The importance of clinical history in the diagnosis of drug hypersensitivity in children

In case of suspected hypersensitivity reactions (HRs) to drugs, a challenging area for pediatricians is detecting relevant elements in the parent‐reported history, in order to reach a definite diagnosis. We analyzed the concordance between the description of the HR and the medical reports documented at the time of the event. Furthermore, we studied any correlation between clinical history variables and the prediction of true allergy.


| INTRODUC TI ON
Detecting elements and peculiar characteristics of parent/patientreported clinical history regarding suspected hypersensitivity reactions (HRs) to drugs is a challenging area for the pediatrician.
Nonetheless, it is of utmost importance because it guides the allergy work-up in identifying the most useful tests to be performed and lead to a definite diagnosis in the simplest and safest way for children.
In 2017, a systematic review 1 based on 53 studies showed that the pooled prevalence of self-reported drug allergy was around 7.2%.
Anyway, the great variability between different studies should be taken into account, and the prevalence was higher in adults than in children, as well as in the medical setting than in general population.
3][4][5][6] Various approaches for de-labeling are reported in the literature, [7][8][9][10] even by exploiting history alone, trying to perform risk stratification based only on the medical history. 11Currently, the greatest limitation of this pathway is the risk of faulty memories or mistakes.In particular, a study reported that elements possibly connected to higher risk of penicillin allergy were as follows: severe delayed reactions (at any point in the past) and severe immediate symptoms (in particular if occurred within 5 years from the event). 11[14][15][16] In this paper, the primary aim was to analyze the concordance between the description of a suspected HR, provided by families during the visit at our Allergy Unit, and the medical reports documented at the time of the event by physicians at the Emergency Department (ED).Furthermore, all the patients underwent a complete drug allergy work-up, which led to a definite diagnosis with confirmation or exclusion of HR.Other purposes were to identify any correlation between clinical history variables, related to the patient or to the reaction, and the prediction of true drug allergy.

| PATIENTS AND ME THODS
We retrospectively consecutively collected 50 charts of children who were referred to the Allergy Unit for evaluation, after a previous access to the ED of Meyer Children's Hospital IRCCS for suspected HRs to drugs.We compared in detail the written description of the reaction and the clinical examination reported by the pediatrician of the ED ("acute phase"), to the history as it was told by parents at the moment of the visit ("allergy work-up").For each patient, we filled two parallel separated Case Report Forms (CRF) at acute phase (CRF1) and at allergy work-up (CRF2), reporting the characteristics of the event, such as the culprit drug, the timing of reaction (including the date of reaction, the latency between drug intake and symptoms, in particular if less than 1 hour, and the day of the course therapy), the clinical manifestations and any required treatment.We also considered the latency between the reaction and the allergy investigations, then we reported the results of the tests (in vitro and in vivo) that eventually led to confirm or rule out HRs to the specific drug.
The data were evaluated and summarized in Table 1.These were all classified as "known" and "unknown" based on the comparison between CRF1 and CRF2.The reactions were also considered "severe" (in case of IgE-mediated anaphylaxis or else in case of Severe Cutaneous Adverse Reaction-SCAR) or "not severe" (all the others).
The definite diagnosis was "confirmed" or "excluded" according to the final results of the complete drug allergy work-up, which were reported on the medical charts.Among patients with a confirmed diagnosis, some had undergone drug provocation tests (DPT) 17,18 in case of non-severe reactions, while, in case of anaphylaxis or of SCARs, the diagnosis was confirmed by a history of severe drug HR and positive in vivo/in vitro tests.The standard procedures performed included in vivo tests (skin prick tests-SPT; intradermal tests-IDT; patch tests-PTs-and DPT) and in vitro tests (specific IgE for the culprit antibiotic, in some cases lymphocyte transformation test-LTT) following the European Network for Drug Allergy (ENDA) Guidelines. 19 correlated the time interval between the suspected HR and the visit at the Allergy Unit, with the reliability of the information given by the parents; moreover, we analyzed the whole time required to reach the diagnosis, to find out whether having a better-defined history since the beginning of the allergy work-up could be a helpful tool to direct the investigations and thus to shorten their duration.

| Statistical analysis
Statistical analyses were performed using the IBM Statistical Package for Social Science software (SPSS, Version 28.0, Chicago,

Key message
The diagnosis of drug allergy in children starts with the medical history collected by the pediatrician at the allergy visit.The history told by the parents can be often affected by mistakes in remembering what happened at the time of the event.In this study, including a pediatric population of 50 children, we compared the medical history documented at the acute phase to that reported by families at the subsequent allergy visit.We also analyzed the possible association between some variables related to the reaction and to the patient, which could direct the diagnostic investigations in the simplest and safest way in order to reach a definite diagnosis.
Illinois, USA).Data were reported as medians and interquartile ranges (IQR) for continuous variables and as percentages for categorical variables.Mood's median test was used for analyzing differences in continuous variables and the chi-squared test was used for categorical variables.Logistic regression analysis was performed to determine the association between demographic and reaction characteristics and the accuracy of reported clinical history or the confirmation of real allergy diagnosis.The results were expressed as odds ratio (OR) and 95% confidence interval (CI).Hosmer-Lemeshow goodness-of-fit statistics were used to assess model fit.p-values of <.05 were considered statistically significant.

| RE SULTS
Fifty children were included, 22 males and 28 females.Median age at reaction was 5.9 years (interquartile range, IQR 2.6-11.4)and 24 (48%) children were older than 6 years at the time of the reaction.
Ten patients (20%) had a positive familiar history of drug allergy.
Regarding the type and timing of the suspected HR, 17 patients (34%) experienced a severe reaction and 18 (36%) an immediate reaction-IR (onset <1 h from the drug intake).Ten out of 18 (55.5%)IR were also severe (anaphylaxis).The culprit drugs were as follows: amoxicillin-clavulanic acid (17/50), ibuprofen ( Analyzing CRF1 and CRF2 for each patient, we compared the concordance of the history reported by the parents: The type of the reaction was known in 37 (74%), the timing was known in 14 (28%), whereas the culprit drug was remembered by almost all the parents-49 (98%).The median time interval between the index reaction and the first allergy visit was 2.5 months.The median duration of the whole allergy work-up was 2 months and, at the end of the investigations, the definite diagnosis of HR was confirmed in 20 (40%)-05/20 by DPT, 15/20 by other in vivo/in vitro tests-and excluded in 30 (60%).The 15 patients who were diagnosed as allergic without undergoing the DPT were those with a history of severe reaction for whom the specific tests were found positive (SPT and/ or IDT and sIgE in case of suggestive anaphylaxis, PT or the LTT in case of suggestive SCARs) or negative/not performed in five cases.
We showed in Table 2 the reactions' characteristics and the results of the allergy investigations in the 20 patients with confirmed hypersensitivity.
We performed logistic regression analysis to find any correlation among variables related to the patients and to the reaction with the accuracy of the clinical history (Table 3).Logistic regression analysis highlighted an association between having had a severe reaction and an increased odds of remembering the timing of reaction (odds ratio, OR 4.94; p-value .038).On the other hand, being older >6 years at the time of reaction and having had an IR were associated with an increased odds of remembering the type of reaction (OR 6.12; p-value .046and OR 5.74; p-value .072respectively).
We also performed logistic regression analysis to find correlation between the accuracy of clinical history and the final diagnostic confirmation (Table 4).An association was found between remembering the timing of reaction and an increased odds of reaching a confirmed diagnosis (OR 6.26; p-value .009).
Most patients with a true HR reached the diagnosis within 12 months from the first allergy visit (17/20, 85%).Diagnostic confirmation was more likely in the case of patients with a shorter diagnostic work-up compared to those with a longer work-up (>12 months elapsed between the first visit and the definite diagnosis), p = .036.Time to diagnostic was significantly lower in patients whose parents remembered the type of reaction (2.67 vs. 6.15 months for the ones who were unable to describe it, p = .051).However, having a familiar history of drug allergy was not statistically correlated with a better recall of the reaction.Moreover, no statistical correlation was found between having a family history of drug allergy and the confirmation of drug hypersensitivity.

| DISCUSS ION
This paper analyses the reliability of clinical history in patients with history of drug reactions highlighting some interesting aspects.First, parents whose children experienced a severe reaction were more likely to remember the timing of reaction, and when timing was remembered, diagnostic confirmation was more frequent.
TA B L E 1 Demographic characteristics of the patients and their history of reaction.Furthermore, the families with children aged >6 years reported a more detailed and reliable history, so we can assume that an older child, being able to remind the event happened to himself, may help to describe the reaction during the first allergy visits and the subsequent medical evaluations.

Characteristics of the patients and their history of reaction
In addition, we demonstrated that a more accurate description of the type of reaction was associated with a significant shorter duration of the allergy work-up, suggesting that the tests performed by physicians were better addressed and led beforehand to a definite diagnosis, in particular in the group of IgE-mediated reactions.It was also interesting that all the parents but one can recall the name of the culprit drug, we can hypothesize that the reason is that in pediatric age, there is a constant need for antibiotic or anti-inflammatory therapies for common infections so that it is essential for parents to know which drug should be avoided and which are the alternatives to be prescribed, until allergy investigations are completed.
Finally, we showed that the diagnostic work-up was shorter in patients with a subsequent confirmed diagnosis.This may highlight the importance of carrying out the allergy investigations in a relative short time, which can allow to confirm the diagnosis of HR to drugs especially if performed within a period of 12 months, taking into account that in our group most of the reactions were IgE-mediated where it is known that detection of positive tests is less probable over the time. 20 must be noted that we found a high percentage (40%) of confirmation of drug hypersensitivity in the cohort of patients enrolled in this study, much higher than the usual incidence reported in literature for the pediatric age. 21,22This apparent lack of agreement can be attributed to the fact that we had included only children who were taken to the ED, in order to study the different reported histories by the parents at the acute phase and at the subsequent allergy visit.Therefore, it is reasonable that the patients selected were those who experienced a probable true HR compared with all the other population with a mild reaction who mostly refer to the general pediatrician.
To date, very few studies investigated the differences between the history of a suspected HR as told by the patients and the medical records during the acute phase.
Vyles et al. 23 published a study including children with parentreported penicillin allergy, who presented to a pediatric ED.Parents were asked to complete an allergy questionnaire, to identify lowor high-risk groups for penicillin allergy.Overall, 76% of patients reported exclusively low-risk allergy symptoms and only 24% one or more high-risk symptoms, so the authors suggested that those symptoms did not reflect a true penicillin HR.In their following study, authors proceeded with tests, showing that 100% of patients were negative and so were de-labeled as nonallergic. 24Miller et al. 25 described 100 children whose parents reported, during a hospital admission, a previous antibiotic allergy.The study underlined that using a standardized allergy questionnaire was a useful tool to better assess the validity of the suspect HR, compared to only history told by the family.8][29] Another retrospective analysis including children with adverse drug reactions highlighted the importance of the electronic health record as a modern solution compared with the paper charts, which could be not properly updated. 30The paper of Strömberg Celind et al. about children with asthma showed a good agreement between the questionnairebased data provided by parents and the National Drug Register. 31very recent study 32 analyzed parent-reported penicillin allergy in children through surveys to the parents, and the authors concluded that most of them were unaware of the meaning and consequences of a true hypersensitivity, which had not been investigated.
Bold values indicate p-values of <.05 are considered statistically significant.Abbreviations: CI, Confidence interval; mo, months; OR, odds ratio.Logistic regression results according to type and timing of reaction.Logistic regression results according to diagnostic confirmation.Bold values indicate p-values of <.05 are considered statistically significant.
including adults and adolescents, aimed to investigate the proportion of self-reported drug allergy at ED and the likelihood of a real hypersensitivity.Patients underwent a structuredTA B L E 3Abbreviations: CI, Confidence interval; mo, months; OR, odds ratio.