Deciles Of Risk Research Articles

Predicting 10-Year Risk of Pancreatic Cancer Using a Combined Genetic and Clinical Model.

Pancreatic cancer has the poorest 5-year survival rate of any major solid tumor, but when diagnosed at an early stage, survival rates improve. Population screening is impractical because pancreatic cancer is rare with a lifetime risk of 1.7%, but accurate risk stratification in the general population could enable health care providers to focus early detection strategies to at-risk individuals. Here, we validate a combined risk prediction model that integrates a polygenic risk score and a clinical risk model. Using the UK Biobank, we conducted a prospective cohort study assessing 10-year pancreatic cancer risks based on a polygenic risk score, a clinical risk score, and a combined risk score. We assessed the association, discrimination, calibration, cumulative hazards, and standardized incidence ratios compared to population incidence rates for the risk scores. We also conducted net reclassification analyses. While all of the risk scores discriminated well between affected and unaffected participants, the combined risk score - with a Harrell's C-index of 0.714 (95% confidence interval [CI]= 0.698, 0.730) - discriminated better than both the polygenic risk score (P= .001) and the clinical risk score (P= .02). In terms of calibration, there was no problem with dispersion for the combined risk score (β= 0.952, 95% CI= 0.865-1.039, P= .3) and overall there was a small overestimation of risk (α=-0.089, 95% CI=-0.156 to-0.021, P= .009). Participants in the top decile of 10-year risk were at 1.413 (95% CI= 1.242-1.607) times population risk. The combined risk score was able to identify individuals at substantially increased risk of pancreatic cancer and to whom targeted screening could be useful.

Development and Validation of a Predictive Risk Algorithm for Bleeding in Individuals on Long-term Hemodialysis: An International Prospective Cohort Study (BLEED-HD).

Individuals with kidney disease are at a high risk of bleeding and as such tools that identify those at highest risk may aid mitigation strategies. We set out to develop and validate a prediction equation (BLEED-HD) to identify patients on maintenance hemodialysis at high risk of bleeding. International prospective cohort study (development); retrospective cohort study (validation). Development: 15 countries (Dialysis Outcomes and Practice Patterns Study [DOPPS] phase 2-6 from 2002 to 2018); Validation: Ontario, Canada. Development: 53 147 patients; Validation: 19 318 patients. Hospitalization for a bleeding event. Cox proportional hazards models. Among the DOPPS cohort (mean age, 63.7 years; female, 39.7%), a bleeding event occurred in 2773 patients (5.2%, event rate 32 per 1000 person-years), with a median follow-up of 1.6 (interquartile range [IQR], 0.9-2.1) years. BLEED-HD included 6 variables: age, sex, country, previous gastrointestinal bleeding, prosthetic heart valve, and vitamin K antagonist use. The observed 3-year probability of bleeding by deciles of risk ranged from 2.2% to 10.8%. Model discrimination was low to moderate (c-statistic = 0.65) with excellent calibration (Brier score range = 0.036-0.095). Discrimination and calibration of BLEED-HD were similar in an external validation of 19 318 patients from Ontario, Canada. Compared to existing bleeding scores, BLEED-HD demonstrated better discrimination and calibration (c-statistic: HEMORRHAGE = 0.59, HAS-BLED = 0.59, and ATRIA = 0.57, c-stat difference, net reclassification index [NRI], and integrated discrimination index [IDI] all P value <.0001). Dialysis procedure anticoagulation was not available; validation cohort was considerably older than the development cohort. In patients on maintenance hemodialysis, BLEED-HD is a simple risk equation that may be more applicable than existing risk tools in predicting the risk of bleeding in this high-risk population.

SMART choice (knee) tool: a patient‐focused predictive model to predict improvement in health‐related quality of life after total knee arthroplasty

Current predictive tools for TKA focus on clinicians rather than patients as the intended user. The purpose of this study was to develop a patient-focused model to predict health-related quality of life outcomes at 1-year post-TKA. Patients who underwent primary TKA for osteoarthritis from a tertiary institutional registry after January 2006 were analysed. The primary outcome was improvement after TKA defined by the minimal clinically important difference in utility score at 1-year post-surgery. Potential predictors included demographic information, comorbidities, lifestyle factors, and patient-reported outcome measures. Four models were developed, including both conventional statistics and machine learning (artificial intelligence) methods: logistic regression, classification tree, extreme gradient boosted trees, and random forest models. Models were evaluated using discrimination and calibration metrics. A total of 3755 patients were included in the study. The logistic regression model performed the best with respect to both discrimination (AUC=0.712) and calibration (intercept=-0.083, slope=1.123, Brier score=0.202). Less than 2% (n=52) of the data were missing and therefore removed for complete case analysis. The final model used age (categorical), sex, baseline utility score, and baseline Veterans-RAND 12 responses as predictors. The logistic regression model performed better than machine learning algorithms with respect to AUC and calibration plot. The logistic regression model was well calibrated enough to stratify patients into risk deciles based on their likelihood of improvement after surgery. Further research is required to evaluate the performance of predictive tools through pragmatic clinical trials. Level II, decision analysis.

Predicting Chronic Opioid Use Among Patients With Osteoarthritis Using Electronic Health Record Data.

To estimate the risk of a patient with osteoarthritis (OA) developing chronic opioid use (COU) within 1 year of a new opioid prescription by using electronic health record (EHR) data and predictive models. We used EHR data from 13 health care organizations to identify patients with OA with an opioid prescription between March 1, 2017 and February 28, 2019 and no record of opioid use in the prior 6 months. We evaluated 4 machine learning models to estimate patients' risk of COU (≥3 prescriptions ≥84 days, maximum gap ≤60 days). We also estimated the transportability of models to organizations outside the training set. The cohort consisted of 33,894 patients with OA, of whom 2,925 (8.6%) developed COU within 1 year. All models demonstrated good discrimination, with the best-performing model (random forest) achieving an area under the receiver operating characteristic curve (AUC) of 0.728 (95% CI 0.711-0.745), but the simplest regression model performed nearly as well (AUC 0.717 [95% CI 0.699-0.734]). Predicted risk deciles spanned a range of 2% risk for the 10th percentile to 18% risk for the 90th percentile for well-calibrated models. Models showed highly variable discrimination across organizations (AUC 0.571-0.842). We found that EHR-based predictive models could estimate the risk of future COU among patients with OA to help inform care decisions. Black-box methods did not have significant advantages over more interpretable models. Care should be taken when extending all models into organizations not included in model training because of a high variability in performance across held-out organizations.

Genetics of myopia: Gene finding and beyond

AbstractThe knowledge on the genetic background of refractive error and myopia has expanded dramatically in the past few years. To date, over 500 genetic loci have been identified for refractive error and myopia, and risk variants mostly carry low risk but are highly prevalent in the general population. Several genes for secondary syndromic myopia overlap with those for common myopia. Polygenic risk scores show overrepresentation of high myopia in the higher deciles of risk. However, also patients with low scores can develop myopia and vice versa. Annotated genes have a wide variety of functions, and all retinal layers appear to be sites of expression. The current genetic findings offer a world of new molecules involved in myopiagenesis. As the missing heritability is still large, further genetic advances including large‐scale, in‐depth genetic studies using complementary big data analytics, consideration of gene–environment effects by thorough measurement of environmental exposures, and focus on subgroups with extreme phenotypes and high familial occurrence are needed. Functional characterization of associated variants is simultaneously needed to bridge the knowledge gap between sequence variance and consequence for eye growth.

Public drug insurance, moral hazard and children's use of mental health medication: Latent mental health risk-specific responses to lower out-of-pocket treatment costs.

Studies have shown that reducing out-of-pocket costs can lead to higher medication initiation rates in childhood. Whether the cost of such initiatives is inflated by moral hazard issues remains a question of concern. This paper looks to the implementation of a public drug insurance program in Québec, Canada, to investigate potential low-benefit consumption in children. Using a nationally representative longitudinal sample, we harness machine learning techniques to predict a child's risk of developing a mental health disorder. Using difference-in-differences analyses, we then assess the impact of the drug program on children's mental health medication uptake across the distribution of predicted mental health risk. Beyond showing that eliminating out-of-pocket costs led to a 3 percentage point increase in mental health drug uptake, we show that demand responses are concentrated in the top two deciles of risk for developing mental health disorders. These higher-risk children increase take-up of mental health drugs by 7-8 percentage points. We find even stronger effects for stimulants (8-11 percentage point increases among the highest risk children). Our results suggest that reductions in out-of-pocket costs could achieve better uptake of mental health medications, without inducing substantial low-benefit care among lower-risk children.

Actualización y validación del riesgo cardiovascular de por vida en España: IBERLIFERISK2

IntroductionThe function to estimate lifetime cardiovascular risk –IBERLIFERISK– in Spanish working population, obtained a satisfactory discrimination; however, there was a slight overestimation of the risk in men and an underestimation of the risk in women. ObjectiveTo recalibrate the current lifetime risk equation after extending the follow-up by 3 years. MethodsRetrospective cohort study. 762.058 workers who underwent an occupational health examination between 2004 and 2007 were included. All episodes of temporary sickness and cardiovascular mortality up to December 2017 were evaluated. Regression models were combined to take into account the presence of competing risks in estimating cardiovascular risk in the derivation cohort. Calibration was performed by graphically representing the proportion of expected and observed events at 10 years of follow-up in the validation cohort, stratifying by risk deciles and calculating the Spiegelhalter Z statistic. Discrimination was evaluated graphically using the Receiver Operating Curve (ROC) and calculating Harrell's C index. ResultsThe mean age was 35.48 years (SD 10.56). 71.14% were men. Harrell's C index was 0.78 (95% CI 0.76-0.79) in men and 0.73 (95% CI 0.69-0.77) in women. In general, there was a slight degree of underestimation in women and overestimation in men in the last decile of risk, although the Spiegelhalter Z statistic was not statistically significant in both sexes (p>0.05). ConclusionsThe updated model continues to discriminate satisfactorily, although the model's calibration has not substantially improved with the new update.

Validation of cardiovascular risk prediction in Type 2 diabetes through federated cohorts in Europe

Abstract Background The SCORE2 risk model has been recommended for cardiovascular risk assessment in individuals aged over 40 years without diabetes in 4 defined risk regions of Europe. We aimed to validate a novel SCORE2-DM model in Type 2 diabetes with additional risk factors mainly based on UK datasets, using federated databases from the EUBIROD network. Methods We defined a full operational protocol to implement a standard procedure for validation across contributing sources in Europe. The manual described inclusion/exclusion criteria (aged 40+, diagnosis of T2 diabetes at 30+, no prior CVD), risk factors measured over a baseline interval (1/2013-62015), target/competing events at follow-up (2015-2019). We specified a common data model with 9 steps required to process longitudinal records and deliver summary cohort data (one record per person). All rules were implemented in R and NeuBIRO, an original tool written in Java/Groovy and H2 SQL (https://github.com/eubirodnetwork/neubiro). Results Software was able to produce the following outputs at each source: distribution of risk factors by sex and 5-year age groups; Harrell's C-index and standard error of SCORE2 and SCORE2-DM by sex and 10-year age groups, C-index differences; average 5-year predictive vs observed risk by risk deciles; adjusted cumulative incidence of 5-year competing risk by sex and 5-year age groups. Code was packaged into a stand-alone bundle, with test data and coefficients ofthe SCORE2-DM model. The procedures allowed either creating cohort data autonomously to run the supplied R code, or let NeuBIRO complete all steps foreseen to deliver. Tests have been successfully completed in the derivation data, with results from federated databases expected to contribute to the final external validation of SCORE2-DM by midyear. Conclusions We defined a collaborative method to validate risk prediction models in high risk subgroups using international pooling of cohort data with privacy protection. Key messages • Collaborative methods to validate risk prediction models can enhance access to real world data for researchers across chronic diseases. • Implementation of flexible and reusable source code can increase opportunities to use prediction models for robust clinical decision making in multimorbidity.

Comparison of the 2016 and 2021 European Society of Cardiology and the 2019 American College of Cardiology/American Heart Association guidelines for primary cardiovascular prevention

Abstract Introduction The European Society of Cardiology (ESC) released in 2021 a new cardiovascular risk prediction model, SCORE2. We aimed to: 1) compare the proportion of individuals included in each category of risk according to 2016 and 2021 ESC and 2019 American Heart Association/American College of Cardiology (AHA/ACC) guidelines on cardiovascular prevention; and 2) assess the discriminative and calibration performances of ESC SCORE1, SCORE2, and AHA/ACC Pooled Cohort Equations (PCE) to predict atherosclerotic cardiovascular disease (ASCVD). Methods We used data from the first follow-up of the CoLaus|PsyCoLaus study, a Swiss population-based cohort, including individuals (40–80 years) recruited between 2009–2012 and having a 10-year follow-up. We included participants without lipid-lowering treatment and free from ASCVD at baseline. We computed SCORE1, SCORE2 (SCORE2-OP in people &amp;gt;70 years) and PCE in individuals eligible for score computation according to each guidelines separately. We assessed the performance of the scores based on discrimination and calibration metrics using first incident ASCVD as outcome. Results Among 4,107 included participants (women, 55.7%), 128 (3.1%) experienced an incident ASCVD during a mean follow-up time of 8.1 (±1.9) years. Statins would be recommended or considered in 40.3% (95% of confident interval [CI], 38.3–42.3), 27.3% (25.4–29.1) and 35.5% (33.6–37.5) of women, and in 62.2% (60.0–64.5), 59.3% (57.0–61.5) and 65.4% (63.2–67.6) of men according to ESC 2016, ESC 2021 and AHA/ACC 2019 guidelines, respectively. Scores were computed in 3,456 (women, 58.2%), 3,318 (women, 57.1%) and 3,384 (women, 56.7%) participants in primary prevention according to ESC 2016, ESC 2021 and AHA/ACC 2019 guidelines, respectively. 50% of women and 17.4% of men developing an incident ASCVD were not eligible for lipid-lowering treatment at baseline according to SCORE2 (compared to 27.5% of women and 14.5% of men using SCORE1, and 42.1% of women and 14.9% of men using PCE). SCORE2 and PCE presented comparable discriminative capacities with area under the receiver operating characteristic (AUROC) of 0.776 (95% CI, 0.730–0.822) and 0.775 (0.729–0.821), respectively. SCORE1 presented a lower AUROC (0.717 [95% CI, 0.665–0.769], p-value=0.0001). All scores underestimated risk in subjects in intermediate deciles of risk and overestimated risk in people in high deciles of risk. SCORE2 was better calibrated in high-risk individuals compared to SCORE1 and PCE. Conclusions Based on ESC 2021 guidelines, if fully implemented, less than a third of women would be eligible for a lipid-lowering therapy, which is lower than according to ESC 2016 and AHA/ACC 2019 recommendations. Among women developing an ASCVD in this sample, half of them were not eligible for a lipid-lowering therapy based on ESC 2021 guidelines. Both SCORE2 and PCE presented good predictive capacities and could be interchangeably used in comparable populations. Funding Acknowledgement Type of funding sources: Public grant(s) – National budget only. Main funding source(s): The CoLaus|PsyCoLaus study was and is supported by research grants from GlaxoSmithKline, the Faculty of Biology and Medicine of Lausanne, and the Swiss National Science Foundation (grants 3200B0–105993, 3200B0-118308, 33CSCO-122661, 33CS30-139468, 33CS30-148401 and 33CS30_177535/1).

Risk for Mood, Anxiety, and Psychotic Disorders in Individuals at High and Low Genetic Liability for Bipolar Disorder and Major Depression

The nature of the genetic relationship between major depression and bipolar disorder remains unclear and might be clarified by considering disorders outside of the mood spectrum. To better understand the relationship between genetic liabilities for major depression (MD) and bipolar disorder (BD). A cohort study was conducted with data for individuals born in Sweden to Swedish parents from 1960 to 1990, with follow-up through December 31, 2018. The data included family genetic risk scores for MD and BD and International Classification of Diseases codes for a range of disorders as reported in primary care, specialist, and hospital registries. Data analysis was conducted from April 2022 to July 2022. High and low genetic liability were defined as being in the upper and lower 2 risk deciles. Risk was compared in individuals at high genetic liability to (1) MD only, (2) BD only, and (3) both MD and BD and those at (4) high genetic liability to BD and low genetic liability to MD and (5) high genetic liability to MD and low genetic liability to BD. Risk for nonpsychotic MD and BD, psychotic MD and BD, anxiety disorders, obsessive-compulsive disorder, schizoaffective disorder (SAD), schizophrenia, and other nonaffective psychosis. Data were included for 2 736 950 individuals with a mean (SD) age at follow-up of 43.9 (9.1) years. High genetic liability to only BD increased risk for nonpsychotic BD, psychotic BD, and SAD. High genetic liability to only MD augmented risk for nonpsychotic MD, anxiety disorders, and nonpsychotic BD. High genetic liability to both BD and MD had the strongest association with risk for nonpsychotic BD, anxiety disorders, and nonpsychotic MD. High genetic liability to BD and low genetic liability to MD increased risk for psychotic BD, nonpsychotic BD, and SAD with no increased risk for nonpsychotic MD or anxiety disorders. High genetic liability to MD and low genetic liability to BD increased risk for nonpsychotic MD, nonpsychotic BD, and anxiety disorders with no increased risk for psychotic BD. In this study, hypotheses that BD and MD are either genetically distinct or genetically closely interrelated were not supported. Both BD and MD were associated with a genetic vulnerability to mood disorders, but even that liability was partially selective. However, compared with individuals at high liability to MD, those at elevated genetic liability for BD had a substantially increased risk for psychosis. Compared with individuals at elevated genetic liability to BD, those at high genetic risk for MD had a considerably augmented risk for anxiety disorders. Clarifying genetic relationships between psychiatric syndromes can be substantially aided by the consideration of profiles of risk for a range of disorders.

External validity of four risk scores predicting 30-day mortality after surgery.

Surgical risk prediction tools can facilitate shared decision-making and efficient allocation of perioperative resources. Such tools should be externally validated in target populations before implementation. Predicted risk of 30-day mortality was retrospectively derived for surgical patients at Royal Perth Hospital from 2014 to 2021 using the Surgical Outcome Risk Tool (SORT) and the related NZRISK (n=44 031, 53 395 operations). In a sub-population (n=31 153), the Physiology and Operative Severity Score for the enumeration of Mortality (POSSUM) and the Portsmouth variant of this (P-POSSUM) were matched from the Copeland Risk Adjusted Barometer (C2-Ai, Cambridge, UK). The primary outcome was risk score discrimination of 30-day mortality as evaluated by area-under-receiver operator characteristic curve (AUROC) statistics. Calibration plots and outcomes according to risk decile and time were also explored. All four risk scores showed high discrimination (AUROC) for 30-day mortality (SORT=0.922, NZRISK=0.909, P-POSSUM=0.893; POSSUM=0.881) but consistently over-predicted risk. SORT exhibited the best discrimination and calibration. Thresholds to denote the highest and second-highest deciles of SORT risk (>3.92% and 1.52-3.92%) captured the majority of deaths (76% and 13%, respectively) and hospital-acquired complications. Year-on-year SORT calibration performance drifted towards over-prediction, reflecting a decrease in 30-day mortality over time despite an increase in the surgical population risk. SORT was the best performing risk score in predicting 30-day mortality after surgery. Categorising patients based on SORT into low, medium (80-90th percentile), and high risk (90-100th percentile) might guide future allocation of perioperative resources. No tools were sufficiently calibrated to support shared decision-making based on absolute predictions of risk.

Interaction of background genetic risk, psychotropic medications, and primary angle closure glaucoma in the UK Biobank.

Background/AimsPsychotropic medications have been reported as a risk factor for angle closure disease. However, the interaction between background genetic risk for primary angle closure glaucoma (PACG) and susceptibility to angle closure disease among psychotropic medication users has not been investigated. Here we demonstrate the utility of a genome-wide polygenic risk score (PRS) in identifying and risk-stratifying subjects with PACG and investigate the association between PACG genetic burden and exposure to psychotropic medications on prevalent angle closure.MethodsThis analysis used the UK Biobank dataset, a prospective cohort study of 502,506 UK residents. We constructed a PACG PRS for participants using genome-wide association study summary statistics from a multiethnic meta-analysis using the Lassosum method.ResultsAmong the 441,054 participants, 959 (0.22%) were identified as PACG cases. Individuals with PACG had higher PRS compared to those without PACG (0.24±1.03 SD vs. 0.00±1.00 SD, p<0.001) and PACG prevalence increased with each decile of higher PRS. Among individuals using psychotropic medication, those with PACG had higher average PRS (0.31±1.00 SD vs. 0.00±1.00 SD, p<0.001) and were more likely to have a PRS in upper deciles of polygenic risk (p = 0.04). At each decile of PRS, psychotropic medication use was associated with increased risk of PACG. These effects were more pronounced and significant in higher deciles.ConclusionWe demonstrate the utility of a PRS for identifying individuals at higher risk of PACG. Additionally, we demonstrate an important relationship where the association between psychotropic medications use and PACG diagnosis varies across the polygenic risk spectrum.

Abstract 2256: Validation of an integrated breast cancer risk model with mammographic density in three prospective cohort studies

Abstract Introduction: Incorporation of mammographic density to breast cancer risk models could improve risk stratification to tailor screening and prevention strategies according to risk. However, robust validation of such models in prospective cohort studies is needed to determine their accuracy in identifying women at different risk levels. Methods: We incorporated Breast Imaging and Reporting Data System (BI-RADS) breast density to a literature-based model with questionnaire-based risk factors and a 313-variant polygenic risk score (PRS). The Individualized Coherent Absolute Risk Estimator (iCARE) tool was used to build and validate a 5-year absolute risk model for breast cancer. The model was evaluated for calibration and discrimination in three prospective cohorts of women of European ancestry (1,468 cases, 19,104 controls): US-based Nurses’ Health Study (NHS I and II) and Mayo Mammography Health Study (MMHS); and Sweden-based Karolinska Mammography Project for Risk Prediction of Breast Cancer (KARMA) study. Analyses were done separately for women younger (NHS II, KARMA) and older than 50 years (NHS I, MMHS, KARMA). Improvements in risk stratification were assessed among US non-Hispanic White women aged 50-70 years. Results: For women younger than 50 years, the model with questionnaire-based risk factors, PRS and BI-RADS was well calibrated across risk deciles in NHS II, but overestimated risk at the highest risk decile in KARMA. For women 50 years or older, the model showed good calibration in all studies, with evidence of slight overestimation at the highest risk decile and underestimation at the lowest risk decile. The model with PRS and BI-RADS was well calibrated for women at high-risk in both age groups. Incorporation of BI-RADS to questionnaire-based risk factors and PRS improved risk discrimination: area under the curves (AUC) 67.0% (95% CI: 63.5-70.6%) vs. 65.6% (95% CI: 61.9-69.3%) for models with and without BI-RADS for younger women and 66.1% (95% CI 64.4-67.8%) vs. 65.5% (95% CI: 63.8-67.2%) for older women. The model with BI-RADS identified 18.4% population of non-Hispanic US women 50-70 years old above 3% 5-year risk (used for recommending risk-reducing medication in the US), with 42.4% of future cases expected to occur in this group. Addition of BI-RADS led to the reclassification of ~8.0% of US non-Hispanic White women aged 50-70 years at the high-risk threshold (3% 5-year risk), resulting in identification of 4.2% of additional future cases. Conclusion: Integrating BI-RADS with questionnaire-based risk factors and PRS resulted in improved risk stratification among women of European ancestry, with evidence of slight overestimation of risk for women at elevated risk. Additional validation of the integrated model in diverse populations is needed prior to considering clinical applications. Citation Format: Charlotta V. Mulder, Yon Ho Jee, Xin Yang, Christopher G. Scott, Chi Gao, Amber N. Hurson, Mikael Eriksson, Per Hall, Peter Kraft, Celine M. Vachon, Antonis C. Antoniou, Gretchen Gierach, Montserrat Garcia-Closas, Parichoy Pal Choudhury. Validation of an integrated breast cancer risk model with mammographic density in three prospective cohort studies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 2256.

Out-of-Sample Validity of the PROLOGUE Score to Predict Neurologic Function after Cardiac Arrest

Background: The clinical value of a prognostic score depends on its out-of-sample validity because inaccurate outcome prediction can be not only useless but potentially fatal. We aimed to evaluate the out-of-sample validity of a recently developed and highly accurate Korean prognostic score for predicting neurologic outcome after cardiac arrest in an independent, plausibly related sample of European cardiac arrest survivors. Methods: Analysis of data from a European cardiac arrest center, certified in compliance with the specifications of the German Council for Resuscitation. The study sample included adults with nontraumatic out-of-hospital cardiac arrest admitted between 2013 and 2018. Exposure was the PROgnostication using LOGistic regression model for Unselected adult cardiac arrest patients in the Early stages (PROLOGUE) score, including 12 clinical variables readily available at hospital admission. The outcome was poor 30-day neurologic function, as assessed using the cerebral performance category scale. The risk of a poor outcome was calculated using the PROLOGUE score regression equation. Predicted risk deciles were compared to observed outcome estimates in a complete-case analysis, a best-case analysis, and a multiple-data-imputation analysis using the Markov chain Monte Carlo method. Results: A total of 1051 patients (median 61 years, IQR 50–71; 29% female) were analyzed. A total of 808 patients (77%) were included in the complete-case analysis. The PROLOGUE score overestimated the risk of poor neurologic outcomes in the range of 40% to 100% predicted risk, involving 63% of patients. The model fit did not improve after missing data imputation. Conclusions: In a plausibly related sample of European cardiac arrest survivors, risk prediction by the PROLOGUE score was largely too pessimistic and failed to replicate the high accuracy found in the original study. Using the PROLOGUE score as an example, this study highlights the compelling need for independent validation of a proposed prognostic score to prevent potentially fatal mispredictions.

Development and validation of a prediction model for conversion of outpatient to inpatient surgery

BackgroundUnplanned hospital admission after intended outpatient surgery is an undesirable outcome. We aimed to develop a prediction model that estimates a patient’s risk of conversion from outpatient surgery to inpatient hospitalization. MethodsThis was a retrospective analysis using the American College of Surgeons National Surgical Quality Improvement Program database, 2005 to 2018. Conversion from outpatient to inpatient surgery was defined as having outpatient surgery and >1 day hospital stay. The Surgical Risk Preoperative Assessment System was developed using multiple logistic regression on a training dataset (2005–2016) and compared to a model using the 26 relevant variables in the American College of Surgeons National Surgical Quality Improvement Program. The Surgical Risk Preoperative Assessment System was validated using a testing dataset (2017–2018). Performance statistics and Hosmer-Lemeshow plots were compared. Two high-risk definitions were compared: (1) the maximum Youden index, and (2) the cohort above the tenth decile of risk on the Hosmer-Lemeshow plot. The sensitivities, specificities, positive predictive values, negative predictive values, and accuracies were compared. ResultsIn all, 2,822,379 patients were included; 3.6% of patients unexpectedly converted to inpatient. The 6-variable Surgical Risk Preoperative Assessment System model performed comparably to the 26-variable American College of Surgeons National Surgical Quality Improvement Program model (c-indices = 0.818 vs. 0.823; Brier scores = 0.0308 vs 0.0306, respectively). The Surgical Risk Preoperative Assessment System performed well on internal validation (c-index = 0.818, Brier score = 0.0341). The tenth decile of risk definition had higher specificity, positive predictive values, and accuracy than the maximum Youden index definition, while having lower sensitivity. ConclusionThe Surgical Risk Preoperative Assessment System accurately predicted a patient’s risk of unplanned outpatient-to-inpatient conversion. Patients at higher risk should be considered for inpatient surgery, while lower risk patients could safely undergo operations at ambulatory surgery centers.

Harnessing Whole Genome Polygenic Risk Scores to Stratify Individuals Based on Cardiometabolic Risk Factors and Biomarkers at Age 10 in the Lifecourse-Brief Report.

In this study, we investigated the capability of polygenic risk scores to stratify a cohort of young individuals into risk deciles based on 10 different cardiovascular traits and circulating biomarkers. We first conducted large-scale genome-wide association studies using data on adults (mean age 56.5 years) enrolled in the UK Biobank study (n=393 193 to n=461 460). Traits and biomarkers analyzed were body mass index, systolic blood pressure, diastolic blood pressure, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglycerides, apolipoprotein B, apolipoprotein A-I, C-reactive protein and vitamin D. Findings were then leveraged to build whole genome polygenic risk scores in participants from the Avon Longitudinal Study of Parents and Children (mean age, 9.9 years) which were used to stratify this cohort into deciles in turn and analyzed against their respective traits. For each of the 10 different traits assessed, we found strong evidence of an incremental trend across deciles (all P<0.0001). Large differences were identified when comparing top and bottom deciles; for example, using the apolipoprotein B polygenic risk scores there was a mean difference of 13.2 mg/dL for this established risk factor of coronary heart disease in later life. Although the use of polygenic prediction in a clinical setting may currently be premature, our findings suggest they are becoming increasingly powerful as a means of predicting complex trait variation at an early stage in the lifecourse.

Risk-Adjusting Mortality in the Nationwide Veterans Affairs Healthcare System

BackgroundThe US Veterans Affairs (VA) healthcare system began reporting risk-adjusted mortality for intensive care (ICU) admissions in 2005. However, while the VA’s mortality model has been updated and adapted for risk-adjustment of all inpatient hospitalizations, recent model performance has not been published. We sought to assess the current performance of VA’s 4 standardized mortality models: acute care 30-day mortality (acute care SMR-30); ICU 30-day mortality (ICU SMR-30); acute care in-hospital mortality (acute care SMR); and ICU in-hospital mortality (ICU SMR).MethodsRetrospective cohort study with split derivation and validation samples. Standardized mortality models were fit using derivation data, with coefficients applied to the validation sample. Nationwide VA hospitalizations that met model inclusion criteria during fiscal years 2017–2018(derivation) and 2019 (validation) were included. Model performance was evaluated using c-statistics to assess discrimination and comparison of observed versus predicted deaths to assess calibration.ResultsAmong 1,143,351 hospitalizations eligible for the acute care SMR-30 during 2017–2019, in-hospital mortality was 1.8%, and 30-day mortality was 4.3%. C-statistics for the SMR models in validation data were 0.870 (acute care SMR-30); 0.864 (ICU SMR-30); 0.914 (acute care SMR); and 0.887 (ICU SMR). There were 16,036 deaths (4.29% mortality) in the SMR-30 validation cohort versus 17,458 predicted deaths (4.67%), reflecting 0.38% over-prediction. Across deciles of predicted risk, the absolute difference in observed versus predicted percent mortality was a mean of 0.38%, with a maximum error of 1.81% seen in the highest-risk decile.Conclusions and RelevanceThe VA’s SMR models, which incorporate patient physiology on presentation, are highly predictive and demonstrate good calibration both overall and across risk deciles. The current SMR models perform similarly to the initial ICU SMR model, indicating appropriate adaption and re-calibration.

P-POSSUM and the NELA Score Overpredict Mortality for Laparoscopic Emergency Bowel Surgery: An Analysis of the NELA Database.

Risk stratification has become a key part of the care processes for patients having emergency bowel surgery. This study aimed to determine if operative approach influences risk-model performance, and risk-adjusted mortality rates in the United Kingdom. A prospectively planned analysis was conducted using National Emergency Laparotomy Audit (NELA) data from December 2013 to November 2018. The risk-models investigated were P-POSSUM and the NELA Score, with model performance assessed in terms of discrimination and calibration. Risk-adjusted mortality was assessed using Standardised Mortality Ratios (SMR). Analysis was performed for the total cohort, and cases performed open, laparoscopically and converted to open. Sub-analysis was performed for cases with ≤ 20% predicted mortality. Data were available for 116 396 patients with P-POSSUM predicted mortality, and 46 935 patients with the NELA score. Both models displayed excellent discrimination with little variation between operative approaches (c-statistic: P-POSSUM 0.801-0.836; NELA Score 0.811-0.862). The NELA score was well calibrated across all deciles of risk, but P-POSSUM over-predicted risk beyond 20% mortality. Calibration plots for operative approach demonstrated that both models increasingly over-predicted mortality for laparoscopy, relative to open and converted to open surgery. SMRs calculated using both models consistently demonstrated that risk-adjusted mortality with laparoscopy was a third lower than open surgery. Risk-adjusted mortality for emergency bowel surgery is lower for laparoscopy than open surgery, with P-POSSUM and NELA score both over-predicting mortality for laparoscopy. Operative approach should be considered in the development of future risk-models that rely on operative data.

Interaction of Cigarette Smoking and Polygenic Risk Score on Reduced Lung Function

The risk of airflow limitation and chronic obstructive pulmonary disease (COPD) is influenced by combinations of cigarette smoking and genetic susceptibility, yet it remains unclear whether gene-by-smoking interactions are associated with quantitative measures of lung function. To assess the interaction of cigarette smoking and polygenic risk score in association with reduced lung function. This UK Biobank cohort study included UK citizens of European ancestry aged 40 to 69 years with genetic and spirometry data passing quality control metrics. Data was analyzed from July 2020 to March 2021. PRS of combined forced expiratory volume in 1 second (FEV1) and percent of forced vital capacity exhaled in the first second (FEV1/FVC), self-reported pack-years of smoking, ever- vs never-smoking status, and current- vs former- or never-smoking status. FEV1/FVC was the primary outcome. Models were used to test for interactions with models, including the main effects of PRS, different smoking variables, and their cross-product terms. The association between pack-years of smoking and FEV1/FVC were compared for those in the highest vs lowest decile of estimated genetic risk for low lung function. We included 319 730 individuals, of whom 24 915 (8%) had moderate-to-severe COPD cases, and 44.4% were men. Participants had a mean (SD) age 56.5 of (8.02) years. The PRS and pack-years were significantly associated with lower FEV1/FVC (PRS: β, -0.03; 95% CI, -0.031 to -0.03; pack-years: β, -0.0064; 95% CI, -0.0064 to -0.0063) and the interaction term (β, -0.0028; 95% CI, -0.0029 to -0.0026). A stepwise increment in estimated effect sizes for these interaction terms was observed per 10 pack-years of smoking exposure. The interaction of PRS with 11 to 20, 31 to 40, and more than 50 pack-years categories were β (interaction) -0.0038 (95% CI, -0.0046 to -0.0031); -0.013 (95% CI, -0.014 to -0.012); and -0.017 (95% CI, -0.019 to -0.016), respectively. There was evidence of significant interaction between PRS with ever- or never- smoking status (β, interaction; -0.0064; 95% CI, -0.0068 to -0.0060) and current or not-current smoking (β, interaction; -0.0091; 95% CI, -0.0097 to -0.0084). For any given level of pack-years of smoking exposure, FEV1/FVC was significantly lower for individuals in the tenth decile (ie, highest risk) than the first decile (ie, lowest risk) of genetic risk. For every 20 pack-years of smoking, those in the tenth decile compared with the first decile of genetic risk showed nearly a 2-fold reduction in FEV1/FVC. COPD is characterized by diminished lung function, and our analyses suggest there is substantial interaction between genome-wide PRS and smoking exposures. While smoking was associated with decreased lung function across all genetic risk categories, the associations were strongest in individuals with higher estimated genetic risk.

Abstract 10210: Automating Risk Assessment of Acute Kidney Injury for Patients Undergoing Percutaneous Coronary Intervention

Background: The risk of acute kidney injury (AKI) for percutaneous coronary intervention (PCI) can inform decision-making and mitigation. However, risk assessment is hampered by the burden of manually abstracting and inputting data required by the established model developed on registry data. Accordingly, there is a need to develop AKI models that use automatically extracted electronic health records (EHR) data and evaluate their performance in comparison to the contemporary registry model. Methods: We used EHR data for PCIs performed at Yale New Haven Hospital (YNHH) and the linked local National Cardiovascular Data Registry (NCDR) CathPCI data to develop and compare the EHR models with the NCDR model. AKI was defined as an increase of ≥0.3 mg/dL or 50% in serum creatinine. To maximize interoperability across systems and usability at the point of care, we only considered structured EHR readily extractable before PCI, and implemented simple but universal data preprocessing. Longitudinal data was aggregated into 24h, 7d and 14d prior to PCI. We used Lasso regression and gradient descent boosting (GDB) to train the models and tested their capacity to approximate the NCDR model predictions and predict AKI independently. We evaluated performance via cross-validation by c-statistic, calibration slope, and predictive range defined by event rate difference in the lowest and highest risk deciles. We used paired t tests for model comparison. Results: The cohort included 9,202 PCIs from December 2012 to November 2019 at YNHH for 8,445 patients and had a mean age of 67.0±12.0 years with 27.6% females and 798 (8.7%) AKIs. Evaluated on the same test sets, the EHR models approximating the NCDR model predictions achieved similar performance to that model (all p&gt;0.05). Further, the EHR model employing GDB to independently predict AKI achieved slightly better performance than the NCDR model in c-statistic (0.84 vs 0.83, p=0.02) and predictive range (41.2% vs 39.5%, p=0.13) with similar calibration (slope 0.98 vs 0.99). Conclusion: The automatically extracted EHR data produced models with similar performance to the registry model. These models can automate AKI risk assessment for PCI, alleviating provider burden and increase uptake of risk assessment.