Behind the numbers: COVID-19 in selected locations on 21 Apr 2020.

RCMI Members News June 2020

When President Mike Hoare asked me to write a piece for this issue on the COVID-19 pandemic, I objected saying that I was not an expert. On reflection, I realize that we have all become experts on the pandemic, not only from living through it, but also from the information constantly being given to us. I offered to write about my struggles trying to understand the numbers that are presented daily, in ever enlarging red dots. It is no surprise to conclude that these numbers mislead us in many ways. This article does aim not reduce our fascination with the numbers but it might teach us simple ways to glean the lessons behind them.

The daily news usually lists the total number of confirmed COVID-19 cases by country or region in a descending order. While Italy dominated for a long time, Spain crept past, and now the United States is firmly in the lead. China, where this all started, is rarely mentioned. Total cases are prefaced with the daily new cases. In medical parlance, the total number of active cases is the prevalence, whereas the new cases are called the daily incidence. We have many concerns about the accuracy of these numbers. The incidence depends heavily on the rate of testing, the number of test false-negatives and the number of asymptomatic infected individuals within the population. While there are wide variations in these factors globally, we can be sure that the incidence reported anywhere is an underestimate. The prevalence is the cumulative daily incidence minus the patients who have either eliminated the virus or have died. Prevalence should be the most interesting number for us, because as it tells us how many cases are active in a community or region. When we talk about the peak, it is the prevalence over time that we should look at. The number of recovered patients is unknown and probably underestimated. The graphs, that are presented to us as total cases, appear to be cumulative incidence without removals. The so-called peak is represented as that point where the upward trend plateaus on these graphs.

We have been told that the majority of people who contract COVID-19 suffer little more than the common cold, which is caused by other coronaviruses. Indeed, population screening of the 4000 souls aboard USS Theodore Roosevelt found many asymptomatic sailors among the 660 crew who tested positive. Captain Crozier, who had been dismissed for raising the alarm, is also reported to be positive. An unknown number of sailors are sick and one, aged 41, has died. More is at stake if a death occurs. We should expect mortality reportage to be more accurate than incidence in our reports. Crude mortality numbers are the next most frequently reported statistic on our news, given as daily and cumulative numbers. However there are concerns that some jurisdictions only report confirmed COVID-19 positive patients who die in hospital, missing an unknown number of people who died of the disease in a nursing home, or in their own home, or without a positive test. Reported deaths are, therefore, also underestimated.

A red circle, scaled to the size of the crude number reported, is the commonest graphic that is currently used. The intention of showing dots on a map is to tell us where the virus hot spots are in the world. However, the impression it leaves with us, especially when the dot is coloured red and scaled, is that the disease itself is worse in the areas where the dot is large. Indeed it is not possible to show the red dot over Detroit without covering a large area of south west Ontario, nor is it possible to see the red dot in Ireland because it is covered by that belonging to the UK.

Despite these concerns, these are the best numbers that we have. I have outlined them in the table below for 21 Apr 2020. The countries are listed in descending order, according to the total number of confirmed COVID-19 cases. As reported in the news, the US leads followed by Spain, Italy, Germany and the UK. However when you look at the total number of cases per million population a different order emerges: Spain, Ireland, Italy before the US. Canada, with less than half the US rate of positive cases, is well down the list. Ultra low outliers include China and South Korea which have only a small fraction of the rate of positive cases reported elsewhere. China's low rate, often a source of complaints of manipulated reportage, is similar to that in South Korea, which is frequently lauded for its response. At the other end of the scale, New York city is off the charts with by far the highest rate of confirmed COVID-19 cases in the world. The rate of positive cases in New York city is 15 times that of Canada.

Behind the numbers: COVID 19 in selected locations on 21 Apr 2020.

Location (population in millions)	Total cases	Total deaths	Total Cases PMP	Total Deaths PMP	Case fatality rate %	Test rate PMP	Test positivity rate %	2019 Annual mortality (monthly) PMP
USA (327.2)	788920	42458	2411	130	5.4	12167	19.8	8782 (732)
Spain (46.7)	204178	21282	4372	456	10.4	19896	21.9	9115 (760)
Italy (60.5)	181228	24114	2996	399	13.3	23122	13	10566 (880)
Germany (82.8)	147593	4869	1783	59	3.3	20629	8.6	11275 (940)
UK (66.4)	125856	16509	1895	249*	13.1	7386	25.6	9398 (783)
China (1404)	83853	4636	60	3	5.5			7261 (605)
Canada (37.6)	37933	1753	1009	47	4.6	14826	6.8	7758 (647)
Republic of Ireland (4.8)	15652	687	3260	143	4.4	18358	17.8	6139 (512)
South Korea (51.3)	10683	237	208	5	2.4	9766	1	6111 (509)
Northern Ireland (1.8)	2486	193	1381	107*	7.8
Michigan (10)		2245		225
New York City (8.6)	134874	14604	15683	1698	10.8
Hubei (58.9)		4512		77
Quebec (8.5)	20126	939	2367	110	4.7
Ontario (14.6)	11735	647	803	44	5.5

Notes: data collected on 21 Apr 2020, from the COVID-19 Dashboard provided by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University and from government websites; PMP stands for per million population; * only deaths-in-hospital reported.

Testing rates may explain some, but not all, of the differences seen in total number of cases. Ireland's high positive rate per million, compared to the UK, reflects its significantly higher testing rate (x 2.5). In contrast, Canada has significantly less positive cases than the US despite a slightly higher rate of testing. The percentage of tests that are positive is considered a better measure of severity of an outbreak. Despite a relatively low test rate per million population, South Korea continues to have a very low test positivity rate of 1.2%. Australia and New Zealand have similarly low test positivity rates. These rates probably reflect the relative isolation of these countries. Canada's test positivity rate, at 6.8%, is considered low worldwide, especially among countries with high test rates. Badly affected countries have test positivity rates around 20% as exemplified by the UK, Spain, US and Ireland.

The top 4 in total deaths in our selection include, in order, the US, Italy, Spain and UK. The ratio of total deaths to total cases, known as the case fatality rate or CFR, is a measure of the severity of the disease combined with the failure of treatment. Badly affected countries have a CFR from 10 – 13%, as seen in Italy, UK, and Spain. CFR in the next cohort of countries is approximately 5% including the China, US, Canada (4.6%) and Ireland. Germany (3.3%) and South Korea (2.4%) have even lower CFR. When deaths are counted per million population the order changes again to Spain, Italy, UK, Ireland, and US. Canada has seen only one third the number, corrected for population, of deaths compared to the US. In Europe, Belgium has the highest mortality with 510 COVID-19 deaths per million population. Spain, Italy and the UK come next. There is concern that the UK is not including patients who die outside of the hospital and may be even worse off. The low outlier in Europe is Germany with a mortality rate one tenth of that seen in Belgium.

Population corrected data favours countries such as China and India with huge populations. COVID-19 originated in Wuhan, Hubei Province. Data from Hubei are more in line with the rest of the world and suggest that China was successful in preventing spread of the virus to the rest of the country. Likewise in the US, not all of the country is affected equally. Most of the US has weathered COVID-19 similarly to Canada. While the numbers from New York City are frightening, densely populated California has fared much better.

Comparisons require us to pair like with like. Some neighbours provide lessons. It is believed that a single person introduced COVID-19 to the Republic of Ireland and Northern Ireland on the same day. Mike Tomlinson, a professor at Queen's University Belfast, compared the jurisdictions where the Republic of Ireland adopted stricter lockdown measures earlier than the UK and found a higher death-in-hospital rate in Northern Ireland. Despite that, Northern Ireland has fared better than the rest of the UK, possibly indicating an influence of the rest of the island, with whom they now plan to cooperate regarding COVID-19. There are other hotspots in the US such as Detroit. Comparing Michigan's mortality rate (225 PMP) to its similar sized neighbour Ontario (44 PMP) is sobering. Quebec, with 101 COVID-19 deaths per million population, is the most affected Province in Canada.

Finally an interesting, if somewhat heartless, perspective is given by the annual all-cause mortality. In Canada, the COVID-19 death rate is currently 47 PMP and this represents just 0.6% of the country's expected annual all-cause mortality. In Spain, where the peak has passed, COVID-19 mortality currently represents 5% of the expected annual all-cause mortality. It may have been these estimates that led some government advisors to aim for herd immunity. In addition, they felt that most of the COVID-19 mortality would occur in people who would die from other causes, in the not too distant future; the blip in mortality caused by COVID-19 would be followed by a dip and things would even out. What they did not contend with was the impact on society of watching COVID-19 devastate nursing homes and healthcare systems.

Crude total case and death rates provide a misleading picture that can have a depressing effect on a captive audience. Very simple calculations can place these data in contexts that we understand. Canada's precautions have protected us. COVID-19 incidence increases exponentially as its prevalence increases. There is little doubt that it could infect up to 80% of a community if left unchecked. Countries that isolated themselves (or were isolated naturally) and applied social measures early, in order to reduce the prevalence of COVID-19, have had a distinctly different experience, up to now, compared to those that did not. We do not know the mortality of COVID-19 in an untreated population. We do know that CFR increases from 2% to 13% in countries with modern healthcare systems that become overwhelmed. If we put the untreated CFR at 20%, Canada was at risk of 30 million cases and 5 million deaths. If Canada had applied protective measures less successfully and we suffered New York city's fate, half a million cases and 100,000 deaths would have been the outcome by now, with more to come. The toll on healthcare workers would have wrecked our healthcare system for a generation.

Finally, what numbers can we apply to safely lift the restrictions of social lockdown? Mortality is probably misleading because it is at least 2 weeks delayed from disease onset and might be declining despite arise in incidence. Total cases reflect the past and prevalence is not available. An accurate incidence would be the best indicator but is also elusive. If we maintain a high test rate and see the test positivity rate come down, we may feel confident enough to relax. Targets are probably for a reduction less than 2% test positivity before a general lifting of restrictions. We may also look within our community and maintain security where test positivity remains high (for example, in nursing homes). An early downward trend in test positivity may permit resumption of low risk activities such as those outdoors among close family (for example park use, camping and cottage life) and of high value services (such as non-COVID hospital care). But life may never be the same. Large gatherings in areas with poor air circulation may not be popular for a generation. We may learn to carry a cloth mask and whip it out at the first sign of sniffles. Who knows, we might beat influenza and the common cold too!