Deaths at home, supermarket infections and the Cobra effect

Episode Summary

Executive Summary: This episode debunks misleading COVID and public-health claims: reported deaths are not the same as deaths occurring that day; home deaths rose largely as a location shift from hospitals rather than a clear surge in non-COVID mortality; isolation payments likely improve compliance despite theoretical perverse incentives; supermarket transmission was overstated from misread data; and Scottish drug-death figures are not artificially inflated by including accidents like car crashes.

Main Topics: COVID death reporting and misleading headlines (Priority: 5/5): David Spiegelhalter explains that daily UK death headlines reflect reporting dates, not actual dates of death, making single-day comparisons highly misleading. Death location shifts from hospitals to homes (Priority: 5/5): The episode examines evidence from England, Wales, and Scotland suggesting many deaths have moved from hospitals to homes during the pandemic, raising questions about end-of-life care and health-seeking behaviour. Self-isolation payments and the Cobra effect (Priority: 4/5): Abby Adam-Prassl discusses whether extending £500 self-isolation payments could backfire by incentivising infection, but argues the net effect is likely positive. Where COVID is really spreading (Priority: 5/5): The programme challenges the claim that supermarkets were the top source of transmission, showing the statistic was a misreading of Public Health England data. Scottish drug-death statistics and misinformation (Priority: 5/5): The Ferret Fact Service checks claims that Scotland’s drug deaths are inflated by broad definitions, concluding that excluded categories like car crashes are explicitly not counted. Limits of statistical certainty in epidemics (Priority: 4/5): Across the episode, experts stress that data can reveal patterns and risks, but often cannot determine exact causation or individual circumstances.

Key Arguments: Reported COVID deaths are delayed administrative counts, so a headline for one day may substantially overstate the number of people who actually died that day. Comparisons between the current COVID death peak and the first wave are distorted because early testing was limited, causing the first-wave totals to be undercounted. Excess deaths at home do not necessarily mean excess non-COVID mortality; many deaths may simply have shifted from hospitals to homes. Scottish data suggest increases in home deaths from cancer, heart disease, and stroke, but not a large rise in overall non-COVID deaths, implying a location shift rather than entirely new mortality. Self-isolation payments can improve compliance because people with low incomes face real costs from staying home when ill. Theoretical perverse incentives from self-isolation payments exist, but there is little empirical evidence that they outweigh the benefits. The supermarket statistic was derived from a table of places visited by infected people, not proof that supermarkets were the source of infection. Contact tracing studies show household and close family/social contact are the highest-risk settings for transmission, while data on supermarkets and similar venues are limited. Scottish drug-death figures are not inflated by counting traffic accidents, drownings, assaults, or AIDS-related deaths; those are explicitly excluded. Claims that Scotland’s drug-death data are incomparable to other countries do not change the overall picture: Scotland’s rate remains far higher than European peers.

Data Points: Reported UK COVID deaths on 20 January: 1,820 - Headline figure discussed as a reporting-day total, not a count of deaths that occurred that day. Deaths actually occurring on that day so far: at least 1,028 - David Spiegelhalter said later registrations indicate the true number was much lower than the headline. Estimated eventual deaths for that day: about 1,100 - Spiegelhalter said the total may rise further but not near 1,820. First-wave UK dashboard peak deaths: 1,073 - Dashboard figure for 8 April, noted as undercounting actual deaths because of limited testing. First-wave actual deaths on that day: 1,456 - ONS-based estimate of deaths registered with COVID on 8 April. Excess deaths in homes in England and Wales: 42,000 - Up to 8 January, showing large increase in deaths at home during the pandemic. COVID deaths among those home excess deaths: 4,000 - Most excess home deaths were not directly assigned to COVID. Fewer non-COVID deaths in hospitals: 42,000 fewer - Mirrors the increase in home deaths, supporting a shift in place of death. Extra deaths in Scotland in 2020: about 7,000 - Compared with roughly 6,000 COVID deaths. COVID deaths in Scotland in 2020: about 6,000 - Used to show that overall excess mortality broadly matched COVID deaths. Extra cancer deaths at home in Scotland: 1,900 - Attributed to a rise in home deaths rather than a clear overall increase in cancer mortality. Extra heart disease and stroke deaths at home in Scotland: about 1,200 - Part of the home-death shift noted by Spiegelhalter. Government-backed self-isolation compliance study: 11% - September 2020 study finding only 11% stayed indoors for the full quarantine period. Self-isolation payment: £500 - Payment introduced for some people asked to self-isolate and discussed as a possible broader policy. Original supermarket transmission claim: 20% - Misread from a Public Health England table; it did not mean 20% of cases caught COVID in supermarkets. Scottish drug deaths in 2019: 1,264 - Highest number since records began 25 years earlier. Year-on-year increase in Scottish drug deaths: 6% - 2019 figure was 6% higher than the previous year. Decade-long increase in Scottish drug deaths: 132% - Shows more than doubling over ten years. Scotland drug deaths per million population: 295 - 2018 comparable statistic used in European comparisons. Sweden drug deaths per million population: 81 - Nearest higher comparator in the European table cited. UK drug deaths per million population: 76 - Used for comparison with Scotland. EU average drug deaths per million population: 24 - Shows Scotland far above the European average.

Pivotal Quotes: "all these headlines I feel were deeply, deeply misleading" — David Spiegelhalter: On media coverage of daily COVID death totals that treated reported deaths as same-day deaths. "The Cobra effect is a term that basically refers to a scenario where the incentives designed to solve a problem not only don't work, but they actually backfire completely" — Abby Adam-Prassell: Explaining the risk that self-isolation payments could unintentionally encourage infection-seeking behaviour. "we just don't know that" — Charlotte MacDonald: On whether supermarket visits were the source of infection, after unpacking the misread Public Health England data.

Implications: Listeners should be cautious with headline statistics: reporting dates, definitions, and table design can radically change meaning. Policymakers need evidence-based communication and targeted public-health incentives, while audiences should resist simplistic claims about transmission, mortality, or drug-death data.