LAUSR

How to estimate population health loss due to COVID-19 has been greatly contested [1]. This has resulted in various approaches being applied to estimate the years of life lost to premature mortality (YLL), which Ferenci discusses in detail [2]. Due to the overwhelming impact of the COVID19 pandemic, YLL methods have become more frequently used and sometimes, unfortunately, misused. Although the YLL measure has been around since the 1940s, the creation of the Global Burden of Disease study and its associated methodological developments have increased awareness and understanding over its application, although many issues are still challenged [3]. Firstly, it is important to articulate that 'true' YLL can never be observed, and as such, language that indicates that YLL estimates have been underor over-estimated may be misplaced. From the perspective of informing public health policy, the counterfactual to be applied in the estimation of YLL is that of an ideal, aspirational, standard based upon desirably low mortality risks. The merits of resulting YLL estimates should be appraised entirely on their data inputs and choice of ageconditional life expectancy valuation. Secondly, the key utility of YLL estimates lays in comparisons, whether with respect to other health outcomes, across time, or between demographic sub-populations or geographic regions. This requires that the same measure of loss of life years be used for a death at a given age, whatever the cause, the subpopulation in which the death occurs, or the time period in which the death occurs. Therefore, one cannot arbitrarily decide that selected causes will be corrected for co-morbidities, as that will impact the validity of any resulting comparison across causes, populations or time periods. It also imposes an enormous and generally unrealistic demand for data on the distribution of all relevant comorbidities in the people who die of COVID-19, presumably stratified by country and time period as well as on the counterfactual (unobservable) risks of death in the absence of COVID-19. And in the absence of data, additional assumptions would be required. Furthermore any proposals for adjustment would also need to be considered from the alternative perspective, that being that a non-COVID-19 death could be causally related to a prior COVID-19 infection. From this perspective, data availability would be even more likely to be sparser. COVID-19 has represented a novel mortality hazard, and researchers have been trying to apply methods to assess its impact on population health. However, although COVID-19 is novel, the methodological situation to estimating YLL in situations of sudden heightened mortality risk is not. The same is true for most sudden spikes in mortality risk that occurs. For example, in individuals with severe pneumonia, or for those suffering a severe road traffic accident, the risk of death would be expected to be greater in those whose health is impaired, compared to those in excellent health. Should a fatal road traffic accident result in fewer YLL in a frail citizen, compared to someone of the same age without underlying health issues? The answer to this is no, because we are describing health outcomes, which on their own cannot capture the accumulation of risks and occurrence of other health outcomes along the life course. * Grant M. A. Wyper [email protected]