key: cord-0688549-3m85gpgw
authors: Pfoh, Elizabeth R.; Hariri, Essa H.; Misra-Hebert, Anita D.; Deshpande, Abhishek; Jehi, Lara; Rothberg, Michael B.
title: Late Diagnosis of COVID-19 in Patients Admitted to the Hospital
date: 2020-06-15
journal: J Gen Intern Med
DOI: 10.1007/s11606-020-05949-1
sha: fbc760eda46fcd2471669fe8c7cd3b76b19a5dcd
doc_id: 688549
cord_uid: 3m85gpgw

nan

Testing for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is vital for identification of cases. Currently, the Centers for Disease Control and Prevention recommends that hospitalized patients with symptoms receive priority for testing. 1 Delayed testing could occur because patients have atypical presentations or the initial test is a false negative. Since test sensitivity depends on the site sampled 2 and hospital protocols usually rely on nasal samples, patients with lower respiratory tract infections can have a false-negative result. 3 Understanding how often the diagnosis of SARS-CoV-2 is missed is essential for infection control. We describe the frequency and characteristics of hospitalized patients with COVID-19 diagnosed after admission.

This retrospective cohort study included adults hospitalized before April 20, 2020, who tested positive for SARS-CoV-2. Testing became available for Cleveland Clinic Health System (CCHS) patients with suspected COVID-19 starting March 6 and was restricted to symptomatic, high-risk patients (e.g., hospitalized patients) on March 10. We identified patients with a positive SARS-CoV-2 test through CCHS's COVID-19 registry. We compared patients whose first positive SARS-CoV-2 test was on the day of admission versus later ("late diagnosis").

We collected demographic and health data from our electronic health record and the registry. We collected common symptoms of COVID-19. We also created a binary indicator of having none of the cardinal symptoms of COVID-19: cough, shortness of breath, or fever.

We report descriptive statistics. To identify factors associated with late diagnosis, we used a multivariable logistic regression model that included presenting symptoms, comorbidities, age, and sex. We used model building techniques to identify variables that were associated with late diagnosis and picked the model with the lowest AIC and BIC. To assess the association between late diagnosis and length of stay, intensive care unit (ICU) admission, discharge disposition, and mortality, we conducted multivariable regressions that included age, sex, and the variables associated with late diagnosis. CCHS's IRB deemed this study exempt.

Our study included 356 patients; 86% with COVID-19 diagnosed on admission and 14% had a late diagnosis (n = 49). Patients diagnosed later presented without fever, shortness of breath, or cough 29% of the time ( Table 1) ; 86% of the time they had ≥ 1 common symptom. They were similar in age, sex, and pre-existing comorbidities to patients diagnosed on admission. In the adjusted model, patients with late diagnosis presented less often with sputum (AOR 0.26, 95% CI 0.08-0.88) and more often with heart failure (AOR 2.36, 95% CI 1.15-4.84) and with none of the cardinal symptoms (AOR 2.89, 95% CI 1.30-6.43). Late diagnosis was not associated with length of stay, ICU admission, death, or discharge home ( Table 2) .

Eleven patients with late diagnosis had an initial negative result and 4 had two prior negative results. Four of 11 had their positive result collected from a non-nasal site. False negatives averaged 3.7 days to a positive result versus 2.2 days for others with late diagnosis (P > 0.05). False negatives were retested due to worsening symptoms (7/11), imaging suggestive of COVID-19 (2/11), and discharge protocol for skilled nursing facility (2/11).

Undiagnosed patients with COVID-19 can increase caregivers' risk of infection, as protective equipment may not be used. We found that 14% of COVID-19 patients were diagnosed after admission and most presented with ≥ 1 common symptom. Few late diagnoses were due to a false-negative test. Retesting of false negatives often occurred after clinical deterioration, but 2 patients were not diagnosed until discharge. This study is limited because we could only identify false negatives if a patient was retested. Further, we could only include routinely collected clinical data. In conclusion, routine Missing death information on 24 people (7%); missing length of stay and discharge destination on 79 people (22%) IQR interquartile range a P values were obtained using an adjusted regression model including the following confounders: age, sex, heart failure, sputum on admission and binary indicator of no cough, shortness of breath, or fever on admission

Centers for Disease Control and Prevention. Evaluating and Testing Persons for Coronavirus Disease 2019 (COVID-19). Available at

Detection of SARS-CoV-2 in Different Types of Clinical Specimens

Chest CT for typical 2019-nCoV pneumonia: relationship to negative RT-PCR testing