More than a year after the acute phase of the COVID-19 pandemic, an international group of researchers met in the Canadian Rockies to ask an uncomfortable question: how much do we actually know about what worked? The Pandemic EVIDENCE Collaboration 2025 International Conference, held May 14–16, 2025 at the Banff Centre for Arts and Creativity under the title “Looking at the Pandemic in the Rearview Mirror: Successes, Failures and Unintended Consequences,” brought together over 120 attendees from 13 countries to examine the evidence base behind the non-pharmaceutical interventions (NPIs) that shaped pandemic responses worldwide. A collection of the conference’s peer-reviewed accepted abstracts, workshop and seminar descriptions has now been published in BMC Proceedings.
The collaboration itself is relatively young. Founded in 2023 with its official launch in April 2024 and funded by the McCall MacBain Foundation, it was initiated by independent academics from institutions including Bond University in Australia, the University of Calgary and McMaster University in Canada, Oslo Metropolitan University in Norway, and the University of Oxford in the United Kingdom. Its stated mission is to identify, develop, and implement strategies for generating high-quality evidence on non-pharmacological interventions during public health emergencies, with the aim of better informing policy and care decisions. The Banff meeting was structured around the collaboration’s three pillars, with daily themes covering diagnostics and transmission, interventions and evidence, and practice and policy.
One of the most methodologically significant contributions came from researchers at the University of Edinburgh, the UK Health Security Agency, and the University of Oxford, who presented a rapid review of NPI effectiveness as implemented in the United Kingdom during the COVID-19 pandemic. The team systematically searched MEDLINE, EMBASE, the NIH Covid Portfolio and Corona Central for UK interventional, observational, and mathematical modelling studies published between March 2020 and January 2024, covering outcomes such as COVID-19 cases, transmission, hospitalisation, and mortality. Ninety-seven studies met their criteria: 85 modelling studies, nine observational studies, and three interventional studies, reporting outcomes across 20 NPI categories.
The review’s findings were sobering. Modelling studies were limited by multiple quality issues, and only half of the non-modelling studies were judged to be of high quality. Because of substantial study heterogeneity, the authors synthesised evidence on the direction of effects rather than their size. Evidence for test-and-release strategies for contacts of positive cases, supported by moderate certainty, and for school-related NPIs and universal lockdown, both at low certainty, was suggestive of a protective effect. For the remaining NPI categories, the certainty of evidence was very low or inconclusive. The authors emphasised that these results do not necessarily mean the packages of NPIs implemented in the UK were ineffective; rather, they highlight how difficult it is to evaluate public health interventions after the fact, and they propose strategies for improving evidence generation in future pandemics.
That difficulty is precisely why several conference presentations argued for a different approach going forward: randomised trials of public health and social measures. Researchers from the Norwegian Institute of Public Health outlined the statistical and design challenges such trials face. Parallel designs may not always satisfy ethics committees, simple designs can be inefficient and risk exacerbating trial fatigue and recruitment difficulties, and trials of suboptimal implementation strategies may produce misleading conclusions about safety and effectiveness. Common outcome measures, such as dichotomous events and risk ratios, may fail to capture the dynamic nature of infections and symptoms. The presentation explored methodological options in trial design, powering, outcome selection, analysis, and reporting, arguing that traditional approaches such as publishing a single effect estimate may limit the utility of trial results for future pandemics in different populations.
The Norwegian group also brought empirical work to the meeting. A cluster-randomised trial is evaluating the national Norwegian tool for observation of infection prevention measures in healthcare, known as NOST, which involves observing and recording compliance with hand hygiene recommendations, reporting results back to hospital wards, and following up with training and improvement measures. Data from more than 3,000 hand hygiene observations across 106 wards in 16 hospitals were collected in November 2024, with preliminary results presented at the conference. Separately, the same institute reported on a survey-based study of Norwegian public attitudes toward government-initiated public health policy experiments, motivated by the observation that authorities rarely initiated randomised implementation of NPIs during the pandemic, partly because public acceptance may be a barrier to generating high-quality evidence.
Randomised evidence on masking was also on the agenda. A pragmatic randomised superiority trial of surgical face masks in public spaces, led by researchers at the Norwegian Institute of Public Health and published in the BMJ in 2024, examined the personal protective effect of wearing surgical masks on self-reported respiratory symptoms in adults. Another team from the University of Calgary and Alberta Health Services presented a systematic review of continuous masking policies in hospital settings, identifying 17 eligible studies from 3,691 screened records. Eight studies assessed the impact of masking policies on infection rates in patients, and all eight reported reductions in transmission; of nine studies assessing healthcare workers, seven reported reductions while two found no statistically significant change. The authors cautioned that the evidence relies heavily on observational designs and that quality assessment was still pending at the time of presentation.
Beyond intervention effectiveness, several abstracts documented the unintended consequences of pandemic restrictions. A mixed-methods study from Bhopal, India, surveyed 256 adolescent girls aged 15 to 24 and conducted focus groups and in-depth interviews with 30 participants. Forty-seven percent of the girls reported experiencing some form of violence, with family income loss and movement restrictions identified as key drivers. Among girls with partners, 69 percent reported an increase in intimate partner violence. Seventy-two percent reported anxiety symptoms, 52 percent showed signs of depression, and 43 percent experienced sleep disturbances. The authors argued that lockdowns exacerbated pre-existing gender inequalities and that NPI design must account for how interventions interact with existing vulnerabilities.
Other presentations examined the digital dimensions of pandemic response. An analysis of India’s CoWIN vaccination platform, based on dashboard data and 26 in-depth interviews, found that while the app improved logistics in urban centres, rural areas with limited digital literacy and internet penetration were disproportionately affected: by May 2021, urban areas received 30 vaccine doses for every 12.7 doses administered in rural India. The study also reported that over 100 million women had not received their second vaccine dose by mid-2022, which the authors attributed in part to gendered digital inequities. A complementary analysis of India’s overall vaccination programme described more than one billion registrations on the CoWIN portal by January 2023 and the “Vaccine Maitri” initiative, which supplied over 250 million doses to more than 100 countries.
The burden of long COVID featured prominently as well. A living systematic review and agent-based modelling project from the University of Ottawa and the Public Health Agency of Canada found that reported prevalence of post-COVID-19 condition ranged widely, from 0.9 to 96.2 percent across studies, though prevalence decreased with time after infection. Most individuals returned to work within six months to a year, with a median return-to-work rate of 93.4 percent, but modelling estimated that 66 percent of healthcare workers with the condition would return to work without modifications within one year, compared with 85 percent of those without it, and that affected workers were twice as likely to have lower productivity, miss shifts, or require leave. The heterogeneity of outcome definitions across studies emerged as a recurring theme throughout the conference, alongside calls for standardised reporting, validated quality assessment tools, and better preparedness for generating reliable evidence when the next pandemic arrives.
Subject of Research: Conference abstracts and workshops on generating high-quality evidence for non-pharmaceutical interventions during pandemics
Article Title: Abstracts, workshops and seminars from the Pandemic EVIDENCE Collaboration 2025 International Conference
Article References: Abstracts, workshops and seminars from the Pandemic EVIDENCE Collaboration 2025 International Conference. (2026). BMC Proceedings, 20(29), 35. https://doi.org/10.1186/s12919-026-00388-y (Author names are not included in the publisher’s citation export.)
Image Credits: AI Generated
DOI: 10.1186/s12919-026-00388-y
Keywords: non-pharmaceutical interventions, COVID-19, pandemic preparedness, evidence synthesis, randomized trials, masking policies, lockdown impacts, long COVID, vaccine equity, public health policy, Abstracts, workshops
News Source: Kristina Jarvis. (October 29, 2026). Pandemic EVIDENCE Collaboration gathers global findings on NPIs at Banff conference. Scienmag.
