Rabu, 15 Februari 2012
Getting Schooled: School Closure, Age Distribution, and Pandemic Mitigation
David N. Fisman, MD, MPH
Author Affiliations
From University of Toronto, Toronto, Ontario M5T 3M7, Canada.
Despite the gains in antimicrobial therapy and vaccines that have come in the past 100 years (1), epidemics and pandemics (synchronized, global epidemics) remain an important source of morbidity, mortality, and costs in high-, middle-, and low-income countries. Epidemics can be thought of as self-perpetuating, exponential growth processes; because infections are communicable, the more cases you have, the more cases you will get, as long as the population contains susceptible persons to infect.
Epidemiologists refer to the key index of this type of growth as the reproductive number of an infectious disease—the number of new (incident) cases created by each old (prevalent) case before the prevalent case recovers (2). Reproductive numbers are the product of 3 core components: how infectious a person is, the duration of infectiousness of a person, and how many contacts that person has. Epidemic mitigation strategies that seek to reduce the latter component of the reproductive number (contact between infectious and susceptible persons) are often referred to as social-distancing measures.
Social-distancing measures may include closing schools, suspending religious services, and canceling large public gatherings. A famous study in contrasts with respect to the implementation of social distancing for influenza pandemic control occurred in St. Louis and Philadelphia during the severe influenza A(H1N1) pandemic in 1918 to 1919 (3). Authorities in Philadelphia declined to impose social-distancing measures (including, famously, not canceling a parade through the center of the city that drew large crowds) until the epidemic was severe. In contrast, St. Louis proactively and aggressively restricted religious and social gatherings and closed schools early in its epidemic, and the effect of influenza seems to have been greatly mitigated (3). Whether the divergent courses of St. Louis and Philadelphia were attributable to social-distancing measures or whether the willingness to implement such measures reflected fundamental differences in public health culture is not known. However, an analysis of more U.S. cities during the pandemic (4) suggests that the speed at which social distancing is implemented plays a role in reducing a pandemic's effect.
The 3 subsequent recognized influenza pandemics (in 1957, 1968, and 2009) have been far less severe than that of 1918 to 1919, with the 2009 pandemic being notably mild (5). However, social distancing was considered a component of pandemic response in both the United States and Canada in 2009. One particularly attractive means of social distancing is school closure, and numerous schools were closed because of concern about the spread of influenza, especially early in the 2009 pandemic (6).
The intuitive attractiveness of school closure relates to the particular epidemiology of younger persons in relation to influenza. Children and teenagers seem to play a key role in the propagation of seasonal influenza epidemics (7, 8), and such age effects are even more marked in pandemics (9, 10). The effect of the 1918 and 2009 pandemics was heavily skewed toward younger persons in the population (9, 10). In both seasonal influenza epidemics and influenza pandemics, interventions (such as immunization) aimed at younger persons seem to reduce attack rates in all persons in the population (11, 12). Natural experiments associated with holiday-related school closures (in France) and labor-related closures (in Israel) support the contention that school closure disrupts influenza transmission in children (13–15).
Regardless, school closure as a social-distancing strategy in pandemics remains controversial. The lack of trials or other experimental data on school closure may make available data on school closure seem less credible to decision makers. These individuals may be concerned that closure will simply redistribute children and teenagers to other venues (such as day care centers or shopping malls). Even if closing schools effectively reduces influenza transmission, questions would remain with regard to cost-effectiveness. The need for parents to leave work to provide child care would result in large societal costs and could disrupt the provision of other essential services, such as health care (14, 16). Furthermore, modeling studies (which assume effectiveness of school closure) note that if closure is triggered by incidence exceeding a predetermined threshold, schools would need to be repeatedly closed and reopened in the absence of a vaccine (17).
Several methodological difficulties are associated with studying the effect of school closure as a means of mitigating the spread of influenza. First, when virologically confirmed influenza is the outcome of interest, sampling intensity may change over time. For example, in Ontario, Canada, influenza virus testing was restricted in June 2009 (18), such that subsequent decreases in observed cases of influenza could have been attributed to school closure or to a decrease in testing. Second, school closure is likely to occur in concert with other disease-control measures, both formal (such as the use of prophylactic antiviral therapy) and informal (such as a tendency to avoid handshaking because of concern related to illness); isolating the effect of school closure from the effect of other contemporaneous changes could be difficult. Finally, in many temperate countries, reproductive numbers for influenza are likely to oscillate seasonally (19), such that the end of the school year in late spring or early summer may coincide with a nadir for the transmissibility of influenza.
Bearing these limitations in mind, the 2009 influenza pandemic has now produced at least 2 natural experiments that strongly support the efficacy of school closure as a means of mitigating pandemic influenza transmission. The first of these occurred in Mexico, where an early April spring break, the subsequent reopening of schools, and a late April emergency school closure order provided Chowell and colleagues (20) with an opportunity to evaluate the within-season effect of school closing and reopening on the dynamics of influenza transmission. These investigators found that the reproductive numbers declined to approximately 1 (which is associated with cessation of epidemic growth) when the schools closed and then increased again when the schools reopened. Consistent with school closures being the driver behind the decrease in transmissibility, the relative decrease in cases was concentrated in school-aged children (20).
The second natural experiment occurred in Alberta, Canada, and is the subject of a well-thought-out study by Earn and colleagues in this issue (21). Because virologic testing continued in Alberta through the first wave and much of the second wave of the 2009 influenza pandemic and because school dismissal dates in the province varied by age group in a manner that did not correlate with (and thus was not confounded by) the level of influenza activity within a given age group or region, Earn and colleagues saw an opportunity to evaluate the effect of school closure. Because the dates of dismissal were within 2 weeks of each other, differences in change in incidence between age groups are unlikely to be due to seasonally varying factors. Indeed, school closure as an exposure has effectively been randomized across age groups. The findings are striking: Incidence within a given age group fell sharply within 1 to 2 incubation periods of the dismissal of classes. A best-fit mathematical model included 2 factors: school dismissal and a temperature variable. The second wave of the pandemic was also easily simulated by using a model that considered school opening dates and decreases in temperature.
Although these studies make a strong case for the effectiveness of school closure as a means of mitigating influenza transmission in a pandemic, it is important to note that the timing of influenza waves has substantially varied in earlier pandemics (9). In the case of Alberta, it is less clear that a pandemic wave that occurred in the middle of the (rather cold) Alberta winter could be as easily mitigated via school closure, not only because the seasonal reproductive number would be higher (and may stay above 1 even with school closure, all other things being equal) but also because outdoor environmental conditions would require that children and teenagers continue to spend substantial time indoors, where transmission might still be enhanced by crowding.
The authors also note that the economic costs associated with school closure are likely to be substantial, but correctly indicate that the question of cost is distinct from that of effectiveness. In the face of a future pandemic, decision makers will probably need to perform cost–benefit calculations that carefully consider the virulence of the newly emerged influenza strain against the economic costs of social-distancing measures (16). In a mild pandemic, such as the one in 2009, the costs of school closure could outweigh the benefits (22); however, given the remarkable variability in clinical severity associated with influenza, this may not be the case in future pandemics. Earn and colleagues have demonstrated that school closure represents a practical and effective means to buy time in a future pandemic while vaccine against the newly emerged viral strain is developed and produced.
Next Section
Article and Author Information
Potential Conflicts of Interest: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M11-2637.
Author Affiliations
From University of Toronto, Toronto, Ontario M5T 3M7, Canada.
Despite the gains in antimicrobial therapy and vaccines that have come in the past 100 years (1), epidemics and pandemics (synchronized, global epidemics) remain an important source of morbidity, mortality, and costs in high-, middle-, and low-income countries. Epidemics can be thought of as self-perpetuating, exponential growth processes; because infections are communicable, the more cases you have, the more cases you will get, as long as the population contains susceptible persons to infect.
Epidemiologists refer to the key index of this type of growth as the reproductive number of an infectious disease—the number of new (incident) cases created by each old (prevalent) case before the prevalent case recovers (2). Reproductive numbers are the product of 3 core components: how infectious a person is, the duration of infectiousness of a person, and how many contacts that person has. Epidemic mitigation strategies that seek to reduce the latter component of the reproductive number (contact between infectious and susceptible persons) are often referred to as social-distancing measures.
Social-distancing measures may include closing schools, suspending religious services, and canceling large public gatherings. A famous study in contrasts with respect to the implementation of social distancing for influenza pandemic control occurred in St. Louis and Philadelphia during the severe influenza A(H1N1) pandemic in 1918 to 1919 (3). Authorities in Philadelphia declined to impose social-distancing measures (including, famously, not canceling a parade through the center of the city that drew large crowds) until the epidemic was severe. In contrast, St. Louis proactively and aggressively restricted religious and social gatherings and closed schools early in its epidemic, and the effect of influenza seems to have been greatly mitigated (3). Whether the divergent courses of St. Louis and Philadelphia were attributable to social-distancing measures or whether the willingness to implement such measures reflected fundamental differences in public health culture is not known. However, an analysis of more U.S. cities during the pandemic (4) suggests that the speed at which social distancing is implemented plays a role in reducing a pandemic's effect.
The 3 subsequent recognized influenza pandemics (in 1957, 1968, and 2009) have been far less severe than that of 1918 to 1919, with the 2009 pandemic being notably mild (5). However, social distancing was considered a component of pandemic response in both the United States and Canada in 2009. One particularly attractive means of social distancing is school closure, and numerous schools were closed because of concern about the spread of influenza, especially early in the 2009 pandemic (6).
The intuitive attractiveness of school closure relates to the particular epidemiology of younger persons in relation to influenza. Children and teenagers seem to play a key role in the propagation of seasonal influenza epidemics (7, 8), and such age effects are even more marked in pandemics (9, 10). The effect of the 1918 and 2009 pandemics was heavily skewed toward younger persons in the population (9, 10). In both seasonal influenza epidemics and influenza pandemics, interventions (such as immunization) aimed at younger persons seem to reduce attack rates in all persons in the population (11, 12). Natural experiments associated with holiday-related school closures (in France) and labor-related closures (in Israel) support the contention that school closure disrupts influenza transmission in children (13–15).
Regardless, school closure as a social-distancing strategy in pandemics remains controversial. The lack of trials or other experimental data on school closure may make available data on school closure seem less credible to decision makers. These individuals may be concerned that closure will simply redistribute children and teenagers to other venues (such as day care centers or shopping malls). Even if closing schools effectively reduces influenza transmission, questions would remain with regard to cost-effectiveness. The need for parents to leave work to provide child care would result in large societal costs and could disrupt the provision of other essential services, such as health care (14, 16). Furthermore, modeling studies (which assume effectiveness of school closure) note that if closure is triggered by incidence exceeding a predetermined threshold, schools would need to be repeatedly closed and reopened in the absence of a vaccine (17).
Several methodological difficulties are associated with studying the effect of school closure as a means of mitigating the spread of influenza. First, when virologically confirmed influenza is the outcome of interest, sampling intensity may change over time. For example, in Ontario, Canada, influenza virus testing was restricted in June 2009 (18), such that subsequent decreases in observed cases of influenza could have been attributed to school closure or to a decrease in testing. Second, school closure is likely to occur in concert with other disease-control measures, both formal (such as the use of prophylactic antiviral therapy) and informal (such as a tendency to avoid handshaking because of concern related to illness); isolating the effect of school closure from the effect of other contemporaneous changes could be difficult. Finally, in many temperate countries, reproductive numbers for influenza are likely to oscillate seasonally (19), such that the end of the school year in late spring or early summer may coincide with a nadir for the transmissibility of influenza.
Bearing these limitations in mind, the 2009 influenza pandemic has now produced at least 2 natural experiments that strongly support the efficacy of school closure as a means of mitigating pandemic influenza transmission. The first of these occurred in Mexico, where an early April spring break, the subsequent reopening of schools, and a late April emergency school closure order provided Chowell and colleagues (20) with an opportunity to evaluate the within-season effect of school closing and reopening on the dynamics of influenza transmission. These investigators found that the reproductive numbers declined to approximately 1 (which is associated with cessation of epidemic growth) when the schools closed and then increased again when the schools reopened. Consistent with school closures being the driver behind the decrease in transmissibility, the relative decrease in cases was concentrated in school-aged children (20).
The second natural experiment occurred in Alberta, Canada, and is the subject of a well-thought-out study by Earn and colleagues in this issue (21). Because virologic testing continued in Alberta through the first wave and much of the second wave of the 2009 influenza pandemic and because school dismissal dates in the province varied by age group in a manner that did not correlate with (and thus was not confounded by) the level of influenza activity within a given age group or region, Earn and colleagues saw an opportunity to evaluate the effect of school closure. Because the dates of dismissal were within 2 weeks of each other, differences in change in incidence between age groups are unlikely to be due to seasonally varying factors. Indeed, school closure as an exposure has effectively been randomized across age groups. The findings are striking: Incidence within a given age group fell sharply within 1 to 2 incubation periods of the dismissal of classes. A best-fit mathematical model included 2 factors: school dismissal and a temperature variable. The second wave of the pandemic was also easily simulated by using a model that considered school opening dates and decreases in temperature.
Although these studies make a strong case for the effectiveness of school closure as a means of mitigating influenza transmission in a pandemic, it is important to note that the timing of influenza waves has substantially varied in earlier pandemics (9). In the case of Alberta, it is less clear that a pandemic wave that occurred in the middle of the (rather cold) Alberta winter could be as easily mitigated via school closure, not only because the seasonal reproductive number would be higher (and may stay above 1 even with school closure, all other things being equal) but also because outdoor environmental conditions would require that children and teenagers continue to spend substantial time indoors, where transmission might still be enhanced by crowding.
The authors also note that the economic costs associated with school closure are likely to be substantial, but correctly indicate that the question of cost is distinct from that of effectiveness. In the face of a future pandemic, decision makers will probably need to perform cost–benefit calculations that carefully consider the virulence of the newly emerged influenza strain against the economic costs of social-distancing measures (16). In a mild pandemic, such as the one in 2009, the costs of school closure could outweigh the benefits (22); however, given the remarkable variability in clinical severity associated with influenza, this may not be the case in future pandemics. Earn and colleagues have demonstrated that school closure represents a practical and effective means to buy time in a future pandemic while vaccine against the newly emerged viral strain is developed and produced.
Next Section
Article and Author Information
Potential Conflicts of Interest: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M11-2637.
Langganan:
Postingan (Atom)
