Individualized treatment rule characterization via a value function surrogate.

Precision medicine is a promising framework for generating evidence to improve health and health care. Yet, a gap persists between the ever-growing number of statistical precision medicine strategies for evidence generation and implementation in real-world clinical settings, and the strategies for closing this gap will likely be context-dependent. In this paper, we consider the specific context of partial compliance to wound management among patients with peripheral artery disease. Using a Gaussian process surrogate for the value function, we show the feasibility of using Bayesian optimization to learn optimal individualized treatment rules. Further, we expand beyond the common precision medicine task of learning an optimal individualized treatment rule to the characterization of classes of individualized treatment rules and show how those findings can be translated into clinical contexts.

Use of machine learning to identify characteristics associated with severe hypoglycemia in older adults with type 1 diabetes: a post-hoc analysis of a case-control study.

INTRODUCTION: Severe hypoglycemia (SH) in older adults (OAs) with type 1 diabetes is associated with profound morbidity and mortality, yet its etiology can be complex and multifactorial. Enhanced tools to identify OAs who are at high risk for SH are needed. This study used machine learning to identify characteristics that distinguish those with and without recent SH, selecting from a range of demographic and clinical, behavioral and lifestyle, and neurocognitive characteristics, along with continuous glucose monitoring (CGM) measures. RESEARCH DESIGN AND METHODS: Data from a case-control study involving OAs recruited from the T1D Exchange Clinical Network were analyzed. The random forest machine learning algorithm was used to elucidate the characteristics associated with case versus control status and their relative importance. Models with successively rich characteristic sets were examined to systematically incorporate each domain of possible risk characteristics. RESULTS: Data from 191 OAs with type 1 diabetes (47.1% female, 92.1% non-Hispanic white) were analyzed. Across models, hypoglycemia unawareness was the top characteristic associated with SH history. For the model with the richest input data, the most important characteristics, in descending order, were hypoglycemia unawareness, hypoglycemia fear, coefficient of variation from CGM, % time blood glucose below 70 mg/dL, and trail making test B score. CONCLUSIONS: Machine learning may augment risk stratification for OAs by identifying key characteristics associated with SH. Prospective studies are needed to identify the predictive performance of these risk characteristics.

Assessing risk of bias in the meta-analysis of round 1 of the Health Care Innovation Awards.

BACKGROUND: Systematic reviews of observational studies can be affected by biases that lead to under- or over-estimates of true intervention effects. Several tools have been reported in the literature that attempt to characterize potential bias. Our objective in this study was to determine the extent to which study-specific bias may have influenced intervention impacts on total costs of care (TCOC) in round 1 of the Health Care Innovation Awards. METHODS: We reviewed 82 statistical evaluations of innovation impacts on Medicare TCOC. We developed five risk-of-bias measures and assessed their influence on TCOC impacts using meta-regression. RESULTS: The majority of evaluations used propensity score matching to create their comparison groups. One third of the non-randomized interventions were judged to have some risk of biased effects due largely to the way they recruited their treatment groups, and 35% had some degree of covariate imbalance remaining after propensity score adjustments. However, in the multivariable analysis of TCOC effects, none of the bias threats we examined (comparison group construction method, risk of bias, or degree of covariate imbalance) had a major impact on the magnitude of HCIA1 innovation effects. Evaluations using propensity score weighting produced larger but imprecise savings effects compared to propensity score matching. DISCUSSION: Our results suggest that it is unlikely that HCIA1 TCOC effect sizes were systematically affected by the types of bias we considered. Assessing the risk of bias based on specific study design features is likely to be more useful for identifying problematic characteristics than the subjective quality ratings used by existing risk tools.

Balancing evidence-based care with patient-centered individualized care.

Weak evidence, when manifested in clinical guidelines, can translate into biased vascular care. In vascular surgery, we have few randomized controlled trials with appropriate representation of females and persons of color, so generalizability of trial results can be problematic. Physicians are required to balance evidenced-based care (which is only as good as the underlying evidence) with personalized treatment recommendations that are often based on demographics, social circumstances, and/or existing therapeutic relationships. Biases, whether implicit or explicit, have an oversized effect on treatment decisions, and patient outcomes. In this commentary, we propose three principles to strengthen the vascular surgery evidence foundation and patient-centered decision-making going forward: (1) generating evidence designed for individualized care, (2) constructing clinical guidelines that are context specific and complexity aware, and (3) strengthening the training and support for surgeons to deliver patient-centered individualized care.

Differential impact of missed initial wound clinic visit on 6-month wound healing by race/ethnicity among patients with chronic limb-threatening ischemia.

Chronic limb-threatening ischemia (CLTI) is associated with significant morbidity, including major limb amputation, and mortality. Healing ischemic wounds is necessary to optimise vascular outcomes and can be facilitated by dedicated appointments at a wound clinic. This study aimed to estimate the association between successful wound care initiation and 6-month wound healing, with specific attention to differences by race/ethnicity. This retrospective study included 398 patients with CLTI and at least one ischaemic wound who scheduled an appointment at our wound clinic between January 2015 and July 2020. The exposure was the completion status of patients' first scheduled wound care appointment (complete/not complete) and the primary outcome was 6-month wound healing (healed/not healed). The analysis focused on how this association was modified by race/ethnicity. We used Aalen-Johansen estimators to produce cumulative incidence curves and calculated risk ratios within strata of race/ethnicity. The final adjustment set included age, revascularization, and initial wound size. Patients had a mean age of 67 ± 14 years, were 41% female, 46% non-White and had 517 total wounds. In the overall cohort, 70% of patients completed their first visit and 34% of wounds healed within 6-months. There was no significant difference in 6-month healing based on first visit completion status for White/non-Hispanic individuals (RR [95% CI] = 1.18 [0.91, 1.45]; p-value = 0.130), while non-White individuals were roughly 3 times more likely to heal their wounds if they completed their first appointment (RR [95% CI] = 2.89 [2.66, 3.11]; p-value < 0.001). In conclusion, non-White patients were approximately three times more likely to heal their wound in 6 months if they completed their first scheduled wound care appointment while White/non-Hispanic individuals' risk of healing was similar regardless of first visit completion status. Future efforts should focus on providing additional resources to ensure minority groups with wounds have the support they need to access and successfully initiate wound care.

Individualized interventions and precision health: Lessons learned from a systematic review and implications for analytics-driven geriatric research.

Older adults are characterized by profound clinical heterogeneity. When designing and delivering interventions, there exist multiple approaches to account for heterogeneity. We present the results of a systematic review of data-driven, personalized interventions in older adults, which serves as a use case to distinguish the conceptual and methodologic differences between individualized intervention delivery and precision health-derived interventions. We define individualized interventions as those where all participants received the same parent intervention, modified on a case-by-case basis and using an evidence-based protocol, supplemented by clinical judgment as appropriate, while precision health-derived interventions are those that tailor care to individuals whereby the strategy for how to tailor care was determined through data-driven, precision health analytics. We discuss how their integration may offer new opportunities for analytics-based geriatric medicine that accommodates individual heterogeneity but allows for more flexible and resource-efficient population-level scaling.

Central concepts for randomized controlled trials and other emerging trial designs.

Randomized controlled trials (RCTs) are widely considered to provide the highest-quality evidence for the comparative efficacy and safety of competing clinical strategies. The strength of using RCTs for causal inference is derived from random treatment assignment and prospective data collection. Randomization eliminates confounding at the time of treatment group assignment, achieving exchangeability of the baseline study groups, such that they are the same, on average, except for the study intervention. Prospective data collection helps ensure that eligibility assessment, treatment assignment, and the start of follow-up are aligned temporally. Temporal alignment prevents biases that are common in observational research (eg, immortal time bias). In ideal settings, the results of an RCT provide the average causal effect of the intervention on the selected outcomes in the study population. Although observational research can estimate similar causal effects, observational designs require more assumptions and more advanced analytic frameworks than an RCT designed to answer the same question. Emerging trial designs, also discussed here, seek to address certain limitations of traditional RCT designs. The purpose of this review was to provide a broad overview of the central concepts in RCT design, implementation, conduct, and data analysis.

Analysis of a Machine Learning-Based Risk Stratification Scheme for Chronic Limb-Threatening Ischemia.

Importance: Valid risk stratification schemes are key to performing comparative effectiveness research; however, for chronic limb-threatening ischemia (CLTI), risk stratification schemes have limited efficacy. Improved, accurate, comprehensive, and reproducible risk stratification models for CLTI are needed. Objective: To evaluate the use of topic model cluster analysis to generate an accurate risk prediction model for CLTI. Design, Setting, and Participants: This multicenter, nested cohort study of existing Project of Ex Vivo Vein Graft Engineering via Transfection (PREVENT) III clinical trial data assessed data from patients undergoing infrainguinal vein bypass for the treatment of ischemic rest pain or ischemic tissue loss. Original data were collected from January 1, 2001, to December 31, 2003, and were analyzed in September 2021. All patients had 1 year of follow-up. Exposures: Supervised topic model cluster analysis was applied to nested cohort data from the PREVENT III randomized clinical trial. Given a fixed number of clusters, the data were used to examine the probability that a patient belonged to each of the clusters and the distribution of the features within each cluster. Main Outcomes and Measures: The primary outcome was 1-year CLTI-free survival, a composite of survival with remission of ischemic rest pain, wound healing, and freedom from major lower-extremity amputation without recurrent CLTI. Results: Of the original 1404 patients, 166 were excluded because of a lack of sufficient feature and/or outcome data, leaving 1238 patients for analysis (mean [SD] age, 68.4 [11.2] years; 800 [64.6%] male; 894 [72.2%] White). The Society for Vascular Surgery Wound, Ischemia, and Foot Infection grade 2 wounds were present in 543 patients (43.8%), with rest pain present in 645 (52.1%). Three distinct clusters were identified within the cohort (130 patients in stage 1, 578 in stage 2, and 530 in stage 3), with 1-year CLTI-free survival rates of 82.3% (107 of 130 patients) for stage 1, 61.1% (353 of 578 patients) for stage 2, and 53.4% (283 of 530 patients) for stage 3. Stratified by stage, 1-year mortality was 10.0% (13 of 130 observed deaths in stage 1) for stage 1, 13.5% (78 of 578 patients) for stage 2, and 20.2% (105 of 521 patients) for stage 3. Similarly, stratifying by stage revealed major limb amputation rates of 4.2% (5 of 119 observed major limb amputations in stage 1) for stage 1, 10.8% (55 of 509 patients) for stage 2, and 18.4% (81 of 440 patients) for stage 3. Among survivors without a major amputation, the rates of CLTI recurrence were 9.2% (11 of 119 observed recurrences in stage 1) for stage 1, 24.9% (130 of 523 patients) for stage 2, and 29.6% (132 of 446 patients) for stage 3. Conclusions and Relevance: The topic model cluster analysis in this cohort study identified 3 distinct stages within CLTI. Findings suggest that CLTI-free survival is an end point that can be accurately and reproducibly quantified and may be used as a patient-centric outcome.

Mapping mental health inequalities: The intersecting effects of gender, race, class, and ethnicity on ADHD diagnosis.

While the effects of social stratification by gender, race, class, and ethnicity on health inequalities are well-documented, our understanding of the intersecting consequences of these social dimensions on diagnosis remains limited. This is particularly the case in studies of mental health, where "paradoxical" patterns of stratification have been identified. Using a Bayesian multi-level random-effects Poisson model and a nationally representative random sample of 138,009 households from the National Survey of Children's Health, this study updates and extends the literature on mental health inequalities through an intersectional investigation of one of the most commonly diagnosed psychiatric conditions of childhood/adolescence: attention-deficit hyperactivity disorder (ADHD). Findings indicate that gender, race, class, and ethnicity combine in mutually constitutive ways to explain between-group variation in ADHD diagnosis. Observed effects underscore the importance and feasibility of an intersectional, multi-level modelling approach and data mapping technique to advance our understanding of social subgroups more/less likely to be diagnosed with mental health conditions.

The Society for Vascular Surgery Objective Performance Goals for Critical Limb Ischemia are attainable in select patients with ischemic wounds managed with wound care alone.

BACKGROUND: To set therapeutic benchmarks, in 2009 the Society for Vascular Surgery defined objective performance goals (OPG) for treatment of patients with chronic limb threatening ischemia (CLTI) with either open surgical bypass or endovascular intervention. The goal of these OPGs are to set standards of care from a revascularization standpoint and to provide performance benchmarks for 1 year patency rates for new endovascular therapies. While OPGs are useful in this regard, a critical decision point in the treatment of patients with CLTI is determining when revascularization is necessary. There is little guidance in the comprehensive treatment of this patient population, especially in the nonoperative cohort. Guidelines are needed for the CLTI patient population as a whole and not just those revascularized, and our aim was to assess whether CLTI OPGs could be attained with nonoperative management alone. METHODS: Our cohort included patients with an incident diagnosis of CLTI (by hemodynamic and symptomatic criteria) at our institution from 2013-2017. The primary outcome measured was mortality. Secondary outcomes were limb loss and failure of amputation-free survival. Descriptive statistics were used to define the 2 groups - patients undergoing primary revascularization and patients undergoing primary wound management. The risk difference in outcomes between the 2 groups was estimated using collaborative-targeted maximum likelihood estimation. RESULTS: Our cohort included 349 incident CLTI patients; 60% male, 51% white, mean age 63 +/- 13 years, 20% Rutherford 4, and 80% Rutherford 5. Most patients (277, 79%) underwent primary revascularization, and 72 (21%) were treated with wound care alone. Demographics and presenting characteristics were similar between groups. Although the revascularized patients were more likely to have femoropopliteal disease (72% vs. 36%), both groups had a high rate of infrapopliteal disease (62% vs. 57%). Not surprisingly, the patients in the revascularization group were less likely to have congestive heart failure (34% vs. 42%), complicated diabetes (52% vs. 79%), obesity (19% vs. 33%), and end stage renal disease (14% vs. 28%). In the wound care group, 2-year outcomes were 65% survival, 51% amputation free survival, 19% major limb amputation, and 17% major adverse cardiac event. The wound care cohort had a 13% greater risk of death at 2 years; however, the risk of limb loss at 2 years was 12% less in the wound care cohort. CONCLUSIONS: A comprehensive set treatment goals and expected amputation free survival outcomes can guide revascularization, but also assure that appropriate outcomes are achieved for patients treated without revascularization. The 2-year outcomes achieved in this cohort provide an estimate of outcomes for nonrevascularized CLTI patients. Although multi-center or prospective studies are needed, we demonstrate that equal, even improved, limb salvage rates are possible.

Limiting adverse birth outcomes in resource-limited settings (LABOR): protocol of a prospective intrapartum cohort study.

Background: Each year, nearly 300,000 women and 5 million fetuses or neonates die during childbirth or shortly thereafter, a burden concentrated disproportionately in low- and middle-income countries. Identifying women and their fetuses at risk for intrapartum-related morbidity and death could facilitate early intervention. Methods: The Limiting Adverse Birth Outcomes in Resource-Limited Settings (LABOR) Study is a multi-country, prospective, observational cohort designed to exhaustively document the course and outcomes of labor, delivery, and the immediate postpartum period in settings where adverse outcomes are frequent. The study is conducted at four hospitals across three countries in Ghana, India, and Zambia. We will enroll approximately 12,000 women at presentation to the hospital for delivery and follow them and their fetuses/newborns throughout their labor and delivery course, postpartum hospitalization, and up to 42 days thereafter. The co-primary outcomes are composites of maternal (death, hemorrhage, hypertensive disorders, infection) and fetal/neonatal adverse events (death, encephalopathy, sepsis) that may be attributed to the intrapartum period. The study collects extensive physiologic data through the use of physiologic sensors and employs medical scribes to document examination findings, diagnoses, medications, and other interventions in real time. Discussion: The goal of this research is to produce a large, sharable dataset that can be used to build statistical algorithms to prospectively stratify parturients according to their risk of adverse outcomes. We anticipate this research will inform the development of new tools to reduce peripartum morbidity and mortality in low-resource settings.

Precision Medicine Enables More TNM-Like Staging in Patients With Chronic Limb Threatening Ischemia.

Introduction: In cancer, there are survival-based staging systems and tailored, stage-based treatments. There is little personalized treatment in vascular disease. The 2019 Global Vascular Guidelines on the Management of CLTI proposed successful treatment hinges upon Patient risk, Limb severity, and ANatomic complexity (PLAN). We sought to confirm a three axis approach and define how increasing severity affects mortality, not just limb loss. Methods: Patients revascularized for incident CLTI at our institution from 2013 to 2017 were included. Outcomes were mortality, limb loss, the composite endpoint of amputation-free survival. Using Bayesian machine learning, specifically supervised topic modeling, clusters of patient features associated with mortality were formed after controlling for revascularization type. Patients were assigned to the cluster they belonged to with highest probability; clusters were characterized by analyzing the characteristics of patients within them. Patient outcomes were used to order the clusters into stages with increasing mortality. Results: We defined three distinct clusters as the basis for patient- and limb-centered stages. Across stages, rates of 1-year mortality were 7.6, 13.8, 18.9% and rates of amputation-free survival were 84.8, 79.3, and 63.2%. Stage one had patients with rest pain and previous revascularization who were less likely to have wounds, diabetes, and renal disease. Stage two had doubled mortality, likely related to diabetes prevalence. Stage three is characterized by high rates of complicated comorbidities, particularly end stage renal disease, and significantly higher rate of limb loss (22.6 vs. 8% in stages one and two). Conclusion: Using precision medicine, we have demonstrated clustering of CLTI patients that can be used toward a robust staging system. We provide empiric evidence for PLAN and detail about how changes in each variable affect survival and amputation-free survival.

Beyond Two Cultures: Cultural Infrastructure for Data-driven Decision Support.

In the twenty years since Dr. Leo Breiman's incendiary paper Statistical Modeling: The Two Cultures was first published, algorithmic modeling techniques have gone from controversial to commonplace in the statistical community. While the widespread adoption of these methods as part of the contemporary statistician's toolkit is a testament to Dr. Breiman's vision, the number of high-profile failures of algorithmic models suggests that Dr. Breiman's final remark that "the emphasis needs to be on the problem and the data" has been less widely heeded. In the spirit of Dr. Breiman, we detail an emerging research community in statistics - data-driven decision support. We assert that to realize the full potential of decision support, broadly and in the context of precision health, will require a culture of social awareness and accountability, in addition to ongoing attention towards complex technical challenges.

Using Electronic Medical Records to Identify Enhanced Recovery After Surgery Cases.

CONTEXT: Enhanced recovery after surgery (ERAS) aims to improve surgical outcomes by integrating evidence-based practices across preoperative, intraoperative, and postoperative care. Data in electronic medical records (EMRs) provide insight on how ERAS is implemented and its impact on surgical outcomes. Because ERAS is a multimodal pathway provided by multiple physicians and health care providers over time, identifying ERAS cases in EMRs is not a trivial task. To better understand how EMRs can be used to study ERAS, we describe our experience with using current methodologies and the development and rationale of a new method for retrospectively identifying ERAS cases in EMRs. CASE DESCRIPTION: Using EMR data from surgical departments at the University of North Carolina at Chapel Hill, we first identified ERAS cases using a protocol-based method, using basic information including the date of ERAS implementation, surgical procedure and date, and primary surgeon. We further examined two operational flags in the EMRs, a nursing order and a case request for OR order. Wide variation between the methods compelled us to consult with ERAS surgical staff and explore the EMRs to develop a more refined method for identifying ERAS cases. METHOD: We developed a two-step method, with the first step based on the protocol definition and the second step based on an ERAS-specific medication definition. To test our method, we randomly sampled 150 general, gynecological, and urologic surgeries performed between January 1, 2016 and March 30, 2017. Surgical cases were classified as ERAS or not using the protocol definition, nursing order, case request for OR order, and our two-step method. To assess the accuracy of each method, two independent reviewers assessed the charts to determine whether cases were ERAS. FINDINGS: Of the 150 charts reviewed, 74 were ERAS cases. The protocol only method and nursing order flag performed similarly, correctly identifying 74 percent and 73 percent of true ERAS cases, respectively. The case request for OR order flag performed less well, correctly identifying only 44 percent of the true ERAS cases. Our two-step method performed well, correctly identifying 98 percent of true ERAS cases. CONCLUSION: ERAS pathways are complex, making study of them from EMRs difficult. Current strategies for doing so are relatively easy to implement, but unreliable. We have developed a reproducible and observable ERAS computational phenotype that identifies ERAS cases reliably. This is a step forward in using the richness of EMR data to study ERAS implementation, efficacy, and how they can contribute to surgical care improvement.

Impact Of Health Care Delivery System Innovations On Total Cost Of Care.

Using delivery system innovations to advance health care reform continues to be of widespread interest. However, it is difficult to generalize about the success of specific types of innovations, since they have been examined in only a few studies. To gain a broader perspective, we analyzed the results of forty-three ambulatory care programs funded by the first round of the Center for Medicare and Medicaid Innovation's Health Care Innovations Awards. The innovations' impacts on total cost of care were estimated by independent evaluators using multivariable difference-in-differences models. Through the first two years, most of the innovations did not show a significant effect on total cost of care. Using meta-regression, we assessed the effects on costs of five common components of these innovations. Innovations that used health information technology or community health workers achieved the greatest cost savings. Savings were also relatively large in programs that targeted clinically fragile patients-clinically complex populations at risk for disease progression. While the magnitude of these effects was often substantial, none achieved conventional levels of significance in our analyses. Meta-analyses of a larger number of delivery system innovations are needed to more clearly establish their potential for patient care cost savings.