Discussion
Spurious BLA labels are dangerous to patients and pose a public health emergency in terms of health care costs and driving ABR. In what we believe to be the first publication on the epidemiology of reported BLA in Africa, we found that the prevalence of reported BLA in hospitalised patients was only 3.3%, a rate considerably lower than HIC rates. The self reported BLA rates were lower in government-funded hospitals than in the privately-funded hospitals. Sadly, very few participants in our study with a BLA label returned for allergy de-labeling.
The lowest rate of self-reported BLA from a HIC inpatient settings is 9.9% of 1738 patients enrolled over a one-year period in Montreal, Canada (9). Our study reports a rate less than half of this low-end rate. Furthermore, more than 80% of our small number of self-reported cases of BLA were considered low risk for true BLA, meaning that the prevalence of confirmed beta-lactam hypersensitivity may in fact be even lower in our South African population. There are several possible contributing factors that may explain both a lower rate of BLA label as well as a possible lower rate of true beta-lactam hypersensitivity in our population. Antibiotic prescribing patterns differ across the world, and involve a complex interplay of social, patient, provider, and economic factors(18). The majority of BLA labels are the result of viral or drug-related exanthems in childhood (2). But difficulties in accessing healthcare in South Africa (particularly in rural areas) resulting in less antibiotics being given(18), may limit the childhood exposure to BLA allergens and so contribute to less identification of true BLA in our setting.
Drug-related (and viral) exanthems, considered to be the predominant driver of incorrect BLA labels, are predominantly T-cell mediated, the majority of which are HLA-restricted. A recent genome-wide association study (GWAS) linked HLAB55:01 with penicillin allergy label. Interestingly, this allele has relatively low frequencies in Black African populations. This contrasts with another recent GWAS amongst confirmed cases of immediate hypersensitivity to beta-lactams, that identify HLA-DRB1*10:01 as a risk allele. HLA-DRB1*10:01 is carried twice as frequently in those of African ancestry (19). Other genetic studies in European populations with confirmed beta-lactam hypersensitivity have linked polymorphisms in cytokine genes (TNFA, IL-13, IL-4, IL4R)(20), which are also likely to be population-specific. Thus, it is possible that genetically African populations may have similar or higher risk for true immediate beta-lactam hypersensitivity, but less likelihood of mild delayed reactions, the major drivers of incorrect BLA labeling in childhood. Finally, skin colour may impact the detection of maculopapular exanthems in those with Fitzpatrick skin type IV, V and VI (Lehloenya North American clinics accepted for publication), as erythema and fine rashes are harder to detect in pigmented skin. Taken together, there may be several genetic, and biological factors that explain self reported BLA prevalence in in African populations.
BLA labels were significantly more common in patients attending private compared to (government funded) hospitals. South Africa is the most inequitable country in Africa with a GINI coefficient of 0.63 in 2015, an inequality reflected in its public versus private healthcare. Race and socioeconomic status are a major factor(21-23). The population utilizing privately funded health care sector in South Africa more closely resembles that of a HIC in that the predominant racial group is white (European ancestry)(22) with a higher income. The population that accesses government-funded health care has a high percentage of black African, Indian, or mixed race patients. Thus, the difference in BLA rates between government and private hospital inpatients may reflect both differences in access to healthcare in childhood with different levels of antibiotic exposure, as well as genetic and skin pigment biological factors related to populations of origin. Further qualitative and basic science research is required to understand these differences.
Using the PEN-FAST tool, most patients with BLA label in our study were considered low risk for true type 1 hypersensitivity to beta-lactams. However, despite several attempts to contact and assist patients to attend our tertiary allergy clinic at GSH, only 1 patient could complete direct oral challenge for de-labeling. Multiple factors play into this failure, including: patient-perceived lack of importance of carrying a BLA label, fear of the testing and procedures or the time involved, changes in contact details and a mobile patient population, or even lack of resources to return for clinic visits (18, 22, 24). This inability to have patients return and attend allergy clinics or elective procedures has been highlighted, even in HICs, and undoubtedly aggravated across the world by the COVID-19 pandemic (25). In paediatric patients recommendations now exclude the use of SPT as a possible barrier to care and with concerns regarding increased false positive test rates(26). This data supports that the only viable option for BLA de-labeling in LMICs will be direct de-labelling or challenges in low-risk patients by non-allergists at the bedside. The development, validation and implementation of risk-stratification tools will be critical to this effort. A comprehensive framework for incorporation of BLA de-labeling in LMIC/LIC has been recently outlined (4).
The inclusion of several hospitals across different levels of the health system, and inclusion of private and government-funded hospitals, improved the generalizability of this data. However, the fact that the study was only performed in one city of SA is a limitation. The COVID-19 pandemic meant that there was a several month gap between the initial survey at GSH in April 2019 and the other surveys in the study, as well as limiting bed occupancy at the time of surveys in several hospitals, which may have impacted results. The lack of reported BLA in patients under 18 years is likely due to the low bed occupancy rates the high percentage of paediatric patients under the age of one year in the cohort, likely limits the generalisability of paediatric data.
In conclusion, this study provides the first data on the epidemiology of BLA in Africa and demonstrates that the overall prevalence of BLA is much lower than that reported in HICs. Furthermore, there is disparity of our results across health sectors, which highlights several of the complex social and biological determinants of both true and incorrectly labeled BLA. The inability to confirm BLA in the majority of cases through skin testing and direct oral challenge in the allergy clinic highlights the challenges of incorporating BLA de-labeling strategies in antibiotic stewardship programs in LMICs, and demands a different solution than HICs, by shifting the site of testing to the bedside. Epidemiology data from other LMIC is required to confirm our findings and help LMIC policy makers decide on the importance of targeting BLA de-labeling in local antibiotic stewardship efforts. Affordable strategies for direct de-labeling or inpatient challenge are supported by this data.