Alejandro Schuler

Cost-effectiveness of Intensive Blood Pressure Management

ImportancenAmong high-risk patients with hypertension, targeting a systolic blood pressure of 120 mm Hg reduces cardiovascular morbidity and mortality compared with a higher target. However, intensive blood pressure management incurs additional costs from treatment and from adverse events.nnnObjectivenTo evaluate the incremental cost-effectiveness of intensive blood pressure management compared with standard management.nnnDesign, Setting, and ParticipantsnThis cost-effectiveness analysis conducted from September 2015 to August 2016 used a Markov cohort model to estimate cost-effectiveness of intensive blood pressure management among 68-year-old high-risk adults with hypertension but not diabetes. We used the Systolic Blood Pressure Intervention Trial (SPRINT) to estimate treatment effects and adverse event rates. We used Centers for Disease Control and Prevention Life Tables to project age- and cause-specific mortality, calibrated to rates reported in SPRINT. We also used population-based observational data to model development of heart failure, myocardial infarction, stroke, and subsequent mortality. Costs were based on published sources, Medicare data, and the National Inpatient Sample.nnnInterventionsnTreatment of hypertension to a systolic blood pressure goal of 120 mm Hg (intensive management) or 140 mm Hg (standard management).nnnMain Outcomes and MeasuresnLifetime costs and quality-adjusted life-years (QALYs), discounted at 3% annually.nnnResultsnStandard management yielded 9.6 QALYs and accrued

Some methods for heterogeneous treatment effect estimation in high dimensions: Some methods for heterogeneous treatment effect estimation in high dimensions

155u202f261 in lifetime costs, while intensive management yielded 10.5 QALYs and accrued

DISCOVERING PATIENT PHENOTYPES USING GENERALIZED LOW RANK MODELS.

176u202f584 in costs. Intensive blood pressure management cost

Performing an Informatics Consult: Methods and Challenges

23u202f777 per QALY gained. In a sensitivity analysis, serious adverse events would need to occur at 3 times the rate observed in SPRINT and be 3 times more common in the intensive management arm to prefer standard management.nnnConclusions and RelevancenIntensive blood pressure management is cost-effective at typical thresholds for value in health care and remains so even with substantially higher adverse event rates.

Comparing methods for estimation of heterogeneous treatment effects using observational data from health care databases

When devising a course of treatment for a patient, doctors often have little quantitative evidence on which to base their decisions, beyond their medical education and published clinical trials. Stanford Health Care alone has millions of electronic medical records that are only just recently being leveraged to inform better treatment recommendations. These data present a unique challenge because they are high dimensional and observational. Our goal is to make personalized treatment recommendations based on the outcomes for past patients similar to a new patient. We propose and analyze 3 methods for estimating heterogeneous treatment effects using observational data. Our methods perform well in simulations using a wide variety of treatment effect functions, and we present results of applying the 2 most promising methods to data from The SPRINT Data Analysis Challenge, from a large randomized trial of a treatment for high blood pressure.

An evaluation of clinical order patterns machine-learned from clinician cohorts stratified by patient mortality outcomes

The practice of medicine is predicated on discovering commonalities or distinguishing characteristics among patients to inform corresponding treatment. Given a patient grouping (hereafter referred to as a phenotype), clinicians can implement a treatment pathway accounting for the underlying cause of disease in that phenotype. Traditionally, phenotypes have been discovered by intuition, experience in practice, and advancements in basic science, but these approaches are often heuristic, labor intensive, and can take decades to produce actionable knowledge. Although our understanding of disease has progressed substantially in the past century, there are still important domains in which our phenotypes are murky, such as in behavioral health or in hospital settings. To accelerate phenotype discovery, researchers have used machine learning to find patterns in electronic health records, but have often been thwarted by missing data, sparsity, and data heterogeneity. In this study, we use a flexible framework called Generalized Low Rank Modeling (GLRM) to overcome these barriers and discover phenotypes in two sources of patient data. First, we analyze data from the 2010 Healthcare Cost and Utilization Project National Inpatient Sample (NIS), which contains upwards of 8 million hospitalization records consisting of administrative codes and demographic information. Second, we analyze a small (N=1746), local dataset documenting the clinical progression of autism spectrum disorder patients using granular features from the electronic health record, including text from physician notes. We demonstrate that low rank modeling successfully captures known and putative phenotypes in these vastly different datasets.

Association of Hemoglobin A1c Levels With Use of Sulfonylureas, Dipeptidyl Peptidase 4 Inhibitors, and Thiazolidinediones in Patients With Type 2 Diabetes Treated With Metformin: Analysis From the Observational Health Data Sciences and Informatics Initiative

Our health care system is plagued by missed opportunities, waste, and harm. Data generated in the course of care are often underutilized, scientific insight goes untranslated, and evidence is overlooked. To address these problems, we envisioned a system where aggregate patient data can be used at the bedside to provide practice-based evidence. To create that system, we directly connect practicing physicians to clinical researchers and data scientists through an informatics consult. Our team processes and classifies questions posed by clinicians, identifies the appropriate patient data to use, runs the appropriate analyses, and returns an answer, ideally in a 48-hour time window. Here, we discuss the methods that are used for data extraction, processing, and analysis in our consult. We continue to refine our informatics consult service, moving closer to a learning health care system.

The Impact of Acute Organ Dysfunction on Long-Term Survival in Sepsis

There is growing interest in using routinely collected data from health care databases to study the safety and effectiveness of therapies in real-world conditions, as it can provide complementary evidence to that of randomized controlled trials. Causal inference from health care databases is challenging because the data are typically noisy, high dimensional, and most importantly, observational. It requires methods that can estimate heterogeneous treatment effects while controlling for confounding in high dimensions. Bayesian additive regression trees, causal forests, causal boosting, and causal multivariate adaptive regression splines are off-the-shelf methods that have shown good performance for estimation of heterogeneous treatment effects in observational studies of continuous outcomes. However, it is not clear how these methods would perform in health care database studies where outcomes are often binary and rare and data structures are complex. In this study, we evaluate these methods in simulation studies that recapitulate key characteristics of comparative effectiveness studies. We focus on the conditional average effect of a binary treatment on a binary outcome using the conditional risk difference as an estimand. To emulate health care database studies, we propose a simulation design where real covariate and treatment assignment data are used and only outcomes are simulated based on nonparametric models of the real outcomes. We apply this design to 4 published observational studies that used records from 2 major health care databases in the United States. Our results suggest that Bayesian additive regression trees and causal boosting consistently provide low bias in conditional risk difference estimates in the context of health care database studies.

A comparison of methods for model selection when estimating individual treatment effects

OBJECTIVEnEvaluate the quality of clinical order practice patterns machine-learned from clinician cohorts stratified by patient mortality outcomes.nnnMATERIALS AND METHODSnInpatient electronic health records from 2010 to 2013 were extracted from a tertiary academic hospital. Clinicians (nu202f=u202f1822) were stratified into low-mortality (21.8%, nu202f=u202f397) and high-mortality (6.0%, nu202f=u202f110) extremes using a two-sided P-value score quantifying deviation of observed vs. expected 30-day patient mortality rates. Three patient cohorts were assembled: patients seen by low-mortality clinicians, high-mortality clinicians, and an unfiltered crowd of all clinicians (nu202f=u202f1046, 1046, and 5230 post-propensity score matching, respectively). Predicted order lists were automatically generated from recommender system algorithms trained on each patient cohort and evaluated against (i) real-world practice patterns reflected in patient cases with better-than-expected mortality outcomes and (ii) reference standards derived from clinical practice guidelines.nnnRESULTSnAcross six common admission diagnoses, order lists learned from the crowd demonstrated the greatest alignment with guideline references (AUROC rangeu202f=u202f0.86-0.91), performing on par or better than those learned from low-mortality clinicians (0.79-0.84, Pu202f<u202f10-5) or manually-authored hospital order sets (0.65-0.77, Pu202f<u202f10-3). The same trend was observed in evaluating model predictions against better-than-expected patient cases, with the crowd model (AUROC meanu202f=u202f0.91) outperforming the low-mortality model (0.87, Pu202f<u202f10-16) and order set benchmarks (0.78, Pu202f<u202f10-35).nnnDISCUSSIONnWhether machine-learning models are trained on all clinicians or a subset of experts illustrates a bias-variance tradeoff in data usage. Defining robust metrics to assess quality based on internal (e.g. practice patterns from better-than-expected patient cases) or external reference standards (e.g. clinical practice guidelines) is critical to assess decision support content.nnnCONCLUSIONnLearning relevant decision support content from all clinicians is as, if not more, robust than learning from a select subgroup of clinicians favored by patient outcomes.

General-purpose validation and model selection when estimating individual treatment effects.

Key Points Question Can the effectiveness of second-line treatment of type 2 diabetes after initial therapy with metformin be characterized via an open collaborative research network? Findings In this analysis of data from more than 246 million patients in multiple cohorts, treatment with dipeptidyl peptidase 4 inhibitors compared with sulfonylureas and thiazolidinediones did not differ in reducing hemoglobin A1c levels or hazard of kidney disorders. In a meta-analysis, sulfonylureas compared with dipeptidyl peptidase 4 inhibitors were associated with a small increased hazard of myocardial infarction and eye disorders in patients with type 2 diabetes. Meaning Large-scale characterization of the effectiveness of type 2 diabetes therapy across nations through an open collaborative research network aligns with the 2017 recommendation of the American Association of Clinical Endocrinologists and American College of Endocrinology in type 2 diabetes management recommending dipeptidyl peptidase 4 inhibitors over sulfonylureas in patients with diabetes for whom metformin was the first-line treatment.