Comparing countries, the risk to NHS staff, and birdsong

Episode Summary

Executive Summary: This episode of More or Less examines COVID-19 statistics, arguing that death data suggests the UK may already have passed its peak, while cautioning that international comparisons are often misleading without adjusting for population, demographics, reporting methods, and excess mortality. It also explains why an NHS infographic on R was confusing, estimates NHS and bus-worker risk, and debunks the idea that birds are singing louder by showing quieter human environments make them easier to hear.

Main Topics: UK COVID-19 death peak and timing of lockdown effects (Priority: 5/5): Jason Oak explains how recalculating deaths by actual date of death rather than announcement date reveals a peak around 8 April in English hospital data, with a slow subsequent decline. The episode links this to earlier distancing measures and suggests infections may have peaked before full lockdown. Why country-to-country COVID comparisons can mislead (Priority: 5/5): Ruth Alexander and Crystal Donnelly discuss how raw deaths and cases comparisons ignore population size, age structure, density, mobility, and differences in death recording. Excess mortality and age-adjusted analysis are presented as better ways to compare countries. Understanding the NHS leaflet on R (Priority: 4/5): Matt Parker resolves a listener puzzle about why an NHS infographic showed 406 cases from an R value of 2.5 over 30 days. The leaflet used cumulative infections, not just the final generation, which explains the mismatch with a simple 2.5^6 calculation. Risk to NHS staff and transport workers (Priority: 4/5): The programme estimates that NHS worker deaths may not be dramatically above what would be expected for a working-age population, though reporting is incomplete and the risk may differ by role. Bus workers in London appear disproportionately affected, with a notably high death rate relative to their workforce size. Birdsong during lockdown (Priority: 3/5): The team investigates whether birds are actually singing louder. Experts say quieter streets, reduced traffic, and spring breeding behaviour make birds more audible; in noisy settings birds may sing louder via the Lombard effect, but overall city birds are likely not louder than usual.

Key Arguments: Recalculating deaths by actual date of death reveals the epidemic curve more accurately than daily announcement totals, which are distorted by reporting delays and weekday effects. The UK’s peak in hospital deaths appears to have occurred around 8 April, implying infections likely peaked weeks earlier and before full lockdown. Raw international comparisons are not fair because countries differ in population size, density, age distribution, interconnection, testing, and how deaths are counted. Excess mortality is a better cross-country metric because it compares a country with its own historical baseline and can be age-adjusted. The NHS leaflet’s R example was internally consistent once cumulative cases, not new cases in the final generation, were used. NHS staff deaths should not be interpreted simply as evidence of extreme occupational risk without accounting for workforce size, age profile, health status, and reporting gaps. London bus workers likely face higher risk than the general working-age population because they have repeated public contact without hospital-grade protection. Birdsong sounds louder mainly because the environment is quieter, not necessarily because birds are increasing volume; in noisy contexts birds adapt their song amplitude to be heard.

Data Points: Peak date for English hospital deaths: 8 April - Estimated by Jason Oak using death-by-date analysis Estimated deaths on peak date: about 800 - Oxford analysis of English hospital death data Deaths one week after peak: around 700 - Shows slow decline after the peak Approximate infection peak inferred from death peak: 18 March - Tim Harford notes a roughly three-week lag from infection to death Full UK lockdown start: 23 March - Compared with earlier distancing measures Earlier distancing guidance: 16 March - Encouraged avoiding travel/contact and working from home School and pub closures: 20 March - Potentially effective measures before full lockdown US COVID cases: 770,000 - Used in country comparison discussion US COVID deaths: more than 41,000 - Used in country comparison discussion UK comparison in deaths per million: 7th - UK ranking after adjustment for population US comparison in deaths per million: 12th - US falls from first in raw totals to 12th per million San Marino population: about 33,000 - Example of why tiny populations can distort per-million rankings Andorra population: 77,000 - Another tiny-country example in per-million rankings Belgium population: 11 million - Large enough population to make high per-million mortality more meaningful NHS leaflet R value: 2.5 - Average infections per infected person in infographic example Days/generation in leaflet example: 5 days - Each transmission generation in the infographic 30-day projection: 6 generations - Used to calculate spread over a month Simple 2.5^6 calculation: just over 244 - What listener Adam expected for new infections Leaflet’s stated total: 406 cases - Discrepancy resolved by cumulative counting UK population: 66 million - ONS estimate used for NHS worker risk calculation Working-age UK population: 42 million - Approximation for comparing with NHS staff Working-age hospital deaths: 2,145 - Used to estimate death risk among working-age people Working-age death rate: a bit over 1 in 19,000 - Back-of-envelope risk estimate NHS staff total: about 1.5 million - Base for occupational risk comparison NHS staff under 65: 1.47 million - About 98% of NHS staff Expected NHS staff deaths if same as working-age population: about 77 - 1.47 million divided by 19,000 Reported NHS staff deaths in newspapers: 100 - Headline figure discussed Estimated NHS staff under-65 deaths: 77 - Working estimate matching expected baseline London transport worker deaths: 26 - Deaths among transport workers in London Bus-company worker deaths: 20 - Subset of London transport worker deaths London bus workforce: just over 31,000 - Transport for London estimate Bus worker death rate: 1 in 1,560 - Computed from bus-worker deaths and workforce size Euston Station and St Paul's noise: many decibels quieter - UCL measurements of city sound levels during lockdown Chif-chaf song level without noise: 61 dB - Field measurement near Manchester airport Plane overflight noise: 72 dB - Measured background noise during bird recording Chif-chaf song level during plane noise: 66 dB - Bird increased song amplitude in response to aircraft noise

Pivotal Quotes: "We think we've already reached the peak. We think that occurred on the 8th of April." — Dr Jason Oak: Explaining recalculated English hospital death data and the likely peak in deaths "So the numbers in the leaflet are not new infections. They're total infections so far." — Matt Parker: Clarifying why the NHS R infographic did not match the listener’s 2.5^6 calculation "Trying to look at who's better or who's worse than us doesn't help us at all." — Professor Crystal Donnelly: Arguing against simplistic international comparisons during an ongoing pandemic

Implications: Listeners should treat COVID statistics carefully: timing, definitions, and baselines matter more than raw headlines. For policymakers and media, excess mortality, adjusted comparisons, and transparent visualizations are essential; for the public, quieter cities can change what we hear, not necessarily what birds are doing.