Reviewing the articles, it became obvious that four of them originate from the same institution (1–3, 6). It remains unclear whether patients were described several times, or whether they were different patients.

The articles described only patients with confirmed COVID-19. In clinical routine, however, it is important to differentiate between COVID-19 and other diseases. It would be very useful to know differences in symptoms and laboratory findings in patients with initially suspected and later confirmed, or excluded, COVID-19 infection.

This issue is mentioned in one article: “The absence of fever in COVID-19 is more frequent than in Severe Acute Respiratory Syndrome Coronavirus (SARS-CoV) and Middle East Respiratory Syndrome Coronavirus (MERS-CoV) infection, so afebrile patients may be missed if the surveillance case definition focuses on fever detection” (5). In a further article, the consequences of atypical presentations are described: “A total of 14 patients (10%) initially presented with diarrhea and nausea 1 to 2 days prior to development of fever and dyspnea…. One patient in the current study presented with abdominal symptoms and was admitted to the surgical department. More than 10 health care workers in this department were presumed to have been infected by this patient…” (4).

The data in all included studies were collected retrospectively. It remains unclear, however, in how many cases information was missing, since just one publication reported missing values in some, not all, tables (5).

Ageusia or anosmia were not reported in the included articles, but only in more recent ones from China (9) and Italy (10, 11). Why were these not reported in the included articles? One answer could be that symptoms were assessed in accordance with the International Severe Acute Respiratory and Emerging Infection Consortium (ISARIC) case record form, as indicated in the methods section of the articles. In this form, there is no space for “other symptoms” (12).

Data about medication before hospital admission are lacking, thus no information can be obtained whether specific drugs render the patients more prone to a severe course of the disease. This issue, relating angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers, and statins has been extensively discussed in the meantime (13, 14).

Data about level of education and occupation, household size and composition are not reported. Neither weight nor body mass index of the patients are reported. The reason might be the above-mentioned ISARIC case record form, as it does not require such information. Among the comorbidities, the terms “obesity” and “malnutrition” “as defined by the clinical staff” are registered in the case record form, but not the body mass index as an objective measure of obesity. Whether obesity influences the prognosis in COVID-19 is a still controversially assessed issue, and it is uncertain whether ethnic and socioeconomic differences might play a role for the discrepant findings (15–17).

We miss urine analysis data, which might help to characterize the type and pathomechanism of renal failure in COVID-19 (18). Is the frequently encountered hypalbuminemia due to renal loss of protein? Again, this lack of data may be related to the ISARIC case record form, where no such data are requested.

“Cardiac injury” is frequently reported in COVID-19 patients. The definition for “cardiac injury” was adopted from a publication in which cardiac biomarkers, electrocardiography (ECG), or echocardiography among hospitalized patients infected with influenza A (H7N9) virus had been reported (19). “Myocarditis” is reported in one publication, however, it remains unclear how it was diagnosed (2). In another publication, arrhythmia is mentioned, but no description is given (4). In yet another publication, “cardiac injury” is only defined by elevations of the troponin (3).

No details about ECG findings are given in any of the included articles. Whereas, echocardiography should only be performed if it is expected to provide clinical benefit (and not as screening investigation), because of close proximity of patient and echocardiographer; this is not the case with ECG (20). An ECG is easy to perform and yields useful information about the cardiac condition, especially in patients with preexisting cardiovascular diseases. It would be of great interest to know how often and which types of arrhythmias were registered, the prevalence and dynamics of ST-elevations or depressions, and their association with cardiac biomarkers like troponin or natriuretic peptides. Up to now, our knowledge about ECG findings in COVID-19 patients from China is limited to small series comprising 63 (21), 102 (22), and 112 patients (23).

The negligence of relevant variables can have important consequences. The variables may be confounders and the omission of these confounders will produce biased results. Moreover, the consequences might be enduring in future analyses, as model-building is often based (and even advised in some textbooks) on results of prior studies (24, 25). The selection of patients is unclear or biased (4, 5). In addition, patients still in follow-up, constituting the major part of observations, were excluded from the sample in two studies (3, 6). This phenomenon, that patients are still hospitalized when the article is written, is not only encountered in publications from China but also from Europe and the U.S.A (17, 26).