gms | German Medical Science

Introduction: SARS-CoV-2 [1] still has continuous impact on everyday life and healthcare systems. Hospital workforce is in the forefront of treating infected persons and takes high risks to be affected themselves. We used the dataset of the published retrospective cohort study “SLKovid” [2] to investigate a possible correlation between clusters of self-reported symptoms and the presence of a SARS-CoV-2 infection. Such correlations could be used to predict the existence of the infection with a currently unknown level of certainty. Furthermore, this information could be useful to prioritize test capacities.

Methods: SLKovid investigated the seroprevalence of antibodies against SARS-CoV-2 among the SLK hospital staff. In total, 3,067 employees took part. Most prevalent age group was 50-59 (28.1%,862/3,067), 66.5% (2,041/3,067) had close contact to patients, 80.0% (2454/3,067) were female, 19.3% (593/3,067) male.

The 14-question questionnaire collected i.a. information about working conditions, self-reported PCR test result and self-reported symptoms from January till July 2020. The list of symptoms consisted of: cough, fever ≥38˚C, congested/running nose, sore throat/hoarseness, shortness of breath, respiratory distress, muscle pain, joint pain, headache, malaise/weakness, diarrhea, nausea/vomiting, stomach pain, loss of appetite, loss of weight, impairment of taste/smell, lymph node swelling, conjunctivitis, skin rash, abnormal sleepiness/drowsiness, apathy, “other symptoms”, and “no symptoms since 01-01-2020”.

In total, 3,067 questionnaires and laboratory data were used for a clustering analysis with the R package klaR (v0.6-15) [3].

Whether a participant was defined SARS-CoV-2 positive was based on the PCR result and performed serology tests. For each symptom-based cluster the positive predictive value (PPV) and negative predictive value (NPV) as well as the 95% CI - in squared brackets - were calculated.

Results: 3.5% of participants were found SARS-CoV-2 positive. We identified nine clusters based on the collected symptom data. The biggest cluster (n=697) has a PPV of 1.3% [0.7;2.4], NPV=95.9% [95.0;96.6]. It includes „running/congested nose“ and „headache“. The cluster (n=110) with the highest PPV (38.2% [29.7;47.5], (NPV=97.8% [97.2;98.3]) includes following symptoms: impairment of taste/smell, cough, running/congested nose, sore throat/hoarseness, headache, malaise/weakness. Of note, impairment of taste/smell alone has a PPV of 35.9%[28.8;43.7], NPV=96.5% [95.8;97.1].

Discussion: The answers provided could be inaccurate due to recall bias, as participants were asked in July 2020 about their symptoms since January 1st. Moreover, time of symptom occurrence cannot be specified. Furthermore, the assumption about the presence of the infection could be false as it is based on self-reported test results.

Regarding the wild type of SARS-CoV-2, results indicate that people with symptoms such as impairment of taste/smell, running/congested nose, cough and sore throat should undergo PCR testing as the possibility of infection is higher within this group of symptoms. People experiencing no impairment of taste/smell but having fever, cough, running nose and sore throat should as well be subjected to PCR testing. However, in case of mild symptoms that are mainly associated with common cold, e.g. running nose/headache, performing a PCR could be less urgent or maybe even unnecessary. In this case, one could await further development and not overburden the laboratories. Similarly, having a cough without any additional symptoms should not be the sole reason for PCR testing.

Conclusions: Findings are consistent with the impairment of taste/smell being a predictive symptom for having a SARS-CoV-2 infection [4]. Symptom information could be useful information to prioritize limited test capacities. Further research in this topic is encouraged.

The authors declare that they have no competing interests.

The authors declare that a positive ethics committee vote has been obtained.