Finding “Zombies” in Your Database by Confirming Vital Status


David K. O’Brien, PhD, GISP
Alaska Cancer Registry, Department of Health and Social Services, Anchorage, AK
Email: david.obrien@alaska.gov, phone: 907-344-5874

 

 

Objective

The Alaska Cancer Registry (ACR) conducted a study in 2015 to identify and correct the vital status of certain cases in its database. These cases were reported as deceased by the original healthcare facility but were not identified as being deceased using routine death resources. Cases incorrectly reported as deceased are referred to as “Zombies.”

Background

Cancer registry databases must have accurate vital status information on their cancer cases. This is important for:

  • Cancer survival analysis
  • Cancer control planning
  • Studies on the outcomes of cancer care
  • Patient follow-back & cancer research studies requiring patient contact

There are several ways cancer registries passively determine the vital status of their cancer cases. This involves linking their database with mortality databases:

  • State mortality database (SMD): Works well for current state residents
  • Social Security Death Index (SSDI): Works well for former state residents who moved to another state, but does not list the cause of death
  • National Death Index (NDI): Works well for former state residents who moved to another state, and includes the cause of death

However, not all deaths are discoverable using these 3 mortality databases due to various factors:

  • People who died out-of-country as a resident of another country will have no state death certificate (usually for foreign-born cases)
  • In-state deaths that were never reported to the state will have no state death certificate
  • Decedents who never qualified for social security will not be in the SSDI
  • Deaths reported late to the state after NDI closes its current death year will never be added to the NDI

Another way cancer registries receive vital status information is from the hospital/healthcare provider that originally reported the case to the cancer registry. The source of that information could be:

  • Copy of death certificate
  • Note from primary physician
  • Follow-up with primary physician
  • Obituary search
  • OR… a typographical error!

Cancer registry cases not linking to SMD, SSDI, or NDI fall into four categories:

  • Truly alive
  • Truly dead as indicated by additional information obtained by the healthcare provider
  • “Immortals”: Reported as alive but are really dead – forever living in the registry
  • “Zombies”: Reported as dead but are really alive – the undead!

Therefore, “Zombies” are cases that were improperly reported as deceased by the original healthcare source and are identified as being alive by consulting additional resources. ACR wanted to identify Zombie candidates in its database, research them, and correct the vital status of those cases that were determined to be true Zombies.

Methods

ACR links its cancer cases annually against the SMD, SSDI, and NDI to determine the current case vital status. After the annual SMD death review in the spring of 2015, ACR identified Zombie candidates by checked all its deceased cases that had never linked to SMD, SSDI, and NDI in the past. Zombie candidates were reassessed after linking with SSDI in the summer and NDI in the fall.

The remaining Zombie candidates were manually reviewed using the Alaska Permanent Fund Dividend (PFD) applicant database. If an individual continued to file PFD applications for several years after their date of death, then they were assumed to be still alive. The state PFD Division reviews all the applications for potential fraud, so the PFD database is considered to be highly reliable. If the individual was not in the PFD database, then they were looked up in the Alaska State Police (Troopers) database, which is linked to the Alaska Department of Motor Vehicles. If the individual had filed for a driver’s license after their date of death, they were assumed to be still alive.

Results

After the initial SMD linkage, 85 potential Zombie cases were discovered (0.2% of the total cases in the database). Of these, 44 (51.2%) were reported to ACR by other central cancer registries, most of them from other state health departments. After the SSDI linkage, 50 cases were verified as deceased, and after the NDI linkage, 15 more cases were verified as deceased. Only 20 potential Zombie cases remained for manual review.

After manual review of the 20 cases using the PFD and Troopers databases, 14 cases (70%) were determined to be true Zombies and verified as alive. Their vital status was changed to alive in the ACR database. The remaining 6 cases were verified as deceased:

  • 3 died out-of-country; death information was documented in the original report from the healthcare facility
  • 2 died in-state with no death certificate; death information was documented in the original report from the healthcare facility
  • 1 died out-of-state; death information verified by that state’s cancer registry

The study results are summarized in the following table:


Type of Case # of Cases
Potential Zombie cases discovered after death clearance 85 Total
     Resolved by SSDI linkage 50
     Further resolved by NDI linkage 15
     Potential Zombie cases remaining 20
Potential Zombie cases for manual review 20 Total
     True Zombie cases, confirmed alive, changed vital status 14
     Died out of country 3
     Died in state 2
     OOS source record 1

Discussion

ACR manually reviewed 20 cases for this study; 70% of them turned out to be true Zombie cases and had their vital status changed from dead to alive. But Alaska has one of the smallest annual cancer caseloads in the country at about 3,000 cases per year. The annual caseloads for most of the other states are much higher, especially for the most populous states of NY, TX, FL, and CA with over 100,000 cancer cases per year and potentially over 1,000 Zombie candidates. It could take a considerable effort to do this type of study in a state with a large population.

This is not just a problem specific to cancer registry databases. This also affects any other database that includes the vital status of a person, such as another type of healthcare database, law enforcement, or a financial institution. We have probably all heard stories of a person who couldn’t access their finances because the bank thought they had died. Other organizations would benefit from this type of review if the vital status of the people in their database is important for that organization.


The opinions expressed in this article are those of the authors and may not represent the official positions of NAACCR.

Share this:

Trackbacks

  1. … [Trackback]

    […] There you will find 33884 more Infos: news.naaccr.org/finding-zombies-in-your-database-by-confirming-vital-status/ […]

Leave a Reply