Translation, transcultural adaptation and validation to Brazilian Portuguese of tools for adverse drug reaction assessment in children

Adverse drug reactions (ADRs) are harmful or unpleasant responses that occur unintentionally after using a drug. They may cause temporary or permanent physical damage, contributing to increased morbidity, mortality, and hospital admission costs [1, 2]. Children are more vulnerable to these reactions due to changes in the body resulting from the growth process and the scarcity of studies in this population [3]. In Brazil, almost two thousand suspected serious ADRs, including deaths, were reported in the National Health Surveillance Agency database (Notivisa) in children between 2008 and 2013 [4].

The causality (likelihood of a drug causing the ADR) and avoidability (possibility of ADR prevention) of an ADR are relevant to the study of drug safety. Such investigations may help develop protocols to aid prescribing, drug monitoring, and appropriate use of pharmacotherapy [5‐7].

The most commonly used tools for assessment of the causality of ADRs, such as the algorithm proposed by Naranjo [8], have limitations for use in the paediatric population [9, 10]. The Naranjo tool, published in 1982, may have less sensitivity for use in children because the overall score obtained can be artificially reduced [11]. In the same way, the most commonly used scale for assessment of ADR avoidability, created by Hallas [12], calls for in-depth knowledge of treatment protocols used which often don’t exist for paediatrics [6, 13]. In Brazil, the Naranjo tool is frequently used in ADR causality assessment, although studies focused on children are scarce, notably, those who assess avoidability.

Because of these limitations, two English-specific tools for the characterisation of ADRs in children have been developed and validated: the Liverpool Causality Assessment Tool (LCAT) [11] and the Liverpool Avoidability Assessment Tool (LAAT) [14]. LCAT and LAAT tools are both flow diagrams which are quick and easy to use enabling a systematic approach to assessing causality and avoidability of ADRs in children. Both tools use dichotomous responses and were developed and validated using retrospective case reports in the United Kingdom [11, 14].

This study aimed to translate and transcultural adapt the LCAT and LAAT to Brazilian-Portuguese and to evaluate the psychometric properties of these tools for the analysis of ADRs in paediatric patients.

Translation and transcultural adaptation were performed in collaboration with the research group responsible for designing and validating the original English language versions of LCAT and LAAT. For an in-depth understanding of the construction and investigation of the psychometric properties of the tools, the corresponding author spent approximately 6 months based in the Paediatric Medicines Research Unit at Alder Hey Children’s Hospital (AHCH) (Liverpool, UK) for study design [15] and practical training.

The validation of the transcultural adaptation of tools was obtained by the functional equivalence (given by the degree of agreement between conceptual and item equivalence, semantic equivalence, operational and measurement equivalence) between the original versions and the translated versions of the tools (Fig. 1) [15, 16].

The conceptual and semantic equivalence of items considered the different terms used by native translators. In addition, a grammatical review was conducted to ensure the correct use of the Brazilian-Portuguese language.

In the analysis of semantic equivalence with the group of 27 specialists, the format and ease of use of LCAT and LAAT were considered higher in comparison with other international algorithms, according to the previous experience of specialists. They highlighted the clarity of the technical terms and expressions used. However, it was suggested that the development of a supporting manual to aid in the application of the tools by inexperienced professionals would be beneficial.

For the LCAT, the main changes were related to the direct translation of questions and statements, including the word “drug”. There was also an exhaustive search for synonyms and the review of concepts related to the terms “lost-lasting disability” and “positive rechallenge” to improve the clarity of tools for the Portuguese language. For LAAT, the same was found for the terms “drug” and “medication”. The translation by the physician used the term “therapeutic conduct”. However, the literal translation was kept (due to the smallest number word count) (Table 1) to maintain the visual format of the English and Portuguese tools.

Version 2 of the tools (Figs. 2 and 3) were applied to the assessment of the twenty-six case reports of suspected adverse drug reactions to anti-infectives used by twenty-two children from IPPMG—UFRJ (Table 2).

Cefepime, vancomycin, amoxicillin, amoxicillin clavulanate, azithromycin, fluconazole, amikacin, ampicillin, cefuroxime, oxacillin, gentamicin, linezolid, crystalline penicillin and trimethoprim-sulfamethoxazole were the anti-infectives identified in case-reports.

The Cronbach’s alpha coefficient, Cohen’s Kappa, Weighted Kappa, percentage of exact agreement (%EA) and extreme disagreement (%ED) between reviewers using LCAT and LATT v2 are presented in Table 3. The %EA for each category of tool as present in Table 4.

This study’s methodological approach allowed the translation, transcultural adaptation, and validation to Brazilian-Portuguese of existing tools for ADR assessment in children. Using the translated versions of each of these tools, the Cronbach’s alpha coefficient value obtained was acceptable (above 0.7), and the kappa values showed a good (avoidability) and very good (causality) reliability [21, 22]. These results indicate that the translated versions of LCAT and LAAT are reliable and valid for assessing the causality of suspected ADRs associated with anti-infectives and whether they could be avoided based on current knowledge and best evidence available in paediatric patients, respectively [23].

Regarding the translation and transcultural adaptation, it is also essential to highlight some issues relevant to the semantic equivalence of tools. In the translated versions of LCAT and LAAT, the term “drug” has been replaced by “medicine” because, although there is no significant difference in English, in the Portuguese language, “drug” is commonly related to “substances of abuse” [24]. Some terms were maintained in their literal translation of LCAT as “lost-lasting disability OR impairment” and “causal ADR mechanism” as well as in LAAT as “Information available” and “management plan”. The word order in sentence formulation did not represent changes in its meaning. It was decided to keep the asterisks present in the original version of LCAT, with a detailed explanation to help understand the question.

Some technical terms in the pharmacovigilance field in the English language have been changed to facilitate understanding. “Positive rechallenge” (observation of the same signs and symptoms with the administration of the same drug and after the reaction suspected to have disappeared [25] was translated to “reexposição positiva” and modified to “reexposição a uma dose subsequente” in the LCAT, to maintain the notion of temporality in the analysis.

The term “avoidance” (LAAT) has two translations: “evitar” or “prevenir”; with the former preferred by the study team. In most cases, these terms are interchangeable as it is considered that if an event can be avoided in the absence of errors, then it can be prevented [26].

Two aspects should be pointed about ADR causality and avoidability assessment and these may have influenced the degree of agreement observed between reviewers. The first is the quality of the writing of the prospective case reports. The causality assessment pre-supposes the availability of necessary information, particularly about event time to onset, rechallenge, dechallenge, allergies, underlying diseases, and other confounding factors. It is essential to investigate and summarize in the case report, for example, all doses of suspected medicine administered, and whether or not there is any mention in the patient’s medical documentation about the occurrence of previous events to ensure a robust analysis [10, 11]. However, the completeness of the patient’s medical record may be insufficient. Thus, it is strongly recommended that the collection of information is prospective and daily, whenever possible.

The second aspect is the ability of reviewers to search, rank, and interpret scientific evidence and information related to treatment with anti-infectives in children. Moreover, the rapid training provided before the first assessment (to mitigate uncertainties and doubts raised by the team of 27 specialists in the previous phase) allowed discussion and consensus on possible limitations in the search, especially regarding the prevention of ADRs. These strategies may be crucial because each reviewer’s perception relies on their professional background and experience [11].

All assessed cases could be classified using LCAT into one of the four outcomes. Most ADRs were considered definite or probable. LCAT has advantages over Naranjo’s algorithm when rating the causality of ADRs in children [11]. Naranjo’s algorithm is a weighted scoring system tool in which some variables (temporality, rechallenge, and non-existence of confounding factors) have a higher weighting (higher possible item score) and potential influence on the total score than others [10, 11]. On the other hand, LCAT uses a system based on binary decisions with some of Naranjo’s questions combined. If there is a record in the case report about the appearance of the reaction after drug administration, dechallenge, and no alternative cause for the event, it is possible to define the reaction with LCAT in cases of positive rechallenge, or the same event reported in the patient history. In addition, information commonly available only in clinical trials as placebo use is not considered in LCAT. Drug toxicity level detection in the blood (therapeutic drug monitoring), another unreliable source of data, is modified and combined with questions related to any objective evidence of ADR for causality assessment in LCAT [11].

One limitation of the present study was the assessment only of ADRs induced by anti-infectives. However, the English versions of these tools have been used to assess other medical groups using the same process, and there is no apparent reason why this would not be the case for the translated versions. In any case, information availability should be considered and may influence the analysis, as discussed above. Besides that, given Brazil’s cultural diversity, it is recommended that the tools are used in other Brazilian centres to confirm validation [27].

LCAT and LAAT are two easy and quick tools to perform the characterization of ADRs in children. We have translated and transcultural adapted both tools into Brazilian-Portuguese with a careful methodology by a multidisciplinary expert specialist committee, including the authors of original tools [15, 26]. Only 5% of the Brazilian population affirm to know the English language [28]. Thus, this version may be applied by professionals (patient safety teams) and researchers in Brazil in groups or by a single reviewer [14].

It seems appropriate to assume that the availability of LCAT and LAAT tools in Brazilian Portuguese will lead to better care by improved characterization of the risks of ADRs to antimicrobials and better prevention and management practices and promoting organizational changes [14, 27].

The methodological approach allowed the translation, transcultural adaptation, and validation to Brazilian-Portuguese of the two new tools (LCAT and LAAT) for ADRs assessment in children. The availability of the translated versions of these tools may motivate paediatric pharmacoepidemiologic studies and drug use safety in this population.