Symptoms and Transmission of SARS-CoV-2 Among Children — Utah and Wisconsin, March–May 2020

In collaboration with state and local health departments and in the setting of a public health emergency response, the US Centers for Disease Control and Prevention (CDC) conducted a household transmission investigation during March to May 2020 in Milwaukee, Wisconsin, and Salt Lake City, Utah, metropolitan areas. Methods from the full investigation are available at Lewis et al¹¹ ; here, we describe a secondary analysis of the overall household transmission investigation, focused on details related to children in the households. In brief, we approached and enrolled a convenience sample of index case patients with laboratory-confirmed SARS-CoV-2 and their household contacts. Index case patients were identified by local health departments through routine surveillance and contact tracing activities. After completing their own investigation of the reported case, local health department staff described the CDC investigation protocol, asked if the index case patient and household members were interested in participating, and, if so, referred them to the CDC team for study enrollment. Households were eligible for enrollment if the index case patient was within 10 days of his or her first positive test result for SARS-CoV-2 by real-time reverse transcription polymerase chain reaction (RT-PCR) and was at home during enrollment (ie, not hospitalized), with at least one additional household member who was willing to participate. All enrolled index case patients and household contacts were followed prospectively for 14 days.

Index case patients and household contacts were interviewed by using standard questionnaires that captured demographics, medical history, previous SARS-CoV-2 testing, and symptoms since the index case patient’s illness onset (Supplemental Information). At study enrollment (day 0), the CDC team visited the household to collect blood and respiratory specimens (nasopharyngeal and self-collected anterior nasal swabs). If a household contact developed new or worsening symptoms, the CDC team conducted an interim visit to collect respiratory specimens from all household members. Five households were selected for intensive swabbing requiring collection of respiratory specimens from all household members during four interim visits regardless of symptom presence. At study close-out (day 14), the CDC team returned to collect blood and respiratory specimens. During the 14-day follow-up, participants completed daily symptom logs, which included temperature checks with study-provided digital thermometers; a parent or guardian used discretion to determine if child(ren) needed assistance.

The City of Milwaukee Health Department Laboratory and the Utah Public Health Laboratory tested respiratory specimens using the CDC 2019-nCoV real-time RT-PCR assay¹²  and processed and shipped blood specimens to the CDC for serology testing using the CDC SARS-CoV-2 spike protein enzyme-linked immunosorbent assay (ELISA).¹³ 

We classified household members as either primary patient or household contact; the primary patient was the household member with earliest symptom onset (and positive SARS-CoV-2 RT-PCR test result). In most (89%) households, the primary patient was the same as the index case patient first identified by the health department. To assess transmission patterns, we excluded households with coprimary patients (with symptom onset ≤2 days of each other), so that each household had one source individual. We categorized household members <18 years as children and classified adult and pediatric household contacts as adult case patients and pediatric case patients, respectively, if they tested positive for SARS-CoV-2 by RT-PCR or had total anti–SARS-CoV-2 antibody titers ≥100 by ELISA at any point during the study period. Given the low community prevalence early in the pandemic and lag time between exposure and study enrollment, seropositivity was used as a marker of recent acute infection.

We described demographics of household contacts, stratified by age. To account for household clustering, we used generalized estimating equations (GEEs) with an independent correlation matrix and logit link to compare secondary infection rates between pediatric and adult contacts. Among pediatric contacts, we calculated unadjusted odds ratios (ORs) and 95% confidence intervals (CIs) using GEEs to assess risk factors for infection. We calculated the serial interval as the number of days between symptom onset of the primary patient and associated secondary case patient.

Among pediatric case patients, we described the timing of RT-PCR and serology results in relation to primary patient and individual symptom onset. To assess the possibility of children transmitting SARS-CoV-2 within households, we calculated the difference in symptom onset in days between all case patients in the same household. On the basis of expected minimum incubation periods, we considered a child as a possible source of transmission if a subsequent case patient developed symptoms >2 days later.^14,15  Community prevalence in these two metropolitan areas was low during this time, and both were under stay-at-home orders, which started soon after study initiation and extended throughout study enrollment, reducing the likelihood of outside exposures.

We categorized symptoms as constitutional (measured or subjective fever, chills, myalgia, and fatigue), upper respiratory (runny nose, nasal congestion, and sore throat), lower respiratory (cough, difficulty breathing, shortness of breath, wheezing, and chest pain), neurologic (headache, loss of taste, and loss of smell), and gastrointestinal (nausea and/or vomiting, diarrhea, and abdominal pain). Among pediatric case patients, we described the frequencies of reporting symptoms and meeting COVID-19 and syndromic surveillance case definitions (coronavirus disease 2019–like illness [CLI]¹⁵ ; Council of State and Territorial Epidemiologists [CSTE] COVID-19¹⁶ ; influenza-like illness [ILI]¹⁷ ; World Health Organization [WHO] acute respiratory infection [ARI]¹⁸ ) and calculated positive predictive values (PPVs), negative predictive values (NPVs), and 95% CIs. We used GEEs to compare the symptom frequency and duration between pediatric and adult case patients; those who remained symptomatic at the end of the study period were treated statistically as if their symptoms resolved on day 14.

This protocol was reviewed by CDC and deemed nonhuman subjects research as part of the COVID-19 public health response. All participants provided written consent (or child assent with parental permission) before participation.

Individuals from 62 households were enrolled; four households with coprimary patients were excluded (Fig 1). Among the 58 households included (UT: n = 34; WI: n = 24), children resided in 33 (57%). One household had a pediatric primary patient. In total, 188 household contacts were enrolled and included; 68 (36%) were children (Table 1). Of 120 adult contacts, 65 (54%) resided in households with children.

Among 68 pediatric contacts, 37 (54%) were female, 34 (50%) were between 5 and 12 years, and 43 (60%) were non-Hispanic white (Table 1). Nine (13%) children had an underlying medical condition, with asthma the most commonly reported. Most (63%) pediatric contacts were children of the primary patient, whereas 21% were siblings of the (adult) primary patient. Among adult contacts, 74% were between 18 and 49 years, and 49 (41%) had an underlying medical condition. A full description of household characteristics and mitigation measures in the household is available at Lewis et al.¹¹ 

Among 68 pediatric contacts, 19 (28%) tested positive for SARS-CoV-2 by RT-PCR or ELISA. Among 120 adult contacts, 36 (30%) tested positive, including 18 of 65 (28%) in households with children and 18 of 55 (33%) in households without childrenunder (Fig 2). There were no significant differences in secondary infection rates between adult and pediatric contacts among all households (OR: 1.11; 95% CI: 0.56 to 2.21) or among households with children (OR: 0.99; 95% CI: 0.51 to 1.90).

Among the 19 pediatric case patients, the median age was 13 years (range: 3‒17; interquartile range [IQR]:10‒15 years), 12 (63%) were female, and 14 (74%) were non-Hispanic white. Children of primary patients had increased odds of acquiring infection compared with children in households in which the primary patient was not their parent (OR: 17.28; 95% CI: 2.36 to 126.8). There were no significant differences in the odds of acquiring infection by age, sex, race–ethnicity, underlying medical conditions, number of household members, household size, or ratio of children to bedrooms and bathrooms. No children were hospitalized, and none died. The median serial interval among pediatric case patients was 5 days (IQR: 4‒9 days), and the median symptom duration was 10 days (IQR: 6‒14 days), although 2 children were still experiencing symptoms at the end of the study period (Table 2).

In Fig 3, we display serial intervals and transmission patterns in households with children. There was potential for transmission from children in 12 households, ie, there were susceptible household contacts >2 days after illness onset of the first infected child in the house (Fig 3A; households A‒F and I‒N). Of 10 households with potential for child-to-adult transmission, 2 (20%; households K and M) represent possible transmission from child to adult. Of 6 households with potential for child-to-child transmission, 1 (17%; household L) represents possible transmission between children. In household L, the first infected child (girl; 14 years) was a caretaker for her infected mother and may have transmitted SARS-CoV-2 to her sibling (girl; 10 years). In households with no pediatric case patients (Fig 3B), 9 of 39 (23%) adult contacts tested positive for SARS-CoV-2.

The single pediatric primary patient had an respiratory specimen positive for SARS-CoV-2 by RT-PCR collected before study enrollment and did not seroconvert during the study period (Fig 4; child A1).

Among the 19 pediatric case patients, 15 (79%) had respiratory specimens positive for SARS-CoV-2 by RT-PCR collected before or during the study period (Fig 4). One child was asymptomatic at the time of first positive test by RT-PCR but developed symptoms 2 days later (Fig 4A; child H2); among the remaining 14 pediatric case patients with respiratory specimens positive for SARS-CoV-2 by RT-PCR, the median time between symptom onset and RT-PCR positive test result was 2 days (IQR: 1‒6 days); specimens from 3 of these children were RT-PCR–positive for SARS-CoV-2 on the day of symptom onset. At day 14, 3 (16%) children had repeat RT-PCR SARS-CoV-2–positive specimens collected, indicating that length of RT-PCR positivity was at least 14 days in those children.

Among the 19 pediatric case patients, 17 had at least 1 available serum specimen; all tested positive for SARS-CoV-2 antibodies by ELISA during the study period (Fig 4B). These children had a median of 13 days between symptom onset and seropositivity (IQR: 7‒16 days) and median of 17 days between primary patient symptom onset (ie, exposure) and seropositivity (IQR: 16‒24 days). Sixteen had paired sera; 9 of 16 (56%) revealed evidence of seroconversion during the study period.

Four pediatric case patients tested positive by serology only (Fig 4; children B1, C1, M1, and N1). All 4 displayed COVID-19 symptoms during or shortly before the study period but after the primary patient’s symptom onset. One (child N1) seroconverted during the study period, consistent with recent acute infection. Two children (B1 and C1) had an initial RT-PCR positive test result 16 days after the primary patients’ symptom onset, and symptoms resolved on the day of or 1 day before study enrollment. Child M1 had serum collected on day 14 only, preceded by symptoms consistent with acute infection.

Among the 19 pediatric case patients, all reported at least one symptom. Upper respiratory and neurologic symptoms (both 84%) were the most commonly reported symptom categories (Fig 5). The most commonly reported individual symptoms were headache (79%), sore throat (68%), rhinorrhea (68%), and nasal congestion (63%). Less than one-half of pediatric case patients reported any gastrointestinal symptoms (42%), cough (42%), loss of smell (32%), loss of taste (21%), or measured fever (16%). In general, the majority of symptoms had low PPVs; the highest PPV was for measured fever (PPV: 100%; 95% CI: 44% to 100%), although it was only reported by 3 pediatric case patients (Table 3). The highest NPVs were in neurologic symptoms (NPV: 93%; 95% CI: 81% to 97%), specifically, headache (NPV: 91%; 95% CI: 78% to 96%) and upper respiratory symptoms (NPV: 90%; 95% CI: 75% to 97%). COVID-19 and syndromic surveillance case definitions had low PPVs and high NPVs (CSTE COVID-19 [NPV: 93%; 95% CI: 81% to 98%]; CLI [NPV: 89%; 95% CI: 77% to 95%]; WHO ARI [NPV: 90%; 95% CI: 74% to 97%]).

When comparing pediatric (n = 19) and adult (n = 36) case patients, children had a lower frequency of reporting most symptoms (Fig 6). Pediatric case patients were significantly less likely to report lower respiratory symptoms (OR: 0.17; 95% CI: 0.04 to 0.82), cough (OR: 0.15; 95% CI: 0.04 to 0.57), loss of taste (OR: 0.21; 95% CI: 0.06 to 0.74), and loss of smell (OR: 0.29; 95% CI: 0.09 to 0.96) than adult case patients and more likely to report sore throat (OR: 3.4; 95% CI: 1.04 to 11.18). Pediatric case patients experienced a shorter duration of symptoms, in comparison with adults (median of 10 vs 16 days; β = −6.5; 95% CI: −10.1 to −2.9); this difference may be more pronounced because 2 (11%) children, compared with 14 (39%) adults, remained symptomatic at the end of the study period.

In this investigation, we followed a cohort of pediatric and adult household contacts of individuals with COVID-19 over time to assess their symptoms, determine if they acquired SARS-CoV-2 infection, and describe household transmission patterns. Our investigation is novel in that we describe outpatient children with COVID-19, who otherwise may not have been tested, and compare them with children who did not acquire infection and adults who did. We found that, compared with adults, children acquired infection at similar rates but developed less severe illness. We also found that, among this pediatric population, being the child of a primary patient increased the risk of acquiring infection in the household. Lastly, we observed that children were a possible source of further transmission in approximately one-fifth of households with the potential for such transmission patterns.

At the time of this investigation and until the availability of widespread testing, symptoms consistent with SARS-CoV-2 infection had been the trigger for testing and diagnosis. Our results are similar to other studies,^4,7,9  in that children with COVID-19 often had mild, nonspecific symptoms (sore throat, rhinorrhea, nasal congestion, and headache) that may be challenging to differentiate from other common childhood viral illnesses. In the absence of features commonly associated with COVID-19 in adults (fever, cough, and shortness of breath), children with COVID-19 may not have sought care or did not meet diagnostic testing criteria earlier in the pandemic, thus resulting in underrepresentation in surveillance data. In contrast to the 8.3% of cases reported among children in the United States,³  35% of household cases in our study population were in children, most of whom were identified because we tested all household contacts regardless of symptoms.

As schools and daycares reopen, home- and school-based symptom screening strategies are being used to limit transmission.^19,20  However, with most symptoms and case definitions having low PPVs, symptom screening checklists may either be too broad, resulting in unnecessarily restricting access, or too narrow, resulting in failed recognition of disease. Additionally, in recent studies, it is suggested that asymptomatic children may transmit SARS-CoV-2^4,7,9,21 ; although none of the children who tested positive for SARS-CoV-2 in our study were asymptomatic, 4 were either presymptomatic or on the day of symptom onset at first positive RT-PCR test result, suggesting that they may have been shedding the virus while not experiencing symptoms. Clinicians and schools should maintain a lower threshold for laboratory testing in children if there is an epidemiological link, especially to a parent with suspected or confirmed COVID-19, including in asymptomatic or mildly symptomatic children. Although challenging, the use of syndromic case definitions and/or use of symptoms with high NPVs for COVID-19 or other infectious diseases of concern may still have utility.

An improved understanding of children’s role in the transmission of SARS-CoV-2 is important to better inform prevention recommendations. In studies conducted early in the pandemic, researchers suggested that children were infrequently the primary source of household SARS-CoV-2 infections.^22,23  However, in a more recent study of 153 pediatric index patients, researchers found that older children (10‒19 years old) transmitted SARS-CoV-2 to 18.6% of household contacts,²⁴  and evidence has emerged that children with COVID-19 have similar or higher viral loads compared with adults.^25–27  We were limited in our ability to observe transmission from children, but, among households with the potential for such patterns, we found possible child-to-adult transmission in 20% and possible child-to-child transmission in 17%. Similarly, a case series of 15 households in Chicago, Illinois, suggested both child-to-child and child-to-adult transmission in 13%.²⁸  We found that children acquired infection at the same rate as adults, at ∼30%. Compared to other respiratory viral infections, this is higher than documented secondary transmission rates for Middle East Respiratory Syndrome (4%)²⁹  and pandemic influenza H1N1 (10.3%‒20.2%),³⁰  the latter of which may have had higher attack rates in children than adults. Seasonal influenza secondary attack rates range widely by year, strain type, and age, making direct comparisons difficult.^31–33 

Our investigation was subject to several limitations. We enrolled a small, convenience sample of households in 2 states, which may not be representative of US households; this may result in selection bias and could affect the generalizability of our findings. Among all primary patients, only 1 was a child, which limited our ability to observe transmission from children and may not be reflective of pediatric infection rates in high-transmission areas or when communities and/or schools are open. Although we cannot be certain that the pediatric case patients in our study acquired infection in the household, these communities were under stay-at-home orders for the majority of the study period, reducing the risk of acquiring infection in the community. Our small sample size of children and the resulting wide CIs may have limited our ability to identify differences. The true infection rate among children may have been higher than reported because we were unable to collect serum specimens on all pediatric contacts. Because our study period was only 14 days, some household contacts may have developed additional symptoms or acquired infection after the observation period ended. All symptoms were self-reported or, for some children, reported by their parent(s), which may be subject to recall error, and some symptoms were not systematically asked about for all participants. Because of the high prevalence (28%) of SARS-CoV-2 infection among pediatric contacts, the PPVs and NPVs reported here may not be representative of all settings.

With our investigation, we provide additional information about the symptoms of children with COVID-19, time line of exposure and symptom onset to laboratory-confirmed infection, and household transmission patterns. Additional studies are warranted to examine pediatric transmission patterns outside the home in other environments in which children congregate, such as schools, and in areas with higher incidence than was observed during our study period. In these studies, researchers should include assessments of the effectiveness of mitigation efforts. Finally, additional studies on previous exposures to other coronaviruses and duration of SARS-CoV-2 antibodies may be done to shed light on protective factors against future SARS-CoV-2 infections in children.

In this cohort of exposed household contacts, children acquired SARS-CoV-2 infection at the same rate as adults, and, among pediatric contacts, being the child of an individual with COVID-19 increased the risk of acquiring infection. Children with COVID-19 commonly reported mild symptoms, including headache, sore throat, rhinorrhea, and nasal congestion. Compared with adults, children were less likely to report lower respiratory symptoms, particularly cough, and had a shorter duration of illness. We observed that children were the potential source of further transmission in approximately one-fifth of households with the opportunity to observe such transmission patterns. With these early pediatric data, we provide insight into the risk of COVID-19 infection and resulting symptom profiles of children with mild illness. Further investigations are warranted to better understand SARS-CoV-2 transmission from children both within and outside the household.

We thank the study participants. We also thank: Davis County Health Department (UT): Heather Gibb and Sara Hall; Summit County Health Department (UT): Carolyn Rose; Salt Lake County Health Department (UT): Ilene Risk, Lee Cherie Booth, Jeff Sanchez, Madison Clawson, and Tara Scribellito; North Shore Health Department (WI): Kala Hardy, Christine Cordova, Kevin Rorabeck, and Kathleen Platt; City of Milwaukee Health Department (WI): Clinical, Epidemiology, and Laboratory teams; Wauwatosa Health Department (WI): Laura Conklin, Paige Bernau, and Emily Tianen; Wisconsin Department of Health Services: COVID-19 Response Team; CDC COVID-19 Response: Mayer Antoine, Adebowale Ojo, Brandi Limbago, S. Michele Owens, Wendi Kuhnert-Tallman, Jeff Johnson, Collette Leaumont, Jenny Milucky, Amy Schumacher, Emily Weston, Jessica C. Smith, Michelle Johnson Jones, Jennifer Y. Huang, Erin Moritz, Aubrey Gilliland, Laura Calderwood, Jennifer Imaa, Kaytlin Renfro, Allison Miller, Katie Bantle, Lucia Pawloski, Margaret Williams, Stacy L. Thorne, Jana Manning, Micha Ghertner, Adriane Niare, Yoonjae Kang, and Charlotte D. Kaboré.

ARI

acute respiratory infection

CDC

US Centers for Disease Control and Prevention

CI

confidence interval

CLI

coronavirus disease 2019–like illness

COVID-19

coronavirus disease 2019

CSTE

Council of State and Territorial Epidemiologists

ELISA

enzyme-linked immunosorbent assay

GEE

generalized estimating equation

IQR

interquartile range

NPV

negative predictive value

OR

odds ratio

PPV

positive predictive value

RT-PCR

reverse transcription polymerase chain reaction

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2

WHO

World Health Organization