Highest Risk Quintile Research Articles

Comparing five-year and ten-year predicted cardiovascular disease risk in Aotearoa New Zealand: national data linkage study of 1.7 million adults.

There is no consensus on the optimal time horizon for predicting cardiovascular disease (CVD) risk to inform treatment decisions. New Zealand and Australia recommend 5 years, whereas most countries recommend 10 years. We compared predicted risk and treatment-eligible groups using 5-year and 10-year equations. Individual-level linked administrative datasets identified 1,746,665 New Zealanders without CVD, aged 30-74 years in 2006, with follow-up to 2018. Participants were randomly allocated to derivation and validation cohorts. Sex-specific 5-year and 10-year risk prediction models were developed in the derivation cohort and applied in the validation cohort. 28,116 (3.2%) and 62,027 (7.1%) first CVD events occurred during 5-years and 10-years follow-up respectively (cumulative risk, derivation cohort). Median predicted 10-year CVD risk (3.8%) was approximately 2.5 times 5-year risk (1.6%) and 95% of individuals in the top quintile of 5-year risk were also in the top quintile of 10-year risk, across age/gender groups (validation cohort). Using common guideline-recommended treatment thresholds (5% 5-year, 10% 10-year risk), approximately 14% and 28% of women and men respectively were identified as treatment-eligible applying 5-year equations compared to 17% and 32% of women and men applying 10-year equations. Older age was the major contributor to treatment eligibility in both sexes. Predicted 10-year CVD risk was approximately 2.5 times 5-year risk. Both equations identified mostly the same individuals in the highest risk quintile. Conversely, commonly used treatment thresholds identified more treatment-eligible individuals using 10-year equations and both equations identified approximately twice as many treatment-eligible men as women. The treatment threshold, rather than the risk horizon, is the main determinant of treatment eligibility.

Point-based risk score for the risk stratification and prediction of hepatocellular carcinoma: a population-based random survival forest modeling study

The precise associations between common clinical biomarkers and hepatocellular carcinoma (HCC) risk remain unclear but hold valuable insights for HCC risk stratification and prediction. We examined the linear and nonlinear associations between the baseline levels of 32 circulating biomarkers and HCC risk in the England cohort of UK Biobank (UKBB) (n=397,702). The participants were enrolled between 2006 and 2010 and followed up to 31st October 2022. The primary outcome is incident HCC cases. We then employed random survival forests (RSF) to select the top ten most informative biomarkers, considering their association with HCC, and developed a point-based risk score to predict HCC. The performance of the risk score was evaluated in three validation sets including UKBB Scotland and Wales cohort (n=52,721), UKBB non-White-British cohort (n=29,315), and the Taizhou Longitudinal Study in China (n=17,269). Twenty-five biomarkers were significantly associated with HCC risk, either linearly or nonlinearly. Based on the RSF model selected biomarkers, our point-based risk score showed a concordance index of 0.866 in the England cohort and varied between 0.814 and 0.849 in the three validation sets. HCC incidence rates ranged from 0.95 to 30.82 per 100,000 from the lowest to the highest quintiles of the risk score in the England cohort. Individuals in the highest risk quintile had a 32-73 times greater risk of HCC compared to those in the lowest quintile. Moreover, over 70% of HCC cases were detected in individuals within the top risk score quintile across all cohorts. Our simple risk score enables the identification of high-risk individuals of HCC in the general population. However, including some biomarkers, such as insulin-like growth factor 1, not routinely measured in clinical practice may increase the model's complexity, highlighting the need for more accessible biomarkers that can maintain or improve the predictive accuracy of the risk score. This work was supported by the National Natural Science Foundation of China (grant numbers: 82204125) and the Science and Technology Support Program of Taizhou (TS202224).

Risk-stratified analysis of long-term clinical outcomes and cumulative costs in Finnish patients with recent acute coronary syndrome or coronary revascularization: a 5-year real-world study using electronic health records.

Risk assessment is essential in the prevention of cardiovascular disease. In patients with recent acute coronary syndrome (ACS) or coronary revascularization, risk prediction tools, like the European Society of Cardiology guideline recommended SMART-REACH risk score, are increasingly used to predict the risk of recurrent cardiovascular events enabling risk-based personalized prevention. However, little is known about the association between risk stratification and the social and healthcare costs at a population level. This study evaluated the associations between baseline SMART-REACH risk scores, long-term recurrent clinical events, cumulative costs, and post-index event LDL-C goal attainment in patients with recent ACS and/or revascularization. This retrospective study used electronic health records and was conducted in the North Karelia region of Finland. The study cohort included all patients aged 45-85 admitted to a hospital for ACS or who underwent percutaneous coronary intervention or coronary artery bypass surgery between 1 January 2017 and 31 December 2021. Patients were divided into quintiles based on their baseline SMART-REACH risk scores to examine the associations between predicted 5-year scores and selected clinical and economic outcomes. In addition, simple age-based stratification was conducted as a sensitivity analysis. The observed 5-year cumulative incidence of recurrent events ranged from 20% in the lowest to 41% in the highest risk quintile, whereas the corresponding predicted risks ranged from 13% to 51%, and cumulative 5-year mean total costs per patient ranged from 15 827 to 46 182€, respectively. Both monitoring and attainment of low LDL-C values were suboptimal. The use of the SMART-REACH quintiles as a population-level risk stratification tool successfully stratified patients into subgroups with different cumulative numbers of recurrent events and cumulative total costs. However, more research is needed to define clinically and economically optimal threshold values for a population-level stratification.

Self-Reported Social Determinants of Health and Area-Level Social Vulnerability

Many health care systems are investing resources in identifying social determinants of health (SDoH) needs and facilitating interventions among the populations they serve. Because self-reported SDoH information is lacking, area-level measures are often used to estimate needs and direct resources. To describe the large-scale deployment of SDoH assessments by a health system and determine the extent to which self-reported SDoH needs identified therein are associated with census tract-level social vulnerability measured using the Social Vulnerability Index (SVI). This cross-sectional study assessed SDoH needs between January 1, 2020, and April 30, 2023, in both payer and clinical care settings. Modalities included telephonic outreach, face-to-face clinical interactions, self-entry into a tablet or kiosk, and web-based survey tools. Participants included individuals who responded to the assessment and had sufficient information for census tract identification. Respondents included both Highmark Health Plan members and nonmembers. Health plan members responded to the assessment through health plan programs or platforms, and both members and nonmembers responded to assessments during inpatient or outpatient encounters with the affiliated health system. Overall and domain-specific SDoH needs self-reported through assessments, and severity and complexity of needs identified. Residential social vulnerability measures included overall SVI and the 4 conceptual themes comprising overall SVI. In total, 841 874 assessments were recorded for 401 697 individuals (55.1% women; median [IQR] age, 55 [41-70] years). Social determinants of health needs were identified in 120 769 assessments (14.3%). Across all SDoH domains, increasing SVI was associated with a higher positivity rate (eg, 11.2% of those residing in the lowest-risk SVI quintile reported a need compared with 22.7% among those residing in the highest-risk quintile). Associations varied by SDoH domain and SVI theme. After adjusting for demographic and screening characteristics, odds of positive screening among those residing in the highest-risk SVI quintile were 1.74 (95% CI, 1.62-1.86) to 3.73 (95% CI, 3.48-4.00) times the odds among those residing in lowest risk quintile. In this cross-sectional study, the overall level of SDoH needs generally corresponded to area-level vulnerability. Some SDoH domains appeared far more sensitive to community characteristics than others. Notably, even among individuals from the highest-risk areas, the positive screening rate was roughly 1 in 4. These findings underscore the importance of individual-level SDoH data for service provision planning and health services research.

318 Discovering Subgroups with Supervised Machine Learning Models for Heterogeneity of Treatment Effect Analysis

OBJECTIVES/GOALS: The goal of the study is to provide insights into the use of machine learning methods as a means to predict heterogeneity of treatment effect (HTE) in participants of randomized clinical trials. METHODS/STUDY POPULATION: Using data from 2,441 participants enrolled in the ASPirin in Reducing Events in the Elderly (ASPREE) randomized controlled trial of daily low-dose aspirin vs placebo in the United States, we developed multivariable risk prediction models for the composite outcome of dementia, disability, or death. We used two machine learning techniques, decision trees and random forests, to develop novel non-parametric outcomes classifiers and generate risk-based subgroups. The comparator method was an extant semi-parametric proportional hazards predictive risk model. We then assessed HTE by examining the 5-year absolute risk reduction (ARR) of aspirin vs placebo in each risk subgroup. RESULTS/ANTICIPATED RESULTS: In the random forest classifier, the ARR at 5 years in the highest risk quintile was 13.7% (95% CI 3.1% to 24.4%). For the semi-parametric proportional hazards model, the ARR in the highest risk quintile was 15.1% (95% CI 4.0% to 26.3%). These results were comparable and provide evidence of the viability of internally developed parsimonious non-parametric machine learning models for HTE analysis. The decision tree model results (5-year ARR = 17.0%, 95% CI= -5.4% to 39.4% in the highest risk subgroup) exhibited more uncertainty in the results. DISCUSSION/SIGNIFICANCE: None of the models detected significant HTE on the relative scale; there was substantial HTE on the absolute scale in three of the models. Treatment benefit on the absolute scale may be regarded as bearing greater clinical importance and may be present even in the absence of benefit on the relative scale.

A polygenic risk score predicts atrial fibrillation in patients with coronary artery disease without severe systolic dysfunction

Abstract Background Genome-wide polygenic risk scores (GPS) comprised of multiple common variants can identify individuals within the general population at significantly higher risk for developing atrial fibrillation AF; however, the clinical utility in general populations is limited by the low absolute risk of AF. Purpose To determine whether a genome-wide polygenic score for AF (GPSAF) can be utilized to identify a patient population at high absolute risk for the development of AF among older patients with established coronary artery disease (CAD) and left ventricular ejection fraction (LVEF) > 30-35%. Methods Participants in the PREDETERMINE study without a prior history of AF who had claims data available in Centers for Medicare and Medicaid Services (CMS) database were followed for incident AF defined as International Classification of Diseases codes 427.3 and I48. A recently validated GPSAF comprised of 1.1 million variants was generated and participants were divided into quintiles based upon the distribution of the GPSAF in the general population. The population was separated into 3 categories of genetic risk, with the top quintile classified as high risk, the second through fourth quintiles classified as intermediate risk, and the bottom quintile classified as low risk. Cumulative incidence curves for incident AF were plotted stratified by the GPSAF categories and the association between GPSAF and incident AF were estimated using Cox proportional hazard models with and without adjusting for established AF risk factors and biomarkers (N-terminal pro-B-type natriuretic peptide, high-sensitivity Troponin I, high sensitivity c-reactive protein). Results Over a median follow-up of 8.0 years, there were 482 incident AF events among 2272 study participants. Overall, median age was 67 [62,73] years and 73.5% were men. Baseline AF risk factors and biomarkers did not differ across GPSAF risk categories. The unadjusted cumulative incidence of AF over the course of the study was 39.2% (29.2%-47.8%) in the high GPSAF risk group, 32.5% (25.8%-38.5%) in the intermediate GPSAF risk group, and 21.6% (13.4%-29.9%) in the low GPSAF risk group (Figure 1). Compared with the participants with low GPSAF risk, participants with high and intermediate GPSAF risk had respectively t 2.32-fold (95% CI, 1.71-3.14; p=0.001) and 1.64-fold (95% CI, 1.27-2.10; p=0.001) increased risk for incident AF, adjusted for age and sex. Both remained significantly elevated after controlling for clinical factors, and biomarkers (Table 1). Conclusion A polygenic risk score for AF is strongly associated with incident AF risk among patients with CAD without severe systolic dysfunction, and clinically meaningful elevations in absolute risk of developing AF were observed in those in the highest quintile of risk. Future research is needed to determine if genetic risk assessment and targeted screening strategies will result in reduce morbidity from AF in this high-risk population.

Predicting All-Cause Mortality in Women With and Without Breast Cancer Using the Schonberg Index: A Women's Health Initiative Study.

When treating older women with breast cancer, life expectancy is an important consideration. ASCO recommends calculating 10-year mortality probabilities to inform treatment decisions. One useful tool is the Schonberg index, which predicts risk-based all-cause 10-year mortality. We investigated the use of this index in women aged ≥65 years with breast cancer in the Women's Health Initiative (WHI). We calculated 10-year mortality risk scores for 2,549 WHI participants with breast cancer ("cases") and 2,549 age-matched breast cancer-free participants ("controls") using Schonberg index risk scoring. Risk scores were grouped into quintiles for comparisons. Risk-stratified observed mortality rates and 95% confidence intervals were compared across cases and controls. Observed 10-year mortality rates in cases and controls were also compared with Schonberg index-based predicted 10-year mortality rates. Compared with controls, cases were more often white (P=.005), had higher income and education levels (P<.001 for both), more often lived with their husband/partner (P<.001), scored higher on subjective health/happiness (P<.001), and needed less assistance in activities of daily living (P<.001). Participants with breast cancer had similar risk-stratified 10-year mortality rates compared with controls (34% vs 33%, respectively). Stratified results showed that cases had slightly higher mortality rates than controls in the lowest risk quintile and lower mortality rates in the 2 highest risk quintiles. Observed mortality rates in cases and controls were similar to Schonberg index-predicted mortality, with model c-indexes of 0.71 and 0.76, respectively. Among women aged ≥65 years with incident breast cancer, the Schonberg index-based risk-stratified 10-year mortality rates were similar to those in women without breast cancer, demonstrating a similar performance of the index among both populations. Along with other health measures, prognostic indexes can help predict survival among older women with breast cancer and support geriatric oncology guidelines that promote using life expectancy calculation tools for shared decision-making.

Cartilage Acidic Protein 1 in Plasma Associates With Prevalent Osteoarthritis and Predicts Future Risk as Well as Progression to Joint Replacements: Results From the UK Biobank Resource.

The level of cartilage acidic protein 1 (CRTAC1) in plasma was recently discovered to be associated with osteoarthritis (OA) risk and progression to joint replacement in Iceland. This study was undertaken to validate these findings in an independent population. In this study, 1,462 plasma proteins were measured in 54,265 participants from the UK Biobank on the Olink Explore platform. We analyzed the association of plasma proteins with prevalent OA, incident OA, and progression to joint replacement. We assessed the specificity of OA association through comparison of associations with inflammatory joint diseases and with previous joint replacement. The CRTAC1 protein showed the strongest association with prevalent knee OA (odds ratio [OR] 1.34 [95% confidence interval (95% CI) 1.27, 1.41]) and was associated with hip OA (OR 1.19 [95% CI 1.11, 1.28]). It predicted incident diagnosis of OA in the knee (hazard ratio [HR] 1.40 [95% CI 1.35, 1.46]) and hip (HR 1.25 [95% CI 1.19, 1.31]), as well as progression to joint replacement (HR 1.20 [95% CI 1.08, 1.33] for the knee and HR 1.22 [95% CI 1.08, 1.38] for the hip), while no association was found with inflammatory joint diseases. Individuals in the highest quintile of risk based on CRTAC1 level, age, sex, and body mass index had a 10-fold risk of knee or hip OA within 5 years compared to those in the lowest quintile. Adding aggrecan core protein (ACAN) and neurocan core protein (NCAN) to the model improved the prediction of OA but not joint replacement. Furthermore, we replicated the association of CUB domain-containing protein 1 with prior joint replacement. Plasma CRTAC1 is a specific biomarker for OA and a predictor of OA risk and progression to joint replacement. Adding ACAN and NCAN protein levels to the CRTAC1 model improved the prediction of OA.

Applying a genetic risk score for prostate cancer to men with lower urinary tract symptoms in primary care to predict prostate cancer diagnosis: a cohort study in the UK Biobank

BackgroundProstate cancer is highly heritable, with >250 common variants associated in genome-wide association studies. It commonly presents with non-specific lower urinary tract symptoms that are frequently associated with benign conditions.MethodsCohort study using UK Biobank data linked to primary care records. Participants were men with a record showing a general practice consultation for a lower urinary tract symptom. The outcome measure was prostate cancer diagnosis within 2 years of consultation. The predictor was a genetic risk score of 269 genetic variants for prostate cancer.ResultsA genetic risk score (GRS) is associated with prostate cancer in symptomatic men (OR per SD increase = 2.12 [1.86–2.41] P = 3.5e-30). An integrated risk model including age and GRS applied to symptomatic men predicted prostate cancer (AUC 0.768 [0.739–0.796]). Prostate cancer incidence was 8.1% (6.7–9.7) in the highest risk quintile. In the lowest quintile, prostate cancer incidence was <1%.ConclusionsThis study is the first to apply GRS in primary care to improve the triage of symptomatic patients. Men with the lowest genetic risk of developing prostate cancer could safely avoid invasive investigation, whilst those identified with the greatest risk could be fast-tracked for further investigation. These results show that a GRS has potential application to improve the diagnostic pathway of symptomatic patients in primary care.

Risk stratification and cost-effectiveness analysis of adult patients receiving extracorporeal membrane oxygenation.

A more effective allocation of critical care resources is important as the cost of intensive care increases. A model has been developed to predict the probability of in-hospital death among patients who received extracorporeal membrane oxygenation (ECMO). Cost-effectiveness analyses (CEA) were performed regarding the relationship between hospitalization expenses and predicted survival outcomes. Adult patients who received ECMO in a medical center in Taiwan (2005-2016) were included. A logistic regression model was applied to a spectrum of clinical measures obtained before and during ECMO institutions to identify the risk variables for in-hospital mortality. CEA were reported as a predictive risk in quintiles and defined as the cost of each quality-adjusted life-year (QALY). The distribution of the cost-effectiveness ratio (CER) was measured by the ellipse and acceptability curve methods. A total of 919 patients (659 males, mean age: 53.7 years) were enrolled. Ten variables emerged as significant predictors of in-hospital death. The area under the receiver operating characteristic curve was 0.75 (95% confidence interval: 0.72-0.79). In-hospital and total follow-up times were 40,366 and 660,205 person-days, respectively. The total in-hospital expense was $31,818,701 USD and the total effectiveness was 1687.3 QALY. For the lowest to the highest risk quintile, the mean mortality risks were 0.30, 0.48, 0.61, 0.75, and 0.88, and mean adjusted CER were $24,230, $43,042, $54,929, $84,973, and $149,095 per QALY, respectively. The efficient allocation of limited and costly resources is most important when one is forced to decide between groups of critically ill patients. The current analyses of ECMO outcomes should assist in identifying candidates with the greatest prospect for survival while avoiding futile treatments.

The CRTAC1 Protein in Plasma Is Associated With Osteoarthritis and Predicts Progression to Joint Replacement: A Large-Scale Proteomics Scan in Iceland.

ObjectiveBiomarkers for diagnosis and progression of osteoarthritis (OA) are lacking. This study was undertaken to identify circulating biomarkers for OA that could predict disease occurrence and/or progression to joint replacement.MethodsUsing the SomaScan platform, we measured 4,792 proteins in plasma from 37,278 individuals, of whom 12,178 individuals had OA and 2,524 had undergone joint replacement. We performed a case–control study for identification of potential protein biomarkers for hip, knee, and/or hand OA, and a prospective study for identification of biomarkers for joint replacement.ResultsAmong the large panel of plasma proteins assessed, cartilage acidic protein 1 (CRTAC1) was the most strongly associated with both OA diagnosis (odds ratio 1.46 [95% confidence interval 1.41–1.52] for knee OA, odds ratio 1.36 [95% confidence interval 1.29–1.43] for hip OA, and odds ratio 1.33 [95% confidence interval 1.26–1.40] for hand OA) and progression to joint replacement (hazard ratio 1.40 [95% confidence interval 1.30–1.51] for knee replacement and hazard ratio 1.31 [95% confidence interval 1.19–1.45] for hip replacement). Patients with OA who were in the highest quintile of risk of joint replacement, based on known risk factors (i.e., age, sex, and body mass index) and plasma CRTAC1 level, were 16 times more likely to undergo knee replacement within 5 years of plasma sample collection than those in the lowest quintile, and 6.5 times more likely to undergo hip replacement. CRTAC1 was not associated with other types of inflammatory arthritis. A specific protein profile was identified in those patients who had undergone joint replacement prior to plasma sample collection.ConclusionThrough a hypothesis‐free approach, we identified CRTAC1 in plasma as a novel promising candidate biomarker for OA that is both associated with occurrence of OA and predictive of progression to joint replacement. This biomarker might also be useful in the selection of suitable patients for clinical trial enrollment.

A Machine Learning Model Approach to Risk-Stratify Patients With Gastrointestinal Cancer for Hospitalization and Mortality Outcomes.

Patients with gastrointestinal (GI) cancer frequently experience unplanned hospitalizations, but predictive tools to identify high-risk patients are lacking. We developed a machine learning model to identify high-risk patients. In the study, 1341 consecutive patients undergoing GI (abdominal or pelvic) radiation treatment (RT) from March 2016 to July 2018 (derivation) and July 2018 to January 2019 (validation) were assessed for unplanned hospitalizations within 30 days of finishing RT. In the derivation cohort of 663 abdominal and 427 pelvic RT patients, a machine learning approach derived random forest, gradient boosted decision tree, and logistic regression models to predict 30-day unplanned hospitalizations. Model performance was assessed using area under the receiver operating characteristic curve (AUC) and prospectively validated in 161 abdominal and 90 pelvic RT patients using Mann-Whitney rank-sum test. Highest quintile of risk for hospitalization was defined as "high-risk" and the remainder "low-risk." Hospitalizations for high- versus low-risk patients were compared using Pearson's χ2 test and survival using Kaplan-Meier log-rank test. Overall, 13% and 11% of patients receiving abdominal and pelvic RT experienced 30-day unplanned hospitalization. In the derivation phase, gradient boosted decision tree cross-validation yielded AUC=0.823 (abdominal patients) and random forest yielded AUC=0.776 (pelvic patients). In the validation phase, these models yielded AUC=0.749 and 0.764, respectively (P < .001 and P=.002). Validation models discriminated high- versus low-risk patients: in abdominal RT patients, frequency of hospitalization was 39% versus 9% in high- versus low-risk groups (P < .001) and 6-month survival was 67% versus 92% (P=.001). In pelvic RT patients, frequency of hospitalization was 33% versus 8% (P=.002) and survival was 86% versus 92% (P=.15) in high- versus low-risk patients. In patients with GI cancer undergoing RT as part of multimodality treatment, machine learning models for 30-day unplanned hospitalization discriminated high- versus low-risk patients. Future applications will test utility of models to prompt interventions to decrease hospitalizations and adverse outcomes.

Automated imaging-based stratification of early-stage lung cancer patients prior to receiving surgical resection using deep learning applied to CTs.

1552 Background: Computed tomography (CT) imaging is an important tool to guide further investigation and treatment in patients with lung cancer. For patients with early stage lung cancer, surgery remains an optimal treatment option. Artificial intelligence applied to pretreatment CTs may have the ability to quantify mortality risk and stratify patients for more individualized diagnostic, treatment and monitoring decisions. Methods: A fully automated, end-to-end model was designed to localize the 36cm x 36cm x 36cm space centered on the lungs and learn deep prognostic features using a 3-dimensional convolutional neural network (3DCNN) to predict 5-year mortality risk. The 3DCNN was trained and validated in a 5-fold cross-validation using 2,924 CTs of 1,689 lung cancer patients from 6 public datasets made available in The Cancer Imaging Archive. We evaluated 3DCNN’s ability to stratify stage I &amp; II patients who received surgery into mortality risk quintiles using the Cox proportional hazards model. Results: 260 of the 1,689 lung cancer patients in the withheld validation dataset were diagnosed as stage I or II, received a surgical resection within 6 months of their pretreatment CT and had known 5-year disease and survival outcomes. Based on the 3DCNN’s predicted mortality risk, patients in the highest risk quintile had a 14.2-fold (95% CI 4.3-46.8, p &lt; 0.001) increase in 5-year mortality hazard compared to patients in the lowest risk quintile. Conclusions: Deep learning applied to pretreatment CTs provides personalised prognostic insights for early stage lung cancer patients who received surgery and has the potential to inform treatment and monitoring decisions.[Table: see text]

Multilevel analysis of geographic variation among correlates of child undernutrition in India.

Prior research has identified a number of risk factors ranging from inadequate household sanitation to maternal characteristics as important determinants of child malnutrition and health in India. What is less known is the extent to which these individual‐level risk factors are geographically distributed. Assessing the geographic distribution, especially at multiple levels, matters as it can inform where, and at what level, interventions should be targeted. The three levels of significance in the Indian context are villages, districts, and states. Thus, the purpose of this paper was to (a) examine what proportion of the variation in 21 risk factors is attributable to villages, districts, and states in India and (b) elucidate the specific states where these risk factors are clustered within India. Using the fourth National Family Health Survey dataset, from 2015 to 2016, we found that the proportion of variation attributable to villages ranged from 14% to 63%, 10% to 29% for districts and 17% to 62% for states. Furthermore, we found that Bihar, Jharkhand, Madhya Pradesh, and Uttar Pradesh were in the highest risk quintile for more than 10 of the risk factors included in our study. This is an indication of geographic clustering of risk factors. The risk factors that are clustered in states such as Bihar, Jharkhand, Madhya Pradesh and Uttar Pradesh underscore the need for policies and interventions that address a broader set of child malnutrition determinants beyond those that are nutrition specific.

Prognostic models for predicting the risk of foot ulcer or amputation in people with type 2 diabetes: a systematic review and external validation study

Aims/hypothesisApproximately 25% of people with type 2 diabetes experience a foot ulcer and their risk of amputation is 10–20 times higher than that of people without type 2 diabetes. Prognostic models can aid in targeted monitoring but an overview of their performance is lacking. This study aimed to systematically review prognostic models for the risk of foot ulcer or amputation and quantify their predictive performance in an independent cohort.MethodsA systematic review identified studies developing prognostic models for foot ulcer or amputation over minimal 1 year follow-up applicable to people with type 2 diabetes. After data extraction and risk of bias assessment (both in duplicate), selected models were externally validated in a prospective cohort with a 5 year follow-up in terms of discrimination (C statistics) and calibration (calibration plots).ResultsWe identified 21 studies with 34 models predicting polyneuropathy, foot ulcer or amputation. Eleven models were validated in 7624 participants, of whom 485 developed an ulcer and 70 underwent amputation. The models for foot ulcer showed C statistics (95% CI) ranging from 0.54 (0.54, 0.54) to 0.81 (0.75, 0.86) and models for amputation showed C statistics (95% CI) ranging from 0.63 (0.55, 0.71) to 0.86 (0.78, 0.94). Most models underestimated the ulcer or amputation risk in the highest risk quintiles. Three models performed well to predict a combined endpoint of amputation and foot ulcer (C statistics >0.75).Conclusions/interpretationThirty-four prognostic models for the risk of foot ulcer or amputation were identified. Although the performance of the models varied considerably, three models performed well to predict foot ulcer or amputation and may be applicable to clinical practice.Graphical abstract

Association Between Patient Social Risk and Physician Performance Scores in the First Year of the Merit-based Incentive Payment System

The US Merit-based Incentive Payment System (MIPS) is a major Medicare value-based payment program aimed at improving quality and reducing costs. Little is known about how physicians' performance varies by social risk of their patients. To determine the relationship between patient social risk and physicians' scores in the first year of MIPS. Cross-sectional study of physicians participating in MIPS in 2017. Physicians in the highest quintile of proportion of dually eligible patients served; physicians in the 3 middle quintiles; and physicians in the lowest quintile. The primary outcome was the 2017 composite MIPS score (range, 0-100; higher scores indicate better performance). Payment rates were adjusted -4% to 4% based on scores. The final sample included 284 544 physicians (76.1% men, 60.1% with ≥20 years in practice, 11.9% in rural location, 26.8% hospital-based, and 24.6% in primary care). The mean composite MIPS score was 73.3. Physicians in the highest risk quintile cared for 52.0% of dually eligible patients; those in the 3 middle risk quintiles, 21.8%; and those in the lowest risk quintile, 6.6%. After adjusting for medical complexity, the mean MIPS score for physicians in the highest risk quintile (64.7) was lower relative to scores for physicians in the middle 3 (75.4) and lowest (75.9) risk quintiles (difference for highest vs middle 3, -10.7 [95% CI, -11.0 to -10.4]; highest vs lowest, -11.2 [95% CI, -11.6 to -10.8]; P < .001). This relationship was found across specialties except psychiatry. Compared with physicians in the lowest risk quintile, physicians in the highest risk quintile were more likely to work in rural areas (12.7% vs 6.4%; difference, 6.3 percentage points [95% CI, 6.0 to 6.7]; P < .001) but less likely to care for more than 1000 Medicare beneficiaries (9.4% vs 17.8%; difference, -8.3 percentage points [95% CI, -8.7 to -8.0]; P < .001) or to have more than 20 years in practice (56.7% vs 70.6%; difference, -13.9 percentage points [95% CI, -14.4 to -13.3]; P < .001). For physicians in the highest risk quintile, several characteristics were associated with higher MIPS scores, including practicing in a larger group (mean score, 82.4 for more than 50 physicians vs 46.1 for 1-5 physicians; difference, 36.2 [95% CI, 35.3 to 37.2]; P < .001) and reporting through an alternative payment model (mean score, 79.5 for alternative payment model vs 59.9 for reporting as individual; difference, 19.7 [95% CI, 18.9 to 20.4]; P < .001). In this cross-sectional analysis of physicians who participated in the first year of the Medicare MIPS program, physicians with the highest proportion of patients dually eligible for Medicare and Medicaid had significantly lower MIPS scores compared with other physicians. Further research is needed to understand the reasons underlying the differences in physician MIPS scores by levels of patient social risk.

Can risk modelling improve treatment decisions in asymptomatic carotid stenosis?

BackgroundCarotid endarterectomy (CEA) is routinely performed for asymptomatic carotid stenosis, yet its average net benefit is small. Risk stratification may identify high risk patients that would clearly benefit from treatment.MethodsRetrospective cohort study using data from the Asymptomatic Carotid Atherosclerosis Study (ACAS). Risk factors for poor outcomes were included in backward and forward selection procedures to develop baseline risk models estimating the risk of non-perioperative ipsilateral stroke/TIA. Baseline risk was estimated for all ACAS participants and externally validated using data from the Atherosclerosis Risk in Communities (ARIC) study. Baseline risk was then included in a treatment risk model that explored the interaction of baseline risk and treatment status (CEA vs. medical management) on the patient-centered outcome of any stroke or death, including peri-operative events.ResultsThree baseline risk factors (BMI, creatinine and degree of contralateral stenosis) were selected into our baseline risk model (c-statistic 0.59 [95% CI 0.54–0.65]). The model stratified absolute risk between the lowest and highest risk quintiles (5.1% vs. 12.5%). External validation in ARIC found similar predictiveness (c-statistic 0.58 [0.49–0.67]), but poor calibration across the risk spectrum. In the treatment risk model, CEA was superior to medical management across the spectrum of baseline risk and the magnitude of the treatment effect varied widely between the lowest and highest absolute risk quintiles (3.2% vs. 10.7%).ConclusionEven modestly predictive risk stratification tools have the potential to meaningfully influence clinical decision making in asymptomatic carotid disease. However, our ACAS model requires target population recalibration prior to clinical application.

Behavior Revealed in Mobile Phone Usage Predicts Credit Repayment

Abstract Many households in developing countries lack formal financial histories, making it difficult for firms to extend credit, and for potential borrowers to receive it. However, many of these households have mobile phones, which generate rich data about behavior. This article shows that behavioral signatures in mobile phone data predict default, using call records matched to repayment outcomes for credit extended by a South American telecom. On a sample of individuals with (thin) financial histories, this article's method actually outperforms models using credit bureau information, both within-time and when tested on a different time period. But the method also attains similar performance on those without financial histories, who cannot be scored using traditional methods. Individuals in the highest quintile of risk by the measure used in this article are 2.8 times more likely to default than those in the lowest quintile. The method forms the basis for new forms of credit that reach the unbanked.

Influence of Risk on Reduction of Readmission and Death by Disease Management Programs in Heart Failure

Disease management programs (DMPs) may reduce short-term readmission or death after heart failure (HF) hospitalization. We sought to determine if targeting of DMP to the highest-risk patients could improve efficiency. Patients (n = 412) admitted with HF were randomized to usual care or an intensive DMP including optimizing intravascular volume status at discharge, increased self-care education, exercise guidance, closer home surveillance, and increased intensity of HF nurse follow-up. Both treatment groups were similar in demographics, medication use, Charlson comorbidity index, ejection fraction, and left ventricular and atrial volumes. Readmission or death occurred in 74/197 (37%) usual care and 50/215 (23%) DMP patients within 30 days (relative risk [RR] 0.62, 95% confidence interval [CI] 0.46-0.84), and 113/197 (57%) usual care and 78/215 (36%) DMP patients within 90 days, (RR 0.63, 9%% CI 0.51-0.78). The predicted risk of death and readmission (estimated from our previously developed risk score) was similar between treatment groups (mean predicted risk 38.6 ± 22.2% vs 39.4 ± 21.9%; P = .73) and similar across categories of predicted risk between the treatment groups. For 30-day readmission or death, patients from the 2 highest risk quintiles showed a benefit from intervention, and there was an interaction between intervention and predicted risk (P = .02). For 90-day readmission or death, most patients-other than those in the lowest-risk quintile-benefited from the intervention. Use of a risk score may permit targeting of DMP to reduce HF admission. Intensive DMP may reduce short-term readmission or death, particularly in high-risk patients.

Food Insecurity in Older Adults in an Integrated Health Care System.

To estimate food insecurity prevalence and develop a statistical prediction model for food insecurity. Retrospective cohort study. Kaiser Permanente Colorado. Adult members who completed a pre-Medicare Annual Wellness Visit survey. Food insecurity was assessed using a single screening question. Sociodemographic and clinical characteristics from electronic health records and self-reported characteristics from the survey were used to develop the prediction model. Of 130,208 older adult members between January 2012 and December 2015, 50,097 (38.5%) completed food insecurity screening, 2,859 of whom (5.7% of respondents) reported food insecurity. The prevalence of food insecurity was 10.0% or greater among individuals who were black or Hispanic, had less than high school education, had Medicaid insurance, were extremely obese, had poor health status or quality of life, had depression or anxiety, had impairments in specific activities of daily living, had other nutritional risk factors, or were socially isolated (all p<.001). A multivariable model based on these and other characteristics showed moderate discrimination (c-statistic = 0.74) between individuals with food insecurity and those without and 14.3% of individuals in the highest quintile of risk reported food insecurity. Food insecurity is prevalent even in older adults with private-sector healthcare coverage. Specific individual characteristics, and a model based on those characteristics, can identify older adults at higher risk of food insecurity. System-level interventions will be necessary to connect older adults with community-based food resources.