Implementation of a Secure Firearm Storage Program in Pediatric Primary Care: A Cluster Randomized Trial.

Importance: Increased secure firearm storage can reduce youth firearm injury and mortality, a leading cause of death for children and adolescents in the US. Despite the availability of evidence-based secure firearm storage programs and recommendations from the American Academy of Pediatrics, few pediatric clinicians report routinely implementing these programs. Objective: To compare the effectiveness of an electronic health record (EHR) documentation template (nudge) and the nudge plus facilitation (ie, clinic support to implement the program; nudge+) at promoting delivery of a brief evidence-based secure firearm storage program (SAFE Firearm) that includes counseling about secure firearm storage and free cable locks during all pediatric well visits. Design, Setting, and Participants: The Adolescent and Child Suicide Prevention in Routine Clinical Encounters (ASPIRE) unblinded parallel cluster randomized effectiveness-implementation trial was conducted from March 14, 2022, to March 20, 2023, to test the hypothesis that, relative to nudge, nudge+ would result in delivery of the firearm storage program to an additional 10% or more of the eligible population, and that this difference would be statistically significant. Thirty pediatric primary care clinics in 2 US health care systems (in Michigan and Colorado) were included, excluding clinics that were not the primary site for participating health care professionals and a subset selected at random due to resource limitations. All pediatric well visits at participating clinics for youth ages 5 to 17 years were analyzed. Interventions: Clinics were randomly assigned in a 1:1 ratio to receive either the nudge or nudge+. Main Outcomes and Measures: Patient-level outcomes were modeled to estimate the primary outcome, reach, which is a visit-level binary indicator of whether the parent received both components of the firearm storage program (counseling and lock), as documented by the clinician in the EHR. Secondary outcomes explored individual program component delivery. Results: A total of 47 307 well-child visits (median [IQR] age, 11.3 [8.1-14.4] years; 24 210 [51.2%] male and 23 091 [48.8%] female) among 46 597 children and 368 clinicians were eligible to receive the firearm storage program during the trial and were included in analyses. Using the intention-to-treat principle, a higher percentage of well-child visits received the firearm storage program in the nudge+ condition (49%; 95% CI, 37-61) compared to nudge (22%; 95% CI, 13-31). Conclusions and Relevance: In this study, the EHR strategy combined with facilitation (nudge+) was more effective at increasing delivery of an evidence-based secure firearm storage program compared to nudge alone. Trial Registration: ClinicalTrials.gov Identifier: NCT04844021.

Investigating racial and gender disparities in virtual randomized clinical trial enrollment: Insights from the BE ACTIVE study.

Randomized clinical trials (RCTs) often suffer from a lack of representation from historically marginalized populations, and it is uncertain whether virtual RCTs (vRCTs) enhance representativeness or if elements of their consent and enrollment processes may instead contribute to underrepresentation of these groups. In this study, we aimed to identify disparities in enrollment demographics in a vRCT, the BE ACTIVE study, which recruited patients within a single health system. We discovered that the proportions of eligible patients who were randomized differed significantly by gender and race/ethnicity (men 1.2%, women 2.0%, P < .001; White 1.8%, Black 1.3%, Hispanic 0.7%, Asian 0.9%; P < .001), and compared with White patients, non-White patients were less likely to have a valid email address on file and were less likely to click on the email link to the study webpage and begin enrollment.

Feedback and Financial Incentives for Reducing Cell Phone Use While Driving: A Randomized Clinical Trial.

Importance: Handheld phone use while driving is a major factor in vehicle crashes. Scalable interventions are needed to encourage drivers not to use their phones. Objective: To test whether interventions involving social comparison feedback and/or financial incentives can reduce drivers' handheld phone use. Design, Setting, and Participants: In a randomized clinical trial, interventions were administered nationwide in the US via a mobile application in the context of a usage-based insurance program (Snapshot Mobile application). Customers were eligible to be invited to participate in the study if enrolled in the usage-based insurance program for 30 to 70 days. The study was conducted from May 13 to June 30, 2019. Analysis was completed December 22, 2023. Interventions: Participants were randomly assigned to 1 of 6 trial arms for a 7-week intervention period: (1) control; (2) feedback, with weekly push notification about their handheld phone use compared with that of similar others; (3) standard incentive, with a maximum $50 award at the end of the intervention based on how their handheld phone use compared with similar others; (4) standard incentive plus feedback, combining interventions of arms 2 and 3; (5) reframed incentive plus feedback, with a maximum $7.15 award each week, framed as participant's to lose; and (6) doubled reframed incentive plus feedback, a maximum $14.29 weekly loss-framed award. Main Outcome and Measure: Proportion of drive time engaged in handheld phone use in seconds per hour (s/h) of driving. Analyses were conducted with the intention-to-treat approach. Results: Of 17 663 customers invited by email to participate, 2109 opted in and were randomized. A total of 2020 drivers finished the intervention period (68.0% female; median age, 30 [IQR, 25-39] years). Median baseline handheld phone use was 216 (IQR, 72-480) s/h. Relative to control, feedback and standard incentive participants did not reduce their handheld phone use. Standard incentive plus feedback participants reduced their use by -38 (95% CI, -69 to -8) s/h (P = .045); reframed incentive plus feedback participants reduced their use by -56 (95% CI, -87 to -26) s/h (P < .001); and doubled reframed incentive plus feedback participants reduced their use by -42 s/h (95% CI, -72 to -13 s/h; P = .007). The 5 active treatment arms did not differ significantly from each other. Conclusions and Relevance: In this randomized clinical trial, providing social comparison feedback plus incentives reduced handheld phone use while individuals were driving. Trial Registration: ClinicalTrials.gov Identifier: NCT03833219.

Optimal refinement of strata to balance covariates.

What is the best way to split one stratum into two to maximally reduce the within-stratum imbalance in many covariates? We formulate this as an integer program and approximate the solution by randomized rounding of a linear program. A linear program may assign a fraction of a person to each refined stratum. Randomized rounding views fractional people as probabilities, assigning intact people to strata using biased coins. Randomized rounding is a well-studied theoretical technique for approximating the optimal solution of certain insoluble integer programs. When the number of people in a stratum is large relative to the number of covariates, we prove the following new results: (i) randomized rounding to split a stratum does very little randomizing, so it closely resembles the linear programming relaxation without splitting intact people; (ii) the linear relaxation and the randomly rounded solution place lower and upper bounds on the unattainable integer programming solution; and because of (i), these bounds are often close, thereby ratifying the usable randomly rounded solution. We illustrate using an observational study that balanced many covariates by forming matched pairs composed of 2016 patients selected from 5735 using a propensity score. Instead, we form 5 propensity score strata and refine them into 10 strata, obtaining excellent covariate balance while retaining all patients. An R package optrefine at CRAN implements the method. Supplementary materials are available online.

Effect of Gamification, Financial Incentives, or Both to Increase Physical Activity Among Patients at High Risk of Cardiovascular Events: The BE ACTIVE Randomized Controlled Trial.

BACKGROUND: Physical activity is associated with a lower risk of major adverse cardiovascular events, but few individuals achieve guideline-recommended levels of physical activity. Strategies informed by behavioral economics increase physical activity, but their longer-term effectiveness is uncertain. We sought to determine the effect of behaviorally designed gamification, loss-framed financial incentives, or their combination on physical activity compared with attention control over 12-month intervention and 6-month postintervention follow-up periods. METHODS: Between May 2019 and January 2024, participants with clinical atherosclerotic cardiovascular disease or a 10-year risk of myocardial infarction, stroke, or cardiovascular death of ≥7.5% by the Pooled Cohort equation were enrolled in a pragmatic randomized clinical trial. Participants received a wearable device to track daily steps, established a baseline, selected a step goal increase, and were randomly assigned to control (n=151), behaviorally designed gamification (n=304), loss-framed financial incentives (n=302), or gamification+financial incentives (n=305). The primary outcome of the trial was the change in mean daily steps from baseline through the 12-month intervention period. RESULTS: A total of 1062 patients (mean±SD age, 67±8; 61% female; 31% non-White) were enrolled. Compared with control subjects, participants had significantly greater increases in mean daily steps from baseline during the 12-month intervention in the gamification arm (adjusted difference, 538.0 [95% CI, 186.2-889.9]; P=0.0027), financial incentives arm (adjusted difference, 491.8 [95% CI, 139.6-844.1]; P=0.0062), and gamification+financial incentives arm (adjusted difference, 868.0 [95% CI, 516.3-1219.7]; P<0.0001). During the 6-month follow-up, physical activity remained significantly greater in the gamification+financial incentives arm than in the control arm (adjusted difference, 576.2 [95% CI, 198.5-954]; P=0.0028), but it was not significantly greater in the gamification (adjusted difference, 459.8 [95% CI, 82.0-837.6]; P=0.0171) or financial incentives (adjusted difference, 327.9 [95% CI, -50.2 to 706]; P=0.09) arms after adjustment for multiple comparisons. CONCLUSIONS: Behaviorally designed gamification, loss-framed financial incentives, and the combination of both increased physical activity compared with control over a 12-month intervention period, with the largest effect in gamification+financial incentives. These interventions could be a useful component of strategies to reduce cardiovascular risk in high-risk patients. REGISTRATION: URL: https://clinicaltrials.gov; Unique Identifier: NCT03911141.

Default Palliative Care Consultation for Seriously Ill Hospitalized Patients: A Pragmatic Cluster Randomized Trial.

Importance: Increasing inpatient palliative care delivery is prioritized, but large-scale, experimental evidence of its effectiveness is lacking. Objective: To determine whether ordering palliative care consultation by default for seriously ill hospitalized patients without requiring greater palliative care staffing increased consultations and improved outcomes. Design, Setting, and Participants: A pragmatic, stepped-wedge, cluster randomized trial was conducted among patients 65 years or older with advanced chronic obstructive pulmonary disease, dementia, or kidney failure admitted from March 21, 2016, through November 14, 2018, to 11 US hospitals. Outcome data collection ended on January 31, 2019. Intervention: Ordering palliative care consultation by default for eligible patients, while allowing clinicians to opt-out, was compared with usual care, in which clinicians could choose to order palliative care. Main Outcomes and Measures: The primary outcome was hospital length of stay, with deaths coded as the longest length of stay, and secondary end points included palliative care consult rate, discharge to hospice, do-not-resuscitate orders, and in-hospital mortality. Results: Of 34 239 patients enrolled, 24 065 had lengths of stay of at least 72 hours and were included in the primary analytic sample (10 313 in the default order group and 13 752 in the usual care group; 13 338 [55.4%] women; mean age, 77.9 years). A higher percentage of patients in the default order group received palliative care consultation than in the standard care group (43.9% vs 16.6%; adjusted odds ratio [aOR], 5.17 [95% CI, 4.59-5.81]) and received consultation earlier (mean [SD] of 3.4 [2.6] days after admission vs 4.6 [4.8] days; P < .001). Length of stay did not differ between the default order and usual care groups (percent difference in median length of stay, -0.53% [95% CI, -3.51% to 2.53%]). Patients in the default order group had higher rates of do-not-resuscitate orders at discharge (aOR, 1.40 [95% CI, 1.21-1.63]) and discharge to hospice (aOR, 1.30 [95% CI, 1.07-1.57]) than the usual care group, and similar in-hospital mortality (4.7% vs 4.2%; aOR, 0.86 [95% CI, 0.68-1.08]). Conclusions and Relevance: Default palliative care consult orders did not reduce length of stay for older, hospitalized patients with advanced chronic illnesses, but did improve the rate and timing of consultation and some end-of-life care processes. Trial Registration: ClinicalTrials.gov Identifier: NCT02505035.

MRI-Based Brain Volume Scoring in Cerebral Malaria Is Externally Valid and Applicable to Lower-Resolution Images.

BACKGROUND AND PURPOSE: Children with cerebral malaria have an elevated risk of mortality and neurologic morbidity. Both mortality and morbidity are associated with initially increased brain volume on MR imaging, as graded by the Brain Volume Score, a subjective ordinal rating scale created specifically for brain MRIs in children with cerebral malaria. For the Brain Volume Score to be more widely clinically useful, we aimed to determine its independent reproducibility and whether it can be applicable to lower-resolution MRIs. MATERIALS AND METHODS: To assess the independent reproducibility of the Brain Volume Score, radiologists not associated with the initial study were trained to score MRIs from a new cohort of patients with cerebral malaria. These scores were then compared with survival and neurologic outcomes. To assess the applicability to lower-resolution MRI, we assigned Brain Volume Scores to brain MRIs degraded to simulate a very-low-field (64 mT) portable scanner and compared these with the original scores assigned to the original nondegraded MRIs. RESULTS: Brain Volume Scores on the new cohort of patients with cerebral malaria were highly associated with outcomes (OR for mortality = 16, P < .001). Scoring of the simulated degraded images remained consistent with the Brain Volume Scores assigned to the original higher-quality (0.35 T) images (intraclass coefficients > 0.86). CONCLUSIONS: Our findings demonstrate that the Brain Volume Score is externally valid in reproducibly predicting outcomes and can be reliably assigned to lower-resolution images.

Sensitivity Analysis for Observational Studies with Recurrent Events.

We conduct an observational study of the effect of sickle cell trait Haemoglobin AS (HbAS) on the hazard rate of malaria fevers in children. Assuming no unmeasured confounding, there is strong evidence that HbAS reduces the rate of malarial fevers. Since this is an observational study, however, the no unmeasured confounding assumption is strong. A sensitivity analysis considers how robust a conclusion is to a potential unmeasured confounder. We propose a new sensitivity analysis method for recurrent event data and apply it to the malaria study. We find that for the causal conclusion that HbAS is protective against malarial fevers to be overturned, the hypothesized unmeasured confounder must be as influential as all but one of the measured confounders.

Marshall Joffe's Contributions to Causal Inference, Biostatistics, and Epidemiology.

We pay tribute to Marshall Joffe, PhD, and his substantial contributions to the field of causal inference with focus in biostatistics and epidemiology. By compiling narratives written by us, his colleagues, we not only present highlights of Marshall's research and their significance for causal inference but also offer a portrayal of Marshall's personal accomplishments and character. Our discussion of Marshall's research notably includes (but is not limited to) handling of posttreatment variables such as noncompliance, employing G-estimation for treatment effects on failure-time outcomes, estimating effects of time-varying exposures subject to time-dependent confounding, and developing a causal framework for case-control studies. We also provide a description of some of Marshall's unpublished work, which is accompanied by a bonus anecdote. We discuss future research directions related to Marshall's research. While Marshall's impact in causal inference and the world outside of it cannot be wholly captured by our words, we hope nonetheless to present some of what he has done for our field and what he has meant to us and to his loved ones.

A mixed model approach to estimate the survivor average causal effect in cluster-randomized trials.

In many medical studies, the outcome measure (such as quality of life, QOL) for some study participants becomes informatively truncated (censored, missing, or unobserved) due to death or other forms of dropout, creating a nonignorable missing data problem. In such cases, the use of a composite outcome or imputation methods that fill in unmeasurable QOL values for those who died rely on strong and untestable assumptions and may be conceptually unappealing to certain stakeholders when estimating a treatment effect. The survivor average causal effect (SACE) is an alternative causal estimand that surmounts some of these issues. While principal stratification has been applied to estimate the SACE in individually randomized trials, methods for estimating the SACE in cluster-randomized trials are currently limited. To address this gap, we develop a mixed model approach along with an expectation-maximization algorithm to estimate the SACE in cluster-randomized trials. We model the continuous outcome measure with a random intercept to account for intracluster correlations due to cluster-level randomization, and model the principal strata membership both with and without a random intercept. In simulations, we compare the performance of our approaches with an existing fixed-effects approach to illustrate the importance of accounting for clustering in cluster-randomized trials. The methodology is then illustrated using a cluster-randomized trial of telecare and assistive technology on health-related QOL in the elderly.

Racial Disparities in Length of Stay Among Severely Ill Patients Presenting With Sepsis and Acute Respiratory Failure.

Importance: Although racial and ethnic minority patients with sepsis and acute respiratory failure (ARF) experience worse outcomes, how patient presentation characteristics, processes of care, and hospital resource delivery are associated with outcomes is not well understood. Objective: To measure disparities in hospital length of stay (LOS) among patients at high risk of adverse outcomes who present with sepsis and/or ARF and do not immediately require life support and to quantify associations with patient- and hospital-level factors. Design, Setting, and Participants: This matched retrospective cohort study used electronic health record data from 27 acute care teaching and community hospitals across the Philadelphia metropolitan and northern California areas between January 1, 2013, and December 31, 2018. Matching analyses were performed between June 1 and July 31, 2022. The study included 102 362 adult patients who met clinical criteria for sepsis (n = 84 685) or ARF (n = 42 008) with a high risk of death at the time of presentation to the emergency department but without an immediate requirement for invasive life support. Exposures: Racial or ethnic minority self-identification. Main Outcomes and Measures: Hospital LOS, defined as the time from hospital admission to the time of discharge or inpatient death. Matches were stratified by racial and ethnic minority patient identity, comparing Asian and Pacific Islander patients, Black patients, Hispanic patients, and multiracial patients with White patients in stratified analyses. Results: Among 102 362 patients, the median (IQR) age was 76 (65-85) years; 51.5% were male. A total of 10.2% of patients self-identified as Asian American or Pacific Islander, 13.7% as Black, 9.7% as Hispanic, 60.7% as White, and 5.7% as multiracial. After matching racial and ethnic minority patients to White patients on clinical presentation characteristics, hospital capacity strain, initial intensive care unit admission, and the occurrence of inpatient death, Black patients experienced longer LOS relative to White patients in fully adjusted matches (sepsis: 1.26 [95% CI, 0.68-1.84] days; ARF: 0.97 [95% CI, 0.05-1.89] days). Length of stay was shorter among Asian American and Pacific Islander patients with ARF (-0.61 [95% CI, -0.88 to -0.34] days) and Hispanic patients with sepsis (-0.22 [95% CI, -0.39 to -0.05] days) or ARF (-0.47 [-0.73 to -0.20] days). Conclusions and Relevance: In this cohort study, Black patients with severe illness who presented with sepsis and/or ARF experienced longer LOS than White patients. Hispanic patients with sepsis and Asian American and Pacific Islander and Hispanic patients with ARF both experienced shorter LOS. Because matched differences were independent of commonly implicated clinical presentation-related factors associated with disparities, identification of additional mechanisms that underlie these disparities is warranted.

Using remotely monitored patient activity patterns after hospital discharge to predict 30 day hospital readmission: a randomized trial.

Hospital readmission prediction models often perform poorly, but most only use information collected until the time of hospital discharge. In this clinical trial, we randomly assigned 500 patients discharged from hospital to home to use either a smartphone or wearable device to collect and transmit remote patient monitoring (RPM) data on activity patterns after hospital discharge. Analyses were conducted at the patient-day level using discrete-time survival analysis. Each arm was split into training and testing folds. The training set used fivefold cross-validation and then final model results are from predictions on the test set. A standard model comprised data collected up to the time of discharge including demographics, comorbidities, hospital length of stay, and vitals prior to discharge. An enhanced model consisted of the standard model plus RPM data. Traditional parametric regression models (logit and lasso) were compared to nonparametric machine learning approaches (random forest, gradient boosting, and ensemble). The main outcome was hospital readmission or death within 30 days of discharge. Prediction of 30-day hospital readmission significantly improved when including remotely-monitored patient data on activity patterns after hospital discharge and using nonparametric machine learning approaches. Wearables slightly outperformed smartphones but both had good prediction of 30-day hospital-readmission.

How Physician Self-Perceptions Affect the Impact of Peer Comparison Feedback on Opioid Prescribing.

Peer comparison feedback is a promising strategy for reducing opioid prescribing and opioid-related harms. Such comparisons may be particularly impactful among underestimating clinicians who do not perceive themselves as high prescribers relative to their peers. But peer comparisons could also unintentionally increase prescribing among overestimating clinicians who do not perceive themselves as lower prescribers than peers. The objective of this study was to assess if the impact of peer comparisons varied by clinicians' preexisting opioid prescribing self-perceptions. Subgroup analysis of a randomized trial of peer comparison interventions among emergency department and urgent care clinicians was used. Generalized mixed-effects models were used to assess whether the impact of peer comparisons, alone or combined with individual feedback, varied by underestimating or overestimating prescriber status. Underestimating and overestimating prescribers were defined as those who self-reported relative prescribing amounts that were lower and higher, respectively, than actual relative baseline amounts. The primary outcome was pills per opioid prescription. Among 438 clinicians, 54% (n = 236) provided baseline prescribing self-perceptions and were included in this analysis. Overall, 17% (n = 40) were underestimating prescribers whereas 5% (n = 11) were overestimating prescribers. Underestimating prescribers exhibited a differentially greater decrease in pills per prescription compared to nonunderestimating clinicians when receiving peer comparison feedback (1.7 pills, 95% CI, -3.2 to -0.2 pills) or combined peer and individual feedback (2.8 pills, 95% CI, -4.8 to -0.8 pills). In contrast, there were no differential changes in pills per prescription for overestimating versus nonoverestimating prescribers after receiving peer comparison (1.5 pills, 95% CI, -0.9 to 3.9 pills) or combined peer and individual feedback (3.0 pills, 95% CI, -0.3 to 6.2 pills). Peer comparisons were more impactful among clinicians who underestimated their prescribing compared to peers. By correcting inaccurate self-perceptions, peer comparison feedback can be an effective strategy for influencing opioid prescribing.

A randomized controlled trial of gamification, financial incentives, or both to increase physical activity among patients with elevated risk for cardiovascular disease: rationale and design of the be active study.

BACKGROUND: Higher levels of physical activity are associated with improvements in cardiovascular health, and consensus guidelines recommend that individuals with or at risk for atherosclerotic cardiovascular disease (ASCVD) participate in regular physical activity. However, most adults do not achieve recommended levels of physical activity. Concepts from behavioral economics have been used to design scalable interventions that increase physical activity over short time periods, but the longer-term efficacy of these strategies is uncertain. STUDY DESIGN AND OBJECTIVES: BE ACTIVE (NCT03911141) is a pragmatic, virtual, randomized controlled trial designed to evaluate the effectiveness of 3 strategies informed by behavioral economic concepts to increase daily physical activity in patients with established ASCVD or 10-year ASCVD risk > 7.5% who are seen in primary care and cardiology clinics affiliated with the University of Pennsylvania Health System. Patients are contacted by email or text message, and complete enrollment and informed consent on the Penn Way to Health online platform. Patients are then provided with a wearable fitness tracker, establish a baseline daily step count, set a goal to increase daily step count by 33% to 50%, and are randomized 1:2:2:2 to control, gamification, financial incentives, or both gamification and financial incentives. Interventions continue for 12 months, with follow-up for an additional 6 months to evaluate the durability of behavior change. The trial has met its enrollment goal of 1050 participants, with a primary endpoint of change from baseline in daily steps over the 12-month intervention period. Key secondary endpoints include change from baseline in daily steps over the 6-month post-intervention follow-up period and change in moderate to vigorous physical activity over the intervention and follow-up periods. If the interventions prove effective, their effects on life expectancy will be compared with their costs in cost-effectiveness analysis. CONCLUSIONS: BE ACTIVE is a virtual, pragmatic randomized clinical trial powered to demonstrate whether gamification, financial incentives, or both are superior to attention control in increasing physical activity. Its results will have important implications for strategies to promote physical activity in patients with or at risk for ASCVD, as well as for the design and implementation of pragmatic virtual clinical trials within health systems.

Hospital Strain and Variation in Sepsis ICU Admission Practices and Associated Outcomes.

To understand how strain-process-outcome relationships in patients with sepsis may vary among hospitals. DESIGN: Retrospective cohort study using a validated hospital capacity strain index as a within-hospital instrumental variable governing ICU versus ward admission, stratified by hospital. SETTING: Twenty-seven U.S. hospitals from 2013 to 2018. PATIENTS: High-acuity emergency department patients with sepsis who do not require life support therapies. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: The mean predicted probability of ICU admission across strain deciles ranged from 4.9% (lowest ICU-utilizing hospital for sepsis without life support) to 61.2% (highest ICU-utilizing hospital for sepsis without life support). The difference in the predicted probabilities of ICU admission between the lowest and highest strain deciles ranged from 9.0% (least strain-sensitive hospital) to 45.2% (most strain-sensitive hospital). In pooled analyses, emergency department patients with sepsis (n = 90,150) experienced a 1.3-day longer median hospital length of stay (LOS) if admitted initially to the ICU compared with the ward, but across the 27 study hospitals (n = 517-6,564), this effect varied from 9.0 days shorter (95% CI, -10.8 to -7.2; p < 0.001) to 19.0 days longer (95% CI, 16.7-21.3; p < 0.001). Corresponding ranges for inhospital mortality with ICU compared with ward admission revealed odds ratios (ORs) from 0.16 (95% CI, 0.03-0.99; p = 0.04) to 4.62 (95% CI, 1.16-18.22; p = 0.02) among patients with sepsis (pooled OR = 1.48). CONCLUSIONS: There is significant among-hospital variation in ICU admission rates for patients with sepsis not requiring life support therapies, how sensitive those ICU admission decisions are to hospital capacity strain, and the association of ICU admission with hospital LOS and hospital mortality. Hospital-level heterogeneity should be considered alongside patient-level heterogeneity in critical and acute care study design and interpretation.

Long-term Effect of Machine Learning-Triggered Behavioral Nudges on Serious Illness Conversations and End-of-Life Outcomes Among Patients With Cancer: A Randomized Clinical Trial.

Importance: Serious illness conversations (SICs) between oncology clinicians and patients are associated with improved quality of life and may reduce aggressive end-of-life care. However, most patients with cancer die without a documented SIC. Objective: To test the impact of behavioral nudges to clinicians to prompt SICs on the SIC rate and end-of-life outcomes among patients at high risk of death within 180 days (high-risk patients) as identified by a machine learning algorithm. Design, Setting, and Participants: This prespecified 40-week analysis of a stepped-wedge randomized clinical trial conducted between June 17, 2019, and April 20, 2020 (including 16 weeks of intervention rollout and 24 weeks of follow-up), included 20 506 patients with cancer representing 41 021 encounters at 9 tertiary or community-based medical oncology clinics in a large academic health system. The current analyses were conducted from June 1, 2021, to May 31, 2022. Intervention: High-risk patients were identified using a validated electronic health record machine learning algorithm to predict 6-month mortality. The intervention consisted of (1) weekly emails to clinicians comparing their SIC rates for all patients against peers' rates, (2) weekly lists of high-risk patients, and (3) opt-out text messages to prompt SICs before encounters with high-risk patients. Main Outcomes and Measures: The primary outcome was SIC rates for all and high-risk patient encounters; secondary end-of-life outcomes among decedents included inpatient death, hospice enrollment and length of stay, and intensive care unit admission and systemic therapy close to death. Intention-to-treat analyses were adjusted for clinic and wedge fixed effects and clustered at the oncologist level. Results: The study included 20 506 patients (mean [SD] age, 60.0 [14.0] years) and 41 021 patient encounters: 22 259 (54%) encounters with female patients, 28 907 (70.5%) with non-Hispanic White patients, and 5520 (13.5%) with high-risk patients; 1417 patients (6.9%) died by the end of follow-up. There were no meaningful differences in demographic characteristics in the control and intervention periods. Among high-risk patient encounters, the unadjusted SIC rates were 3.4% (59 of 1754 encounters) in the control period and 13.5% (510 of 3765 encounters) in the intervention period. In adjusted analyses, the intervention was associated with increased SICs for all patients (adjusted odds ratio, 2.09 [95% CI, 1.53-2.87]; P < .001) and decreased end-of-life systemic therapy (7.5% [72 of 957 patients] vs 10.4% [24 of 231 patients]; adjusted odds ratio, 0.25 [95% CI, 0.11-0.57]; P = .001) relative to controls, but there was no effect on hospice enrollment or length of stay, inpatient death, or end-of-life ICU use. Conclusions and Relevance: In this randomized clinical trial, a machine learning-based behavioral intervention and behavioral nudges to clinicans led to an increase in SICs and reduction in end-of-life systemic therapy but no changes in other end-of-life outcomes among outpatients with cancer. These results suggest that machine learning and behavioral nudges can lead to long-lasting improvements in cancer care delivery. Trial Registration: ClinicalTrials.gov Identifier: NCT03984773.

Instrumented difference-in-differences.

Unmeasured confounding is a key threat to reliable causal inference based on observational studies. Motivated from two powerful natural experiment devices, the instrumental variables and difference-in-differences, we propose a new method called instrumented difference-in-differences that explicitly leverages exogenous randomness in an exposure trend to estimate the average and conditional average treatment effect in the presence of unmeasured confounding. We develop the identification assumptions using the potential outcomes framework. We propose a Wald estimator and a class of multiply robust and efficient semiparametric estimators, with provable consistency and asymptotic normality. In addition, we extend the instrumented difference-in-differences to a two-sample design to facilitate investigations of delayed treatment effect and provide a measure of weak identification. We demonstrate our results in simulated and real datasets.

Spatial Disadvantage and Racial Disparities in Gun Homicides.

A spatially disadvantaged census tract is one that is surrounded by disadvantaged tracts. More spatially disadvantaged neighborhoods may experience more violence, independent of their own level of disadvantage, and majority Black middle-class neighborhoods are more likely to be spatially disadvantaged than majority white neighborhoods. The purpose of this paper is to study how much of the racial difference in gun homicide rates between majority Black and majority white middle-class neighborhoods can be explained by differences in spatial disadvantage. To study this, comparable majority Black and majority white tracts were matched to understand how gun homicide rates differ in neighborhoods with similar levels of disadvantage. Further matching on spatial disadvantage reduced the disparity in gun homicides between majority Black and majority white middle-class neighborhoods, suggesting that spatial disadvantage accounts for some but not all of the disparity.

Automated analysis of low-field brain MRI in cerebral malaria.

A central challenge of medical imaging studies is to extract biomarkers that characterize disease pathology or outcomes. Modern automated approaches have found tremendous success in high-resolution, high-quality magnetic resonance images. These methods, however, may not translate to low-resolution images acquired on magnetic resonance imaging (MRI) scanners with lower magnetic field strength. In low-resource settings where low-field scanners are more common and there is a shortage of radiologists to manually interpret MRI scans, it is critical to develop automated methods that can augment or replace manual interpretation, while accommodating reduced image quality. We present a fully automated framework for translating radiological diagnostic criteria into image-based biomarkers, inspired by a project in which children with cerebral malaria (CM) were imaged using low-field 0.35 Tesla MRI. We integrate multiatlas label fusion, which leverages high-resolution images from another sample as prior spatial information, with parametric Gaussian hidden Markov models based on image intensities, to create a robust method for determining ventricular cerebrospinal fluid volume. We also propose normalized image intensity and texture measurements to determine the loss of gray-to-white matter tissue differentiation and sulcal effacement. These integrated biomarkers have excellent classification performance for determining severe brain swelling due to CM.

Among-Hospital Variation in Intensive Care Unit Admission Practices and Associated Outcomes for Patients with Acute Respiratory Failure.

Rationale: We have previously shown that hospital strain is associated with intensive care unit (ICU) admission and that ICU admission, compared with ward admission, may benefit certain patients with acute respiratory failure (ARF). Objectives: To understand how strain-process-outcomes relationships in patients with ARF may vary among hospitals and what hospital practice differences may account for such variation. Methods: We examined high-acuity patients with ARF who did not require mechanical ventilation or vasopressors in the emergency department (ED) and were admitted to 27 U.S. hospitals from 2013 to 2018. Stratifying by hospital, we compared hospital strain-ICU admission relationships and hospital length of stay (LOS) and mortality among patients initially admitted to the ICU versus the ward using hospital strain as a previously validated instrumental variable. We also surveyed hospital practices and, in exploratory analyses, evaluated their associations with the above processes and outcomes. Results: There was significant among-hospital variation in ICU admission rates, in hospital strain-ICU admission relationships, and in the association of ICU admission with hospital LOS and hospital mortality. Overall, ED patients with ARF (n = 45,339) experienced a 0.82-day shorter median hospital LOS if admitted initially to the ICU compared with the ward, but among the 27 hospitals (n = 224-3,324), this effect varied from 5.85 days shorter (95% confidence interval [CI], -8.84 to -2.86; P < 0.001) to 4.38 days longer (95% CI, 1.86-6.90; P = 0.001). Corresponding ranges for in-hospital mortality with ICU compared with ward admission revealed odds ratios from 0.08 (95% CI, 0.01-0.56; P < 0.007) to 8.89 (95% CI, 1.60-79.85; P = 0.016) among patients with ARF (pooled odds ratio, 0.75). In exploratory analyses, only a small number of measured hospital practices-the presence of a sepsis ED disposition guideline and maximum ED patient capacity-were potentially associated with hospital strain-ICU admission relationships. Conclusions: Hospitals vary considerably in ICU admission rates, the sensitivity of those rates to hospital capacity strain, and the benefits of ICU admission for patients with ARF not requiring life support therapies in the ED. Future work is needed to more fully identify hospital-level factors contributing to these relationships.