• This interactive graph is based on Office for National Statistics data Provisional counts of the number of deaths registered in England and Wales, including deaths involving the coronavirus (COVID-19), by local authority, health board and place of death in the latest weeks for which data are available. , released each Tuesday. A 'Covid' death is where Covid-19 has been mentioned anywhere on the death certificate.
• The numbers for each Local Authority change substantially from week to week due to chance variation, but when aggregated over England and Wales they show a distinctive pattern.

• The data can be shown either by week of registration, or week of occurrence.
  • Registrations relate to when the death was officially recorded. These are subject to varying amounts of time-lag, caused by weekends, bank holidays and bureaucratic processes. Once published, these numbers do not change.
  • Occurrences relate to when the deaths actually happened. Numbers for recent weeks will change as more deaths get added.
• The final week ends on the previous Friday but one: eg data released on June 9th covered registrations and occurrences up to Week 22, ending May 29th: occurrences include those registered up to three days before the release, eg June 6th for the release on June 9th.
• In a declining epidemic, ‘Week of Occurrence’ tends to produce lower numbers of all-cause deaths, and hence fewer excess deaths, than ‘Week of Registration’. This is because (a) Registrations reflect the number of deaths occurring further back in the 'spike', (b) further deaths will be added to the Occurrence data.

• It is impossible to use these figures to estimate how many deaths might have occurred if different policies had been adopted, eg either earlier or no lockdown.
• This interactive graph was written by Harry Giles in collaboration with David Spiegelhalter. Please note that while we will try and answer technical queries regarding the methods, we cannot discuss events in specific Local Authorities.

WARNING: The numbers in the summary table can be extremely sensitive to start/stop dates, and whether looking at Registrations or Occurrences. Have a play around with the app to develop your understanding of this relationship.

Once the app appears, please allow a few seconds for the app to load its data and render its display. Go to full page version of the app.

• The 'baseline' for registrations is taken from the tables accompanying the section on Deaths registered by place of occurrence in the ONS Weekly death registration bulletin.
• The 'baseline' for occurrences is derived from 'five-year' average data for each week and place of death in England and Wales, scaled in proportion to the total number of deaths in each place in that LA in the first 10 weeks of the year, so this is not a five-year average for that specific LA. Since the observed national deaths for weeks 1-10 in 2020 are below the 5-year average, the observed counts in each LA are generally below the baseline, right back to the start of the year.
• There is no 'correct' baseline for calculating excess deaths, which are essentially constructed rather than counted. For example, the Institute and Faculty of actuaries Mortality Monitor uses 2019 as a 'non-epidemic' year baseline from which to calculate excess deaths using registration data.