Rationale for school closure
Three different rationales for closing schools during a public health emergency emerged in the 2009 H1N1 outbreak: (1) limiting spread of the virus in the community, (2) protecting vulnerable children, and (3) reacting to staff shortages or children kept at home because of infection or parents' fears of infection.
In the United States as in Europe [
15], the most common rationale for school closure was to limit spread of the H1N1 virus in the community. This is based on the idea that schools provide an ideal context for spread of infectious diseases because children are more susceptible to infection, less likely to adopt behavioral changes that reduce disease spread, and more likely to sustain person-to-person contact for lengthy periods. Closing schools, therefore, may limit spread among children as well as to their families and the general community, and can be an important component of a community's "social distancing" efforts [
16]. Reducing community transmission might have the effect of reducing cumulative incidence in a community, but, more likely, would spread the epidemic curve in a community, slowing the accumulation of cases while alternative control measures are employed and reducing peak incidence to a level more manageable by the health care system [
17]. Closing schools was an important component of Mexico's highly-regarded social distancing efforts in response to 2009 H1N1 [
18]. U.S. school administrators and public health officials typically justified the costs of school closure on these grounds [
19‐
21].
Evidence of the impact of school closure on the spread of influenza, however, is limited and mixed. Historical analyses and epidemiologic modeling studies for influenza and other respiratory diseases suggest that social distancing measures, especially if implemented early in an outbreak and sustained, can substantially reduce both the total number of cases and the peak attack rate [
22‐
26]. On the other hand, a recent systematic review found only 19 studies with primary empirical data on the impact of school closure
per se, many of which had significant methodological challenges [
27]. Some mathematical models suggest that school closure has a limited impact on cumulative case counts, since children not attending school are free to transmit infection to their families and others in the community [
28]. Other modeling studies suggest combining school closure with sequestering children in the home may curb pandemic spread and peak incidence [
29]. Differences are largely attributable to assumptions underlying the models, reflecting uncertainties about epidemiology and behavior that may be better understood as more data become available. At minimum, though, it is clear that closure's effectiveness is reduced if students congregate in large number in other settings, and this proved to be a challenge in 2009 [
30]. It also appears, based on modeling studies, that significant reductions in cumulative incidence may require extended periods of closure (with eight to sixteen weeks resulting in the greatest reductions in peak incidence) [
31]. We did not identify any school that closed for this long, except in Hong Kong, where schools were closed and not reopened before summer break, with a reduction in cases occurring simultaneously [
32]. This is the only epidemiological analysis of which we are aware to suggest that school closings actually had the effect of limiting spread of H1N1 in 2009.
Although advocated as a social distancing measure that imposes less societal costs than workplace closures or public disruptions, it became evident during the Spring 2009 H1N1 pandemic wave that school closures imposed substantial costs in some instances. Parents complained of the difficulty of finding child care or the financial costs associated with finding someone to take care of their children or staying home [
33]. New York City officials originally based their decision to close schools one-at-a-time on the need to balance public health interests "with the child-care and educational needs of families" [
34]. These officials noted, however, that closures were compromised as a control measure if "the kids don't go to school and instead go to the shopping mall or go to the park" [
35] as library officials in Queens reported a large number of children congregating after their schools were closed [
36].
Indeed, formal studies of students in Boston and Pittsburgh have documented this behavior. High school students in the Winsor School, a private girls' school in Boston that closed from May 20-26 following a sudden increase in absenteeism, reported the average number of days during this period on which they participated in the following activities: shopping 1.47 days, visiting a friend 2.21 days, using public transport 1.89 days, eating out 2.44 days, and outdoor activities 3.42 days [
37]. During a one-week closure of an elementary school in southwestern Pennsylvania that closed during this period, 69% of students report having visited at least one location outside their home [
38]. These findings are consistent with pre-pandemic findings about student behavior during closures for seasonal influenza [
39].
As data emerged to suggest that school-aged children experienced a higher attack rate than other age groups and an unusually high rate of complications from the 2009 H1N1 virus [
40], protecting children, especially those who may be particularly vulnerable to complications, became a second rationale for school closure during the 2009 outbreak. In New York City, where officials continued closing schools even after CDC stopped recommending it, Mayor Michael Bloomberg eventually clarified that school closure would not slow transmission and that closure "has absolutely nothing to do with the spread of the disease." Rather, the justification for closures shifted to preventing secondary cases among particularly vulnerable school contacts [
41].
The third rationale was purely practical: staff shortages, whether due to actual illness or concerns about infection or children kept at home because parents feared they would become infected simply made it impossible to keep schools open. We did not observe this rationale in the Spring 2009 H1N1 outbreak, however closure due to very high absenteeism became common in the Fall resurgence. For example, several Connecticut schools closed in October because of high student absenteeism. "You can't teach with one- to two-thirds of the class absent," explained the superintendant of a district with closed schools [
42]. In the last week of October 2009, about 350 schools were closed nationwide, a high proportion of which appeared to be in response to high absenteeism [
43].
Triggers for school closings
Modeling studies consistently show that, to be effective, school closure to limit community-wide transmission requires an early trigger, such as before 1-2% of the population is infected [
44,
45]. These results reflect the assumption that influenza transmission can occur before patients develop symptoms or in the presence of mild symptoms, so there can be a substantial number of infections before a sizable number are identified. Detection of a novel strain usually occurs after it is well-established in a locale, so the community-wide benefit of school closures is likely substantially diminished at that point [
46]. Unless closure is very fast following the first identified cases or case detection is very efficient, it may be difficult for schools to close in time to be maximally effective. Indeed, in 2009 transmission was often not identified in a locale until clusters had already been detected in schools [
47].
For epidemiological purposes H1N1 cases are defined (with ascending degrees of certainty) as "suspected," "probable," or "laboratory confirmed" and the initial 2009 H1N1 case definitions emphasized contact with other known cases [
48]. In particular, a suspected case, based on symptoms, became a probable case if there were other cases in a child's school. As such, these definitions created a degree of circularity, since suspected cases in a school were sometime enough to make others into probable cases.
Most U.S. schools did not close in the Spring until the surfacing of a probable case or until higher than normal rates of influenza-like-illness were observed at the school [
49,
50]. In New York City, a private high school with students who had recently returned from Mexico became an early focal point for infections, and was closed shortly after cases were confirmed, but after at least eight children were ill. In Texas, however, the Fort Worth school district closed all 144 schools after one confirmed and three others suspected school-aged cases were reported [
51]. In Montgomery County, Maryland, school officials closed a public high school after a single probable case was identified. At the same time, other Maryland schools with probable cases remained open after consulting local health officials, though this decision appears to have been driven by the belief that transmission within the schools had not occurred and secondary cases were unlikely [
52].
We did not observe schools adopting formal thresholds for closure (for example a certain percentage of students identified as ill or absent), although these were used in Japan in Spring 2009 [
53].
Authority and decision-making process for school closures
Much of the variation in school closure decisions in 2009 was due to differences from one jurisdiction to another in whom the legal and practical authority for making decisions was vested. Depending on the jurisdiction, the legal authority to close schools in response to a public health threat may rest with school or health officials, at the state or local level. Additionally, in some jurisdictions, closure authority changes if an emergency has been declared, potentially in different ways depending on the form of the declared emergency [
54]. Many states include school closure measures in their pandemic plans [
55,
56], but the plans are often vague about who has the authority to make the decision. It is not surprising, therefore, that there was substantial variation in decisions to close schools during April and May 2009 and conflict between authorities in some jurisdictions.
In Cibolo, Texas, the first American jurisdiction to close schools, the decision was made by state health officials [
57]. In Fort Worth, Texas, on the other hand, local school officials made the decision to close schools district-wide, on the basis of advice from the local health department [
58]. In Montgomery County, Maryland, shortly after health officials decided to close schools, the School Superintendent protested the decision in a memo to the county school board stating, "We do not believe that this is the right decision given the lack of compelling evidence for continued closure provided to us by state and county health officials" [
59].
In New York City, public schools were closed by the city Schools Chancellor in consultation with the New York City Department of Health and Mental Hygiene [
60]. In addition, the local Roman Catholic archdiocese independently closed some schools on suspicion of 2009 H1N1 cases [
61]. It is unclear whether city officials had the authority to order private schools closed, but they did recommend closure [
62].
In some instances, public health and school officials faced contradictory concerns. One frequent issue that arose dealt with laws mandating the number of instruction days schools must provide to receive state funding. In several states, schools that closed for public health purposes risked losing state education funding or incurring significant costs by extending the school year. Different states responded to this issue in different ways. Rhode Island law authorizes the state's education commissioner to issue waivers of the instruction days requirement for schools closed due to emergencies, and such waivers were granted to schools closed for 2009 H1N1 [
63]. New York law authorized waivers for schools closed due to weather and other disruptions but not epidemics [
64]. The state legislature responded by passing a law authorizing a waiver similar to that in Rhode Island [
65]. Tennessee's legislature passed a similar bill [
66]. Connecticut law authorized a waiver for "extreme circumstances" which the state education commissioner did not include closures for influenza, and the state legislature did not enact a proposed statutory exemption. Closed Connecticut schools had to reschedule classes to meet for 180 days [
67]. State education officials in Alabama similarly indicated that any school closed for influenza would lose a portion of its state funding if it did not reschedule enough classes to meet the minimum number of instruction days [
68].
Official Guidance
Early in the Spring outbreak, state and local health departments followed CDC guidance that districts "consider adopting school dismissal." This guidance largely was influenced by early reports suggesting a high case-fatality rate from 2009 H1N1 and that youth may be at greater than average risk. CDC's initial guidance also suggested that schools with cases stay closed for 14 days [
69], reflecting concerns about the risks of reopening while disease was still being transmitted. Such "guidance," often transmitted through state and local health departments, carries substantial weight in local decision-making. In part because state and local officials do not have the same epidemiological knowledge as CDC, guidance is often regarded as a recommendation.
CDC subsequently revised its guidance, announcing that schools closed under the prior guidance could reopen, but included the caveat that "decisions about school closure should be at the discretion of local authorities based on special circumstances and local considerations, including public concern and the impact of school absenteeism and staffing shortages" [
70].
When additional data showed that the novel 2009 H1N1 was not especially severe, CDC changed its recommendation to keeping ill children at home [
71]. Some schools, however, continued closing "to be on the safe side" [
72].
Officials were often frustrated by frequent changes in the CDC guidance. For example, shortly after CDC increased its closure recommendation from 7 to 14 days, a Fort Worth, Texas, official stated, "The CDC is changing its plans and guidance on a daily basis" [
73]. However, especially at the beginning of a disease outbreak, knowledge about disease severity, transmissibility, and the extent to which people with various underling conditions are at increased risk of complications is necessarily based on limited data. It should not be surprising for this information and the resulting guidance to be revised as more cases accumulate. Indeed in CDC's August 2009 school closure guidance, the agency notes that while it did not currently recommend closure, this could change if the disease's severity increased [
74].
CDC's August 2009 school closure guidance suggests that local authorities make school closure decisions by balancing "the risks of keeping the students in school with the social disruption that school dismissal can cause" [
75]. The guidance notes that "the potential benefits of preemptively dismissing students from school are often outweighed by negative consequences," but also that "school dismissals may be warranted, depending on the disease burden and other conditions." Recognizing that the severity may change, CDC has alternate guidance to be followed in the event of more serious disease [
76].