Validation of Mortality Information From a Real-World Data Source: Can Administrative Claims Alone be Trusted?

Speaker(s)

Palmer L, Brady B, Ross R
Merative, Ann Arbor, MI, USA

OBJECTIVES: While mortality is a key outcome in real-world evidence (RWE) generation, accessing death data is challenging; validity of this information can also be variable depending on the data source. This analysis evaluates validity of death information coming from employer-based sources relative to that contained in the Centers for Disease Control and Prevention’s National Death Index (NDI).

METHODS: A random sample of 6,000 adults selected from the MarketScan® by MerativeTM Commercial and Medicare Database (MSN) were linked to the NDI. To optimize the potential to identify and validate deaths, and to exclude active pandemic years, the sample focused on individuals whose employer-sponsored insurance benefits ended during 2019 or 2022. Employee and dependent death information was captured solely from employer data feeds. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were analyzed (%, 95% confidence interval (CI)). Date concordance, based on NDI reported death date, was also evaluated.

RESULTS: Overall, 1,504 and 652 deaths were identified in the NDI and MSN respectively, during either 2019 or 2022. Specificity, PPV, and NPV were all high. Specificity, in this case the ability to correctly identify individuals that did not die, was very high (99.07% [98.73, 99.33]). PPV was 93.56% (91.39, 95.23), while NPV was 83.28% (82.25, 84.27). Sensitivity, the ability to comprehensively identify individuals who died, was modest at 40.56% (38.06, 43.09). Concordance around reported date of death was high, with 97.28% of MSN death dates +/- 1 day of the NDI reported date.

CONCLUSIONS: Although death information coming exclusively from employer data feeds may under report death, information captured - presence of death and specific date of death - are very robust. The inclusion of secondary death data source (e.g., Social Security Death Master file) may be considered to increase the comprehensive capture of death for RWE generation.

Code

MSR209

Topic

Study Approaches

Disease

No Additional Disease & Conditions/Specialized Treatment Areas