National Health Insurance Sharing Service Research Articles

Discontinuation of antiplatelet therapy after stent-assisted coil embolization for cerebral aneurysms.

Dual antiplatelet therapy (DAPT) is usually temporarily used after stent-assisted coil embolization (SACE), and is commonly converted to mono antiplatelet therapy (MAPT) for indefinitely. In this study, we aimed to find the possibility of discontinuing MAPT, and to determine the proper period of DAPT use. We used the Standard Sample Cohort DB dataset from the National Health Insurance Sharing Service. Among approximately 1 million people in the dataset, SACE was performed in 214 patients whose data this study analyzed. The relationship between discontinuation of antiplatelet therapy and intracranial hemorrhage or cerebral infarction was analyzed using multiple logistic regression, considering all confounding variables. The survival rate according to the continuation of antiplatelet therapy was obtained using Kaplan-Meier analysis, and the difference in survival rate according to the continuation of antiplatelet therapy was verified using the log-rank test. The hazard ratio according to continuation of antiplatelet therapy was obtained using the Cox proportional hazards model. The analysis was conducted by applying the same statistical method to the duration of DAPT use. Among 214 patients who underwent SACE, 50, 159 and five patients continued, discontinued and did not use antiplatelet therapy (except at the time of procedure), respectively. In multiple logistic regression analysis, discontinuation of antiplatelet agents (including aspirin) and the period of DAPT use did not affect the occurrence of intracranial hemorrhage or cerebral infarction, considering various confounding factors. In the survival analysis according to the continuation of antiplatelet agents, patients who continued to use antiplatelet agents had a higher survival rate than those in other groups (p=0.00). The survival rate was higher in the rest of the group than in the group that received DAPT for three months (p=0.00). Continuation of antiplatelet agents or the period of DAPT use did not affect the occurrence of intracranial hemorrhage or cerebral infarction. Considering the survival rate, it would be better to maintain at least three months of antiplatelet therapy and it might be recommended to continue DAPT use for 12 months.

Depression, antidepressant use, and the risk of type 2 diabetes: a nationally representative cohort study.

Previous studies have reported that depression can increase the risk of type 2 diabetes. However, they did not sufficiently consider antidepressants or comorbidity. The National Health Insurance Sharing Service database was used. Among the sample population, 276,048 subjects who had been diagnosed with depression and prescribed antidepressants (DEP with antidepressants group) and 79,119 subjects who had been diagnosed with depression but not prescribed antidepressants (DEP without antidepressants group) were found to be eligible for this study. Healthy controls (HCs) were 1:1 matched with the DEP with antidepressants group for age and sex. We followed up with them for the occurrence of type 2 diabetes. In the group of DEP with antidepressants, although the risk of type 2 diabetes increased compared to HCs in a crude analysis, it decreased when comorbidity was adjusted for. In the group of DEP without antidepressants, the risk of type 2 diabetes decreased both in the crude model and the adjusted models. The risk varied by age group and classes or ingredients of antidepressants, with young adult patients showing an increased risk even in the fully adjusted model. Overall, those with depression had a reduced risk of type 2 diabetes. However, the risk varied according to the age at onset, comorbidity, and type of antidepressants.

How to Use Health Insurance Data Effectively for Healthcare Research

Objectives: The purpose of this study is to present an understanding of the structure and characteristics of health insurance data, which should be basically known in order to use this data more effectively to derive more accurate research results across health and medical care, and an effective use method based on it. Methods: The National Health Insurance Sharing Service and Healthcare Big data open system was reviewed to suggest the structure and characteristics of health insurance data and an efficient use method, and Google Scholar and the Korean Studies Information Service System were used to review the overall application method according to the characteristics of healthcare big data and related literature. Results: The types and characteristics of each research data for public use provided by National Health Insurance Service (NHIS), Health Insurance Review & Assessment Service (HIRA), and the differences between the data of the two institutions were compared and presented. In addition, for the efficient use of health insurance data in healthcare research, understanding the structure and characteristics of health insurance data based on the health insurance system, clear operational definition of analysis variables, adjust bias between comparison groups in case-control study, and the interpretation of analysis results suggest that both statistical and clinical significance should be considered. Conclusions: It is very important to understand the structure and characteristics of health insurance data based health insurance system. And if we consider in more depth ways to efficiently utilize the health insurance data presented in this study, it will be of great help to future researchers and to improve the quality of the research.

Sex-Specific U-Shaped Relationships Between High-Density Lipoprotein Cholesterol Levels and 10-year Major Adverse Cardiovascular Events: A Nationwide Cohort Study of 5.7 Million South Koreans.

BackgroundHigh-density lipoprotein cholesterol (HDL-C) is a well-known predictor of atherosclerotic cardiovascular diseases (ASCVD). We explored the relationships between HDL-C levels and 10-year major adverse cardiovascular events (MACE) and provided sex-specific upper reference limits for HDL-C levels.MethodsBased on the Korean National Health Insurance Sharing Service, we identified 5,703,897 subjects (women, 48%) with age ≥40 years, eligible HDL-C results, and no prior ASCVD in 2009. We investigated the distribution of 10-year MACE according to HDL-C levels in 10 mg/dL (0.26 mmol/L) intervals and in three HDL-C groups (low men <40 mg/dL [1.03 mmol/L], women <50 mg/dL [1.29 mmol/L]; high between low and extremely high levels; and extremely high >90 mg/dL [2.33 mmol/L]).ResultsThere were U-shaped relationships between HDL-C levels and 10-year MACE with later inflection in women than in men (nadir 80-99 mg/dL [2.07-2.56 mmol/L] and 50-59 mg/dL [1.29-1.53 mmol/L], respectively). In men, the extremely high HDL-C group showed significantly higher 10-year MACE than the high group (28.1% vs. 24.6%, P<0.0001). In women, the extremely high group showed the lowest 10-year MACE; if the extremely high starting point was raised to 130 mg/dL, it became similar to that in men and showed higher 10-year MACE than the high group (25.6% vs. 20.1%, P<0.0001).ConclusionsThe 10-year MACE showed U-shaped relationships with HDL-C levels, and extremely high HDL-C level at 90 mg/dL (2.33 mmol/L) in men was corresponding in risk to 130 mg/dL (3.36 mmol/L) in women.

Nationwide population-based incidence of cancer among patients with HIV/AIDS in South Korea

Cancers are the leading cause of death among people living with HIV/AIDS (PLWHA); however, nationwide studies on cancer incidence are limited. We aimed to determine the trends in the incidence rates of AIDS-defining cancers (ADCs) and non-AIDS-defining cancers (NADCs) among Korean PLWHA. Data from the National Health Insurance Sharing Service from 2004 to 2017 were collected. Age- and sex-adjusted standardized incidence ratios (SIRs) for various cancer types relative to the general population were calculated. Of the 11,737 PLWHA followed-up for 65,052 person-years (PYs), 445 (ADCs, 130 and NADCs, 298) developed cancer. The incidence rate of ADCs decreased, whereas that of NADCs remained unchanged. PLWHA were at an increased risk of ADCs (SIR: 12.6, 95% CI: 10.6–15.0), including Kaposi’s sarcoma, non-Hodgkin’s lymphoma, and cervical cancer, and some NADCs, including anal cancer, lung cancer, liver cancer, and oropharyngeal cancer. Of the 396 patients who received antiretroviral therapy (ART), 215 with optimal adherence had lower incidence rates for ADCs and NADCs than those with non-optimal adherence. The 5-year survival rate of PLWHA with NADCs was 57.8%. Close surveillance and routine screening of cancers and improvement in ART adherence are required to improve the clinical outcomes of PLWHA.

Cohort Profile: The Korean Vietnam War Veterans' Health Study Cohort (KOVECO).

During the Vietnam War, many troops and citizen were exposed to large amounts of Agent Orange (AO), and the hazardous effects of AO are continuously being researched and reported. The Korean Vietnam War Veterans’ Health Study Cohort (KOVECO) is a retrospective cohort to demonstrate the health status of the Korean Vietnam War veterans and their second-generation offsprings. The KOVECO is a collaboration of data from the Ministry of Patriots and Veterans Affairs and the National Health Insurance Sharing Service from 2002 to 2018. The study participants were all Korean Vietnam War veterans and their second-generation offsprings, and the references were the general population in which gender and region were matched with the participants. As of 2002, 191,272 Vietnam War veterans (1,000,320 comparisons) and 1,963,402 s-generations (1,173,061 references) were included in the cohort. The KOVECO consists of personal information, medical facility visit information, and general health examination information. The KOVECO could act as a health surveillance system, which would be able to detect long-term health effects caused by exposure to AO and provide a direction for policy making through academic research.

The Relationship between Parkinson's Disease and Acute Myocardial Infarction in Korea : A Nationwide Longitudinal Cohort Study.

ObjectiveThe goal of the following statewide age and gender-coordinated cohort study in Korea is to find out if there is a link between acute myocardial infarction (AMI) and Parkinson’s disease (PD). MethodsUtilizing the National Health Insurance Sharing Service cohort, patient data were collected. Six thousand four hundred seventy-five individuals with PD were distinguished by utilizing the International Classification of Diseases 10 code G20 and have enrolled in the PD group. The number of participants decreased to 5259 after excluding 1039 patients who were hospitalized less than one time or who visited an outpatient clinic less than twice. Then, 26295 individuals were selected as part of the control group after case control matching was conducted through 1 : 5 age- and gender-coordinated matching. The Cox proportional hazard regression analysis and Kaplan-Meier method were utilized to analyze the likelihood of AMI in PD. ResultsAfter controlling for age and gender, the hazard ratio of AMI in the PD group was 3.603 (95% confidence interval [CI], 2.837–4.577). After that, the following hazard ratio of AMI in the PD group was modified against for co-morbid medical disorders, resulting in 3.551 (95% CI, 2.795–4.511). According to a subgroup analysis, in males and females aged <65 and aged ≥65 and in the non-diabetes and diabetes, hypertension and non-hypertension, dyslipidemia and non-dyslipidemia subgroups, the AMI incidence rates were dramatically higher in the PD group compared to that of the control. ConclusionIndividuals with PD have a greater chance of AMI, according to this cross-national study.

Measuring Patient Similarity on Multiple Diseases by Joint Learning via a Convolutional Neural Network.

Patient similarity research is one of the most fundamental tasks in healthcare, helping to make decisions without incurring additional time and costs in clinical practices. Patient similarity can also apply to various medical fields, such as cohort analysis and personalized treatment recommendations. Because of this importance, patient similarity measurement studies are actively being conducted. However, medical data have complex, irregular, and sequential characteristics, making it challenging to measure similarity. Therefore, measuring accurate similarity is a significant problem. Existing similarity measurement studies use supervised learning to calculate the similarity between patients, with similarity measurement studies conducted only on one specific disease. However, it is not realistic to consider only one kind of disease, because other conditions usually accompany it; a study to measure similarity with multiple diseases is needed. This research proposes a convolution neural network-based model that jointly combines feature learning and similarity learning to define similarity in patients with multiple diseases. We used the cohort data from the National Health Insurance Sharing Service of Korea for the experiment. Experimental results verify that the proposed model has outstanding performance when compared to other existing models for measuring multiple-disease patient similarity.

Significant Physical and Exercise-Related Variables for Exercise-Centred Lifestyle: Big Data Analysis for Gynaecological Cancer Patients

This study investigated the characteristics of gynaecological cancers and is aimed at identifying significant risk variables using the National Health Insurance Sharing Service database to develop practical interventions for affected patients. Data regarding patients with uterine and ovarian cancer from the National Health Insurance Sharing Service database were collected and analysed using Student's t-test, logistic regression, and receiver operating characteristic curve analyses. Student's t-test analyses revealed that age, body mass index, blood pressure, and waist variables differed significantly among patients with uterine cancer. Gamma-glutamyl transpeptidase levels were higher in patients with ovarian cancer than in patients with uterine cancer. Physical fitness function tests reflected the status of patients with cancer. Moreover, physical disability was associated with an increased incidence of ovarian cancer. Intensive exercise for 20 min more than 1 time per week must be avoided to prevent uterine cancer. Receiver operating characteristic curve analyses showed that the optimal cutoff value for one-leg standing time, a prognostic and preventive factor in ovarian cancer, was 9.50 s (sensitivity, 94.9%; specificity, 96.9%). Controlling significant variables for each gynaecological cancer type in an individualised and optimised manner is recommended, including by maintenance of an adjusted exercise-centred lifestyle.

Development of a Suicide Prediction Model for the Elderly Using Health Screening Data.

Suicide poses a serious problem globally, especially among the elderly population. To tackle the issue, this study aimed to develop a model for predicting suicide by using machine learning based on the elderly population. To obtain a large sample, the study used the big data health screening cohort provided by the National Health Insurance Sharing Service. By applying a machine learning technique, a predictive model that comprehensively utilized various factors was developed to select the elderly aged > 65 years at risk of suicide. A total of 48,047 subjects were included in the analysis. Individuals who died by suicide were older, and the number of men was significantly greater. The suicide group had a more prominent history of depression, with the use of medicaments significantly higher. Specifically, the prescription of benzodiazepines alone was associated with a high suicide risk. Furthermore, body mass index, waist circumference, total cholesterol, and low-density lipoprotein level were lower in the suicide group. We developed a model for predicting suicide by using machine learning based on the elderly population. This suicide prediction model can satisfy the performance to some extent by employing only the medical service usage behavior without subjective reports.

Epidemiological study of post-traumatic ankle osteoarthritis after ankle sprain in 195,393 individuals over middle age using the National Health Insurance Database: A retrospective design

ObjectivesThis study aimed to determine the risk of ankle OA onset after an incidence of sprain, relative to the risk of onset in healthy population, and to investigate the effect of gender, age, BMI, and exercise on the development of ankle OA after sprain. DesignRetrospective cohort study. MethodsUsing a sample cohort dataset from 2002 to 2013 provided by the Korean National Health Insurance Sharing Service, we calculated the mean survival time and cumulative incidence rate in sprained and healthy populations using Kaplan-Meier analysis. A Cox proportional hazards model was used to analyze the adjusted hazard ratio (HR) of sprain for the development of OA with 95% confidence intervals (CIs). Adjusted HRs of gender, age, BMI, and exercise (yes/no) were analyzed in the sprained group. ResultsAmong the selected population of 195,393 individuals, 40,876 (20.9%) were diagnosed with an ankle sprain, and 1543 (3.85%) of those individuals developed ankle OA. Of the 154,517 healthy individuals (79.1%), 4062 (2.66%) cases had progressed to OA. The sprained group had 46% (HR, 1.46; 95% CI, 1.38–1.55) greater rate of progression to ankle OA than did healthy individuals. In the sprain group, individuals who were female, obese, and overweight had 40% (HR, 1.40; 95% CI, 1.26–1.55), 43% (HR, 1.43; 95% CI, 1.12–1.82), and 22% (HR, 1.22; 95% CI, 1.10–1.35) higher incidence to develop ankle OA, respectively, compared to those who were male, underweight and normal. ConclusionsThis study found that ankle sprain was a significant risk factor for a diagnosis of early OA. Female gender and high BMI increased the incidence of ankle OA after sprain.

Big Data Analysis of the Risk of Intracranial Hemorrhage in Korean Populations Taking Low-Dose Aspirin

ObjectivesAspirin has traditionally been used as an analgesic and anti-inflammatory drug; however, low-dose aspirin is known to increase the risk of gastrointestinal and intracranial hemorrhage. In this study, the risk of intracranial hemorrhage in patients taking low-dose aspirin was assessed. Materials and MethodsWe used the Standard Sample Cohort DB dataset from the National Health Insurance Sharing Service of Korea. This dataset includes details of medical care and prescriptions for patients who used hospital services during a 14-year period throughout Korea. Of approximately 1 million total patients, data from 746,703 adults over the age of 30 years were included for analysis. An Χ2 test was performed to assess the effect of low-dose aspirin on intracranial hemorrhage. In addition, the relationship between use of low-dose aspirin and intracranial hemorrhage was analyzed using multiple logistic regression with consideration of all confounding variables. ResultsThe results revealed no significant positive correlations between the use of low-dose aspirin and intracranial hemorrhage requiring hospitalization. ConclusionsBig data analysis of 746,703 patients in Korea over a period of 14 years showed that serious intracranial hemorrhage requiring hospitalization was unrelated to low-dose aspirin use. Moreover, low-dose aspirin use reduced the risk of intracranial hemorrhage in Korean populations.

Fluoroquinolone and no risk of Achilles-tendinopathy in childhood pneumonia under eight years of age-a nationwide retrospective cohort.

BackgroundThe emergence of macrolide-resistant Mycoplasma pneumoniae pneumonia (MRMP) has made its treatment challenging. A few guidelines have recommended fluoroquinolones (FQs) as second-line drugs of choice for treating MRMP in children under the age of eight, but concerns about potential adverse events (i.e., Achilles tendinopathy; AT) have been raised. The aim of this study was to investigate the relationship between the use of FQs and the risk of AT in pneumonia in children under eight years of age.MethodsChildren hospitalized with pneumonia (total of 2,213,807 episodes) from 2002 to 2017 were enrolled utilizing the Korean National Health Insurance Sharing Service (NHISS) database. The independent risk of FQs for AT was analyzed by a generalized estimating equation with adjustment for age, sex, and underlying diseases.ResultsAmong 2,213,807 episodes of pneumonia hospitalization, children in a total of 6,229 episodes (0.28%) were treated with FQs (levofloxacin 40.9%, ciprofloxacin 36.1%, moxifloxacin 11.6%, and others 11.4%). The FQ-exposure group showed a 0.19% (12/6,229) incidence of AT within 30 days after the first administration of FQ. The use of FQs increased the risk of AT (OR 3.00; 95% CI: 1.71–5.29), but became null after adjusting for age, sex, and underlying diseases (aOR 0.85; 95% CI: 0.48–1.51). All AT related to the use of FQs occurred after the use of ciprofloxacin or levofloxacin, and not in children under eight years of age.ConclusionsAT was a rare adverse event of FQ use for childhood pneumonia, particularly under eight years of age. Clinicians could consider using FQs as a second-line option in the treatment of childhood pneumonia when there are no alternative therapeutic options.

Time-Varying Risk of Atrial Fibrillation in Patient With Medically and Surgically Treated Primary Aldosteronism: A Nationwide Cohort Study

Evidence of increased cardiovascular risk, especially atrial fibrillation, has been accumulating among patients with primary aldosteronism (PA), but there is still limited information about long-term prognosis related to different treatment strategy. The aim of this study was not only to investigate the incidence of atrial fibrillation, but also to evaluate their time-dependent changes after adrenalectomy (surgery group) or mineralocorticoid receptor antagonists (drug group) for patients with PA compared to those with essential hypertension (EH). From a nationwide cohort in Korea (2003–2017), PA were individually matched for sex, age (±10 years), and index year in a 1:5 ratio with EH. The primary end point was the time-varying risk of new-onset atrial fibrillation (NOAF) among PA according to treatment mode compared to EH. The secondary end points were the risks of major adverse cardiovascular events (composite of non-fatal myocardial infarction, non-fatal stroke, and death from cardiovascular causes), hospitalization for heart failure, and all-cause mortality. Cox proportional-hazards analysis or time-dependent Cox analysis based on the Schoenfeld residuals testing were performed. We enrolled 1,418 PA patients (755 in PA surgery group and 663 in PA drug group), and matched theses with 7,090 EH controls with a median of 5 years. The risk of incident NOAF was statistically higher in patients with PA (both surgery and drug groups) within the three years after diagnosis (adjusted hazard ratio, 3.02; p<0.001), whereas there was no statistically significance after the three years compared to EH (adjusted hazard ratio, 0.50; p=0.053). Patients in the PA drug group had higher risk of non-fatal stroke during the total followed up period (adjusted hazard ratio, 1.53, p=0.031), although the PA surgery group didn’t. In contrast, patients with PA had no statistically significant difference in risks for other secondary cardiovascular outcomes. In conclusion, this propensity cohort study of adults with PA demonstrated the changeable risk of NOAF over time possibly due to the residual effect of inappropriate aldosterone levels. These findings can provide clinically relevant guidance in the monitoring the cardiovascular complications, especially NOAF and non-fatal stroke, even after treatment among patients with PA. Acknowledgements: This study was supported by Collaborative Research Project of Korean Endocrine Society and National Health Insurance Sharing Service (NHIS-2019-4-005). We also thank Minheui Yu and Doori Cho, the members of the SENTINEL (Severance ENdocrinology daTa scIeNcE pLatform) team for technical assistance in searching and summarizing the relevant literature (4-2018-1215).

Risk of cancer in patients with recurrent aphthous stomatitis in Korea: A nationwide population-based study.

The relationship between recurrent aphthous stomatitis (RAS), a common mucosal lesion, and cancer has not been demonstrated. This study investigated the risk for developing cancer in patients with RAS, based on data from Korea's National Health Insurance Sharing Service (NHISS). Nationwide population-based cohort data from 2005 to 2009 provided by the NHISS was used. The group diagnosed with RAS for 5 years and an undiagnosed control group were constructed through 1:1 propensity score matching (PSM). The experimental design compared the incidence rate of a cancer diagnosis from 2010 to 2015 between these 2 groups. After identifying 13,808 people that met our inclusion criterion from a 1 million cohort group, 13,808 controls were included in the study through PSM. Among all cancers, pancreatic cancer had an adjusted hazard ratio of 1.26 (95% confidence interval: 1.01–1.57, P < .041). For the rest of the cancers, there was no significant incidence rate. RAS was associated with an increased risk of pancreatic cancer in the analysis using large population-based cohort data. Further long-term follow-up studies are needed.

An Investigation of the Association between Health Screening and Dental Scaling in Korea.

Dental disease is one of the most prevalent chronic diseases worldwide, and its expenditure is continuously increasing. Periodontal disease is increasing as a chronic non-communicable disease in adults and older people. Health screening has been shown to be cost-effective and improves the quality of life through the early detection of diseases. This study aimed to analyze the relationship between national health screening and dental scaling as a preventive service for periodontal disease. The study used sample cohort data from 2002 to 2015 provided by the National Health Insurance Sharing Service in South Korea. A logistic regression analysis of the utilization of dental scaling was performed to identify the independent effects of national health screening. People who underwent health screening showed a higher tendency to undergo dental scaling. Additionally, disparities in utilization according to socioeconomic status were reduced among those who underwent screening. The intervention to extend dental coverage could be more beneficial when combined with health screening, encouraging more people to participate and reducing inequalities in utilization.

Optimal treatment recommendations for diabetes patients using the Markov decision process along with the South Korean electronic health records

The extensive utilization of electronic health records (EHRs) and the growth of enormous open biomedical datasets has readied the area for applications of computational and machine learning techniques to reveal fundamental patterns. This study’s goal is to develop a medical treatment recommendation system using Korean EHRs along with the Markov decision process (MDP). The sharing of EHRs by the National Health Insurance Sharing Service (NHISS) of Korea has made it possible to analyze Koreans’ medical data which include treatments, prescriptions, and medical check-up. After considering the merits and effectiveness of such data, we analyzed patients’ medical information and recommended optimal pharmaceutical prescriptions for diabetes, which is known to be the most burdensome disease for Koreans. We also proposed an MDP-based treatment recommendation system for diabetic patients to help doctors when prescribing diabetes medications. To build the model, we used the 11-year Korean NHISS database. To overcome the challenge of designing an MDP model, we carefully designed the states, actions, reward functions, and transition probability matrices, which were chosen to balance the tradeoffs between reality and the curse of dimensionality issues.

Changes in Antibiotic Prescription After Tooth Extraction: A Population-Based Study from 2002 to 2018

ObjectiveMicrobial resistance due to antibiotic misuse is a worldwide problem. This study aimed to examine the trends in antibiotic prescription after tooth extraction in Korea. MethodsFrom the database of National Health Insurance Sharing Service, patients who underwent tooth extraction from 2002 to 2018 were selected as subjects, and 10% of them were selected via stratified sampling based on sex and age. Overall, 15,838,529 cases were observed for annual antibiotic prescription and broad-spectrum antibiotic prescription patterns. Additionally, standardized annual antibiotic use was calculated using a defined daily dose. ResultsPrescriptions were issued in 13,429,770 (84.8%) of the tooth extraction cases, of which 12,179,185 (90.7%) included antibiotics. Logistic regression analysis revealed that the likelihood of prescribing antibiotics after tooth extraction decreased in 2003 compared to 2002 (odds ratio, 0.95) but increased from 2004 to 2018. In the case of the broad-spectrum antibiotic prescription rate, there was no clear trend between 2002 and 2012 (odds ratio, 0.89-1.07); however, over the last 5 years, the likelihood of broad-spectrum antibiotic prescription has steadily increased. The value of antibiotics based on the defined daily dose of 1000 patients per day was calculated to be 4.39 in 2002, exhibiting a steady increase later and reaching 6.97 in 2018, whereas that of broad-spectrum antibiotics was 1.68 in 2002 and has since been on the rise; the highest was 3.82 in 2018. ConclusionsAntibiotic use after tooth extraction increased over the last 17 years; additionally, the rate of prescribing broad-spectrum antibiotics has increased over the last 5 years.

Predicting the Risk of Chronic Kidney Disease (CKD) Using Machine Learning Algorithm

Background: Creatinine is a type of metabolite of blood that is strongly correlated to glomerular filtration rate (GFR). As measuring GFR is difficult, creatinine value is used for indirectly determining GFR and then the stage of chronic kidney disease (CKD). Adding a creatinine test into routine health examination could detect CKD. As more items for comprehensive examination means higher cost, creatinine testing is not included in the routine health examination in many countries. An algorithm based on common test results, without creatinine test, to evaluate the risk of CKD will increase the chance of its early detection and treatment. Methods: In this study, we used open source data containing 1 million samples. These data contain 23 health-related features, including common diagnostic test results provided by National Health Insurance Sharing Service (NHISS). A low GFR indicates possible chronic kidney disease (CKD). As is commonly accepted in the medical community, a GFR of 60 mL/min is used as the threshold, below which is considered to have CKD. In this study, the first step aims to build a regression model to predict the value of creatinine from 23 features, and then combine the predicted value of creatinine with the original 23 features to evaluate the risk of CKD. We will show by simulation that by the proposed method we can achieve better prediction results compared to direct prediction from 23 features. The data is extremely unbalanced for predicting the target variable creatinine. We used undersampling method and proposed a new cost-sensitive mean-squared error (MSE) loss function to deal with the problem. Regrading model selection, this work used three machine learning models: a bagging tree model named Random Forest, a boosting tree model named XGBoost, and a neural network based model named ResNet. To improve the result of the creatinine predictor, we averaged results from eight predictors, a method known as ensemble learning. Finally, the predicted creatinine and the original 23 features is used to predict the risk of CKD. Results: We optimized results of R-Squared (R2) value to select the appropriate undersampling strategy and the regression model for the regression stage of creatinine prediction. Ensembled model achieved the best performance of R2 of 0.5590. The six factors from 23 are selected from the top of the list of how strongly they affect the creatinine value. They are sex, age, hemoglobin, the level of urine protein, waist circumference, and habit of smoking. Using the predicted value of creatinine, an area under Receiver Operating Characteristic curve (AUC) of 0.76 is achieved while classifying samples for CKD. Conclusions: Using commonly available health parameters, the proposed system can assess the risk of CKD for public health. High-risk subjects can be screened and advised to take a creatinine test for further confirmation. In this way, we can reduce the impact of CKD on public health and facilitate early detection for many, where a blanket test of creatinine is not available for all.

Atrial fibrillation burden and the risks of ischemic stroke and intracranial hemorrhage: comparisons of catheter ablation, medical therapy, and non-atrial fibrillation population

Abstract Background Although atrial fibrillation (AF) burden was known to be related to the risk of ischemic stroke (IS), clinical evidence regarding intracranial hemorrhage (ICH) or comparison of the patients after AF catheter ablation (AFCA) and the non-AF population are limited. Objective We explored the risks of IS and ICH after AFCA or medical therapy (Med) in the AF population and the matched non-AF population. Methods We compared 1,629 with AFCA (Yonsei AF cohort), 3,258 with Med (Korean National Health Insurance Sharing Service [KNHISS] database), and 3,258 non-AF population (KNHISS database) after 1:2:2 propensity-score match in terms of clinical characteristics and medications (57±12 years old, 22.1% female). IS and ICH were determined by ICD code, brain imaging, and hospital admission in Med and non-AF groups. All AFCA patients underwent regular rhythm follow-up, and the medications including antithrombotic therapies were evaluated for 52±23 months. Results 1. Among AFCA group, the annualized IS rate was significantly higher in the patients with sustaining recurrence of AF/atrial tachycardia (AT) after the last ablation procedure than those remaining in sinus rhythm (SR) (0.87% vs. 0.24%, p=0.017; HR 4.87 [1.36–17.49], p=0.015; log rank p=0.003). 2. The annualized ICH rate was 0% in SR group and 0.06% in sustaining recurrent AF/AT group after AFCA (p=0.361, log rank p=0.545). 3. The annualized IS rate was significantly higher in Med group (1.09%) than in AFCA group (0.30%, p&amp;lt;0.001, log rank p&amp;lt;0.001) or non-AF group (0.34%, p&amp;lt;0.001, log rank p&amp;lt;0.001). There was no significant difference in annualized IS rate between AFCA and non-AF groups (p=0.673, log rank p=0.874) 4. The annualized ICH rates were 0.17% in Med group, 0.06% AFCA group (p=0.023, log rank p=0.042 vs. Med; p=0.172, log rank p=0.193 vs. non-AF), and 0.12% in non-AF group (p=0.226, log rank p=0.241 vs. Med), respectively. Conclusion Post-procedural AF burden influence the risk of IS. AFCA significantly reduces the risks of both IS and ICH to the extent of non-AF population compared to Med group. Funding Acknowledgement Type of funding source: None