Excess of mortality by COVID-19

• Why excess of mortality?
• Data limitations
• A data framework
• An example: Peru
• Discussion

Modelling robust estimates of mortality is important towards:

• assessing real magnitude of pandemics
• provides insights on the impact across age, sex, geography, socio-economic status, and other dimensions
• to support informed decision-making and policy development to develop mitigation strategies and allocation of resources.

Measuring the death toll of COVID-19 presents diverse challenges, specially among low and middle and income countries:

• Varying criteria adopted by governments and physicians to catalogue a death as caused by COVID-19:
  • 'Non-COVID-19 excess deaths occur predominantly in older age groups (…) undiagnosed COVID-19 could help explain the rise in these deaths' (ONS, 2020).
• Availability, timing and quality of diagnostic accuracy:
  • 34.7% of patients with positive chest CT findings had negative RT-PCR results of throat swab samples (Lippi et al., 2020)

• Lack of proper identification of death causes:
  • Unusable COD: 18%, ranging from 14% in Australia and Canada to 25% in Japan
  • 17% of all deaths in the 70-plus age group in Japan due to 'old age' (Mikkelsen et al., 2020)
• Under-registration of deaths that varies by age and location:
  • Countries with relatively low levels of mortality and young populations can have completeness rates close to 100% (e.g. some Gulf states), while those with higher mortality/older populations may have rates less than 30% (Adair and Lopez, 2018)

Measuring weekly excess deaths has been suggested as a gold-standard of estimating COVID-19 deaths (Leon,2020):

• Potentially provides the most objective and comparable way of assessing the consequences of the pandemic.

Excess of mortality estimates a expected value of number of additional deaths occurred across a given time period in comparison to the usually expected deaths for the same period based on recent historical data.

Relies on the idea of a counterfactual scenario (a potential outcome model) where COVID-19 had not occurred.