Toward a comprehensive cancer registration in Germany

Population-Based Clinical Cancer Registration in Germany

INTRODUCTION

In 2013, a new federal law obligated all German federal states to collect additional clinical data in population-based cancer registries as an active tool for monitoring and improving the quality of cancer care, increasing transparency and promoting health research. Now, 10 years later, the current status of the expanded cancer registration is presented, including current figures on cancer in Germany.

METHODS

Reporting of cancer is mandatory for physicians, and about 5 to 10 reports from different healthcare providers are expected for each case. A uniform national dataset of about 130 items is used, and reports are usually sent electronically to the registry. We used the most recent data available from cancer registries up to the year of diagnosis in 2019. We calculated incidence rates and 5-year relative survival (5YRS) for common cancers. Data on clinical outcomes and benchmarking based on quality indicators (QIs) from guidelines were provided by the Cancer Registry Schleswig-Holstein (CR SH).

RESULTS

All federal state cancer registries met most of the previously defined national eligibility criteria. Approximately 505,000 cancer cases were registered in 2019, with breast, prostate, colorectal and lung cancer being the most common cancers. The age-standardised cancer incidence has slightly decreased during the last decade. and spatial heterogeneity can be observed within Germany. 5YRS for all cancers was 67% and 63% for women and men, respectively. Therapy data for rectal cancer in 2019-2021 from the CR SH are shown as an example: 69% of the registered patients underwent surgery, mostly with curative intent (84%) and tumour-free resection (91%). Radiotherapy was given to 33% of the patients, and chemotherapy was given to 40%. Three selected QIs showed differences between involved healthcare providers.

DISCUSSION

The implementation of population-based clinical cancer registration can be considered a success. Comprehensive recording of diagnosis, treatment and disease progression and the use of registry data for quality assurance, benchmarking and feedback have been implemented.

Unsupervised anomaly detection of implausible electronic health records: a real-world evaluation in cancer registries

BACKGROUND

Cancer registries collect patient-specific information about cancer diseases. The collected information is verified and made available to clinical researchers, physicians, and patients. When processing information, cancer registries verify that the patient-specific records they collect are plausible. This means that the collected information about a particular patient makes medical sense.

METHODS

Unsupervised machine learning approaches can detect implausible electronic health records without human guidance. Therefore, this article investigates two unsupervised anomaly detection approaches, a pattern-based approach (FindFPOF) and a compression-based approach (autoencoder), to identify implausible electronic health records in cancer registries. Unlike most existing work that analyzes synthetic anomalies, we compare the performance of both approaches and a baseline (random selection of records) on a real-world dataset. The dataset contains 21,104 electronic health records of patients with breast, colorectal, and prostate tumors. Each record consists of 16 categorical variables describing the disease, the patient, and the diagnostic procedure. The samples identified by FindFPOF, the autoencoder, and a random selection-a total of 785 different records-are evaluated in a real-world scenario by medical domain experts.

RESULTS

Both anomaly detection methods are good at detecting implausible electronic health records. First, domain experts identified [Formula: see text] of 300 randomly selected records as implausible. With FindFPOF and the autoencoder, [Formula: see text] of the proposed 300 records in each sample were implausible. This corresponds to a precision of [Formula: see text] for FindFPOF and the autoencoder. Second, for 300 randomly selected records that were labeled by domain experts, the sensitivity of the autoencoder was [Formula: see text] and the sensitivity of FindFPOF was [Formula: see text]. Both anomaly detection methods had a specificity of [Formula: see text]. Third, FindFPOF and the autoencoder suggested samples with a different distribution of values than the overall dataset. For example, both anomaly detection methods suggested a higher proportion of colorectal records, the tumor localization with the highest percentage of implausible records in a randomly selected sample.

CONCLUSIONS

Unsupervised anomaly detection can significantly reduce the manual effort of domain experts to find implausible electronic health records in cancer registries. In our experiments, the manual effort was reduced by a factor of approximately 3.5 compared to evaluating a random sample.

Proposal for a descriptive and differentiated presentation of the longitudinal impact of the new organized cancer screening guideline and HPV vaccination in Germany

INTRODUCTION

Since 01/01/2020, the cervical cancer screening in Germany has been carried out due to the organized early cancer diagnosis guideline (oKFE-RL). In 2007, HPV vaccination was initiated in Germany. The main goal of both initiatives is to further reduce the incidence of invasive cervical cancer. To assess the effect of the new screening strategy in a timely manner, monitoring of short-term changes need to be considered. Ideally, the effects of both prevention methods would be presented together in one model.

MATERIALS AND METHODS

Because no change in the incidence of invasive cervical cancer is initially expected, the incidence of CIN 3 is used as a surrogate parameter to assess the effects of the prevention efforts. Based on expected additional effects of vaccination and co-testing, a model-based estimation of the expected CIN 3 incidence during the evaluation of the screening program is performed using the CIN 3 incidence in the Saarland population.

MODELING RESULTS

The oKFE-RL provides for two groups: Primary cytodiagnosis continues until 35 years of age. Here, in the next few years, CIN 3 incidence will be reduced not by the oKFE-RL but by the increasing proportion of vaccinated women. In the group over 35 years, co-testing was introduced with a stringent algorithm. Due to the higher sensitivity of the HPV test, significantly more CIN 3 are detected in the first round of 3 years and thus, the CIN 3 incidence initially increases. As these CIN 3 are absent in the second round, significantly fewer CIN 3 cases will be detected then. These effects suggest a global decrease in CIN 3 incidence of 25.8% after 6 years.

CONCLUSION

Observation of the age distribution curve of CIN 3 allows both effects of prevention to be assessed in a timely manner and separately. In the future, data from epidemiologic cancer registries should be incorporated into the model to replace modeling with real data.

Routine practice data of three cancer entities: Comparison among cancer registry and health insurance data

INTRODUCTION

Claims data and cancer registry data are valuable secondary data sources for addressing health service research questions. This study provides a thorough insight into the comparability of data from health insurance companies and cancer registries in Germany regarding breast, prostate, and lung cancer patients and their treatment.

METHODS

For this study claims data of the InGef database and data of the Cancer Registry of Rhineland-Palatinate were used to identify patients living in Rhineland-Palatinate with an incident breast, prostate, or lung cancer diagnosis between Jan. 1, 2018 and Dec. 31, 2019. Both datasets were compared for patient and tumour characteristics as well as treatment strategy. For the descriptive analysis of tumour localisation and treatment all patients were followed up for a maximum of two years.

RESULTS

A total of 1,470 incident cancer cases were identified in the InGef database and 1,694 in the Cancer Registry. Data on sex, age, and tumour localisation matched well for all cancer entities in the cohorts. Data for early UICC stages I+II varied between the cohorts for prostate cancer (84% InGef, 66% Cancer Registry) and lung cancer (29% InGef, 20% Cancer Registry). Larger deviations were found for antihormonal treatment (breast 54% vs. 44%, prostate 32% vs. 18%). Significant differences were found for surgery (breast and lung) and radiation (breast and prostate), respectively.

DISCUSSION

Age at diagnosis, tumour localisation, and treatment for breast cancer was well documented in both databases. Tumour-specific deviations were observed for tumour localisations (lung cancer), UICC stage (prostate and lung cancer) and treatment options.

CONCLUSION

Both databases show very good completeness across cancer entities, but at the same time have minor limitations where they could readily complement each other. Individual linkage of claims and registry data could be an important step to improve oncological studies with routine practice data and to overcome the limitations identified.

Initial and ten-year treatment patterns among 11,000 breast cancer patients undergoing breast surgery-an analysis of German claims data

BACKGROUND

We aimed to explore the potential of German claims data for describing initial and long-term treatment patterns of breast cancer patients undergoing surgery.

METHODS

Using the German Pharmacoepidemiological Research Database (GePaRD, ~ 20% of the German population) we included patients with invasive breast cancer diagnosed in 2008 undergoing breast surgery and followed them until 2017. We described initial and long-term treatment patterns and deaths. Analyses were stratified by stage (as far as available in claims data), age at diagnosis, and mode of detection (screen-detected vs. interval vs. unscreened cases).

RESULTS

The cohort comprised 10,802 patients. The proportion with neoadjuvant therapy was highest in patients < 50 years (19% vs. ≤ 8% at older ages). The proportion initiating adjuvant chemotherapy within four months after diagnosis decreased with age (< 50 years: 63%, 50-69: 46%, 70-79: 27%, 80 + : 4%). Among women < 69 years, ~ 30% had two breast surgeries in year one (70-79: 21%, 80 + : 14%). Treatment intensity was lower for screen-detected compared to interval or unscreened cases, both in year one (e.g., proportion with mastectomy ~ 50% lower) and within 2-10 years after surgery (proportions with radiotherapy or chemotherapy about one third lower each).

CONCLUSIONS

This study illustrates the potential of routine data to describe breast cancer treatment and provided important insights into differences in initial and long-term treatment by mode of detection and age.

[Oral and pharyngeal cancer: incidence, mortality, and survival in Germany]

Hintergrund

Die Gruppe der Lippen‑, Mundhöhlen- und Pharynxkarzinome (ICD-10: C00–C14) beschreibt eine heterogene Gruppe bösartiger Tumoren, deren Inzidenz, Mortalität und Überleben sich nach Entität stark unterscheiden.

Ziel der Arbeit

Diese Arbeit gibt eine detaillierte Übersicht epidemiologischer Maßzahlen für diese Tumorentitäten unter Berücksichtigung der Heterogenität bezüglich Alter, Geschlecht, Lokalisation und Stadium.

Material und Methoden

Inzidenz- und Mortalitätsdaten für Deutschland für die Jahre 1999–2016 wurden aus der interaktiven Datenbank des Zentrums für Krebsregisterdaten (ZfKD) extrahiert. Alters- und Stadienverteilungen und altersstandardisiertes relatives 5‑Jahres-Überleben wurden auf dem gepoolten ZfKD-Datensatz (Diagnosejahre 1999–2017) berechnet.

Ergebnisse

Im Jahr 2016 lagen die Inzidenz und Mortalität für alle Entitäten insgesamt bei 17,6 bzw. 7,0 pro 100.000 Männern und 6,5 bzw. 1,8 pro 100.000 Frauen. Das relative 5‑Jahres-Überleben für 2015–2017 lag bei 53 % bzw. 63 %. Es zeigten sich deutliche Unterschiede in der Überlebensrate und in der Alters- und Stadienverteilung zwischen den Entitäten. Im zeitlichen Verlauf zeigte sich ein Anstieg des Alters bei Diagnose, insbesondere bei Männern, aber keine Veränderung in der Stadienverteilung. Trotzdem stieg das relative 5‑Jahres-Überleben zwischen 1999–2002 und 2013–2017 von 45 % (Männer) bzw. 59 % (Frauen) auf 52 % bzw. 63 %.

Diskussion

Die starke Heterogenität der untersuchten Tumoren verdeutlicht die Notwendigkeit einer nach Geschlechtern und Lokalisationen getrennten Betrachtung für eine aussagekräftige Interpretation der epidemiologischen Kennzahlen. Mit Ausbau der klinischen Krebsregistrierung werden in Zukunft zusätzliche Analysen unter Einbezug weiterer wichtiger klinischer Faktoren möglich sein.

Zusatzmaterial online

Zusätzliche Informationen sind in der Online-Version dieses Artikels (10.1007/s00103-021-03368-z) enthalten.

Small-area analysis on socioeconomic inequalities in cancer survival for 25 cancer sites in Germany

Socioeconomic inequalities in cancer survival have been reported in various countries but it is uncertain to what extent they persist in countries with relatively comprehensive health insurance coverage such as Germany. We investigated the association between area-based socioeconomic deprivation on municipality level and cancer survival for 25 cancer sites in Germany. We used data from seven population-based cancer registries (covering 32 million inhabitants). Patients diagnosed in 1998 to 2014 with one of 25 most common cancer sites were included. Area-based socioeconomic deprivation was assessed using the categorized German Index of Multiple Deprivation (GIMD) on municipality level. We estimated 3-month, 1-year, 5-year and 5-year conditional on 1-year age-standardized relative survival using period approach for 2012 to 2014. Trend analyses were conducted for periods between 2003-2005 and 2012-2014. Model-based period analysis was used to calculate relative excess risks (RER) adjusted for age and stage. In total, 2 333 547 cases were included. For all cancers combined, 5-year survival rates by GIMD quintile were 61.6% in Q1 (least deprived), 61.2% in Q2, 60.4% in Q3, 59.9% in Q4 and 59.0% in Q5 (most deprived). For most cancer sites, the most deprived quintile had lower 5-year survival compared to the least deprived quintile even after adjusting for stage (all cancer sites combined, RER 1.16, 95% confidence interval 1.14-1.19). For some cancer sites, this association was stronger during short-term follow-up. Trend analyses showed improved survival from earlier to recent periods but persisting deprivation differences. The underlying reasons for these persisting survival inequalities and strategies to overcome them should be further investigated.

Disclosing progress in cancer survival with less delay

Cancer registration plays a key role in monitoring the burden of cancer. However, cancer registry (CR) data are usually made available with substantial delay to ensure best possible completeness of case ascertainment. Here, we investigate empirically with routinely available data whether such a delay is mandatory for survival analyses or whether data can be used earlier to provide more up-to-date survival estimates. We compared distributions of prognostic factors and period relative survival estimates for three population-based CRs in Germany (Schleswig-Holstein (SH), Rhineland-Palatinate (RP), Saarland (SA)) computed on datasets extracted one (DY+1) to 5 years after the year of diagnosis (DY+5; reference). Analyses were conducted for seven cancer sites and various survival analyses scenarios. The proportion of patients registered in the datasets at a given time varied strongly across registries with 57% (SH), 2% (RP) and 26% (SA) registered in DY+1 and >93% in all registries in DY+3. Five-year survival estimates for the most recent three-year period were comparable to estimates from the reference dataset already in DY+1 (mean absolute deviations = 0.2-0.6% units). Deviations >1% units were only observed for pancreatic and lung cancer in RP and leukemia in SA (all ≤1.5% units). For estimates of 1-year survival based on the most recent 1-year period only, slightly longer delays were required, but reasonable estimates were still obtained after 1-2 years, depending on the CR and cancer site. Thus, progress in cancer survival could be disclosed in a more timely manner than commonly practiced despite delays in completeness of registration.

Area-Based Socioeconomic Inequalities in Colorectal Cancer Survival in Germany: Investigation Based on Population-Based Clinical Cancer Registration

Background: Socioeconomic inequalities in colorectal cancer survival have been observed in many countries. To overcome these inequalities, the underlying reasons must be disclosed.

Methods: Using data from three population-based clinical cancer registries in Germany, we investigated whether associations between area-based socioeconomic deprivation and survival after colorectal cancer depended on patient-, tumor- or treatment-related factors. Patients with a diagnosis of colorectal cancer in 2000–2015 were assigned to one of five deprivation groups according to the municipality of the place of residence using the German Index of Multiple Deprivation. Cox proportional hazards regression models with various levels of adjustment and stratifications were applied.

Results: Among 38,130 patients, overall 5-year survival was 4.8% units lower in the most compared to the least deprived areas. Survival disparities were strongest in younger patients, in rectal cancer patients, in stage I cancer, in the latest period, and with longer follow-up. Disparities persisted after adjustment for stage, utilization of surgery and screening colonoscopy uptake rates. They were mostly still present when restricting to patients receiving treatment according to guidelines.

Conclusion: We observed socioeconomic inequalities in colorectal cancer survival in Germany. Further studies accounting for potential differences in non-cancer mortality and exploring treatment patterns in detail are needed.

Socioeconomic differences and lung cancer survival in Germany: Investigation based on population-based clinical cancer registration

OBJECTIVES

Studies from several countries reported socioeconomic inequalities in lung cancer survival. Hypothesized reasons are differences in cancer care or tumor characteristics. We investigated associations of small-area deprivation and lung cancer survival in Germany and the possible impact of differences in patient, tumor or treatment factors.

MATERIALS AND METHODS

Patients registered with a primary tumor of the lung between 2000-2015 in three German population-based clinical cancer registries were included. Area-based socioeconomic deprivation on municipality level was measured with the categorized German Index of Multiple Deprivation. Association of deprivation with overall survival was investigated with Cox regression models.

RESULTS

Overall, 22,905 patients were included. Five-year overall survival from the least to the most deprived quintile were 17.2%, 15.9%, 16.7%, 15.7%, and 14.4%. After adjustment for patient and tumor factors, the most deprived group had a lower survival compared to the least deprived group (Hazard Ratio (HR) 1.06, 95% confidence interval (CI) 1.01-1.11). Subgroup analyses revealed lower survival in the most deprived compared to the least deprived quintile in patients with stage I-III [HR: 1.14, 95% CI: 1.06-1.22]. The association persisted when restricting to patients receiving surgery but was attenuated for subgroups receiving either chemotherapy or radiotherapy.

CONCLUSION

Our results indicate differences in lung cancer survival according to area deprivation in Germany, which were more pronounced in patients with I-III stage cancer. Future research should address in more detail the underlying reasons for the observed inequalities and possible approaches to overcome them.

Risk of loco-regional recurrence and distant metastases of patients with invasive breast cancer up to ten years after diagnosis - results from a registry-based study from Germany

BACKGROUND

Population-based estimates of the long-term risk of loco-regional recurrence and distant metastases of breast cancer (BRC) patients are scant, as most published studies used hospital-based cohorts or participants of clinical trials. This work aims to extend available knowledge by providing population-based long-term estimates of the cumulative risk of BRC recurrence up to 10 years after diagnosis.

METHODS

Data from the population-based Saarland Cancer Registry were used and included 9359 female patients with primary invasive BRC diagnosed between 1999 and 2009. Estimates of the cumulative incidence (CI) of BRC recurrence were derived for patients who had received local surgery with free resection margins by type of recurrence and stratified by age, tumor characteristics and major treatment options, taking into account mortality from any cause as a competing risk.

RESULTS

The 10-year CI of BRC recurrence was 16%. For loco-regional recurrence and distant metastases alone it was 8 and 11%, respectively. The estimates showed substantial variation and were particularly increased if tumors were advanced (T1/2N+ 23%, T3/4N0 24%, T3/4N+ 34%), of high grade (23%), or of 'HER2/neu positive' (28%) or 'triple negative' subtype (23%), respectively.

CONCLUSIONS

The derived estimates reflect the risk of 'real world' patients and may therefore extend available knowledge. These data are thus of great relevance for clinicians, their patients and researchers. The study likewise demonstrated the usefulness of cancer registries for a population-based monitoring of the effectiveness of cancer care in terms of disease recurrence as a major treatment related outcome measure.