鈴木 隆弘

OBJECTIVE: We compared the incidences of four opportunistic infections (OI) in patients with rheumatoid arthritis (RA) treated with molecular-targeted drugs from big claims data. MATERIALS AND METHODS: We identified 205 906 patients with RA who were prescribed molecular-targeted drugs 2010-2017 from the National Database of Japan, and calculated the incidence of four OIs (Pneumocystis pneumonia [PCP], tuberculosis [TB], nontuberculous mycobacterial infection [NTM], and herpes zoster [HZ]). RESULTS: The total number of PCP, TB, NTM, and HZ patients with biological disease-modifying antirheumatic drugs (bDMARDs) or tofacitinib treatment history in RA were 765, 1158, 834, and 18 336, respectively. The incidence rates (IRs) of each OI for all bDMARDs were 0.14, 0.14, 0.09, and 2.40 per 100 person-years, respectively; while for tofacitinib they were 0.22, 0.22, 0.07, and 7.00 per 100 person-years. No big difference was observed among bDMARDs. All OIs showed higher incidence in those older than 65 years; but PCP, NTM and HZ showed no difference between those 65-74 years old and those over 75 years old. The median of occurrence was the third, seventh, ninth, and thirteenth month after treatment, respectively. CONCLUSION: We counted real IRs of OIs for the whole nation from big claims data.

PURPOSE: We aimed to develop a reliable identification algorithm combining diagnostic codes with several treatment factors for inpatients with acute ischemic stroke (AIS) to conduct pharmacoepidemiological studies using the administrative database MID-NET® in Japan. METHODS: We validated 11 identification algorithms based on 56 different diagnostic codes (International Classification of Diseases, Tenth Revision; ICD-10) using Diagnosis Procedure Combination (DPC) data combined with information on AIS therapeutic procedures added as "AND" condition or "OR" condition. The target population for this study was 366 randomly selected hospitalized patients with possible cases of AIS, defined as relevant ICD-10 codes and diagnostic imaging and prescription or surgical procedure, in three institutions between April 1, 2015 and March 31, 2017. We determined the positive predictive values (PPVs) of these identification algorithms based on comparisons with a gold standard consisting of chart reviews by experienced specialist physicians. Additionally, the sensitivities of them among 166 patients with the possible cases of AIS at a single institution were evaluated. RESULTS: The PPVs were 0.618 (95% confidence interval [CI]: 0.566-0.667) to 0.909 (95% CI: 0.708-0.989) and progressively increased with adding or limiting information on AIS therapeutic procedures as "AND" condition in the identification algorithms. The PPVs for identification algorithms based on diagnostic codes I63.x were >0.8. However, the sensitivities progressively decreased to a maximum of ~0.2 after adding information on AIS therapeutic procedures as "AND" condition. CONCLUSIONS: The identification algorithms based on the combination of appropriate ICD-10 diagnostic codes in DPC data and other AIS treatment factors may be useful to studies for AIS at a national level using MID-NET®.

OBJECTIVES: The objective of the study was to determine the seasonal changes in the initiation of biological disease-modifying antirheumatic drugs (bDMARDs) and methotrexate (MTX) using big claims data. METHODS: We counted the monthly number of initial administrations of each bDMARD and MTX in patients with rheumatoid arthritis (RA) between April 2010 and March 2017. Data were collected from the National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB). This database covers more than 95% of Japanese citizens. Seasonal changes in the number of initiations were determined. Patient claims were also classified according to drugs, districts, gender and ages. RESULTS: The initiation of bDMARDs and MTX administration varied according to the season in a sine curve shape, with the highest numbers in May to July and the lowest numbers in November to January. The same changing pattern was observed among each bDMARD, district, gender and age groups particularly when the number was on the higher side. CONCLUSION: We noted an apparent seasonal change in the number of bDMARDs initiated, with a peak during spring, suggesting an exacerbation of RA in the spring in Japan. These changes are overlooked in daily practice and are only visible using big data.

OBJECTIVES: To describe the real-world prescription and treatment retention of molecular-targeted drugs for rheumatoid arthritis (RA) in Japan. MATERIALS AND METHODS: 204,416 patients with RA prescribed at least one of the eight molecular-targeted drugs in 7 years from the National Database of Health Insurance Claims and Specific Health Checkups of Japan covering 98.3% of the Japanese population. The retention rate of each drug as well as head-to head comparisons were estimated by Kaplan-Meier method. RESULTS: 121,131 RA patients were prescribed any molecular-targeted drug for the first time, while 36,633 uses of molecular-targeted drug switched from another (switch use). The overall retention rates of molecular-targeted drugs at 12, 36, and 60 months were 0.64, 0.42, and 0.32 for the naïve use and 0.59, 0.40, and 0.31 for the switch use, respectively. Non-tumor necrosis factor (TNF)-inhibitor molecular-targeted drugs, particularly tocilizumab and tofacitinib, had higher retention rates than TNF inhibitors for both naïve and switch uses regardless of the previous drug, and showed higher retention rates in head-to-head comparisons between eight molecular-targeted drugs. CONCLUSIONS: Our data reveal that the real-world drug retention is overall lower than previously reported and higher with non-TNF inhibitors than with TNF inhibitors.

1. 研究目的 本研究では，アンケート調査により音声認識システムに対するニーズの高い診療領域・診療場面を明らかにし，それらの領域・場面での辞書構築を図ることを目的とした．

PURPOSE: To establish a new medical information database network (designated MID-NET® ) to provide real-world data for drug safety assessments in Japan. METHODS: This network was designed and developed by the Ministry of Health, Labour and Welfare and the Pharmaceuticals and Medical Devices Agency in collaboration with 23 hospitals from 10 healthcare organizations across Japan. MID-NET® is a distributed and closed network system that connects all collaborative organizations through a central data center. A wide variety of data are available for analyses, including clinical and administrative information. Several coding standards are used to standardize the data stored in MID-NET® to allow the integration of information originating from different hospitals. A rigorous and consistent quality management system was implemented to ensure that MID-NET® data are of high quality and meet Japanese regulatory standards (good post-marketing study practice and related guidelines). RESULTS: MID-NET® was successfully established as a reliable and valuable medical information database and was officially launched in April 2018. High data quality with almost 100% consistency was confirmed between original data in hospitals and the data stored in MID-NET® . A major advantage is that approximately 260 clinical laboratory test results are available for analysis. CONCLUSIONS: MID-NET® is expected to be a major data source for drug safety assessments in Japan. Experiences and best practices established in MID-NET® may provide a model for the future development of similar database networks.

PURPOSE: To examine the potential role of Medical Information Database Network (MID-NET® ), a newly established Japanese medical information database network, in postmarketing drug safety assessments through the characterization of its advantages and limitations in five pilot studies. METHODS: The pilot studies were designed to address three major objectives in postmarketing drug safety assessments, ie, the examination of actual drug utilization, the impact of safety-related regulatory actions, and drug-associated risks. The five studies were conducted on the following topics: (a) utilization of codeine-containing products and its relationship with respiratory depression, (b) impact of a Dear Healthcare Professional letter on hypocalcemia incidence associated with denosumab (Ranmark® ) use, (c) risk of acute myocardial infarction associated with antidiabetic drug use, (d) risk of glucose metabolism disorders associated with atypical antipsychotic drug use, and (e) prospective monitoring of abnormal laboratory test results during new drug prescriptions. RESULTS: The pilot studies were successfully conducted and demonstrated the value of MID-NET® in postmarketing drug safety assessments. In particular, the ability to utilize laboratory test results as objective clinical indicators is a major advantage of this database. Potential limitations include a relatively small sample size and a lack of patient-level data linkages among hospitals. CONCLUSIONS: MID-NET® was confirmed to be a valuable database for postmarketing drug safety assessments. The use of laboratory test results to define clinical outcomes may allow more objective and accurate analyses to be conducted. These studies furthered our understanding of the characteristics of MID-NET® , including its advantages and limitations.

OBJECTIVES: The objective of this study is to compare the effectiveness of biological disease-modifying antirheumatic drugs (bDMARDs) by analyzing claims data of 13 Japanese national university hospitals. METHODS: We evaluated 4970 cases of rheumatoid arthritis treated with bDMARDs from the Clinical Information Statistical Analysis database, which has collected and integrated 13 Japanese national university hospitals' claims data for 10 years. We surveyed the medications and calculated the retention rates of bDMARDs using the Kaplan-Meier method and differentiated the effectiveness between the two bDMARDs by comparing the retention rates after switching from one drug to another. RESULTS: Of the 4970 cases, 1364 switched bDMARDs at least once. Tocilizumab (TCZ) reported the highest retention rate, whereas abatacept (ABT) revealed a similar rate compared with only naïve cases. The retention rate curves were higher in cases on TCZ that switched from the other bDMARDs than those in the reversed cases. Following TCZ, ABT and etanercept indicated better results than the other bDMARDs. CONCLUSION: We could compare the effectiveness among bDMARDs by differentiating the retention rates from big claims data. TCZ reported higher retention rates in both naïve and switched cases than other bDMARDs.

BACKGROUND: In this study, differences in older patients between drug use as reported in clinical practice and in clinical trials for regulatory approval were examined. METHODS: Electronic medical record (EMR) data such as patient background (age, sex), concomitant drugs, data on laboratory tests, and prescribed doses of drugs from outpatients with rheumatoid arthritis, diabetes, high blood pressure, or depression at Chiba University Hospital were obtained for the period from January 2003 to December 2012. These data were compared with data from relevant clinical trials for regulatory approval in order to examine differences in drug use. RESULTS: There were 5134 eligible patients. The prescribed doses of drugs were lower than the standard approved doses for depression and rheumatoid arthritis but were generally within the approved dose range for type 2 diabetes mellitus and hypertension. When comparing the characteristics of older patients taking tacrolimus, 5.6% to 17.0% of those would not be able to participate in clinical trials because of liver or renal abnormality, and the incidence rates of some adverse drug events (ADEs) differed significantly between clinical practice and clinical trials. CONCLUSIONS: Appropriate doses of drugs for older patients may differ from approved doses in certain diseases. Complex situations such as a lot of polypharmacy, comorbidity, and functional impairment in older patients in clinical practice make it difficult to evaluate safety based on data from clinical trials. In the future, utilization of a database created from the EMR of older patients should be considered for assessment of drug safety in older patients in clinical practice.

Worldwide populations are aging and countries have to prepare for the effects of demographic change in health care. Health information exchange (HIE), which is the process of moving patient information across health care providers electronically, can help overcome health data fragmentation and open opportunities to improve patient care in terms of quality, economy and efficiency. Since Japan and Germany are among the first countries strongly impacted by demographic changes of aging populations, we report on current developments about health information systems carrying out HIE based on case studies in both countries. Four projects that address the improvement of HIE within a defined region have been selected and investigated: the German project of the Lower Saxony Bank of Health and the Japanese projects Chiba ITnet, Nagasaki AjisaiNet and the National Disaster and Backup System of Japan. The project descriptions are based on relevant English publications, on-site visits and interviews with developers and users. The projects are introduced in terms of their basic architecture and implementation, their present status and future objectives. The projects' developments are still in progress and all have to cope with significant challenges before they will be able to provide a fully working trans-institutional health network solution.

By the diffusion of Hospital Information Systems, many medical documents have been computerized. In addition, most of paper documents before computerization have been also scanned and archived as document images. These were usually converted to text data by using document analysis techniques and Optical Character Reader (OCR) and archived for medical document retrieval. However, the resolutions of some documents are not sufficient for character recognition because of storage spaces, scanning regulations and so on. Therefore, we cannot search desired keywords in the documents, as a result, these documents are not still used effectively in medical document retrieval systems. In this study, we discuss a keyword detection and extraction methods for these document images. As the first step of this study, this paper proposes a method to detect and extract desired words from these documents by using weighted dissimilarity and character sequence. Evaluation experiments using actual medical documents are conducted to discuss the effectiveness of the proposed method.

We performed the multi-year project to collect discharge summary from multiple hospitals and made the big text database to build a common document vector space, and developed various applications. We extracted 243,907 discharge summaries from seven hospitals. There was a difference in term structure and number of terms between the hospitals, however the differences by disease were similar. We built the vector space using TF-IDF method. We performed a cross-match analysis of DPC selection among seven hospitals. About 80% cases were correctly matched. The use of model data of other hospitals reduced selection rate to around 10%; however, integrated model data from all hospitals restored the selection rate.

We constructed a simulation model with a geographic information system (GIS) to predict the future shortage of beds in the Tokyo Metropolitan Area. With a grid square method, we calculated patient numbers for every 500 square meters of the Tokyo Metropolitan Area until 2040 and estimated whether those in need could be admitted to hospitals within an hour's drive from their homes. The simulation demonstrates that after 2025 many patients may not be able to find hospitals within this time framework. The situation will be especially serious in the center of Tokyo and along the railway lines, where many senior citizens reside. We can now apply this innovative GIS method in many fields and especially for the precise estimation of future demands for and supply of medical assistance.

We started a multi-year project to collect discharge summaries from multiple hospitals and create a big text database to build a common document vector space, and develop various applications such as the autoselection of the disease. As the first step, we extracted discharge summary from two hospitals. Using a text mining method, we carried out a DPC selection. There was a difference in term structure and number of terms between the discharge summaries from both hospitals. Nevertheless, the selection rate of the disease is resembled closely.

Chiba University Hospital website has a lot of information and helps visitors to find all kinds of information about the hospital, while too much information makes it difficult to find a suitable doctor as a specialist for them. Visitors have to search every department website one by one. This task is very troublesome, especially large hospital. To solve this problem, we developed a specialist search engine. By entering a disease name, the engine will find the specialists for the disease. The engine can control synonyms by using International Statistical Classification of Diseases and Related Health Problems (ICD), Japanese Standard Disease-Code Master and arrange the specialists in order of the experience. © 2012 IEEE.

Recently, Electronic Medical Record (EMR) systems have become popular in Japan, and numerous discharge summaries are being stored electronically, although they have not yet been reutilized. We performed text mining by using the term frequencyinverse document frequency method along with a morphological analysis of the discharge summaries from 3 hospitals (the Chiba University Hospital, St. Luke's International Hospital, and the Saga University Hospital). We found differences in the styles of the summaries between hospitals, while the rates of properly classified Diagnosis Procedure Combination (DPC) codes were almost the same. Beyond the different styles for the discharge summaries, the text mining method was able to obtain appropriate extracts of the proper DPC codes. An improvement was observed by using the integrated model data between the hospitals. It appeared that a large database containing data from many hospitals could improve the precision of text mining.

OBJECTIVES: The prevalence of electronic medical record in Japan varies according to the size of the hospital which is 62.5% in major hospitals, 21.7% in medium, 9.1% in small size hospitals, and 16.5% in clinics. The complete paperless system is very limited, though some major hospitals are aiming at this system. Several regional network systems which connect different platforms of EMRs, have been developing in many districts, while the final picture of a regional network has not been clearly proposed. To develop a whole electronic health record or personal health records system from the regional network data, we have several obstacles to overcome such as standardization, a privacy act, unique national health number. METHODS: Some experimental trials have just been started. The reuse of the accumulated data has also just been initiated. We exploited text mining systems (term frequency-inverse document frequency method) to find similar cases and auto-audit Japanese diagnosis related group (DRG) coding by using discharge summaries. RESULTS: The same or even a more extreme phenomenon of huge data accumulation is occurring in genetic research and confluence of multi-disciplines of informatics is the next step, which has an enormous accumulation of data and discoveries of the relations beyond the dimension of each informatics. CONCLUSIONS: We need another approach to science apart from the conventional method, and data-driven approach with data mining techniques must be brought in for each field. Informaticians have new important roles as coordinators to link up numerous phenomena over dimensions.

Recently, electronic medical record (EMR) systems have become popular in Japan, and number of discharge summaries is stored electronically, though they have not been reutilized yet. We performed text mining with Tf-idf method and morphological analysis in the discharge summaries from three Hospitals (Chiba University Hospital, St. Luke's International Hospital and Saga University Hospital). We showed differences in the styles of summaries, between hospitals, while the rate of properly classified DPC (Diagnosis Procedure Combination) codes were almost the same. Beyond different styles of the discharge summaries, text mining method could obtain proper extracts of proper DPC codes. Improvement was observed by using integrated model data between the hospitals. It seemed that huge database which contains the data of many hospitals can improve the precision of text mining.

Recent investigations suggest that the AKT/glycogen synthase kinase 3 (GSK3) signaling cascade may be associated with the pathophysiology of schizophrenia and methamphetamine (METH) use disorder. One important molecule related to this cascade is beta-arrestin 2 (ARRB2). We therefore conducted a genetic case-control association analysis of the gene for ARRB2 with schizophrenia and METH use disorder in a Japanese population (547 people with schizophrenia, 177 with METH use disorder and 546 controls). A possible association of 'tag single nucleotide polymorphisms (SNPs)' was found in METH use disorder (rs1045280: P(genotype) = 0.0118, P(allele) = 0.00351; rs2036657: P(allele) = 0.0431; rs4790694: P(genotype) = 0.0167, P(allele) = 0.0202), but no association was found with schizophrenia. We also evaluated the gene-gene interactions among ARRB2, AKT1, and GSK3B, which we previously reported for each of these diseases. However, no interaction was seen in our samples. This is the first association analysis of ARRB2, and our results indicate that ARRB2 may play a role in the pathophysiology of METH use disorder.

To analyze the laboratory data by data mining, user-centered universal tools have not been available in medicine. We analyzed 1,565,877 laboratory data of 771 patients with viral hepatitis in order to find the difference of the temporal changes in laboratory test data between Hepatitis B and Hepatitis C by the combination of temporal abstraction and data mining. The data for one patient is temporal for more than 5 years. After pretreatment the data was converted to abstract patterns and then selected into sets of data combination and rules to identify Hepatitis B or C by D2MS and LUPC which were originally produced by ourselves. Not only data pattern, but also temporal relations were considered as a part of the rules. In the course of evaluating the results by domain experts, even though there were not so remarkable hypotheses, visualization tools made it easier for them to understand the relations of the complicated rules.