Prescription histories have long served as valuable evidence for assessing the mortality risk of insurance applicants. The introduction of electronic prescription histories now enables this information to be accessed in a more structured, verifiable, and cost-effective manner.
Idea in Brief
This article presents the mortality experience from a large block of insurance applicants for whom prescription drug histories were obtained at the time of application. Mortality differentials by drug severity classification and fill frequency provide key insights into the promise this tool offers for life insurance risk assessment.
Methods and Data
Milliman IntelliScript, a leading provider of prescription histories to the insurance industry, provided the data used for this mortality research study. For its IntelliScript product, Milliman gathers prescription histories on insurance applicants who have authorized the release of their records. A network of pharmacy benefit management companies is queried electronically and the results for the applicant are returned to the insurance carrier. The query may not find all of an individual’s prescriptions; however, for this study, some prescription information was known on over 70% of the applicants. The study data included information on 1.1 million applicants who had applied for insurance from 2005-2007. Over 21 million distinct prescription fills were included among these applicants.
Each applicant was matched by Social Security Number and validated by last name and birth date to the Social Security Administration’s March 2008 Death Master File (SDMF) to identify applicants who had died during the study period 2005-2007. Actuarial exposure by count was calculated for each applicant within each calendar year in the study period. Seriatim tabular mortality and actual-to-tabular ratios were calculated from the 2001 VBT Mortality table ultimate rates. Information on tobacco usage was not available. Relative mortality ratios were calculated by dividing the actual-to-tabular ratio for each category by the grand total actual-to-tabular ratio. The graphs throughout this document display these relative mortality ratios bounded by approximate 95% confidence intervals based on the number of deaths.
As a validation and alternate method of analysis, randomized control groups were generated and compared to the cohort of deceased applicants. The ratios of these distributions provide insights into the relative mortality differentials by prescription history classification similar to the mortality study analysis.
Read More +