Rapid Review of NHS England’s statistics on daily Covid-19 deaths in hospitals

Dear Mark,

CORONAVIRUS (COVID-19) DAILY DEATHS

I am writing to endorse the approach you have taken to develop the daily deaths statistics that NHS England has published in response to the coronavirus (COVID-19) pandemic. The statistics report on the deaths of patients who have died in hospitals in England and had either tested positive for COVID-19; or where COVID-19 was mentioned on the death certificate. This information has proved critical for decision-makers and scientists, for example, in managing resource throughout the pandemic and reviewing lockdown measures in England.

My team has conducted a rapid regulatory review of these statistics. We have reviewed the extent to which they have been produced in accordance with the Code of Practice’s Trustworthiness, Quality and Value pillars, while taking account of the pressures that NHS England has faced to deliver timely statistics about a rapidly evolving national emergency. I really appreciate your colleagues’ positive engagement throughout this process. A summary of our findings, including recommendations, is set out below.

Value

  • We welcome the rapid development of these statistics so they could meet the national demands for mortality data and inform decision-making, for example through their use in the UK Government’s COVID-19 daily briefings. These statistics have proven to be invaluable to both policy colleagues, the media and the general public as a timely, robust source of information on hospital deaths in England.
  • NHS England has striven to improve and innovate these statistics since their first publication, both in response to national data requirements and to enquiries from the media and general public. For example, NHS England improved the presentation of its data and statistics by using charts and graphs to illustrate findings. NHS England has also started publishing detailed weekly tables which include breakdowns by ethnicity, gender, age groups and pre-existing conditions, which reflects a positive response to user demand. Whilst we consider these statistics and data to be incredibly valuable given their high level of insight, no time series data are available in this form, only overall totals. We ask you to talk with your users and reflect on how time series data might add value for users of the data about ethnicity, gender, age groups and pre-existing conditions.
  • Whilst the COVID-19 daily deaths publication is easily accessible through NHS England’s website; it is only available in xlsx format which limits its ability to comprehensively meet user needs. For example, this format is not machine-readable which limits users in producing their own analysis of the data and statistics. We encourage you to reflect on the accessibility of the data and statistics to ensure they meet a wide range of user needs.
  • We also welcome NHS England’s collaboration with the University of Bristol and its collection of data on the deaths of people with learning disabilities or autism which is due to be published shortly. This will offer additional, valuable insight into the impact of COVID-19 on these groups and help inform decision making and public debate.
  • A potential source of confusion for users has been around understanding the impact of any lags in recording deaths on the statistics. We appreciate that the data are presented on the basis of the date of death rather than the date of recording, and that in your charts you illustrate clearly the impact of any later recording on the data series. As mentioned earlier, the availability of the data behind the charts, in machine-readable form would aid accessibility and transparency in order to aid further analysis and insight.

Quality

  • In March 2020, NHS England moved quickly, from its Incident Control Centre collecting operational incident response data about COVID-19 deaths in hospitals through its network of regional contacts, to establish the COVID-19 Patient Notification System. Developing this standardised collection tool so quickly was an important and impressive endeavour, as was establishing a dedicated team to work closely with yourself, as Head of Operational Information.
  • It is very important that the strengths and limitations of these statistics are transparent, and that users can be assured of their quality. We appreciate the headline data notes that your team has published alongside the statistics, though we consider more information about the quality of the breakdowns by ethnicity, gender, age groups and pre-existing conditions, which evolved later, should also be included to provide users with further reassurances around these data.
  • We welcome the recent move to publish the data collection user guide which includes information about how the data are collected and validated. It was good to hear from your team about how it works with its national and regional contacts to assure the quality of the data, and that the team sense checks with other sources. We have encouraged your team to reflect on their own quality documentation in line with the Government Statistical Service guidance on urgent quality assurance of data.
  • During the pandemic, the data landscape has evolved and become more complex, bringing with it the greater risk of presenting a confusing picture for users. The key example being the early shift in focus from examining hospital deaths to Department for Health and Social Care (DHSC) starting to publish daily statistics on deaths in all settings. We welcome the recent information published alongside the statistics that provides some clarity about the different methodologies used to report on COVID-19 deaths by the DHSC and the Office for National Statistics (ONS). We think it would also be helpful if you could briefly explain for users what statistics are available about COVID-19 deaths in hospitals for the other countries of the UK.
  • As England has moved through the first wave of the pandemic, questions naturally arise about the future of these statistics – for example, “for how long will daily reporting continue to make sense?” and “what potential lessons can be learned that will inform how to operate during a potential second wave?” We were very encouraged to learn that your team has held a pause and review with stakeholders, including representatives of those in the supply chain, to reflect and learn lessons. Involving users and partners closely in decisions on future changes will be important to the development of these statistics.

Trustworthiness

  • NHS England’s orderly and timely release of these data has been invaluable in allowing users to act quickly and has reflected the pace of change during the COVID-19 pandemic. Where occasional errors have arisen in the data, these have been reported quickly and transparently and the team assured us that lessons have been learned.
  • We welcome that NHS England has published a pre-release access list alongside these statistics. We understand that the list of those with early access has reduced significantly in number since the broader DHSC statistics became the headline government figures. For clarity, we would advise that you update the published pre-release access list with the now reduced list and indicate how far in advance access is granted.

We look forward to seeing these statistics continue to develop. Based on the findings of our review you can include a statement in your release such as “These statistics have been produced quickly in response to developing world events. The Office for Statistics Regulation, on behalf of the UK Statistics Authority, has reviewed them against several key aspects of the Code of Practice for Statistics and regards them as consistent with the Code’s pillars of Trustworthiness, Quality and Value.”

I am copying this letter to Nalyni Shanmugathasan and Paul Shafee, the responsible analysts at NHS England and NHS Improvement, and Stephen Balchin, Head of Profession for Statistics at DHSC.

Yours sincerely

Ed Humpherson
Director General for Regulation