Covid: there is still a war of numbers over lethality


The calculation of the infection fatality rate is proving to be one of the most controversial aspects of this pandemic. Among scientific fights, models, serological analyzes, the proposed figures are still very far apart, and waiting for certain data, all that remains is to decide who to trust

While the world struggles to come to terms with the limitations imposed by lockdown and from stages 2, the fight against the epidemic begins to bully into the political arena. This is the case in the United States, the country that is experiencing the most deaths in this pandemic, where opinions on COVID-19 they are becoming more and more divisive: on the one hand the right Trumpiana, which tends to reduce epidemic risks to avoid unpopular choices and to please the anarchoid tendencies of the republican electorate; on the other, the Democrats, tending to be more tolerant of the common good, and more sensitive to the dangers of a disease that America (much more than in countries like ours) ends up preferentially affecting the weaker sections of the population. The result is a polarization of opinions that begins to make itself felt also in the scientific field. A good example is the calculation of the lethality of Covid 19: the Education serological in progress around the world are in fact providing the first data on the real spread of the virus, with which we should soon be able to review the initial estimates of lethality, by force of things destined to overestimate (or underestimate) the danger of the disease. But pending reliable results, controversy is currently on the agenda.

Cfr and Ifr

The general concept is probably clear to everyone now. The lethality of a disease (in English Case Fatality Ratio, or Cfr) during an epidemic it is normally calculated as the ratio between dead is cases make certain. Obviously, it has several limits as an indicator, not least that of relying on partial data: the confirmed cases of disease can never reach 100% of the real ones. Especially in the case of a disease like Covid, which is asymptomatic in a large percentage of patients (exactly what is still under discussion) and in many cases still produces light symptoms, easy to confuse with other seasonal ailments such as flu or colds.

A more effective method of assessing the risk posed by the virus is the so-called Infection Fatality Rate (or Ifr), or the relationship between deaths is infections Total. A parameter that is necessarily lower than the previous one (or even higher, to tell the truth, if the dead turned out to be more than the known ones), and provides a concrete estimate of the number of people who they risk dying during an epidemic. But which at the same time represents an ideal index, as it is impossible to really know how many people have been infected in an epidemic, unless you have a 100% accurate diagnostic test and subject the entire population to analysis .

If at the beginning of a new epidemic the only available data is the Cfr, with the passing of the weeks, and the months, it is normal to revise the downward estimates, when you start to have some reliable data on which to base the estimates of the IFR . In our case, the main tool is the investigation of seroprevalence, like the one launched by the Ministry of Health these days: studying the spread of specific antibodies against Covid-19 in the population it is possible to extrapolate the percentage of people who should have contracted the disease, including cases not intercepted by the health system. And refine the accordingly calculation of lethality. Not all seroprevalence studies, however, are created equal: an account is an investigation, like the one we are carrying out these days, designed to recruit participants statistically representative of the entire population. Obviously, it is quite different if the tests are performed on a cohort not selected for statistical purposes, as has happened in most of the studies available to date, the results of which are therefore more difficult to extend to the entire population.

Ioannidis at the center of controversy

Among the scientists who look with greater interest to the results of seroprevalence studies there is certainly John Ioannidis, epidemiologist and biomedical data science expert from Stanford who in an interview some time ago had told us his doubts about the reliability of the data available on Covid-19: few, heterogeneous, unreliable, too little in short to be able to plan the next stages of this epidemic without running the risk of doing damage worse than that caused by the virus. The risk – Ioannidis wrote in an article published in March on Statnews – is of “end up like an elephant, which ends up falling into a ravine to escape a mouse“. It is clear that the Stanford professor believes that the measures taken in recent months have been excessive, and in the absence of reliable data with which to assess their risks and benefits, he has recently decided to take the matter head-on, obtaining them himself. With his research team he organized a study of seroprevalence centered in county Santa Clara, in California, to check the prevalence of infections, compare it with official diagnoses, and consequently calculate more accurately the Covid IFR in the area.

Covid in Santa Clara

The researchers have recruited 3,300 people on Facebook, performed the tests and adjusted the results to account for the demographic characteristics of the area, the sensitivity and specificity of the test used, and other potential confounding elements. Thus arriving to estimate a prevalence of 2.8% in the population of Santa Clara: 54 thousand people have already come into contact with the virus, against i 956 cases make certain at the time the study was conducted. At this point, the researchers move on to calculating theiFR of Covid in the area: i dead ascertained (always at the time the study was carried out) were 94, for a Infection Fatality Rate which just reaches it 0.17%. Extremely encouraging results therefore, which speak of a disease that is, after all, manageable, outside the great epidemic outbreaks such as that of New York, or like ours in Lombardy. Always obviously that they are reliable, something on which many, at least among American experts, seem to have different doubts.

Research continues

Ioannidis however he is convinced he is on the right track. After having published the results of the study on Santa Clara in preprint, he also carried out a sort of review of the researches of the same type carried out since the beginning of the epidemic, also published for now without peer review (and therefore to be taken with the pliers awaiting a formal publication in a scientific journal). Based on a series of parameters (minimum sample size, population studied, etc …) he selected 12 searches from all over the world, tried to statistically account for the differences in the representativeness of the samples analyzed, and then calculated the IFR at the time the studies were carried out. Also in this case, albeit with wide variability, the results are much lower than the Cfr calculated on official infections: it goes from one 0.02% registered to Kobe, in Japan, at 0.04% of a French studio, to reach a maximum of 0.5% emerged from a research carried out in Geneva.

For Ioannidis this would be excellent news: Covid’s average lethality would be in line with that of a bad one season flu (about 0.1-0.2%). And this while not changing much compared to the dangerousness of a pandemic that has already killed over 350 thousand people all over the world, it can be very useful in choosing the next moves with which to face it. Covid-19, second Ioannidis, kills a lot in certain setting (nursing homes, large epidemic outbreaks in which hospitals go haywire, areas with large populations of frail and elderly people), and much less in others, and it is therefore desirable to thoroughly study the variables that influence its lethality, to avoid to repeat total lockdowns as in previous months, and instead identify strategies aimed at protecting only the sections of the population at risk.

A shower of criticism

The results of the team Stanford, as we said, they have not been particularly welcomed by the American scientific community. And if the media related to Republican party they used them to diminish the danger posed by Covid and press for the suspension of the lockdowns, those of the democratic area instead demolished them, passing, in a short time, from accusations of little scientific rigor to those of real scam. Leaving aside the technical criticisms, a matter for specialists, it is worth mentioning the scoop (or presumed such) published by Buzzfeed May 15th. Citing confidential information obtained from a whistleblower, the article talks about dozens of emails exchanged between researchers and the patron of the JetBlue Airways, David Neeleman, regarding the results of the Santa Clara study.

A research that the tycoon would have financed with five thousand dollars, and then exploited to bring water to his mill, having all the interest in a rapid conclusion of the lockdowns given his commitment to civil aviation. Ioannidis and colleagues dismissed the allegations, recalling that Stanford has an office that collects donations which are then distributed anonymously to researchers at the university. And ensuring that he hasn’t been pressured by Neeleman on how to conduct research. Obviously it is difficult to know who is right. But what is certain, judging by the tenor of the controversy, is that the atmosphere in America is not the best. And that research aimed at better understanding the spread and lethality of Sars-Cov-2 at the moment is still an extremely divisive topic.

What do the other studies say?

Regarding Covid 19, in the past few months we have seen a bit of everything. The initial WHO estimates spoke of an (average) lethality equal to 3.4%. L’THEmperial College, whose reports have played a key role in convincing countries like the United Kingdom and the United States to implement the lockdown, initially estimated Covid’s Ifr around 0.9%, to then go down to the 0.6% (data also used in the models of our technical scientific committee). Enough to speculate hundreds of thousands of deaths in Italy in the absence of energetic containment measures. It is important to consider that in such a situation even a single decimal can radically change the scenario: making a very rough calculation on the Italian population, in the case of a carpet infection with a lethality of the 0.6% we would see 360 thousand dead, going down to the 0.4% the dead become 240 thousand, with the 0.2% we come to 120 thousand. Many, too many, obviously. But there is no denying that there is a big difference. In this case, however, we speak of theoretical calculations, carried out with epidemiological models.

One of the most thorough attempts to bring order to the myriad of field research carried out in recent months is to Gideon Meyerowitz-Katz is Lea Merone, two Australian epidemiologists affiliated to the University of Wollongong and James Cook University respectively. Their research, also in this case in preprint, is one systematic review of the scientific literature on the subject, which he analyzed 25 investigations serological made in recent months. Covid-19’s Infection Fatality Rate, calculated by making the weighted average among those described in the analyzed studies, was equal to 0.64%. However, the two authors point out that the results of the research analyzed are many heterogeneous, and that therefore the estimate made by them should be considered a hypothesis rather than a certainty. It is probable – they write – that the heterogeneity of the results is linked at least in part to the different response capacities of the countries to this disease, and also to the demographic characteristics of each nation.

Three months after the start of the pandemic, in short, there are still Covid more mysteries than certainties. If all goes as hoped we should soon have concrete data on the spread of the infections in our country, with which to calculate the lethality of the virus with precision. Going back to America for now, the CDCs seem to have taken the word Ioannidis, establishing a lethality of 0.4% for symptomatic patients, and at the same time underlining that there is a portion of patients who do not develop symptoms, evidently destined to further lower the Infection Fatality Rate of the disease. Whether it is a real scientific belief, or rather perhaps the result of pressure from the Trump administration, at the moment, unfortunately, it is impossible to establish.

It may also interest you

Source link


Please enter your comment!
Please enter your name here