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

Abstract

ObjectiveTo adapt a Canadian algorithm for the identification of female cases of breast cancer (BC) deaths to German health insurance claims data and to test and validate the algorithm by comparing results with official cause of death (CoD) data on the individual and the population level.DesignValidation study, secondary data, medical claims.SettingClaims data of two statutory health insurance providers (SHIs) for inpatient and outpatient care, CoD added via record linkage with epidemiological cancer registry (ECR).ParticipantsAll women insured with the two SHIs and who deceased in the period 2006–2013, were residents of North Rhine Westphalia (NRW) and were linked with ECR data: n=22 413.Main outcome measuresBased on inpatient and outpatient diagnoses in the year before death, six algorithms were derived and the accordance of the algorithm-based CoD with the official CoD was evaluated calculating specificity, sensitivity, negative and positive predictive values (NPV, PPV). Furthermore, algorithm-based age-specific BC mortality rates covering several calendar years were calculated for the entire insured female population and compared with official national rates.ResultsOur final algorithm, derived from the NRW subsample, comprised codes indicating the presence of BC, metastases, a terminal illness phase and the absence of codes for other tumours. Overall, specificity, sensitivity, NPV and PPV of this algorithm were 97.4%, 91.3%, 98.9% and 81.7%, respectively. In the age range 40–80 years, sensitivity and PPV slightly decreased with increasing age. Algorithm-based age-specific BC mortality rates agreed well with official rates except for the age group 85 years and older.ConclusionsThe algorithm-based identification of BC deaths in German claims data is feasible and valid, except for higher ages. The algorithm to ascertain BC mortality rates in an epidemiological study seems applicable when information on the official CoD is not available in the original database.

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

“…PPV/sensitivity of 79%/81%, 88%/72%, and 93%/77%, respectively, for incident cases of lung ( n = 665), colorectal ( n = 796), and breast ( n = 897) cancer was reported using definitions based on hospital discharge codes 3 . Validation studies conducted in the EU using commercial insurance databases reported PPV of 69% and sensitivity of 66% for lymphoma identified by diagnostic codes ( n = 340), 4 and PPV of 82% with negative predictive value of 99%, specificity of 97%, and sensitivity of 91% for the identification of deaths due to breast cancer using algorithms adapted from those previously validated with Canadian data ( n = 22 413) 19 …”

“… 3 Validation studies conducted in the EU using commercial insurance databases reported PPV of 69% and sensitivity of 66% for lymphoma identified by diagnostic codes ( n = 340), 4 and PPV of 82% with negative predictive value of 99%, specificity of 97%, and sensitivity of 91% for the identification of deaths due to breast cancer using algorithms adapted from those previously validated with Canadian data ( n = 22 413). 19 …”

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

“…The algorithm has been developed in a sample for which both claims data and the official cause of death were directly linked. The initial version of the algorithm, described by Langner et al, 27 showed a sensitivity of 91.3% and a specificity of 97.4%, and is currently being further optimized, eg by also considering information on cancer treatment. For study participants living in the federal states of North Rhine-Westphalia, Bavaria, and Lower Saxony official cause of death records will be directly available by linkage to the cancer registry of the respective federal state.…”

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

“…It holds that greater severity of the endpoint is associated with higher validity, while less relevant observations have lower validity (inpatient sensitivity to hypertension 65%, cancer 91% and acute myocardial infarction 94% [44], outpatient sensitivity to back pain 74% and hypertension 81% [19]). Good validity can especially be assumed for mortality-related endpoints among the vascular patient cohort over the age of 65 years [15,16,25,32,38].…”