusions of Macy et al. (2004) who found that only 42.5% of patients with histories of penicillin allergy were accurately coded with respect to history of allergy and penicillin skin testing. In the study by Irawati (2003), 14.8% (4 of 27) of alleged penicillin allergics received a penicillin after their allergy was diagnosed. 3/4 experienced no adverse reaction. 66.7% had received a penicillin related compound. 88.2% experienced no adverse reactions. In 13% of patients with charts labelled with penicillin allergy the label could be removed using information based solely on a thorough history (Tripp & Brown 1993). However, ignoring the patients' statement of penicillin allergy may have legal implications in some countries and could be considered unethical.

Only in one case (of 15 consenting to investigation) we could verify penicillin allergy. This patient demonstrated a positive IDT to benzyl penicilloyl polylysine and challenge was therefore omitted. Whether this patient would have been challenge positive remains unknown especially since the two IgE positive patients were challenge negative. Only 13/96 patients underwent testing for their allergy, which greatly incapacitates this part of the study. Most patients had agreed to be invited to elective investigations during hospitalization, but later cancelled or did not respond to repeated letters of notification. The most common excuse among those who cancelled was that following a troublesome hospitalisation they did not have the energy or interest to spend a day in hospital for scientific reasons arguing that should they ever need it there are alternative drugs to penicillin.

In Denmark, it is recommended to use cefuroxime in penicillin allergics. Indeed, cefuroxime is now the drug of choice for antibiotic prophylaxis for many surgical procedures in non-allergics. 8.2% of penicillin allergics have been reported to have reactions to cephalosporins, whereas only 1.7% of non-allergics react to cephalosporin administration (Irawati 2003). In the present study only one patient with a known allergy to cefuroxime had a rash to cefuroxime treatment.

We find 2.3 times higher acquisition costs of antibiotics in penicillin allergics compared to non-allergics. The average antibiotic cost in alleged penicillin allergics receiving penicillins were 1.3 times higher than it would have been had they not been allergic. If only cases where penicillins were either given or indicated according to guidelines the cost of the treatment given were 1.5 respectively 1.7 times higher according to the hospital antibiotic guidelines and the Danish Medical Association guidelines. However, the actual costs of pharmacologic treatment are not only the acquisition costs of the drugs. Several other factors, such as administering drugs, monitoring for their side effects, treating side effects, providing nursing and medical care cost several fold higher than the purchase price (Eisenberg et al. 1988; Kerr et al. 1992). Actually, total costs range from 1.2 to almost 8 times the acquisition cost (Plumridge 1990). The factor by which to multiply the acquisition costs have never been assessed in a Danish hospital. Cost-effectiveness of drug allergy testing compared to acquisition costs is of little practical use, may be misunderstood, and has therefore not been calculated in the present study. It seems plausible that the factor is somewhere between 1.2 and 8, but the outcome of cost-effectiveness analysis of drug allergy testing would vary significantly depending on whether the factor is closer to 1.2 or 8.

Irawati (2003) showed that alleged penicillin allergics with severe community acquired pneumonia are hospitalised longer than non-allergics (12.5±2.98 days versus 8.0±2.13 days), and that average cost of antibiotic treatment (17.8% in mild/moderate disease; 238.9% in severe disease) and average total cost of admission (1.1% in mild/moderate disease; 55.5% in severe disease) is higher in alleged penicillin allergics.

A decrease in antibiotic use in patients undergoing penicillin skin tests has been observed. Alleged penicillin allergics received antibiotics costing an average of 15.5% more per course compared with the average, and received three times more courses of antibiotics per year during the year before they were tested for penicillin allergy, than the average (Macy 1998). The mean antibiotic cost for penicillin allergics has been found to be 37% higher than for non-allergics ($ 26.81 versus $ 16.28) (MacLaughlin et al. 2000). The median total cost of antibiotics and drug administration in community acquired pneumonia was US$ 12.90 for outpatients, and US$ 228.70 for inpatients (Gilbert et al. 1998). This difference is hardly explained by differences in acquisition costs between parenteral and non-parenteral outpatients, and US$ 228.70 for inpatients (Gilbert et al. Although only acquisition costs were calculated, the present study confirms that antibiotic treatment in penicillin allergics are more expensive than in non allergics.

The accurate diagnosis of penicillin allergy requires a drug challenge test, which is a resource-demanding test procedure. Therefore, an important preventive approach to reduce the number of suspected penicillin-related skin reactions will include the physician's consideration not to withdraw antibiotic treatment too early in the course of an infectious disease with development of skin changes. It may be an infectious exanthema or infection that is a co-factor. However, this discussion must in each individual case be balanced with the risk of the patient developing a more severe skin reaction.