Over the past few decades, millions of people worldwide have been affected by natural, manmade, and technological disasters as well as complex humanitarian emergencies.Reference Nager1 Disasters can disproportionately affect children with consequences including being separated from their families and caregivers during the event. Unaccompanied children may face secondary injuries such as abduction, neglect, physical and sexual abuse, and long-term psychological distress.Reference Nager1–Reference Baker and Cormier3 In order to minimize the potential for harm and protect displaced children in disaster situations, identification and reunification of these children with their families should be considered a top priority.Reference Nager1, Reference Brandenburg, Watkins and Brandenburg2
During past disasters, reunification has been found to be very challenging, especially when the event involved a large number of displaced children.Reference Brandenburg, Watkins and Brandenburg2 After Hurricane Katrina in 2009, 5068 children were separated from their families; many were transported to different shelters across the country from their parents without an adequate way to track their location. With the assistance of a nongovernmental organization, the National Center for Missing and Exploited Children (NCMEC), the final child was reunited with their family 6 months later.Reference Brandenburg, Watkins and Brandenburg2, Reference Baker and Cormier3 Reunification is a complex process that requires organization and coordination, communication, and sharing of information between public and private local, regional, and national agencies.Reference Nager1, Reference Brandenburg, Watkins and Brandenburg2, Reference Chung and Blake4 Various reunification approaches have been attempted to collect information and track displaced individuals during disasters, including national programs such as the Red Cross Safe and Well program and the NCMEC’s reunification program. However, as of 2018, a standardized community reunification system does not exist in the United States.
Having a standardized community reunification system would help shorten the time to reunification. Such a system would need a centralized database able to combine information on unidentified children with information provided by guardians seeking lost children. However, use of a centralized database would require support from parents and legal guardians, as only they are legally able to share their child’s personal information. A general increase in concern for protecting private health and personal information has arisen in recent years, as cyber attacks have become more frequent and publicly reported.5–7 Past studies have found that parents/guardians are reluctant to share their child’s information, depending on the content as well as the entity with which the information would be shared.Reference Weitzman, Kelemen and Kaci8
The primary purpose of this study is to determine legal guardians’ willingness to share their children’s personal information in a centralized reunification software system. Secondary aims are to identify guardians’ attitudes and beliefs regarding the sharing of their children’s personal information to determine possible obstacles to development of a shared reunification system, identify perceived trust in agencies that might manage a reunification system, and ascertain perceived ability to use and trust various reunification system formats.
METHODS
An anonymous online surveyReference Harris, Taylor and Thielke9 administered via RedCap®was offered to a convenience sample of adults presenting with a child to the emergency department at Boston Children’s Hospital, Boston, Massachusetts, and Cardinal Glennon Children’s Hospital, St Louis, Missouri, during October 2017 through April 2018. Adults were excluded if they were not able to read or speak English or if they did not have at least 1 child aged 14 years or younger.
Survey Questionnaire
This questionnaire was based on research related to reunification of unaccompanied minors during a disaster; the American Academy of Pediatrics Disaster Preparedness Advisory Council Reunification subcommittee’s developing toolkit, Family Reunification Following Disasters: A Planning Tool for Health Care Facilities; and existing programs used to reunite families in disasters, such as NCMEC.5 The primary outcome was the study participants’ willingness to share elements of their children’s personal information, such as the child’s name, physical description (hair, eye, and skin color), language, distinguishing physical characteristics (eg, birthmarks, scars, tattoos, or piercings), photographs, and videos. A secondary outcome includes identification of parents and legal guardians’ attitudes and beliefs regarding the sharing of their children’s personal information that may influence their willingness to share such information, such as perceived concern about data security breaches, perceived importance of having a shared community reunification system, and perceived trustworthiness of various agencies. Willingness to share, attitude/belief questions, and perceived trust in agencies were measured on a 5-point Likert-type scale (very willing to very unwilling, strongly disagree to strongly agree, and very untrustworthy to very trustworthy, respectively). Perceived ability to use and trust in the 3 reunification system formats (telephone call, smartphone/tablet app, and internet site) were measured as yes/no.
The questionnaire was pilot-tested with a group of 10 guardians of children at each site. The study was approved by the Saint Louis University Institutional Review Board and the Boston Children’s Hospital Institutional Review Board.
Data Analysis
An a priori power analysis was conducted to determine an appropriate sample size. Both a multivariate linear regression (outcome variable, sharing information score that can range from 0 to 19) and a multivariate logistic regression (outcome variable, willing to share even a single item of information versus unwilling to share anything) were planned, with possible stratification by hospital. The desired sample size was calculated with G*Power6 to be 323 subjects (about 162 per site), with the assumption of a 95% confidence interval and a margin of error of 5%. The study aimed to include a total sample of 350 subjects: 175 from each of the 2 participating hospitals. The Statistical Package for the Social Sciences (SPSS®) 24.0 and R 3.5.0 were used for all analyses.7 Attitude/belief Likert-scale questions were dichotomized (agree somewhat and agree strongly = 1; neutral, disagree somewhat, and disagree strongly = 0). An overall sharing information score was calculated by assigning 1 point for each piece of personal information that the parent/guardian reported they would be willing to share about their children for reunification purposes (ie, indicated they were somewhat willing or very willing to share). The highest possible sharing information score was 19 (ie, 1 point for each of the 19 personal information items). The full list of items is available in Figure 1.
Percentage of Parents/Guardians Willing to Share Each Piece of Their Child(ren)’s Personal Information for Reunification Purposes
Descriptive statistics were used to describe the percentage of parents/guardians who were willing to share each item of their children’s personal information, their attitudes/beliefs about the safety of sharing such information, perceived trustworthiness of agencies, and preferred formats for reunification systems. Proportions tests were used to evaluate differences in agreement between attitudes/beliefs, perceived trustworthiness of agencies, and preferred formats for reunification systems. Chi-squares were used to assess racial differences and attitudes/beliefs and perceived trustworthiness of agencies. Multivariate linear regression was used to determine factors associated with parents/guardians’ willingness to share more pieces of their children’s personal information for reunification purposes. The Hosmer and Lemeshow goodness-of-fit test was used to assess overall model fit for the regression. Univariate analyses consisting of independent samples t tests and analysis of variance were conducted prior to the regression analysis, with all demographic variables and attitude/belief items as possible predictors. Only variables that were significant in univariate analysis (with a critical P value of .05) were included in the multivariate analysis. Variables that were significant on univariate analysis but nonsignificant on multivariate analysis were dropped from the model. Only the final model is reported.
RESULTS
In all, 363 individuals participated (response rate, 80.3%); 51.8% (n = 188) were from Boston and 48.2% (n = 175) from St Louis. Most were female (78.0%, n = 276; Table 1). A little more than half (57.4%, n = 193) were Caucasian, and about a third (34.6%, n = 116) were African American. A full list of participant demographics is provided in Table 1. Participants in Boston were older, had received more education, had higher income, and were more likely to be male and white than were the St Louis participants (Table 1).
Demographics of Respondents
Abbreviation: NS, not significant.
a Denominator varies due to missing data (some respondents chose not to identify their gender, race, or other demographic data).
b Determined by the chi-square test.
Fewer than 1% of parents/guardians (0.8%, n = 3) reported having ever been physically separated from children during a past disaster. Almost a quarter (19.0%, n = 69) reported having been separated from their children when out in public, such as at a sports stadium, fair, the mall, or amusement park. About a third of parents/guardians (32.0%, n = 116) reported that they have at least 1 child under the age of 15 years whose medical history would be necessary for medical providers to know in order to prevent harm to the child if separated during a disaster. Of the parents/guardians who have a child with a critical medical history (n = 116), 62.1% (n = 72) reported that this makes them more willing to share their child’s personal information for reunification purposes, and another third (31.9%, n = 37) indicated that it does not affect their willingness to share personal information; only 6.0% (n = 7) reported that their child’s medical history makes them less willing to share their child’s personal information for reunification purposes.
Parents/Guardians’ Willingness to Share Their Children’s Personal Information for Reunification
Overall, 4.4% (n = 16) of parents/guardians were unwilling to share any of their children’s personal information for reunification purposes. The remainder (95.6%, n = 347) were willing to share an average of 16 items (SD, 5.4; range, 1-19). The only significant predictor of willingness to share one’s child’s personal information was perceiving that having a community reunification system is important (P < .001); all other demographics, such as age, gender, income, or child’s age, and all other attitude and belief questions were nonsignificant.
Most parents/guardians were willing to share every personal information item assessed. Agreement to share each piece of personal information ranged from 72.3% (video) to 88.6% (birthmarks). Figure 1 outlines the percentages of parents/guardians who were willing to share each piece of their children’s personal information for reunification purposes. Parents/guardians were significantly more willing to share information about their children’s birthmarks, piercings, tattoos, hair color/description, and eye color than they were to share their children’s school name (P < .02), home address (P < .01), photo of child’s identifying marks (P < .01), video of child (P < .01), or DNA (P < .001).
Parents/Guardians’ Attitudes and Beliefs About Sharing Their Children’s Personal Information for Reunification
Most parents/guardians believed it is important to have a community reunification system (87.9%, n = 319), though significantly fewer believed it would be necessary to share their own child’s data (80.4%, n = 292; P < .01; Table 2). Approximately half were concerned about the protection or misuse of their child’s information to either claim their child or for another purpose (55.4%, 53.2%, and 52.3%, respectively; Table 2). Just under half (46.6%, n = 169) were concerned that if they shared their child’s information, it would stay in the database permanently (Table 2). About a quarter (24.0%, n = 87) were concerned that their child’s personal information would be shared with child protective services (Table 2). Non-Caucasian parents/guardians were significantly more concerned than Caucasian parents/guardians about the protection or misuse of their child’s information (Table 2).
Parents/Guardians’ Attitudes and Beliefs Regarding Sharing Their Child(ren)’s Personal Information for Reunification Purposes
Abbreviation: NS, not significant.
a Denominator is fewer than all respondents because some respondents did not report their race.
b Determined by the X 2 test.
Perceived Trust in Agencies to Manage Reunification Information
Parents/guardians were asked whether they would trust 10 different agencies to manage a reunification system/program. Figure 2 outlines the percentages of parents/guardians who trust each agency with their children’s personal information for reunification purposes. Hospitals and the NCMEC were the 2 most trusted agencies to manage reunification information (86.0% and 78.8%, respectively; Figure 2). Universities and a state government other than the state in which the parent/guardian lives were the least trusted (57.3% and 47.9%, respectively; Figure 2). Significantly more parents/guardians reported trusting hospitals than any other agency (P < .001 for all comparisons). Significantly fewer non-Caucasian parents/guardians reported trusting local (P < .01), state (P < .05), and federal governments (P < .01) and public health (P < .05) than Caucasian parents/guardians.
Percentage of Parents/Guardians Who Trust the Agency with Reunification Information
Parents/Guardians’ Perceived Ability to Use and Trust in Telephone, Smartphone, Tablet, or Internet Reunification System Formats
Parents/guardians were asked about their perceived ability to use and their trust in a reunification system implemented through a smartphone or tablet app, an internet site, or a telephone call. Most parents/guardians reported that they would be able to use a telephone call, smartphone/tablet, or internet site for reunification (96.4%, 94.5%, and 93.4%, respectively). Although most parents/guardians reported trusting a telephone call, smartphone/tablet, or internet site (80.7%, 77.7%, and 69.7%, respectively) for relaying reunification data, they were significantly more likely to report being able to use these systems than trusting in those systems (P < .05 for all comparisons). Significantly more parents/guardians reported trusting a telephone call for reunification than a smartphone/tablet or internet site (P < .001 for both comparisons).
DISCUSSION
This study found that the vast majority of parents are willing to share at least some personal information on their child in order to rapidly reunify with them during a disaster. This indicates that parents/guardians are open to interacting with a reunification system that utilizes personal information. While parents/guardians are willing to share information to enable faster reunification, privacy concerns pervade.
While rapid and effective reunification is universally acknowledged as an integral part of disaster management and recovery, reunification planning remains underdeveloped, with plans often existing in silos. Hospitals, which will be the site for family reunifications if there are injuries, are lagging in preparing for this scenario; a recent survey shows only 47% of US emergency departments have disaster plans that involve children.Reference Gausche-Hill, Ely and Schmuhl10 The 2010 National Commission of Children and Disasters recommended that the Department of Homeland Security lead the way in developing technology capable of tracking and reunifying children in a disaster.11 While there are multiple systems capable of performing some aspects of disaster preparedness, including those of NCMEC, the Red Cross, and even systems developed by social media sites such as Twitter and Facebook, a complete dual portal system able to match information submitted by parents/guardians with information submitted by those in custody of an unidentified child does not yet exist on a national level. Parents seeking their missing children may need to access various systems to find their children.Reference Blake and Stevenson12
The development of a broader community reunification system would also relieve the burden of reunification from already overwhelmed first responders and guardians. It is essential that such a system combine information gathered about the child from the guardian with information from agencies that have unaccompanied minors in their custody. However, guardians may be reluctant to share their child’s personal information due to potential privacy concerns, especially if the data is collected electronically.Reference Weitzman, Kelemen and Kaci8 This extends to seemingly irrelevant but unique and useful information, such as a school or pet name. This survey sought to identify guardians’ concerns regarding the sharing of their child’s personal information in a disaster setting. Identifying information guardians are unwilling to share allows a reunification system to be designed that will address and minimize such concerns. Additionally, though gathering more information on children would potentially allow easier and faster identification, it is critical to restrict information-gathering to only those variables guardians are comfortable sharing, as this will increase trust and the likelihood of the system being utilized during a disaster.
After Hurricane Katrina, the use of children’s photographs was found to be the most effective means of reunification.Reference Brandenburg, Watkins and Brandenburg2, Reference Broughton, Allen and Hannemann13 Chung and Shannon proposed a system in which digital images of children separated from their parents could be uploaded into a centralized system. Using advanced imaging and feature extraction algorithms, the system would automatically index facial features, such as skin or eye color. Parents trying to find their children could enter their child’s facial features into the system and receive a reduced set of images for identification, allowing for rapid reunification of the family.Reference Chung and Shannon14 In a survey of emergency management professionals, Chung et al. found that participants preferred a system that displayed unedited photographs of missing children, and over 50% of participants were willing to adopt a photo-based reunification system if the system could only reunite 10% of the families in a large-scale disaster.Reference Chung, Monteiro and Ziniel15 In their pilot prototype, Chung et al. showed that such an image-based reunification system reduced the number of images reviewed before parents identified their child. Our study shows that requesting photos or videos of children or collecting DNA samples is likely to cause discomfort in guardians, although most (72%–83%) parents/guardians are still willing to share this information.Reference Chung, Mario Christoudias and Darrell16 While some of this information may be extremely useful, such as photographs for use with facial recognition software, special framing for these requests may be helpful to encourage guardians to share this particular data.Reference Broach, Yong and Manuell17, Reference Kling, Egeland and Herrera Piñero18
Findings from this study indicate that a reunification system needs to clearly delineate with whom the child’s personal information will be shared, have clear parameters for the removal of information, and outline the steps used to protect information. Half of the respondents expressed concerns about these topics. Events in which information is hacked, such as the 2017 WannaCry ransomware attack to the UK National Health Service or the September 2018 breach of over 40 000 patient records in Hawaii, generate distrust among guardians.19 Advanced security measures are necessary to protect the personal information of such a vulnerable population.Reference Gausche-Hill, Ely and Schmuhl10 These concerns were particularly prominent in the minority population responding to this survey. In particular, prevention or detection of privacy leaks and security breaches in any aspect of a health care system are active areas of research in computer security. Effective measures to guarantee privacy of children’s information in database entries exist. For example, confidentiality and live authentication can be achieved with the latest encryption technologies, access control rules can be changed dynamically, or differential privacy techniques can be implemented on database records.Reference Stallings20 However, even if the most effective security and privacy mechanisms are adopted, it is critical to educate those who will have access to this sensitive data to ensure the safety of the information.
Findings from this and previous studies indicate that one significant way to improve guardians’ trust, and hence utilization of a reunification system, would be to have a hospital manage the system. One study showed baseline trust in confidentiality with personal health information to be highest with hospitals (85%), followed by universities (73%), and local government (39%).Reference Willison, Steeves and Charles21 A UK study suggested that universities were the most highly trusted with health information (after the National Health Service), and private organizations were least trusted.Reference Ritchie, Reid and Smith22 Guardians were significantly more likely to trust a hospital with their child’s personal information. Interestingly, in this current study, universities, while often affiliated with hospitals through medical schools (including both hospitals used as sites for this study), were among the groups least trusted. Academic hospitals implementing a reunification system may find it more successful to emphasize the hospital name as being the managing agency.
A final factor to consider when planning how to develop a successful reunification system is how users will access the system. Though guardians in this study reported being capable of using multiple modalities, telephone calls were perceived as most trustworthy compared to using an app or internet site to enter their child’s information. This has significant implications for development and implementation of a reunification system, as the resources needed to deploy phone line(s) are much higher than those needed to deploy a passive app. Further research is needed to determine if guardians would tolerate a reunification system that either minimized or did not use a telephone line. It is likely that a reunification system used in a future disaster would need to use a mix of modalities to ensure flexibility and comprehensive coverage. For example, a successful system may need to encourage guardians to use an app or another internet-based service, but also have a backup phone system that would take calls.
This study has some limitations. A convenience sampling methodology was used, which can introduce sampling error and potential bias. It was conducted in 2 hospitals in 2 different geographical regions of the United States to gain opinions from diverse populations. However, the sample may be biased toward those who live in urban and suburban areas and have access to a pediatric hospital. Those who live in rural areas or live somewhere other than St Louis or Boston may not share the same opinions as those in this study. This study also excluded non-English speakers, and their willingness to share their children’s personal information may differ from this sample. Additionally, it is possible that trust in hospitals was biased by the recruitment approach used in this study (ie, approaching families seeking care in a hospital emergency room). Recreating this survey in a nonhospital setting would be helpful to determine the reproducibility of these findings.
CONCLUSION
There is near-universal willingness on the part of guardians to share their children’s personal information to facilitate reunification after a disaster. The potential for misuse of a database of children’s information is of concern, so selection of the managing agency, appropriate handling of sensitive information, and clear guidelines on how and when information is purged from the system are critical to ensure parental trust. Findings from this study indicate that a hospital would be the ideal agency to manage a reunification system. In addition, a multimethod approach involving an app, internet-based system, and/or telephone line to collect data for the reunification system would be best to ensure success. To be most effective, a reunification system will need to take all of these factors into account in its design and implementation.
Financial Support
This work was supported by the Dr Michael Shannon Emergency Medicine Award (Boston Children’s Hospital) to Dr Sarita Chung.
