Prediction of pancreatic cancer in patients with new onset hyperglycemia: A modified ENDPAC model.

BACKGROUND/OBJECTIVES: The Enriching New-Onset Diabetes for Pancreatic Cancer (ENDPAC) model relies primarily on fasting glucose values. Health systems have increasingly shifted practice towards use of glycated hemoglobin (HbA1c) measurement. We modified the ENDPAC model using patients with new onset hyperglycemia. METHODS: Four cohorts of patients 50-84 years of age with HbA1c results ≥6.2-6.5 % in 2011-2018 were identified. A combine cohort was formed. A widened eligibility criterion was applied to form additional four individual cohorts and one combined cohort. The primary outcome was the diagnosis of pancreatic cancer within 3 years after the first elevated HbA1c testing. The performance of the modified ENDPAC model was evaluated by AUC, sensitivity, positive predictive value, cases detected, and total number of patients screened. RESULTS: The individual and combined cohorts consisted of 39,001-79,060 and 69,334-92,818 patients, respectively (mean age 63.5-65.0 years). The three-year PC incidence rates were 0.47%-0.54 %. The AUC measures were in the range of 0.75-0.77 for the individual cohorts and 0.75 for the combined cohorts. When the four individual cohorts were combined, more PC cases can be identified (149 by the combined vs. 113-116 by individual cohorts when risk score was 5+). Performance measures were compromised in nonwhites. Asian and Pacific islanders had lower sensitivity compared to other racial and ethnic groups (29 % vs. 50-60 %) when risk score was 5+. CONCLUSIONS: The modified ENDPAC model targets a broader population and thus identifies more high-risk patients for cancer screening. The differential performance needs to be considered when the model is applied to non-white population.

MASTering systemic mastocytosis: Lessons learned from a large patient cohort.

Background: Systemic mastocytosis (SM), a rare condition affecting about 32,000 individuals in the United States, is often misdiagnosed or underdiagnosed owing to its nonspecific symptoms and the need for invasive biopsies. Objective: Our aim was to identify, classify, and characterize the natural history of patients with SM. Methods: In a retrospective cohort study, administrative data from a large managed care organization was used to identify patients with confirmed SM, based on World Health Organization criteria. Demographic data, delay to diagnosis, disease progression, and health care resource utilization were examined. Results: Of 116 patients with confirmed SM, 77% had indolent SM, 2% had smoldering SM, 12% had SM with associated hematologic neoplasm, 9% had aggressive SM, and none had mast cell leukemia. In all, 5 patients were misclassified as having a less advanced SM subtype initially and 3 were completely undiagnosed (missed diagnosis). The average delay to diagnosis of SM was 58.3 plus or minus 73.1 months. In all, 18% of patients progressed from a nonadvanced form of SM (indolent or smoldering SM) to an advanced form of SM (aggressive SM, SM with associated hematologic neoplasm, or mast cell leukemia) over an average of 88.3 plus or minus 82.7 months. Patients with SM had increased health care utilization, including increases in their numbers of hospital admissions, emergency room visits, urgent care visits, and specialty provider visits, after diagnosis versus before. Conclusions: Rare diseases such as SM would benefit from increased understanding and awareness to improve diagnostic accuracy. Prospective studies are needed to better characterize this patient population and determine the type of follow-up needed to recognize advanced forms of SM so that appropriate treatment can be implemented.

Pretransplant cardiac stress testing and transplant wait time in kidney transplantation candidates.

OBJECTIVE: Routine screening for cardiovascular disease before kidney transplantation remains controversial. This study aims to compare cardiac testing rates in patients with end-stage renal disease, referred and not referred for transplantation, and assess the impact of testing on transplant wait times. METHODS: This is a retrospective cohort study of 22 687 end-stage renal disease patients from 2011 to 2022, within an integrated health system. Cardiac testing patterns, and the association between cardiac testing and transplant wait times and post-transplant mortality were evaluated. RESULTS: Of 22 687 patients (median age 66 years, 41.1% female), 6.9% received kidney transplants, and 21.0% underwent evaluation. Compared with dialysis patients, transplant patients had a 5.6 times higher rate of stress nuclear myocardial perfusion imaging with single-photon emission (rate ratio (RR) 5.64, 95% CI 5.37 to 5.92), a 6.5 times higher rate of stress echocardiogram (RR 6.51, 95% CI 5.65 to 7.51) and 16% higher cardiac catheterisation (RR 1.16, 95% CI 1.06 to 1.27). In contrast, revascularisation rates were significantly lower in transplant patients (RR 0.46, 95% CI 0.36 to 0.58). Transplant wait times were longer for patients who underwent stress testing (median 474 days with no testing vs 1053 days with testing) and revascularisation (1796 days for percutaneous intervention and 2164 days for coronary artery bypass surgery). No significant association was observed with 1-year post-transplant mortality (adjusted OR 1.99, 95% CI 0.46 to 8.56). CONCLUSIONS: This study found a higher rate of cardiac testing in dialysis patients evaluated for kidney transplants. Cardiac testing was associated with longer transplant wait time, but no association was observed between testing and post-transplant mortality.

Identifying the Severity of Heart Valve Stenosis and Regurgitation Among a Diverse Population Within an Integrated Health Care System: Natural Language Processing Approach.

BACKGROUND: Valvular heart disease (VHD) is a leading cause of cardiovascular morbidity and mortality that poses a substantial health care and economic burden on health care systems. Administrative diagnostic codes for ascertaining VHD diagnosis are incomplete. OBJECTIVE: This study aimed to develop a natural language processing (NLP) algorithm to identify patients with aortic, mitral, tricuspid, and pulmonic valve stenosis and regurgitation from transthoracic echocardiography (TTE) reports within a large integrated health care system. METHODS: We used reports from echocardiograms performed in the Kaiser Permanente Southern California (KPSC) health care system between January 1, 2011, and December 31, 2022. Related terms/phrases of aortic, mitral, tricuspid, and pulmonic stenosis and regurgitation and their severities were compiled from the literature and enriched with input from clinicians. An NLP algorithm was iteratively developed and fine-trained via multiple rounds of chart review, followed by adjudication. The developed algorithm was applied to 200 annotated echocardiography reports to assess its performance and then the study echocardiography reports. RESULTS: A total of 1,225,270 TTE reports were extracted from KPSC electronic health records during the study period. In these reports, valve lesions identified included 111,300 (9.08%) aortic stenosis, 20,246 (1.65%) mitral stenosis, 397 (0.03%) tricuspid stenosis, 2585 (0.21%) pulmonic stenosis, 345,115 (28.17%) aortic regurgitation, 802,103 (65.46%) mitral regurgitation, 903,965 (73.78%) tricuspid regurgitation, and 286,903 (23.42%) pulmonic regurgitation. Among the valves, 50,507 (4.12%), 22,656 (1.85%), 1685 (0.14%), and 1767 (0.14%) were identified as prosthetic aortic valves, mitral valves, tricuspid valves, and pulmonic valves, respectively. Mild and moderate were the most common severity levels of heart valve stenosis, while trace and mild were the most common severity levels of regurgitation. Males had a higher frequency of aortic stenosis and all 4 valvular regurgitations, while females had more mitral, tricuspid, and pulmonic stenosis. Non-Hispanic Whites had the highest frequency of all 4 valvular stenosis and regurgitations. The distribution of valvular stenosis and regurgitation severity was similar across race/ethnicity groups. Frequencies of aortic stenosis, mitral stenosis, and regurgitation of all 4 heart valves increased with age. In TTE reports with stenosis detected, younger patients were more likely to have mild aortic stenosis, while older patients were more likely to have severe aortic stenosis. However, mitral stenosis was opposite (milder in older patients and more severe in younger patients). In TTE reports with regurgitation detected, younger patients had a higher frequency of severe/very severe aortic regurgitation. In comparison, older patients had higher frequencies of mild aortic regurgitation and severe mitral/tricuspid regurgitation. Validation of the NLP algorithm against the 200 annotated TTE reports showed excellent precision, recall, and F1-scores. CONCLUSIONS: The proposed computerized algorithm could effectively identify heart valve stenosis and regurgitation, as well as the severity of valvular involvement, with significant implications for pharmacoepidemiological studies and outcomes research.

Neighborhood Deprivation and Crime on Asthma Exacerbations and Utilization in Mild Adult Asthma.

BACKGROUND: Social determinants of health have a significant impact on asthma outcomes, and factors such as income level and neighborhood environment have crucial roles. OBJECTIVE: This study aimed to assess the impact of the Neighborhood Deprivation Index (NDI) and Total Crime Index (TCI) on acute asthma exacerbation (AAE) and asthma-related emergency department and urgent care (ED/UC) visits in adults with mild asthma. METHODS: This retrospective cohort study used administrative data from Kaiser Permanente Southern California among 198,873 adult patients with mild asthma between January 1, 2013 and December 31, 2018. We employed robust Poisson regression models, adjusted for age and sex, to investigate the associations of NDI and TCI with AAE and asthma-related ED/UC visits. Data analysis included subgroup assessments by race and ethnicity and body mass index categories to explore potential disparities in asthma outcomes. RESULTS: Among the cohort, 12,906 patients (6.5%) experienced AAE in 1 year, and Black patients had the highest AAE percentage (7.1%). Higher NDI quintiles were associated with increased AAE risk (adjusted risk ratio = 1.11-1.27), with similar trends across body mass index categories and race or ethnicity, except for Black patients. The TCI showed weaker associations with AAE. Regarding ED/UC visits, 5.0% had such visits within 1 year. Higher NDI quintiles were associated with higher ED/UC visit risk (adjusted risk ratio = 1.23-1.75) whereas TCI associations were weaker. CONCLUSION: Addressing socioeconomic disparities, as indicated by NDI, may be crucial in mitigating asthma exacerbations and reducing health care use, highlighting the importance of incorporating social determinants into asthma management strategies even in patients with mild asthma.

Racial and Ethnic Differences in the Prevalence of Severe Aortic Stenosis by Echocardiography.

BACKGROUND: Understanding the burden of aortic stenosis (AS) across diverse racial and ethnic populations is important to ensure equitable resource allocation. This study explored whether severe AS rate varies by race and ethnicity. METHODS: The rates of severe AS, stratified by race and ethnicity, were calculated among 615,038 adults with a transthoracic echocardiogram. Logistic regression analysis was performed to identify factors associated with severe AS. RESULTS: Severe AS rates ranged from 0.08% in adults < 50 years old to 3.8% in those ≥ 90 years old. Compared to non-Hispanic White and Asian American [adjusted odds ratio (aOR) = 0.47, 95% confidence interval (CI): 0.42-0.53] and non-Hispanic Black (aOR = 0.44, 95% CI: 0.39-0.50) patients were less likely to have severe AS, whereas Hispanic patients (aOR = 0.91, 95% CI: 0.87-0.98) had near similar likelihood. Age was the strongest risk factor for severe AS (compared to age < 50 years, aOR = 21.8, 95% CI: 17.8-26.6 for age 80-89 years, and aOR = 43.8, 95% 35.5-54.0 for age ≥ 90 years). Additional factors associated with severe AS included male sex (aOR = 1.38, 95% CI: 1.30-1.46) and diabetes (aOR = 1.23, 95% CI: 1.15-1.31). CONCLUSIONS: Asian American and non-Hispanic Black adults had lower rates of severe AS compared to White and Hispanic patients. The rate of severe AS progressively increases with age in all racial and ethnic groups, with higher rates in men compared with women. With a demographic shift toward an aging and more diverse population, the burden of AS is anticipated to rise. Ensuring adequate allocation of resources to meet the evolving needs of a diverse population remains a shared health care imperative.

Identifying Symptoms Prior to Pancreatic Ductal Adenocarcinoma Diagnosis in Real-World Care Settings: Natural Language Processing Approach.

BACKGROUND: Pancreatic cancer is the third leading cause of cancer deaths in the United States. Pancreatic ductal adenocarcinoma (PDAC) is the most common form of pancreatic cancer, accounting for up to 90% of all cases. Patient-reported symptoms are often the triggers of cancer diagnosis and therefore, understanding the PDAC-associated symptoms and the timing of symptom onset could facilitate early detection of PDAC. OBJECTIVE: This paper aims to develop a natural language processing (NLP) algorithm to capture symptoms associated with PDAC from clinical notes within a large integrated health care system. METHODS: We used unstructured data within 2 years prior to PDAC diagnosis between 2010 and 2019 and among matched patients without PDAC to identify 17 PDAC-related symptoms. Related terms and phrases were first compiled from publicly available resources and then recursively reviewed and enriched with input from clinicians and chart review. A computerized NLP algorithm was iteratively developed and fine-trained via multiple rounds of chart review followed by adjudication. Finally, the developed algorithm was applied to the validation data set to assess performance and to the study implementation notes. RESULTS: A total of 408,147 and 709,789 notes were retrieved from 2611 patients with PDAC and 10,085 matched patients without PDAC, respectively. In descending order, the symptom distribution of the study implementation notes ranged from 4.98% for abdominal or epigastric pain to 0.05% for upper extremity deep vein thrombosis in the PDAC group, and from 1.75% for back pain to 0.01% for pale stool in the non-PDAC group. Validation of the NLP algorithm against adjudicated chart review results of 1000 notes showed that precision ranged from 98.9% (jaundice) to 84% (upper extremity deep vein thrombosis), recall ranged from 98.1% (weight loss) to 82.8% (epigastric bloating), and F1-scores ranged from 0.97 (jaundice) to 0.86 (depression). CONCLUSIONS: The developed and validated NLP algorithm could be used for the early detection of PDAC.

Stomach Cancer Prediction Model (SCoPM): An approach to risk stratification in a diverse U.S. population.

BACKGROUND AND AIMS: Population-based screening for gastric cancer (GC) in low prevalence nations is not recommended. The objective of this study was to develop a risk-prediction model to identify high-risk patients who could potentially benefit from targeted screening in a racial/ethnically diverse regional US population. METHODS: We performed a retrospective cohort study from Kaiser Permanente Southern California from January 2008-June 2018 among individuals age ≥50 years. Patients with prior GC or follow-up <30 days were excluded. Censoring occurred at GC, death, age 85 years, disenrollment, end of 5-year follow-up, or study conclusion. Cross-validated LASSO regression models were developed to identify the strongest of 20 candidate predictors (clinical, demographic, and laboratory parameters). Records from 12 of the medical service areas were used for training/initial validation while records from a separate medical service area were used for testing. RESULTS: 1,844,643 individuals formed the study cohort (1,555,392 training and validation, 289,251 testing). Mean age was 61.9 years with 53.3% female. GC incidence was 2.1 (95% CI 2.0-2.2) cases per 10,000 person-years (pyr). Higher incidence was seen with family history: 4.8/10,000 pyr, history of gastric ulcer: 5.3/10,000 pyr, H. pylori: 3.6/10,000 pyr and anemia: 5.3/10,000 pyr. The final model included age, gender, race/ethnicity, smoking, proton-pump inhibitor, family history of gastric cancer, history of gastric ulcer, H. pylori infection, and baseline hemoglobin. The means and standard deviations (SD) of c-index in validation and testing datasets were 0.75 (SD 0.03) and 0.76 (SD 0.02), respectively. CONCLUSIONS: This prediction model may serve as an aid for pre-endoscopic assessment of GC risk for identification of a high-risk population that could benefit from targeted screening.

Risk Factors for Acute Asthma Exacerbations in Adults With Mild Asthma.

BACKGROUND: Although individuals with mild asthma account for 30% to 40% of acute asthma exacerbations (AAEs), relatively little attention has been paid to risk factors for AAEs in this population. OBJECTIVE: To identify risk factors associated with AAEs in patients with mild asthma. METHODS: This was a retrospective cohort study. We used administrative data from a large managed care organization to identify 199,010 adults aged 18 to 85 years who met study criteria for mild asthma between 2013 and 2018. An asthma-coded qualifying visit (index visit) was identified for each patient. We then used information at the index visit or from the year before the index visit to measure potential risk factors for AAEs in the subsequent year. An AAE was defined as either an asthma-coded hospitalization or emergency department visit, or an asthma-related systemic corticosteroid administration (intramuscular or intravenous) or oral corticosteroid dispensing. Poisson regression models with robust SEs were used to estimate the adjusted risk ratios for future AAEs. RESULTS: In the study cohort, mean age was 44 years and 64% were female; 6.5% had AAEs within 1 year after the index visit. In multivariate models, age, sex, race, ethnicity, smoking status, body mass index, prior acute asthma care, and a variety of comorbidities and other clinical characteristics were significant predictors for future AAE risk. CONCLUSION: Population-based disease management strategies for asthma should be expanded to include people with mild asthma in addition to those with moderate to severe disease.

Early features of pancreatic cancer on magnetic resonance imaging (MRI): a case-control study.

PURPOSE: Magnetic resonance imaging has been recommended as a primary imaging modality among high-risk individuals undergoing screening for pancreatic cancer. We aimed to delineate potential precursor lesions for pancreatic cancer on MR imaging. METHODS: We conducted a case-control study at Kaiser Permanente Southern California (2008-2018) among patients that developed pancreatic cancer who had pre-diagnostic MRI examinations obtained 2-36 months prior to cancer diagnosis (cases) matched 1:2 by age, gender, race/ethnicity, contrast status and year of scan (controls). Patients with history of acute/chronic pancreatitis or prior pancreatic surgery were excluded. Images underwent blind review with assessment of a priori defined series of parenchymal and ductal features. We performed logistic regression to assess the associations between individual factors and pancreatic cancer. We further assessed the interaction among features as well as performed a sensitivity analysis stratifying based on specific time-windows (2-3 months, 4-12 months, 13-36 months prior to cancer diagnosis). RESULTS: We identified 141 cases (37.9% stage I-II, 2.1% III, 31.4% IV, 28.6% unknown) and 292 matched controls. A solid mass was noted in 24 (17%) of the pre-diagnostic MRI scans. Compared to controls, pre-diagnostic images from cancer cases more frequently exhibited the following ductal findings: main duct dilatation (51.4% vs 14.3%, OR [95% CI]: 7.75 [4.19-15.44], focal pancreatic duct stricture with distal (upstream) dilatation (43.6% vs 5.6%, OR 12.71 [6.02-30.89], irregularity (42.1% vs 6.0%, OR 9.73 [4.91-21.43]), focal pancreatic side branch dilation (13.6% vs1.6%, OR 11.57 [3.38-61.32]) as well as parenchymal features: atrophy (57.9% vs 27.4%, OR 46.4 [2.71-8.28], focal area of signal abnormality (39.3% vs 4.8%, OR 15.69 [6.72-44,78]), all p < 0.001). CONCLUSION: In addition to potential missed lesions, we have identified a series of ductal and parenchymal features on MRI that are associated with increased odds of developing pancreatic cancer.

Mortality Outcomes in a Large Population with and without Covert Cerebrovascular Disease.

Covert cerebrovascular disease (CCD) is frequently reported on neuroimaging and associates with increased dementia and stroke risk. We aimed to determine how incidentally-discovered CCD during clinical neuroimaging in a large population associates with mortality. We screened CT and MRI reports of adults aged ≥50 in the Kaiser Permanente Southern California health system who underwent neuroimaging for a non-stroke clinical indication from 2009-2019. Natural language processing identified incidental covert brain infarcts (CBI) and/or white matter hyperintensities (WMH), grading WMH as mild/moderate/severe. Models adjusted for age, sex, ethnicity, multimorbidity, vascular risks, depression, exercise, and imaging modality. Of n=241,028, the mean age was 64.9 (SD=10.4); mean follow-up 4.46 years; 178,554 (74.1%) had CT; 62,474 (25.9%) had MRI; 11,328 (4.7%) had CBI; and 69,927 (29.0%) had WMH. The mortality rate per 1,000 person-years with CBI was 59.0 (95%CI 57.0-61.1); with WMH=46.5 (45.7-47.2); with neither=17.4 (17.1-17.7). In adjusted models, mortality risk associated with CBI was modified by age, e.g. HR 1.34 [1.21-1.48] at age 56.1 years vs HR 1.22 [1.17-1.28] at age 72 years. Mortality associated with WMH was modified by both age and imaging modality e.g., WMH on MRI at age 56.1 HR = 1.26 [1.18-1.35]; WMH on MRI at age 72 HR 1.15 [1.09-1.21]; WMH on CT at age 56.1 HR 1.41 [1.33-1.50]; WMH on CT at age 72 HR 1.28 [1.24-1.32], vs. patients without CBI or without WMH, respectively. Increasing WMH severity associated with higher mortality, e.g. mild WMH on MRI had adjusted HR=1.13 [1.06-1.20] while severe WMH on CT had HR=1.45 [1.33-1.59]. Incidentally-detected CBI and WMH on population-based clinical neuroimaging can predict higher mortality rates. We need treatments and healthcare planning for individuals with CCD.

Association of endoscopic biopsy sampling methods with detection of precursor lesions of gastric cancer.

BACKGROUND AND AIMS: The yield of various endoscopic biopsy sampling methods for detection of precursor lesions of noncardia gastric cancer in a real-world setting remains unclear. Our objective was to evaluate the association of endoscopic biopsy sampling methods with detection of gastric intestinal metaplasia (GIM) and gastric dysplasia (GD). METHODS: We conducted a case-control study of adult patients who underwent EGD with biopsy sampling between 2010 and 2021 in a racially and ethnically diverse U.S. healthcare system. Cases were patients with histopathologic findings of GIM and/or GD. Control subjects were matched 1:1 by age, procedure date, and medical center. We compared the detection of GIM and GD using 4 different biopsy sampling methods: unspecified, specified stomach location, 2+2, and the Sydney protocol. Additionally, we assessed trends in use of sampling methods (Cochrane-Armitage) and identified patient and endoscopist factors associated with their use (logistic regression). RESULTS: We identified 20,938 GIM and 455 GD matched pairs. A greater proportion of GIM cases were detected using 2+2 (31.3% vs 25.3%, P < .0001) and the Sydney protocol (9.1% vs 1.0%, P < .0001) compared with control subjects. Similarly, a greater proportion of GD cases were detected using the Sydney protocol (15.6% vs .4%, P < .0001). We observed an increasing trend in the use of the Sydney protocol during the study period (3.8%-16.1% in cases, P < .0001; 1%-1.1% in control subjects, P = .005). Male and Asian American patients were more likely to undergo 2+2 or the Sydney protocol, whereas female and Hispanic endoscopists were more likely to perform sampling using these protocols. CONCLUSIONS: The application of the Sydney protocol is associated with an increased detection of precursor lesions of gastric cancer in routine clinical practice.

Dexamethasone Suppression Testing in a Contemporary Cohort with Adrenal Incidentalomas in Two U.S. Integrated Healthcare Systems.

Autonomous cortisol secretion (ACS) from an adrenal adenoma can increase the risk for comorbidities and mortality. The dexamethasone suppression test (DST) is the standard method to diagnose ACS. A multi-site, retrospective cohort of adults with diagnosed adrenal tumors was used to understand patient characteristics associated with DST completion and ACS. Time to DST completion was defined using the lab value and result date; follow-up time was from the adrenal adenoma diagnosis to the time of completion or censoring. ACS was defined by a DST > 1.8 µg/dL (50 nmol/L). The Cox proportional hazards regression model assessed associations between DST completion and patient characteristics. In patients completing a DST, a logistic regression model evaluated relationships between elevated ACS and covariates. We included 24,259 adults, with a mean age of 63.1 years, 48.1% obese, and 28.7% with a Charlson comorbidity index ≥ 4. Approximately 7% (n = 1768) completed a DST with a completion rate of 2.36 (95% CI 2.35, 2.37) per 100 person-years. Fully adjusted models reported that male sex and an increased Charlson comorbidity index were associated with a lower likelihood of DST completion. Current or former smoking status and an increased Charlson comorbidity index had higher odds of a DST > 1.8 µg/dL. In conclusion, clinical policies are needed to improve DST completion and the management of adrenal adenomas.

Stratifying future stroke risk with incidentally-discovered white matter disease severity and covert brain infarct site.

BACKGROUND: Covert cerebrovascular disease (CCD) includes white matter disease (WMD) and covert brain infarction (CBI). Incidentally-discovered CCD is associated with increased risk of subsequent symptomatic stroke. However, it is unknown whether the severity of WMD or the location of CBI predicts risk. OBJECTIVES: To examine the association of incidentally-discovered WMD severity and CBI location with risk of subsequent symptomatic stroke. METHOD: This retrospective cohort study includes patients 50 years old in the Kaiser Permanente Southern California health system who received neuroimaging for a non-stroke indication between 2009-2019. Incidental CBI and WMD were identified via natural language processing of the neuroimage report, and WMD severity was classified into grades. RESULTS: 261,960 patients received neuroimaging; 78,555 (30.0%) were identified to have incidental WMD, and 12,857 (4.9%) to have incidental CBI. Increasing WMD severity is associated with increased incidence rate of future stroke. However, the stroke incidence rate in CT-identified WMD is higher at each level of severity compared to rates in MRI-identified WMD. Patients with mild WMD via CT have a stroke incidence rate of 24.9 per 1,000 person-years, similar to that of patients with severe WMD via MRI. Among incidentally-discovered CBI patients with a determined CBI location, 97.9% are subcortical rather than cortical infarcts. CBI confers a similar risk of future stroke, whether cortical or subcortical, or whether MRI- or CT-detected. CONCLUSIONS: Increasing severity of incidental WMD is associated with an increased risk of future symptomatic stroke, dependent on the imaging modality. Subcortical and cortical CBI conferred similar risks.

Research data warehouse: using electronic health records to conduct population-based observational studies.

Background: Electronic health records and many legacy systems contain rich longitudinal data that can be used for research; however, they typically are not readily available. Materials and methods: At Kaiser Permanente Southern California (KPSC), a research data warehouse (RDW) has been developed and maintained since the late 1990s and widely extended in 2006, aggregating and standardizing data collected from internal and a few external sources. This article provides a high-level overview of the RDW and discusses challenges common to data warehouses or repositories for research use. To demonstrate the application of the data, we report the volume, patient characteristics, and age-adjusted prevalence of selected medical conditions and utilization rates of selected medical procedures. Results: A total of 105 million person-years of health plan enrollment was recorded in the RDW between 1981 and 2018, with most healthcare utilization data available since early or middle 1990s. Among active enrollees on December 31, 2018, 15% were ≥65 years of age, 33.9% were non-Hispanic white, 43.3% Hispanic, 11.0% Asian, and 8.4% African American, and 34.4% of children (2-17 years old) and 72.1% of adults (≥18 years old) were overweight or obese. The age-adjusted prevalence of asthma, atrial fibrillation, diabetes mellitus, hypercholesteremia, and hypertension increased between 2001 and 2018. Hospitalization and Emergency Department (ED) visit rates appeared lower, and office visit rates seemed higher at KPSC compared to the reported US averages. Discussion and conclusion: Although the RDW is unique to KPSC, its methodologies and experience may provide useful insights for researchers of other healthcare systems worldwide in the era of big data analysis.

Machine learning versus regression for prediction of sporadic pancreatic cancer.

BACKGROUND/OBJECTIVES: There is currently no widely accepted approach to identify patients at increased risk for sporadic pancreatic cancer (PC). We aimed to compare the performance of two machine-learning models with a regression-based model in predicting pancreatic ductal adenocarcinoma (PDAC), the most common form of PC. METHODS: This retrospective cohort study consisted of patients 50-84 years of age enrolled in either Kaiser Permanente Southern California (KPSC, model training, internal validation) or the Veterans Affairs (VA, external testing) between 2008 and 2017. The performance of random survival forests (RSF) and eXtreme gradient boosting (XGB) models were compared to that of COX proportional hazards regression (COX). Heterogeneity of the three models were assessed. RESULTS: The KPSC and the VA cohorts consisted of 1.8 and 2.7 million patients with 1792 and 4582 incident PDAC cases within 18 months, respectively. Predictors selected into all three models included age, abdominal pain, weight change, and glycated hemoglobin (A1c). Additionally, RSF selected change in alanine transaminase (ALT), whereas the XGB and COX selected the rate of change in ALT. The COX model appeared to have lower AUC (KPSC: 0.737, 95% CI 0.710-0.764; VA: 0.706, 0.699-0.714), compared to those of RSF (KPSC: 0.767, 0.744-0.791; VA: 0.731, 0.724-0.739) and XGB (KPSC: 0.779, 0.755-0.802; VA: 0.742, 0.735-0.750). Among patients with top 5% predicted risk from all three models (N = 29,663), 117 developed PDAC, of which RSF, XGB and COX captured 84 (9 unique), 87 (4 unique), 87 (19 unique) cases, respectively. CONCLUSIONS: The three models complement each other, but each has unique contributions.

Prediction of persistent chronic cough in patients with chronic cough using machine learning.

Introduction: The aim of this study was to develop and validate prediction models for risk of persistent chronic cough (PCC) in patients with chronic cough (CC). This was a retrospective cohort study. Methods: Two retrospective cohorts of patients 18-85 years of age were identified for years 2011-2016: a specialist cohort which included CC patients diagnosed by specialists, and an event cohort which comprised CC patients identified by at least three cough events. A cough event could be a cough diagnosis, dispensing of cough medication or any indication of cough in clinical notes. Model training and validation were conducted using two machine-learning approaches and 400+ features. Sensitivity analyses were also conducted. PCC was defined as a CC diagnosis or any two (specialist cohort) or three (event cohort) cough events in year 2 and again in year 3 after the index date. Results: 8581 and 52 010 patients met the eligibility criteria for the specialist and event cohorts (mean age 60.0 and 55.5 years), respectively. 38.2% and 12.4% of patients in the specialist and event cohorts, respectively, developed PCC. The utilisation-based models were mainly based on baseline healthcare utilisations associated with CC or respiratory diseases, while the diagnosis-based models incorporated traditional parameters including age, asthma, pulmonary fibrosis, obstructive pulmonary disease, gastro-oesophageal reflux, hypertension and bronchiectasis. All final models were parsimonious (five to seven predictors) and moderately accurate (area under the curve: 0.74-0.76 for utilisation-based models and 0.71 for diagnosis-based models). Conclusions: The application of our risk prediction models may be used to identify high-risk PCC patients at any stage of the clinical testing/evaluation to facilitate decision making.

An Analysis of the Clinical Medication Rules of Traditional Chinese Medicine for Polycystic Ovary Syndrome Based on Data Mining.

Objective: The aim of the present study is to investigate the rules and characteristics of the clinical administration of traditional Chinese medicine (TCM) in the treatment of polycystic ovary syndrome (PCOS) using data mining methods. Method: Medical cases of well-known contemporary TCM doctors treating PCOS were collected from the China National Knowledge Infrastructure, Chinese Biomedical Literature Service System, Wanfang, Chinese Scientific Journals Database, and PubMed; the data were then characterized, and a standardized database of medical cases was built. This database was used to (1) count the frequency of syndrome types and the herbs used in medical cases by data mining methods and (2) analyze drug association rules and systematic clustering methods. Results: A total of 330 papers were included, involving 382 patients and a total of 1,427 consultations. The most common syndrome type was kidney deficiency; sputum stasis was the core pathological product and causative factor. A total of 364 herbs were used. Among them, 22 herbs were used >300 times, including Danggui (Angelicae Sinensis Radix), Tusizi (Semen Cuscutae), Fuling (Poria), Xiangfu (Nutgrass Galingale Rhizome), and Baizhu (Atractylodis Macrocephalae Rhizoma). Additionally, 22 binomial associations were obtained from the analysis of association rules; five clustering formulae were obtained via the analysis of high-frequency drug clusters; and 27 core combinations were obtained by k-means clustering of formula. Conclusion: In the treatment of PCOS, TCM is primarily employed as a combination approach involving tonifying the kidneys, strengthening the spleen, eliminating damp and dissolving phlegm, activating blood circulation, and resolving blood stasis. The core prescription is primarily a compound intervention based on the Cangfu Daotan pill, Liuwei Dihuang pill, and Taohong Siwu decoction.

Stratifying Future Stroke Risk with Incidentally Discovered White Matter Disease Severity and Covert Brain Infarct Site.

BACKGROUND: Covert cerebrovascular disease (CCD) includes white matter disease (WMD) and covert brain infarction (CBI). Incidentally discovered CCD is associated with increased risk of subsequent symptomatic stroke. However, it is unknown whether the severity of WMD or the location of CBI predicts risk. OBJECTIVES: The aim of this study was to examine the association of incidentally discovered WMD severity and CBI location with risk of subsequent symptomatic stroke. METHOD: This retrospective cohort study includes patients aged ≥50 years old in the Kaiser Permanente Southern California health system who received neuroimaging for a nonstroke indication between 2009 and 2019. Incidental CBI and WMD were identified via natural language processing of the neuroimage report, and WMD severity was classified into grades. RESULTS: A total of 261,960 patients received neuroimaging; 78,555 patients (30.0%) were identified to have incidental WMD and 12,857 patients (4.9%) to have incidental CBI. Increasing WMD severity is associated with an increased incidence rate of future stroke. However, the stroke incidence rate in CT-identified WMD is higher at each level of severity compared to rates in MRI-identified WMD. Patients with mild WMD via CT have a stroke incidence rate of 24.9 per 1,000 person-years, similar to that of patients with severe WMD via MRI. Among incidentally discovered CBI patients with a determined CBI location, 97.9% are subcortical rather than cortical infarcts. CBI confers a similar risk of future stroke, whether cortical or subcortical or whether MRI- or CT-detected. CONCLUSIONS: Increasing severity of incidental WMD is associated with an increased risk of future symptomatic stroke, dependent on the imaging modality. Subcortical and cortical CBI conferred similar risks.

Risk Prediction of Pancreatic Cancer in Patients With Recent-onset Hyperglycemia: A Machine-learning Approach.

BACKGROUND: New-onset diabetes (NOD) has been suggested as an early indicator of pancreatic cancer. However, the definition of NOD by the American Diabetes Association requires 2 simultaneous or consecutive elevated glycemic measures. We aimed to apply a machine-learning approach using electronic health records to predict the risk in patients with recent-onset hyperglycemia. MATERIALS AND METHODS: In this retrospective cohort study, health plan enrollees 50 to 84 years of age who had an elevated (6.5%+) glycated hemoglobin (HbA1c) tested in January 2010 to September 2018 with recent-onset hyperglycemia were identified. A total of 102 potential predictors were extracted. Ten imputation datasets were generated to handle missing data. The random survival forests approach was used to develop and validate risk models. Performance was evaluated by c -index, calibration plot, sensitivity, specificity, and positive predictive value. RESULTS: The cohort consisted of 109,266 patients (mean age: 63.6 y). The 3-year incidence rate was 1.4 (95% confidence interval: 1.3-1.6)/1000 person-years of follow-up. The 3 models containing age, weight change in 1 year, HbA1c, and 1 of the 3 variables (HbA1c change in 1 y, HbA1c in the prior 6 mo, or HbA1c in the prior 18 mo) appeared most often out of the 50 training samples. The c -indexes were in the range of 0.81 to 0.82. The sensitivity, specificity, and positive predictive value in patients who had the top 20% of the predicted risks were 56% to 60%, 80%, and 2.5% to 2.6%, respectively. CONCLUSION: Targeting evaluation at the point of recent hyperglycemia based on elevated HbA1c could offer an opportunity to identify pancreatic cancer early and possibly impact survival in cancer patients.