Main

Twelve patients with confirmed COVID-19 were identified in six states. Five did not require hospitalization and were isolated at home (patients 1–5) and seven were hospitalized for clinical and public health reasons (patients 6–12) (Fig. 1). Median age was 53 years (range: 21–68); eight patients were male (Table 1). Four of five patients with underlying medical conditions were hospitalized (Table 1,2).

Fig. 1: Timeline of illness onset, SARS-CoV-2 RNA detection, hospitalization, oxygen requirement and reported symptom resolution among the first 12 patients with COVID-19 in the United States, January to February 2020.
figure 1

Patients 1–5 were not hospitalized and patients 6–12 were hospitalized. Days are sequential from day of symptom onset (day 1). Light blue bars indicate time patients were under home isolation. Yellow bars indicate duration of hospitalization. Dark blue bars indicate duration of supplemental oxygen administration in hospital. The orange dashed bar indicates duration of intensive care unit (ICU)-level monitoring for patient 9. The black ‘+’ indicates collection date of the earliest sample that tested positive for SARS-CoV-2 by rRT–PCR. The black asterisk indicates collection date of the latest sample tested by CDC with a positive result for SARS-CoV-2 by rRT–PCR. The orange diamond indicates date of last report of symptoms. The black square indicates the last day of isolation precautions. Patient 9 was isolated in a healthcare facility for the full time period; they were discharged from the first healthcare facility on day 27 and subsequently transferred to a second healthcare facility for public health purposes. Patient 1 reported a cough with initial onset in mid-December before the patient traveled to China. The patient reported no change in the cough from the initial onset until reported resolution on day 14. Because onset date was difficult to determine for this patient, we have used date of detection as day 1 to assess viral RNA detection.

Table 1 Demographic characteristics, exposure history and clinical characteristics of the first 12 patients with COVID-19 in the United States, January to February 2020

Dates of illness onset ranged from 14 to 29 January 2020. Ten patients traveled to mainland China in the 2 weeks before illness onset, including nine to Wuhan City. Two patients were contacts of US COVID-19 patients in this series2. Among all patients, the duration of potential exposure ranged from 5 d to over 1 month; time between last date of possible exposure and illness onset ranged 0–5 d.

The most commonly reported initial signs or symptoms were cough (n = 8) and fever (n = 7) (Table 1). Two patients reported neither fever nor cough initially, though they developed them subsequently: one reported sore throat as their initial symptom and the other reported diarrhea (one day before fever and cough); the patient with diarrhea had recently traveled outside the United States before illness onset and later tested positive for Giardia and Clostridiodes difficile.

Over the course of illness, patients reported cough (n = 12), subjective or measured fever (n = 9), diarrhea (n = 3) and vomiting (n = 2). Three patients who never reported fever were never hospitalized and remained on home isolation. Of these, one patient reported only cough and rhinorrhea; one reported only cough, which began before travel to China and did not change from the initial onset until resolution; and one reported cough, chills, fatigue, headache and nausea.

The clinical course for each hospitalized patient is described in the Supplementary Information and Extended Data (Extended Data Figs. 17). All hospitalized patients were managed using standard, contact and airborne precautions, including eye protection. Median duration of fever was 9 d (range: 2–11). Peak body temperature during hospitalization (range: 99.1–102.9 °F) occurred at a median of illness day 9 (range: 4–10) (Fig. 2). All hospitalized patients had oxygen saturation <94% on room air at some point during their illness, with oxygen saturation nadir (range: 86–93%) occurring at a median of illness day 12 (range: 4–23) (Fig. 2). Five patients reported difficulty in breathing and four received supplemental oxygen (Table 2 and Fig. 1). Patient 9 required high-flow nasal cannula oxygen supplementation and intensive care monitoring. No patients required mechanical ventilation.

Fig. 2: Clinical and laboratory values by illness day for the first seven patients hospitalized with COVID-19 in the United States, January to February 2020.
figure 2

Clinical and laboratory values collected during hospitalization are shown for seven hospitalized patients by illness day. Values include oxygen saturation (SpO2), supplemental oxygen requirement (O2 req), maximum body temperature (Tmax), AST, ALT and white blood cell (WBC) count. Pink shading indicates days of remdesivir administration for three patients. Dotted lines show fever threshold of 100.4 °F (Tmax), 40 U l−1 for AST and ALT, 4,000 cells per µl for WBC and 150 and 250 103 mm3 for platelets.

Table 2 Clinical characteristics of the first seven patients hospitalized with COVID-19 in the United States, January to February 2020a

Two patients received a short course (≤3 d) of corticosteroids. Three, including one who received corticosteroids, received the investigational antiviral remdesivir (Gilead Sciences) under expanded access (compassionate use) for a duration of 4–10 d. Following remdesivir initiation, all had transient gastrointestinal symptoms, including nausea, vomiting, gastroparesis or rectal bleeding. No other post-remdesivir symptoms were observed. Patient 9 reported loose stool and rectal bleeding; as noted above, this patient had recently traveled outside the United States before illness onset and their stool later tested positive for Giardia and Clostridiodes difficile. Remdesivir was discontinued after improvement in each patient’s respiratory symptoms.

Blood cultures were negative in all six hospitalized patients tested, including those obtained from four patients treated empirically for bacterial pneumonia. Molecular testing for influenza A and B on respiratory specimens was negative and multipathogen respiratory PCR panels were negative for all targets in all hospitalized patients (Table 2).

Six of seven hospitalized patients had leukopenia (<4,000 cells per µl) and the white blood cell count nadir occurred at a median of illness day 9 (range: 4–15) (Fig. 2). Procalcitonin levels were <0.25 ng ml−1 in all six patients tested. Aminotransferase levels were elevated in all hospitalized patients: aspartate aminotransferase (AST) levels peaked (median peak value 129 U l−1, range 46–190 U l−1) at a median of illness day 13 (range 7–19) and alanine aminotransferase (ALT) levels peaked (median peak value 136 U l−1, range 66–389 U l−1) at a median of illness day 14 (range: 6–23). Three of seven hospitalized patients had mildly elevated alkaline phosphatase levels >115 U l−1 (maximum value 163 U l−1). Elevated lactate dehydrogenase levels >600 U l−1, coinciding with clinical deterioration, were observed in two patients tested. No major elevations in serum total bilirubin (seven patients tested) or prolongations in prothrombin time (four patients tested) were identified. Among the three remdesivir recipients, aminotransferase elevation developed in patient 6, 1 d after starting remdesivir and in patient 8, 4 d after starting remdesivir. Patient 9 had an aminotransferase peak before starting remdesivir and a second peak 5 d after starting remdesivir.

Unilateral or bilateral opacities were seen on chest imaging for all hospitalized patients (Table 2). Four hospitalized patients had normal initial chest radiographs (illness day range: 4–9). One had an abnormal chest computed tomography scan on the day of the normal chest radiograph.

All 12 patients had initial respiratory specimens collected between illness days 1–9 (median, day 4) and all but one tested positive in ≥2 respiratory specimen types (Fig. 3). Viral culture was attempted on initial respiratory specimens from nine patients and was successful for all nine, including two patients who were not hospitalized (Fig. 3); viable SARS-CoV-2 was cultured at day 9 of illness (patient 10), but was not attempted on later specimens. SARS-CoV-2 real-time PCR with reverse transcription (rRT–PCR) cycle threshold (Ct) values of virus isolated from the first tissue culture passage were 12.3–35.7 and for one patient, virus isolated from tissue culture passage 3 had a titer of 7.75 × 106 median tissue culture infectious dose per ml (Supplementary Table 1); these data were likely more reflective of growth in tissue culture than patient viral load.

Fig. 3: SARS-CoV-2 rRT–PCR results by specimen type and day among the first 12 patients with COVID-19 in the United States, January to February 2020.
figure 3

Specimen types tested include NP swab, OP swab, sputum, serum, stool and urine. Days are sequential from day of symptom onset (day 1). Viral culture was attempted on selected respiratory specimens collected early in the course of illness. rRT–PCR results were reported as positive (all three targets positive), negative (all three targets negative) or inconclusive (only one or two positive targets). Black-filled circles indicate rRT–PCR-positive specimens. Black-outlined circles indicate rRT–PCR-negative specimens. Gray-filled circles indicate specimens with inconclusive rRT–PCR results. Red squares surrounding black-filled circles indicate rRT–PCR-positive specimens from which viral culture was successful. Gray squares surrounding black-filled circles indicate rRT–PCR-positive specimens from which viral culture was unsuccessful. An asterisk indicates patients who required supplemental oxygen.

Overall, 448 specimens were collected from the 12 patients throughout the course of illness and tested at the US Centers for Disease Control and Prevention (CDC). All 12 patients had SARS-CoV-2 RNA detected in at least one nasopharyngeal (NP) swab, 11 of 12 in an oropharyngeal (OP) swab, 6 of 6 in sputum, 1 of 11 in serum, 7 of 10 in stool and 0 of 10 in urine (Fig. 3). Among 117 pairs of simultaneous NP and OP specimens, 45 (38%) had discordant results. Among 32 discordant pairs with one positive specimen, the NP specimen was positive in 21 (66%). Thirteen additional discordant pairs had one negative and one inconclusive specimen. SARS-CoV-2 RNA was detected at a maximum of day 32 in NP specimens, day 36 in OP specimens, day 29 in sputum and day 25 in stool (Fig. 3). Two patients provided sputum specimens after NP and/or OP specimens tested negative and sputum continued to be positive in both patients. In patient 7, viral RNA was detected in sputum 17 d after the last positive OP specimen and ≥2 weeks after reported symptom resolution. In seven patients who had SARS-CoV-2 RNA detected in stool, most detections occurred when viral RNA was still detectable in the respiratory tract. Among three patients who reported diarrhea, all had viral RNA detected in stool.

Mean Ct values in positive specimens were 17.0–39.0 for NP, 22.3–39.7 for OP and 24.1–39.4 for stool. Ct values of upper respiratory tract specimens were lower in the first week of illness than the second in most patients (Extended Data Fig. 8); in some patients, low Ct values continued into the second and third week of illness. SARS-CoV-2 rRT–PCR results turned positive in serum of patient 9 in the second week of illness at the time of rapid clinical deterioration.

Ct values and duration of RNA detection in the upper respiratory tract did not seem to differ by hospitalization status or oxygen requirement.

All patients reported symptom resolution (Fig. 1). Eleven patients reported cough (often intermittent) as the last symptom. Median symptom duration was 14 d (range: 6–37). SARS-CoV-2 RNA was detected after reported symptom resolution in 7 of 12 patients, including in NP (n = 6), OP (n = 3), sputum (n = 1) and stool (n = 3) specimens. Home isolation or Transmission-Based Precautions were discontinued for all patients per CDC criteria4; the last respiratory specimens tested at CDC with a positive test result were collected from these patients on days 8–36 (median = day 19).

Complete genome sequences were generated from respiratory specimens from all 12 patients. The sequences had >99% nucleotide identity to 85 reference sequences of SARS-CoV-2 genomes; phylogenetic tree analysis identified a few distinct subgroups (Extended Data Fig. 9) that were not divergent from each other, suggesting that these patients were identified during an early stage of the outbreak.

We describe the first 12 patients with confirmed COVID-19 in the United States, including clinical course of the first 7 hospitalized patients. Ten patients had traveled to China, including nine to Wuhan City and two had close contact with a US COVID-19 patient. Illness ranged from mild to moderately severe and hospitalized patients showed signs of clinical worsening in the second week. All patients recovered and three patients tolerated treatment with the investigational antiviral remdesivir. SARS-CoV-2 RNA was detected in upper and lower respiratory specimens, stool and serum. The highest viral RNA levels in the upper respiratory tract were typically detected in the first week of illness, and viable SARS-CoV-2 was cultured from early respiratory specimens. Viral RNA was detected after reported symptom resolution for seven patients, although the implications for infectiousness and transmission later in illness are unclear. SARS-CoV-2 genome sequencing and phylogenetic analysis from these 12 patients’ respiratory tract specimens support a recent zoonotic transmission event and subsequent human-to-human transmission.

The clinical manifestations described here reflect the milder end of the full disease spectrum. Severe illness and death has since been reported in the United States5,6,7. Among hospitalized patients in this report, the second week of illness was characterized by clinical or laboratory signs of worsening including hypoxemia or elevation of aminotransferases. Although some patients received empiric antibiotics for possible secondary bacterial pneumonia, we found definitive evidence of bacterial co-infection. Worsening in the second week of illness is consistent with previous reports8,9 and highlights the importance of close monitoring beyond the first week, even in patients with mild illness or no initial radiographic abnormalities.

Patient 9, the most severely ill among this series, experienced clinical deterioration late in the second week of illness. This was the only patient with SARS-CoV-2 RNA detected in serum and detection in serum was temporally related to clinical deterioration. Similar observations have been described previously10,11. Increased proinflammatory cytokines have been observed in patients with COVID-198 and it is possible that cytokine dysregulation and endothelial dysfunction contribute to both clinical worsening and SARS-CoV-2 RNA detection in serum.

We detected viral RNA and cultured virus from upper respiratory specimens, even from patients with predominantly lower respiratory tract illness. Ct values in upper respiratory tract specimens typically were lowest during the first week of illness (suggesting high RNA levels), consistent with previous reports11,12,13,14. SARS-CoV-2 RNA was detected in upper respiratory tract specimens for 2–3 weeks in most patients and for as long as 36 d. Sputum specimens were less frequently available; in two patients with a productive cough and available sputum, viral RNA was detected in sputum longer than in NP or OP specimens. We detected SARS-CoV-2 RNA in stool of multiple patients and in the serum of one hospitalized patient.

SARS-CoV-2 RNA levels and duration of RNA detection in the upper respiratory tract did not seem to vary by illness severity and viral RNA was detected after reported symptom resolution in several patients. More data are needed to better understand how duration of RNA detection, RNA levels and presence of viable virus are related to symptom progression, illness severity and infectiousness.

Three hospitalized patients received the investigational antiviral remdesivir under expanded access (compassionate use) at the time of clinical worsening based upon a decision by each patient’s clinician. Remdesivir inhibits viral replication through premature termination of RNA transcription15,16. In vitro studies have demonstrated that remdesivir inhibits SARS-CoV-2 replication in nonhuman cells17. Because remdesivir use was not given as part of a randomized controlled trial, we are unable to assess effectiveness or safety. Randomized controlled trials of remdesivir are underway18,19,20. Two hospitalized patients received corticosteroids. The World Health Organization and CDC advise against use of corticosteroids unless indicated for another reason21.

Our investigation has several limitations. Our patient sample is small, and results may not be generalizable. Information from patient interviews may have been subject to response bias. The threshold for hospitalization in these early cases was likely low because of uncertainty about COVID-19 clinical course. Illness resolution dates may be imprecise due to nonspecific lingering symptoms or symptoms from chronic or unrelated conditions. Clinical laboratory tests and radiographic studies were ordered as a part of routine patient care and were not collected systematically. SARS-CoV-2 RNA detection does not necessarily reflect the presence of infectious virus and rRT–PCR Ct values may have varied due to specimen collection or handling.

Characterization of the first 12 COVID-19 patients identified in the United States provides insight into the epidemiology, clinical characteristics and natural history of SARS-CoV-2 infection. Although duration of infectiousness is unclear, these early data show viable virus can be cultured readily from upper respiratory tract specimens soon after illness onset; further studies on infectious period and risk factors for transmission are needed. Clinicians should anticipate that some patients may worsen in the second week of illness, but appropriate monitoring of these patients will present challenges as healthcare systems work to meet the increasing demands. Studies are urgently needed to better characterize risk factors for and early indicators of severe disease. Randomized controlled trials of therapeutic options and their effects on clinical outcomes and infectiousness are critical to guide clinical and public health management. Additional investigations to understand clinical course, immunological response, SARS-CoV-2 RNA detection, viral culture and transmission, will inform clinical management and public health strategies to prevent the spread of disease.

Methods

CDC’s Human Research Protection Office determined that this work was exempt from human participants’ research regulations as it involved identification, control or prevention of disease in response to an immediate public health threat. Patient consent was waived. Forms were approved under Office of Management and Budget, number 0920-1011.

Local health departments, in consultation with clinicians, identified patients under investigation (PUIs) for COVID-19 beginning 17 January 2020. PUI testing criteria changed during this period but included the presence of fever and/or lower respiratory symptoms (for example, cough or shortness of breath) and at least one epidemiological risk factor in the 2 weeks before symptom onset. Between 17 and 31 January 2020, epidemiological risk factors were travel from Wuhan City, close contact with a patient with laboratory-confirmed COVID-19 or currently under investigation for COVID-19 (refs. 22,23). Beginning 1 February 2020, epidemiological risk factors changed to close contact with a patient with confirmed COVID-19 or history of travel from mainland China24. During both time periods, close contact was defined as being within 6 feet for a prolonged period of time25 or contact with respiratory secretions26. Close contacts of patients with confirmed COVID-19 were identified by local and state health departments and monitored for illness. Specimens from PUIs, including symptomatic close contacts, were tested for SARS-CoV-2 at CDC27.

Upper respiratory tract (NP and OP) and available lower respiratory tract (sputum) specimens were collected and tested for SARS-CoV-2 RNA by rRT–PCR28. A case of COVID-19 was defined as identification of laboratory-confirmed SARS-CoV-2 in ≥1 specimen from a patient. We included COVID-19 patients confirmed by CDC between 20 January 2020 and 5 February 2020; the size of this convenience sample was not predetermined and we aimed to describe the first cases as soon as possible. Patients sought outpatient or inpatient care at local facilities; some were transferred or referred to facilities that had specifically prepared to receive patients with COVID-19.

Patients were interviewed by public health officials about demographics, exposures, travel history and symptoms, including signs or symptoms before presentation. For all patients, available medical records were reviewed. For hospitalized patients, clinicians systematically abstracted data from the medical record.

Illness day 1 was defined as the first day of reported COVID-19 signs and symptoms; collection date of the first SARS-CoV-2-positive specimen was used for one patient with no clear symptom onset date. When prehospital symptoms or onset dates in the medical record differed from those reported from the public health interview, the latter were used. Results for virological tests were reported relative to illness day 1. Duration of potential exposure to SARS-CoV-2 was defined as dates of travel to China or dates of first to last exposure to a US patient with COVID-19. Fever was defined as feeling feverish or measuring body temperature ≥100.4 °F. We requested collection of NP swabs, OP swabs, sputum (if available), serum, urine and stool every 2–3 d throughout infection.

Multiple clinical and laboratory observations were included from each of the 12 patients. No associations between clinical or laboratory data and outcomes were tested statistically. Data were analyzed and visualized using Excel in Microsoft Office 365, SAS v.9.4, R v.3.6.2 and Python v.3.7.3 (refs. 29,30,31,32).

CDC evaluated specimens using SARS-CoV-2 RNA detection, viral culture, whole genome sequencing and phylogenetic analysis.

All patient specimens were tested for the presence of SARS-CoV-2 RNA by rRT–PCR targeting three regions of the gene encoding the nucleocapsid protein28. The rRT–PCR results were reported as positive (all targets positive), negative (all targets negative) or inconclusive (only one or two positive targets). If an initial result was inconclusive, the specimen was re-tested; if both tests were inconclusive, the final result was reported as inconclusive. For inconclusive results, a positive detection of SARS-CoV-2 could not be definitively ruled out. If available, Ct values, which are approximately inversely related to the RNA levels in each specimen33, were reported as the mean of three reported Ct values. Results from serial rRT–PCR testing were not immediately available to inform clinical management.

Viral culture was attempted from early SARS-CoV-2-positive respiratory specimens (NP, OP or sputum) with Ct values <33. Specifically, 100 µl of clinical specimens were diluted twofold across a 96-well plate in serum-free DMEM supplemented with 2× penicillin–streptomycin and 2× amphotericin B (Sigma). Vero CCL-81 cells were trypsinized and resuspended in DMEM + 10% FBS + 2× penicillin–streptomycin + 2× amphotericin B at 2.5 × 105 cells per ml. A 100-µl cell suspension was added directly to the clinical specimen dilutions and mixed gently by pipetting. The inoculated cultures were grown in a humidified 37 °C incubator with 5% CO2 and observed for cytopathic effect daily. When cytopathic effect was observed, presence of SARS-CoV-2 was confirmed by rRT–PCR.

Nucleic acid was extracted from respiratory specimens (NP, OP or sputum) positive for SARS-CoV-2 rRT–PCR and used for whole genome sequencing on both Sanger and Oxford Nanopore MinION sequencing platforms. For Sanger sequencing, 37 sets of individual nested PCR assays spanning the entire 2019-nCoV genome were designed based on the reference sequence, GenBank accession number NC045512. PCR amplicons were sequenced in both directions using Big Dye 3.1 cycle sequencing kits (Thermo Fisher Scientific) on an ABI 3730 Automated Capillary Sequencer (Thermo Fisher Scientific) with PCR primers and additional internal primers. The consensus sequences were generated from both sequencing directions using Sequencher 5.4.6 (Gene Codes Corporation). For Nanopore sequencing, individual PCR amplicons were pooled and barcoded by sample, which were used for library preparation using the Ligation Sequencing kit (Oxford Nanopore Technologies). Libraries were run on a MinION sequencer and consensus sequences were generated using minimap v.2.17 and samtools v.1.9 (refs. 34,35).

Full genome sequences from the 12 confirmed cases in this report and 85 full genome sequences (as of 11 February 2020) available from GenBank and the Global Initiative on Sharing All Influenza Data database were aligned using MAFFT v.7.450 (ref. 36). Sequences with obvious early stop codons were excluded. Phylogenetic trees were then inferred with the maximum likelihood method using the Hasegawa Kishino Yano nucleotide substitution model with γ-distributed rate variation among sites (HKY + G) and 1,000 bootstrap replicates implemented in Geneious Prime (Biomatters) and MEGA X37,38.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.