Counterfactual Modeling Reframes School Closures as a Moving Target in COVID-19 Control
A new counterfactual analysis from the Netherlands makes a blunt point: school closure policies during the COVID-19 pandemic did not operate as a single, fixed intervention. Their impact shifted as transmission moved across age groups, as immunity accumulated unevenly, and as the school system itself changed from one period to the next.
The study used a stochastic age-stratified transmission model fitted to epidemiological and contact survey data to examine how alternative closure and reopening strategies might have altered the pandemic trajectory from early 2020 through late 2021. The modeling framework focused on elementary and secondary schools, then tested how different policy choices could have changed transmission and healthcare burden in the Netherlands.
In practical terms, the findings argue against treating school closure as a uniform lever that produces the same effect at every phase of an outbreak. School policy moved through an environment shaped by changing contact networks, shifting susceptibility, and evolving immunity. The viral pressure on households and communities was not static, so the effect of closing a classroom in one period could differ sharply from the effect of closing it months later.
Transmission Pressure Shifted From Adults to Adolescents to Younger Children
The model captured a clear reordering of who drove transmission over time. Adults dominated early in the pandemic, adolescents carried more of the burden in late 2020 and early 2021, and children became more important in late 2021.
This sequence matters because school policy only affects transmission meaningfully when the age group most responsible for spread is also the age group most affected by the intervention. When adults dominated early transmission, school-based restrictions had a more limited upstream effect than they would later. When adolescents emerged as a larger driver, secondary school policies carried more weight. By late 2021, elementary schools had become more relevant because younger children represented a larger share of the remaining susceptible population.
The epidemiologic logic is straightforward. School settings increase structured contact among large numbers of susceptible individuals, and those contacts can amplify community spread when viral circulation is already high. Once population immunity rises in some age groups but not others, the same school environment can amplify transmission in a different part of the age spectrum. The model reflects that biological and social coupling rather than assuming schools act as a fixed source of spread.
Secondary Schools Mattered More for Hospitalizations in 2020
During 2020, secondary schools had a greater impact on national hospitalizations than elementary schools. That pattern aligns with the model’s finding that adolescents played a larger role in transmission during that phase of the pandemic.
Secondary schools are not simply larger versions of elementary schools. Older students tend to have broader contact networks, more movement between peer groups, and more opportunities to carry infection between school, household, and community settings. Those features can increase the downstream effect on hospitalizations when community transmission is active and immunity remains limited. In the model, the contribution of secondary schools was therefore more visible in the first year of the pandemic, when infection pressure was still largely shaped by uncontrolled spread and limited protection.
A single sentence captures the consequence: when adolescents drove spread, secondary schools moved the national hospitalization curve more than elementary schools did.
Elementary Schools Became More Important in 2021
The balance changed in 2021. Elementary schools gained importance for national hospitalizations, in part because children had lower infection-induced immunity and negligible vaccine-induced immunity compared with adolescents.
That difference is biologically and policy-relevant. Once vaccination reached older age groups, the relative susceptibility of younger children increased. If adolescents were partially protected while younger children remained largely immunologically naive, elementary schools could become a more prominent site of ongoing transmission. The model suggests that closure or reopening decisions made in 2021 would have needed to account for this uneven immunity profile rather than rely on assumptions from earlier phases of the pandemic.
This is one of the most consequential findings in the analysis. School policy is often debated as if the classroom were a constant entity, but the risk pool inside that classroom changes over time. Age-specific immunity, not just school structure, determines which institution sits closer to the center of transmission.
Policy Timing, School Type, and Immunity Determined Effectiveness
The counterfactual simulations showed that school closures could mitigate transmission and reduce healthcare burden, but their effectiveness depended heavily on context. School type, contact patterns, policy timing, and population immunity all influenced the result.
That means the intervention was neither universally strong nor uniformly weak. A closure introduced when an age group was driving spread could have a larger epidemiologic payoff than one imposed after transmission had already shifted elsewhere. Likewise, reopening decisions made after partial population immunity had accrued could carry a different risk profile than those made at the beginning of the pandemic. The model’s time-varying elasticity analysis reinforced this idea by showing that the transmission contribution of school-age groups changed through the study period.
For public health planning, the lesson is not simply whether schools should close or stay open. The more useful question is which school level is amplifying spread, which age group is most susceptible, and how much community immunity already exists when the policy is enacted.
What This Means for Future Pandemic School Policy
The study supports a more targeted and adaptive approach to school policy in future pandemics. Broad closures may still have value, but the model indicates that precision matters: the same action will not have the same effect across all phases of an outbreak.
Targeting the relevant school level may reduce collateral disruption while preserving some public health benefit. If secondary schools are the dominant contributor to transmission in one period, policy should not default to elementary school closure. If younger children become the more important susceptible group later, the strategy may need to pivot. This kind of age-stratified response is especially important when healthcare capacity is under pressure and policymakers are trying to reduce hospitalizations without imposing unnecessary social cost.
The study also reinforces a basic feature of respiratory epidemic control: immunity distribution changes the geometry of risk. Infection-induced immunity, vaccine-induced immunity, and age-related contact patterns all shape whether schools function as transmission accelerators, neutral spaces, or relatively minor contributors to spread.
The model is strong on epidemiologic inference, but its conclusions remain tied to the Netherlands and to the assumptions built into the counterfactual framework. The most clinically useful output is not a universal directive to close or keep open schools. It is a warning that school policy during a pandemic has to move in step with age-specific transmission, immunity patterns, and the timing of viral waves.