Implementation of an algorithm for the identification of breast cancer deaths in German health insurance claims data: a validation study based on a record linkage with administrative mortality data

Enrichment of health insurance claims data with official death certificate information from three German cancer registries: Proportions of successful linkages and differences by region, year, and age

Evaluating breast cancer mortality in the German mammography screening program with health insurance claims data requires the availability of claims data with information on causes of death. This work aimed to determine the proportions of successful cause-of-death linkages between the second-largest German statutory health insurance fund and three federal cancer registries and to investigate whether linked proportions differed by region, year, and age. Women aged 40-90 years whose insurance was terminated between 2006 and 2018 were included. Proportions successfully linked to the official death certificate databases of all individuals (available in one registry) and of registered cancer cases (available in three registries) were calculated. Of 150,369 women whose insurance was terminated due to death, 90.0% were linked to the database including all deceased women. Regarding the databases including only registered cancer cases, 35.9% of 150,369, 38.6% of 47,472, and 20.1% of 65,893 deceased women were linked. Linked proportions increased from 2006 to 2018 and peaked in age group 60-69 years. The data will be used for the evaluation of the German Mammography screening program. Since causes of death were not linked for all deceased women and the proportions of linkages differed by region, year, and age, claims-based algorithms will also be considered to complement claims data with causes of death.

The EU project Real4Reg: unlocking real-world data with AI

BACKGROUND

The use of real-world data is established in post-authorization regulatory processes such as pharmacovigilance of drugs and medical devices, but is still frequently challenged in the pre-authorization phase of medicinal products. In addition, the use of real-world data, even in post-authorization steps, is constrained by the availability and heterogeneity of real-world data and by challenges in analysing data from different settings and sources. Moreover, there are emerging opportunities in the use of artificial intelligence in healthcare research, but also a lack of knowledge on its appropriate application to heterogeneous real-world data sources to increase evidentiary value in the regulatory decision-making and health technology assessment context.

METHODS

The Real4Reg project aims to enable the use of real-world data by developing user-friendly solutions for the data analytical needs of health regulatory and health technology assessment bodies across the European Union. These include artificial intelligence algorithms for the effective analysis of real-world data in regulatory decision-making and health technology assessment. The project aims to investigate the value of real-world data from different sources to generate high-quality, accessible, population-based information relevant along the product life cycle. A total of four use cases are used to provide good practice examples for analyses of real-world data for the evaluation and pre-authorization stage, the improvement of methods for external validity in observational data, for post-authorization safety studies and comparative effectiveness using real-world data. This position paper introduces the objectives and structure of the Real4Reg project and discusses its important role in the context of existing European projects focussing on real-world data.

DISCUSSION

Real4Reg focusses on the identification and description of benefits and risks of new and optimized methods in real-world data analysis including aspects of safety, effectiveness, interoperability, appropriateness, accessibility, comparative value creation and sustainability. The project's results will support better decision-making about medicines and benefit patients' health. Trial registration Real4Reg is registered in the HMA-EMA Catalogues of real-world data sources and studies (EU PAS number EUPAS105544).

Implementation and Evaluation of a Breast Cancer Disease Model Using Real-World Claims Data in Germany from 2010 to 2020

Breast cancer is the leading cause of cancer-related mortality among women in Germany and worldwide. This retrospective claims data analysis utilizing data from AOK Baden-Wuerttemberg, a major statutory German health insurance provider, aimed to construct and assess a real-world data breast cancer disease model. The study included 27,869 female breast cancer patients and 55,738 age-matched controls, analyzing data from 2010 to 2020. Three distinct breast cancer stages were analyzed: Stage A (early breast cancer without lymph node involvement), Stage B (early breast cancer with lymph node involvement), and Stage C (primary distant metastatic breast cancer). Tumor subtypes were estimated based on the prescription of antihormonal or HER2-targeted therapy. The study established that 77.9% of patients had HR+ breast cancer and 9.8% HER2+; HR+/HER2- was the most common subtype (70.9%). Overall survival (OS) analysis demonstrated significantly lower survival rates for stages B and C than for controls, with 5-year OS rates ranging from 79.3% for stage B to 35.4% for stage C. OS rates were further stratified by tumor subtype and stage, revealing varying prognoses. Distant recurrence-free survival (DRFS) analysis showed higher recurrence rates in stage B than in stage A, with HR-/HER2- displaying the worst DRFS. This study, the first to model breast cancer subtypes, stages, and outcomes using German claims data, provides valuable insights into real-world breast cancer epidemiology and demonstrates that this breast cancer disease model has the potential to be representative of treatment outcomes.

The Lack of a Standardized Definition of Chronic Dialysis Treatment in German Statutory Health Insurance Claims Data—Effects on Estimated Incidence and Mortality

BACKGROUND

Chronic kidney failure (CKF) is often treated with dialysis, which is invasive and costly and carries major medical risks. The existing studies of patients with CKF requiring dialysis that are based on claims data from German statutory health insurance (SHI) carriers employ varying definitions of this entity, with unclear consequences for the resulting statistical estimates.

METHODS

We carried out a cohort study on four random samples, each consisting of 62 200 persons aged 70 or above, from among the insurees of the SHI AOK Nordost, with one sample for each of the years 2012, 2014, 2016, and 2018. The prevalence, incidence, mortality, and direct health care costs of CKF requiring dialysis were estimated and compared on the basis of four different definitions from literature and a new definition developed by the authors in reference to billing data.

RESULTS

The different definitions led to variation in 12-month prevalences (range: 0.33-0.61%) and 6-month incidences (0.058-0.100%). The percentage of patients with prior acute kidney injury (AKI) ranged from 27.6% to 61.8%. Among incident patients, three-month survival ranged from 70.2% to 88.1%, and six-month survival from 60.5% to 81.3%. In CKF patients without prior AKI, the survival curves differed less across definitions (80.2-91.8% at three months, 70.7-84.4% at six months). The monthly health care costs ranged from €6010 to €9606, with marked variability across definitions in the costs of inpatient and outpatient care.

CONCLUSION

The lack of a standardized definition of CKF requiring dialysis in German SHI claims data leads to variability in the estimated case numbers, mortality, and health care costs. These differences are most probably in part due to the variable inclusion of inpatients who received short-term dialysis after AKI.

Editor's Choice - Long Term Outcomes After Invasive Treatment of Carotid Artery Stenosis: a Longitudinal Study of German Health Insurance Claims

OBJECTIVE

There is a paucity of observational data including long term outcomes after invasive treatment for carotid artery stenosis.

METHODS

This retrospective study used nationwide insurance claims from the third largest provider in Germany, DAK-Gesundheit. Patients who underwent inpatient carotid endarterectomy (CEA) or carotid artery stenting (CAS) between 1 January 2008 and 31 May 2017 were included. The Elixhauser comorbidity scores from longitudinally linked hospital episodes were used. Kaplan-Meier analysis and the log rank test were used to determine long term stroke free survival. Multivariable regression models were developed to adjust for confounding.

RESULTS

A total of 22 637 individual patients (41.6% female, median age 72.5 years) were included, of whom 15 005 (66.3%) were asymptomatic and 17 955 (79.3%) underwent CEA. After a median of 48 months, 5 504 any stroke or death events were registered. The mortality rate varied between 0.4% (CEA for asymptomatic stenosis) and 2.1% (urgent CAS for acute stroke patients) at 30 days, and between 4.1% and 8.4% at one year, respectively. The rate for any stroke varied between 0.6% (CEA for asymptomatic stenosis) and 2.5% (CAS for symptomatic patients) at 30 days, and between 2.5% and 6.4% at one year, respectively. The combined rate for any stroke and mortality at one year was 6.3% (CEA for asymptomatic stenosis), 8.7% (CAS for asymptomatic stenosis), and 12.5% (urgent CAS for acute stroke patients). After five years, the overall stroke rate was 7.4% after CEA and 9.0% after CAS. In adjusted analyses, both older age and van Walraven comorbidity score were associated with events, while treatment of asymptomatic stenosis was associated with lower event rates.

CONCLUSION

The current study revealed striking differences between previous landmark trials and real world practice. It further suggested excess deaths among invasively treated asymptomatic patients.

Novel Algorithm for the Estimation of Cancer Incidence Using Claims Data in Japan: A Feasibility Study

PURPOSE

We developed algorithms to identify patients with newly diagnosed cancer from a Japanese claims database to identify the patients with newly diagnosed cancer of the sample population, which were compared with the nationwide cancer incidence in Japan to assess the validity of the novel algorithms.

METHODS

We developed two algorithms to identify patients with stomach, lung, colorectal, breast, and cervical cancers: diagnosis only (algorithm 1), and combining diagnosis, treatments, and medicines (algorithm 2). Patients with newly diagnosed cancer were identified from an anonymized commercial claims database (JMDC Claims Database) in 2017 with two inclusions/exclusion criteria: selecting all patients with cancer (extract 1) and excluding patients who had received cancer treatments in 2015 or 2016 (extract 2). We estimated the cancer incidence of the five cancer sites and compared it with the Japan National Cancer Registry incidence (calculated standardized incidence ratio with 95% CIs).

RESULTS

The number of patients with newly diagnosed cancer ranged from 219 to 17,840 by the sites, algorithms, and exclusion criteria. Standardized incidence ratios were significantly higher in the JMDC Claims Database than in the national registry data for extract 1 and algorithm 1, extract 1 and algorithm 2, and extract 2 and algorithm 1. In extract 2 and algorithm 2, colorectal cancer in male and stomach, lung, and cervical cancers in females showed similar cancer incidence in the JMDC and national registry data.

CONCLUSION

The novel algorithms are effective for extracting information about patients with cancer from claims data by using the combined information on diagnosis, procedures, and medicines (algorithm 2), with 2-year cancer-treatment history as an exclusion criterion (extract 2).

Effectiveness of Mammography Screening on Breast Cancer Mortality – A Study Protocol for Emulation of Target Trials Using German Health Claims Data

Background

The efficacy of mammography screening in reducing breast cancer mortality has been demonstrated in randomized trials. However, treatment options - and hence prognosis – for advanced tumor stages as well as mammography techniques have considerably improved since completion of these trials. Consequently, the effectiveness of mammography screening under current conditions is unclear and controversial. The German mammography screening program (MSP), an organized population-based screening program, was gradually introduced between 2005 and 2008 and achieved nation-wide coverage in 2009.

Objective

We describe in detail a study protocol for investigating the effectiveness of the German MSP in reducing breast cancer mortality in women aged 50 to 69 years based on health claims data. Specifically, the proposed study aims at estimating per-protocol effects of several screening strategies on cumulative breast cancer mortality. The first analysis will be conducted once 10-year follow-up data are available.

Methods and Analysis

We will use claims data from five statutory health insurance providers in Germany, covering approximately 37.6 million individuals. To estimate the effectiveness of the MSP, hypothetical target trials will be emulated across time, an approach that has been demonstrated to minimize design-related biases. Specifically, the primary contrast will be in terms of the cumulative breast cancer mortality comparing the screening strategies of “never screen” versus “regular screening as intended by the MSP”.

Ethics and Dissemination

In Germany, the utilization of data from health insurances for scientific research is regulated by the Code of Social Law. All involved health insurance providers as well as the responsible authorities approved the use of the health claims data for this study. The Ethics Committee of the University of Bremen determined that studies based on claims data are exempt from institutional review. The findings of the proposed study will be published in peer-reviewed journals.

Development of algorithms for identifying patients with Crohn's disease in the Japanese health insurance claims database

Background

Real-world big data studies using health insurance claims databases require extraction algorithms to accurately identify target population and outcome. However, no algorithm for Crohn’s disease (CD) has yet been validated. In this study we aim to develop an algorithm for identifying CD using the claims data of the insurance system.

Methods

A single-center retrospective study to develop a CD extraction algorithm from insurance claims data was conducted. Patients visiting the Kitasato University Kitasato Institute Hospital between January 2015–February 2019 were enrolled, and data were extracted according to inclusion criteria combining the Tenth Revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10) diagnosis codes with or without prescription or surgical codes. Hundred cases that met each inclusion criterion were randomly sampled and positive predictive values (PPVs) were calculated according to the diagnosis in the medical chart. Of all cases, 20% were reviewed in duplicate, and the inter-observer agreement (Kappa) was also calculated.

Results

From the 82,898 enrolled, 255 cases were extracted by diagnosis code alone, 197 by the combination of diagnosis and prescription codes, and 197 by the combination of diagnosis codes and prescription or surgical codes. The PPV for confirmed CD cases was 83% by diagnosis codes alone, but improved to 97% by combining with prescription codes. The inter-observer agreement was 0.9903.

Conclusions

Single ICD-code alone was insufficient to define CD; however, the algorithm that combined diagnosis codes with prescription codes indicated a sufficiently high PPV and will enable outcome-based research on CD using the Japanese claims database.

Validity of claims-based algorithms for selected cancers in Japan: Results from the VALIDATE-J study

PURPOSE

Real-world data from large administrative claims databases in Japan have recently become available, but limited evidence exists to support their validity. VALIDATE-J validated claims-based algorithms for selected cancers in Japan.

METHODS

VALIDATE-J was a multicenter, cross-sectional, retrospective study. Disease-identifying algorithms were used to identify cancers diagnosed between January or March 2012 and December 2016 using claims data from two hospitals in Japan. Positive predictive values (PPVs), specificity, and sensitivity were calculated for prevalent (regardless of baseline cancer-free period) and incident (12-month cancer-free period; with claims and registry periods in the same month) cases, using hospital cancer registry data as gold standard.

RESULTS

22 108 cancers were identified in the hospital claims databases. PPVs (number of registry cases) for prevalent/incident cases were: any malignancy 79.0% (25 934)/73.1% (18 119); colorectal 84.4% (3519)/65.6% (2340); gastric 87.4% (3534)/76.8% (2279); lung 88.1% (2066)/79.9% (1636); breast 86.4% (4959)/59.9% (3185); pancreatic 87.1% (582)/80.4% (508); melanoma 48.7% (46)/42.9% (36); and lymphoma 83.6% (1457)/77.8% (1035). Specificity ranged from 98.3% to 100% (prevalent)/99.5% to 100% (incident); sensitivity ranged from 39.1% to 67.6% (prevalent)/12.5% to 31.4% (incident). PPVs of claims-based algorithms for several cancers in patients ≥66 years of age were slightly higher than those in a US Medicare population.

CONCLUSIONS

VALIDATE-J demonstrated high specificity and modest-to-moderate sensitivity for claims-based algorithms of most malignancies using Japanese claims data. Use of claims-based algorithms will enable identification of patient populations from claims databases, while avoiding direct patient identification. Further research is needed to confirm the generalizability of our results and applicability to specific subgroups of patient populations.

Machine-learning model to predict the cause of death using a stacking ensemble method for observational data

OBJECTIVE

Cause of death is used as an important outcome of clinical research; however, access to cause-of-death data is limited. This study aimed to develop and validate a machine-learning model that predicts the cause of death from the patient's last medical checkup.

MATERIALS AND METHODS

To classify the mortality status and each individual cause of death, we used a stacking ensemble method. The prediction outcomes were all-cause mortality, 8 leading causes of death in South Korea, and other causes. The clinical data of study populations were extracted from the national claims (n = 174 747) and electronic health records (n = 729 065) and were used for model development and external validation. Moreover, we imputed the cause of death from the data of 3 US claims databases (n = 994 518, 995 372, and 407 604, respectively). All databases were formatted to the Observational Medical Outcomes Partnership Common Data Model.

RESULTS

The generalized area under the receiver operating characteristic curve (AUROC) of the model predicting the cause of death within 60 days was 0.9511. Moreover, the AUROC of the external validation was 0.8887. Among the causes of death imputed in the Medicare Supplemental database, 11.32% of deaths were due to malignant neoplastic disease.

DISCUSSION

This study showed the potential of machine-learning models as a new alternative to address the lack of access to cause-of-death data. All processes were disclosed to maintain transparency, and the model was easily applicable to other institutions.

CONCLUSION

A machine-learning model with competent performance was developed to predict cause of death.

Quality of care in surgical/interventional vascular medicine: what can routinely collected data from the insurance companies achieve?

The complexity and diversity of surgical/interventional vascular medicine necessitate innovative and pragmatic solutions for the valid measurement of the quality of care in the long term. The secondary utilization of routinely collected data from social insurance institutions has increasingly become the focus of interdisciplinary medicine over the years. Owing to their longitudinal linkage and pan-sector generation, routinely collected data make it possible to answer important questions and can complement quality development projects with primary registry data. Various guidelines exist for their usage, linkage, and reporting. Studies have shown good validity, especially for endpoints with major clinical relevance. The numerous advantages of routinely collected data face several challenges that require thorough plausibility and validity procedures and distinctive methodological expertise. This review presents a discussion of these advantages and challenges and provides recommendations for starting to use this increasingly important source of data.