There are two significant obstacles to a comprehensive understanding of the impact of Covid-19 on mortality in Ireland. One is the statistical challenge while the other is an unwillingness to pursue comparative study on a north/south basis, writes Mike Tomlinson, Emeritus Professor of Social Policy, Queen’s University Belfast.
As the first wave subsides, I suspect we have lost our appetite for numbers. There is something morally questionable about reducing so many lives cut short to an overall death toll or a death rate, especially when this comes across as an unseemly international league table. Exactly the same sentiment arises over north/south comparisons on the island of Ireland. Northern Ireland’s Minister of Health recently described such comparisons as ‘crude and often ghoulish’. But compare we must, as this is the way we learn from others.
Several months ago, the Irish Times published an opinion piece (22 April) in which I suggested that the Covid-19 death rate in Northern Ireland was considerably higher than in the Republic: ‘one island with two very different death rates’ ran the headline. Different policies, on the pace of lockdown and on testing and tracing, appeared to have resulted in different outcomes.
The response to this was not what I expected. It ranged from the First Minister’s special adviser referring to the piece publicly as ‘the now widely dismissed report’, to the London-based think-tank Policy Exchange devoting a 15-page pamphlet to the topic. More surprisingly, two days after publication, Facebook branded the Irish Times piece as ‘partly false’. This ensured that the article was relegated in the news feed, and distribution reduced. People sharing the content were referred to a report by a third party fact-checker which had rated the statistical comparison as ‘unsubstantiated’. The false news tag remained in place for a full week until Facebook reclassified the piece as ‘opinion’ on 1 May.
Reactions to north/south Covid-19 comparisons fall into two broad camps. First there are those who suggest that it is ‘impossible’ to compare Northern Ireland with the Republic as the published statistics count different things. Then there are those who say that north/south comparisons are ‘undesirable’, although typically this is implied rather than stated outright.
It is certainly difficult to compare the data on deaths. For Northern Ireland there are four figures to choose from. The lowest of the four is the daily reported number of Covid-19 deaths based on those tested positive for the disease and published by the Department of Health. Then there are three statistics published weekly by NISRA (the Northern Ireland Statistics and Research Agency). Two of these are a count of death certificates which mention Covid-19: by date registered and by date of occurrence. Finally, we have a figure for ‘excess deaths’, a measure of the spike in deaths from all causes over and above what would normally be expected at this time of year, based on an average for the preceding five years. This is the highest figure and the one that is regarded by most epidemiologists as the key statistic.
Three of these four counts are shown in Figure 1 (I’m indebted to Chris Giles at the Financial Times for the design of this graph). The top set of columns represent an estimate of the excess deaths occurring between the weekly published figure.
For the Republic the daily reported deaths include those tested positive for the disease and also ‘probable’ Covid-19 deaths referred to the Coroner (see Figure 2). No other data will be available until the mortality statistics for the first quarter of 2020 are published.
Where does this leave us for making comparisons? At the time of writing, the daily reported figures indicate a death rate for the Republic of 35.1 per 100,000 population and 28.9 for Northern Ireland. How is it that Northern Ireland, with later lockdowns and more restrictive testing, is recording a lower crude death rate? This is contrary to what one might expect from the international experience where countries following World Health Organisation guidelines and acting quickly, have recorded much lower Covid-19 death tolls.
The daily counts receive the most publicity and are the most often quoted. But they are not comparable for two reasons. First the Republic’s daily total includes some deaths (about 250) that have not been tested positive for Covid-19 (see Figure 2). Secondly Northern Ireland’s daily figure – the black line in Figure 1 – consists mostly of hospital deaths whereas these are only a minority (about 40 per cent) in the Republic’s daily total. This is a direct result of different testing policies. In the Republic, testing and contact tracing, whatever the difficulties, were expanded from early March. But from 12 March the British Government decided to stop contact tracing and to restrict testing to hospital patients.
Within a matter of days, fell behind the Republic in the confirmation of cases. By 21 March the Republic’s case rate was double Northern Ireland’s and the later expansion of testing in Northern Ireland did not change the overall picture: The Republic’s rate at 494 per 100,000 population (20 May) was more than double that in Northern Ireland (235). Put another way, towards the end of May about 1 in 50 had been tested in Northern Ireland compared to one in 20 in the Republic.
One way of approaching the comparison is to look at excess deaths. This is regarded as the gold standard for assessing the impact of an epidemic or other exceptional event on mortality. As Spiegelhalter (Cambridge Professor of Statistics) says, excess mortality “is currently a reasonable measure of the total impact of the epidemic, both the virus and the measures taken against it”.
What about excess deaths in the Republic? While weekly mortality statistics are not published, there is a proxy for excess deaths that several academics are now using to assess the impact of Covid-19. Seamus Coffey, University College Cork, has shown that there is a remarkably close relationship between the official count of registered deaths and death notices posted on RIP.ie going back over a number of years. Indeed, the Chief Medical Officer Tony Holohan has stated that the RIP.ie data captures over 98 per cent of deaths registered in the Republic. This means we can compare north/south excess deaths using the RIP.ie data.
There is a significant difference north/south as to how the excess deaths figures compare to the other measures. In the Republic the total of Covid-19 positive and ‘probable’ Covid-19 deaths now exceed the estimate of excess deaths from all causes (Figure 2), unlike the position in the Northern Ireland (Figure 1) where the NISRA registered deaths were 84 per cent of the excess by 19 June, while the Department of Health daily figure was only 56 per cent. But taking the excess deaths up to 29 May (when excess deaths ceased in the Republic) we find that the crude death rate per 100,000 population in Northern Ireland is 58 per cent above that in the Republic. Standardising for age, using the European standard population, brings the figure down to 32 per cent.
As the policy discussion turns to the relaxation and possible re-introduction of lockdown measures, there is a danger that efforts to suppress the virus will be undermined if the two jurisdictions do not follow similar policies. Two areas stand out. First, testing, tracing and isolating practices need to be aligned to prevent the infection circulating more freely in one part of the island relative to the other. Secondly, as Gabriel Scally and others have argued, health surveillance at points of entry to the island needs to be the same either side of the border in order to control the import of the virus.
Ideally, in future, the Republic will shorten the time allowed for a death to be registered from three months to five days and will publish mortality statistics weekly as in Northern Ireland and on a similar basis. The need to compare outcomes does not disappear with the first wave.