Causal Inference Approach Research Articles

Can open peer review improve uptake of preprints into policies? Evidence from a causal inference

Abstract The advantage of no publication time lag had led to the use of preprints as research evidence for public policy development that required a rapid response. However, the opposite side of the lack of publication time lag for preprints was their usual lack of peer review, which was the main reason why preprints were criticized as low quality and unreliable. This study aimed to investigate the effects of open peer review on the uptake of preprints into policies (measured by preprints being cited in policy documents). The coronavirus disease 2019 (COVID-19) preprints from bioRxiv and medRxiv were used as a sample (n = 20,959). A causal inference approach, namely, propensity score matching analysis, was used to examine the dataset. Our study found that open peer review significantly increased the uptake of preprints by policies. Limitations of this study were also discussed. This study could bring insights to researchers and preprint servers in improving the uptake of preprints into policies.

Propensity Score Weighting Analysis of Survival Outcomes Using Pseudo-Observations

Survival outcomes are common in comparative effectiveness studies and require unique handling because they are usually incompletely observed due to right-censoring. A ``once for all'' approach for causal inference with survival outcomes constructs pseudo-observations and allows standard methods such as propensity score weighting to proceed as if the outcomes are completely observed. For a general class of model-free causal estimands with survival outcomes on user-specified target populations, we develop corresponding propensity score weighting estimators based on the pseudo-observations and establish their asymptotic properties. In particular, utilizing the functional delta-method and the von Mises expansion, we derive a new closed-form variance of the weighting estimator that takes into account the uncertainty due to both pseudo-observation calculation and propensity score estimation. This allows valid and computationally efficient inference without resampling. We also prove the optimal efficiency property of the overlap weights within the class of balancing weights for survival outcomes. The proposed methods are applicable to both binary and multiple treatments. Extensive simulations are conducted to explore the operating characteristics of the proposed method versus other commonly used alternatives. We apply the proposed method to compare the causal effects of three popular treatment approaches for prostate cancer patients.

Time-varying living arrangements and suicide death in the general population sample: 14-year causal survival analysis via pooled logistic regression.

While past research suggested that living arrangements are associated with suicide death, no study has examined the impact of sustained living arrangements and the change in living arrangements. Also, previous survival analysis studies only reported a single hazard ratio (HR), whereas the actual HR may change over time. We aimed to address these limitations using causal inference approaches. Multi-point data from a general Japanese population sample were used. Participants reported their living arrangements twice within a 5-year time interval. After that, suicide death, non-suicide death and all-cause mortality were evaluated over 14 years. We used inverse probability weighted pooled logistic regression and cumulative incidence curve, evaluating the association of time-varying living arrangements with suicide death. We also studied non-suicide death and all-cause mortality to contextualize the association. Missing data for covariates were handled using random forest imputation. A total of 86,749 participants were analysed, with a mean age (standard deviation) of 51.7 (7.90) at baseline. Of these, 306 died by suicide during the 14-year follow-up. Persistently living alone was associated with an increased risk of suicide death (risk difference [RD]: 1.1%, 95% confidence interval [CI]: 0.3-2.5%; risk ratio [RR]: 4.00, 95% CI: 1.83-7.41), non-suicide death (RD: 7.8%, 95% CI: 5.2-10.5%; RR: 1.56, 95% CI: 1.38-1.74) and all-cause mortality (RD: 8.7%, 95% CI: 6.2-11.3%; RR: 1.60, 95% CI: 1.42-1.79) at the end of the follow-up. The cumulative incidence curve showed that these associations were consistent throughout the follow-up. Across all types of mortality, the increased risk was smaller for those who started to live with someone and those who transitioned to living alone. The results remained robust in sensitivity analyses. Individuals who persistently live alone have an increased risk of suicide death as well as non-suicide death and all-cause mortality, whereas this impact is weaker for those who change their living arrangements.

Physician Effects in Critical Care: A Causal Inference Approach Through Propensity Weighting with Parametric and Super Learning Methods

Physician performance is critical to caring for patients admitted to the intensive care unit (ICU), who are in life-threatening situations and require high level medical care and interventions. Evaluating physicians is crucial for ensuring a high standard of medical care and fostering continuous performance improvement. The non-randomized nature of ICU data often results in imbalance in patient covariates across physician groups, making direct comparisons of the patients’ survival probabilities for each physician misleading. In this article, we utilize the propensity weighting method to address confounding, achieve covariates balance, and assess physician effects. Due to possible model misspecification, we compare the performance of the propensity weighting methods using both parametric models and super learning methods. When the generalized propensity or the quality function is not correctly specified within the parametric propensity weighting framework, super learning-based propensity weighting methods yield more efficient estimators. We demonstrate that utilizing propensity weighting offers an effective way to assess physician performance, a topic of considerable interest to hospital administrators.

Investigating environmental effects on phonology using diachronic models.

Previous work has proposed various mechanisms by which the environment may affect the emergence of linguistic features. For example, dry air may cause careful control of pitch to be more effortful, and so affect the emergence of linguistic distinctions that rely on pitch such as lexical tone or vowel inventories. Criticisms of these proposals point out that there are both historical and geographic confounds that need to be controlled for. We take a causal inference approach to this problem to design the most detailed test of the theory to date. We analyse languages from the Bantu language family, using a prior geographic-phylogenetic tree of relationships to establish where and when languages were spoken. This is combined with estimates of humidity for those times and places, taken from historical climate models. We then estimate the strength of causal relationships in a causal path model, controlling for various influences of inheritance and borrowing. We find no evidence to support the previous claims that humidity affects the emergence of lexical tone. This study shows how using causal inference approaches lets us test complex causal claims about the cultural evolution of language.

Estimating Soybean Yields using Causal Inference and Deep Learning Approaches with Satellite Remote Sensing Data

Estimating Soybean Yields using Causal Inference and Deep Learning Approaches with Satellite Remote Sensing Data

Bayesian-Based Causal Structure Inference With a Domain Knowledge Prior for Stable and Interpretable Soft Sensing.

Due to the high-stakes nature of industrial processes, there is an immediate and pressing need on soft sensors for stability and interpretability. In this regard, causality-inspired modeling aims to learn causal features corresponding to the direct causes of quality variables, exhibiting great potential in terms of both stability and interpretability. However, most existing causality-inspired methods overlook temporal modeling and domain knowledge integration, which hinders their real-world application in industrial soft sensing. To this end, this article proposes a novel causality-inspired stable long short-term memory (Stable-LSTM), which leverages Bayesian-based causal structure inference and incorporates domain knowledge as a prior to enhance the performance stability and physical interpretability of soft sensors. After extracting temporal features via long short-term memory (LSTM), a Bayesian-based causal structure inference approach is developed by leveraging variational inference to learn the underlying hidden causal structure within the industrial processes. Through a hidden explanation of domain knowledge, a prior distribution is placed on the hidden causal structure, which will greatly enhance the physical interpretability and facilitate the exploration for true causality. Moreover, we also introduce a global sample reweighting strategy to remove spurious correlations and reveal causal effects between time series hidden features and quality variables. Finally, the performance stability and physical interpretability of the proposed Stable-LSTM are verified using a three-phase flow facility and a m-phenylenediamine distillation process. The results show that the Stable-LSTM achieves the highest soft sensing accuracy under distribution shift, and the inferred causal structure exhibits the greatest consistency with the domain knowledge, when compared with the seven existing methods.

Estimating and Mitigating the Congestion Effect of Curbside Pick-ups and Drop-Offs: A Causal Inference Approach

Curb space is one of the busiest areas in urban road networks. Especially in recent years, the rapid increase of ride-hailing trips and commercial deliveries has induced massive pick-ups/drop-offs (PUDOs), which occupy the limited curb space that was designed and built decades ago. These PUDOs could jam curbside utilization and disturb the mainline traffic flow, evidently leading to significant negative societal externalities. However, there is a lack of an analytical framework that rigorously quantifies and mitigates the congestion effect of PUDOs in the system view, particularly with little data support and involvement of confounding effects. To bridge this research gap, this paper develops a rigorous causal inference approach to estimate the congestion effect of PUDOs on general regional networks. A causal graph is set to represent the spatiotemporal relationship between PUDOs and traffic speed, and a double and separated machine learning (DSML) method is proposed to quantify how PUDOs affect traffic congestion. Additionally, a rerouting formulation is developed and solved to encourage passenger walking and traffic flow rerouting to achieve system optimization. Numerical experiments are conducted using real-world data in the Manhattan area. On average, 100 additional units of PUDOs in a region could reduce the traffic speed by 3.70 and 4.54 miles/hour (mph) on weekdays and weekends, respectively. Rerouting trips with PUDOs on curb space could respectively reduce the system-wide total travel time (TTT) by 2.44% and 2.12% in Midtown and Central Park on weekdays. A sensitivity analysis is also conducted to demonstrate the effectiveness and robustness of the proposed framework. Funding: The work described in this paper was supported by the National Natural Science Foundation of China [Grant 52102385], grants from the Research Grants Council of the Hong Kong Special Administrative Region, China [Grants PolyU/25209221 and PolyU/15206322], a grant from the Otto Poon Charitable Foundation Smart Cities Research Institute (SCRI) at the Hong Kong Polytechnic University [Grant P0043552], and a grant from Hong Kong Polytechnic University [Grant P0033933]. S. Qian was supported by a National Science Foundation Grant [Grant CMMI-1931827]. Supplemental Material: The e-companion is available at https://doi.org/10.1287/trsc.2022.0195 .

Perceived autonomy support from healthcare professionals and physical activity among breast cancer survivors: A propensity score analysis.

The majority of women treated for breast cancer are physically inactive although physical activity (PA) could attenuate many adverse effects of cancer and treatment. Autonomy support from healthcare professionals may improve PA initiation, adherence and maintenance. This study aimed to determine, using a causal inference approach, whether or not perceived autonomy support (PAS) from healthcare professionals is associated with light, moderate, and vigorous intensity PA among women treated for breast cancer. Data were drawn from the longitudinal study "Life After Breast Cancer: Moving On" (n = 199). PAS was measured with the Health Care Climate Questionnaire and PA was assessed using GT3X triaxial accelerometers. Associations between PAS and PA were estimated with linear regressions and adjusted estimations were obtained using propensity score-based inverse probability of treatment weights (IPTW). Results reveal no association between PAS and PA of light ([Formula: see text](95%CI) = -0.09 (-0.68, 0.49)), moderate ([Formula: see text] (95%CI) = -0.03 (-0.17, 0.11)), or vigorous ([Formula: see text](95%CI) = 0.00 (-0.03, 0.02)) intensity. Different forms of engagement and support by healthcare professionals should be explored to identify the best intervention targets to encourage women to adopt and maintain regular PA in the cancer continuum.

Causal Analysis of Influence of the Solar Cycle and Latitudinal Solar-Wind Structure on Co-Rotation Forecasts

Studying solar-wind conditions is central to forecasting the impact of space weather on Earth. Under the assumption that the structure of this wind is constant in time and co-rotates with the Sun, solar-wind and thereby space-weather forecasts have been made quite effectively. Such co-rotation forecasts are well studied with decades of observations from STEREO and near-Earth spacecraft. Forecast accuracy is primarily determined by three factors: i) the longitudinal separation of spacecraft from Earth determines the corotation time (and hence forecast lead time) [δ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$\\delta$\\end{document}t] over which the solar wind must be assumed to be constant, ii) the latitudinal separation (or offset) between Earth and spacecraft [δθ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$\\delta\ heta$\\end{document}]] determines the degree to which the same solar wind is being encountered at both locations, and iii) the solar cycle, via the sunspot number (SSN), acts as a proxy for both how fast the solar-wind structure is evolving and how much it varies in latitude. However, the precise dependencies factoring in uncertainties are a mixture of influences from each of these factors. Furthermore, for high-precision forecasts, it is important to understand what drives the forecast accuracy and its uncertainty. Here we present a causal inference approach based on information-theoretic measures to do this. Our framework can compute not only the direct (linear and nonlinear) dependencies of the forecast mean absolute error (MAE) on SSN, Δθ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$\\Delta \ heta $\\end{document}, and Δt\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$\\Delta t$\\end{document}, but also how these individual variables combine to enhance or diminish the MAE. We provide an initial assessment of this with the potential of aiding data assimilation in the future.

Causal inference on strategies to increase conversational engagement for socially isolated older adults

AbstractBackgroundSocial isolation, characterized by a small social network and lack of contact, negatively impact the quality of life. Older adults are at an increased risk for social isolation, which is found to be a risk factor of cognitive decline and dementia. Enhancing social connectedness or interactions has been shown to be an effective intervention for slowing age‐associated cognitive decline and possibly delaying the onset of dementia. Here we aim to develop conversational chatbots which can provide engaging conversations with older adults with mild cognitive impairment (OAwMCI) to mitigate social isolation. To resemble human human interactions, chatbots have to adapt certain human strategies, such as empathy, during conversation.MethodOur study utilized causal structural learning and causal inference approaches to examine the dialogue strategy for engaging OAwMCI. We analyzed transcriptions from a clinical trial, I‐CONECT (ClinicalTrials.gov #: NCT02871921), which involved semi‐structured conversations between facilitators and participants. By using emotion recognition in conversation and dialogue act modeling, we produced a time series data of state (participant’s emotions and length of utterance per conversation turn), intervention (facilitator’s conversational strategies), and outcome (participant’s following emotions and engagement). A R package, pcalg, was used to identify causal relationships between the state, intervention, and outcome variables.ResultGraphical models implying causal relationships between certain factors were generated. Our analysis revealed that facilitators adjusted their dialogue strategies according to the participants' emotions and utterance. For example, facilitators expressed appreciation when a participant showed . Additionally, we found that the dialogue acts from facilitators influenced participants' emotions and engagement in the conversation. For instance, non‐opinion statements and yes/no questions from facilitators may affect participants’ verbal engagement and level of anger during the conversation.ConclusionFindings from this study are in line with real‐world human‐human conversation, and our next step is to explore how to use these findings to improve conversational chatbots. Future work shall also consider other crucial features into the state variable that might influence the dynamics of participants' status during the conversation, such as cognitive capacity and dialogue coherence. Our findings have the potential to inform the development of conversational tools and strategies to reduce social isolation among OAwMCI.

Branched chain amino acids harbor distinct and often opposing effects on health and disease.

The branched chain amino acids (BCAA) leucine, isoleucine, and valine are essential nutrients that have been associated with diabetes, cancers, and cardiovascular diseases. Observational studies suggest that BCAAs exert homogeneous phenotypic effects, but these findings are inconsistent with results from experimental human and animal studies. Hypothesizing that inconsistencies between observational and experimental BCAA studies reflect bias from shared lifestyle and genetic factors in observational studies, we used data from the UK Biobank and applied multivariable Mendelian randomization causal inference methods designed to address these biases. In n = 97,469 participants of European ancestry (mean age = 56.7 years; 54.1% female), we estimate distinct and often opposing total causal effects for each BCAA. For example, of the 117 phenotypes with evidence of a statistically significant total causal effect for at least one BCAA, almost half (44%, n = 52) are associated with only one BCAA. These 52 associations include total causal effects of valine on diabetic eye disease [odds ratio = 1.51, 95% confidence interval (CI) = 1.31, 1.76], valine on albuminuria (odds ratio = 1.14, 95% CI = 1.08, 1.20), and isoleucine on angina (odds ratio = 1.17, 95% CI = 1.31, 1.76). Our results suggest that the observational literature provides a flawed picture of BCAA phenotypic effects that is inconsistent with experimental studies and could mislead efforts developing novel therapeutics. More broadly, these findings motivate the development and application of causal inference approaches that enable 'omics studies conducted in observational settings to account for the biasing effects of shared genetic and lifestyle factors.

The impact of confirmed cases of COVID-19 on residents’ traditional Chinese medicine health literacy: A survey from Gansu Province of China

Since the outbreak of the new crown epidemic in China in early 2020, the number of confirmed cases of COVID-19 has continued to increase, and the Chinese government’s policy of "static management" in the first round of the epidemic may affect the health behavior adjustment of Chinese residents. Using survey data on the TCM health literacy of 4016 residents in China (Gansu Province), a causal inference approach was used to explore the impact of the emergence of confirmed cases of COVID-19 on residents’ TCM health literacy. We found that the emergence of confirmed cases can increase by 3.5%-7.0% in residents’ TCM health literacy. Among them, the TCM health literacy of uneducated residents has not improved significantly, and the residents with secondary education have increased significantly by 8%. For those with higher education, the number of residents increased significantly by 6%. At the same time, the emergence of confirmed cases will increase the residents’ practical TCM health literacy and decrease theoretical TCM health literacy. Through heterogeneity analysis, we explored the impact mechanism of confirmed cases on residents’ TCM health literacy. We believe that the emergence of confirmed cases will make residents more inclined to participate in TCM-free clinics, theme activities, and other ways to acquire TCM knowledge.

Abstract 17900: Target Trial Emulation: Evaluating Cardiac Rehabilitation After TAVR

Introduction: Effectiveness of cardiac rehabilitation (CR) after transcatheter aortic valve replacement (TAVR) has not been examined in randomized trials. We used a novel causal inference approach called target trial emulation to evaluate whether participation in CR after TAVR reduces all-cause mortality. Methods: We emulated a target trial among Medicare beneficiaries age≥ 65 years who had undergone a TAVR from October 2016 through September 2019. The intervention was starting CR within 90 days of discharge (time 0) and the outcome was death from any cause. We created two exact copies of each patient’s data (“clones”) and assigned one clone to the control arm and the other to the intervention arm. This eliminated baseline confounding and immortal time bias. Each clone was censored when not consistent with the assigned treatment strategy (e.g., when a control clone initiated CR). Inverse probability weights accounted for potentially informative censoring. Falsification endpoints examined residual confounding. Results: Among 62,628 Medicare beneficiaries eligible for CR after TAVR, mean age was 80 years, and 43% were women. Of these, 28% initiated CR within the 90-day grace period. After adjusting for demographic, clinical, social, and economic differences, all-cause mortality at 2 years was lower in the CR arm than in the no CR arm (HR 0.74, 95% CI 0.63-0.87, p<0.01) (Figure). Falsification endpoints suggested some residual confounding but absolute differences between the two groups were small. Conclusion Fewer than 1 in 3 eligible patients receive CR after TAVR. In the first-ever target trial emulation of CR in patients undergoing TAVR, patients who initiate CR had lower rates of all-cause death. Ongoing work will examine additional approaches to account for residual confounding and use alternative quasi-experimental study designs. If replicated, our findings suggest that innovation in CR access and delivery is needed to enhance uptake after TAVR.

Causal effects of place, people, and process on rooftop solar adoption through Bayesian inference

While previous studies have established correlations between rooftop solar adoption and various factors, a comprehensive understanding of the underlying causal mechanisms has been limited by the intricate interrelationships among these variables. To address this gap, we propose a Bayesian causal inference approach that examines the interplay of various factors influencing rooftop solar adoption across multiple cities. By employing post-phenomenology, we uncover latent variables encompassing place, people, and process, shedding light on how they shape public responses to emerging energy technologies. We analyze the causal effects of these factors and highlight the significance of housing and built environment attributes in determining energy expenditure and rooftop solar adoption, emphasizing the need for policies that target energy equity. Additionally, we reveal the influence of neighborhood spillovers on adoption, indicating the role of social norms and information diffusion. The observed city-level variability underscores the importance of local contexts and location-specific factors in the adoption process. Furthermore, we highlight the need to consider causal relationships and the indirect effects of people-related attributes mediated through place-related attributes. Overall, these findings contribute to a deeper understanding of the factors shaping rooftop solar adoption via causal modeling and underscore the importance of tailored policies to promote adoption.

Wildfire precursors show complementary predictability in different timescales

In most of the world, conditions conducive to wildfires are becoming more prevalent. Net carbon emissions from wildfires contribute to a positive climate feedback that needs to be monitored, quantified, and predicted. Here we use a causal inference approach to evaluate the influence of top-down weather and bottom-up fuel precursors on wildfires. The top-down dominance on wildfires is more widespread than bottom-up dominance, accounting for 73.3% and 26.7% of regions, respectively. The top-down precursors dominate in the tropical rainforests, mid-latitudes, and eastern Siberian boreal forests. The bottom-up precursors dominate in North American and European boreal forests, and African and Australian savannahs. Our study identifies areas where wildfires are governed by fuel conditions and hence where fuel management practices may be more effective. Moreover, our study also highlights that top-down and bottom-up precursors show complementary wildfire predictability across timescales. Seasonal or interannual predictions are feasible in regions where bottom-up precursors dominate.

A Resampling Approach for Causal Inference on Novel Two-Point Time-Series with Application to Identify Risk Factors for Type-2 Diabetes and Cardiovascular Disease

AbstractTwo-point time-series data, characterized by baseline and follow-up observations, are frequently encountered in health research. We study a novel two-point time-series structure without a control group, which is driven by an observational routine clinical dataset collected to monitor key risk markers of type-2 diabetes (T2D) and cardiovascular disease (CVD). We propose a resampling approach called “I-Rand” for independently sampling one of the two-time points for each individual and making inferences on the estimated causal effects based on matching methods. The proposed method is illustrated with data from a service-based dietary intervention to promote a low-carbohydrate diet (LCD), designed to impact risk of T2D and CVD. Baseline data contain a pre-intervention health record of study participants, and health data after LCD intervention are recorded at the follow-up visit, providing a two-point time-series pattern without a parallel control group. Using this approach we find that obesity is a significant risk factor of T2D and CVD, and an LCD approach can significantly mitigate the risks of T2D and CVD. We provide code that implements our method.

Exploring disease-causing traits for drug repurposing in critically ill COVID-19 patients: A causal inference approach

Despite recent development of vaccines to prevent SARS-CoV-2 infection, treatment of critically ill COVID-19 patients remains an important goal. In principle, genome-wide association studies (GWASs) provide a shortcut to the clinical evidence needed to repurpose existing drugs; however, genes identified frequently lack a causal disease link. We report an alternative method for finding drug repurposing targets, focusing on disease-causing traits beyond immediate disease genetics. Sixty blood cell types and biochemistries, and body mass index, were screened on a cohort of critically ill COVID-19 cases and controls that exhibited mild symptoms after infection, yielding high neutrophil cell count as a possible causal trait for critical illness. Our methodology identified CDK6 and janus kinase (JAK) inhibitors as treatment targets that were validated in an exvivo neutrophil extracellular trap (NET) formation assay. Our methodology demonstrates the increased power for drug target identification by leveraging large disease-causing trait datasets.

Long-term particulate matter exposure and the risk of neurological hospitalization: Evidence from causal inference of a large longitudinal cohort in South China

With limited evidence on the neurological impact of particulate matter (PM) exposure in China, particularly for PM1 which is smaller but more toxic, we conducted a large Chinese cohort study using causal inference approaches to comprehensively clarify such impact. A total of 36,271 participants in southern China were recruited in 2015 and followed up through 2020. We obtained the neurological hospitalizations records by linking the cohort data to the electronic reports from 418 medical institutions across the study area. By using high-resolution PM concentrations from satellite-based spatiotemporal models and the cohort data, we performed marginal structural Cox models under causal assumptions to assess the potential causal links between time-varying PM exposure and neurological hospitalizations. Our findings indicated that increasing PM1, PM2.5, and PM10 concentrations by 1 μg/m³ were associated with higher overall neurological hospitalization risks, with hazard ratios (HRs) of 1.10 (95% confidence interval (CI) 1.04–1.16), 1.09 (95% CI 1.04–1.14), and 1.03 (95% CI 1.00–1.06), respectively. PM1 appeared to have a stronger effect on neurological hospitalization, with a 1% and 7% higher impact compared to PM2.5 and PM10, respectively. Additionally, each 1-μg/m3 increase in the annual PM1 concentration was associated with an elevated risk of hospitalizations for ischemic stroke (HR: 1.15; 95% CI, 1.06–1.26), which tended to be larger than the estimates for PM2.5 (HR: 1.13, 95% CI, 1.04–1.23) and PM10 (HR: 1.05, 95% CI, 1.00–1.09). Furthermore, never-married or female individuals tended be at a greater risk compared with their counterparts. Our study provides important insights into the health impact of particles, particularly smaller particles, on neurological hospitalization risk and highlights the need for clean-air policies that specifically target these particles.

Wildfire exposure and academic performance in Brazil: A causal inference approach for spatiotemporal data

As the frequency and intensity of wildfires are projected to globally amplify due to climate change, there is a growing need to quantify the impact of exposure to wildfires in vulnerable populations such as adolescents. In our study, we applied rigorous causal inference methods to estimate the effect of wildfire exposure on the academic performance of high school students in Brazil between 2009 and 2015. Using longitudinal data from 8,183 high schools across 1,571 municipalities in Brazil, we estimated that the average performance in most academic subjects decreases under interventions that increase wildfire exposure, e.g., a decrease of 1.8 % (p = 0.01) in the natural sciences when increasing the wildfire density from 0.0035 wildfires/km2 (first quantile in the sample) to 0.0222 wildfires/km2 (third quartile). Furthermore, these effects considerably worsened over time. Our findings highlight the adverse impact of wildfires on educational outcomes.