outcome measures modelled as a dependent variable, >500 data points were included. In fact, this count should be further multiplied by the models tested in each study, i.e. allele and genotype association, dominant, recessive, codominant, and by the number of tests. However, we abstracted for each variant the statistics showing the larger effect size or the most significant result, if applicable. Populations The 32 articles included for data abstraction analysed 26 independent cohorts—i.e. 3 cohorts were described in 4, 3 and 2 studies, respectively (Table 1). Consistently with the outbreaks experienced in the last decades and the period considered for inclusion, SARS infection was by far the most frequent endpoint. Therefore, all studies investigated patients of Asian ancestry— Chinese Han and Vietnamese—except the only study which focused on MERS in Saudi Arabia and one recent analysis on individuals of European ancestry based in the UK Biobank. Patient series were ascertained according to the current diagnostic criteria and recruited retrospectively. No randomisation procedure was found in eligible studies. The healthcare setting—i.e. whether in-patient or out-patient clinic, follow-up, etc.—was seldom reported. The control series resulted to be consistent for ancestry. No paired case-control enrolment was applied.The discovery samples were heterogeneous in size across studies, ranging from 44 to 323,570, while the patient series count was from 20 to 817 (Table 1). Outcomes Out of the 32 studies, 27 considered the disease status as the primary outcome measure, 3 seropositivity or positive nasopharyngeal shedding, 1 used disease-associated biomarkers (serum LDH level, white-blood count) and 1 measured COVID-19 severity (Table 1). A large proportion of association studies considered the severity of the disease course as a secondary outcome, as measured by admittance to intensive care unit, administration of oxygen therapy, or death. The clinical variables analysed as secondary outcomes were ascertained on a subset of the patient series and modelled as binary variables (Table 1). Designs All studies implemented a case-control design. If the association with clinical outcome was tested, a subset of patients was analysed and classified in a binary variable. Genotype and allele frequencies were compared by using univariate parametric statistics. Several studies applied also multivariate statistics including age and sex as covariate; one considered lifestyle-related risk factors; one corrected for putatively associated variants. Di Maria et al. Human Genomics (2020) 14:30 Page 4 of 19 Table 1 Articles included as eligible for data abstraction, ordered by year of publication and first author. For each study, main features, genes/loci examined and summary of the main findings are reported Author, year Population description Country Disease Primary outcome Other outcomes Sample N Cases N Controls N Notes on cohorts Gene/locus Conclusions Ref. Lin M, 2003 SARS cases admitted to Taipei Hospital, Taiwan Taiwan SARS Manifest disease Severity 134 33 101 HLA-A, HLA-B, HLA-DRB1 HLA-B*4601 nominally associated with severity (vs larger control group), not with infection after P correction [28] Chiu RWK, 2004 SARS patients admitted at Hong Kong Chinese University Hospital Hong Kong SARS Manifest disease Severity 496 168 328 ACE2 Negative results [29] Itoyama S, 2004 Vietnamese patients with SARS Vietnam SARS Manifest disease Hypoxemia 147 44 103 # ACE ACE nominally associated with severity [30] Ng MHL, 2004 SARS patients admitted at Hong Kong Chinese University Hospital Hong Kong SARS Manifest disease Severity 18864 90 18774 HLA-A, HLA-B, HLA-DR, HLA-DQ HLA-B*0703 and -B*0301 nominally associated with the disease; no association with severity [31] Chan KC, 2005 SARS patients previously admitted at Hong Kong Hospitals Hong Kong SARS Manifest disease Severity 466 140 326 ACE Negative results [32] Hamano E, 2005 SARS patients Vietnam SARS Manifest disease Severity (oxygen therapy) 147 44 103 OAS-1, MxA, PKR Nominal association of OAS-1 with disease but not severity; nominal association of MxA with severity only [33] Ip WK, 2005 SARS patients previously admitted at 5 Hong Kong Hospitals Hong Kong SARS Manifest disease Serum MBL level; severity (death) 1757 569 1188 MBL Association of -221 Y allele with disease and MBL level; no association with mortality [34] Itoyama S, 2005 Vietnamese patients with SARS Vietnam SARS Manifest disease 147 44 103 # ACE2 Negative results [35] Yuan FF, 2005 SARS patients admitted at Prince of Wales Hospital, Hong Kong Hong Kong SARS Manifest disease Severity (ICU or death) 380 180 200 § FcγRIIA, MBL No association with MBL; FcγRIIA nominally associated with severity [36] Zhang H, 2005 SARS patients from Bejing, China SARS Manifest disease Serum MBL level 744 352 392 MBL Association of nt 54 B allele with disease and MBL level [37] Di Maria et al. Human Genomics (2020) 14:30 Page 5 of 19 Table 1 Articles included as eligible for data abstraction, ordered by year of publication and first author. For each study, main features, genes/loci examined and summary of the main findings are reported (Continued) Author, year Population description Country Disease Primary outcome Other outcomes Sample N Cases N Controls N