How do mortality data in the Healthcare Integrated Research Database compare to the gold standard?

Compared to the National Death Index, considered the gold standard of mortality data, the mortality data in the Healthcare Integrated Research Database (HIRD®) provides an accurate and more reliable assessment of death outcomes than any source alone.

Enterprise Analytics Core domain(s): Quality of care, oncology

Summary

Background

Mortality ascertainment is challenging in observational studies due to the lack of complete death data in administrative claims or electronic medical records. Though the National Death Index is considered the gold standard of sources in the US, it is time-lagged, costly, and burdensome to access. Carelon Research (formerly HealthCore) developed an algorithm to integrate death data from various sources, including Social Security Administration Death Master File (DMF), Center for Medicare and Medicaid Services (CMS), online obituary data, inpatient discharge status, termination reasons, and utilization management claims from Elevance Health. Researchers assessed the sensitivity and accuracy of the Carelon Research mortality algorithm and each death source compared to the NDI, which have not been evaluated to date.

Methods

This study utilizes an established NDI linkage for 40,690 advanced cancer patients in the HIRD between January 2010 and December 2018. Carelon Research -death date was assigned if at least one death date from any individual death source within the study period was identified. Similarly, the NDI-death date was determined through the established NDI linkage. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) metrics were calculated with a 95% confidence interval (CI). The gap (in days) between the NDI-death and Carelon Research -death dates was assessed.

Results

Among 40,690 advanced cancer patients disenrolled from their health plan, the NDI linkage identified deaths for 27,602 (68%). Dates of death between NDI and HIRD differed by an average of 1.83 days (1.25, 2.40). Among all patients who died according to the NDI, 89%, 95%CI (88.4%, 89.2%) were captured by the Carelon Research -integrated death, i.e., sensitivity. Among all death cases captured by the Carelon Research -integrated death, 93%, 95%CI (93.0%, 94.0%) were actual death, i.e., PPV (Figure 1).
When comparing each death data source, on average, each source identified less than 50% of deaths events; however, their accuracy was similar to the Carelon Research -integrated dates of death, i.e., 93% (data not shown).

Figure 1. Annual Validation Metrics of the Healthcare Integrated Death Data in the HIRD 2010-2018*
Mortality Study Figure 1.jpg

*NDI data for 2018 were incomplete.

Key takeaways

Our study demonstrates that integrating death data from multiple data sources provides an accurate and more reliable assessment of death dates than any source alone. These data are essential in many critical therapeutic areas, such as oncology, liver failure, and heart disease, where having the correct information on the date of death is needed for drug safety and effectiveness research.

Publication

Aziza Jamal-Allial, Todd Sponholtz, Anahit Papazian, Shiva K. Vojjala, Biruk Eshete, Seyed Hamidreza Mahmoudpour, Patrice Verpillat, Daniel C. Beachler. Evaluation of Mortality data sources compared to the National Death Index in the HealthCore Integrated Research Database (2022). Pharmacoepidemand Dr S. 31(S2): 3-628 [#206 P86].

Carelon Research Project Team: Daniel Beachler, Biruk Eshete, Aziza Jamal-Allial, Anahit Papazian, Todd Sponholtz, Shiva Vojjala

For more information on a specific study or to connect with the Actionable Insights Committee, contact us at [email protected].

This study was conducted by Carelon Research, a subsidiary of Elevance Health. Dissemination and sharing of the Newsletter is limited to Elevance Health and its subsidiaries and included findings and implications are for Elevance Health and its affiliates’ internal use only.

Ready to get started? Sign up now!

Highlighted Studies

A real-world study of immunoglobulin (IG) reveals multiple cost-saving opportunities

Hub Domain(s): Immunology, oncology, costs of care, IngenioRx, policy guidance
Summary:

Background: Immunoglobulins (IG) are costly blood-derived products accounting for more than $500m in annual spending in the Anthem membership
Methods: HealthCore examined IG utilization using claims data to evaluate appropriateness of use, the extent of site-of-care optimization and non-preferred IG use.
Results: The findings revealed significant off-label use of IGs (30%), including 22% for indications that were neither approved by FDA nor covered by Anthem, but were accepted by other health plans; and 8% for indications that were neither approved by FDA nor covered by any health plan

Additionally, 25% of the IG use was observed in the most expensive site of administration of hospital outpatient (HOPD) and non-preferred IG use accounted for 71% of the total IG use

Key Takeaways: Examining the IG uses from various dimensions presents several opportunities to IngenioRx and Anthem:

Reducing off-label IG use represented an estimated $27.8m annual savings opportunity
Shifting IG utilization from HOPD to a less expensive site of care could drive a further estimated $6.3m in annual savings
The study also identified opportunities to increase the share of preferred IG products used by our patients.

Major depressive disorder (MDD) patients highly impacted by COVID-19 face daunting challenges

Hub Domain(s): COVID, member experience
Summary:

Background: Disruption to daily life and lockdown measures that were imposed in an attempt to contain the spread of COVID-19 presented new hurdles for at-risk populations, such as patients with major depressive disorder (MDD)
Methods: Conducted a 2020 internet survey to understand reasons for antidepressant continuation, switching and discontinuation in patients with MDD; COVID-19 pandemic questions were developed and added to survey.

COVID-19 Pandemic Impact (CPI) score was calculated and, based on their CPI scores, patients were categorized as experiencing low (LPI), medium (MPI) or high pandemic impact (HPI).

Results: Compared with LPI and MPI patients, HPI patients reported:

Significantly more problems accessing healthcare; few reported not having health insurance benefits due to loss of/reduced health insurance coverage (Figure 1).
More challenges obtaining depression prescription medications in terms of:

Filling new/current prescription(s)
Receiving samples from healthcare providers
Paying for prescription(s) due to loss of/reduced health insurance coverage and/or pharmacy benefits

Since start of COVID-19 pandemic engaged in significantly more:

Alcohol, drug, and tobacco use
Unhealthy mindless eating
Sedentary behavior

Key Takeaways: Our findings are consistent with literature showing that traumatic events, e.g. pandemics, can negatively impact health behaviors.

More research is needed to explore the COVID-19 impact in other patient populations.
Education opportunities exist for health plans to increase awareness among providers and patients of possible short and long-term COVID-19 effects on health behaviors.

Figure 1: Impact of COVID-19 pandemic on accessing healthcare/mental healthcare

Publications: Poster presented at the Virtual American Society of Clinical Psychopharmacology (ASCP) Annual Meeting, June 1-4, 2021

Consectetur adipiscing elit...

Joanna C.

"Et harum quidem rerum facilis est et expedita distinctio!"

Stanley T.

"Nam libero tempore, cum soluta nobis est eligendi."

Danielle W.

"Temporibus autem quibusdam et aut officiis debitis!"

Actionable Insights Newsletter
Translating research to power whole health

Issue 4 | November 2022