A cohort study of sodium-glucose cotransporter-2 inhibitors after acute kidney injury among Veterans with diabetic kidney disease.

Sodium-glucose cotransporter-2 inhibitors (SGLT2i) reduce the risk for several adverse outcomes among patients with diabetic kidney disease. Yet, optimal timing for SGLT2i after acute kidney injury (AKI) is uncertain, as are the providers responsible for post-AKI SGLT2i initiation. Using a retrospective cohort of United States Veterans with diabetes mellitus type 2 and proteinuria, we examined encounters by provider specialty before SGLT2i initiation and subsequent all-cause mortality after hospitalization with AKI, defined by a 50% or more rise in serum creatinine. Covariates included recovery, defined by return to a 110% or less of baseline creatinine, and time since AKI hospitalization. Among 21,330 eligible Veterans, 7,798 died (37%) and 6,562 received a SGLT2i (31%) over median follow-up of 2.1 years. Post-AKI SGLT2i use was associated with lower mortality risk [adjusted hazard ratio 0.63 (95% confidence interval 0.58-0.68)]. Compared with neither SGLT2i use nor recovery, mortality risk was similar with recovery without SGLT2i use [0.97 (0.91-1.02)] but was lower without recovery prior to SGLT2i use [0.62 (0.55-0.71)] and with SGLT2i use after recovery [0.60 (0.54-0.67)]. Finally, the effect of SGLT2i was stable over time (P for time-interaction 0.19). Thus, we observed reduced mortality with SGLT2i use after AKI among Veterans with diabetic kidney disease whether started earlier or later or before or after observed recovery. Hence, patients with diabetic kidney disease who receive a SGLT2i earlier after AKI experience no significant harm impacting mortality and experience a lower mortality risk than those who do not.

A flexible Bayesian framework for individualized inference via adaptive borrowing.

The explosion in high-resolution data capture technologies in health has increased interest in making inferences about individual-level parameters. While technology may provide substantial data on a single individual, how best to use multisource population data to improve individualized inference remains an open research question. One possible approach, the multisource exchangeability model (MEM), is a Bayesian method for integrating data from supplementary sources into the analysis of a primary source. MEM was originally developed to improve inference for a single study by asymmetrically borrowing information from a set of similar previous studies and was further developed to apply a more computationally intensive symmetric borrowing in the context of basket trial; however, even for asymmetric borrowing, its computational burden grows exponentially with the number of supplementary sources, making it unsuitable for applications where hundreds or thousands of supplementary sources (i.e., individuals) could contribute to inference on a given individual. In this article, we propose the data-driven MEM (dMEM), a two-stage approach that includes both source selection and clustering to enable the inclusion of an arbitrary number of sources to contribute to individualized inference in a computationally tractable and data-efficient way. We illustrate the application of dMEM to individual-level human behavior and mental well-being data collected via smartphones, where our approach increases individual-level estimation precision by 84% compared with a standard no-borrowing method and outperforms recently proposed competing methods in 80% of individuals.

An intersectional framework for counterfactual fairness in risk prediction.

Along with the increasing availability of health data has come the rise of data-driven models to inform decision making and policy. These models have the potential to benefit both patients and health care providers but can also exacerbate health inequities. Existing "algorithmic fairness" methods for measuring and correcting model bias fall short of what is needed for health policy in two key ways. First, methods typically focus on a single grouping along which discrimination may occur rather than considering multiple, intersecting groups. Second, in clinical applications, risk prediction is typically used to guide treatment, creating distinct statistical issues that invalidate most existing techniques. We present novel unfairness metrics that address both challenges. We also develop a complete framework of estimation and inference tools for our metrics, including the unfairness value ("u-value"), used to determine the relative extremity of unfairness, and standard errors and confidence intervals employing an alternative to the standard bootstrap. We demonstrate application of our framework to a COVID-19 risk prediction model deployed in a major Midwestern health system.

Feature selection for support vector regression using a genetic algorithm.

Support vector regression (SVR) is particularly beneficial when the outcome and predictors are nonlinearly related. However, when many covariates are available, the method's flexibility can lead to overfitting and an overall loss in predictive accuracy. To overcome this drawback, we develop a feature selection method for SVR based on a genetic algorithm that iteratively searches across potential subsets of covariates to find those that yield the best performance according to a user-defined fitness function. We evaluate the performance of our feature selection method for SVR, comparing it to alternate methods including LASSO and random forest, in a simulation study. We find that our method yields higher predictive accuracy than SVR without feature selection. Our method outperforms LASSO when the relationship between covariates and outcome is nonlinear. Random forest performs equivalently to our method in some scenarios, but more poorly when covariates are correlated. We apply our method to predict donor kidney function 1 year after transplant using data from the United Network for Organ Sharing national registry.

Machine Learning Identifies Higher Survival Profile In Extracorporeal Cardiopulmonary Resuscitation.

OBJECTIVES: Extracorporeal cardiopulmonary resuscitation (ECPR) has been shown to improve neurologically favorable survival in patients with refractory out-of-hospital cardiac arrest (OHCA) caused by shockable rhythms. Further refinement of patient selection is needed to focus this resource-intensive therapy on those patients likely to benefit. This study sought to create a selection model using machine learning (ML) tools for refractory cardiac arrest patients undergoing ECPR. DESIGN: Retrospective cohort study. SETTING: Cardiac ICU in a Quaternary Care Center. PATIENTS: Adults 18-75 years old with refractory OHCA caused by a shockable rhythm. METHODS: Three hundred seventy-six consecutive patients with refractory OHCA and a shockable presenting rhythm were analyzed, of which 301 underwent ECPR and cannulation for venoarterial extracorporeal membrane oxygenation. Clinical variables that were widely available at the time of cannulation were analyzed and ranked on their ability to predict neurologically favorable survival. INTERVENTIONS: ML was used to train supervised models and predict favorable neurologic outcomes of ECPR. The best-performing models were internally validated using a holdout test set. MEASUREMENTS AND MAIN RESULTS: Neurologically favorable survival occurred in 119 of 301 patients (40%) receiving ECPR. Rhythm at the time of cannulation, intermittent or sustained return of spontaneous circulation, arrest to extracorporeal membrane oxygenation perfusion time, and lactic acid levels were the most predictive of the 11 variables analyzed. All variables were integrated into a training model that yielded an in-sample area under the receiver-operating characteristic curve (AUC) of 0.89 and a misclassification rate of 0.19. Out-of-sample validation of the model yielded an AUC of 0.80 and a misclassification rate of 0.23, demonstrating acceptable prediction ability. CONCLUSIONS: ML can develop a tiered risk model to guide ECPR patient selection with tailored arrest profiles.

A Cluster-Randomized Evaluation of the SuperShelf Intervention in Choice-Based Food Pantries.

BACKGROUND: Interventions in food pantry settings have the potential to improve health among clients at risk of diet-related disease. PURPOSE: This study evaluates whether a cluster-randomized, behavioral intervention in food pantries resulted in improved client outcomes. METHODS: Sixteen Minnesota food pantries were randomized to an intervention (n = 8) or control condition (n = 8). The intervention offered pantries technical assistance to improve healthy food supply and implement behavioral economics strategies to promote healthy food selection. A convenience sample of adult clients were enrolled (paired sample, 158 intervention, 159 control) and followed for 1 year. Additional clients were enrolled at follow-up to assess food selection (follow-up sample, 85 intervention, 102 control). Analysis was limited to data from 11 pantries (5 intervention, 6 control) due to COVID-19. Outcome measures included Healthy Eating Index-2015 (HEI-2015) total and subcomponent scores for 24-hr dietary recalls and client cart selections, and Life's Simple 7 (LS7) total and subcomponent scores. Multilevel mixed-effects models tested whether client outcomes differed by intervention condition. RESULTS: In adjusted models, there were no statistically significant differences by intervention condition in HEI-2015 or LS7 scores. Clients in intervention food pantries had improved Refined Grain subcomponent scores (p = .004); clients in control pantries had worsened Saturated Fat subcomponents scores (p = .019) and improved physical activity scores (p = .007). CONCLUSIONS: The intervention did not result in improved diet quality or cardiovascular health as measured by HEI-2015 or LS7. Coordinated efforts across settings are needed to address health risks facing this population.

Food pantries are an optimal setting to address health and diet quality among clients experiencing food insecurity. This study tests whether a food pantry intervention resulted in improved dietary and cardiovascular outcomes among clients. Sixteen Minnesota food pantries were randomized to either receive an intervention or a delayed intervention. The intervention offered food pantries technical assistance to improve healthy food supply and "nudge" clients toward healthy choices. Due to the COVID-19 pandemic, measures were completed 11 pantries (5 intervention, 6 control). Outcome measures included diet quality of food selected by clients, diet quality of food consumed by clients, and Life's Simple 7 measure of cardiovascular health. The intervention did not result in improved diet quality or cardiovascular health. Coordinated efforts across community settings are needed to address health risks facing this population.

Potential impact of blood cholesterol guidelines on statin treatment in the U.S. population using interrupted time series analysis.

BACKGROUND: The 2013 ACC/AHA Guideline was a paradigm shift in lipid management and identified the four statin-benefit groups. Many have studied the guideline's potential impact, but few have investigated its potential long-term impact on MACE. Furthermore, most studies also ignored the confounding effect from the earlier release of generic atorvastatin in Dec 2011. METHODS: To evaluate the potential (long-term) impact of the 2013 ACC/AHA Guideline release in Nov 2013 in the U.S., we investigated the association of the 2013 ACC/AHA Guideline with the trend changes in 5-Year MACE survival and three other statin-related outcomes (statin use, optimal statin use, and statin adherence) while controlling for generic atorvastatin availability using interrupted time series analysis, called the Chow's test. Specifically, we conducted a retrospective study using U.S. nationwide de-identified claims and electronic health records from Optum Labs Database Warehouse (OLDW) to follow the trends of 5-Year MACE survival and statin-related outcomes among four statin-benefit groups that were identified in the 2013 ACC/AHA Guideline. Then, Chow's test was used to discern trend changes between generic atorvastatin availability and guideline potential impact. RESULTS: 197,021 patients were included (ASCVD: 19,060; High-LDL: 33,907; Diabetes: 138,159; High-ASCVD-Risk: 5,895). After the guideline release, the long-term trend (slope) of 5-Year MACE Survival for the Diabetes group improved significantly (P = 0.002). Optimal statin use for the ASCVD group also showed immediate improvement (intercept) and long-term positive changes (slope) after the release (P < 0.001). Statin uses did not have significant trend changes and statin adherence remained unchanged in all statin-benefit groups. Although no other statistically significant trend changes were found, overall positive trend change or no changes were observed after the 2013 ACC/AHA Guideline release. CONCLUSIONS: The 2013 ACA/AHA Guideline release is associated with trend improvements in the long-term MACE Survival for Diabetes group and optimal statin use for ASCVD group. These significant associations might indicate a potential positive long-term impact of the 2013 ACA/AHA Guideline on better health outcomes for primary prevention groups and an immediate potential impact on statin prescribing behaviors in higher-at-risk groups. However, further investigation is required to confirm the causal effect of the 2013 ACA/AHA Guideline.

Renin-Angiotensin-Aldosterone System Blockade after AKI with or without Recovery among US Veterans with Diabetic Kidney Disease.

SIGNIFICANCE STATEMENT: Among patients with CKD, optimal use of angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers after AKI is uncertain. Despite these medications' ability to reduce risk of mortality and other adverse outcomes, there is concern that ACEi/ARB use may delay recovery of kidney function or precipitate recurrent AKI. Prior studies have provided conflicting data regarding the optimal timing of these medications after AKI and have not addressed the role of kidney recovery in determining appropriate timing. This study in US Veterans with diabetes mellitus and proteinuria demonstrated an association between ACEi/ARB use and lower mortality. This association was more pronounced with earlier post-AKI ACEi/ARB use and was not meaningfully affected by initiating ACEis/ARBs before versus after recovery from AKI. BACKGROUND: Optimal use of angiotensin-converting enzyme inhibitors (ACEis) or angiotensin II receptor blockers (ARBs) after AKI is uncertain. METHODS: Using data derived from electronic medical records, we sought to estimate the association between ACEi/ARB use after AKI and mortality in US military Veterans with indications for such treatment (diabetes and proteinuria) while accounting for AKI recovery. We used ACEi/ARB treatment after hospitalization with AKI (defined as serum creatinine ≥50% above baseline concentration) as a time-varying exposure in Cox models. The outcome was all-cause mortality. Recovery was defined as return to ≤110% of baseline creatinine. A secondary analysis focused on ACEi/ARB use relative to AKI recovery (before versus after). RESULTS: Among 54,735 Veterans with AKI, 31,146 deaths occurred over a median follow-up period of 2.3 years. Approximately 57% received an ACEi/ARB <3 months after hospitalization. In multivariate analysis with time-varying recovery, post-AKI ACEi/ARB use was associated with lower risk of mortality (adjusted hazard ratio [aHR], 0.74; 95% confidence interval [CI], 0.72 to 0.77). The association between ACEi/ARB use and mortality varied over time, with lower mortality risk associated with earlier initiation ( P for interaction with time <0.001). In secondary analysis, compared with those with neither recovery nor ACEi/ARB use, risk of mortality was lower in those with recovery without ACEi/ARB use (aHR, 0.90; 95% CI, 0.87 to 0.94), those without recovery with ACEi/ARB use (aHR, 0.69; 95% CI, 0.66 to 0.72), and those with ACEi/ARB use after recovery (aHR, 0.70; 95% CI, 0.67 to 0.73). CONCLUSIONS: This study demonstrated lower mortality associated with ACEi/ARB use in Veterans with diabetes, proteinuria, and AKI, regardless of recovery. Results favored earlier ACEi/ARB initiation.

External validation of the Norwegian anterior cruciate ligament reconstruction revision prediction model using patients from the STABILITY 1 Trial.

PURPOSE: A machine learning-based anterior cruciate ligament (ACL) revision prediction model has been developed using Norwegian Knee Ligament Register (NKLR) data, but lacks external validation outside Scandinavia. This study aimed to assess the external validity of the NKLR model (https://swastvedt.shinyapps.io/calculator_rev/) using the STABILITY 1 randomized clinical trial (RCT) data set. The hypothesis was that model performance would be similar. METHODS: The NKLR Cox Lasso model was selected for external validation owing to its superior performance in the original study. STABILITY 1 patients with all five predictors required by the Cox Lasso model were included. The STABILITY 1 RCT was a prospective study which randomized patients to receive either a hamstring tendon autograft (HT) alone or HT plus a lateral extra-articular tenodesis (LET). Since all patients in the STABILITY 1 trial received HT ± LET, three configurations were tested: 1: all patients coded as HT, 2: HT + LET group coded as bone-patellar tendon-bone (BPTB) autograft, 3: HT + LET group coded as unknown/other graft choice. Model performance was assessed via concordance and calibration. RESULTS: In total, 591/618 (95.6%) STABILITY 1 patients were eligible for inclusion, with 39 undergoing revisions within 2 years (6.6%). Model performance was best when patients receiving HT + LET were coded as BPTB. Concordance was similar to the original NKLR prediction model for 1- and 2-year revision prediction (STABILITY: 0.71; NKLR: 0.68-0.69). Concordance 95% confidence interval (CI) ranged from 0.63 to 0.79. The model was well calibrated for 1-year prediction while the 2-year prediction demonstrated evidence of miscalibration. CONCLUSION: When patients in STABILITY 1 who received HT + LET were coded as BPTB in the NKLR prediction model, concordance was similar to the index study. However, due to a wide 95% CI, the true performance of the prediction model with this Canadian and European cohort is unclear and a larger data set is required to definitively determine the external validity. Further, better calibration for 1-year predictions aligns with general prediction modelling challenges over longer periods. While not a large enough sample size to elicit the true accuracy and external validity of the prediction model when applied to North American patients, this analysis provides more support for the notion that HT plus LET performs similarly to BPTB reconstruction. In addition, despite the wide confidence interval, this study suggests optimism regarding the accuracy of the model when applied outside of Scandinavia. LEVEL OF EVIDENCE: Level 3, cohort study.

Predictors of Human Milk Feeding and Direct Breastfeeding for Infants with Single Ventricle Congenital Heart Disease: Machine Learning Analysis of the National Pediatric Cardiology Quality Improvement Collaborative Registry.

OBJECTIVE: To identify factors that support or limit human milk (HM) feeding and direct breastfeeding (BF) for infants with single ventricle congenital heart disease at neonatal stage 1 palliation (S1P) discharge and at stage 2 palliation (S2P) (∼4-6 months old). STUDY DESIGN: Analysis of the National Pediatric Cardiology Quality Improvement Collaborative (NPC-QIC) registry (2016-2021; 67 sites). Primary outcomes were any HM, exclusive HM, and any direct BF at S1P discharge and at S2P. The main analysis involved multiple phases of elastic net logistic regression on imputed data to identify important predictors. RESULTS: For 1944 infants, the strongest predictor domain areas included preoperative feeding, demographics/social determinants of health, feeding route, clinical course, and site. Significant findings included: preoperative BF was associated with any HM at S1P discharge (OR = 2.02, 95% CI = 1.74-3.44) and any BF at S2P (OR = 2.29, 95% CI = 1.38-3.80); private/self-insurance was associated with any HM at S1P discharge (OR = 1.91, 95% CI = 1.58-2.47); and Black/African-American infants had lower odds of any HM at S1P discharge (OR = 0.54, 95% CI = 0.38-0.65) and at S2P (0.57, 0.30-0.86). Adjusted odds of HM/BF practices varied among NPC-QIC sites. CONCLUSIONS: Preoperative feeding practices predict later HM and BF for infants with single ventricle congenital heart disease; therefore, family-centered interventions focused on HM/BF during the S1P preoperative time are needed. These interventions should include evidence-based strategies to address implicit bias and seek to minimize disparities related to social determinants of health. Future research is needed to identify supportive practices common to high-performing NPC-QIC sites.

Variation in local Ryan White HIV/AIDS program service use and impacts on viral suppression: informing quality improvement efforts.

The U.S. Ryan White HIV/AIDS Program (RWHAP) funds comprehensive services for people living with HIV to support viral suppression (VS). We analyzed five years of RWHAP data from the Minneapolis-St. Paul region to (1) assess variation and (2) evaluate the causal effect of each RWHAP service on sustained VS by race/ethnicity. Sixteen medical and support services were included. Descriptive analyses assessed service use and trends over time. Causal analyses used generalized estimating equations and propensity scores to adjust for the probability of service use. Receipt of AIDS Drug Assistance Program and financial aid consistently showed higher probabilities of sustained VS, while food aid and transportation aid had positive impacts on VS at higher levels of service encounters; however, the impact of services could vary by race/ethnicity. For example, financial aid increased the probability of sustained VS by at least 3 percentage points for white, Hispanic and Black/African American clients, but only 1.6 points for Black/African-born clients. This study found that services addressing socioeconomic needs typically had positive impacts on viral suppression, yet service use and impact of services often varied by race/ethnicity. This highlights a need to ensure these services are designed and delivered in ways that equitably serve all clients.

Early results of a natural experiment evaluating the effects of a local minimum wage policy on the diet-related health of low-wage workers, 2018-2020.

OBJECTIVE: The current study presents results of a midpoint analysis of an ongoing natural experiment evaluating the diet-related effects of the Minneapolis Minimum Wage Ordinance, which incrementally increases the minimum wage to $15/h. DESIGN: A difference-in-difference (DiD) analysis of measures collected among low-wage workers in two U.S. cities (one city with a wage increase policy and one comparison city). Measures included employment-related variables (hourly wage, hours worked and non-employment assessed by survey questions with wages verified by paystubs), BMI measured by study scales and stadiometers and diet-related mediators (food insecurity, Supplemental Nutrition Assistance Program (SNAP) participation and daily servings of fruits and vegetables, whole-grain rich foods and foods high in added sugars measured by survey questions). SETTING: Minneapolis, Minnesota and Raleigh, North Carolina. PARTICIPANTS: A cohort of 580 low-wage workers (268 in Minneapolis and 312 in Raleigh) who completed three annual study visits between 2018 and 2020. RESULTS: In DiD models adjusted for time-varying and non-time-varying confounders, there were no statistically significant differences in variables of interest in Minneapolis compared with Raleigh. Trends across both cities were evident, showing a steady increase in hourly wage, stable BMI, an overall decrease in food insecurity and non-linear trends in employment, hours worked, SNAP participation and dietary outcomes. CONCLUSION: There was no evidence of a beneficial or adverse effect of the Minimum Wage Ordinance on health-related variables during a period of economic and social change. The COVID-19 pandemic and other contextual factors likely contributed to the observed trends in both cities.

Limited clinical utility of a machine learning revision prediction model based on a national hip arthroscopy registry.

PURPOSE: Accurate prediction of outcome following hip arthroscopy is challenging and machine learning has the potential to improve our predictive capability. The purpose of this study was to determine if machine learning analysis of the Danish Hip Arthroscopy Registry (DHAR) can develop a clinically meaningful calculator for predicting the probability of a patient undergoing subsequent revision surgery following primary hip arthroscopy. METHODS: Machine learning analysis was performed on the DHAR. The primary outcome for the models was probability of revision hip arthroscopy within 1, 2, and/or 5 years after primary hip arthroscopy. Data were split randomly into training (75%) and test (25%) sets. Four models intended for these types of data were tested: Cox elastic net, random survival forest, gradient boosted regression (GBM), and super learner. These four models represent a range of approaches to statistical details like variable selection and model complexity. Model performance was assessed by calculating calibration and area under the curve (AUC). Analysis was performed using only variables available in the pre-operative clinical setting and then repeated to compare model performance using all variables available in the registry. RESULTS: In total, 5581 patients were included for analysis. Average follow-up time or time-to-revision was 4.25 years (± 2.51) years and overall revision rate was 11%. All four models were generally well calibrated and demonstrated concordance in the moderate range when restricted to only pre-operative variables (0.62-0.67), and when considering all variables available in the registry (0.63-0.66). The 95% confidence intervals for model concordance were wide for both analyses, ranging from a low of 0.53 to a high of 0.75, indicating uncertainty about the true accuracy of the models. CONCLUSION: The association between pre-surgical factors and outcome following hip arthroscopy is complex. Machine learning analysis of the DHAR produced a model capable of predicting revision surgery risk following primary hip arthroscopy that demonstrated moderate accuracy but likely limited clinical usefulness. Prediction accuracy would benefit from enhanced data quality within the registry and this preliminary study holds promise for future model generation as the DHAR matures. Ongoing collection of high-quality data by the DHAR should enable improved patient-specific outcome prediction that is generalisable across the population. LEVEL OF EVIDENCE: Level III.

Survival Modelling For Data From Combined Cohorts: Opening the Door to Meta Survival Analyses and Survival Analysis using Electronic Health Records.

Non-parametric estimation of the survival function using observed failure time data depends on the underlying data generating mechanism, including the ways in which the data may be censored and/or truncated. For data arising from a single source or collected from a single cohort, a wide range of estimators have been proposed and compared in the literature. Often, however, it may be possible, and indeed advantageous, to combine and then analyze survival data that have been collected under different study designs. We review non-parametric survival analysis for data obtained by combining the most common types of cohort. We have two main goals: (i) To clarify the differences in the model assumptions, and (ii) to provide a single lens through which some of the proposed estimators may be viewed. Our discussion is relevant to the meta analysis of survival data obtained from different types of study, and to the modern era of electronic health records.

Associations Between Psychosocial Needs, Carbohydrate-Counting Behavior, and App Satisfaction: A Randomized Crossover App Trial on 92 Adults With Diabetes.

To examine whether psychosocial needs in diabetes care are associated with carbohydrate counting and if carbohydrate counting is associated with satisfaction with diabetes applications' usability, a randomized crossover trial of 92 adults with type 1 or 2 diabetes requiring insulin therapy tested two top-rated diabetes applications, mySugr and OnTrack Diabetes. Survey responses on demographics, psychosocial needs (perceived competence, autonomy, and connectivity), carbohydrate-counting frequency, and application satisfaction were modeled using mixed-effect linear regressions to test associations. Participants ranged between 19 and 74 years old (mean, 54 years) and predominantly had type 2 diabetes (70%). Among the three tested domains of psychosocial needs, only competence-not autonomy or connectivity-was found to be associated with carbohydrate-counting frequency. No association between carbohydrate-counting behavior and application satisfaction was found. In conclusion, perceived competence in diabetes care is an important factor in carbohydrate counting; clinicians may improve adherence to carbohydrate counting with strategies designed to improve perceived competence. Carbohydrate-counting behavior is complex; its impact on patient satisfaction of diabetes application usability is multifactorial and warrants consideration of patient demographics such as sex as well as application features for automated carbohydrate counting.

Machine learning algorithm to predict anterior cruciate ligament revision demonstrates external validity.

PURPOSE: External validation of machine learning predictive models is achieved through evaluation of model performance on different groups of patients than were used for algorithm development. This important step is uncommonly performed, inhibiting clinical translation of newly developed models. Machine learning analysis of the Norwegian Knee Ligament Register (NKLR) recently led to the development of a tool capable of estimating the risk of anterior cruciate ligament (ACL) revision ( https://swastvedt.shinyapps.io/calculator_rev/ ). The purpose of this study was to determine the external validity of the NKLR model by assessing algorithm performance when applied to patients from the Danish Knee Ligament Registry (DKLR). METHODS: The primary outcome measure of the NKLR model was probability of revision ACL reconstruction within 1, 2, and/or 5 years. For external validation, all DKLR patients with complete data for the five variables required for NKLR prediction were included. The five variables included graft choice, femur fixation device, KOOS QOL score at surgery, years from injury to surgery, and age at surgery. Predicted revision probabilities were calculated for all DKLR patients. The model performance was assessed using the same metrics as the NKLR study: concordance and calibration. RESULTS: In total, 10,922 DKLR patients were included for analysis. Average follow-up time or time-to-revision was 8.4 (± 4.3) years and overall revision rate was 6.9%. Surgical technique trends (i.e., graft choice and fixation devices) and injury characteristics (i.e., concomitant meniscus and cartilage pathology) were dissimilar between registries. The model produced similar concordance when applied to the DKLR population compared to the original NKLR test data (DKLR: 0.68; NKLR: 0.68-0.69). Calibration was poorer for the DKLR population at one and five years post primary surgery but similar to the NKLR at two years. CONCLUSION: The NKLR machine learning algorithm demonstrated similar performance when applied to patients from the DKLR, suggesting that it is valid for application outside of the initial patient population. This represents the first machine learning model for predicting revision ACL reconstruction that has been externally validated. Clinicians can use this in-clinic calculator to estimate revision risk at a patient specific level when discussing outcome expectations pre-operatively. While encouraging, it should be noted that the performance of the model on patients undergoing ACL reconstruction outside of Scandinavia remains unknown. LEVEL OF EVIDENCE: III.

Comparison of 3-factor versus 4-factor prothrombin complex concentrate for emergent warfarin reversal: a systematic review and meta-analysis.

BACKGROUND: Patients requiring emergent warfarin reversal (EWR) have been prescribed three-factor prothrombin complex concentrate (PCC3) and four-factor prothrombin complex concentrate (PCC4) to reverse the anticoagulant effects of warfarin. There is no existing systematic review and meta-analysis of studies directly comparing PCC3 and PCC4. METHODS: The primary objective of this systematic review and meta-analysis was to determine the effectiveness of achieving study defined target INR goal after PCC3 or PCC4 administration. Secondary objectives were to determine the difference in safety endpoints, thromboembolic events (TE), and survival during the patients' hospital stay. Random-effects meta-analysis models were used to estimate the odds ratios (OR), and heterogeneity associated with the outcomes. The Newcastle-Ottawa Scale was used to assess study quality, and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed. RESULTS: Ten full-text manuscripts and five abstracts provided data for the primary and secondary outcomes. Patients requiring EWR had more than three times the odds of reversal to goal INR when they were given PCC4 compared to PCC3 (OR = 3.61, 95% CI: 1.97-6.60, p < 0.001). There was no meaningful clinical association or statistically significant result between PCC4 and PCC3 groups in TE (OR = 1.56, 95% CI: 0.83-2.91, p = 0.17), or survival during hospital stay (OR = 1.34, 95% CI: 0.81-2.23, p = 0.25). CONCLUSION: PCC4 is more effective than PCC3 in meeting specific predefined INR goals and has similar safety profiles in patients requiring emergent reversal of the anticoagulant effects of warfarin.

Iterated multisource exchangeability models for individualized inference with an application to mobile sensor data.

Researchers are increasingly interested in using sensor technology to collect accurate activity information and make individualized inference about treatments, exposures, and policies. How to optimally combine population data with data from an individual remains an open question. Multisource exchangeability models (MEMs) are a Bayesian approach for increasing precision by combining potentially heterogeneous supplemental data sources into analysis of a primary source. MEMs are a potentially powerful tool for individualized inference but can integrate only a few sources; their model space grows exponentially, making them intractable for high-dimensional applications. We propose iterated MEMs (iMEMs), which identify a subset of the most exchangeable sources prior to fitting a MEM model. iMEM complexity scales linearly with the number of sources, and iMEMs greatly increase precision while maintaining desirable asymptotic and small sample properties. We apply iMEMs to individual-level behavior and emotion data from a smartphone app and show that they achieve individualized inference with up to 99% efficiency gain relative to standard analyses that do not borrow information.

Reducing potentially preventable health events among patients with asthma through multi-state, multi-center quality improvement program.

INTRODUCTION: Enhancing Care for Patients with Asthma is a multi-state, multi-center quality improvement program developed to augment guideline-based practice among health care providers through Plan-Do-Study-Act cycle. This study examined the association between the implementation of the guideline-based quality improvement program and subsequent changes in asthma-related emergency room visits and hospitalizations. METHODS: This retrospective, interrupted time-series study used administrative claims data from a private insurer that provided coverage to patients receiving care from participating health centers (15 centers in New Mexico, Oklahoma, Texas, and Illinois). The 12-month implementation period started in January 2013 for centers in Cohort 1 and October 2013 for centers in Cohort 2. The claims of 1,828 patients with asthma from January 2012 to May 2015 were analyzed. The data included 12-month pre-program implementation, 12-month program implementation, and 5-month post-program completion periods. RESULTS: The average number of asthma-related emergency room visits and hospitalizations decreased from 2.22 to 1.38 and 1.97 to 1.04 per 100 patients per month, respectively, in the 12-month pre-implementation period as compared to 12-month implementation period. The results of three-level generalized linear mixed models found that during the 12-month implementation period, patients had 37.7% and 47.1% lower rates of emergency room visits and hospitalizations, respectively, compared to the 12-month pre-implementation period (p < 0.001 in both comparisons). CONCLUSIONS: Enhancing Care for Patients with Asthma is an effective quality improvement program that was successfully executed in diverse geographical states and associated with reductions in potentially preventable health events. The findings support the widespread use of the program in other settings.

The determinants of racial disparities in obesity: baseline evidence from a natural experiment.

This paper uses baseline data from an observational study to estimate the determinants of racial and gender disparities in obesity. Samples of low-income workers in Minneapolis and Raleigh reveal that respondents in Minneapolis have lower Body Mass Indices (BMIs) than respondents in Raleigh. There are large, statistically significant race and gender effects in estimates of BMI that explain most of the disparity between the two cities. Accounting for intersectionality - the joint impacts of being Black and a woman - reveals that almost all the BMI gaps between Black women in Minneapolis and Raleigh can be explained by age and education differences.