Study design and ethical considerations
We conducted a longitudinal observational study on COVID-19 patients in Osaka, Japan. Data from the active epidemiological investigation for COVID-19 under the Infectious Diseases Control Law  were used in this study, and data was collected between 29 January 2020 and 31 May 2020. Public health nurses working at public health centres throughout Osaka Prefecture collected the data through active epidemiological investigation, such as patient age, sex, comorbidity, date of onset, symptoms, and PCR testing which was collected by telephone or electronic-based worksheets. Informed consent was not required as data analyses were performed in accordance with the Infectious Diseases Control Law; however, all data were anonymized to uphold confidentiality and patient privacy. This study was approved by the ethical committee of Osaka University (T20114).
Osaka Prefecture is in the central area of western Japan and covers an area of 1905 km2. Osaka, the third most populous prefecture in Japan, had an estimated population of 8,819,226 as of 1 April 2020 . The first laboratory-confirmed COVID-19 case in Japan was detected on 16 January 2020; thereafter, the first case in Osaka was confirmed on 29 January 2020. Our study population included all laboratory-confirmed COVID-19 patients in Osaka between 29 January 2020 and 31 May 2020; 1783 patients were treated across 62 hospitals and three non-hospital recuperation facilities as well as homes. The last follow-up date was 30 June 2020. All patients were diagnosed via PCR tests on specimens of sputum, nasopharyngeal swabs, or both, according to laboratory guidance from the National Institute of Infectious Diseases, Japan .
We restricted patients involved in this study to those who were classified as ‘non-severe’ cases throughout inpatient or non-hospital recuperation. Patient severity was defined by categorization provided by the Japanese Ministry of Health, Labour, and Welfare  wherein ‘mild (with respiratory failure)’ indicated difficulty breathing and SpO2 levels ≤93% needing oxygen therapy, and ‘severe’ indicated being admitted to the ICU or needing mechanical ventilation. ‘Non-severe’ patients were those who were not categorized in any of these criteria . For asymptomatic patients, the date of specimen collection for a positive test was considered the date of onset, as in the same manner as the guideline principle. As of 31 May 2020, the release criteria for these patients were over 72 h of symptom clearance and two consecutive negative PCR tests on specimens of either sputum or nasopharyngeal swabs, taken 24 h apart. In Osaka, non-severe patients were first admitted to hotel recuperation facilities on 14 April 2020. Admission to non-hospital recuperation facilities was decided by the Infected Persons Follow-up Centre in Osaka Prefecture as requested by the director of public health centres and based on the patients’ vital signs, symptoms, age, and comorbidities. The admission criteria for these patients included non-severe patients, those who maintained independence in activities of daily living (ADLs), and those who had no known active comorbidities requiring medical care.
Clinical characteristics included the following: age, sex, comorbidities (pre-categorized by: diabetes, respiratory disease, coronary heart disease, immunodeficiency, cancer, hypertension, and other), symptoms at the time of admission (pre-categorized by: fever, cough, dyspnea, digestive symptoms, olfactory and taste disorders, fatigue, and other), radiological findings of pneumonia in chest examination of X-ray or CT, location of recuperation, days to admission, and detailed data of PCR testing. The date of symptom onset was defined as the date on which the patients reported the first noticeable symptom. Days to admission was defined as the date from symptom onset to date of admission. Information on comorbidities was self-reported and collected based on interviews with public health nurses. At the time of writing, a definitive treatment for the novel coronavirus was still under investigation. Data on pharmaceutical methods of treatment were not obtained in this study, as all patients were considered non-severe and did not require intensive medical intervention such as intubation or ECMO.
Duration of viral clearance was calculated as the number of days from the date of symptom onset to the date of specimen collection of the first negative PCR test (out of two consecutive tests). Specimens from nasopharyngeal swabs were used in hospitals for PCR tests; sputum specimens were used to test for negativity at hotels or homes. Prolonged duration of viral clearance was defined as the top quartile of the duration and days after 14 days of symptom onset.
Continuous variables were presented as median and interquartile range (IQR). Categorical variables were presented as numbers and percentages. Comparisons were determined by Mann-Whitney U test for continuous variables and chi-square test or Fisher’s exact test for categorical variables as appropriate. Univariate and multivariate logistic regression analyses were performed to explore the association of sex, age groups, location of recuperation, comorbidities, and symptoms with prolonged duration of viral clearance. Prolonged duration of viral clearance was defined as more than 29 days from symptom onset, the top quartile of the overall duration of viral clearance. Logistic regression analysis was used to calculate odds ratios. Linear regression analysis was also performed to examine the days of viral clearance after 14 days of symptom onset, according to each potential predictor. These analyses were stratified by the location of recuperation, that is, hospitals and non-hospitals. Statistical analyses were performed using Stata version 15.1 (StataCorp, College Station, TX, USA). All p values were two-tailed with p < 0.05 considered statistically significant.