Semiparametric Bayesian inference for optimal dynamic treatment regimes via dynamic marginal structural models.

Considerable statistical work done on dynamic treatment regimes (DTRs) is in the frequentist paradigm, but Bayesian methods may have much to offer in this setting as they allow for the appropriate representation and propagation of uncertainty, including at the individual level. In this work, we extend the use of recently developed Bayesian methods for Marginal Structural Models to arrive at inference of DTRs. We do this (i) by linking the observational world with a world in which all patients are randomized to a DTR, thereby allowing for causal inference and then (ii) by maximizing a posterior predictive utility, where the posterior distribution has been obtained from nonparametric prior assumptions on the observational world data-generating process. Our approach relies on Bayesian semiparametric inference, where inference about a finite-dimensional parameter is made all while working within an infinite-dimensional space of distributions. We further study Bayesian inference of DTRs in the double robust setting by using posterior predictive inference and the nonparametric Bayesian bootstrap. The proposed methods allow for uncertainty quantification at the individual level, thereby enabling personalized decision-making. We examine the performance of these methods via simulation and demonstrate their utility by exploring whether to adapt HIV therapy to a measure of patients' liver health, in order to minimize liver scarring.

Effect modification of perinatal exposure to air pollution and childhood asthma incidence.

Perinatal exposure to ambient air pollution has been associated with childhood asthma incidence, however, less is known regarding the potential effect modifiers in this association. We examined whether maternal and infant characteristics modified the association between perinatal exposure to air pollution and development of childhood asthma.761 172 births occurring between 2006 and 2012 were identified in the province of Ontario, Canada. Associations between exposure to ambient air pollutants and childhood asthma incidence (up to age 6) were estimated using Cox regression models.110,981 children with asthma were identified. In models adjusted for postnatal exposures, second trimester exposures to particulate matter with a diameter ≤2.5 µm (PM2.5) (Hazard Ratio (HR) per interquartile (IQR) increase=1.07, 95% CI: 1.06-1.09) and nitrogen dioxide (NO2) (HR per IQR increase=1.06, 95% CI: 1.03-1.08) were associated with childhood asthma development. Enhanced impacts were found among children born to mothers with asthma, those who smoked during pregnancy, boys, those born preterm, of low birth weight and among those born to mothers living in urban areas during pregnancy.Prenatal exposure to air pollution may have a differential impact on the risk of asthma development according to maternal and infant characteristics.

Annual Trends in Follow-Up Visits for Pediatric Concussion in Emergency Departments and Physicians' Offices.

OBJECTIVE: To assess whether children and youth with concussion receive follow-up visits in accordance with the recommended guidelines. STUDY DESIGN: We conducted a retrospective, population-based study using linked health administrative data from all concussion-related visits to emergency department and physician offices by children aged 5 through 18 years (range, 5.00-18.99) in Ontario between 2003 and 2013. We analyzed the percentage of children and youth seen for follow-up. The Mann-Kendall test for trends was used to assess a monotonic increasing trend over time in concussion follow-up visits. RESULTS: A total of 126 654 children and youth were evaluated for an index concussion visit. The number of children and youth assessed for concussion follow-up (N = 45 155) has increased significantly over time (P < .001). In 2003, 781 of 7126 patients (11.0%; 95% CI, 10.3-11.7) with an index visit for concussion had a follow-up assessment. By 2013, 6526 of 21 681 (30.1%; 95% CI, 29.5-30.7) patients received follow-up care. CONCLUSIONS: The proportion of children and youth receiving follow-up after an acute concussion has significantly increased between 2003 and 2013. Nevertheless, more than two-thirds of all patients do not seek medical follow-up or clearance as recommended by current concussion guidelines, suggesting that ongoing efforts to improve and monitor compliance with recommended guidelines by patients and physicians are important.

Annual and Seasonal Trends in Ambulatory Visits for Pediatric Concussion in Ontario between 2003 and 2013.

OBJECTIVE: To investigate annual and seasonal trends in physician office and emergency department (ED) visit rates for pediatric concussion in Ontario between 2003 and 2013. STUDY DESIGN: A retrospective, population-based study was conducted using linked health administrative data from all concussion-related visits to ED and physician office by children aged 5 through 18 years. Time series analysis was used to assess whether periodic components exist in the monthly number of concussion-related visits. RESULTS: Over the 11-year study period, there were 176 685 pediatric visits for concussion in EDs and physician offices in Ontario. Standardized concussion-related visits showed a 4.4-fold (95% CI 4.37-4.45) increase per 100 000 from 2003 to 2013, with nearly 35 000 total visits in 2013. Concussion-related visits demonstrated a steep increase from 2010 onward. The greatest increases in standardized visits were in females (6.3-fold, 95% CI 6.23-6.46 vs 3.6-fold, 95% CI 3.56-3.64 in males) and 13-18.99 year olds (5.0-fold, 95% CI 4.93-5.08 vs 4.1-fold, 95% CI 3.99-4.27 in 9-12 years and 2.3-fold, 95% CI 2.23-2.42 in 5-8 years). A strong seasonal variability (R2autoreg = 0.87, P < .01) in the number of concussion-related visits was present, with most occurring in fall and winter. CONCLUSIONS: Pediatric concussion-related ED and physician office visit rates have greatly increased in the last decade, particularly since 2010. Prevention strategies may be targeted at those most at risk and at seasonal-related activities carrying the greatest risk of concussion.

Bayesian inference for optimal dynamic treatment regimes in practice.

In this work, we examine recently developed methods for Bayesian inference of optimal dynamic treatment regimes (DTRs). DTRs are a set of treatment decision rules aimed at tailoring patient care to patient-specific characteristics, thereby falling within the realm of precision medicine. In this field, researchers seek to tailor therapy with the intention of improving health outcomes; therefore, they are most interested in identifying optimal DTRs. Recent work has developed Bayesian methods for identifying optimal DTRs in a family indexed by ψ via Bayesian dynamic marginal structural models (MSMs) (Rodriguez Duque D, Stephens DA, Moodie EEM, Klein MB. Semiparametric Bayesian inference for dynamic treatment regimes via dynamic regime marginal structural models. Biostatistics; 2022. (In Press)); we review the proposed estimation procedure and illustrate its use via the new BayesDTR R package. Although methods in Rodriguez Duque D, Stephens DA, Moodie EEM, Klein MB. (Semiparametric Bayesian inference for dynamic treatment regimes via dynamic regime marginal structural models. Biostatistics; 2022. (In Press)) can estimate optimal DTRs well, they may lead to biased estimators when the model for the expected outcome if everyone in a population were to follow a given treatment strategy, known as a value function, is misspecified or when a grid search for the optimum is employed. We describe recent work that uses a Gaussian process ( G P ) prior on the value function as a means to robustly identify optimal DTRs (Rodriguez Duque D, Stephens DA, Moodie EEM. Estimation of optimal dynamic treatment regimes using Gaussian processes; 2022. Available from: https://doi.org/10.48550/arXiv.2105.12259). We demonstrate how a G P approach may be implemented with the BayesDTR package and contrast it with other value-search approaches to identifying optimal DTRs. We use data from an HIV therapeutic trial in order to illustrate a standard analysis with these methods, using both the original observed trial data and an additional simulated component to showcase a longitudinal (two-stage DTR) analysis.