Search

Background: Aortic root smooth muscle cells (SMC) develop from both the second heart field (SHF) and neural crest. Disparate responses to disease-causing Fbn1 variants by these lineages are proposed to promote focal aortic root aneurysm formation in Marfan syndrome (MFS), but lineage-stratified SMC analysis in vivo is lacking. Methods: We generated SHF lineage-traced MFS mice and performed integrated multiomic (single-cell RNA and assay for transposase-accessible chromatin sequencing) analysis stratified by embryological origin. SMC subtypes were spatially identified via RNA in situ hybridization. Response to TWIST1 overexpression was determined via lentiviral transduction in human aortic SMCs. Results: Lineage stratification enabled nuanced characterization of aortic root cells. We identified heightened SHF-derived SMC heterogeneity including a subset of Tnnt2 (cardiac troponin T)-expressing cells distinguished by altered proteoglycan expression. MFS aneurysm-associated SMC phenotypic modulation was identified in both SHF-traced and nontraced (neural crest–derived) SMCs; however, transcriptomic responses were distinct between lineages. SHF-derived modulated SMCs overexpressed collagen synthetic genes and small leucine-rich proteoglycans while nontraced SMCs activated chondrogenic genes. These modulated SMCs clustered focally in the aneurysmal aortic root at the region of SHF/neural crest lineage overlap. Integrated RNA-assay for transposase-accessible chromatin analysis identified enriched Twist1 and Smad2/3/4 complex binding motifs in SHF-derived modulated SMCs. TWIST1 overexpression promoted collagen and SLRP gene expression in vitro, suggesting TWIST1 may drive SHF-enriched collagen synthesis in MFS aneurysm. Conclusions: SMCs derived from both SHF and neural crest lineages undergo phenotypic modulation in MFS aneurysm but are defined by subtly distinct transcriptional responses. Enhanced TWIST1 transcription factor activity may contribute to enriched collagen synthetic pathways SHF-derived SMCs in MFS.

Background Epidemiologists are often concerned with unobserved biases that produce confounding in population-based studies. We introduce a new design approach—‘full matching incorporating an instrumental variable (IV)’ or ‘Full-IV Matching’—and illustrate its utility in reducing observed and unobserved biases to increase inference accuracy. Our motivating example is tailored to a central question in humanitarian emergencies—the difference in sexual violence risk by displacement setting. Methods We conducted a series of 1000 Monte Carlo simulations generated from a population-based survey after the 2010 Haitian earthquake and included earthquake damage severity as an IV and the unmeasured variable of ‘social capital’. We compared standardized mean differences (SMDs) for covariates after different designs to understand potential biases. Mean risk differences (RDs) were used to assess each design’s accuracy in estimating the oracle of the simulated data set. Results Naive analysis and pair matching equivalently performed. Full matching reduced imbalances between exposed and comparison groups across covariates, except for the unobserved covariate of ‘social capital’. Pair and full matching overstated differences in sexual violence risk when displaced to a camp vs a community [pair: RD = 0.13, 95% simulation interval (SI) 0.09–0.16; full: RD = 0.11, 95% SI 0.08–0.14). Full-IV Matching reduced imbalances across observed covariates and importantly ‘social capital’. The estimated risk difference (RD = 0.07, 95% SI 0.03–0.11) was closest to the oracle (RD = 0.06, 95% SI 0.4–0.8). Conclusion Full-IV Matching is a novel approach that is promising for increasing inference accuracy when unmeasured sources of bias likely exist.

An important step for any causal inference study design is understanding the distribution of the treated and control subjects in terms of measured baseline covariates. However, not all baseline variation is equally important. In the observational context, balancing on baseline variation summarized in a propensity score can help reduce bias due to self-selection. In both observational and experimental studies, controlling baseline variation associated with the expected outcomes can help increase the precision of causal effect estimates. We propose a set of visualizations which decompose the space of measured covariates into the different types of baseline variation important to the study design. These ``assignment-control plots'' and variations thereof visually illustrate core concepts of causal inference and suggest new directions for methodological research on study design. As a practical demonstration, we illustrate one application of assignment-control plots to a study of cardiothoracic surgery. While the family of visualization tools for studies of causality is relatively sparse, simple visual tools can be an asset to education, application, and methods development., 17 pages, 8 figures

OBJECTIVE: To evaluate the effect of perioperative allogeneic packed red blood cell (RBC) transfusion during aortic root replacement. METHOD: We reviewed patients undergoing aortic root replacement at our institution between March 2014 and April 2020. In total, 760 patients underwent aortic root replacement, of whom 442 (58%) received a perioperative RBC transfusion. Propensity score matching was used to account for baseline and operative differences resulting in 159 matched pairs. All-cause mortality was assessed with Kaplan–Meier curves. Data were obtained from our institutional Society of Thoracic Surgeons database and chart review. RESULTS: After propensity score matching, the RBC-transfused and -nontransfused groups were similar for all preoperative characteristics. Cardiopulmonary bypass time, crossclamp time, and lowest operative temperature were similar between the transfused and nontransfused groups (standardized mean difference

We consider how to merge a limited amount of data from a randomized controlled trial (RCT) into a much larger set of data from an observational data base (ODB), to estimate an average causal treatment effect. Our methods are based on stratification. The strata are defined in terms of effect moderators as well as propensity scores estimated in the ODB. Data from the RCT are placed into the strata they would have occupied, had they been in the ODB instead. We assume that treatment differences are comparable in the two data sources. Our first "spiked-in" method simply inserts the RCT data into their corresponding ODB strata. We also consider a data-driven convex combination of the ODB and RCT treatment effect estimates within each stratum. Using the delta method and simulations, we identify a bias problem with the spiked-in estimator that is ameliorated by the convex combination estimator. We apply our methods to data from the Women's Health Initiative, a study of thousands of postmenopausal women which has both observational and experimental data on hormone therapy (HT). Using half of the RCT to define a gold standard, we find that a version of the spiked-in estimator yields lower-MSE estimates of the causal impact of HT on coronary heart disease than would be achieved using either a small RCT or the observational component on its own.

Objective To develop prediction models for intensive care unit (ICU) vs non-ICU level-of-care need within 24 hours of inpatient admission for emergency department (ED) patients using electronic health record data. Materials and Methods Using records of 41 654 ED visits to a tertiary academic center from 2015 to 2019, we tested 4 algorithms—feed-forward neural networks, regularized regression, random forests, and gradient-boosted trees—to predict ICU vs non-ICU level-of-care within 24 hours and at the 24th hour following admission. Simple-feature models included patient demographics, Emergency Severity Index (ESI), and vital sign summary. Complex-feature models added all vital signs, lab results, and counts of diagnosis, imaging, procedures, medications, and lab orders. Results The best-performing model, a gradient-boosted tree using a full feature set, achieved an AUROC of 0.88 (95%CI: 0.87–0.89) and AUPRC of 0.65 (95%CI: 0.63–0.68) for predicting ICU care need within 24 hours of admission. The logistic regression model using ESI achieved an AUROC of 0.67 (95%CI: 0.65–0.70) and AUPRC of 0.37 (95%CI: 0.35–0.40). Using a discrimination threshold, such as 0.6, the positive predictive value, negative predictive value, sensitivity, and specificity were 85%, 89%, 30%, and 99%, respectively. Vital signs were the most important predictors. Discussion and Conclusions Undertriaging admitted ED patients who subsequently require ICU care is common and associated with poorer outcomes. Machine learning models using readily available electronic health record data predict subsequent need for ICU admission with good discrimination, substantially better than the benchmarking ESI system. The results could be used in a multitiered clinical decision-support system to improve ED triage.

Aims Since 1968, heart transplantation has become the definitive treatment for patients with end-stage heart failure. We aimed to summarize our experience in heart transplantation at Stanford University since the first transplantation performed over 50 years ago. Methods and results From 6 January 1968 to 30 November 2020, 2671 patients presented to Stanford University for heart transplantation, of which 1958 were adult heart transplantations. Descriptive analyses were performed for patients in 1968–95 (n = 639). Stabilized inverse probability weighting was applied to compare patients in 1996–2006 (n = 356) vs. 2007–19 (n = 515). Follow-up data were updated through 2020. The primary endpoint was all-cause mortality. Prior to weighting, recipients in 2007–19 vs. those in 1996–2006 were older and had heavier burden of chronic diseases. After the application of stabilized inverse probability weighting, the distance organ travelled increased from 84.2 ± 111.1 miles to 159.3 ± 169.9 miles from 1996–2006 to 2007–19. Total allograft ischaemia time also increased over time (199.6 ± 52.7 vs. 225.3 ± 50.0 min). Patients in 2007–19 showed superior survival than those in 1996–2006 with a median survival of 12.1 vs. 11.1 years. Conclusion In this half-century retrospective descriptive study from one of the largest heart transplant programmes in the USA, long-term survival after heart transplantation has improved over time despite increased recipient and donor age, worsening comorbidities, increased technical complexity, and prolonged total allograft ischaemia time. Further investigation is warranted to delineate factors associated with the excellent outcomes observed in this study.

Supplemental Digital Content is available in the text., Background: Many graft configurations are clinically used for valve-sparing aortic root replacement, some specifically focused on recapitulating neosinus geometry. However, the specific impact of such neosinuses on valvular and root biomechanics and the potential influence on long-term durability are unknown. Methods: Using a custom 3-dimenstional–printed heart simulator with porcine aortic roots (n=5), the anticommissural plication, Stanford modification, straight graft (SG), Uni-Graft, and Valsalva graft configurations were tested in series using an incomplete counterbalanced measures design, with the native root as a control, to mitigate ordering effects. Hemodynamic and videometric data were analyzed using linear models with conduit as the fixed effect of interest and valve as a fixed nuisance effect with post hoc pairwise testing using Tukey’s correction. Results: Hemodynamics were clinically similar between grafts and control aortic roots. Regurgitant fraction varied between grafts, with SG and Uni-Graft groups having the lowest regurgitant fractions and anticommissural plication having the highest. Root distensibility was significantly lower in SG versus both control roots and all other grafts aside from the Stanford modification (P≤0.01 for each). All grafts except SG had significantly higher cusp opening velocities versus native roots (P5-fold higher in the anticommissural plication, Stanford modification, and Valsalva graft groups. Conclusions: In this ex vivo modeling system, clinically used valve-sparing aortic root replacement conduit configurations have comparable hemodynamics but differ in biomechanical performance, with the straight graft most closely recapitulating native aortic root biomechanics.

Observational studies often benefit from an abundance of observational units. This can lead to studies that -- while challenged by issues of internal validity -- have inferences derived from sample sizes substantially larger than randomized controlled trials. But is the information provided by an observational unit best used in the analysis phase? We propose the use of `pilot design,' in which observations are expended in the design phase of the study, and the post-treatment information from these observations is used to improve study design. In modern observational studies, which are data rich but control poor, pilot designs can be used to gain information about the structure of post-treatment variation. This information can then be used to improve instrumental variable designs, propensity score matching, doubly-robust estimation, and other observational study designs. We illustrate one version of a pilot design, which aims to reduce within-set heterogeneity and improve performance in sensitivity analyses. This version of a pilot design expends observational units during the design phase to fit a prognostic model, avoiding concerns of overfitting. Additionally, it enables the construction of `Assignment-Control (AC) plots,' which visualize the relationship between propensity and prognostic scores. We first show some examples of these plots, then we demonstrate in a simulation setting how this alternative use of the observations can lead to gains in terms of both treatment effect estimation and sensitivity analyses of unobserved confounding., Comment: 21 pages, 7 figures

ImportanceTesting for coronary heart disease (CHD) in asymptomatic kidney transplant candidates before transplant is widespread and endorsed by various professional societies, but its association with perioperative outcomes is unclear.ObjectiveTo estimate the association of pretransplant CHD testing with rates of death and myocardial infarction (MI).Design, Setting, and ParticipantsThis retrospective cohort study included all adult, first-time kidney transplant recipients from January 2000 through December 2014 in the US Renal Data System with at least 1 year of Medicare enrollment before and after transplant. An instrumental variable (IV) analysis was used, with the program-level CHD testing rate in the year of the transplant as the IV. Analyses were stratified by study period, as the rate of CHD testing varied over time. A combination of US Renal Data System variables and Medicare claims was used to ascertain exposure, IV, covariates, and outcomes.ExposuresReceipt of nonurgent invasive or noninvasive CHD testing during the 12 months preceding kidney transplant.Main Outcomes and MeasuresThe primary outcome was a composite of death or acute MI within 30 days of after kidney transplant.ResultsThe cohort comprised 79 334 adult, first-time kidney transplant recipients (30 147 women [38%]; 25 387 [21%] Black and 48 394 [61%] White individuals; mean [SD] age of 56 [14] years during 2012 to 2014). The primary outcome occurred in 4604 patients (5.3%; 2063 [2.6%] death, 2329 [2.9%] acute MI). During the most recent study period (2012-2014), the CHD testing rate was 56% in patients in the most test-intensive transplant programs (fifth IV quintile) and 24% in patients at the least test-intensive transplant program (first IV quintile, P Conclusions and RelevanceThe results of this cohort study suggest that pretransplant CHD testing was not associated with a reduction in early posttransplant death or acute MI. The study findings potentially challenge the ubiquity of CHD testing before kidney transplant and should be confirmed in interventional studies.

Background. Prior studies have found that residential proximity to upstream oil and gas production is associated with increased risk of adverse health outcomes. Emissions of ambient air pollutants from oil and gas wells in the preproduction and production stages have been proposed as conferring risk of adverse health effects, but the extent of air pollutant emissions and resulting nearby pollution concentrations from wells is not clear.Objectives. We examined the effects of upstream oil and gas preproduction (count of drilling sites) and production (total volume of oil and gas) activities on concentrations of five ambient air pollutants in California.Methods. We obtained data on approximately 1 million daily observations from 314 monitors in the EPA Air Quality System, 2006-2019, including daily concentrations of five routinely monitored ambient air pollutants: PM2.5, CO, NO2, O3, and VOCs. We obtained data on preproduction and production operations from Enverus and the California Geographic Energy Management Division (CalGEM) for all wells in the state. For each monitor and each day, we assessed exposure to upwind preproduction wells and total oil and gas production volume within 10 km. We used a panel regression approach in the analysis and fit adjusted fixed effects linear regression models for each pollutant, controlling for geographic, seasonal, temporal, and meteorological factors.Results. We observed higher concentrations of PM2.5 and CO at monitors within 3 km of preproduction wells, NO2 at monitors at 1-2 km, and O3 at 2-4 km from the wells. Monitors with proximity to increased production volume observed higher concentrations of PM2.5, NO2, and VOCs within 1 km and higher O3 concentrations at 1-2 km. Results were robust to sensitivity analyses.Conclusion. Adjusting for geographic, meteorological, seasonal, and time-trending factors, we observed higher concentrations of ambient air pollutants at air quality monitors in proximity to preproduction wells within 4 km and producing wells within 2 km.

Separated both in academics and practice since the Rockefeller Foundation effort to "liberate" public health from perceived subservience to clinical medicine a century ago, research in public health and clinical medicine have evolved separately. Today, translational research in population health science offers a means of fostering their convergence, with potentially great benefit to both domains. Although evidence that the two fields need not and should not be entirely distinct in their methods and goals has been accumulating for over a decade, the prodigious efforts of biomedical and social sciences over the past year to address the COVID-19 pandemic has placed this unifying approach to translational research in both fields in a new light. Specifically, the coalescence of clinical and population-level strategies to control disease and novel uses of population-level data and tools in research relating to the pandemic have illuminated a promising future for translational research. We exploit this unique window to re-examine how translational research is conducted and where it may be going. We first discuss the

Introduction Suboptimal and differential participant engagement in randomized trials—including retention at primary outcome assessments and attendance at intervention sessions—undermines rigor, internal validity, and trial conclusions. Methods First, this study describes Methods-Motivational Interviewing approach and strategies for implementation. This approach engages potential participants before randomization through interactive, prerequisite orientation sessions that illustrate the scientific rationale behind trial methods in accessible language and use motivational interviewing to diffuse ambivalence about participation. Then, this study examines the potential improvements in retention (proportion of participants assessed at follow-up visits) and attendance (e.g., mean percentage of intervention sessions attended, percentage of participants who attended 0 sessions) in 3 randomized weight-management trials that quickly added prerequisite orientations to their protocols following early signs of suboptimal or differential participant engagement (Supporting Health by Integrating Nutrition and Exercise [2009–2013, n=194]; Get Social [2016–2020, n=217]; GestationaL Weight Gain and Optimal Wellness [2014–2018, n=389]). Using a pre–post analytical design, adjusted estimates from regression models controlling for condition and assessment timepoint (analyses from 2020) are reported. Results After adding prerequisite orientations, all 3 trials attained higher participant engagement. Retention at assessments was 11.4% and 17.3% higher (Get Social and Supporting Health by Integrating Nutrition and Exercise, respectively). Mean percentage of attendance at intervention sessions was 8.8% higher (GestationaL Weight Gain and Optimal Wellness), and 10.1% fewer participants attended 0 intervention sessions (Get Social). Descriptively, all the remaining retention and attendance outcomes were consistently higher but were nonsignificant. Across the trials, adding prerequisite orientations did not impact the proportion of eligible participants enrolled or the baseline demographics. Conclusions The Methods-Motivational Interviewing approach shows promise for increasing the rigor of randomized trials and is readily adaptable to in-person, webinar, and conference call formats. Trial Registration All 3 trials are registered at www.clinicaltrials.gov (Supporting Health by Integrating Nutrition and Exercise: NCT00960414; Get Social: NCT02646618; and GestationaL Weight Gain and Optimal Wellness: NCT02130232).

Background The Centers for Disease Control and Prevention identify antibiotic prescribing stewardship as the most important action to combat increasing antibiotic resistance. Clinicians balance broad empiric antibiotic coverage vs. precision coverage targeting only the most likely pathogens. We investigate the utility of machine learning-based clinical decision support for antibiotic prescribing stewardship. Methods In this retrospective multi-site study, we developed machine learning models that predict antibiotic susceptibility patterns (personalized antibiograms) using electronic health record data of 8342 infections from Stanford emergency departments and 15,806 uncomplicated urinary tract infections from Massachusetts General Hospital and Brigham & Women’s Hospital in Boston. We assessed the trade-off between broad-spectrum and precise antibiotic prescribing using linear programming. Results We find in Stanford data that personalized antibiograms reallocate clinician antibiotic selections with a coverage rate (fraction of infections covered by treatment) of 85.9%; similar to clinician performance (84.3% p = 0.11). In the Boston dataset, the personalized antibiograms coverage rate is 90.4%; a significant improvement over clinicians (88.1% p Conclusions Precision empiric antibiotic prescribing with personalized antibiograms could improve patient safety and antibiotic stewardship by reducing unnecessary use of broad-spectrum antibiotics that breed a growing tide of resistant organisms.

The empirical science of measuring and preventing sexual assault is in its infancy, especially when considering adolescents in developing nations. We analyze pre-intervention data collected in a two-arm cluster-randomized controlled trial of a classroom-based sexual assault prevention program deployed to Class 6 students around Nairobi, Kenya. We estimate that 7.2% of girls were raped in the prior 12 months. We identify school- and individual-level risk factors for rape. We isolate, as much as possible, variation in probability of rape attributable to a subset of these risk factors. We discuss statistical challenges and solutions in each of these domains.

Background Sexual violence against adolescents is prevalent worldwide and results in significant physical and mental injuries as well as loss of economic and personal potential. Urban informal settlements such as those around Nairobi, Kenya have been shown to have especially high incidences of violence. Research has shown that empowerment interventions for female adolescents can reduce sexual assault. However, these interventions have had limited testing in urban informal settlements, with young adolescents, or in coordination with complementary programs for male adolescents. Methods/design This study was a two-arm, parallel, cluster-randomized trial testing a combination of a previously-tested girls’ intervention, IMPower, and a newly revised boys’ intervention, Source of Strength. Clusters were defined as schools within the informal settlements; participants were adolescent girls and boys in class 6, generally between the ages of 10–14 at baseline. Data collection began in January 2016 and continued through December 2018. The primary outcome was the change in incidence of self-reported sexual assault among girls from baseline, compared to a life skills standard of care intervention. Secondary outcomes included experiences of physical and emotional violence, as well as determining the effects of the intervention on self-efficacy, self-esteem, and gender attitudes and beliefs, and how those effects led to changes in experience of sexual assault. For the primary outcome and several of the secondary outcomes, we used an intention to treat estimand. Discussion This was the first randomized controlled trial with longitudinal follow-up of an empowerment self-defense approach to violence prevention for adolescents in informal settlements. The large size and rigorous design supported analysis to understand multiple subgroup experiences in the hypothesized reduction in sexual assault. The study was also unique in its focus on young (10–14 years of age) adolescents and in engaging both boys and girls in separate but coordinated curriculums. The focus on a highly vulnerable and understudied population will make it a significant contribution to the literature on violence prevention. Trial registration Clinical Trials.gov #NCT02771132. Version 3.1 registered May 2017, first participant enrolled January 2017. Retrospectively registered.

OBJECTIVES Neurodevelopmental disability is the most common complication among congenital heart surgery survivors. The Bayley scales are standardized instruments to assess neurodevelopment. The most recent edition (Bayley Scales of Infant and Toddler Development 3rd Edition, Bayley-III) yields better-than-expected scores in typically developing and high-risk infants than the second edition (Bayley Scales of Infant Development 2nd Edition, BSID-II). We compared BSID-II and Bayley-III scores in infants undergoing cardiac surgery. METHODS We evaluated 2198 infants who underwent operations with cardiopulmonary bypass between 1996 and 2009 at 26 institutions. We used propensity score matching to limit confounding by indication in a subset of patients (n = 705). RESULTS Overall, unadjusted Bayley-III motor scores were higher than BSID-II Psychomotor Development Index scores (90.7 ± 17.2 vs 77.6 ± 18.8, P CONCLUSIONS The Bayley-III yielded higher scores than the BSID-II and classified fewer children as severely impaired. The systematic bias towards higher scores with the Bayley-III precludes valid comparisons between early and contemporary cardiac surgery cohorts.

Currently, the most successful prevention interventions against sexual violence (SV) on United States college campuses target modifications at the individual and interpersonal levels. Community-level interventions have been under-developed for college campuses. To address this gap, we employ a citizen science model for understanding campus community factors affecting SV risk. The model, called Our Voice, starts by engaging groups of college students to collect data in their own communities, identifying factors they view as increasing the risk of SV. In facilitated meetings, participants then review and analyze their collective data and use it to generate actionable community-level solutions and advocate for them with local decision-makers. We share findings from a first-generation study of the Our Voice model applied to SV prevention on one college campus, and include recommendations for further research.

In a block-randomized controlled trial, individuals are subdivided by prognostically important baseline characteristics (e.g., age group, sex, or smoking status), prior to randomization. This step reduces the heterogeneity between the treatment groups with respect to the baseline factors most important to determining the outcome, thus enabling more precise estimation of treatment effect. The stratamatch package extends this approach to the observational setting by implementing functions to separate an observational data set into strata and interrogate the quality of different stratification schemes. Once an acceptable stratification is found, treated and control individuals can be matched by propensity score within strata, thereby recapitulating the block-randomized trial design for the observational study. The stratification scheme implemented by stratamatch applies a “pilot design” approach (Aikens, Greaves, and Baiocchi 2019) to estimate a quantity called the prognostic score (Hansen 2008), which is used to divide individuals into strata. The potential benefits of such an approach are twofold. First, stratifying the data enables more computationally efficient matching of large data sets. Second, methodological studies suggest that using a prognostic score to inform the matching process increases the precision of the effect estimate and reduces sensitivity to bias from unmeasured confounding factors (Aikens et al. 2019; Leacy and Stuart 2014; Antonelli, Cefalu, Palmer, and Agniel 2018). A common mistake is to believe reserving more data for the analysis phase of a study is always better. Instead, the stratamatch approach suggests how clever use of data in the design phase of large studies can lead to major benefits in the robustness of the study conclusions.

This study investigates the prevalence and potential etiologies of citation inaccuracies in surgical literature by systematically characterizing influential surgical articles and references they cited.

Background: Nitrate is a widespread groundwater contaminant and a leading cause of drinking water quality violations in California. Associations between nitrate exposure and select adverse birth outcomes have been suggested, but few studies have examined gestational exposures to nitrate and risk of preterm birth (before 37 wk gestation). Objective: We investigated the association between elevated nitrate in drinking water and spontaneous preterm birth through a within-mother retrospective cohort study of births in California. Methods: We acquired over 6 million birth certificate records linked with Office of Statewide Health Planning and Development hospital discharge data for California births from 2000–2011. We used public water system monitoring records to estimate nitrate concentrations in drinking water for each woman’s residence during gestation. After exclusions, we constructed a sample of 1,443,318 consecutive sibling births in order to conduct a within-mother analysis. We used separate conditional logistic regression models to estimate the odds of preterm birth at 20–31 and 32–36 wk, respectively, among women whose nitrate exposure changed between consecutive pregnancies. Results: Spontaneous preterm birth at 20–31 wk was increased in association with tap water nitrate concentrations during pregnancy of 5 to

Background Early career researchers face a hypercompetitive funding environment. To help identify effective intervention strategies for early career researchers, we examined whether first-time NIH R01 applicants who resubmitted their original, unfunded R01 application were more successful at obtaining any R01 funding within 3 and 5 years than original, unfunded applicants who submitted new NIH applications, and we examined whether underrepresented minority (URM) applicants differentially benefited from resubmission. Our observational study is consistent with an NIH working group’s recommendations to develop interventions to encourage resubmission. Methods and findings First-time applicants with US medical school academic faculty appointments who submitted an unfunded R01 application between 2000–2014 yielded 4,789 discussed and 7,019 not discussed applications. We then created comparable groups of first-time R01 applicants (resubmitted original R01 application or submitted new NIH applications) using optimal full matching that included applicant and application characteristics. Primary and subgroup analyses used generalized mixed models with obtaining any NIH R01 funding within 3 and 5 years as the two outcomes. A gamma sensitivity analysis was performed. URM applicants represented 11% and 12% of discussed and not discussed applications, respectively. First-time R01 applicants resubmitting their original, unfunded R01 application were more successful obtaining R01 funding within 3 and 5 years than applicants submitting new applications—for both discussed and not discussed applications: discussed within 3 years (OR 4.17 [95 CI 3.53, 4.93]) and 5 years (3.33 [2.82–3.92]); and not discussed within 3 years (2.81 [2.52, 3.13]) and 5 years (2.47 [2.22–2.74]). URM applicants additionally benefited within 5 years for not discussed applications. Conclusions Encouraging early career researchers applying as faculty at a school of medicine to resubmit R01 applications is a promising potential modifiable factor and intervention strategy. First-time R01 applicants who resubmitted their original, unfunded R01 application had log-odds of obtaining downstream R01 funding within 3 and 5 years 2–4 times higher than applicants who did not resubmit their original application and submitted new NIH applications instead. Findings held for both discussed and not discussed applications.

OBJECTIVE: To assess both the feasibility and potential impact of predicting preventable hospital readmissions using causal machine learning applied to data from the implementation of a readmissions prevention intervention (the Transitions Program). DATA SOURCES: Electronic health records maintained by Kaiser Permanente Northern California (KPNC). STUDY DESIGN: Retrospective causal forest analysis of postdischarge outcomes among KPNC inpatients. Using data from both before and after implementation, we apply causal forests to estimate individual‐level treatment effects of the Transitions Program intervention on 30‐day readmission. These estimates are used to characterize treatment effect heterogeneity and to assess the notional impacts of alternative targeting strategies in terms of the number of readmissions prevented. DATA COLLECTION: 1 539 285 index hospitalizations meeting the inclusion criteria and occurring between June 2010 and December 2018 at 21 KPNC hospitals. PRINCIPAL FINDINGS: There appears to be substantial heterogeneity in patients’ responses to the intervention (omnibus test for heterogeneity p = 2.23 × 10(−7)), particularly across levels of predicted risk. Notably, predicted treatment effects become more positive as predicted risk increases; patients at somewhat lower risk appear to have the largest predicted effects. Moreover, these estimates appear to be well calibrated, yielding the same estimate of annual readmissions prevented in the actual treatment subgroup (1246, 95% confidence interval [CI] 1110‐1381) as did a formal evaluation of the Transitions Program (1210, 95% CI 990‐1430). Estimates of the impacts of alternative targeting strategies suggest that as many as 4458 (95% CI 3925‐4990) readmissions could be prevented annually, while decreasing the number needed to treat from 33 to 23, by targeting patients with the largest predicted effects rather than those at highest risk. CONCLUSIONS: Causal machine learning can be used to identify preventable hospital readmissions, if the requisite interventional data are available. Moreover, our results suggest a mismatch between risk and treatment effects.