We need a slew of public health measures to contain Covid-19 and keep the total deaths in our population low even as the count of cases rises
By K Srinath Reddy & Surabhi Pandey
Here’s a quiz. Which country has the highest number of Covid-19 deaths currently? No prize for answering: the US. Which of the three North American countries is faring best in death rates? Here, the prize will go to the correct answer on which statistic the quizmaster is using. If we use the Case Fatality Rate (CFR), which many reports publish, the US (5.96) seems to be doing better than Canada (7.45) and Mexico (10.51). Before NAFTA’s Covid-19 mortality prize is awarded to the US, it will be useful to look at another mortality indicator—of the recorded Covid-19 deaths per million population (DPM). Here, Mexico (35) fares far better than Canada (145) and the US (263), as reported till May 15, 2020. In terms of actual Covid-19 deaths, that is the order that represents the reality of death burdens. If you have some doubts, look at Nicaragua, with a horrific CFR of 32 and a very reassuring DPM of only 1 at present.
To make the differences in these two mortality indicators, we examined global data of 53 countries which have a CFR higher than the US (see table). Some of them have higher population level mortality rates than the US, while several have much lower values. Why this discordance and why are two measures needed anyway?
CFR measures the deaths among those diagnosed to have the disease, by whatever clinical or lab test criteria used to define a case. It estimates the severity of the disease in terms of its lethal effects and varies between diseases. For example, Ebola and MERS have very high CFRs compared to Covid-19 or common flu. CFR also varies, at the same time, between different population groups. For example, the elderly and persons with co-morbidities have higher CFR than the young. Differences in CFR have been noted between men and women in Europe and between different racial groups in the US. Different treatments and competencies of healthcare systems, too, can yield different CFRs for the same disease.
DPM measures the extent to which the disease has caused deaths across the whole population, at any given time. It is a function of the spread of the infection, the severity of illness, and the comprehensiveness and competence of the response which combines both containment and care. A disease with a high CFR can still be limited to a low DPM if it can be quickly contained and localised. A disease with a low CFR can cumulatively kill many and have a high DPM, if it spreads fast and far—as is true of Covid-19, when compared to Ebola, MERS or SARS.
Both CFR and DPM are dependent on the same numerator (recorded and ascribed deaths). Problems of undercounted or misclassified deaths are common to both. Verbal autopsy techniques can fill in the gap related to diagnosis of out-of-hospital deaths. The denominator, however, differs between the two indicators. In CFR, the denominator can be large or small depending on testing rates, but the case mix of diagnosed cases can also vary according to testing criteria. Large-scale testing expands the denominator and often adds more mild cases, as compared to low levels of testing, which yield a small denominator often comprising the more severely ill cases who have been prioritised for testing. Even if the case mix is similar, increased case detection expands the denominator more than the numerator, since only a small fraction of the additionally detected cases would die. Though the numerator is fixed, large-scale testing yields a lower CFR than limited testing which yields a high CFR, for these reasons. Many Covid-19 infected persons may remain asymptomatic and hence untested, giving lower than accurate denominator and hence higher than accurate CFR. As testing is restricted in the initial stage of the new epidemic, we get a high number for CFR but as testing ramps up to higher levels, CFR has a lower number. Comparing countries on CFR, without taking variations in testing rates and testing criteria, can be very misleading.
The denominator for DPM is fixed and is not variable. It is the population of the whole country in millions. So, unlike CFR, which has variability in both numerator and denominator, DPM is a more stable measure of Covid-19 deaths as experienced across the whole population (exposed and unexposed). DPM will increase over time, as the epidemic advances, as more persons get exposed and some of them die. However, it is a true measure of disease-related mortality in the whole population. It is this figure we need to track and contain, as we move further on.
CFR and DPM have great discordance when the testing rates are low, but even when testing rates are high they do not relate well. Do we need to test much more just to push CFR down? That is not rational, when CFR is not correlated to absolute mortality in the population. Testing criteria, especially in resource constrained health systems, have to optimally use the testing kits available without being singularly focused on testing numbers. It is pertinent to cite a recent statement from the Covid-19 Group at the Imperial College, UK: “Testing is essential for pandemic surveillance but its direct contribution to the prevention of transmission is likely to be limited to patients, HCWs (healthcare workers) and other high-risk groups.” Testing has to be guided and aided by symptom-based syndromic surveillance. Containment also calls for other measures like physical distancing, masks and hand hygiene.
We need a slew of public health measures to contain Covid-19 and keep the total deaths in our population low even as the count of cases rises. A DPM of 2 indicates we are doing well so far. Can the rate of rise be kept slow even after the lockdown ends? We need as much agility in our public health interventions as the virus has exhibited in its spread, and as much ability in our clinical care systems as the virus has displayed in striking at multiple locations in the body. Perhaps even more.