Medicare Paying for Surgeries for Dead Patients?

The Centers for Medicare and Medicaid Services (CMS) paid millions of dollars in benefits for dead Americans between 2004 and 2008 according to a new study by PearlDiver Technologies, Inc.

PearlDiver’s analysts found that between 2004 and 2008, CMS paid for 142, 000 procedures by 2, 119 hospitals or clinics on approximately 4, 980 dead patients. Eventually, CMS paid $33 million for those claims.

The type of orthopedic procedures performed on patients coded as “dead” included arthrocentesis aspiration of a major joint, total knee replacements, total hip replacements, knee arthroscopy surgery, fracture repair, shoulder surgeries and spine surgeries.

This isn’t the first time that the Centers for Medicare and Medicaid Services have been found to pay for the dead.

Paying Dead Doctors

In July 2008 the Senate Permanent Subcommittee on Investigations reported that Medicare paid tens of millions of dollars to suppliers improperly using identification numbers of doctors who died years earlier.

According to news reports, the subcommittee’s 2008 report stated: “From 2000 to 2007, Medicare paid 478, 500 claims containing identification numbers that were assigned to deceased physicians…. The total amount paid for these claims is estimated to be between $60 million and $92 million. These claims contained identification numbers for an estimated 16, 548 to 18, 240 deceased physicians.”

In 16% of these cases, the report said suppliers used identification numbers of doctors who had been dead for more than ten years. In one case, Medicare paid more than 2, 000 claims totaling $479, 000 for services provided from 2002 to 2007, even though the doctor had died in 1999.

Another doctor died in 2001, but his identification number was used in more than 3, 800 claims from 2002 to 2007, with payments totaling more than $354, 000.

Herb Kuhn, deputy administrator of CMS, responded to the subcommittee’s report by, first, acknowledging the problem and then laying out CMS’s plan for correcting the risk of improper ID numbers. He testified:

“With increasing expenditures, expanding Federal benefits, and a growing beneficiary population, the importance and the challenges of safeguarding CMS programs are greater than ever. Fraud, waste, and abuse schemes have become increasingly complex, and are quick to adapt and stump even the latest oversight strategies of Congress, CMS, and our law enforcement partners.”

CMS has, in fact, reduced the rate of fee-for-service error rates. For example, in FY 2005 CMS set as its goal a medical documentation error rate of 7.9% and actually achieved an error rate of 5.2%. In FY 2006, the goal was 5.1% and the actual error rate was 4.4%.

In 2008 more than one billion claims from providers, physicians and suppliers for products and services were sent to Medicare for processing. So, it’s really no surprise that errors would occur.

Fixing the Dead Doctor Dilemma

As a result of the OIG audit, CMS set about updating, correcting and, where necessary, deactivating provider ID numbers with invalid addresses and/or no claims activity for one year. In addition, CMS subcontracted with the American Medical Association (AMA) to obtain provider data extract files which contain physician dates of death on a bi-weekly basis. On a monthly basis, CMS claims payment contractors were sent a deceased provider identification numbers list and notified to update their physician records.

Five years after the OIG audit, in October 2006, CMS initiated a systematic deactivation of inactive physician ID codes. All told, by 2008 CMS had deactivated 1.5 million provider ID numbers.

Finally, CMS agreed to an information exchange agreement with Social Security Administration (SSA) on July 1, 2008 which provided CMS with monthly updates of the SSA’s Death Master File and unrestricted state death data. 

In theory, with the SSA death data, CMS is supposed to be able to match deaths with their own provider enrollment database, provider enrollment, chain and ownership system.

But somehow, it appears, CMS forgot to look for dead patients.

But Wait, Even Social Security is Paying Dead People

Ironically, while Medicare was setting up a system to use SSA’s Death Master File to correct its death records, a new problem was emerging with SSA’s own databases. 

Earlier this summer the Office of Inspector General reported that the Social Security Administration erroneously coded approximately 14, 000 Americans as dead.

When those errors occur it can have a devastating effect on living beneficiaries. As reported by CNN; “Laura Brooks, of Spotsylvania, Virginia, discovered she had been declared dead when she stopped receiving her disability checks, and her rent and student loan payments unexpectedly bounced. She went to her bank and a representative said her account had been closed because she was dead. Brooks, a 52-year-old mother of two, was already on permanent disability because of a severe depressive disorder, so hearing this turned her already difficult world completely upside down.”

In its report, the OIG said that such “erroneous death entries can lead to benefit termination, cause severe financial hardship and distress to affected individuals.” In Southern California and elsewhere last year, OIG investigators reviewed 305 Social Security beneficiaries who were recorded as deceased in their Social Security Administration files. At least 140 of them were still alive.

All told, investigators say, more than 6, 000 current Social Security beneficiaries are recorded as being deceased. An untold number of them are still, in fact, alive.

Staff members for Rep. Jim Costa’s office (D-California) said that their office handled about ten cases in the past four years in which the Social Security Administration incorrectly classified constituents as dead.

Social Security officials have agreed to investigate the correct status of 6, 733 potentially deceased individuals identified in the OIG investigation’s audit.

The OIG found that $2 million in improper payments were made to the 88 deceased Social Security beneficiaries. The investigators further identified 6, 733 Social Security benefit recipients whose master files “contained a date of death.” Extrapolating from their smaller sample, investigators estimated that more than $40 million may have been paid out improperly to deceased beneficiaries.

As a result of the new study, at least three dozen potential criminal cases have been forwarded to the agency’s Office of Investigations for further inquiry and possible prosecution.

Paying for Surgeries on Dead Patients

There are 1, 908, 435 patients coded as “dead” by healthcare providers and then appeared in PearlDiver’s CMS records from 2004 to 2008. Of those patients, according to the PearlDiver analysis, approximately 1 out of every 400 deceased Medicare patients had one or more further medical procedures. Every single patient included in this PearlDiver study received their follow-on surgeries three months or longer after their demise. So, for example, if a patient died in March, 2006 but then had a hip replacement 90 days or longer after death (say, in July, 2006), they were included in the PearlDiver study. 

How did PearlDiver discover these needles in the CMS data haystack?

In the words of its President, Benjamin Young (brother of the author): “PearlDiver employs Big Data third generation longitudinal tracking algorithms which allow it to mine the entire 2 billion record data base rapidly and efficiently. As a result, in a matter of minutes PearlDiver was able to mine the entire Medicare database and thereby discover payments to patients who had “died.” More typical methods of data mining which employ sampling techniques would not have found these anomalies. “

PearlDiver

PearlDiver is a four-year-old medical data aggregator which started with about 20 million private payer records and is today one of the largest medical data sources in the world with more than 2 billion U.S. patient records. Underlying PearlDiver’s data is a search and organization engine that allows researchers to mine entire databases. The ability to mine an entire database is also known as Big Data capability and it is extremely unusual. More typically, data mining companies employ sampling techniques to handle extremely large databases.

Because PearlDiver can rapidly mine virtually any size database, it can uncover such anomalies as, for example, payments for dead Medicare patients. A sampling approach, on the other hand, which examines perhaps 1 in 100 records, would miss an event that occurs in 1 in 400 records—which these payments did.

For clinical researchers, the ability to mine an entire database lets them examine patient outcomes, costs, complications, comorbidities and then, because the mining occurs rapidly, to “play” with the data by varying age, gender and really any other factor. Recently, researchers from UCLA used the PearlDiver data mining capabilities to do a study of adult scoliosis. The resulting paper won the North American Spine Society (NASS) Editor’s Choice Award as the best paper of 2010. The researchers will receive the award at the upcoming NASS annual meeting.

In many ways, the future of healthcare is all about improving ways that data is mined, interpreted and used to guide clinical and resource decisions. The curious fact that Medicare is paying for surgeries on dead people (or patients who’d been coded as “dead”) is a stark illustration that the current data mining infrastructure for medicine is not yet prepared for the complexities of a changing and increasingly constrained healthcare system.