key: cord-0775576-rtqtru8k
authors: Jacobson, Karen B.; Purington, Natasha; Parsonnet, Julie; Andrews, Jason; Balasubramanian, Vidhya; Bonilla, Hector; Edwards, Karlie; Desai, Manisha; Singh, Upinder; Hedlin, Haley; Jagannathan, Prasanna
title: Inflammatory but not respiratory symptoms are associated with ongoing upper airway viral shedding in outpatients with uncomplicated COVID-19
date: 2021-11-28
journal: Diagn Microbiol Infect Dis
DOI: 10.1016/j.diagmicrobio.2021.115612
sha: eb87fcc581c7f383e0130be9c8f78b3ba088be7d
doc_id: 775576
cord_uid: rtqtru8k

Although the vast majority of SARS-CoV-2 infections are uncomplicated, our understanding of predictors of symptom resolution and viral shedding cessation remains limited. We characterized symptom trajectories and oropharyngeal viral shedding among 120 outpatients with uncomplicated COVID-19 enrolled in a clinical trial of Peginterferon Lambda, which demonstrated no clinical or virologic benefit compared with placebo. In the combined trial cohort, objective fever was uncommon, inflammatory symptoms (myalgias,fatigue) peaked at 4-5 days post symptom onset, and cough peaked at 9 days. The median time to symptom resolution from earliest symptom onset was 17 days (95% CI 14-18). SARS-CoV-2 IgG seropositivity at enrollment was associated with hastened resolution of viral shedding (HR 1.80, 95% CI 1.05-3.1, p=0.03), but not with symptom resolution. Inflammatory symptoms were associated with a significantly greater odds of oropharyngeal SARS-CoV-2 RNA detection; respiratory symptoms were not. These findings have important implications for COVID-19 screening approaches and trial design.

median time to symptom resolution from earliest symptom onset was 17 days (95% CI 14-18). SARS-23

CoV-2 IgG seropositivity at enrollment was associated with hastened resolution of viral shedding (HR 24 1.80, 95% CI 1.05-3.1, p=0.03), but not with symptom resolution. Inflammatory symptoms were 25 associated with a significantly greater odds of oropharyngeal SARS-CoV-2 RNA detection; respiratory 26 symptoms were not. These findings have important implications for COVID-19 screening approaches and 27 trial design. fever and shortness of breath are common (3-5). In contrast, outpatients with more mild disease report 43 less fever, but hyposmia is more common(6, 7). Data on symptom trajectories and clusters is limited, as 44 studies in outpatients have largely been limited to cross-sectional surveys. Furthermore, the relationship 45 of these syndromes to viral shedding and SARS-CoV-2 antibody responses have not been described in 46 detail. 47

In this study, we leveraged a cohort of participants with uncomplicated COVID-19 enrolled in a 49 randomized, placebo-controlled trial of Peginterferon Lambda to perform a detailed examination of the 50 symptomatology and natural history of uncomplicated SARS-CoV-2 infection. We aimed to 1) 51 characterize symptom trajectories and clusters; 2) assess predictors of symptom resolution and 52 Spearman's correlation coefficient and exploratory factor analysis (EFA) were used to assess pairwise 100 correlations between symptom severity and whether symptoms clustered at various days from 101 symptom onset, respectively. EFA was carried out on symptoms reported at symptom onset and on days 102 4, 7, 10, 14, and 21 after symptom onset. Only symptoms with more than 5% overall study prevalence 103 were chosen for examination (Supplementary Methods). Changes in factors derived and symptoms 104 contributing to each factor were compared descriptively across time points of interest. The joint 105 trajectory of symptoms was assessed to determine whether participant clusters persist over time 106

(Supplementary Methods). Fisher's exact test and the Kruskal-Wallis rank sum test were used to test for 107 differences in baseline characteristics and demographics by cluster assignment. 108

Time to first symptom resolution among those reporting symptoms was defined in two ways: 1) as time 111 from reported symptom onset, and 2) as time from enrollment, until the first day when no symptoms 112 were reported, given that vital signs and laboratory tests were not measured prior to enrollment. 113

Participants who did not experience resolution were censored on the day of their last reported 114 symptom. Cox proportional hazards models were fit as a function of covariates of interest. Both models 115 included age at randomization, sex, body mass index (BMI), race/ethnicity, and treatment arm. For all models, crude and adjusted hazard ratios, 95% confidence intervals, and Kaplan-Meier estimates 128 were reported. 129

Mixed effect logistic regression models were fit to PCR result (positive vs negative) as a function of 132 symptom reported (yes vs no), days from randomization, continuous age at randomization, sex, 133 treatment arm, and a random effect for participant to account for repeated measures over time within a 134 participant. This model was fit to each of the 16 symptoms as defined in Figure 1 . Days from enrollment 135 was used in order to align the longitudinal PCR and symptom data. Only days when both the symptom 136 questionnaire and PCR results were available were included. Odds ratios, 95% CIs, and p-values for 137 symptom were reported from each model. 138

Fisher's exact test and the Kruskal-Wallis rank sum test were used to test for differences in baseline 141 characteristics and demographics. 142 143 P-values <0.05 were considered statistically significant. As these were secondary analyses, we did not 144 adjust for multiplicity, but instead consider our findings to be hypothesis generating. 145

We enrolled 120 participants between April 25 and July 17, 2020 ( Figure S1 ). The median age was 36 148 years, 50 participants (41.7%) were female, and 75 (62.5%) were LatinX ethnicity (Table S1) Table S4 ). However, we found 165 that the distributions over time of diarrhea, rash, runny nose, fatigue, myalgias, and chest 166 pain/pressure differed between placebo and lambda, with these symptoms peaking later in 167 Lambda vs. placebo (Wilcoxon rank sum test p<0.05; Figure S4A -D; Table S4 ). Therefore, while 168 the overall prevalence of symptoms was similar between the two arms, there were some 169 individual symptoms that lasted longer in the Lambda arm than the placebo arm. Of 2730 at-home daily oral temperature measurements among 118 participants (average 23.1 172 measurements/participant), only 7 (0.3%) measurements in 6 participants were >100.4 F. However, 173 oral temperatures decreased significantly after enrollment, with mean temperatures declining from 174 97.9F (95% CI 97.8-98.0) between days 1-6 to 97.6F (95% CI 97.5-97.7) between days 7-28 (coef 0.75, 175 95% CI 0.66-0.84, P<0.001, Figure 1C ). In parallel, at home oxygen saturation measurements increased 176 from 96.9% (95% CI 96.6-97.2) from days 1-<7 to 97.4% (95% CI 97.2-97.6) from days 7-28 (coef 0.46, 177 95% CI 0.21-0.72, P<0.001, Figure 1D and/or later peak severity than cluster A, especially chest pain/pressure, fatigue, and myalgias. 193

However, cluster A (n=74) had slightly higher prevalence of runny nose and sore throat compared to 194 cluster B. There were no differences between Clusters A and B in terms of demographics, BMI, initial lab 195 values, or baseline seropositivity (Table S3 ). While participants in cluster B had slightly shorter time to 196 viral shedding cessation, these differences were not statistically significant. In this study, we leveraged a comprehensive and detailed dataset among outpatients with 237 uncomplicated COVID-19 enrolled in a Phase 2 clinical trial to carefully define symptom prevalence, 238 longitudinal emergence and resolution of individual symptoms and symptom clusters, and associations 239 between symptoms, viral shedding, and SARS-CoV-2 seropositivity. Although objective fever was 240 uncommon, inflammatory symptoms (myalgias, chills, fatigue) were common and peaked early; these 241 symptoms were also associated with ongoing viral shedding. In contrast, respiratory symptoms were not 242 associated with viral shedding. Though SARS-CoV-2 seropositivity at enrollment was associated with 243 more rapid viral shedding cessation, it was not associated with more rapid time to symptom resolution. 244

Participants were enrolled early in the course of disease, with excellent participant retention and little 245 missing data. Only 4 participants were hospitalized, and no deaths observed. 246 247 Importantly, though the most prevalent symptoms observed were consistent with previous reports,(3, 4, 248 6, 11, 12) we observed different prevalence peaks per symptom. The most common initial symptoms 249 were headache, myalgias and chills; these symptoms peaked at 4 days post symptom onset. Shortness 250 of breath and cough peaked at 6 and 9 days, respectively. These data differ from a prediction model 251 which suggested that cough would precede headache and myalgias.(13) Anosmia was common and is 252 likely specific for COVID-19(14), but was most prominent one week post symptom onset. Thus, patients 253 may be shedding infectious virus for up to one week prior to the onset of "typical" COVID-19 symptoms 254 such as cough, shortness of breath, and anosmia. 255 256 Only 3 participants had fever >100.4 F at enrollment, and 6 had fever at any timepoint during follow-257 up, consistent with other studies which show relative lack of fever in mild/moderate cases compared 258 with severe/critical cases.(3-7) However, overall temperature did decrease in the first week after 259 enrollment, after which temperatures remained stable at a mean temperature of 97.6°F. These 260 observations are consistent with recent reports suggesting that "normal" body temperatures are well 261 below the established "normal" of 98.6°F,(15, 16) and suggest that temperatures above 97.6°F may 262 represent clinically significant physiologic changes within the context of uncomplicated COVID-19. 263 264 LatinX participants were significantly more likely to have symptoms resolve at any point during the study 265 compared to non-LatinX, white participants. This was after controlling for key demographic/clinical 266 characteristics, including age, gender, and BMI. Although it is possible that biologic differences may 267 underlie these differences in symptom resolution, prior reports have suggested that race or ethnic-268 Myalgia and joint pain were the only symptoms significantly associated with oropharyngeal PCR 281 positivity. These are also some of the earliest symptoms to appear, when viral loads are higher early in 282 disease course and patients are most likely to be infectious. Therefore, screening for respiratory 283 symptoms may miss the most infectious patients, who may attribute their symptoms to other causes. 284 285 There were limitations to this study. This cohort was part of a randomized, placebo-controlled trial, 286

where half the cohort received Lambda and the other half received placebo. The duration of symptoms 287 and viral shedding was similar between the two arms in the primary analysis(8), so we included the 288 entire cohort in order to improve statistical power to perform our analyses. In this analysis of individual 289 symptoms, there were some symptoms that lasted longer in the Lambda vs. placebo arm, although 290 adjustment for treatment arm did not significantly alter our time to event models. Only 4 (3%) 291 participants in this cohort needed hospitalization and 10 (8.3%) went to the ED for evaluation, compared 292 with an incidence of hospitalization of 2.9-11% in other outpatient trials.(7, 23) This is likely related to 293 study selection criteria, and that recruitment occurred during non-surge conditions when even 294 moderately ill patients were being hospitalized. Median participant age was 36 years, and the cohort 295 was predominantly LatinX, reflecting our local population of COVID-19 cases(24); however, our findings 296 may not be generalizable to other settings, including those with other circulating SARS-CoV-2 variants 297 and/or post-vaccination infections. Though there was high retention in the study, 34 participants missed 298 two consecutive daily symptom surveys and were excluded from the symptom cluster analysis. Though 299 we collected symptom data daily for 28 days after enrollment, time of symptom onset was reported 300 retrospectively by participants at enrollment and thus subject to recall bias. Finally, this is a secondary 301 analysis of data from a clinical trial that was not designed for these analyses, so this study should be 302 considered hypothesis-generating. 303

In this cohort of participants with uncomplicated COVID-19, distinct symptom clusters and trajectories 305 were identified and inflammatory symptoms were associated with ongoing viral replication. In contrast, 306 cough, loss of taste/smell, and elevated temperature, which are often used for screening programs, are 307 not always earliest to present and not associated with viral replication, and may therefore not be ideal 308 for screening approaches. Data from this and other longitudinal studies will be critical in shaping our 309 understanding of disease pathogenesis in patients, and for planning large, Phase 3 outpatient studies 310 based on clinical and virologic endpoints. 311

We would like to thank the Lambda study participants, the entire study team and the Clinical and 313 of runny nose and sore throat. Cluster B (teal, n=7 participants) characterized by greater symptom 423 severity and/or later peak severity (chest pain/pressure, fatigue, and myalgias.) Y-axis represents the 424 severity of the symptom (0=Not reported; 1=Mild; 2=Mild to Moderate; 3=Moderate; 4=Moderate to 425 Severe; 5=Severe for the scores ranging from 0-5 and 0=Not reported to 1=Reported for scores ranging 426 from 0-1) while the x-axis represents days from symptom onset. Participants with more than 2 427 consecutive days of missing symptom data (n=39) were excluded. 428 

Faroe Islands -a longitudinal study among non-hospitalized patients

Clinical Features of 69 Cases With Coronavirus Disease 341 2019 in Wuhan, China

Characteristics and 343 Outcomes of Coronavirus Disease Patients under Nonsurge Conditions

Temporal profiles of viral load in 346 posterior oropharyngeal saliva samples and serum antibody responses during infection by SARS-CoV-2: 347 an observational cohort study

Clinical characteristics of asymptomatic and 349 symptomatic patients with mild COVID-19

Hydroxychloroquine in Nonhospitalized Adults With Early COVID-19 : A Randomized Trial

Peginterferon Lambda-1a for treatment of outpatients with uncomplicated COVID-19: a randomized 355 placebo-controlled trial

R: A language and environment for statistical computing. Vienna, 357 Austria: R Foundation for Statistical Computing

A simple test of the proportional hazards assumption

Clinical Characteristics of Coronavirus 361 Disease 2019 in China

Signs and symptoms 363 to determine if a patient presenting in primary care or hospital outpatient settings has COVID-19 364 disease

Modeling the Onset of Symptoms of COVID-366 19

368 Symptoms and symptom clusters associated with SARS-CoV-2 infection in community-based 369 populations: Results from a statewide epidemiological study

Decreasing human body temperature in the 371 United States since the industrial revolution

Individual differences in normal body temperature: 373 longitudinal big data analysis of patient records

Presenting Symptoms and Outcomes Differ by Race/Ethnicity Among Hospitalized Patients with COVID-376 19 Infection? Experience in Massachusetts

Racial/Ethnic Disparities in Hospitalization and Mortality in Patients With COVID-19 in

Viral dynamics in the Upper 381 Respiratory Tract (URT) of SARS-CoV-2

Centers for Disease Control and Prevention. Isolate If You Are Sick: National Center for 385 Immunization and Respiratory Diseases (NCIRD), Division of Viral Diseases

Kinetics and Innate and Adaptive Immune Response Assessment in COVID-19: a Case Series. mSphere

Febuxostat therapy in outpatients with suspected COVID-19: A clinical trial

396 397 treatment arm, and time from first positive result. Odds ratios (ORs) in each row

Color represents the severity of the symptom reported. 'Unknown severity' includes both symptoms in 407 which severity was not asked and symptoms reported as starting prior to enrollment. B) Summary of the 408 peak day of which each symptom was reported with corresponding percentage (%) of participants 409 reporting the symptom on the peak day, and the number and percentage of participants reporting each 410 symptom at least once during the study period. For symptoms present prior to enrollment, severity 411 data was not collected. Temperature (C) and oxygen saturation (D) was measured daily at home using 412 study-provided equipment. Shown are mean temperature and mean percent oxygen saturation with 413 95% confidence intervals (marginal estimates from GEE models) on each day after enrollment. 414