Models for mortality require tailoring in the context of the COVID-19 pandemic – Authors' reply

We thank Arielle Lasry and Roberta Horth for their comments on our study¹ and agree that long-term mortality models, applicable to people with different underlying conditions, have an important ongoing role in the COVID-19 response, and that they require further development to include wider determinants of health.

A fundamental question to any patient with a condition would be, we believe: how have my chances of surviving 1 year changed as a result of the COVID-19 emergency? Clinicians and policy makers currently have no consistent way of answering this question; at best, current approaches provide risk estimates for one disease at a time. Lasry and Horth point out the potential benefits of answering this question, and our Article has provided a prototype for development.¹ A strength of our approach is that we estimated absolute risk in 5-year age bands; increasing age has now been established as the most important risk factor for severe or fatal COVID-19. For example, people aged 80 years or older have a 20 times increased risk compared with those aged 50–59 years.² We agree with Lasry and Horth that it is important to consider the health service, societal, psychological, and economic consequences of the emergency; each of which might affect all-cause mortality more than any association with COVID-19.

Our approach of providing absolute risk information on people similar to a given patient across a range of diverse health conditions is novel and might have uses, irrespective of COVID-19; previous approaches to risk are focused on one or a small number of related diseases. We have shown the high prevalence of multimorbidity for cancer³ and cardiovascular diseases⁴ in relation to COVID-19 excess mortality.

The OurRisk.CoV calculator accompanying our Article explicitly allows the user to choose different relative risks for each condition. Although evidence is emerging of how the short-term (eg, 90 day) risks of the specific outcome of COVID-19 varies across approximately 50 underlying conditions,² there is still little information on how long-term (≥1 year), all-cause mortality has been affected in people with each condition.OurRisk.CoV has received more than 1·3 million visits since its release on May 12, 2020 (660 000 unique users). Despite the calculator being only a prototype for researchers to explore data, we believe that its use, and user feedback, strongly supports the public need to understand risk, tailored to age, sex, and a much wider⁵ range of underlying conditions. The real challenge is not only estimating risk in a more granular way, as we have attempted to do, but also in communicating the concept of risk to populations and individuals.