Introduction

Throughout the COVID-19 pandemic, there has been a paucity of research among school-age children, even though many mitigation strategies disproportionately affect youth (e.g., closing schools but opening non-essential businesses in some settings) [1, 2]. Despite vaccination rolling out among those ≥12 years of age in the fall of 2021, cases among children and adolescents continue to rise, especially in unvaccinated children [3]. The impact of COVID-19 in pediatric populations likely depends on host susceptibility, symptoms, viral load, inoculum, social contact patterns, and behaviors. Studies assessing age-dependent susceptibility to and transmission of COVID-19 have multiple limitations [4]. Research with stored pediatric biospecimens has unique potential to inform key questions about immune responses to COVID-19 infection and inform further research about drivers of the pandemic. In contrast to adult biorepositories, pediatric biorepositories face distinct challenges to consent and thus representative inclusion in research; participants are minors and are considered a vulnerable population, and third-party approval is needed (i.e., from legal guardians) [5]. Enrolling participants that are representative of the pediatric population is essential for the validity, reliability, and interpretation of studies using collected samples. Most research efforts have focused on the effects of the SARS-CoV-2 virus on adult and elderly populations experiencing higher rates of severe illness and mortality [6, 7]. However, pediatric populations are susceptible to severe complications of COVID-19 including multisystem inflammatory syndrome and death, as well as morbidity from long-term COVID-19 symptoms [8].

At the start of the COVID-19 pandemic, a pediatric biorepository was established at Massachusetts General Hospital (MGH) in Boston, Massachusetts [9]. This biorepository collected, processed, and stored donated biospecimens including nasopharyngeal and oropharyngeal swabs, blood, sputum, stool, and urine. We hypothesized that among those approached for enrollment, due to normative development, school-age children (5–12 years) might be less likely to enroll in the biorepository or subsequently donate biospecimens compared to young children and infants (< 5 years) and adolescent and young adults (> 12 years). We also hypothesized that overall willingness to enroll may be higher than in pediatric biorepository studies conducted outside of a pandemic setting. We describe predictors of patient enrollment to a pediatric biorepository to identify strengths and gaps in COVID-19 research on pediatric biospecimens.

Methods

Study population

The study population included children and young adults (0–25 years) in community or academic outpatient or inpatient settings with suspected or confirmed COVID-19 infection. Enrollment procedures have been previously described and were conducted verbally (English/Spanish or with the assistance of an interpreter) in the setting of the pandemic [9], based on outpatient clinic appointment schedules or emergency department and hospital admission lists. Individuals were approached for enrollment to the pediatric biorepository at MGH regardless of later COVID-19 test results. To offset the surge in medical demand from older adults during the early phases of the COVID-19 pandemic, pediatric clinics evaluated young adults through age 25. An institutional community research program was also in place at community enrollment sites.

Clinical and laboratory data

Data were derived from manual chart extractions and automatic electronic data downloaded at presentation to care for both enrollees and non-enrollees who were approached during the first wave of the COVID-19 pandemic in Boston between April 12, 2020, to May 28, 2020. Presentation to care was considered a visit via telephone, video, or in-person. Data included recruitment site, demographics, epidemiological data, and clinical characteristics (vital signs and medical history). Epidemiologic risk was defined by chart review and included international or domestic travel, occupational exposure (non-healthcare), having a household contact, being undomiciled, having received healthcare in a facility with known COVID-19 cases, or other risk factors.

Study outcomes

The primary outcome variable was whether the participant enrolled in the biorepository. Secondary outcomes included biospecimens donated by enrollees including nasopharyngeal and oropharyngeal swabs, blood, stool, and urine. Urine and stool samples were only collected from inpatients.

Predictors

Demographic and clinical predictors were evaluated as well as three additional variables: US Centers for Disease Control and Prevention (CDC) hotspot county definition (a minimum of 5 criteria based on prevalence and changes in incidence of COVID-19) [10], COVID-19 county risk level defined by cases per 100,000 people per day by the Harvard Global Health Institute (low < 1; mild: 1–9; moderate: 10–24; and high: ≥25) [11], and Neighborhood Deprivation Index as outlined by Messer et al, which used variables indicative of neighborhood-level socio-economic status based on zip code [12]. The estimated Neighborhood Deprivation Index is a standardized score that has a mean of zero and standard deviation of one. Higher values indicate a more affluent “less deprived” neighborhood (i.e., a neighborhood that has Neighborhood Deprivation Index score of 0.8 indicates a less deprived neighborhood than one with a score of 0.2; likewise a neighborhood with a score of − 0.2 indicates a less deprived neighborhood than one with a score of − 0.8.). Details of these variables are outlined in the Supplemental Methods.

Statistical analysis

Data are presented as percentages for categorical variables and as median ± inter-quartile range (IQR) for continuous variables. We performed bivariate associations to determine the association between predictors and outcome variables. We compared categorical variables using Chi-squared or Fisher exact tests and continuous variables using Kruskal Wallis tests.

We used multivariable Poisson regression models to assess relationships between outcomes and potential predictors, reporting relative risks (RR), 95% confidence intervals (CI), and p-values. Model selection procedure is described in the Supplemental Methods. We conducted principal component analysis to estimate Neighborhood Deprivation Index (Supplemental Methods). All statistical analyses were conducted using SAS version 9.4.

Institutional review board approval

Research was approved by the Mass General Brigham Human Research Committee (Protocol 2020P003588 and IRB#2020P000955) and the Partners/Massachusetts General Hospital Institutional Biosafety Committee (IBC#2020B000061).

Results

Characteristics of the study population

Among 457 individuals approached for enrollment, 46% were female, 28% self-reported as White/non-Hispanic, 6% Black/non-Hispanic, 9% Other, 12% not reported; 44% reported Hispanic ethnicity, and 68% reported English as a primary language (Table 1). The age strata of those approached for enrollment included: 35% 0–5-years-old, 27% 6–12-years-old, 22% 13–18-years-old, and 17% 19–25-years-old. The median (IQR) of Neighborhood Deprivation Index among those approached for enrollment was − 0.6 (− 0.8, 0.7). Most participants were either in home daycare or cared for by a nanny (21%) or in grade school (39%). An epidemiologic risk factor for COVID-19 was noted on chart review among 53% at the time of enrollment; of those, 48% recorded a household contact, with 30% and 8% of household contacts being with parents and siblings, respectively. Only 15% lived in a COVID-19 hotspot as defined by CDC, and none lived in a high COVID-19-risk level county per the Harvard Global Health Institute definition; 87% lived in low COVID-19-risk level counties.

Table 1 Baseline characteristics among youth ages 0 to 25 years with suspected COVID-19 approached to participate in a pediatric biorepository in the greater Boston area
Full size table

Most individuals who were approached for enrollment presented to an outpatient clinic (83%); only 10% of those approached were admitted to the hospital. At the time of presentation, 81% were symptomatic and 35% reported fever. Only 2% presented with oxygen saturation ≤ 94%. Most (53%) reported having any past medical history and 43% reported being prescribed any medication. Among those with a cardiac or metabolic medical history (20%), 79% had obesity. Among those with pulmonary medical history (26%), 61% reported asthma. Eighty-four percent had development appropriate for age recorded, and 20% had a documented neurodevelopmental condition. Of those with a history of neurologic events (8%), most had seizure (47%) and/or headache (19%). There were few individuals prescribed immunosuppressing medications (≤1%) or inhaled steroids (4%). Among those approached for enrollment, COVID-19 testing was conducted for 88%; of these individuals who received a COVID-19 test, 29% had a positive COVID-19 test. Among those who were symptomatic or asymptomatic at presentation to care, 30 and 26% were COVID-19 positive, respectively.

Characteristics of enrollees versus non-enrollees

Enrollees and non-enrollees had similar distributions of sex assigned at birth, race/ethnicity, primary language spoken, categorical age, oxygen saturation at presentation to care, and most measures of past medical history and COVID-19 testing and positivity (Table 2). Of those approached, 47% enrolled. Enrollees were more likely than non-enrollees to be older (median 10.3 years vs. 8.7 years, p = 0.03). Those in preschool/kindergarten (52%) were more likely to enroll in the biorepository than other daycare or school types; those belonging to group daycare (26%) were least likely to enroll (p < 0.001). Participants living in a COVID-19 hotspot county were also more likely to enroll (99% vs. 37%, p < 0.0001). Participants living in counties at higher COVID-19 risk levels were also more likely to enroll than those at lower risk levels (moderate: 100%; mild: 93%; low: 39%, p < 0.0001). Individuals presenting to the Emergency Department or who were transferred from an outside hospital or elsewhere (63% and 75%) were more likely to enroll than individuals presenting to outpatient clinics (42%, p < 0.0001). Those admitted to the hospital were more likely to enroll than those who were not admitted (75% vs. 43%, p < 0.001). Those who had any symptoms or temperature > 98.6 degrees Fahrenheit at presentation to care were more likely to enroll than those without symptoms (51% vs. 31%, p < 0.001; 57% vs. 41%, p = 0.004). In terms of past medical history and medication, participants who had asthma (54% vs. 35% p = 0.04), took asthma medications (61% vs. 45%, p = 0.02) or inhaled steroids (70% vs. 46% p = 0.03) were more likely to enroll. Sex assigned at birth, Neighborhood Deprivation Index, and primary language spoken at home were not associated with enrollment.

Table 2 Comparison of characteristics of enrollees and non-enrollees in a pediatric biorepository
Full size table

Multivariable regression

Three variables were predictors of enrollment: living in a COVID-19 hotspot [2.4 (1.8, 3.2)] (p < 0.0001), hospital admission [1.8 (1.2, 2.7)] (p = 0.002), and whether the participant was symptomatic at presentation to care [1.8 (1.2, 2.8)] (p = 0.006) (Fig. 1). Although age and Neighborhood Deprivation Index were hypothesized to affect enrollment, these factors were not significant predictors in the Poisson regression model.

Fig. 1
figure 1

Relative risks of predictors of enrollment. The forest plot shows the relative risk (horizontal axis) and 95% confidence interval (horizontal bars) of each variable (vertical axis) included in the Poisson regression model: symptomatic at presentation to care, living in a COVID-19 hotspot, hospital admission Neighborhood Deprivation Index and age

Full size image

Predictors of biospecimen donation

Among people who enrolled, 79% donated any biospecimen; 97 individuals donated nasopharyngeal swabs, 119 donated oropharyngeal swabs, 105 donated blood, 16 donated urine, and 16 donated stool specimens (urine and stool were only collected from hospitalized patients) (Table 3). Among enrollees, living in a COVID-19 hotspot (p = 0.01), whether a COVID-19 test was conducted (p = 0.05), and positive COVID-19 test result (p = 0.02) were associated with donating nasopharyngeal specimens. Known epidemiological risk (p = 0.04) was associated with donating oropharyngeal specimens. Age, sex, race, ethnicity, and zip code-based Neighborhood Deprivation Index and other variables did not predict biospecimen donation.

Table 3 Characteristics of donated biospecimens (N = 204)
Full size table

Discussion

In this study conducted at the beginning of the first surge of SARS-CoV-2 in greater Boston, Massachusetts, 47% of all individuals approached were willing to enroll in a pediatric biorepository for COVID-19. Individuals living in areas at high risk of COVID-19, with symptoms at presentation, or who were admitted to the hospital were the most likely to enroll. These factors may have contributed to a higher perceived risk in individuals that ultimately motivated them to enroll in the biorepository. Even though enrollees tended to be older than non-enrollees, all age categories of interest were represented among biorepository enrollees. Enrollment sites were among the few available places where children could be evaluated in person and tested for COVID-19 during the early pandemic. While we hypothesized that enrollment would be higher than in pediatric biorepositories in non-pandemic settings, these data are lacking for pediatric biorepositories with as broad a focus. Few studies have examined demographics and factors associated with declining enrollment at the time of approach for enrollment, either for pediatric clinical trial research [13], or pediatric biorepositories [14,15,16,17]. Furthermore, few existing studies on pediatric biorepositories enroll patients from a general pediatric population and rather focus on pediatric patients impacted by rare diseases such as congenital heart disease and cancer [13, 16, 18, 19]; enrollment rates in such biorepositories range from 55-97% [14, 18,19,20,21]. To our knowledge, this is the first study of demographic and epidemiologic predictors of enrollment in a pediatric biorepository examining characteristics of pediatric patients themselves (rather than their parents/guardians) or conducted in a pandemic setting [14, 15, 22].

Previous studies within and outside the US context have noted that parents willing to enroll their children in a biorepository are more likely to be highly educated, have a higher income, and experience more trust and positive attitudes towards clinical research [14, 17, 23]. Parents are generally willing to enroll children in clinical trials [24,25,26]. Parents agree to their children's participation of there is a higher perceived benefit to their children [15, 23]. Parents tend to overestimate the benefits of research on the well-being of their children and underestimate any risks in donating samples [26, 27]. While in the setting of this study, more deprived communities tended to be COVID-19 hotspots, it is notable that the community-level measurement of neighborhood socioeconomic status – Neighborhood Deprivation Index – did not predict enrollment. Participants and their parents may have had higher perceived risk in deprived neighborhoods, and therefore a tendency to agree to participate; furthermore, the establishment of community health research infrastructure antecedent to the pandemic may have facilitated research participation in these settings.

In the present study, parents were informed that donating samples would deepen our understanding of how COVID-19 impacts children. Although parents in the US and elsewhere may be less willing to enroll their children than themselves, parents’ beliefs related to their child’s enrollment generally mirror concerns they have for themselves [24, 27,28,29,30,31,32]. Most people are aware of the potential risks of biorepository participation but do not feel concern about being harmed [27]. Overall, pre-pandemic, people have reported high levels of trust in public health and clinical research institutions [16, 28, 30]. Systemic or institutional trust is critical to address common concerns about biorepository enrollment, including biospecimen misuse and unacceptable research in the future (specifically human cloning), risk of identification or identity theft, insurance discrimination, and potential for corporate exploitation [28,29,30,31]. Participants are comfortable donating if they feel they are adequately informed about how their biospecimens will be used. Those who feel the negative impact of biospecimens donation outweigh any positive impact tend to feel that people generally cannot be trusted, are most concerned about re-identification and have worries about information theft [30]. Concerns of data breaches and lack of confidentiality are deterrents to enrollment in biorepositories [33]. In certain studies, women and minorities are not as willing to donate their biospecimens, perhaps due to a historic lack of trustworthiness of institutions conducting clinical research (e.g., unconsented use of Henrietta Lacks’ cancer cells) [33,34,35]. In contrast to prior reports from countries like the US or Australia, in which parents who are white or non-immigrants are most comfortable enrolling children into biorepositories [14, 35], in this study, in which 44% identified as Hispanic, we found that no one racial or ethnic group was more likely to enroll than another.

We hypothesized that school-age children would be least likely to provide nasopharyngeal swabs and blood. Parents are less willing to enroll their children when they are concerned about the physical pain or discomfort that their children may face, if their children are more seriously ill, or if they are concerned about their children’s privacy and confidentiality [13, 15, 16]. We found no differences by age or other factors regarding participants’ willingness to donate biospecimens, which suggests that efforts in this population should be most focused on enrollment itself rather than consenting for specific biospecimen provision.

This study had important limitations. First, by design, this analysis was conducted by chart review and electronic medical record download. We were, therefore, unable to explore reasons for declining biorepository enrollment at the time of being approached; while there are many studies detailing parental willingness to participate in a pediatric biorepository, few have examined motivations for declining enrollment at the time of enrollment [14, 19]. Given reports of high willingness to enroll, there may be important gaps between intention to enroll, enrollment completion, and biospecimen donation. Qualitative surveys of parental perceptions of clinical research institutions, corporate exploitation, religiosity, and data breaches would bolster the interpretation of our study findings. Second, we lacked data on individual-level parental highest level of education and socioeconomic status; to address this, we incorporated Neighborhood Deprivation Index. We also did not collect information on religiosity; in one study lack of religiosity has been associated with willingness to donate biospecimens [32]; another study noted that some religious individuals consider their tissue to be 'sacred' or part of their identity and found that higher religiosity made parents less willing to consent their children for biorepository participation [16]. Finally, our analysis was conducted in Massachusetts and so we were not able to investigate how willingness to consent and enroll differed by state.

Living in a COVID-19 hotspot, the presence of symptoms at presentation, and hospital admission predicted a higher likelihood of pediatric biorepository enrollment among children with suspected or confirmed COVID-19 infection. While overall enrollment appeared to be low compared to previous reports of pediatric biorepository enrollment, the population who consented to biorepository participation reflected those seeking care: children of all age and Neighborhood Deprivation Index categories were represented, as well as patients recording Hispanic ethnicity, who comprised a major part of the first surge in Boston [36]. Understanding whether enrollment is representative of the COVID-19 pandemic is critical to interpreting analyses conducted on donated specimens. Furthermore, understanding factors related to non-enrollment may also permit research to encourage enrollment among key demographics.