Incidence Rate of Acute Respiratory Infections in General Medicine as a Tool to Correcting Official Data of Covid-19 in Places where the Tests of Polymerase Chain Reaction are not Accessible

Diagnostic Tests and Clinical or Epidemiological Utility

At present, a confirmed case of COVID-19 requires a positive PCR test for SARS-CoV-2, based on nasopharyngeal or throat swab. WHO is urging governments to conduct more tests for COVID-19 due to concerns about the failure to report cases in many countries around the world; As more and more nations have introduced stringent measures to try to delay the spread of the virus, the WHO cautions that evaluating the impact of these measures will only be possible with accurate data on the disease. It also warns that a lack of data on how many people have the disease could undermine containment and mitigation efforts in many countries [20].

But, from the clinical and epidemiological point of view, it must be taken into account that there may be false negative results of the PCR tests. PCR tests provide little diagnostic value immediately after exposure to SARS- CoV-2. The probability of false negative results is 100% on the day of exposure, falling to 38% when symptoms begin approximately 4 days later, and then to 20% 3 days after the onset of symptoms. Then it starts to increase again. Therefore, care should be taken in interpreting the PCR tests for SARS- CoV-2 infection, done early in the course of infection. If the clinical suspicion is high, the infection should not be ruled out based only on PCR, but the clinical and epidemiological situation should be carefully considered [34].

Furthermore, the importance of antibody testing lies in providing information to help track the spread of the virus at the community level, and assess the impact of public health efforts. Knowing how widespread the infection has been in a population could guide research and possibly public health decisions [35]. However, the reality is that there is a lack of diagnostic tests (PCR and antibody) or that in many places Health Authorities have a limited ability to test. So, in many countries such as Spain, or the CDC in the United States, the criteria to determine who is tested remain extremely strict, at least during March and April 2020: only people who had recently traveled or had contact with someone who had the virus, or people with a clinical picture of acute respiratory infection admitted to the hospital, or respiratory infection of any degree in health personnel; and similarly, routine diagnostic tests are not performed on contacts [36, 37, 38].

But, in addition there is the problem of test validity. There appears to be no known antibody test to date that has been validated for the diagnosis of SARS-CoV-2 infection. Tests measure IgM or IgG antibodies, but IgM antibodies may not develop at all, and IgG antibodies generally do not develop until later in the disease process. Therefore, the use of such tests to diagnose COVID-19 will bypass some infections [39]. No test is better than a bad test; a high-quality test should achieve 99% or more sensitivity and specificity. That means that testing should turn up only about 1 false positive and 1 false negative for every 100 true positive and true negative results. Without reliable tests, it can end up doing more harm than good [40].

Incidence and Prevalence are two ways of expressing the morbidity rate. Incidence is new cases in a certain time unit as numerator, and the population estimated at the center of the period as denominator; it is the healthy-sick transit within the specified period; 100,000 is generally used as the coefficient in open populations, but in other populations another coefficient may be used. The Incidence Rate can be useful to intuitively measure the weight of a disease outbreak [41].

Epidemiological data when all possible cases cannot be tested. Since the tests are not available to everyone, the morbidity rates do not accurately reflect the extent of transmission in communities around the world. For Spain, the official incidence figures (213,435 positive as of April 30, 2020) are therefore false; perhaps, one could speak of “minimal incidence” [36]. Very recently it has been reported that examining reliable data from influenza surveillance offers a simple approach to indirectly assess the overall effectiveness of population control strategies for COVID-19 infection [24]. In this sense, taking into account the great accessibility of patients to the GP, and the fact of the continuity of care that characterizes their work (which has important epidemiological connotations) [25], an indirect indicator of real COVID-19 cases (new cases of ARI with diagnosis based exclusively on symptoms), it will be even easier to use, as since the safe diagnosis of influenza requires a positive result in samples of the respiratory tract analyzed in the laboratory. Our proposal is to use clinical data, not serological, of ARI cases in the GP consultation, which is much easier and accessible.

Thus, new cases of ARI in GP practice (where COVID-19 cases without diagnostic tests are included) indicate the maximum incidence. A conservative estimate may be to consider that half of the ARI cases during the COVID-19 outbreak are actually COVID-19 cases (although this “maximum COVID-19 incidence does not include asymptomatic or presymptomatic or paucisymptomatic cases; therefore, considering the actual incidence rate of COVID-19 as half the incidence rate of ARI in a given period is probably too conservative).

The Incidence/Prevalence Rate of COVID-19 in the Community

Currently, the infection rate is unknown. Further, the percentage of infected individuals who seroconvert is still an open question [42]. Of course, it is suspected that the official data on those infected is low; We also know that some new coronavirus infections are asymptomatic. All of this means that confirmed case counts are low, but we don’t know exactly how low. Immunological studies could give us an idea of the true infection rates in the community. But only very recently have general population tests been carried out in different places to detect the virus. On the other hand, in some places, there are reservations about the precision of the tests: serological tests have the problem of their validity (false positives and false negatives); Furthermore, the studies must be well designed and have no sampling biases [43].

The first studies of the number of cases in the community begin to be published. These data, usually, show the number of participants who have had COVID-19 (the prevalence); although some refer only to the number of participants who were positive for COVID-19 at the time of swabbing (the positivity rate). Among the preliminary data known are the following: In a Boston suburb, a sample of people tested on the street showed that about a third of the 64 people tested were exposed to the virus [43]. As of April 20, the European Center for Disease Prevention and Control counted more than 300 cases of COVID-19 per million people worldwide, or about 0.03% of the world population. In the United States, nearly 2300 people per million have been infected, about 0.2% of the population nationwide, but the proportion of people who have had COVID-19 varies by state. Cumulative COVID-19 cases range from less than 45 infected people per 100,000 in Alaska (less than half a percent of the population), passing 0.8% in a screening of 6% of the Icelandic population, using PCR on nasopharyngeal and oropharyngeal samples, to nearly 1220 per 100,000 in New York (about 1.2% of the state population), according to US Centers for Disease Control and Prevention data. On the other hand, the WebMD survey WebMD poll of 6,343 readers to the homepage on desktop and mobile, April 20-21, 2020 suggested that around 26% of residents of the City of New York are positive for COVID-19, which show that infection is more widespread than current data suggests. But these data refer to confirmed cases with diagnostic tests. However, the confirmed case count depends on the availability -and the reliability- of diagnostic and antibody testing, which can’t be taken for granted [44, 45, 46].

According to the so-called Heinsberg study, the authors assume that up to 1.8 million people with the new coronavirus may be infected in Germany (an amount ten times higher than the 163,000 official records as of May 4). In Gangelt, Heinsberg district (North Rhine-Westphalia), a carnival event took place on February 15; the study of a random sample of Gangelt resulted in 15.5% (95% CI: 12.3% -19.0%) were infected; a total of 5 times more residents were infected than was officially reported. 22% of patients who tested positive in Heinsberg’s study were asymptomatic [47, 48]. 5% of the Spanish population has been infected with SARS-CoV-2, which represents around 2.3 million citizens (with “official” data of infected as of May 14, of 229,540 cases). However, the impact of the virus varies greatly according to the Spanish province, from Scoria where the percentage of people who have been infected is 14.2%. Madrid reaches

11.3%, Barcelona the prevalence reaches 7.1%, but in Huelva it is 1.5%, Murcia 1.4%, Tarragona 1.6% or La Corona 1.8%. Thus, it can be said regarding Spain, that the official data indicates that they have barely detected around 10% of the total number of actual cases in the community. On the other hand, this study in Spain indicates that 33% of the respondents positive in SARS-CoV-2 are asymptomatic.

The European Center for Disease Control has compiled the first seroprevalence studies published in Europe, and by now almost all are below 5%: 1.9% in Denmark, 3.4% in Helsinki, 3% in France 4.4%, in Oise (France), 3.2% in the Netherlands, 1% in Scotland. A study among the general population, carried out on patients from a health center in Barcelona (Spain), also indicated that 5% of those analyzed (311) through rapid tests had antibodies against the virus [49, 50]. The community prevalence of antibodies to SARS- CoV-2 was 4.65% in Los Angeles County in early April. Based on this prevalence, the authors estimate that approximately 367,000 county residents had antibodies to SARS-CoV-2. The official number of confirmed infections was only 8,430. However, the authors caution that there is likely to be selection bias [51]. In summary, most seroprevalence studies show that less than 5% of the population has antibodies.

What the ARI Incidence Rate Contributes to the Knowledge of the Epidemiology of COVID-19

Our calculations indicate a 3-5% incidence of COVID-19 infections in the population in March-April 2020. This coincides with some data from other places and with the mathematical models built from the case fatality rate of COVID-19 (the number of confirmed deaths due to COVID-19 divided by the total number of confirmed cases of infections with SARS-CoV-2). Other experts have stated that we must multiply by 27, at least, those detected, to approximate the actual number of infected [52, 53, 54, 55, 56].

In our study, using the conservative estimate that half of the ARI are cases of COVID-19 in the GP clinic during the study period, the incidence rate can be assessed as “real” or “reasonable”, which will reach 3,000 per 100,000 inhabitants. In other words, the incidence rate of cases with COVID-19 “official” PCR must be multiplied by 3 or 5 to obtain the real or reasonable rate of COVID-19. Thus, we estimate a number of “real” cases for Spain in the study period of between 540,000- 1,000,000 cases. This figure coincides with other authors. It has been estimated at 900,000 the number of cases with a clinical picture of COVID-19, with mild symptoms and without confirmation that GPs have attended in Spain, and that they have not been mostly referred to hospitals [57, 58]. Based on a survey, the Spanish Society of Family and Community Medicine estimated 65,149 positive cases for COVID-19 that was being monitored by health centers as March 27. Thus, it was calculated that each GP controls two people diagnosed with COVID-19 and 36 people who they suspect to have COVID-19. Other authors had calculated similar figures, which could be between 26 and approximately 40 hidden real cases in each Spanish general medicine consultation, as of April 9, 2020 (12, 36), which again coincides with our conservative estimates of 50% ARI cases: each GP can have between 20-30 cases of COVID-19.

Limitations and Strengths of the Study and the Proposal to Use the ARI Rates to Know the Incidence Rates of COVID-19 in the Community

Limitations

• There is an underestimation based only on symptomatic cases. The proportion of undetected, mild or asymptomatic cases, which could represent a substantial proportion of community transmission, remains largely uncertain. More than half of a population (residents of a skilled nursing facility) who tested positive is reported to be asymptomatic. These results not only indicate that asymptomatic individuals are playing an important role in the transmission of SARS-CoV-2, but that symptom- based detection alone was unable to detect a high proportion of infectious cases and were not sufficient to control the transmission in this context [59]. Therefore, the estimates of the incidence of COVID-19 based on the new cases of ARI in the GP consultation, that is, on the symptoms of ARI, do not include asymptomatic patients.
• In our study, the “raw” rates are shown, that is, those that refer to the population as a whole. Consequently, its interpretation, especially for comparison purposes, should be done with caution, since there is a danger of reaching erroneous conclusions. But presenting the rates more specifically, with very small population bases (the list of patients in the consultation; 2,000 people), would increase the random variation [40].
• Random variation. The study was carried out in a single GP office, with a demographic base small (2,000 people). In epidemiological analyzes of small populations, it is advisable to take the demographic base of about 10,000 people, that is, about 5 GPs practices [60, 61].
• The epidemics curves of different populations are not comparable and do not necessarily approximate the underlying transmission dynamics [24].
• The ecological and descriptive nature of our proposal for analysis between ARI cases and COVID-19 cases, assuming that it be used to assess the results of public health interventions, it must be taken into account that association or correlation does not indicate causality [24].

But, the study has some strength

• Prevalence studies can be expensive: they require good design, valid testing, and the infrastructure and logistics to perform them; in addition in seroprevalence studies, it should be borne in mind that in a population whose infection rate is 5%, a test that is 90 percent accurate could deliver a false positive nearly 70 percent of the time [62]. Instead, the detection of ARI by GPs is part of their daily work and does not need any additional infrastructure or cost, and may be more accurate.
• The convergence of results: There is a similarity of results with other population studies aimed at obtaining the frequency of COVID-19 in the population, which gives consistency to the method.
• It is required alternative options to evaluate the real incidence, and performance of infection control strategies, in the situation of lack of diagnostic tests, and the estimation of incidence rates of COVID-19, based on the ARI incidence rate based on clinical criteria, it is a very feasible and simple alternative; even more than using serological diagnoses of other diseases, as for example influenza as rate marker, as other authors suggest [24].
• Epidemiological plausibility: ARI (such as the flu) share similar clinical presentation (mainly as acute respiratory infection) and routes of transmission (mainly through droplets and fomites) with COVID-19.
• Validity: Although the fact of relying on a single GP practice has drawbacks, on the other hand, it has the advantage that since the same GP performs and records all ARI diagnoses, this increases its homogeneity and consistency.
• Probably the data is accurate: During a pandemic, patients may stop (or be told to stop) attending clinics for minor problems, and the amount of tests ordered for influenza surveillance may also be reduced. But, it is unlikely to happen in the GP consultation, in the conditions of the gateway to the public and universal health system, and within its role of continuity of care; and considering that consultations can be made by phone, without a face-to-face visit, and that it is very likely that all respiratory symptoms will be consulted during the COVID-19 pandemic, given that it is widely reported that it produces respiratory symptoms and patients need to, among other things, sick leave that your GP must authorize [63].
• The underestimation based on symptomatic cases can be corrected by taking higher proportions of IRA as an indicator of the COVID-19 model. In our study we present a conservative estimate of 50% of ARI as cases of COVID-19, but other figures could be used, such as, for example, the interval between 50-75% of ARI cases as an indicator of cases of COVID-19, for correct the possible underestimation for asymptomatic cases of COVID-19 (which obviously have not consulted the GP as IRA).

Finally, it is suggested to reproduce this approach to the knowledge of the epidemiology of COVID-19 from large computerized and validated databases of GP practices to validate or refute the proposal that emerges from our study.