The Covid-19 outbreak is providing us with a stark reminder of the devastation that can result from the sudden and unexpected appearance of a new virus on an unprepared planet. As the global death toll from this new coronavirus continues to rise, attention is now focusing on the differences in death rates as reported from different countries, sometimes even in close geographical proximity to each other. Widely publicised league tables are giving us daily updates of the latest numbers of people tested for, those testing positive, and those dying as a result of, Covid-19 infection. This information is then being used by scientists and politicians to contrast and compare the relative success of individual countries in dealing with the pandemic, as well to decide the best time to ease up on lockdown measures implemented to control disease transmission.
Unfortunately, the measures of death that are being compared are either inaccurate or incomplete, or even both. Although there can be no uncertainty about death itself, the problem arises when there are differences in the ways in which information about deaths is collected, attributed to Covid-19 infection, and then analysed. It may so happen that people who have not been tested, but still suspected of harbouring Covid-19, are either included in, or excluded from, the numbers of Covid-19 cases. The numbers of deaths attributed to Covid-19 may thus depend on whether only confirmed, or both confirmed and suspected, cases are being counted.
The death rate, or, more accurately, the case fatality rate, refers to the number of deaths among the total numbers of confirmed cases-a number that solely depends on how widely testing has been carried out. If only people with symptoms are tested, a much larger group of asymptomatic Covid-19-positive people are likely to be left out. This exclusion inevitably leads to an overestimate of the actual death rate. Reported death rates are thus not readily comparable between countries, as testing strategies may differ widely, depending on both actual public health policy and the availability of diagnostic tests. An alternative might be to count Covid-19 deaths in every 100,000 or every million of the population of the country. But this then assumes that the cases are uniformly distributed within the population, while in most countries Covid-19 outbreaks mainly consist of several clusters, the so-called “hot spots”.
In this context, it is worth looking at the data coming out of the UK. The first Covid-19 death in the UK was reported on 5 March 2020, at the Royal Berkshire Hospital in Reading. That number had reached 28, 131 by 2 May 2020. Initial death counts in the UK were deliberately incomplete. This happened because only deaths in hospital of those already tested positive for Covid-19, so-called “confirmed cases”, were being recorded. It soon became apparent that many Covid-19 deaths in the community were unaccounted for. Due to a lack of community testing, people dying in private homes, in care homes and in hospices were falling under the radar. More people were, however, dying than would be normally expected for the time of year, when compared with a baseline seasonally-adjusted average. This increase in numbers of those dying was a sign that Covid-19 was indeed contributing to an unexpected and unseasonal spike in deaths.
In England and Wales, deaths outside the hospital require a licensed medical practitioner, usually a GP, to issue a medical cause of death certificate. Covid-19 can be included in the information that is sent to the Registrar of Births, Deaths and Marriages, based on the signing doctor’s suspicions. This information is then processed by the Office for National Statistics (ONS) to yield national death figures. The process of certification and analysis can cause a delay, or lag, of up to ten days from the time of death for it to appear in ONS returns. Once the government decided to include the ONS data, the death counts from Covid-19 rose sharply, giving a more accurate picture of what was going on. It is of course possible that some of the additional Covid-19 deaths may have been caused by coexisting or coincidental medical illnesses, with the coronavirus merely an innocent bystander.
In England and Wales, Covid-19 deaths have disproportionately affected older people, those from socially deprived areas, and, for reasons that are yet unclear, people from some black, Asian and minority ethnic (BAME) communities. Similar trends have been reported from America. The insights from this information may contribute to future social and public health policy, to the benefit of those groups of people bearing the brunt of this pandemic.
Comparisons between the death rates in different countries have led to the identification of a so-called ‘German paradox’ (a lower than expected number of deaths given the number of confirmed cases). Significant differences in death rates have been noted within the Nordic countries, where many more people have died in Sweden than in either Denmark or Norway. These differences are being explained in terms of differences in the intensity of lockdown and other public health measures adopted by the respective countries, although this is yet to be confirmed. At a more fundamental level, there may also be concerns about the validity of some reported data, as for example in the suggestion that there has been an under-reporting of both cases and deaths within China.
It seems tempting, although somewhat premature, to read too much into reported differences in death rates between different countries. This is not the time to come to sweeping conclusions. All that can be said with some certainty is that the absolute numbers of deaths are less useful than trends in mortality. These trends can help identify when the outbreak has peaked and reached a plateau prior to a gradual fall in the number of new cases, and of deaths. There seems little doubt that, in due course, national differences in outcomes will provide valuable lessons in future pandemic planning. In the meantime, the priority is to contain the spread of Covid-19, using measures that are tailored to national patterns of spread and available resources.
Ashis Banerjee