The effect of sex hormone on COVID-19: analysis of laboratory-conrmed 5061 patients in South Korea

Limited data describing the effect of sex hormone on coronavirus disease (COVID-19) is available. We evaluated the effect of sex hormone on prevalence and clinical outcomes of COVID-19. Retrospective cohort study was performed using the nationwide claims data of 5061 adult patients with laboratory-conrmed COVID-19 in South Korea, from January 20 to April 8, 2020. COVID-19 was most prevalent in women of the 20-39 age group (1250 [44.14%]). Men were more likely to receive oxygen therapy (144 [6.46%] vs 131 [4.63%], P=0.004), be admitted to the intensive care unit (60 [2.69%] vs 53 [1.87%], P=0.049), and stay longer after admission to the intensive care unit (19.70±11.80 vs 14.75±9.23, P=0.016). However, there was no signicant difference in mortality rates between men and women. In multivariable Cox analysis, independent risk factors for mortality were older age and underlying comorbidities, rather than sex. To evaluate the effect of HRT among women, subgroup analysis was implemented using age-matched case-control data with a 1:3 ratio of females receiving HRT to those who did not. HRT did not have statistically signicant association with clinical outcomes. This study suggests that sex hormone may not affect prevalence and clinical outcome of COVID-19 in South Korea.


Introduction
In South Korea, since the identi cation of the rstcon rmedcase on January 20, 2020, coronavirus disease (COVID-19)casesoccurred sporadically during the initial stages. AfterFebruary 19,a large number of patients was identi ed fromreligious groups andlong-term care facilities. Subsequently, weexperienced anexponential rise in the number of con rmedcasesnationwide 1 . The spread of the infection hasbeenmitigated by adherence to social-distancing measures,mask-wearing,contact-tracing, and isolation of con rmed cases.Although thenumber of con rmed cases reached 10752 with 244 deathsas of April28, the acute upward slope appears to havebeen attened 2 .
Sex has already been a factor of marked importance in theshorthistory of COVID-19. According to several studies, men havehighermorbidity and mortality to COVID-19 than women 3,4 .Somestudies suggest that sex hormone may affect the susceptibilitytosevere acute respiratory syndrome coronavirus 2(SARS-CoV-2)infection and subsequent mortality [3][4][5][6] . Basedon these ndings, a clinical trial is recruiting COVID-19 patientstoevaluate the effect of female sex hormone on clinical outcomesofpatients as administering oestrogen to adult men orpost-menopausalwomen who have low level of female hormone(ClinicalTrials.govIdenti er: NCT04359329). Presently, only fewliterature describingthe effect of sex hormone on morbidity andmortality of COVID-19 isavailable. Understanding the effects of sexhormone on COVID-19 mayassist in the identi cation of high-riskpatients, prognosisprediction, and triage of patients forhospitalization during thecurrent pandemic.
Epidemiologic ndings in South Korea were different toothercountries. Con rmed cases are more common in women, andCOVID-19is most prevalent in young adults 7 . Here, weevaluatesex-speci c characteristics of patients withlaboratory-con rmedCOVID-19 in this epidemiologic setting of SouthKorea usingnationwide claims data and aim to investigate theeffects of thefemale sex hormone on the clinical course and diseaseseverity ofCOVID-19.

Patient data
We performed a retrospective cohort study using theHealthInsurance Review and Assessment (HIRA) dataset consisting ofadultpatients (age≥ 20) with laboratory-con rmed COVID-19betweenJanuary 20, 2020 and April 8, 2020, in South Korea. TheSouthKorean government operates a mandatory nationwide insurancesystem(National Health Insurance) which covers all forms ofhealthservices, including hospitalization, ambulatory care,andpharmaceutical services. All health care utilization The KCD-7 codes for the diagnosis of COVID-19 andsubjects'comorbidities are described in Supplementary Table 1.Underlyingcomorbidities were de ned as the reimbursement of theKCD-7 codeof study diseases, within 5 years prior to the diagnosisofCOVID-19. Drug codes of oestrogen therapy, oestrogen +progesteronetherapy, and tibolone were used to identify the use ofhormonereplacement therapy (HRT) in female patients. Recent use ofHRT wasde ned as the prescription of HRT drug codes for more than1 monthafter 2019. Never use of HRT was de ned as no prescriptionof HRTdrug code for the 5 years covered by the database.

Statistical analysis
Descriptive analysis was used to compare thesex-speci ccharacteristics of COVID-19 patients.
Continuousvariables werecompared using the independent t-test, andcategorical variableswere compared using the chi-squared test orFisher's exact test.Cox proportional hazard model was used toanalyse the risk factorsfor mortality. The effect of HRT on theclinical outcomes of womenwas evaluated using subgroup analysis ofage-matched case-controldata with a 1:3 ratio of women who had beenprescribed HRT for thepast year to those who did not. Allstatistical analyses wereperformed using SAS Enterprise Guide (SASInstitute Inc., Cary, NC,USA). Analysis results with P < 0.05were consideredstatistically signi cant. The study protocol wasapproved by theInstitutional Review Board (3-2020-0072) of YonseiUniversityCollege of Medicine, Gangnam Severance Hospital (Seoul,Republic ofKorea), and the protocol adhered to the tenets of theDeclarationof Helsinki. Since the study was retrospective and thestudyparticipants were anonymized, the institutional review boardwaivedthe requirement for written consent from thepatients.

Data availability
The datasets generated or analysed during the current studyareavailable from the corresponding author on reasonablerequest. Figure 1 showed the trend of patients withlaboratory-con rmedCOVID-19 between January 20, 2020 and April 8,2020, in SouthKorea. During the period, a total of 10384 patientsinfected withSARS-CoV-2 were reported to Korea Centers for DiseaseControl andPrevention. Of those, 5061 adult patients with con rmedCOVID-19were included in our study. Clinical characteristics andoutcomesof the patients were reported in Table 1

Discussion
To our knowledge, this study is the rst report with alargenumber of con rmed cases evaluating the effect of sex hormoneonthe clinical outcome of COVID-19. In this study, con rmedcaseswere most prevalent in young women. Older age andunderlyingcomorbidities rather than sex had a stronger associationwithmortality. HRT did not have a statistically signi cantassociationwith clinical outcomes in peri-and post-menopausalwomen. These ndings suggest that female sex hormone may not have aprotectiverole in morbidity and mortality of COVID-19.
Male dominant susceptibility to viral respiratory infectionswaspreviously suggested 9,10 . Several studies havereporteda male-biased sex ratio in cases con rmedwithCOVID-19 [11][12][13] . In the Middle East respiratorysyndromecoronavirus (MERS-CoV) outbreak in South Korea, the numbero nfected men was higher than that of infected women (59.7%vs.40.3%) 14 . However, epidemiologic studies of severeacuterespiratory syndrome coronavirus (SARS-CoV) infection showedthatinfected female patients are more common inseveralcountries (55.7% in Hong Kong, 61.0% in Toronto, 66.0%inSingapore, and 63.2% in Guanzhou) 15 . Wealsodemonstrated that the number of patients infected withSARS-CoV-2were more common in women. In South Korea, the largestnumber ofcases of COVID-19 is associated with speci creligiousgroups 1 . The predominance of women in thosereligiousgroups could be re ected in the female-biased sex ratiowithin ourdomestic COVID-19 outbreak. In addition, the occupationalhazardsof a crowded workplace could be a risk factor forSARS-CoV-2infection. In South Korea, the COVID-19 outbreak emergedat a callcentre, where the workers were at a high risk of coming incontactwith each other 16 . In this call centre, theaffectedindividuals were predominantly women. We believe that thesesocialand cultural factors could have led to sex-speci cdisparities inCOVID-19 morbidity in South Korea.
In contrast to the studies conducted in other countries,ourstudy showed that COVID-19 was most prevalent in women of the20-39age group. This nding is also seen in another study usingdatareported to the Korea Centers for Disease ControlandPrevention 7,17 . In 2015, the MERS-CoV outbreak inSouthKorea spread through in-hospital infections 14,18 .Hence,during the COVID-19 outbreak, clinicians activelyperformeddiagnostic tests to prevent in-hospital transmission. Inaddition,the Korean government established a wide range o ndications fordiagnostic testings, so that any individual wantingto be testedcan undergo the test, even if symptoms are not severe.Therefore,many healthy young adults with mild or no symptoms couldbe testedand diagnosed. Our results may represent the realepidemiology ofCOVID-19 encompassing asymptomatic to severecases. Data from several countries showed a male-biased sex ratioinCOVID-19 mortality 11,13,19 . In previous micemodels,female hormones had a protective effect on mortality duetoSARS-CoV infection 3 . McCoy et al. suggestedthatandrogens, which could play an important role in SARS-CoV-2entryinto the host cell, seemed to be implicated inCOVID-19mortality 4 . However, in our study, male sex wasnot anindependent risk factor for mortality, and there wasnostatistically signi cant association between HRT and theclinicaloutcomes of peri-and post-menopausal women. Although ourstudy didnot corroborate a role of female hormones againstSARS-CoV-2infection, further studies, which include emerging casesacrossseveral countries, are required to determine the effect ofsexhormones during COVID-19 infection.
This study has several limitations. First, due to the natureofHIRA data, it was impossible to investigate detailedpatientinformation, such as smoking history, viral loads,laboratoryresults, and radiologic ndings, which could in uenceclinicaloutcomes. Second, we cannot rule out the possibilityofunderestimated number of male patients due to asymptomatic ormildinfections in COVID-19. A cohort study of patients withsubclinicalmanifestations is needed to determine the sex ratio ofentireSARS-CoV-2 infections. Third, because of the limited numberofwomen receiving HRT, we could not fully evaluate the effect ofsexhormone on clinical outcomes of COVID-19. In addition,menopausalstatus of female patients could not be evaluated.However,considering that the mean age of menopause, 49.3±0.1 years,isrelatively younger in Korea than that ofothercountries 20 , the patients included in subgroupanalysismay represent peri-and post-menopausal women.
In conclusion, this study may provide indirectepidemiologicevidence on the effect of female sex hormone on theprevalence andclinical outcome of COVID-19. Our report could assistin theeffective planning of valuable health care resource use andpatientmanagement during the current pandemic.