The purpose of this study was to evaluate the diagnostic performance of the symptomatology in detecting the SARS-CoV-2 infection. Recently, we reported the diagnostic accuracy of smell disorders ,detected by psychophysical tests and by the application of a self-perception questionnaire, in the Mexican population (sensitivity of 19.44% and specificity 95.52% [P=0.007]; sensitivity of 50% and specificity 80.59% [P<0.001], respectively). (13) However, so far there are no studies done within the Mexican population to evaluate the above-mentioned objective.
In Mexico, some social conditions have been observed that cause a greater severity of COVID-19, such as: belonging to an indigenous population, a low socioeconomic level or living in the southern states of the country .(14) The lack of universal access to ancillary tests in the medical units, to quickly identify the probable cases of SARS-CoV-2, has led to the adoption of algorithms based on clinical data to help guide the decision-making process.(7) Correctly identifying patients with a high suspicion of infection by SARS-CoV-2 is paramount to the emergency services, for the control of infectious outbreaks.(15,16) Some authors have developed predictive models for the diagnosis of the SARS-CoV-2 infection, to be used in settings where ancillary tests may not be available for first contact physicians. .(17–19) However, the results of the studies published as of today are inconclusive, partly because of their great heterogeneity.(10)
Peyrony et al. (20) carried out a prospective observational study in a French hospital involving 391 patients, out of whom 225 (57.66%) tested positive for SARS-CoV-2 by RT-PCR. In this group, 53 (23.6%) presented gastrointestinal symptoms such as vomiting, diarrhea, or abdominal pain; 147 (65.6%) had a temperature below 38°C and 97 (43.3%) below 37.5°C upon arrival at the emergency department. The symptomatic prevalence in the SARS-CoV-2 positive group was as follows: fever 176 (78.2%), cough 158 (70.2%), dyspnea 131 (58.2%), myalgia 71 (31.6%), rhinitis-pharyngitis 19 (8.4%), anosmia 31 (13.8%), headache 15 (6.7%), fatigue 34 (15.1%). Furthermore, the following symptoms were evaluated for their sensitivity and specificity: fever had a sensitivity of 78% and a specificity of 50%; dyspnea had a sensitivity of 32% and a specificity of 87%; anosmia had a sensitivity of 14% and a specificity of 98%; and oxygen saturation below 95% had a sensitivity of 17% and a specificity of 91% for the SARS-CoV-2 diagnosis.
Tostmann et al.,(18) studied 803 health-worker patients through a questionnaire to evaluate their symptoms associated to COVID-19; 112 patients were positive for SARS-CoV-2 infection, out of which 90 answered the instrument. The analytical model that included all variables (general non-respiratory, respiratory, and gastrointestinal symptoms) excluding fever and cough, reached an AUC of 0.75 (95% CI:0.66-0.84), with a sensitivity of 82.4% and a specificity of 59.2%. The second analytical model that included symptoms significantly associated with the SARS-CoV-2 infection (3 or more symptoms) such as anosmia, myalgia, asthenia, headache, eye pain, and malaise, yielded a sensitivity of 91.2% and specificity of 55.6% for SARS-CoV-2 positivity.
Salmon et al.,(21) performed a prospective, multicenter observational study at 5 hospitals in Paris to determine the frequency of SARS-CoV-2 positive patients with a loss of sense of smell; and to analyze the diagnostic accuracy of olfactory and gustatory dysfunction for the diagnosis of COVID-19. A total of 1,824 patients were included in the second phase of the study, out of whom 849 (46.5%) tested positive for SARS-CoV-2. The positive predictive value (PPV) of olfactory and gustatory dysfunction was 78.5% (95% CI: 76.6%-80.3%), with a sensitivity of 40.8% (95% CI: 38.5%-43.0%), a specificity of 90.3% (95% CI: 88.9%- 91.6%), and a negative predictive value (NPV) of 63.6% (95% CI: 61.4%-65.8%). Cough obtained a sensitivity of 70.4% (95% CI: 68.3%-72.5%), a specificity of 32.4% (95% CI: 30.2%-34.5%), a PPV of 47.5% (95% CI: 45.2%-49.8%) and a NPV of 65.2% (95% CI: 53.5%-58.0%).
In our series, we found a prevalence of SARS-CoV-2 infection of 53.72%. The most prevalent symptoms were: asthenia, headache and cough (63.36%, 60.22% and 60. 08% respectively), similar to those reported in other series(18,21,22). The symptomatology that was significantly associated to the SARS-CoV-2 infection was the presence of anosmia, with an OR of 3.23 ([95% CI: 2.52-4.17], P<0.001), fever OR of 2. 98 ([95% CI: 2.47-3.58], P<0.001), dyspnea OR of 2.9 ([95% CI: 2.39-3.51], P<0.001), cough OR of 2.73 ([95% CI: 2.27-3.28], P<0.001), SO2<93% OR of 2.73 ([95% CI: 1.89-4.00], P<0.001) and myalgia with an OR of 1.66 ([95% CI:1. 39-1.98], P<0.001); similar to what was reported by Lan et al. .(22) who found an OR of 6.5 (95% CI: 2.89-14.51) for anosmia, fever OR of 3.34 (95% CI: 2.07-5.41), myalgia OR of 2.41 (95% CI:1.50-3.89). Similarly, Tostmann et al (18) reported an OR for anosmia of 23 (95% CI: 8.2-64.8), fever OR of 2.7 (95% CI: 1.7-4.2) and myalgia OR of 6.9 (95% CI: 4.2-11.3). In a multivariate model where symptomatology was adjusted according to other predictive variables, an OR for fever of 1.96 ( 95% CI: 1. 58-2.41], P<0.001), cough OR of 1.95 ([95% CI: 1.58-2.41],P<0.001), anosmia OR of 2.96 ([95% CI: 2.27-3.87], P<0.001) and dyspnea OR of 1.48 ([95% CI: 1.17-1.87], P<0.001) were found.
Combining cough-fever and cough-anosmia resulted in an OR of 2.79 ([95% CI: 2.12-3.69], P<0.001) and 2.48 ([95% CI: 1.65-3.74, P<0.001) respectively; something similar to what was reported by Lan et al.(22)
In our study, a symptomatology model was created where 6 symptoms were combined obtaining for ≥ 2 symptoms, a sensitivity of 83.45% (95% CI: 81.17%-85.55%) and a specificity of 32.86% (95% CI: 29.94%-35.89%); and an association with the presence of SARS-CoV-2 with an OR of 2.46 ([95% CI: 2.00-3.04, P<0.001). Similarly, Tostmann et al. reported a sensitivity of 91.2% and a specificity of 55.6%, considering a cut-off point for 3 or more symptoms.(18)
The findings of our study suggest that the symptomatology (anosmia, fever, dyspnea and cough) by itself have a close relationship with the presence of the SARS-CoV-2 infection. Given the symptomatology complexity presented in this type of condition, the combination of symptoms, reported in different series, allows for a greater accuracy in the presumptive diagnosis of the SARS-CoV-2 infection (fever, cough, dyspnea, anosmia, SO2<93%). Finally, the combination of symptoms significantly associated to SARS-CoV-2 infection, integrated in a predictive model, will allow for a faster and more accurate final diagnosis, when limited ancillary resources are available.
From the physio-pathological viewpoint, the presence of dyspnea and hypoxemia has its explanation in the lung damage caused by the virus. The high expression of ACE2 in the apical lung cells,(23) promotes adhesion, penetration and destruction of lung tissue, causing a diffuse interstitial and alveolar inflammatory exudate production, as well as edema.(24) Regarding anosmia, some theories have emerged based on the findings of the neuro-invasion mechanism of SARS-CoV and MERS-CoV, due to the great genetic similarity that these viruses have with SARS-CoV-2 (89.1% similarity with SARS-CoV).(25,26) Stemming from the above-mentioned, three routes of SARS-CoV-2 invading the nervous system have been proposed: 1) the hematogenous route, 2) the direct route (through the cribriform plate via the olfactory neuro epithelium) and 3) a retrograde axonal transport to the central nervous system.(1,27,28) On the other hand, it has been hypothesized that the increase of bradykinins (secondary to SARS-CoV2 infection), specifically DABK, activates the BK1 receptors of the centers in charge of the sense of taste and smell located in the medulla oblongata, which results in the alteration of these senses.(29,30)
The strengths of this study include its contribution to the world´s information around the prevalence of the symptomatology in patients with COVID-19, studied in a significant amount of symptomatic patients, which allows for the findings and inferences to be relevant; its prospective nature improving its internal validity.
Some of our limitations were: the use of RT-PCR as a reference test, since its diagnostic performance has not been accurately determined (31) and its inherent technical limitations.(32) Some authors have reported a low sensitivity, such as Wang et al. who obtained a sensitivity of 60% for the detection of SARS-CoV-2 in nasopharyngeal exudate samples.(5) Another limitation was not including gastrointestinal symptoms and taste alterations, which have been reported in other studies.(20,33) More prospective studies regarding the symptomatology associated to COVID-19, that weigh in the symptomatology strategy in the diagnosis of the disease, are required; and finally, the non-inclusion of the 880 patients who did not undergo the SARS-CoV-2 RT-PCR.