Frontiers in Public Health

Coronavirus disease 2019 (COVID-19) has put an unprecedented burden to world health, economy and social life with possible long-term consequences. The velocity and mass of this infection pandemic had already overwhelmed every robust health care system in the world. The evidence pertaining to this novel infection pandemic is evolving, so are the challenges in terms of adequate preparedness and response. In this review, we enumerate the strategic and operational domains and build a functional framework for the management of hospital mass infection incidents due to COVID-19 and similar future pandemics. This functional framework could assist health policy maker and health care worker to implement, innovate, and translate preparedness and response to save valuable life and resources.

Frontline rescue workers during the COVID-19 pandemic faced heightened psychological risks, yet existing tools like PHQ-9 and GAD-7 lack crisis-specific relevance and brevity. This study aimed to develop and validate a simplified risk probability scale (SRPS) tailored for rapid self-assessment of depression and anxiety in frontline rescue teams. A cross-sectional survey of 273 frontline workers in Lanzhou, China, was conducted using snowball sampling. Participants completed questionnaires integrating socio-economic, occupational, and psychological metrics, analyzed via univariate, multivariate, and ROC curve analyses. Key findings revealed middle-income earners (3000-4000 CNY/month) faced elevated depression risk [OR=3.666, 95%CI (1.085-12.385), p=0.036], while work stress intolerance strongly predicted anxiety [OR=14.258, 95%CI (4.213-58.983), p<0.001]. The SRPS demonstrated moderate predictive accuracy (depression AUC=0.572, sensitivity=58.2%, specificity=53.7%; anxiety AUC=0.662, sensitivity=72.5%, specificity=64.8%)] but prioritized brevity (10 items) and contextual relevance over diagnostic precision. The tools integration into mobile health platforms offers real-time screening potential, enabling targeted interventions for high-risk groups. This study highlights the necessity of context-specific mental health tools in crisis settings and provides a foundation for scalable, dynamic risk assessment in future public health emergencies.

Lung transplantation recipients (LTx) were susceptible to severe acute respiratory syndrome-corona virus-2 (SARS-Cov-2) and suffered a higher mortality risk than healthy subjects. Here we aim to analyze whether it was appropriate or and valuable to maintain lung transplant programs in medical institutions accepting coronavirus disease 2019 (COVID-19) patients. In this study, the clinical characteristics, laboratory testing and epidemiology survey results of 10 LTx recipients undergoing allograft lung transplantation surgeries in the First Affiliated Hospital of Zhengzhou University during the COVID-19 pandemic were collected. A web-based epidemiology questionnaire was used to collect the information of LTx recipients after discharge. Up to now, none of the LTx recipients or their family members get infected with SARS-CoV-2 during the novel coronavirus pandemic. In conclusion, under the premise of taking appropriate preventive measures during hospitalization and after discharge, the lung transplant program can be maintained in the medical institution that accepts patients with COVID-19.

In the present study, tuberculosis specialists were surveyed to rate the most effective strategies to eliminate TB as a public health problem by 2050. Then were investigated the most promising emerging technologies for the prevention, diagnosis and treatment of tuberculosis (TB) expected to reach the market by 2035. This Technology Foresight study was specifically carried out by means of a web survey closed questionnaire, which was sent to 29,988 TB specialists worldwide. Of these, 2,657 answers were obtained and analysed. Respondents had demonstrated a high level of academic training (PhD), more than 10 years of professional experience, and a great diversity of both areas of knowledge and geographic reach. In the view of experts, the strategies with the greatest potential impact on epidemic TB were a) shorter time between diagnosis and start of treatment of DS and MDR-TB; b) strengthening tuberculosis control actions in the most vulnerable populations; c) shorter and less expensive regimens for drug resistant MDR/XDR-TB. Regarding the strategies with the highest potential for eliminating TB, our data suggests that the biomedical paradigm is the strongest among the specialists. The most promising technologies expected to reach the market by 2035 selected by the specialists were: (1) new drugs of known chemical classes or new chemical classes; (2) new point-of-care diagnostic tests for DS-TB, drug resistant or multidrug resistant (MDR/XDR)-TB and TB Infection (TBI). We contribute by discussing the most promising technologies and strategies for the elimination of TB in light of social determinants of health models and forecasting studies. We conclude by suggesting that the expected emerging technologies ongoing development will not suffice to end TB by 2050.

With respect to reopening the economy as a result of the COVID-19 restrictions, governmental response and messaging have been inconsistent, and policies have varied by state as this is a uniquely polarizing topic. Considering the urgent need to return to normalcy, a method was devised to determine the degree of progress any state has made in containing the spread of COVID-19. Using various measures for each state including mortality, hospitalizations, testing capacity, number of infections and infection rate has allowed for the creation of a composite COVID -19 Reopening Readiness Index. This index can serve as a comprehensive reliable and simple-to-use metric to assess the level of containment in any state and to determine the level of risk in further opening. As states struggle to contain the outbreak and at the same time face great pressure in resuming economic activity, an index that provides a data-driven and objective insight is urgently needed. BackgroundWe are in the midst of a once-in-a-lifetime pandemic. All levels of society and governments are working together to "flatten the curve" of the infection and slow the spread of COVID-19. The universal goal is to mitigate its adverse effects on everyday life across the globe and to reduce the number of fatalities. While a vaccine is being developed, the aim is to limit the number of hospitalizations as not to overwhelm healthcare systems in any given city or country. It is well documented that certain regions and localities are more affected than others. It is imperative that containment efforts utilize state and local data at their disposal to understand the readiness of any given area prior to opening its economy, and the level of restrictions that are needed.

The COVID-19 pandemic caused a significant shift in peoples travel behaviors and distractions while driving. This paper aims to investigate the impacts of the COVID-19 pandemic on distracted driving by comparing their behavior before and during the pandemic (from 3/1/2019 to 3/1/2021) in the state of Maryland using a stated preference online survey. Some 158 people were recruited for the survey. Participants were asked about their risky driving behaviors and self-reported distraction both before and during the pandemic. To analyze the results, the Chi-square and posthoc tests with the Bonferroni adjustment were applied. The results showed that during the pandemic, distraction dropped from 25% to 21%. The highest reported distracted driving behavior during the pandemic was using hands-free cell phones (64%), using GPS (75%), and eating or drinking (57%). The respondents daily trips have significantly decreased - about 44% below prepandemic rates. Moreover, using a binary logistic regression, it was revealed that the odds of becoming distracted among participants who used a handheld cell phone before and during the pandemic were 4.5 and 6.6 times higher than others, respectively. The findings of this study shed light on the causes of distraction before and during the pandemic.

Dental quality measures objectively measure the efficiency and performance of dental providers and organizations. While these measures in conjunction with established benchmarks are used routinely for self-assessment, forecasting them ahead of time in a data-driven and evidence-based manner has the potential to assist in assessing future dental treatment needs, oral disease burden, care utilization patterns, and strategic decision making for sustained performance improvement complementing traditional descriptive visualization dashboards. The present study modeled the temporal trends of four key preventive dental quality measures related to caries prevention (Adult New Caries, Sealants (6-9yrs.), Sealants (12-15yrs.), Fluoride Varnish) sampled monthly from (Dec. 2010 to July 2017) averaged across ten Family Health Center Dental Centers (FQHC), Wisconsin, using auto-regressive integrated moving average time series models. Five-month ahead forecasts along with their 95% confidence levels and mean absolute percentage error were determined across the four measures (Adult New Caries: 1.8%, Sealants (6-9yrs.): 0.90%, Sealants (12-15yrs.): 0.30%, Fluoride Varnish: 0.15%). Model diagnostics revealed auto-regressive integrated moving average models to sufficiently capture the temporal patterns of these measures and the forecast estimates of Adult New Caries and Sealants (12-15yrs.) revealed the need for increased efforts for improved preventive care utilization. Forecasting preventive dental quality measures can provide insights into expected treatment needs ahead of time and can assist in optimal resource and staff allocation with potential to prescribe suitable interventions to shift the trajectory from predicted outcomes to desired outcomes in a targeted manner. While the present study investigated organization level preventive dental quality measures, the time series approach presented is as such generic and expected to translate across similar settings.

BackgroundLung transplantation is the final option for end-stage respiratory diseases. Postoperative monitoring of patients physical condition and performance of appropriate interventions for any abnormalities are important to improve the long-term success of lung transplantation. In Japan, patients handwritten self-management charts are widely used to record data for home spirometry, vital signs, and medication dosages. However, handwritten data are not suitable for assessment. We developed an internet-based real-time monitoring system (LT-FollowUp) that can easily assess patient data and detect any abnormalities that arise without delay. The aim of this pilot study was to examine the acceptability of LT-FollowUp to patients. MethodsThis was a prospective single-arm pilot cohort study. Lung transplant patients were recruited consecutively at regular outpatient visits from October 2020. Patients were instructed to enter their self-measurements (pulmonary functions and vital signs) and immunosuppressant dosages twice a day. Acceptability was assessed by the data filling rate. The changes in filling rate over time were evaluated using a linear mixed-effects model for repeated measurements. ResultsA total of 19 patients were included in the study. There were no significant differences in the filling rates between the handwritten charts and LT-FollowUp. ConclusionLT-FollowUp is an acceptable system for patients. Further development of the LT-FollowUp system may lead to better long-term clinical outcomes of lung transplantation.

ObjectiveKidney transplantation requires a multidisciplinary approach to achieve optimal outcomes. Healthcare fragmentation can negatively impact clinical outcomes; however, this issue remains understudied in low- and middle-income countries (LMICs). This study aimed to assess healthcare fragmentation in kidney transplant patients during their first post-transplant year and evaluate its association with three-year survival among patients enrolled in Colombias contributory healthcare scheme. MethodsA retrospective cohort study was conducted using administrative data from Colombias contributory healthcare scheme. The cohort included kidney transplant recipients (2012-2016) who survived the first post-transplant year. Healthcare fragmentation was measured by the number of unique providers involved in the first year. Patients were categorised into high- and low-fragmentation groups based on the 75th percentile of provider distribution. The primary outcome was three-year survival, analysed using multivariate Cox regression to estimate hazard ratios (HRs), adjusted for age, sex, Charlson Comorbidity Index (CCI), insurer, region, and transplant year. ResultsThe cohort comprised 2,028 kidney transplant patients, with a mean age of 47.7 years (SD: 13.4), 38.7% female, and 68.7% presenting a CCI[&le;]3. Healthcare fragmentation ranged from 1 to 34 providers, with a mean of 8.94 (SD: 6.77). High fragmentation ([&ge;]11 providers) was observed in 30.2% of patients. Three-year mortality was significantly higher in the high-fragmentation group (18%) compared to the low-fragmentation group (12%) (p=0.04). High fragmentation was associated with a 49% increased mortality risk (adjusted HR: 1.49; 95% CI: 1.12-1.97; p=0.01). ConclusionHealthcare fragmentation significantly reduces three-year overall survival in kidney transplant recipients in Colombia. These findings underscore the importance of integrated care models and improved coordination among providers to enhance patient outcomes, particularly in resource-limited settings.

BackgroundThe Canadian province of Nova Scotia recently became the first jurisdiction in North America to pass deemed consent organ donation legislation. This change in legislation generated substantial online discussion, which we analyzed to provide insights on public perception. MethodsWe performed directed content analysis on 2663 user-generated comments appearing on two widely-shared Canadian Broadcasting Company (CBC) articles published online in April 2019. We determined levels of support and opposition in comments and described the specific rhetoric used for doing so. We also performed one-way ANOVA and Pearson chi-square tests to determine how the comments were being received and engaged by other users. ResultsA range of commentary was present in both support and opposition to the changes in legislation. There were more negative than positive comments, and negative commentary generated more replies. Positive comments were received more positively by other users, while negative comments were received more negatively. The total sum of negative comments was greatly influenced by a small number of very active participants. Negative commentary focused more on broad concepts and principles related to government, power, and individual rights rather than specific issues in the Nova Scotian context. Substantial issues of trust in the government and healthcare system were evident. ConclusionsThere were strong positive and negative sentiments expressed in the comments, but the total sum of negativity in the comments was significantly influenced by a small number of commentators. Analysis on the presumed consent concerns can be helpful to inform public outreach efforts. Implication StatementThis research is the first to examine the robust online commentary regarding reactions to Nova Scotias recently-passed deemed consent organ donation law. Our analysis provides extensive detail on the commentary in favour and in opposition to the new law, which has relevance for understanding public perception and designing public outreach.

Background and ObjectiveLung transplantation remains the only definitive treatment for patients with end-stage respiratory failure; however, it is burdened by a substantial risk of post-operative mortality. Current risk stratification methods, such as the Lung Transplant Risk Index, offer limited predictive performance and interpretability. This study introduces a novel predictive model based on topological transformers to assess mortality risk following lung transplantation. The objective is to improve predictive accuracy by capturing complex temporal patterns in clinical data while ensuring model interpretability to inform clinical decisions. MethodsA retrospective cohort study was conducted using clinical data from lung transplant recipients. The model integrates both static and time-dependent clinical variables through a transformer-based architecture that incorporates topological features derived from patients temporal trajectories. Model performance was compared to established methods using a held-out test set. The evaluation metrics included accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curve. Model interpretability was assessed using Shapley Additive explanations to identify and rank the most influential predictors of mortality. ResultsThe proposed model demonstrated superior predictive performance compared to the Lung Transplant Risk Index and other benchmark models. On the test dataset, it achieved an accuracy of 87.4%, sensitivity of 84.1%, and specificity of 89.6%. The model consistently outperformed existing approaches across different subgroups, including age, underlying disease, and transplant type. Shapley-based interpretability analysis revealed that dynamic variables such as early post-operative oxygenation trends, immunosuppressive load, and inflammatory markers were among the most critical contributors to mortality risk. ConclusionsThe integration of topological features within a transformer-based framework significantly enhances the prediction of post-transplant mortality risk. By offering both improved predictive power and model transparency, this approach supports more precise and personalised risk stratification in lung transplantation. These findings highlight the potential of topological transformers as a valuable tool in the broader context of precision medicine and clinical decision support.

Hidradenitis suppurativa (HS) is a chronic condition characterized by painful, inflamed lumps under the skin in intertriginous areas. While social media platforms like TikTok are increasingly raising public awareness about HS, the quality of the information presented remains challenging to assess. Our study aims to analyze the top one hundred TikTok videos focusing on HS clinical information, treatment options, and patient experiences from 2020 to 2024. We evaluate user engagement data, content characteristics, and assess the quality of health information using a validated DISCERN score. Overall, our study suggests that while TikTok serves as a valuable resource for raising awareness and accessing new information about HS, content from healthcare professionals and scientific research garners higher user engagement and contributes to greater awareness of HS.

Ambulatory Care Sensitive Conditions (ACSCs) are conditions for which effective and timely primary health care (PHC) can prevent hospitalizations. They are widely used as a proxy indicator of access to and quality of PHC. Despite their relevance, evidence from Central America remains scarce. This study aimed to quantify the burden, describe the epidemiological profile, and assess temporal trends of ACSCs hospitalizations in Honduras from 2014 to 2024. We conducted a retrospective observational study using national administrative hospital discharge data from all Ministry of Health hospitals. ACSCs were defined using a standardized list of 20 diagnostic groups based on ICD-10 codes. We estimated percentages and sex-age-standardized hospitalization rates per 10,000 inhabitants. Clinical indicators included length of stay (LOS) and in-hospital fatality rates. Temporal trends were evaluated using joinpoint regression models to estimate annual percent changes (APC). Analyses included stratification by age, sex, and disease category. A total of 4,023,944 hospitalizations were analyzed, of which 547,486 (13.6%) were classified as ACSCs. The overall sex-age-standardized rate was 54.1 per 10,000 inhabitants. ACSCs' standardized rates increased between 2014 and 2018 (APC: 2.7%; 95% CI: -2.4; 15.2), declined sharply between 2018 and 2021 (APC: -17.8%; 95% CI: -30.6; -10.3), and increased again between 2021 and 2024 (APC: 15.9%; 95% CI: 4.6; 37.6). Despite this rebound, rates remained below pre-pandemic levels. ACSCs were concentrated among children under 5 years (27.7%) and adults aged 60 years and older (29.9%). Noncommunicable diseases accounted for 56.8% of cases, with diabetes mellitus as the leading cause. Compared with non-ACSCs hospitalizations, ACSCs were associated with longer LOS (4.9 vs. 3.9 days; p <0.001) and higher in-hospital fatality rates (2.4% vs. 1.7%; p <0.001). ACSCs hospitalizations constitute a substantial burden in Honduras and reflect persistent gaps in PHC performance. Strengthening PHC resilience and capacity, particularly for chronic disease management and vulnerable populations, is essential to reduce avoidable hospitalizations and improve health system efficiency and equity.

The integration of Large Language Models (LLMs) like GPT-4 and GPT-3.5 into clinical diagnostics has the potential to transform patient-doctor interactions. However, the readiness of these models for real-world clinical application remains inadequately tested. This paper introduces the Conversational Reasoning Assessment Framework for Testing in Medicine (CRAFT-MD), a novel approach for evaluating clinical LLMs. Unlike traditional methods that rely on structured medical exams, CRAFT-MD focuses on natural dialogues, using simulated AI agents to interact with LLMs in a controlled, ethical environment. We applied CRAFT-MD to assess the diagnostic capabilities of GPT-4 and GPT-3.5 in the context of skin diseases. Our experiments revealed critical insights into the limitations of current LLMs in terms of clinical conversational reasoning, history taking, and diagnostic accuracy. Based on these findings, we propose a comprehensive set of guidelines for future evaluations of clinical LLMs. These guidelines emphasize realistic doctor-patient conversations, comprehensive history taking, open-ended questioning, and a combination of automated and expert evaluations. The introduction of CRAFT-MD marks a significant advancement in LLM testing, aiming to ensure that these models augment medical practice effectively and ethically.

Rare diseases (RD) arent rare collectively, affecting around 300 million people globally and 96 million in India. In low- and middle-income countries like India, policies addressing these diseases have only recently been enacted. In 2021, India launched its first functional RD policy. This study comprehensively maps all stakeholders in the RD ecosystem in India to understand their power positions, influence, and needs, thereby enabling better implementation strategies for the RD policy. We conducted in-depth interviews with various stakeholders to understand their perspectives and supplemented the study with media analysis to reach those who did not respond to interview invitations. Our findings suggest a lack of awareness and knowledge about RDs among healthcare professionals who do not specialize in RDs. Encouraging and formalizing the involvement of RD patient organizations in policy-making is crucial due to their high knowledge, interest, and constructive critical capabilities despite their low power. Another important stakeholder group, local companies, can drive innovation and make treatments accessible for RDs but have much lower power than multinational companies, potentially leading to policies that do not favor local needs.

Because of the fading immunity to COVID-19 and continuous evolution of the SARS-CoV-2 Omicron variants, the next epidemic wave of COVID-19 is inevitable. The Omicron variant has been the cause of several waves of the COVID-19 epidemics in the majority of countries. Thus, lessons from other countries may provide guidance regarding the timing and magnitude of the next COVID-19 wave of the pandemic in China. In this study, the COVID-19 surveillance data from 189 countries that experienced two or more waves of the SARS-CoV-2 Omicron variant were analysed. The median peak timing between the first and second/third waves of the SARS-CoV-2 Omicron variant was 164/243 days. The peaks of the second and third waves were much lower than that of the first wave. The median relative peaks of the second and third compared with the first waves were 14.5% and 11.2%, respectively. The time window between the peak timings of the first and second waves showed no significant rank correlation with the five socioeconomic factors included in this study. However, the relative peak of the second wave increased significantly with gross domestic product per capita (P<0.001), urbanisation rate (P=0.003), population density (P=0.007), and proportion of older adults >65 years (P<0.001), although decreased significantly with the proportion of 0-14 teenagers (P<0.001). In summary, the historical situations and progression of COVID-19 outbreaks in other countries may inform the risk assessment of incoming outbreaks in mainland China; however, the timing and magnitude of the next COVID-19 wave may also be influenced by several unknown factors, including rapid viral evaluation of SARS-CoV-2

Several countries have implemented lockdowns to control their COVID-19 epidemic. However, questions like "where" and "when" still require answers. We assessed the impact of national and regional lockdowns considering the French first epidemic wave of COVID-19 as a case study. In a regional lockdown scenario aimed at preventing intensive care units (ICU) saturation, almost all French regions would have had to implement a lockdown within 10 days and 96% of ICU capacities would have been used. For slowly growing epidemics, with a lower reproduction number, the expected delays between regional lockdowns increases. However, the public health costs associated with these delays tend to grow exponentially with time. In a quickly growing pandemic wave, defining the timing of lockdowns at a regional rather than national level delays by a few days the implementation of a nationwide lockdown but leads to substantially higher morbidity, mortality and stress on the healthcare system.

This study conducted a questionnaire survey among 1,605 citizens across 32 provinces in China to examine public risk perception and panic psychology regarding COVID-19 pandemic information. A predictive model of risk cognition and psychological behaviors was established to analyze public risk perception characteristics and behavioral responses during the late stages of the pandemic. Findings revealed that government preventive measures had the strongest impact on public risk perception, significantly surpassing the effects of infection-related information, recovery-related data, and personally relevant information. Compared to the 2020 risk perception map, dimensions such as "cure rates" and "post-recovery contagiousness" shifted from unfamiliar and uncontrollable to familiar and controllable, indicating substantial improvements in public risk cognition. While public risk awareness and fear levels decreased significantly compared to the initial pandemic phase, perceived control over "post-recovery health impacts" remained insufficient. Empirical analysis demonstrated that both positive and negative pandemic information could elevate public risk perception, which subsequently influenced psychological behaviors through the mediating role of risk cognition. Furthermore, the moderating effect of psychological tension underwent significant changes as the pandemic evolved, highlighting the dynamic adaptability of public socio-psychological and behavioral patterns during major public health crises. This study provides empirical evidence and scientific insights for risk communication strategies in the post-pandemic era.

Based on the 2023 Global Burden of Disease (GBD) database, this study analyzed the global burden of preterm birth from 1990 to 2023 and predicted its development trend by 2050, while exploring the disparities in disease burden across regions with different Socio-demographic Index (SDI) levels, income groups and countries. A retrospective trend analysis was conducted to collect data on preterm birth incidence, prevalence, death and disability-adjusted life years (DALYs) in 204 countries and regions worldwide from 1990 to 2023 from the GBD 2023 database. ARIMA model (p=2,d=1,q=1) and grey prediction model (GM(1,1)) were combined to predict the preterm birth burden from 2023 to 2050. In 2023, preterm birth was the primary cause of the global neonatal disease burden, with its four core indicators significantly higher than other neonatal diseases. From 1990 to 2023, the global incidence, death and DALYs of preterm birth decreased to 0.91, 0.44 and 0.52 times of the 1990 levels respectively, while the prevalence increased to 1.54 times of the baseline. Projection results showed that by 2050, the incidence, death and DALYs of preterm birth would drop to 0.79, 0.08 and 0.32 times of the 2023 levels, and the prevalence would rise to 1.23 times of 2023. Low SDI regions, lower-middle income countries, as well as India and Nigeria, bore the heaviest disease burden. Over the past three decades, the global acute health burden of preterm birth such as death has decreased notably, but the continuous rise in prevalence and severe regional and age disparities remain prominent public health challenges. The 0-6 days and 6-11 months age groups are the key time windows for preterm birth intervention. It is urgent to implement targeted prevention and control measures for low SDI regions and lower-middle income countries to reduce the global burden of preterm birth.

BackgroundWith growing efforts aimed at optimizing health care services by reducing "low value care", medical deprescribing represents a critical policy challenge at the intersection of clinical quality, fiscal sustainability, and environmental stewardship. Despite growing evidence of its benefits, France lacks a comprehensive national framework for systematic medication review and deprescribing implementation. ObjectiveTo identify areas of consensus and divergence among key French stakeholders using an adapted Policy Delphi approach to inform national deprescribing policy development. MethodsAn exploratory survey was conducted among stakeholders across five groups (healthcare professionals, patients, academia, policymakers, and the pharmaceutical industry). Consensus levels were assessed using graded Likert scales and analysed across policy domains, including knowledge and training, collaboration, resources, policy support, and sustainability opportunities. ResultsHigh consensus emerged around knowledge gaps, the need for interprofessional collaboration, and clinical benefits of deprescribing. Moderate consensus existed regarding resource constraints and environmental sustainability. Divergence was observed between professionals/academia and policymakers/industry regarding financial incentives and regulatory readiness. A policy Delphi heatmap revealed specific alignment patterns that could serve as policy entry points. ConclusionsMulti-stakeholder consensus mapping provides an innovative governance tool for identifying actionable policy opportunities and contributes to recent tools aimed at reducing low-value care. High-consensus domains, including training, patient safety, and sustainability, offer immediate entry points for coalition-building. On the contrary, areas of divergence require structured dialogue and iterative policy learning among Frances fragmented governance structures to translate stakeholder alignment into systematic deprescribing implementation.