There has been a steady flow of daily statisics for Covid-19 cases in the UK, however, these are most probably understatements of true numbers. John Houston, University of Stirling, provides an in-depth analysis of Covid-19 cases and 'excess' death rates to allow comparisons between Scotland, England and Wales.
While there has been a steady flow of daily statistics of Covid-19 cases, all these are most probably understatements of the true numbers. This blog compares both the magnitude and trajectory of the number of ‘excess deaths’ that have occurred in Scotland with those in England and Wales during the course (to date) of the Covid-19 Pandemic, as well as how those differences appear to depend upon the extent to which account is taken of their different ‘normal’ mortalities.
Outline Summary of the First Wave of the Covid-19 Pandemic in Great Britain (GB)
It will be a matter of future study to determine the importance of the interval between the first manifestation of a virus like Covid-19, but it already widely speculated that Great Britain, in particular England, wasted precious time in moving to lockdown only 18 days after the first death there. The extent to which this delay may explain its subsequent high case- and death rates may be known in the fullness of time. Table One summaries the key dates that led to lockdowns in the three GB Nations.
It took England 74 days to reach its peak (14th April), 19 days more than it took Scotland (21st April) and 11 more than Wales (30th April). Figure One charts the trajectories of the official Covid-19 daily death statistics expressed as a percentage of each nation’s first-wave own peak against the number of days elapsed since each one’s first official case.
Figure One charts the trajectories of the official Covid-19 daily death statistics expressed as a percentage of each nation’s first-wave own peak against the number of days elapsed since each one’s first official case. Other than England’s more leisurely climb towards its peak, some differences can be seen in the three trajectories. Scotland had a cleaner ascent than Wales, but then descended more slowly though persistently. At the time of writing, Scotland’s daily death count has continued to fall, while Wales’ has risen from the end of June. England’s descent from its peak was similar to Scotland’s, though an upturn in its figures around days 125-130 is evident. This will likely be followed by a resumption on a downward track. Scotland has brought down its daily death count the most in Great Britain (at the time of writing).
Figure Two charts the trajectories of each nation’s Covid-19 ‘death rate’, defined here as the cumulative number of official Covid-19 deaths divided by the cumulative number of official Covid-19 cases. The rapid rise in the rates from early March to mid-April can be seen, with England’s rate being well above Wales’ and Scotland’s, and Scotland’s below Wales’. England’s rate began to fall from early May, while Wales’ rate and Scotland’s continued to rise, albeit at a slower rate until around 15th June.
Excess Deaths
There are issues surrounding the official statistics for cases and deaths. Even where we might have some confidence in Great Britains’s ability to collect and publish accurate statistics, it is apparent that re-definitions of what counts as a ‘Covid-19 case’, or a ‘death’ are being argued for and, on occasion, being implemented. There is also the ever-present reality of dependence upon medics deciding whether or not to certify a death as Covid-19 related. Rather than be subject to these vagaries, an alternative way of counting deaths is to estimate the unexpected, or excess deaths that occur in a given week and to ascribe them to the Pandemic. These excess deaths are simply the difference between all deaths recorded in that week and those that were expected. Assuming that the system to record all deaths timeously is working properly, the focus is on deriving the number of deaths that were expected in the corresponding week. Figure Three charts the total monthly deaths recorded in Scotland between January 2000 and December 2019.
There are various ways to establish weekly ‘expected’ or ‘baseline’ deaths, varying in statistical complexity and data granularity. A simple method is to consider the percentage of all deaths that occurred in each week over a number of years and to apply the average percentages to a feasible forecast of total deaths for 2020. The Scottish and English/Welsh weekly deaths data for 2015 to 2019 were used to compute the average percentages for Scotland and England/Wales and are shown in Figure Four.
There are some differences in the distributions between Scotland and England/Wales: in particular, Scotland does not enjoy the same degree of ‘summer premium’, due perhaps, to its cooler, shorter summers. It also has harsher winters, and thus proportionately more deaths in January and in late-December. Conversely, it fares better then England and Wales through February and the Spring months. These inter-annual differences are sufficiently marked to justify applying separate percentages for Scotland and England/Wales in establishing each one’s baselines.
Table Two records the actual and estimated/predicted total recorded deaths (all categories) and mid-year populations for Scotland and England/Wales from 2015 to 2020.
The baseline deaths are computed from the percentages shown in Figure Four applied to the 2020 predictions in the last row of Table Two. These baselines are subtracted from the actual weekly registered deaths to give the excess in those weeks.
As Figure Five indicates, in those weeks when the first wave of the Pandemic was at its height in Great Britain (14-20), the numbers of excess deaths were actually greater than the number of official Covid-19 deaths. If we believe that all these excess deaths were actually Covid-19-related (though not certified as such), it appears that the numbers dying with it were considerably greater than the official statistics showed. This is particularly the case in Scotland, where the Excess deaths exceeded the official Covid-19 counts in weeks 14 to 17 and continued to be significant in weeks 18 to 20. In England and Wales, they only exceeded the official Covid-19 count in week 17 but were otherwise significant through to week 20.
It is also interesting to note that following the peak period (week 21 onwards), the numbers of excess deaths in both Scotland and England/Wales were actually negative. Bearing in mind that the excess deaths are based on the expected average of the last five years, it could be that all of GB had actually benefitted from a ‘good‘ health season that would have happened anyway, and/or it could be that a side-effect of the Lockdown has been to shield people from other fatal illnesses, and also slow the transmission of other viruses via Social Distancing, increased handwashing, etc.
Figure Five, of course, is comparing two very differently-sized territories: one with a population of 5.5 million, the other with a combined total of almost 60 million. Some differences between Scotland and England/Wales can be seen in week 1 when Scotland’s traditionally worst week was better than expected, but England/Wales’ was worse (Figure 4). Thereafter, it appears that all of Great Britain had a relatively mild Winter and early Spring. It might be concluded, by some, that the Pandemic had relatively less impact in Scotland than the rest of Great Britain, perhaps as a consequence of the shorter elapsed time between first death and lockdown (Table One), and/or a better resourced/managed healthcare system.
Some differences between Scotland and England/Wales can be seen in week 1 when Scotland’s traditionally worst week was better than expected, but England/Wales’ was worse (Figure 4). Thereafter, it appears that all of Great Britain had a relatively mild Winter and early Spring. It might be concluded, by some, that the Pandemic had relatively less impact in Scotland than the rest of Great Britain, perhaps as a consequence of the shorter elapsed time between first death and lockdown (Table One), and/or a better resourced/managed healthcare system.
However, as Table Two showed, the death rate in Scotland is typically 0.2% higher than in England and Wales. If the separate deaths rates were equalized to the British average (0.91%), then Scotland’s 2020 death count would be around 9,000 lower, England/Wales’ 9,000 higher. Scotland’s lower expected deaths means that its excess deaths would be higher, if it had had the GB’s death rate, while England/Wales’ excess deaths would have been less. If the Great British rate were applied to both, and the resulting 50,000 (Scotland) and 542,000 (England and Wales) deaths applied in the same percentages as before, the resulting excess deaths (as a percentage of the revised expectations) are as depicted in Figure Seven.
The effect on Scotland in the weeks before the peak is to turn the negative excess deaths into positive ones, of greater relative magnitude than those in England and Wales. In the peak weeks, 14 and 15, Scotland’s excess deaths would have been significantly more significant than England and Wales, peaking close to the 110% seen in Figure Six for the latter. This was maintained in Scotland in week 16 to equal England and Wales. Thereafter, the trajectories are generally downwards across Great Britian, but Scotland is always above England and Wales after week 17: exactly the opposite of the story in Figure Six. Of course, Scotland does have the higher annual death rate used to derive Figure Six, but the point here is that its fundamentally poor(er) health record works to create the impression that its Covid-19 performance was actually superior to England and Wales, when the opposite might actually be true.
Conclusions
Computing excess deaths is used widely to estimate better the true progress of fatalities: in the case of Great Britain, the number of excess deaths actually exceeded the number of official Covid-19 deaths implying that official death rates will have been significantly underestimated. Scotland’s apparently better performance than England’s and Wales’ may be partly illusory, given its historically poorer health performance, and should be accounted for in making such inter-nation comparisons.
Photo by Bill Mackie on Unsplash
