Global incidence and mortality of severe fungal disease

We read, with great interest, David W Denning's estimates on the global incidence and mortality of severe fungal disease.1Denning DW Global incidence and mortality of severe fungal disease.Lancet Infect Dis. 2024; (published online Jan 12.)https://doi.org/10.1016/S1473-3099(23)00692-8Summary Full Text Full Text PDF PubMed Scopus (21) Google Scholar We agree there is a crucial need for accurate estimates of fungal diseases, the absence of which contribute to a lack of awareness, underdiagnosis, equity concerns, and inadequate research funding. However, we have several concerns with the estimation approach and representation of estimates presented in the study. First, the author frequently uses a single global estimate informed by heterogenous data throughout the estimation process that fails to account for the geographic heterogeneity, differences in age groups included in studies, and temporal trends that are present in the underlying data. Although there are many examples, a specific one is the approach to Candida. The ability to detect and appropriately treat candidaemia differs across geography2Kaur H Chakrabarti A Strategies to reduce mortality in adult and neonatal candidemia in developing countries.J Fungi (Basel). 2017; 3: 41Crossref PubMed Scopus (49) Google Scholar—yet the mortality estimate presented used only the mean case fatality rate (CFR) and a global ratio of treated to untreated cases. Although we appreciate that data sparsity and computational demands make country-specific estimates challenging, an estimate at the regional level would be more justifiable than the global approach. Second, the author makes several anti-conservative assumptions that risk overestimation of burden. An example is the CFR of invasive aspergillosis in acute leukaemia after haematopoietic stem cell transplantation, of which the meta-analysis used to inform the estimate found a crude CFR of 29% with CFR ranging from 12% with prophylaxis to 45% without prophylaxis. The author uses the 45% mortality globally as opposed to a more nuanced approach based on geography, or simply using the crude estimate or lower bound from the meta-analysis. Similarly, the CFR for invasive aspergillosis in chronic obstructive pulmonary disease (COPD) describes a range from 43% to 72% and uses the 72% figure for the mortality estimate. In the appendix of the Article,1Denning DW Global incidence and mortality of severe fungal disease.Lancet Infect Dis. 2024; (published online Jan 12.)https://doi.org/10.1016/S1473-3099(23)00692-8Summary Full Text Full Text PDF PubMed Scopus (21) Google Scholar it is noted that an abstract described a CFR of 4% and the rationale for its exclusion is not described, as other abstracts were used in the estimation process. Additionally, the crude CFR data inputs often include studies describing CFR in the intensive care unit, such as the Hohmann study describing 77·3% mortality among 22 patients with candidaemia in an intensive care unit;3Hohmann FB Chaves RCF Olivato GB et al.Characteristics, risk factors, and outcomes of bloodstream Candida infections in the intensive care unit: a retrospective cohort study.J Int Med Res. 2023; 51 (3000605221131122)Crossref PubMed Scopus (2) Google Scholar the crude CFR based on these data are subsequently applied to any case of the fungal disease of interest. The use of this kind of non-representative data almost certainly inflates the CFR and thereby increases the estimated mortality of severe fungal infections. Third, we are concerned that many of the results would not be reproducible if another group was tasked with the same methodological approach, because many of the key steps rely on expert opinion from a single expert. Although we do not call into question Denning's expertise in the field, a more robust approach would have been to use the global action for fungal infections network to query a wider group of experts and present a range of plausible estimates. This is supported in structured expert judgement protocols as well, which were designed to enhance the reliability of model assumptions by underscoring the importance of seeking judgments from multiple experts and encouraging experts to quantify their uncertainty when expressing their assessments.4Swallow B Birrell P Blake J et al.Challenges in estimation, uncertainty quantification and elicitation for pandemic modelling.Epidemics. 2022; 38100547Crossref Scopus (12) Google Scholar Finally, the representation of mortality and incidence as point estimates belies the uncertainty underlying every step in the estimation process. Some quantitative description of uncertainty should be expected for an estimate of this kind and the justification for their omission failed to convince these authors that it was the appropriate decision. Similarly, a sensitivity analysis using the lower bounds would be a welcome exercise and would help describe the floor of plausible burden caused by severe fungal disease. A consequence of these limitations is that it is difficult to confidently compare the estimates presented to other global estimates of diseases or specified pathogens for priority setting. Recent estimates from the Global Burden of Disease describe 1·2 million (95% UI 1·1–1·3 million) deaths due to tuberculosis in 2019;5GBD 2019 Diseases and Injuries CollaboratorsGlobal burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019.Lancet. 2020; 396: 1204-1222Summary Full Text Full Text PDF PubMed Scopus (7590) Google Scholar this number would be dwarfed by the estimated 1·8 million deaths associated with aspergillus and even the 1·3 million deaths directly attributable to aspergillus estimated by Denning. If this figure is true, there ought to be a reckoning within global health priorities, which is why there is an urgent need for reproducible and comparable estimates of severe fungal infections to address this often underappreciated public health need. We declare no competing interests.