Two wrongs don’t make a right – excess deaths

The ONS has published the first weekly deaths data for 2023 (“Deaths registered weekly in England and Wales, provisional: week ending 6 January 2023“). It confirms that they are both, as with 2022, using a sub-optimal methodology for calculating the five-year average against which to measure excess deaths and failing to offer any logic for their choice. I have written to the Office for Statistics Regulation and the National Statistician to ask them to review the figures. Hopefully the regulator will remove the National Statistics imprimatur from the series and encourage ONS to produce more meaningful numbers.

Advertisement

Excess deaths: ONS is too high

Excess deaths was meant to be the preferred way to measure the impact of the pandemic and to allow the best, while still imperfect, country comparisons. While there have indeed been large numbers of excess deaths in the last three years we are left unsure about how many, and what the “true” toll is of Covid-19. The core issue is that the figures put out by the ONS – the most visible in the media – use a simplistic methodology that gives an estimate of excess deaths that is far too high. In addition, the inclusion of 2021 in the base from which to judge excess deaths in 2022 was not only inexplicable but has distorted the in-year pattern ONS is showing. The bottom line is that the ONS methodology for choosing the number of expected deaths is not fit for purpose – it should be reviewed and amended, revising past data. More widely, there are too many public sector sources of excess deaths numbers. They are hard to find and then give different estimates of different variables based on different methodologies. These need to be whittled down or merged, or at least properly explained. If the bodies can’t do it themselves, and UKSA fails even to try, ministers should step in. After the millions spent on Covid-19 statistics, and the Hallett Inquiry under way, the nation deserves to have the best data readily accessible. The different figures showing different trends leave us unsure about the impact on deaths in 2022 of both Covid-19 and NHS issues – the hot topic of the month.

Trendspotting – guess the next number

Many people in their personal or professional lives have to do a bit of forecasting. You know the sort of thing: given this run of data what’s the next number? We’ve done it since we were kids: 2, 4, 6, 8 ….., yes, 10. Join me on a brief journey to guess the best numbers to come next in this series. This exercise matters as I think that a group of people who ought to know better have made a mistake that harms their credibility and the structure they work within, and is affecting both our trust in supposedly independent experts and decisions we all take in our lives.

Media coverage of excess deaths

I feel the ONS calculation of excess deaths is flawed and too high but, not surprisingly, most reporting media run with the high numbers – other estimates of excess death are available but get less attention. Here is a selection of comments from the mainstream media on the latest release of 10 January 2023 – universally the outlets present the story as an omni-crisis in the NHS. Social media either follows that line or takes a more extreme view – essentially the anti vaxxer line that the mysterious, unaccounted for deaths are a consequence of having taken the vaccine. This shows how important it is that ONS does make it easy for its releases to give the wrong impression. It also shows how the supposed system checks and balances – be that the UK Statistics Authority board, the OSR regulator, or fact checkers (at the BBC or elsewhere) – do not seem to be on top of such topics.

The excess death failure of ONS

The weekly death data from ONS, which include the excess death figures, have been hugely misleading in 2022. It is impossible to know if it is institutional, academic or political pressure to keep the death figures high or the ONS just slavishly following naïve practices that are not fit for the times we live in. It seems very unfortunate that the ONS excess death figures suggest that 10s of thousands more have died during the pandemic than best estimates might suggest. It is a terrible indictment of the ONS and its regulator, the OSR – just when the Hallett Inquiry needs some accurate figures they produce questionable data.

Producer price errors

For those of us who have had a ringside seat for the RPI debacle over the last decade it is worth noting the current Producer Price Index (PPI) issues. In statistical terms producer prices and consumer prices face many common issues so a sign of problems in one always sends a chill down the spine – though ONS says that these issues are affecting only the PPI. It’s not yet clear what the problem is though it’s serious enough to have the next month’s figures pulled. While any mistake is unfortunate, it does look this time as if the ONS has identified the problem, been (fairly) honest and open about it, and withheld publication. This is good. If only, many will say, they had done the same with the RPI a decade ago. So much pain could have have been spared. Meanwhile we will have to do without PPI data for 2022 on oil and food – two of the more interesting sectors – until some point the new year.

Churn, vacancies and job confusion in ONS?

A close look at the various org charts shows not only a lack of attention paid to keeping basic facts up to date (as discussed before) but a degree of churn and missing personel in key jobs that must be a cause for concern. To be without key personnel on population (when the census is being published and there’s much worry about the levels of immigration) and lack of leadership on data science seems careless.

The UKSA review

A number of my recent posts have mentioned the promised review of UKSA, and latterly the delay in producing it. Activity seemed to have gone quiet but progress is now being made. A letter was sent last week from Baroness Neviille-Rolfe, the Cabinet Office minister responsible for “Sponsorship of UK Statistics Authority”, to the chair of PACAC, the parliamentary committee that inherited the responsibility of UKSA oversight. It says that “this is the right time to press ahead with a public-facing review to enable the UKSA to prepare for the strategic challenges ahead”. The provisional themes it will cover are efficacy, governance, accountability and efficiency. This is a suitably broad and wide-ranging draft remit. We await news on who will chair the review and its time scale.

UKSA and GSS hierarchy

Who are the big cheeses of the government statistics world? I have no idea which is a shame as that would seem to be the field of plausible “internal” candidates in the race to become National Statistician. Three possible places are the upper reaches of UKSA, the heads of profession in the wider Government Statistical Service (GSS) and existing non-execs. The non-execs are easy to find, the UKSA org charts are messy and it seems that there is no published list of leaders in the GSS so we can not comment on who might be there. As an aside, a close look at the UKSA org charts shows ambiguity and misinformation, and a comparison with the annual accounts and ONS “Leadership team” web page endorses that view. UKSA chooses not to publish its org chart in the same place as many other government departments and equivalent bodies. If it did, the information it published would need to be more accurate. (Blog updated 28 November to add info on leadership page.)

(Re)new National Statistician

I recently wrote about the mechanics of the hunt for the new National Statistician (NS) but there is speculation that the result is already all but settled. It seems that the incumbent, Sir Ian Diamond, will apply. This matters as such competitions ought to be a level playing field and fair for all candidates. Yet having made his intention clear it is hard for others to throw their hat into the ring and have a realistic hope of success. This will be the latest twist in the rise of Sir Ian Diamond who was appointed to the position out of competition. This leaves two further questions. First, if he wants to stay on, why did he not get another extension under the process that has him in post now? Second, if there was pressure to run a competition for the role, why did it not happen sooner? Whatever the outcome we should hope that the UKSA secretariat can get the office’s paperwork in order – it seems like the annual report and CSC submissions, both important public documents, tell an inconsistent story. It’s odd to see such sloppy number work in a statistics office.