SARS-CoV-2 seroprevalence screening study of a children’s hospital health care workers

Abstract

Background

Health care workers (HCWs) are front line responders to the COVID-19 pandemic, but limited data is available for pediatric HCWs, as the research response has largely focused on adult patients and medical personnel that treat these patients.

Methods

We conducted a cross-sectional study of SARS-CoV-2 seroprevalence and risk factors in HCWs at a Children’s Hospital in CO, USA from September 2020 to April 2021. Pediatric HCWs were defined as clinical care providers and administrative staff. Seroprevalence was determined using the Epitope SARS-CoV-2 anti-Nucleocapsid IgG assay (San Diego, CA) and the Euroimmun SARS-CoV-2 anti-Spike Protein IgG assay. Risk factors and vaccination status were assessed via questionnaire.

Results

Overall, 110 HCWs were enrolled, 79 subjects were positive for anti-S antibodies and negative for anti-N antibodies, indicating COVID-19 vaccination. 31 subjects had neither anti-N or anti-S antibodies, indicating no exposure to SARS-CoV-2 and no vaccination. 3/110 had a nucleocapsid serology consistent with a SARS-CoV-2 prior infection. Seroprevalence was observed at 2.7%. It was noted that asthma requiring medication was associated with positive serostatus.

Conclusions

During the winter 2020/21, SARS-CoV-2, we found a 2.7% seroprevalence of pediatric HCW at a children’s hospital in Colorado. We compared this with publicly available seroprevalence data for seroprevalence rates of pediatric HCWs globally. This suggests that this specific children’s hospital COVID-19 personal protective equipment (PPE) and infection control guidelines were effective in limiting SARS-CoV-2 in hospital transmission at the children’s hospital prior to the presence of the Delta variant.

Introduction

Health care workers (HCWs) are in the front line of the global response to COVID-19. HCWs have been found to be at an increased risk of becoming infected while caring for COVID-19 patients and being in close contact with infected asymptomatic co-workers and family members of patients (Buonafine et al., 2020; Galanis et al., 2021; Gómez-Ochoa et al., 2021; Mostafa et al., 2021; Xiao et al., 2020). The frequency of asymptomatic COVID-19 has varied with the emergence and spread of SARS-CoV-2 variants. As well as the subsequent changes in the immunological state of susceptible individuals, extent and timing of vaccination, and population prior infection rate (Gandhi et al., 2020; Hall et al., 2021; Johansson et al., 2021; North et al., 2021; Tang et al., 2021). Initially, many seroprevalence studies focused on HCWs, mostly in general and adult hospital facilities due to COVID-19 hospitalizations disproportionately affecting adults, especially the elderly (Gandhi et al., 2020; Hall et al., 2021; Johansson et al., 2021). However, the focus shifted over the course of the pandemic to include children and adolescents. School and other community-based child and adolescent surveillance efforts became crucial, in part, because of the importance of keeping schools open for the functioning of society. Fortunately, the American Academy of Pediatrics recognized the lack of attention to pediatric COVID-19 infections and hospitalizations early and began collecting and publishing this data (American Academy of Pediatrics, 2020). As a result, it became evident that the number of children being infected increased with the surges of both the Delta and Omicron variants and as cases of Multisystem Inflammatory Disease (MID) and myocarditis became more widely recognized (Butt et al., 2022; Marks et al., 2022; Moreira et al., 2021; Most et al., 2021). High-risk children included those that were unvaccinated and had complex medical conditions. Their infections increased the risk of exposure for HCWs in pediatric hospital settings. In response, the infection control policies and procedures and the availability and correct use of personal protective equipment (PPE) became essential to protect HCWs and other staff from becoming infected at children’s hospitals. Furthermore, given the high prevalence of non-specific signs and symptoms and the lack and variation of common symptoms in our investigation, increased vigilance, innovative screening, and frequent testing is suggested among school-going children and their immediate contacts.

The initial phase of this study prospectively collected epidemiological and clinical data from a variety of health care workers at a children’s hospital in Colorado, including self-reported exposure to patients with the SARS-CoV-2 virus. In Phase 2 of the study, we intend to collect a follow-up sample approximately 12 months after the initial sample, however, due to the low enrollment have suspended efforts.

We hypothesized that higher-risk exposures and/or other factors such as co-morbid conditions would be associated in the initial sample with higher seroprevalence or incident SARS-CoV-2 infections. We explored the seroprevalence found in pediatric health care workers at a children’s hospital in Colorado with other pediatric health care workers globally, as well as with the background seroprevalence of the general population in the state of Colorado.

Materials and methods

Study population

110 HCWs employed at a children’s hospital in Colorado were enrolled in this study from September 2020 to April 2021. The study was carried out in the biggest pediatric hospital in the Rocky Mountain Region. The Number of beds is 427 and maximum tertiary care is provided, including BMT. The number of ICU beds is 56. The HCWs were defined broadly to include clinical care providers and administrative staff, including but not limited to physicians, advanced practitioners (nurses and physician assistants), nurses, respiratory therapists, physical and occupational therapists (PT/OT), medical assistants, and other staff in the healthcare facility (for example, administrative staff, food service workers, and cleaners). Whether or not the HCW was directly exposed to patients with the SARS-CoV-2 virus was determined via questionnaire.

Sample size

The initial sample size calculated was 500 participants under the assumption of a seroprevalence of around 5%. However, enrollment was delayed and slower than anticipated due to pandemic-related delays out of the studies control. At the point of this analysis, 110 participants were enrolled and given the low seroprevalence, and enrollment was suspended.

Recruitment, enrollment, and consent

We did not allow any active outreach to the faculty or hospital staff to ensure that HCWs would not be coerced to participate in the study. Therefore, this study represents a self-selected sample of eligible HCWs who volunteered to participate after the information about the study was posted on the hospital COVID-19 research website. HCWs who did not have current signs and symptoms of acute infection were eligible for enrollment. HCWs with evidence of current infection were asked to wait until at least 2 weeks after their illness has resolved. Ethical approval was obtained by the Colorado Multiple Institutional Review Board (Protocol #20-1903).

Questionnaires

Subjects were consented in person at the time of their blood draw and completed an on-line REDCap questionnaire on demographics, contact information, occupation, environmental exposures (e.g., smoking), COVID-19 exposures, protective measures, mental health, medical history, and COVID-19 vaccination history.

Ten questions were selected from the Pittsburgh Sleep Quality Index, Trauma Addendum (see supplemental file for full questionnaire) (Germain et al., 2005). These well validated items focused on sleep, stress, and functioning.

Blood sample collection, testing, and reporting

Anti-N and anti-S antibody analysis and interpretation

Serological testing was conducted against the SARS-CoV-2 IgG nucleocapsid (N) antigen and the Spike (S) antigen, using commercially available tests (Epitope SARS-CoV-2 IgG assay, EDI (San Diego, CA) that used the nucleocapsid antigen as the target [specificity of 99%]; Euroimmun SARS-CoV-2 IgG assay (Lubeck, Germany) that uses the S1 subunit of the spike antigen [specificity of 97.2%]). Positive, negative, and borderline anti-N results were reported based on the average of triplicate optical density (OD₄₅₀) values for the negative calibrator (xNC). The positive cut-off for each assay was calculated as 1.1*(xNC + 0.18) and the negative cut-off value as 0.9*(xNC + 0.18). Results greater than the positive cut-off value (generally OD₄₅₀ of 0.3 and greater) were reported as “positive,” those below the negative cut-off values were reported “negative,” and those in between the positive and negative cut-off values were reported as “borderline.” Anti-S antibody results were interpreted based on the ratio of sample OD₄₅₀ to the calibrator OD₄₅₀. Samples with ratios ≥1.1 were reported “positive,” ratios ≤0.8 were reported “negative” and those in between 0.8 and 1.1 were reported as “borderline.”

Serology result reporting

Whole blood (6 mL) was collected from each HCW. Serum was separated and stored at 4°C until analysis. Serological testing was conducted against the SARS-CoV-2 IgG nucleocapsid (N) antigen and the Spike (S) antigen, using commercially available tests (Epitope SARS-CoV-2 IgG assay (San Diego, CA) that used the nucleocapsid antigen as the target [specificity of 99%]; Euroimmun SARS-CoV-2 IgG assay (Lubeck, Germany) that uses the S1 subunit of the spike antigen [specificity of 97.2%]). Qualitative results (positive, negative, or borderline) were reported to the subjects as a formal laboratory report using CDC, FDA, and other regulatory agency required language. We also included an explanation as to the up-to-date diagnostic accuracy of the test, and caveats that false negatives and false positivity occur, and that a positive antibody test is not known to confer immunity against future infection. Vaccinated subjects who were not previously infected with SARS-CoV-2 were expected to be negative for anti-N antibodies and positive for anti-S antibodies, whereas subjects who had recovered from COVID-19 were likely to have antibodies to both S and N antigens. This information was included on the SARS-CoV-2 antibody test clinical report. Subjects were provided with contact information to reach out to study personnel if they have additional questions on their results. Positive tests were reported to governmental authorities in keeping with regulations. For our purposes, a positive serological result was based on the result of the Nucleocapsid ELISA results only.

Statistical analysis

Descriptive statistics were carried out using SAS (SAS Institute Inc, Cary, NC). We conducted a statistical comparison of demographic and self-reported risk factors between those who tested positive for SARS-CoV-2, and those that tested negative for SARS-CoV-2. The prevalence of anti-SARS-CoV2 N-specific IgG antibodies was compared by characteristics of the participants (e.g., demographics, comorbidities, exposure categories such as occupation and type of work exposures).t test for continuous variables and Fisher’s Exact tests for categorical variables were used for descriptive statistics.

Retrieving publicly available seroprevalence data for comparison of pediatric healthcare workers globally

We retrieved publicly available seroprevalence data from the SeroTracker living dataset (Arora et al., 2021). There were 3323 cohort data points in the database as of April 25, 2022. We applied an inclusion criterion of all data points with a sample frame (groups of interest) of health care workers and caregivers. We explicitly excluded all other sample frames, including blood donors, household and community samples, essential non-healthcare workers, childcare facilities, and residual sera. This produced a comprehensive list consisting of 724 cohort data points from 634 individual studies. For each of the 634 studies that met the inclusion criteria, we reviewed the associated publication to assess whether the paper focused on a pediatric healthcare facility and was published in English. This was conducted by utilizing a keyword search and find within each publication using terms: children, child, pediatric, and paediatric. Of the 634 studies, there are 40 pediatric cohort data points from 32 individual studies that were included in the analysis. The included pediatric studies include healthcare worker seroprevalence data from pediatric hospitals, pediatric-specific departments including the pediatric ICU, and a pediatric dialysis unit.

Retrieving publicly available seroprevalence data for comparison in Colorado

Using publicly available seroprevalence data from the SeroTracker living dataset (Arora et al., 2021), we applied an inclusion criterion of all data points with any sample frame (groups of interest) as well as a subset of location for Colorado, USA (see supplemental file for comprehensive list). We explicitly excluded all other points of geography as well as any studies that were not published in English. This produced a comprehensive list consisting of 19 cohort data points from 12 individual studies as of April 25, 2022.

Results

Overall, our HCW study population included 44 (41.4%) nurses, 33 (30.8%) physicians, and 11 (10%) nurse practitioners/physician assistants (Table 1). Among these participants, one nurse, one physician, and one nurse practitioner (3/110) had a nucleocapsid serology consistent with a SARS-CoV-2 prior infection. There were 19 additional types of healthcare workers that included PT/OT, technicians, dieticians, medical assistants, and administrative staff. None of these workers had serology results consistent with a prior infection.

Table 1.

Demographic characteristics of health care worker participants at a children’s hospital by SARS-CoV-2 serology results.

Demographic variable name	SARS-CoV-2 negative (N = 107)	SARS-CoV-2 positive (N = 3)	p-value*
	Mean (SD) or N (%)	Mean (SD) or N (%)
Age	39.6 (10.7)	44.0 (6.9)	0.48
Gender = Female	91 (86.7%)	3 (100%)	0.99
Race / Ethnicity			0.21
White	98 (93.3%)	2 (66.7%)
Asian	4 (3.8%)	0 (0%)
Other/Unknown	3 (2.9%)	1 (33.3%)
Ethnicity = Hispanic	5 (5.0%)	0 (0%)	0.99
Housing type			0.58
Apartment	13 (12.4%)	0 (0%)
Condo/Duplex/Triplex	12 (11.4%)	1 (33.3%)
Single family home	79 (75.2%)	2 (66.7%)
Other	1 (1.0%)	0 (0%)
# people who live with you	2.3 (1.2)	2.5 (0.7)	0.80
Health care occupation detail:			0.59
Diagnostic medical sonographer	1 (1.0%)	0 (0%)
Dietician	2 (1.9%)	0 (0%)
Nurse practitioner	6 (5.8%)	1 (33.3%)
Other health technician	1 (1.0%)	0 (0%)
Other health care staff with direct patient contact	9 (8.7%)	0 (0%)
Other non-medical staff with direct patient contact	1 (1.0%)	0 (0%)
Other non-medical staff with no patient contact	1 (1.0%)	0 (0%)
PT/OT	2 (1.9%)	0 (0%)
Physician	32 (30.8%)	1 (33.3%)
Physician’s assistant	4 (3.9%)	0 (0%)
Radiology technician	2 (1.9%)	0 (0%)
Registered nurse	43 (41.4%)	1 (33.3%)
Cared for COVID-19 patients	81	2	0.54
Aerosol generating procedures?	1 (20%)	0 (0%)	n/c

*p-value for t-test for continuous variables and Fisher’s exact test for categorical variables, as appropriate.

n/c = not calculable.

We have investigated the results stratified by occupation, demographic characteristics, potential behavioral risk factors, and mental health stressors (Table 1). The overwhelming majority of participants were women (94/110), three of whom had SARS-CoV-2 nucleocapsid serology results consistent with a prior infection.

One of four non-White non-Asian non-Hispanic female participants had a result consistent with a prior infection. Overall, race/ethnicity was not statistically significantly associated with SARS-CoV-2 seropositivity in our dataset. We also categorized ethnicity to a binary Hispanic versus non-Hispanic status and again did not find any statistically significant association.

We could not corroborate any association of housing density with positive serostatus. Two of the three participants with evidence of a prior infection occurred in the 81% of the participants living in a single-family home with a mean of 2.3 people living together with the participant.

The low number of participants with serological results consistent with a prior infection greatly limits our ability to assess the impact of behavioral risk factors and pre-existing medical conditions on the likelihood of a positive serology. However, despite the small number of participants with evidence of prior infection, asthma requiring medication was associated with positive serostatus infections (Table 2). There were three women who indicated that they were pregnant, none of whom had evidence of a prior infection. Only one participant identified as a current smoker and 89 indicated that they exercise regularly at least 4 days a week.

Table 2.

Health history and behavioral risk factors of participating health care workers at a children’s hospital.

Health history and behavioral risk factors variable name	SARS-CoV-2 negative (N = 107)	SARS-CoV-2 positive (N = 3)	p-value*
	Mean (SD) or N (%)	Mean (SD) or N (%)
Cancer	6 (5.7%)	0 (0%)	n/c
Diabetes	3 (3.3%)	0 (0%)	n/c
Hypertension	8 (7.6%)	0 (0%)	n/c
Asthma requiring medication	22 (21.0%)	2 (66.7%)	0.03
Allergic rhinitis	29 (27.6%)	2 (66.7%)	0.11
Obstructive sleep apnea	1 (1.0%)	0 (0%)	n/c
Autoimmune or rheumatologic disease	7 (6.7%)	1 (33.3%)	0.23
Rheumatoid arthritis	1 (14.3%)	0 (0%)	n/c
Ankylosing spondylitis	1 (14.3%)	0 (0%)	n/c
Inflammatory bowel disease	2 (1.9%)	1 (33.3%)	0.11
Chronic blood disorder	1 (1.0%)	0 (0%)	n/c
Chronic neurological impairment	0 (0%)	1 (33.3%)	n/c
Other pre-existing medical condition	14 (13.3%)	0 (0%)	0.99

*p-value for t-test for continuous variables and Fisher’s exact test for categorical variables, as appropriate. n/c = not calculable.

Serology results

Of the 110 subjects included in the study, only three were positive for anti-N antibodies, indicating previous infection. Seventy-nine subjects were positive for anti-S antibodies and negative for anti-N antibodies, indicating a response to COVID-19 vaccination. Thirty-one subjects had neither anti-N or anti-S antibodies, indicating that they had not been exposed to SARS-CoV-2 and had not been vaccinated (Figure 1).

Figure 1.

Scatterplot of Anti-N and Anti-S SARS-CoV-2 antibodies (dotted lines represent cutoff thresholds).

We compared the seroprevalence obtained in the HCWs from the children’s hospital in Colorado with (a) population estimates within Colorado, and (b) pediatric HCWs globally (see supplemental file for comprehensive list). Other Colorado-based estimates are in line with the figures we present from our study at the children’s hospital in Colorado. Figure 2(a) presents the seroprevalence rates from Colorado-based seroprevalence studies, including six other sampling frames (groups of interest) besides HCWs. Most of these studies estimate the seroprevalence around or under 8%, except for three studies. Two were conducted in persons experiencing homelessness in Denver (N = 334 and 271) with a seroprevalence of 17 and 16%, respectively (McCormick et al., 2020; Rowan et al., 2022). One study, conducted using first responders (essential non-healthcare workers sampling frame), had a sample size of 783 and a seroprevalence of 20% (Montague et al., 2022).

Figure 2.

(a) Seroprevalence rates among various demographic groups in colorado-based seroprevalence studies (b) seroprevalence data among pediatric healthcare workers from International and United States settings.

We compared the seroprevalence obtained in the HCWs from the children’s hospital in Colorado with pediatric HCWs nationally in the US and with international pediatric facilities. Figure 2(b) presents seroprevalence data from both international and United States pediatric HCWs. Most of the studies included a similar sampling frame within a varying 4-month window and resulted in a seroprevalence estimate of under 20% apart from a seroprevalence reported out of Wuhan, China, amounting to 43% (sample size N = 325) (Tu et al., 2021). We discovered substantial heterogeneity between the studies with a wide range of sample sizes (34–4927). Other studies with higher seroprevalence estimates among pediatric HCWs included an estimate of 35% in pediatric dialysis unit in Indiana, United States (sample size N = 35) (Canas et al., 2021) and a seroprevalence of 30% in children’s hospital of Tehran, Iran (sample size N = 475) (Armin et al., 2020) (Figure 2(b)).

Discussion

The 2.7% COVID-19 seroprevalence estimate at children’s hospital in Colorado is considered low, and thus more representative of the general population in Colorado than indicating any additional risk in the hospital during the time frame. In a surveillance study conducted in 264 first responders in Arapahoe County in 2020, 4.2% of the participants were seropositive (Preti et al., 2020; Sabourin et al., 2021). Overall, seroprevalence in Colorado showed some variability due to different testing methods, sampling frames and populations, ranging from 0.4% to 20%, but below 8% in the general population as mentioned above.

The assessment of the effectiveness of infection control protocols and use of PPE in Children’s Hospitals is paramount to prevent the transmission of SARS-CoV-2 within clinical settings. The low seroprevalence in HCWs at a children’s hospital in Colorado speaks to the success of infection control within the hospital setting prior to the introduction and spread of the more transmissible Delta and Omicron variants. See supplemental file (CDPHE, 2022) which displays the graphed variant data for Colorado as well as case rates (per 100k) for the combined Denver metro counties. Only 2 out of 83 HCWs who directly cared for COVID-19 patients in the past month showed evidence of SARS-CoV-2 infection. In addition, we did not find evidence that individuals with higher-risk exposures were more likely to show evidence of a SARS-CoV-2 infection. We intend to obtain follow-up blood samples following a booster vaccine to document asymptomatic and symptomatic breakthrough infections.

The number of pre-existing conditions reported by the participants reinforces that a significant number of healthcare workers are at an increased risk for SARS-CoV-2. Asthma requiring medications was the only comorbidity that was statistically significantly associated with positive serostatus (p-value 0.03, see Table 2), supporting findings from an earlier study in Colorado among first responders (Sabourin et al., 2021). Overall, comorbidities were relatively common among HCWs surveyed at a children’s hospital in Colorado, highlighting the importance of inclusion and further assessment of these potential risk factors during the COVID-19 pandemic.

The COVID-19 pandemic has impacted the fears and concerns of HCWs in the workplace for their personal safety as well as their families, directly affecting household members and interpersonal relationships. It is well documented the psychosocial impact of past epidemic/pandemic outbreaks on HCWs, further supporting this concern (Preti et al., 2020).

There is a scarcity of studies on pediatric HCWs among COVID-19 HCW seroprevalence studies in general. Out of the 3323 cohort data points as of April 25, 2022, from the Serotracker database (Arora et al., 2021), only 40 pediatric data points of seroprevalence estimates were identified (see supplemental file for comprehensive list). This represents a lack of pediatric HCW study data globally.

This study has several limitations. The sample size and the low seroprevalence did not allow for any multivariable regression analysis to thoroughly explore risk factors for SARS-CoV-2 infection in this setting. Since participation was voluntary and participants self-selected to enroll into the study (because only minimal outreach was allowed by the hospital), the study population may not be representative of the HCWs at a children’s hospital in Colorado. This may have resulted in an overestimate or underestimate of the true seroprevalence as the participants might have engaged in high-risk behavior and then select to be tested or be very careful from the start.

We decided to stop enrollment when the majority of HCWs became vaccinated. The most likely selection bias would be that HCWs with higher perceived exposure (e.g., providing direct care to COVID-19 patients) would be more likely to participate. Therefore, the current estimate is likely to be an overestimate, which again highlights the fact that a) seroprevalence is low (as of May 2021) and that community transmission rather than transmission in hospital is driving the epidemic.

The low seroprevalence did not allow further investigation of clustering of infections by occupation. Similarly, the analysis of demographic factors associated with SARS-CoV-2 positivity was limited by the fact that a large majority of the study population were white females, most of whom live in single-family homes. A larger and more diverse study population may have demonstrated a higher seroprevalence in the non-white HCW population.

Conclusions

Seroprevalence of pediatric HCWs at a children’s hospital in Colorado was low in the winter 2020/21, and comparable to the estimates from the surrounding community. This suggests that the COVID-19 PPE of the hospital included in the study and infection control guidelines were effective in limiting SARS-CoV-2 in hospital transmission. Importantly, granular seroprevalence data is needed to evaluate the evolving risk in health care workers as well as general population burden and immunity over time.

ORCID iD

Olivia Pluss https://orcid.org/0000-0003-1978-6261