Methods

Participants and data collection
This retrospective study included all patients who were admitted to the No. 3 People's Hospital of Hubei province (one of the designated hospitals during the COVID-19 outbreak in Wuhan, China) between 31 January 2020 and 9 March 2020. Eligible patients had laboratory-confirmed SARS-CoV-2 infection and had available RNA virological data to estimate the duration of viral shedding. Demographic, clinical, laboratory, treatment and successive virological data were extracted from electronic medical records using a standardised data collection sheet that was modified based on the World Health Organization/International Severe Acute Respiratory and Emerging Infection Consortium case record form. We assessed the severity of illness according to the Chinese management guideline for COVID-19 (sixth version) [10], and used four classes of severity, as follows. 1) “Mild” represented patients with mild clinical symptoms and no pneumonia on chest imaging. 2) “General” represented patients with clinical symptoms (i.e. fever and respiratory tract symptoms) and pneumonia on chest imaging. 3) “Severe” represented adults who met any of the following criteria: respiratory rate ≥30 breaths·min−1, resting oxygen saturation ≤93% while breathing room air, arterial oxygen tension (PaO2)/inspiratory oxygen fraction (FIO2) ≤300 mmHg, and/or >50% lesion progression within 24–48 h on chest imaging. 4) “Critical” represented adults who met any of the following criteria: development of respiratory failure that required mechanical ventilation, occurrence of shock, and/or other organ failure requiring admission to the intensive care unit. The duration of temperature recovery was defined as the duration between the date of symptom onset and the date when the patient's axillary temperature returned to 37.3°C or below without an increase thereafter. The duration of radiological recovery was defined as the duration between the date of symptom onset and the date when the patient's radiological abnormalities were improved without appearance of new radiological lesions at other sites. We also collected the dose and duration of LPV/r treatment. All data were double-checked by two physicians (D. Yan and X-Y. Liu), with any discrepancy being resolved by consensus discussion (D. Yan, X-Y. Liu and Y-H. Gao). The institutional review board of the No. 3 People's Hospital of Hubei province approved the study, and patient-level informed consent was waived.

Virological investigations
Laboratory identification of SARS-CoV-2 infection was made at Wuhan Jinyintan Hospital and No. 3 People's Hospital of Hubei province using a real-time reverse transcription PCR (RT-PCR) assay. The same methods as previously described were adopted, in which the detection reagents were provided by the local Center for Disease Control [2, 3]. After admission, throat-swab specimens were collected and sent for the re-detection of SARS-CoV-2 RNA using RT-PCR assays every 2 days after symptom remission (including fever, cough and dyspnoea). Quantitative virological data were not available. The duration of viral shedding was defined as the interval from the date of symptom onset to the date when SARS-CoV-2 was undetectable from two consecutive throat-swab specimens (≥24 h apart), without converting positive thereafter. Corticosteroid treatment referred to the administration of methylprednisolone at a dose equivalent to ≥25 mg·day−1 during hospitalisation.

Lopinavir/ritonavir treatment
LPV/r (400 mg and 100 mg, orally, twice daily) was administered to the laboratory-confirmed cases at the discretion of the attending physicians at every isolation ward. The treatment duration was ≥10 days, according to the recommendations from the Chinese management guideline for COVID-19 (sixth version) [10]. Exposure to LPV/r was defined as having received at least one dosage of LPV/r.

Statistical analysis
Data are presented as median (interquartile range (IQR)) for continuous variables and number (percentage) for categorical variables. We employed the Mann–Whitney U-test or Kruskal–Wallis test for analysis of continuous variables, and the Chi-squared test or Fisher exact test for analysis of categorical variables. Univariate and adjusted multivariate logistic regression analyses were used to identify the risk factors associated with prolonged duration of SARS-CoV-2 RNA shedding. Prolonged viral shedding was defined as the duration of SARS-CoV-2 RNA shedding being >23 days. The cut-off was determined a priori based on the median duration of viral shedding. Outcomes were defined as the initial time-point from symptom onset to SARS-CoV-2 RNA assay negativity, which required an undetectable RNA from two consecutive throat-swab specimens (≥24 h apart) and without conversion to a positive test thereafter. Variables used for the analysis of prolonged viral shedding included age, sex, smoking status, comorbidities (including diabetes, hypertension and cardiac disease), corticosteroid treatment and LPV/r treatment. Analysis was also performed using a Cox proportional hazard model to assess the risk factors for prolonged SARS-CoV-2 RNA shedding. We employed Kaplan–Meier survival analysis to estimate the cumulative SARS-CoV-2 RNA-negativity rate and the stratified log-rank statistic to compare the difference in SARS-CoV-2 RNA clearance. All statistical analysis was performed using SPSS version 22.0 software (IBM, Armonk, NY, USA). A two-tailed p-value of <0.05 was regarded as statistically significant.