Discussion
Using our population-wide data, CEDAR was able to generate a registry of comprising of more than 7,337,778 individuals. To our knowledge, this is the largest population-wide drug epidemiology study to date; allowing detailed characterisation of the absolute-prevalence and -incidence, individual drug culprits, as well as the age of incident drug allergies of our entire population.
Our study is also the first to investigate the longitudinal incidence of drug allergy labelling on a population-wide basis. CEDAR revealed that the annual incidence of reported drug allergy Hong Kong remained relatively stable over the 5-year study period (331-352 per 100,000 population), until a significant 16.3% drop in 2020. This was likely due to the COVID-19 pandemic resulting in a population-wide reduction in medical visits, drug prescriptions, or allergy reporting.21 It would be interesting to see if the incidence of drug allergy will rebound back to pre-COVID figures following relaxation of social distancing restrictions. Our study also confirmed that about 1 in 18 (5.61%) of Hong Kong’s population have physician-reported drug allergies, which differs quite substantially with other regions (especially in Western cohorts), as well as our own previous hospital-based studies.8,22-26 Although drug allergy prevalence varies across time and region (likely due to combination of differing prescribing practices and biological differences) we postulate discrepancies may also arise due to previous type I errors from inadequate sampling, which would be minimised from this population-wide study. This needs to be confirmed by collaborative inter-regional and -ethnic big data studies in the future.
Interestingly, our unified physician-reported drug allergy system revealed that most drug allergies were reported among middle-aged individuals, with disproportionately higher incidence among those above 40. This is contrary to the traditional dogma reflected from the previous experience from Western cohorts, suggesting the majority of individuals acquire drug allergy labels during childhood and carry them into adulthood.27 We postulate this stark difference may be due to either genuine inter-population differences, or a paradigm shift that has remained unnoticed due to the paucity of prior systematic or large population-based drug allergy studies. This emphasises the importance of region-specific big data studies, which can discover unique features and trends of drug allergy labelling among different populations. Furthermore, identifying target subpopulations would inform the best strategy for allergy education and proactive delabelling at a population level. For example, there are significantly fewer adult allergists than paediatric allergists in Hong Kong (1:2,800,000 vs 1:540,000 per population), which is disproportionate to the incidence and burden of drug allergy in our locality.28 Hence, we advocate for more active drug allergy delabelling initiatives among adults, further resources and positions for training of adult allergists, as well as strengthening the ability of frontline healthcare professionals to discern symptoms of genuine allergy – especially among adult patients.
This study also confirms the severity of a penicillin-dominated drug allergy landscape in Hong Kong. Within anti-infectives, beta-lactams constituted the majority of all reported drug allergy, with penicillins contributing to almost 82.9% of all beta-lactam labels. Penicillins were among the most commonly reported culprits of drug-induced anaphylaxis and Steven-Johnson syndrome. However, our previous studies revealed that only 10-13.8% of reported penicillin allergy are found to be correct after evaluation.1,10 Incorrect labels lead to unnecessary penicillin avoidance, posing immense challenges in antimicrobial stewardship. This is especially relevant in Hong Kong, where there has been an upsurge of various multi-drug resistant organisms.29,30 Penicilin allergy labels have also shown to affect geriatric and immunocompromised patients, associated with a multitude of adverse clinical outcomes, including increased healthcare costs, more frequent and longer hospital stays, and even death.22,31-33 Despite the severe lack of allergists in the Asia-Pacific region, previous multi-disciplinary initiatives have shown much promise in tackling drug allergies.10,34,35For example, a nurse-led penicillin allergy delabelling initiative demonstrated superior outcomes compared to traditional allergist evaluation in Hong Kong and similar strategies employing trained pharmacists are underway.10 Further research on wider applications of similar multidisciplinary drug allergy initiatives is warranted.
Given the direct and indirect consequences of genuine or perceived excipient allergy – from rare cases of excipient-associated anaphylaxis to inappropriate fear of excipient allergy leading to vaccine hesitancy – the ability to accurately disprove excipient allergy is of paramount importance.13-17 However, diagnosis of excipient allergy is near impossible when excipient information is not available. This is a common problem for many regions where no pharmaceutical legislations exist to mandate excipient disclosure in registered medications. Despite our best efforts to construct a excipient registry, we were unable to obtain excipient information from the vast majority of drug formulations. Using PEG as an example, we were able to confirm its presence or absence among only 724 of 2,070 formulations investigated, only representing a mere 36% included in the registry. This made it impossible to exclude or confirm presence of PEG in all formulations of any of the top 50 reported drug allergy culprits in Hong Kong. Lack of transparent drug information impedes in comprehensive evaluations in drug allergy, and undermines the safety and public confidence of population-wide vaccination programmes. Despite past efforts to advocate for regulartory improvements, ingredient disclosure remains to be a persistent issue in our locality and legislative changes have yet to materialise.16,17
Given its observational nature, this study has several limitations. First, only limited patient data was available, and more detailed clinical data such as comorbidities, medication use or hospitalisation records were not available for further subgroup analysis. Second, we did not include reported drug allergies which were entered as ‘free text’ into the patient’s electronic health record that could not be manually converted into structured item (often due to incomplete information). Third, drug allergy labels in Hong Kong were physician-reported, and albeit more preferred than patient-reported adverse reactions, may still be inaccurate and do not reflect only genuine and confirmed drug allergy. This highlights the importance of dedicated and interventional studies in the future.
In conclusion, population-specific drug allergy research is needed to inform the directions of appropriate public health initiatives in an evidence-based manner. The use of big data can unlock new dimensions of drug allergy research - such as drug allergy incidence and comparative age distribution of reported drug allergies. Using our population-based data, we were able to report the longitudinal incidence of reported drug allergy, specific allergy culprits as well as identifying disproportionate drug allergy labelling among middle-aged patients. Future inter-regional and -ethnic big data studies will be required to confirm external validity of our findings to other populations. Lastly, excipient registries remain ineffective and proper allergy evaluation cannot be conducted in countries where excipient information are unavailable. International and collaborative efforts should be united to advocate for universal recommendations toward ingredient disclosure for all registered medications.