Study design, setting and participants
All residents of Denmark have free access to health care. Using the Danish National Patient Register (DNPR), which contains detailed information on all hospital contacts nationwide, we identified all individuals in the entire population of 5.8 million  who had received a COVID-19 diagnosis . Diagnoses in DNPR are coded according to the International Classification of Diseases 10th revision (ICD-10). All hospital contacts between January 1st and July 8th with an ICD-10 diagnosis code of COVID-19 severe acute respiratory syndrome (B972A) and/or COVID-19 infection unspecified sites (B342A) regardless of duration were considered eligible for inclusion. Patients without a unique personal registration number (PNR) in Denmark were excluded.
We defined a hospital admission as any hospital contact that lasted more than 12 hours or a shorter contact which resulted in transfer to ICU and/or death. Several hospital contacts within 24 hours were considered as one event. When admissions were less than one hour apart, the patient was considered to be hospitalised the period in between as well, thus counting towards the 12-hour hospital stay. The start of the first hospital admission to which the first COVID-19 diagnosis could be linked and which was not more than 14 days before or 30 days after the date of first COVID-19 diagnosis, was defined as the index date and was used to determine baseline characteristics as well as the start of follow-up (77.2% were admitted on the day of diagnosis).
All COVID-19 diagnoses were based on a positive polymerase chain reaction (PCR) test for SARS-CoV-2 on respiratory specimens collected by naso- or oropharyngeal swab, sputum expectoration, or tracheal suction.
Data Sources And Collection
The Danish Civil Registration system (CRS) assigns all residents a unique PNR which enables accurate linkage between the Danish national registries [20, 21]. We extracted information from the following specific registries: 1) The CRS , which comprise date of birth, sex, vital status, date of death, emigration, area of residence; 2) The DNPR , which includes dates of admission and discharge, admitting departments, and all primary and secondary discharge diagnoses and procedure codes from hospital contacts; 3) Register of Laboratory Results for Research (RLRR) , which contains nationwide laboratory information (except Midtjylland Region, population 1.3 million) using Nomenclature for Property and Unit (NPU) codes (13). Subsequent analyses were performed on pseudo anonymized data.
Variables And Outcomes
The primary outcomes were death within 7 and 30 days from the index date. Need for ICU admission (including admission to an ICU ward or the codes for use of invasive mechanical ventilation (IMV), dialysis, inotropic/vasopressor therapy and extracorporeal membrane oxygenation (ECMO) ), were used as secondary outcomes (Supplementary Table S1). Age, sex, underlying medical conditions, and laboratory parameters were included in the analysis. We calculated the Charlson comorbidity index modified by Quan (CCI Quan) at time of admission, by retrieving ICD-10 coded hospital discharge diagnoses from the previous 10 years, and categorized patients with low (0), moderate (1–2) or high (> 2) levels of comorbidity .
Laboratory parameters were collected from admitted patients within 24 hours of the index date. If a measurement was repeated within 24 hours, we used an average. Patients were followed from index date until the date of death, completion of 30 days of follow-up, or July 8th, 2020, whichever first.
Baseline characteristics are presented as medians with interquartile range (IQR) for continuous variables, and numbers and percentages for categorical variables. All-cause 30-day mortality stratified by sex, age intervals, and CCI were investigated with Kaplan-Meier (KM) analyses. Prognostic factors for the primary outcomes were evaluated in a univariate and multivariate Cox proportional hazard regression and presented as unadjusted and adjusted hazard ratios (HRs) with 95% confidence intervals (CI). Covariates included age, sex, CCI and laboratory parameters (when available). Interaction between covariates where examined. Cumulative incidence rates of ICU admission were analysed with competing risk survival analysis (Aalen Johansen estimator with death as a competing event) using Grays test for statistical significance. Laboratory parameters underwent additional univariate analysis for differences between subgroups, using t-test for statistical significance. The analyses were stratified by sex, and primary and secondary outcomes.
Statistical significance level was set at 0.05. All tests of significance were two-sided. Statistical analyses were performed using Stata (Stata Corporation LP, Texas, USA) and R (R Foundation for Statistical Computing, Vienna, Austria).
The study was approved by the Danish Health and Medicines Authority (ID: 31-1521-263) and by the Danish Data Protection Agency (P-2020-375). The study was conducted according to the STROBE statement .