Research report: Better data for better public health decision-making: perspectives on quantifying the intended and unintended impacts of public service modifications

When schools shut, trains stop running, or hospitals postpone routine appointments, the health consequences ripple far beyond the immediate crisis. Yet according to a perspective article published in BMC Global and Public Health, the data needed to understand those consequences are almost never collected in any systematic way. A team of researchers spanning institutions in the United Kingdom and Europe argues that this blind spot has left policymakers unable to answer basic questions about whether public service modifications, whether planned interventions or unplanned disruptions, actually achieve their goals, and at what cost.

Healthcare services have direct effects on morbidity and mortality, while public transport plays a critical role in enabling access to care. Schools, they argue, constitute a form of social infrastructure with well-established links to health and wellbeing. The paper uses the term public service modification to describe changes in the availability, accessibility, capacity, timing, or mode of service provision, whether those changes arise from measures introduced to manage a public health threat or from acute external shocks such as floods or heatwaves.

The case for better data rests partly on the persistent uncertainty surrounding one of the most consequential decisions of the COVID-19 pandemic. School closure was deployed widely as a public health and social measure, and policymakers, researchers, and civil society have repeatedly emphasised that schools should be the last to close and first to open. But the authors ask bluntly whether, in a future pandemic, we would know whether to close schools at all, given the persistent uncertainty surrounding the immediate and longer-term impacts of both illness and school closures. Their answer is no. A primary reason, they write, is that we lack robust evidence on the effectiveness of the measures deployed in the past, and the existing literature, while substantive, remains fragmented and dominated by low-certainty findings. Much of that evidence has come from scenario-based modelling analyses, which, while essential, often rely on limited empirical data beyond routinely collected epidemiological indicators.

The scientific evidence on the effectiveness of school closures for outbreak control, and on their long-term societal consequences, has long been equivocal, particularly in low- and middle-income countries. Without systematic data collection, the authors argue, every new deployment of such measures represents a missed opportunity to learn whether, when, and for whom the disruption works. The same gap applies to unplanned disruptions. Floods and heatwaves can lead to the closure of health facilities, missed clinical appointments, and interruptions to routine services including vaccination, yet service interruptions are rarely recorded systematically, making their health impacts difficult to quantify or compare across events.

The perspective illustrates the range of modifications that matter. During infectious disease outbreaks, measures aimed at containing pathogens that spread through air, droplets, or direct contact alter the operation of shared settings, from closing sports and entertainment venues to limiting the frequency of or density on public transport, implementing travel restrictions, and mandating remote education and work from home. Children are a key consideration because of their enhanced susceptibility to some respiratory pathogens, the practical difficulty of sustaining extensive hygiene measures, and the extensive social mixing that occurs in educational and childcare settings. Beyond closures, other modifications remain part of pandemic preparedness plans, including reducing class sizes and staggering start times or class dismissals. Public service infrastructure can also be mobilised to meet new demands: during the 2009 H1N1 pandemic, pharmacies in England served as collection points for antivirals prescribed through the National Pandemic Flu Service, and schools have routinely been used for targeted vaccination campaigns.

Heatwaves exert their own pressures. Heatwaves in England in 2019, 2022, and 2026 led to the issuance of do-not-travel warnings, speed restrictions, and reduced services for railways, and on some occasions flights were suspended due to runway surface defects. During the 2022 and 2026 heatwaves, many schools in England closed early or modified daily schedules to help pupils and staff avoid the hottest parts of the day. Even without official measures, heat-related absenteeism can disrupt service continuity so thoroughly that, for service users, the operational consequences become effectively indistinguishable from deliberate top-down modification. Increased use of air-conditioning places unprecedented demands on electricity grids, and hospitals have in the past cancelled routine appointments to cope with surges in heat-related ambulance and emergency call-outs.

The unintended consequences can be severe and long-lasting. During the COVID-19 pandemic, surveys consistently showed widespread avoidance of routine healthcare, with over a third of adults in some settings reporting missed check-ups or chronic disease reviews due to movement restrictions or fear of infection. In England, the delivery of routine diabetes care processes fell sharply in the first year of the pandemic, contributing to excess non-COVID-19 mortality among people with chronic conditions. Emerging evidence also shows increased rates of missed appointments during heatwaves. On the education side, an extensive and growing body of literature links unscheduled school closures, even short ones, to measurable learning loss. A 21-week closure, the average duration across 41 countries during the COVID-19 pandemic, is estimated to translate into the loss of almost an entire year’s worth of learning, with projected losses of lifetime earnings and poorer long-term health outcomes that fall disproportionately on pupils in the most deprived communities. School closures also cause productivity loss through parental absenteeism, especially in health and care sectors where women comprise a large proportion of the workforce.

The evidence that does exist has been assembled piecemeal. In 2010, lacking uniform reporting requirements, the Department for Education in England estimated snow-related disruption from local-authority and media websites; closures are now captured through pupil-attendance codes, but published data do not provide an event-level record of individual school closures. Researchers examining the educational effects of COVID-19-related closures in Sicily coded individual regional administrative acts because no harmonised dataset of school-closure days was available, while researchers in the United States have assembled closure lists by emailing school representatives for historical records. By contrast, the isolated instances where data were collected systematically, notably during the COVID-19 pandemic, demonstrated substantial value, underpinning analyses ranging from tracking population behavioural changes before policy announcements to the effect of school closures on parental labour supply.

The pandemic also prompted a proliferation of expert-led initiatives such as ACAPS, the Covid-19 pandemic policy monitor, and Response2covid19, which developed detailed codebooks and applied consistent coding of government measures across countries, often relying on sustained manual curation. Volunteer-driven efforts, most notably the Oxford COVID-19 Government Response Tracker and CoronaNet, mobilised large coding teams and enabled robust comparative evaluations of policy timing, intervention effectiveness, and health-system responses. Crucially, once established, these datasets rapidly generated value well beyond their original remit: researchers in economics, governance, education, mobility, and environmental science repurposed the same structured data to investigate firm resilience, learning losses, travel-behaviour changes, democratic accountability, and even short-term air-quality improvements.

The authors catalogue the obstacles that any sustained system would face. Defining a modification implicitly requires a baseline of normal service delivery, which is difficult where provision is patchy and inconsistent; in such contexts they suggest collecting longitudinal data on service provision so that baselines and deviations can be inferred from observed patterns. Modifications also vary enormously in scale, from the closure of a single school to national transport restrictions, and structured formats risk stripping away crucial context such as whether a service was partially reduced or fully suspended. They propose tiered, modular data collection frameworks in which coarse and detailed observations coexist, and they argue that uncertainties in timing, location, and scope should be explicitly represented, for example by recording time intervals rather than fixed dates and approximate geographic boundaries. Information is scattered across government websites, legal documents, news reports, and social media, and modifications may be announced, revised, reversed, or clarified over time, complicating verification.

Their proposed solution is a reusable data ecosystem centred on a systematic registry of public service modifications, extending recent World Health Organization guidance on monitoring public health and social measures. Each record would capture what the modification was, where and when it occurred, why, how long it lasted, the populations affected, baseline provision where possible, data provenance, quality indicators, and uncertainty. Much of the raw material already exists in distributed surveillance streams such as pupil attendance codes and transport timetables, and common metadata and identifiers could make these sources more findable, accessible, interoperable, and reusable. Analytical power would come from linkage with disease surveillance, socioeconomic indicators, household surveys, vaccination records, and hospital episode statistics, potentially within Trusted Research Environments that protect privacy. The authors propose a three-pronged strategy borrowed from the Coalition for Epidemic Preparedness Innovations: prepare foundational protocols and governance structures, transform the response to future crises by building analytical pipelines with feedback loops between data, modelling, and decision-making, and connect stakeholders across academia, public agencies, industry, media, civil society, and affected communities so the systems are operational before emergencies arise. Small-scale pilots, they suggest, could first demonstrate the value and provide a basis for long-term investment.

Subject of Research: Data infrastructure for quantifying the public health impacts of planned and unplanned public service modifications

Article Title: Better data for better public health decision-making: perspectives on quantifying the intended and unintended impacts of public service modifications

Article References: Bhatia, S., Althobaity, Y. M. S., Chapman, L. A. C., Chemutai, J., Delfin S. Estadilla, C., Gansberger, L., Janoušková, E., Karega, P. N., McCabe, R., Murray-Watson, R. E., Naidoo, T. M., Pritchard, M., She, B., Hallett, T. B., Petherick, A., Nouvellet, P., Cori, A., & Mavrodaris, A. (2026). Better data for better public health decision-making: perspectives on quantifying the intended and unintended impacts of public service modifications. BMC Global and Public Health, 4(1), Article 101. https://doi.org/10.1186/s44263-026-00333-3

Image Credits: AI Generated

DOI: 10.1186/s44263-026-00333-3

Keywords: public health, data infrastructure, school closures, heatwaves, pandemic preparedness, public health and social measures, health services, public transport, registry, COVID-19, evidence gaps, policy evaluation