[What can and cannot be achieved by registries : Perspective of the registry working group of the German Network of Health Services Research]. / Was Register leisten können und was nicht : Sicht der AG Register des Deutschen Netzwerks Versorgungsforschung e. V. (DNVF).

In addition to clinical trials, registries and cohort studies are the fundamental basis of patient-orientated research. The importance of registries is increasing because more questions involving patient care under routine conditions (real world data) need to be answered. This article supplies answers to the questions: what can be achieved with registries and what are the limitations? Starting with a consensus definition of a registry from the German Network of Health Services Research (DNVF), the question of existing registries was examined and it was concluded that there was a lack of transparency. Consequently, a registry of registries similar to clinical trials registries is urgently needed as well as an evaluation of the quality of existing registries. Criteria are deduced that allow an assessment of the quality of a registry and which comprehensive possibilities registries can provide are discussed in eight different areas of interest to clinicians. The limitations of registries compared to randomized clinical trials and cohort studies are emphasized and discussed in this article. In the future, the use of registry-based randomized clinical trials (RRCT) will allow data related to efficacy as well as to effectiveness to be collated.

Evaluation of a Binary Semi-supervised Classification Technique for Probabilistic Record Linkage.

BACKGROUND: The process of merging data of different data sources is referred to as record linkage. A medical environment with increased preconditions on privacy protection demands the transformation of clear-text attributes like first name or date of birth into one-way encrypted pseudonyms. When performing an automated or privacy preserving record linkage there might be the need of a binary classification deciding whether two records should be classified as the same entity. The classification is the final of the four main phases of the record linkage process: Preprocessing, indexing, matching and classification. The choice of binary classification techniques in dependence of project specifications in particular data quality has not extensively been studied yet. OBJECTIVES: The aim of this work is the introduction and evaluation of an automatable semi-supervised binary classification system applied within the field of record linkage capable of competing or even surpassing advanced automated techniques of the domain of unsupervised classification. METHODS: This work describes the rationale leading to the model and the final implementation of an automatable semi-supervised binary classification system and the comparison of its classification performance to an advanced active learning approach out of the domain of unsupervised learning. The performance of both systems has been measured on a broad variety of artificial test sets (n = 400), based on real patient data, with distinct and unique characteristics. RESULTS: While the classification performance for both methods measured as F-measure was relatively close on test sets with maximum defined data quality, 0.996 for semi-supervised classification, 0.993 for unsupervised classification, it incrementally diverged for test sets of worse data quality dropping to 0.964 for semi-supervised classification and 0.803 for unsupervised classification. CONCLUSIONS: Aside from supplying a viable model for semi-supervised classification for automated probabilistic record linkage, the tests conducted on a large amount of test sets suggest that semi-supervised techniques might generally be capable of outperforming unsupervised techniques especially on data with lower levels of data quality.

[Robustness of Hospital Benchmarking with the Hospital Standardized Mortality Ratio (HSMR): An Analysis of Secondary Data from 37 German Hospitals]. / Robustheit eines Krankenhausvergleichs mit der Hospital Standardized Mortality Ratio (HSMR): eine Sekundärdatenanalyse von 37 deutschen Krankenhäusern.

Introduction: Internationally, the hospital standardized mortality ratio is increasingly used as a risk-adjusted simple measure for quality control. Goodness of fit of different risk models in Germany and robustness of hospital comparisons were evaluated with a secondary data analysis. Methods: Anonymized routine data from the year 2012 of 37 hospitals of the association Quality Indicators for Ecclesiastical Hospitals were used. 2 independent risk models and the observed mortality were compared, the risk models considered both the original and the adapted forms. Results: The risk models showed an area under curve between 0.906 [95% CI 0.904-0.908] and 0.920 [0.918-0.922]. There was a significant correlation between the risk models and the observed mortality with a correlation coefficient between 0.388 (p<0.05) and 0.936 (p<0.01). 26 hospitals had an identical assessment in all risk models comparing their HSMR with the group. 2 hospitals achieved a positive and a negative assessment taking into account the observed mortality. Conclusion: The quality of the risk models is high and the hospital comparison with the HSMR remained stable. However, it is unclear whether the differences are caused by quality-related issues or by different structures and case-mix. Therefore, the HSMR is primarily intended for quality management purposes within German hospitals.

[Trend of quality of hospital care in German hospitals between 2008 and 2011: a study from national quality assurance data]. / Entwicklung der Qualität der stationären Versorgung zwischen 2008 und 2011: Eine Studie aus Daten der externen Qualitätssicherung.

BACKGROUND AND OBJECTIVES: According to § 137 social code book (SGB) V, data from the national quality assurance are published annually. Objective of this study was to analyse the long-term development of the quality of hospital care. METHODS: On the basis of national data and results of another study covering the years 2004-2008, quality indicators with comparable results of at least 3 subsequent years were included. 292 quality indicators of 26 service groups and 14 medical fields were included. We identified a trend with linear regression and continuous change. RESULTS: 95 % of the performance measures showed either an improvement of quality (105 measures, 36 %) or a stable condition (173 measures, 59 %). Five percent of the quality indicators (14 measures) showed a deterioration. Eleven of the 14 measures assessed quality in coronary angiography and percutaneous coronary intervention. CONCLUSION: Evaluated with performance measures from the national quality assurance, the quality of hospital care improved from 2008-2011, as already from 2004-2008. The deterioration of quality in 5 % of measures shows the need for action in particular situations of care.

[A registry of registries and cohorts: recommendations for metadata and policies]. / Ein Register für Register und Kohorten: Empfehlungen zu Metadaten und Verfahrensregeln.

Registries and cohort studies play a central role in patient-oriented medical research, in particular in health services research. In order to increase the transparency about ongoing registries and cohort studies in Germany, and to promote communication and cooperation between the drivers in the field a so-called register portal should be established. Metadata are characteristics that are used to describe registries and cohort studies in the register portal. A limited set of characteristics, the core set, should correctly describe the projects on the one hand while reducing workload for data capture and data administration on the other hand. The core set consists of 26 data elements that had been defined in a Delphi-consensus process involving experts from the working group registries of the German Network for Health Services Research (DNVF) and the working group IT infrastructure and quality management of the Technology, Methods, and Infrastructure for Networked Medical Research (TMF). Transparent policies are required to guarantee traceability and reliability of the portal's services. Six so-called top-level-tasks and 37 use cases were defined in an interim report so far. The metadata have been agreed upon by most of the member associations of the DNVF. Metadata and rules of procedures are the starting point for the practical implementation of the register portal in the next future.

[Quality of hospital care in Germany: analysis of the trend between 2004 and 2008 from external quality assurance data]. / Qualität der stationären Versorgung in Deutschland. Eine Analyse der Entwicklung zwischen 2004 und 2008 aus Daten der externen vergleichenden Qualitätssicherung.

BACKGROUND AND OBJECTIVES: Introducing fixed fees for inpatient hospital care has raised fears of a negative influence on treatment quality. To address these concerns, an external quality assurance was implemented. The results of the quality indicators should indicate changes in the quality of inpatient care between 2004 and 2008. METHODS: On the basis of the evaluation data delivered by the Federal Quality Assurance Agency, quality indicators with similar results of at least 3 subsequent years were included. A trend was identified with linear regression as well as with a constant change in the results. 204 quality indicators of 19 health care sectors of 6 diagnosis groups were included. RESULTS: Half of the quality indicators showed a particularly striking quality improvement in the categories "indication" and "process". The lowest number of quality indicators with a quality improvement occurred in the cardiac surgery. To this diagnosis group also belong 13 of the 18 quality indicators with deterioration. Sixteen of the 18 quality indicators rank among the category "result". CONCLUSION: The quality of inpatient care improved substantially from 2004 through 2008, measured against the quality indicators of the external comparable quality assurance. This corresponds to other analyses of a similar time period in Germany of the same or different data sources. Those analyses showed additionally that the subjective perceptions of clinicians and of nursing staff are significantly worse.

Development of national competency-based learning objectives "Medical Informatics" for undergraduate medical education.

OBJECTIVES: The aim of this project is to develop a catalogue of competency-based learning objectives "Medical Informatics" for undergraduate medical education (abbreviated NKLM-MI in German). METHODS: The development followed a multi-level annotation and consensus process. For each learning objective a reason why a physician needs this competence was required. In addition, each objective was categorized according to the competence context (A = covered by medical informatics, B = core subject of medical informatics, C = optional subject of medical informatics), the competence level (1 = referenced knowledge, 2 = applied knowledge, 3 = routine knowledge) and a CanMEDS competence role (medical expert, communicator, collaborator, manager, health advocate, professional, scholar). RESULTS: Overall 42 objectives in seven areas (medical documentation and information processing, medical classifications and terminologies, information systems in healthcare, health telematics and telemedicine, data protection and security, access to medical knowledge and medical signal-/image processing) were identified, defined and consented. CONCLUSION: With the NKLM-MI the competences in the field of medical informatics vital to a first year resident physician are identified, defined and operationalized. These competencies are consistent with the recommendations of the International Medical Informatics Association (IMIA). The NKLM-MI will be submitted to the National Competence-Based Learning Objectives for Undergraduate Medical Education. The next step is implementation of these objectives by the faculties.

Visualization of medical data based on EHR standards.

BACKGROUND: To organize an efficient interaction between a doctor and an EHR the data has to be presented in the most convenient way. Medical data presentation methods and models must be flexible in order to cover the needs of the users with different backgrounds and requirements. Most visualization methods are doctor oriented, however, there are indications that the involvement of patients can optimize healthcare. OBJECTIVES: The research aims at specifying the state of the art of medical data visualization. The paper analyzes a number of projects and defines requirements for a generic ISO 13606 based data visualization method. In order to do so it starts with a systematic search for studies on EHR user interfaces. METHODS: In order to identify best practices visualization methods were evaluated according to the following criteria: limits of application, customizability, re-usability. The visualization methods were compared by using specified criteria. RESULTS: The review showed that the analyzed projects can contribute knowledge to the development of a generic visualization method. However, none of them proposed a model that meets all the necessary criteria for a re-usable standard based visualization method. The shortcomings were mostly related to the structure of current medical concept specifications. CONCLUSION: The analysis showed that medical data visualization methods use hardcoded GUI, which gives little flexibility. So medical data visualization has to turn from a hardcoded user interface to generic methods. This requires a great effort because current standards are not suitable for organizing the management of visualization data. This contradiction between a generic method and a flexible and user-friendly data layout has to be overcome.

[Neurological complex treatment: impact on mortality in the first year and costs for statutory health insurance]. / Neurologische Komplexbehandlung - Auswirkungen auf die Sterblichkeit im ersten Jahr und die Kosten der gesetzlichen Krankenkassen.

BACKGROUND: The purpose of this study was to compare standard treatment of strokes with a complex form, defined by the German procedure classification. PATIENTS AND METHODS: 3827 patients in 2005 were compared with 4315 patients in 2007 concerning mortality within hospital and 30, 90 and 360 days after hospital admission. Furthermore, the costs for the statutory health insurances were linked to the mortality rates. 2005 was set as proxy for the standard treatment and 2007 as proxy for the neurological complex treatment. RESULTS: There were significant differences in the 1-year mortality with a 3.11 percentage point reduction (p < 0.0373). For the other periods, there were no significant deviations. The cost for the statutory health insurance per case rose with an average of 0.4809 cost weights. CONCLUSION: There are hints of an improvement in the long-term mortality after the introduction of complex neurological treatment. For the health insurance significant additional costs ensue, which are in an international published range. On the other hand, savings by improving the morbidity must be taken into account.

[Hospital admission due to adverse drug events (ADE): an analysis of German routine hospital data of 2006]. / Stationäre Aufnahmen wegen unerwünschter Arzneimittelereignisse (UAE): Analyse der DRG-Statistik 2006.

BACKGROUND: In developed countries 1-5% of all hospital admissions are due to adverse drug events (ADE). An ADE is defined as an injury resulting from medical intervention related to a drug. The established reporting systems and study designs only capture selective data. The objective of the current analysis was to evaluate the rate, distribution and correlations of ADE related admissions by using German routine data. METHODS: ADEs were identified by an array of 502 specified codes of the ICD-10-GM. The evaluation included only verified codes and was carried out by remote queries of the German DRG-Statistics 2006. Hospital admission due to an ADE was identified via the primary diagnosis. RESULTS: Of all hospital admissions 0.92% were revealed to be certainly caused by an adverse drug event. The average age between affected and non-affected was nearly identical for women 53.48 vs. 53.67 years, for men it was reduced by 4 years (48.38 years). The average hospital stay was lower for cases with an ADE, being reduced by 1.3 days for women (6.26 days vs. 7.55 days) and 1.5 days for men (5.91 days vs. 7.42 days). While mortality with an odds ratio (OR) of 0.59 (95% CI 0.57-0.62) was lower in ADE cases, the rate of emergency admissions due to ADE was increased, the OR being 3.10 (95% CI 3.07-3.13). The wards with excess rates of ADE cases were internal medicine, paediatrics, dermatology, intensive care and neurology. CONCLUSIONS: Younger age, reduced hospital stay and lower mortality of ADE cases are contrary to findings in the relevant literature. The DRG-Statistics also comprise populations which often are excluded in established study designs, in particular, children and cases due to medication errors, overdose, poisoning and allergic reactions. As these cases respond easily to prevention and are of significant interest to pharmacovigilance, the use of routine data is valuable for more intense research of ADE.

[Memorandum registry for health services research]. / Memorandum Register für die Versorgungsforschung.

On August 30, 2010, the German Network for Health Services Research [Deutsches Netzwerk Versorgungsforschung e. V. (DNVF e. V.)] approved the Memorandum III "Methods for Health Services Research", supported by their member societies mentioned as authors and published in this Journal [Gesundheitswesen 2010; 72: 739-748]. Registries in Health Services Research vary in their aims and research questions as well as in their designs, methods of data collection, and statistical analyses. This paper aims to provide both a methodological guideline for developers to ensure a high quality of a planned registry and, to provide an instrument for users of data from registries to assess their overall quality. First, the paper provides a definition of registries and presents an overview of objectives in Health Services Research where registries can be useful. Second, several areas of methodological importance for the development of registries are presented. This includes the different phases of a registry (i. e., conceptual and preliminary design, implementation), technical organisation of a registry, statistical analysis, reporting of results, data protection, and ethical/legal aspects. From these areas, several criteria are deduced to allow the assessment of the quality of a registry. Finally, a checklist to assess a registry's quality is presented.

Registers for Networked Medical Research in Germany: Situation and prospects.

BACKGROUND: Several disease specific registers are operated by members of the 'TMF - Technology, Methods, and Infrastructure for Networked Medical Research', an umbrella organization of research networks in Germany. OBJECTIVE: To describe the coverage and the current state as well as financial and organizational issues of registers operated by member networks of the TMF, to identify their requirements and needs, and to recommend best practice models. METHODS: A survey with a self-completion questionnaire including all 55 TMF member networks was carried out in winter 2007/2008. Interviews focusing on technological issues were conducted and analyzed in summer 2009 with a convenience sample of 10 registers. RESULTS: From 55 TMF member networks, 11 provided information about 14 registers. Six registers address diseases of the circulatory system with more than 150,000 registered patients. The interviews revealed a typical setting of "research registers". Research registers are an important mean to generate hypotheses for clinical research, to identify eligible patients, and to share data with clinical trials. Concerning technical solutions, we found a remarkable heterogeneity. The analysis of the most efficient registers revealed a structure with five levels as best practice model of register management: executive, operations, IT-management, software, hardware. CONCLUSION: In the last ten years, the TMF member networks established disease specific registers in Germany mainly to support clinical research. The heterogeneity of organizational and technical solutions as well as deficits in register planning motivated the development of respective recommendations. The TMF will continue to assist the registers in quality improvement.

Inpatient management of patients with skin cancer in Germany: an analysis of the nationwide DRG-statistic 2005-2006.

BACKGROUND: In 2004, Germany introduced a nationwide DRG (Diagnosis Related Groups) based recompensation system for hospitals. The aim of this study was to provide nationwide quantitative information about the in-hospital management of skin cancer patients in Germany based on the DRG statistic of the years 2005 through 2006. METHODS: We analysed the DRG statistic of the years 2005 and 2006. For each hospitalisation, diagnoses and procedures codes were analysed. The unit of analysis was the hospital admission with a diagnosis of skin melanoma (MEL) or nonmelanoma skin cancer (NMSC) including 151.144 hospitalisations. RESULTS: In 60% and 28% of all MEL- and NMSC-related hospitalisations respectively, skin cancers were surgically excised. Local therapies other than surgical excision were more prevalent among hospitalisations for NMSC than for MEL (9% vs. 4%). 22% and 1% of all MEL-related and NMSC-related hospitalisations respectively included a systemic chemotherapy. Plastic surgery was more common among NMSC-related hospitalisations (56%) than among MEL-related hospitalisations (30%). Native CT or MRI scans were coded in 11% and 2% of all MEL- and NMSC-related hospitalisations respectively. Age-standardized hospitalisation rates for MEL and NMSC varied considerably across the 16 Federal States of Germany. However, these rates were neither associated with the number of dermatology hospital beds or number of registered dermatologists. DISCUSSION: We provide for the first time nationwide quantitative data on the in-hospital management of skin cancer patients. The observed differences in the management of MEL and NMSC most likely reflect differences of the biology, epidemiology and therapeutic modalities of these cancers.

[Frequency of hospital-acquired pneumonia--comparison between electronic and paper-based patient records]. / Häufigkeit der nosokomialen Pneumonie im Krankenhaus--Vergleich von rechnergestützter Basisdokumentation und papierbasierter Krankenakte.

BACKGROUND: The aim of the study was to check the validity of the electronic patient records for hospital-acquired pneumonia and to estimate their reliability. PATIENTS AND METHODS: On reviewing 23,356 inpatients with admission from 1st April 2005 and discharge up to the 30th September 2005, we identified 211 cases with hospital-acquired pneumonia in the electronic patient records whereby 70 cases taken at random were included in the calculation of the positive predictive value. A second random sample of 130 cases under risk was used to calculate its sensitivity. Positive predictive value and sensitivity were combined to estimate the true rate of hospital-acquired pneumonia. RESULTS: In 64.3 % of the cases hospital-acquired pneumonia was confirmed in the paper-based patient records (45 cases, 95 % confidence interval 51.9 - 75.4 %). Sensitivity was calculated as 42.9 % and specificity as 99.0 %. Estimation for the University Clinics Essen revealed a true rate of hospital-acquired pneumonia of 1.32 %. The estimated true rate was higher than the measured rate of 0.90 % which is covered, nevertheless, by the 95 % confidence interval. CONCLUSIONS: Results from electronic and paper-based patient records are similar for hospital-acquired pneumonia, in spite of relevant differences on the level of the individual cases. However, data quality is not sufficient for special claims of supervision, clinical hygiene and prevention, for which a further optimisation is required.

Nationwide hospitalization costs of skin melanoma and non-melanoma skin cancer in Germany.

INTRODUCTION: Non-melanoma skin cancer (NMSC), in contrast to skin melanoma (MEL), is considered a negligible health problem because mortality of NMSC is low. The aim of this study was to provide insights into the burden of NMSC and MEL by analysing nationwide skin cancer hospitalization data and data from a dermatologist panel of Germany. We wanted to estimate hospitalization costs due to skin cancer in Germany. MATERIAL AND METHODS: We analysed the most recent nationwide hospitalization data from 2003 and estimated hospitalization costs due to MEL and NMSC. We estimated the annual number of private dermatologist practice visits in Germany due to skin cancer. RESULTS: In 2003, 20 455 melanoma-related and 41 929 NMSC-related hospitalizations occurred in Germany. Age-standardized hospitalization rates for NMSC were 2.5-fold and 1.8-fold higher among men and women than the rates for MEL, respectively. The age-specific proportions of hospitalizations for NMSC in relation to all cancer-related hospitalizations increased within the age range of 65 years and more. Among people aged 90 years and more, 14% of all cancer-related hospitalizations were due to NMSC. Estimated annual hospitalization costs for MEL were euro50 to 60 million, and those for NMSC were euro105 to 130 million. The estimated number of private dermatologist practice visits in Germany 2003 is considerably higher for NMSC than MEL CONCLUSIONS: Analyses of hospitalizations data and data from private dermatologists give NMSC higher public health relevance than can be obtained from consideration of death statistics.

Integrated information systems for translational medicine.

OBJECTIVES: Translational medicine research needs a two-way information highway between 'bedside' and 'bench'. Unfortunately there are still weak links between successfully integrated information roads for bench, i.e. research networks, and bedside, i.e. regional or national health information systems. The question arises, what measures have to be taken to overcome the deficiencies. METHODS: It is examined how patient care-related costs of clinical research can be separated and shared by health insurances, whether quality of patient care data is sufficient for research, how patient identity can be maintained without conflict to privacy, how care and research records can be archived, and how information systems for care and research can be integrated. RESULTS: Since clinical trials improve quality of care, insurers share parts of the costs. Quality of care data has to be improved by introducing minimum basic data sets. Pseudonymization solves the conflict between needs for patient identity and privacy. Archiving patient care records and research records is similar and XML and CDISC can be used. Principles of networking infrastructures for care and research still differ. They have to be bridged first and harmonized later. CONCLUSIONS: To link information systems for care (bed) and for research (bench) needs technical infrastructures as well as economic and organizational regulations.

[Quality of coding in acute inpatient care]. / Die Kodierqualität in der stationären Versorgung.

Routine data in the electronic patient record are frequently used for secondary purposes. Core elements of the electronic patient record are diagnoses and procedures, coded with the mandatory classifications. Despite the important role of routine data for reimbursement, quality management and health care statistics, there is currently no systematic analysis of coding quality in Germany. Respective concepts and investigations share the difficulty to decide what's right and what's wrong, being at the end of the long process of medical decision making. Therefore, a relevant amount of disagreement has to be accepted. In case of the principal diagnosis, this could be the fact in half of the patients. Plausibility of coding looks much better. After optimization time in hospitals, regular and complete coding can be expected. Whether coding matches reality, as a prerequisite for further use of the data in medicine and health politics, should be investigated in controlled trials in the future.

[Frequency of decubitus ulcer in patients of a university medical center. Combination of routine documentation and cross-sectional study]. / Häufigkeit des Dekubitus in einem Universitätsklinikum. Kombination von Routinedokumentation und Querschnittstudie.

BACKGROUND AND OBJECTIVE: Pressure ulcer is a relevant issue for quality management and cost containment of hospitals. Cross-sectional studies are the typical design to estimate the frequency of pressure ulcers. The derived point prevalence rate is not as good for a case related reference value as the period prevalence rate. The interdisciplinary pressure ulcer project at the University Clinics in Essen combined a routine documentation with a cross-sectional survey, thus providing both measurements for the first time. PATIENTS AND METHODS: The routine and computer-based collection of information about pressure ulcers started in March 2003, using the patient administration system medico//s from Siemens. Findings are presented from 49,904 admissions, starting on 91/03/2004, discharged by 31/03/2004. The mean age was 48.7 +/- 22.4 years; 51.2 % were males, 48.8 % females. Additionally, a decubitus team examined patients from randomly selected wards each work-day. The real period prevalence rate was calculated using the observed period prevalence rate, the sensitivity and the point prevalence rate. RESULTS: In the routine documentation, 700 pressure ulcers had been recorded from 49,904 inpatient cases (period prevalence rate 1.4 %). The decubitus team did 5,415 examinations and 294 times detected at least one pressure ulcer (point prevalence rate 5.4 %). Estimated results in three different period prevalence rates: 3.0 % using sensitivity of the routine documentation, 3.7 % using sensitivity stratified for departments, and 2.3 % using the point prevalence rate for cases with short, medium, and long length of stay. CONCLUSIONS: The project presents for the first time reference values for pressure ulcer frequency in university clinics. A comparison with international rates is hindered by unpublished sensitivity values. In view of the higher point prevalence rate of 10 % present in the literature, a period prevalence rate of 5 % is a realistic reference value.