Background

Beta-lactam allergies are common, and the label is often associated with inappropriate antibiotic prescribing [1]. The risk of a cross-reactive allergy between penicillins and cephalosporins was traditionally thought to be as high as 10%, but is likely to be closer to 1–2% [2]. The removal of cephalosporin prescribing alerts for patients with penicillin allergies has been shown to increase prescribing of cephalosporins without an increase in allergic reactions or anaphylaxis [3].

The likelihood of cross-reactive allergy if the index drug is penicillin relates to the beta-lactam core and/or side-chain, whereas if the index drug is a cephalosporin, it relates to the R1 side chain in most cases. It has been suggested that ‘cephalosporin allergy’ is a misleading term because allergy to cephalosporins as a group is rarely if ever seen [4].

Cefalexin is a commonly-prescribed oral first-generation cephalosporin. It has a similar side chain to amoxicillin, ampicillin and cefaclor. Accordingly, an individual with cefalexin allergy has an appreciable risk of cross-reactive allergy with these other side-chain related antibiotics. Patients who have had a severe reaction to cefalexin (anaphylaxis or severe cutaneous adverse reaction) should avoid other beta-lactam antibiotics, at least pending expert evaluation and opinion, since even with a low likelihood, a dangerous reaction is significant. However, a low-risk cefalexin allergy or intolerance should typically not preclude the use of other cephalosporins or penicillins [5].

There are multiple common indications for cefalexin use, including cellulitis and urinary tract infection. It can be used in the treatment of Gram-positive bacteria as well as certain Gram-negative bacteria including Escherichia coli, Proteus mirabilis and Klebsiella pneumoniae [6]. The unnecessary avoidance of cefalexin and use of non-cephalosporin second-line antibiotics promotes the development of antimicrobial resistance [7].

Our previous work has shown that penicillin allergy labels are common in electronic health records in the Australian setting, with inconsistent use of the terms allergy and intolerance [8]. More recently, we have demonstrated that the presence of a penicillin allergy label is associated with a significant increase in the prescribing of second-line antibiotics including moxifloxacin and vancomycin [1]. However, the characteristics of patients with cefalexin labels and whether this affects prescribing practices is unclear.

The aims of this study were to (1) describe the frequency and nature of cefalexin adverse reaction (AR) records in a medical inpatient cohort, (2) evaluate the accuracy of cefalexin allergy/intolerance labels based on the reaction descriptions in the electronic medical record (EMR) and (3) examine the association between a low-risk cefalexin allergy label and the antibiotics prescribed/received during an inpatient admission.

Methods

Study design, setting and population

A single-centre retrospective cohort study was conducted including consecutive general medicine inpatient admissions at two tertiary hospitals using a single EMR over a 1-year period (October 2020—October 2021). A cohort of medical rather than surgical inpatients was chosen, as antibiotic use in the latter group tends to be more protocolised, and thus, the prescribing patterns would likely reflect the protocol used rather than individual prescriber practice. Multiple admissions of an individual patient were included separately, as were the antibiotics that they were prescribed during each admission.

Data collection

Patient age, sex and Charlson comorbidity index were collected from the EMR which is a modification of the Allscripts Sunrise product (Allscripts, Chicago, IL, USA). Pre-existing AR (or AR where the onset time was not specified), as well as antibiotics prescribed during a given admission, were also collected from the EMR. For some patients, it was unclear whether the AR was first recorded during a given admission, or prior to it. Case note review was undertaken for cases in which cefalexin was administered to a patient with a cefalexin allergy label to determine whether a reaction occurred. Case note review was also undertaken for cases in which amoxicillin or piperacillin-containing antibiotics were administered to patients with high-risk cefalexin allergies to determine whether a reaction occurred, and if so, the nature of the reaction.

Cefalexin allergy gold-standard classification

Cefalexin allergy descriptions were reviewed by medical officers using expert criteria to determine whether the described reaction was most consistent with allergy or intolerance, consistent with a validated assessment tool (Table 1; [9]). If the reaction was consistent with allergy, the criteria were applied to determine whether the reaction was a high-risk or a low-risk allergic reaction (Table 2). Gold-standard classifications were then compared to existing electronic medical record classifications to determine the accuracy of the existing labels. Cases in which the only description of the reaction was “Unknown reaction” were excluded from this analysis.

Table 1 Preset criteria for cefalexin adverse reaction (AR) categorisation—allergy versus intolerance
Full size table
Table 2 Preset criteria for cefalexin adverse reaction (AR) categorisation based on risk level
Full size table

Statistical analysis

Descriptive statistics were used to describe the frequency and nature of cefalexin AR labels. The accuracy of existing cefalexin AR labels, as compared to the gold-standard classifications, were also analysed using descriptive statistics. Odds ratios (OR) were calculated for the likelihood of being prescribed a given antibiotic for a group with a specific cefalexin AR relative to the group with no recorded AR. OR were not calculated to compare between other groups (such as high-risk and low-risk AR), due to sample size precluding meaningful analyses. OR and their associated confidence intervals were calculated as standard. In instances in which zero values limited assessment, 0.5 was added to all values prior to confidence interval calculation [10]. Tests for statistical significance were not conducted to avoid multiple hypothesis testing, and given the descriptive focus of the study. Analyses were conducted in Python (CreateSpace, Scotts Valley, California, United States) and R (R Core Team, Vienna, Austria).

Ethical approval

This study received approval from the Central Adelaide Local Health Network Human Research Ethics Committee (HREC/19/CALHN/209), with a waiver of individual consent.

Results

Patient characteristics

There were 12,134 individual admissions included in the study. The cohort had a mean age of 71.7 years (standard deviation [SD] 18.6) and there were 6090 females (50.2%). The median Charlson comorbidity index was 4 (interquartile ratio [IQR] 3–6). Characteristics of the patients categorised by subgroups are detailed below.

Frequency and nature of cefalexin AR labels

Cefalexin AR labels were present in 224 admissions (1.9%). Of these, the number of patients classified as allergy was 196 (87.5%), intolerance was 28 (12.5%).

Frequency and nature of cefalexin AR gold-standard classifications

With respect to the reclassified gold-standard AR labels, there were 59 (52.7%) with high-risk allergies, 53 (47.3%) with low-risk allergies and 69 with intolerance. There were 43 individuals who could not be further classified with respect to the nature of their AR as the only recorded information was “unknown reaction”. There were 11,910 without an AR. The characteristics of these patients are summarised in Table 3.

Table 3 Characteristics of patients by cefalexin adverse reaction (AR) subgroups
Full size table

Accuracy of cefalexin AR labels when compared to gold-standard

Of those labelled with an allergy in the EMR, 112 (57%) met gold-standard criteria for true allergic reaction, whereas 41 (21.9%) were consistent with intolerance and 43 (19.2%) could not be further evaluated since the only recorded information was “unknown reaction”.

Association between EMR label cefalexin AR and antibiotic prescribing

The total number of antibiotics prescribed in the cohort are described in Table 4. For those with a cefalexin EMR intolerance label, cefuroxime and trimethoprim were the only antibiotics prescribed significantly more frequently. By contrast, in those with a cefalexin allergy label, the beta-lactams amoxicillin, cefazolin, ceftriaxone and cefalexin were prescribed significantly less frequently, whereas trimethoprim, clindamycin, moxifloxacin and ciprofloxacin prescription rates were significantly higher (Table 5).

Table 4 Antibiotics prescribed in cohort
Full size table
Table 5 Antibiotics prescribed as stratified by EMR allergy label
Full size table

Association between gold-standard cefalexin AR and antibiotic prescribing

For those with a high-risk cefalexin allergy, clindamycin, moxifloxacin and ciprofloxacin were significantly more likely to be prescribed, and ceftriaxone was significantly less likely to be prescribed (Table 6). For individuals with a low-risk cefalexin allergy, clindamycin and ciprofloxacin were significantly more likely to be prescribed. In contrast, individuals with a cefalexin intolerance were significantly more likely to receive trimethoprim. In patients in the “unknown” reaction group, trimethoprim and moxifloxacin were significantly more likely to be prescribed, whereas amoxicillin and ceftriaxone were significantly less likely to be prescribed.

Table 6 Antibiotics prescribed as stratified by gold-standard allergy label
Full size table

Administration of amoxicillin- and piperacillin-containing antibiotics in high-risk cefalexin allergy

Of the 13 patients with a high-risk cefalexin allergy (according to their gold-standard classification) who were prescribed amoxicillin, 11 received amoxicillin, and 2 patients had adverse drug reactions to amoxicillin (both pruritic, delayed rashes). Of the 5 patients with a high-risk cefalexin allergy that received piperacillin-containing antibiotics, 2 patients had adverse reactions (one delayed rash and one immediate rash).

Discussion

This study has shown that individuals with low-risk cefalexin allergies are more likely to be prescribed second-line non-beta-lactam antibiotics including clindamycin and ciprofloxacin. This is aligned with previous literature suggesting that penicillin allergy labels (but not intolerance labels) to be associated with an increased likelihood of being prescribed second-line antibiotics including vancomycin and moxifloxacin [1].

In this study, some patients with high-risk cefalexin allergies (according to gold-standard classification) were prescribed ceftriaxone (5.1%), amoxicillin-containing penicillins (22.0%) or piperacillin-containing penicillins (8.5%). Our study demonstrated several cases of cross-reactivity in which patients with high-risk cefalexin allergies were administered amoxicillin- or piperacillin-containing antibiotics, resulting in a cutaneous AR. In situations where the index reaction to cefalexin as anaphylaxis or a severe cutaneous AR, specialist advice or investigation is recommended prior to administering other cephalosporins or penicillins [4].

As expected, the rates of cephalosporin and penicillin prescribing was higher in the low-risk cefalexin allergy group (0% to 13.2% for cephalosporins; 7.5% to 30.2% for penicillins) compared to the high-risk group (0% to 5.1% for cephalosporins; 0% to 22.0% for penicillins; Table 6).

Cross-reactivity within the cephalosporin family is very limited as the reaction is more likely to relate to the side chain rather than the core structure; as such, for low-risk cefalexin allergies, it is appropriate to prescribe other cephalosporins [4]. If this principle is followed, the prescribing rates of cephalosporins (other than cefalexin) should ideally be similar in the low-risk cefalexin allergy group, compared to the background rate. However, this study found that in the low-risk cefalexin allergy group, the prescribing rate for cephalosporins (other than cefalexin) was inappropriately lower (1.9% to 3.8%; Table 6) compared to patients with no cefalexin ARs (1.6% to 16.9%; Table 6).

Second-line non-penicillin antibiotics are associated with increased cost to the health system due to higher medication costs and increased rate of adverse effects [11]. For example, ciprofloxacin may be associated with prolongation of the QT interval, torsade de pointes [12] and tendon rupture [13], as well as drug–drug interactions. The use of clindamycin and ciprofloxacin is also associated with a greater risk of Clostridium difficile compared to first-generation cephalosporins such as cefalexin [14]. Furthermore, in order to reduce the potential for antimicrobial resistance, efforts must be made to limit prescribing of antibiotics such as fluoroquinolones and clindamycin to cases of proven need.

A prior study investigated the cross-reactivity between beta-lactams in patients sensitised to cephalosporins by conducting diagnostic investigations with alternative beta-lactam antibiotics. It included 24 patients with immediate allergic reactions to cephalosporins, 2 of whom had a positive skin test to penicillin reagents, whilst 22 had negative skin tests and tolerated penicillin G challenges [15]. This study, alongside several others in the literature, suggest that in some situations, penicillins may be safely administered in selected patients with cephalosporin allergies and negative skin tests to penicillin determinants. However, in cases of anaphylaxis, a drug provocation test may be required.

Our findings highlight the need for educational interventions in order to improve prescribing practices for patients with low-risk cefalexin allergies. Several articles in the literature have described the successful implementation of educational interventions to increase cephalosporin prescribing in penicillin-allergic patients in a perioperative setting [16]. A similar educational intervention may be used for cefalexin-allergic patients to promote the safe prescribing of penicillins and other cephalosporins. Blumenthal et al. described the implementation of an inpatient antibiotic prescribing guideline for patients with cephalosporin or penicillin allergies, resulting in an almost 7‑fold increase in the number of test doses to beta-lactams without increased AR; consequently, a significant decrease in alternative antibiotic exposure was observed [17]. Macy et al. also proposed a rapidly implantable system-level intervention which involved the removal of the alert in their EMR that warns against cephalosporin use in patients with penicillin allergies, with findings of increased cephalosporin prescribing as a result [3]. Similarly, the removal of alerts from the EMR to the prescription of penicillins and other cephalosporins in patients with cefalexin allergies, may be useful for improving prescribing practices.

In addition, this study has demonstrated that cefalexin allergy labels recorded in the electronic medical record frequently reflect cefalexin intolerances as opposed to true allergies. The literature has previously demonstrated that the accuracy of penicillin AR classification is limited by the extent of the history available, the accuracy of the reaction description, and the level of training of the healthcare professional in differentiating between AR types [8]. It is likely that the accuracy of cefalexin AR labelling is also limited by these factors.

A limitation of this study is that it examined medication orders, but was unable to examine medication administration. It is possible that at times a medication may be ordered, and then cancelled before administration (which may occur in instances of apparent order of a contraindicated medication in a patient with a high-risk allergy that is later cancelled). Although conducted at two centres, this study focussed on medical inpatients and may not be generalisable to other cohorts. Examination of antibiotic prescribing practices in the setting of cefalexin allergies in surgical inpatients and community-dwelling adults may also be beneficial. While cefalexin is a commonly prescribed cephalosporin, examination of allergies to other cephalosporins would be useful to improve the generalisability of these findings.

The findings of our study have implications for in-hospital antibiotic prescribing and the appropriate evaluation of cefalexin allergy labels may be a tool to improve antimicrobial stewardship. Additional studies should be conducted to further evaluate the influence of other antibiotic labels, including other cephalosporin allergies, on contemporary inpatient prescribing practices. Further education and system-based interventions are required to address the potentially misleading term ‘cephalosporin allergy’ and its effect on antibiotic prescribing practices. Further studies evaluating interventions for this issue should focus on patient or system-oriented outcomes, including the financial impact of increased use of second-line antibiotics and their associated complications.