State law should be amended to create an Investigations and Enforcement Office within the Health and Human Services Commission, and require the commission to use advanced computer software to identify and deter provider fraud.
In 1995, the Texas Senate Committee on Health and Human Services and the Health and Human Services Subcommittee of the House Appropriations Committee held hearings on alleged fraud by dentists and others serving Medicaid-eligible children. Although the vast majority of Texas' Medicaid providers are honest and ethical, news reports have recounted stories of children receiving services they did not need and of Medicaid paying for services that were never delivered.
Medicaid fraud cases prosecuted in Texas include:
- a psychotherapist who billed for more than 24 hours worth of sessions in a single day;
- a radiologist who billed for multiple x-rays but performed single x-rays;
- a radiologist who added lab and x-ray services to the original request of the referring physician;
- a dentist who submitted falsified claims for services that had already been paid to, and delivered by, another provider;
- an adult day care center that billed for services rendered to patients who were not present;
- a hearing aid vendor who billed for new hearing aids after delivering used hearing aids;
- a pharmacist who billed for drugs never dispensed, submitted claims for brand names when generic drugs were dispensed, and indicated a larger quantity of drugs than was actually provided;
- a physician who billed for office visits for an entire family when the doctor actually saw a single member of the family; and,
- an agency that billed for home health services for patients who were in the hospital.
The $9.6 billion in state and federal funds Texas spent on Medicaid in fiscal 1996--almost a quarter of the entire state budget--represents a 128 percent increase over the $4.2 billion spent only five years ago. Although caseload growth and inflation in medical services contributed to the growth in program spending, caseloads grew by only 44 percent over the same period. Texas processed more than 26.8 million Medicaid acute care claims in 1995, an average of 550,000 claims a week from more than 121,000 Medicaid providers.
While no one knows the exact amount of losses due to health care fraud, national health care industry experts estimate these losses at 5 percent per year. The U.S. General Accounting Office estimates that losses may be as high as 10 percent. With $6 billion in state and federal funds spent on acute care medical services and another $1.3 billion spent on nursing home services, Texas' potential losses due to Medicaid fraud may be between $365 million and $730 million annually.
Texas also faces losses due to client error or fraud in its Aid to Families with Dependent Children (AFDC) and food stamp programs. In fiscal 1995, Texas spent $523 million on AFDC and received nearly $2.3 billion in federal food stamp benefits. As much as $222.4 million was spent in error in 1995; more than 36 percent of this erroneous spending was attributable to recipient fraud.
Texas welfare recipients have been caught:
- falsifying employment information;
- receiving duplicate benefits under different names and addresses;
- falsifying the number of dependent children in the household or other information about family composition;
- selling their food stamp benefits for cash;
- falsifying information on income or bank account assets; and,
- receiving welfare and other benefits under false names.
Controlling Medicaid fraud
Federal law requires each state participating in the Medicaid program to designate a single agency to oversee the state's program and to direct all related policy. In Texas, that agency is the Health and Human Services Commission (HHSC).
HHSC is responsible for ensuring the integrity of the Texas Medicaid program. Toward that end, HHSC executed a memorandum of understanding with the Texas Department of Health (TDH) to move the Medicaid Provider Sanctions Division from TDH to HHSC in February 1996. Because HHSC oversees all health and human services programs but operates none, it is in a unique position to focus on efforts to eliminate fraud and abuse.
HHSC's Sanctions Division, TDH, and the Office of the Attorney General's (OAG's) Medicaid Fraud Control Unit (MFCU) and Elder Law and Public Health (Elder Law) Division all play a role in Texas' Medicaid fraud control and enforcement efforts. The Sanctions Division, staffed by eight full-time equivalent employees (FTEs), has primary responsibility for the detection, investigation, and sanction of Medicaid provider abuse. When the Sanctions Division suspects provider fraud, it refers the case to the MFCU for criminal investigation and possible prosecution. The Sanctions Division is also responsible for referring cases to Elder Law for civil prosecution. The Sanctions Division spent $294,729 to carry out these responsibilities in fiscal 1995 and is expected to spend approximately $467,000 in fiscal 1996.
TDH is responsible for all Medicaid purchased health care or acute care services, such as hospital, physician, dentist, ambulance, maternity, and early periodic screening diagnosis and treatment services.
Several TDH sections perform tasks that could, if used properly, enhance Texas' efforts to control Medicaid fraud. The Bureau of Statistics and Analysis operates the Texas Computerized Medicaid Claims Processing Assessment System, and its Policy and Analysis group, with six FTEs, oversees the state's Medicaid claims administrator's contract. TDH's Medicaid Vendor Drug Bureau uses a system of automated edits and audits and drug utilization review to ensure appropriate use of prescriptions and to prevent fraud and abuse in the Vendor Drug Program. TDH's Medical Appeals Section conducts desk reviews of specific claims in response to appeals and/or adjustments requested by medical professionals and others who bill the Texas Medicaid program. In addition, the Medical Appeals section makes medical necessity determinations related to long-term care services.
None of these sections receives fraud detection training; they averaged less than five referrals each to the Sanctions Division in fiscal 1996.
Controlling welfare fraud
The Texas Department of Human Services' (DHS) Office of Inspector General (OIG) oversees Texas' AFDC and food stamp fraud control efforts. In fiscal 1995, OIG had a budget of $7.9 million and a staff of 210. While these numbers declined somewhat for fiscal 1996, OIG's budget remains significant at $7.2 million and an authority for 182 FTEs, 132 investigative and 50 audit staff. In fiscal 1996, OIG received 44 percent of its funds from the state and the remaining 56 percent from the federal government.
Reviews of OIG operations by TPR in 1993, by TDHS' Internal Audit Division in 1995 and by an independent consultant in June 1996, found that OIG staff receive little training, were subject to counterproductive performance measures, and receive little attention for their ideas or concerns. The latest report calls fraud detection training for investigators "minimal," "sporadic," and "not tailored to the various levels of experience and different backgrounds of the staff." Furthermore, it states that current performance measures steer staff away from higher-dollar fraud cases, discourage teamwork, and make poor use of investigator and auditor experience or skills.
Part of the problem stems from OIG's current organizational structure. While overall staffing for OIG declined in the past year, the Dallas, Houston, and San Antonio offices retained two regional supervisor levels. Under the present structure, field investigators and auditors working on a project report to separate supervisors, causing confusion and conflict. Another problem is that performance measures place emphasis on quantity rather than quality of investigations performed.
Current fraud detection tools inadequate
To be effective, any detection system must be flexible and powerful enough to keep up with the dynamic nature of fraud.
None of the current tools available to health and human services agencies to control fraud are adequate. The two primary tools are the Computerized Medicaid Claims Processing Assessment System (COMPAS) and the Surveillance Utilization Review Subsystem (SURS). All of the agencies involved in recovering inappropriately paid Medicaid funds or in fraud control use SURS to some degree to research specific claims or allegations of provider misappropriation. The federal government intended SURS to be a major tool in each state's Medicaid fraud control arsenal. TDH, through its contract with National Heritage Insurance Company (NHIC), the state's Medicaid claims administrator, produces federally required SURS reports that retrospectively analyze six to 12 months of the medical services delivered by Medicaid providers.
While SURS profiles providers to identify suspicious provider activity, COMPAS reviews each individual claim paid by NHIC and helps TDH's Bureau of Statistics and Analysis (BSA) evaluate the appropriateness of payments. Once COMPAS identifies an inappropriate payment, BSA can take corrective action to recover the payment made in error.
Because COMPAS takes a "snapshot in time," BSA cannot use it to identify fraudulent or abusive billing patterns. BSA must look at each transaction individually and determine whether a simple billing error exists or something more is involved. When BSA suspects errors due to fraud or abuse, it refers those claims to DHS' Utilization and Assessment Review Division and the Sanctions Division. In fiscal 1996, however, BSA made no referrals to the Sanctions Division.
Providers report their medical specialty and type to the Medicaid program. By federal mandate, 0.5 percent of all providers making the "ranking exception list" must be reviewed. SURS weighs each provider's claims information against his or her peers-- all the other providers who reported themselves to be of the same specialty and type--and ranks providers according to their "weights."
TPR found several instances in which reviews showed that doctors made the ranking list because the medical specialty they reported caused them to stand out from their peers, resulting in a "false positive" by SURS. For example, a surgeon who reported himself as a general practitioner would accumulate a lot of weight because, compared to other general practitioners, he performs many surgeries and conducts many in-patient hospital visits.
SURS sometimes groups different specialties and subspecialties together. For instance, pediatric dentists and general dentists are grouped together, as are pediatric oncologists (doctors who care for children with cancer) and adult oncologists. These groups may practice very differently from one another. In addition, TPR found that providers with low volumes of services may accrue high weights if the few services they provide fall outside their peer group norm.
DHS uses several fraud-control tools. In October 1995, OIG began testing a new Recovery Redesign system that allows on-line referrals from the eligibility worker to the Recovery Unit or OIG. The Recovery Unit handles unintentional violations of program rules or agency errors; OIG handles fraud cases. This system is operating only in Houston, San Antonio, and Tyler. In those locations, the eligibility worker enters the referral in the Recovery Redesign system and electronically forwards the referral with recipient data and benefit information to the appropriate DHS division. Further testing is needed before the system can be used statewide.
OIG uses the Accounts Receivable and Tracking System (ARTS) to help track the status of claims established. By tracking claims established, client notices issued, and cash and coupon benefit recoupment, ARTS should help reduce investigative work. It also maintains three years of information on any uncollected balance from former recipients in case they return to the program. ARTS does not, however, make referrals or otherwise help identify suspicious activities.
In addition, OIG works with the U.S. Department of Agriculture and the Comptroller's office to analyze electronic benefits transfer data to identify patterns of inappropriate food stamp redemption or violation of other state and federal law.
New tools developed
Neural network technology has been used successfully in the private sector to deter and detect fraud. In 1992, Visa International lost a staggering $719 million to credit card fraud. Shortly thereafter, Visa contracted with three software developers using different technologies to design automated fraud detection systems. Each system was given a sample of credit card transactions that included both good and fraudulent transactions. The expert system, a rule-based system that uses experts to identify the rules, correctly identified about 6 percent of the fraud; a statistical system identified about 9 percent; and the neural network system, about 18 percent.
As a "learning" system, the neural network improved over time. By January 1994, this system, called the Cardholder Risk Identification System (CRIS), correctly identified 45 percent of the fraudulent transactions. In 1995, Visa experienced a 24.7 percent reduction in fraud, due in large part to its CRIS system.
The Internal Revenue Service (IRS) and the federal Health Care Financing Administration (HCFA) are looking at this technology to solve fraud-related problems. Both contracted with Los Alamos National Laboratory to employ neural networks and other learning technologies to attack fraud. IRS will use this technology to reduce electronic filing fraud, while HCFA plans to use it to reduce Medicare fraud.
Comptroller's Medicaid fraud detection project
In November 1995, following a recommendation in TPR's Against the Grain, the Comptroller enlisted the aid of the Intelligent Technologies Corporation (ITC), an Austin-based technology firm. Principals in this firm helped develop Visa's CRIS.
With help from the Comptroller, HHSC, TDH, NHIC, and OAG, ITC developed a software system using neural network technology to detect fraud, abuse, and waste in Texas' Medicaid program. Such systems tackle complex problems the way the human brain does--by absorbing information, identifying patterns, and learning how to solve them. The more data and the more examples of "good" and "bad" claims, the better a neural network performs, because it "learns" by example.
The Medicaid Fraud Detection (MFD) project used two years of Medicaid claims and client and provider data to establish patterns used to identify suspicious providers. Data obtained from NHIC and other agencies were used to create the MFD database. While all parties worked aggressively to meet the MFD project's deadlines for data retrieval and transfer, timely data were not always available.
Florida's Public Integrity Unit, equivalent to the Sanctions Division of HHSC, has access to three years of data. This enables Florida investigators to produce ad-hoc reports and review database information. In the case of the MFD project, access to information in-house would enable Texas investigators to review provider information as the system produces it.
The MFD system does three critical things the existing systems cannot. First, it can look at many Medicaid claim details simultaneously to identify subtleties in the data that are not evident when looking at one piece of information at a time. Second, it can combine and analyze data from different databases and apply that information to the problem. Lastly and perhaps most importantly, it generalizes from its previous "learning experience" so that it can identify new schemes as they appear.
With adequate data, the MFD system identified unusual billing patterns that can be tied to a specific set of claims, providers, or recipients. Once the system began identifying these patterns, the agencies involved in the project formed a team to validate suspicious Medicaid provider profiles.
In July 1996, ITC began a series of baseline tests of the neural network software against SURS to compare the capability of each system to correctly identify suspicious provider activity. The Comptroller's staff identified a random sample of providers identified by the MFD system and took a comparable sample from the SURS exception list. In a blind test, the team assigned to review profiles from the two sources determined whether the profiles warranted further investigation.
To date, almost 39 percent of profiles from the MFD system were determined to warrant further investigation, compared to 14 percent of profiles from the SURS system. MFD system profiles were referred to the Sanctions Division for further investigation. The Sanctions Division then teamed with the Comptroller's office and TDH to perform field investigations of the providers identified by the MFD system. These investigations are ongoing.
House Bill 1863, enacted in 1995, requires fingerprint imaging as a condition of eligibility for AFDC and food stamp benefits. Fingerprint imaging is designed to prevent unqualified applicants from receiving duplicate benefits. TDHS is on schedule to begin operating the Lone Star Image System with two fingerprint imaging pilot projects in Guadalupe and Bexar counties on October 23, 1996.
One commercially available software allows banks to authenticate the identities of their customers. It performs a match against the name and social security number, and against thousands of variations on that name and number. The Federal Bureau of Investigation uses a similar tool to check identities. Other commercial software allows companies to check the legitimacy of addresses. This kind of information can be used to ensure that only eligible applicants receive welfare benefits.
Lack of a coordinated strategy
TDH's primary responsibility is to improve the health of Texans. DHS' primary responsibility is to provide temporary assistance to Texans in need. Neither agency focuses on fraud detection or control.
To improve the health of Texans, TDH must ensure that adequate numbers of Medicaid providers participate in the Medicaid program. This may result in a reluctance to investigate medical professionals suspected of wrongdoing. TDHS faces a similar dilemma in that its mission is to help needy Texans rather than to prosecute or investigate recipients. HHSC, however, is charged with ensuring the integrity of Texas' health and human services programs.
As discussed elsewhere in this report, little coordination exists among the agencies involved in Texas' Medicaid fraud control effort. Coordination between OIG and the Medicaid program is virtually nonexistent, although many Medicaid fraud schemes require the participation of welfare recipients.
Several states, including New York, Oregon, and North Dakota, combine their investigative and enforcement functions under an umbrella agency and use their resources to investigate both Medicaid providers and welfare recipients. Texas' OIG conducts some Medicaid fraud investigations when the case involves food stamp or AFDC recipients.
Georgia includes a portion of collections in the method of finance for its OIG as an incentive to increase collections. About $1 million of the OIG's $6 million budget comes from recovered AFDC and food stamp program money. This represents about half of the $2 million in projected collections in fiscal 1997. Recovered money is used as the state match for the federal funding received by the program.
About half of the Sanctions Division's annual expenditures come from state general revenue and the other half from federal funds. Of all the money recovered by the Sanctions Division, only the investigative costs awarded to the state from the settlement of civil cases are used to fund the program. In fiscal 1995, recovered investigative costs totaled $24,482. Recovered Medicaid overpayments are used to pay other Medicaid service claims, and fines and penalties are deposited in the General Revenue Fund. In fiscal 1995, recovered Medicaid overpayments totaled nearly $1.4 million, and recovered fines and penalties totaled nearly $1.5 million.
Slightly more than half of OIG's expenditures come from federal funds and the remainder from state general revenue. Currently, none of the money recovered by OIG is used to fund its operations.
A. State law should be amended to direct the Health and Human Services Commission (HHSC) to begin using learning or neural network technology statewide to identify and deter fraud in Texas' Medicaid program no later than September 1, 1997.
The neural network MFD system will significantly improve the quality of cases referred for enforcement and prosecution of fraud and abuse. Improved coordination in the referral process will ensure that the cases are routed to the most appropriate program for enforcement.
HHSC's Sanctions Division should handle the cases that most lend themselves to administrative penalties. Cases with the greatest potential for civil monetary penalty recoveries should be referred to the Elder Law Division of the Office of the Attorney General (OAG) for civil prosecution. Cases for which criminal prosecution is most appropriate should be referred to the OAG's Medicaid Fraud Control Unit.
HHSC should have two years of data accessible in-house to facilitate its investigations. The "learning system" should be able to stand alone to protect the system's integrity.
Should HHSC fail to meet the implementation date in this recommendation, state law should authorize the Comptroller's office to contract for and oversee operation of learning or neural network technology to be used to detect Medicaid fraud. In addition, the law should require that HHSC and the Comptroller execute a memorandum of understanding to ensure the Comptroller receive all necessary data and resources needed to operate the system.
B. State law should be amended to direct HHSC to use appropriate technology, including authentication software and other technologies, to address recipient fraud.
HHSC should explore commercially available authentication technology and other tools to prevent ineligible recipients from receiving welfare benefits.
C. State law should be amended to transfer the Texas Department of Human Services' Office of the Inspector General (OIG) and the Texas Department of Health's (TDH) Policy and Analysis group to HHSC on September 1, 1997, combine them with HHSC's Sanctions Division, and rename the new office the Investigations and Enforcement Office.
The combined budget for the new Investigations and Enforcement Office would be $6.3 million with 147 full-time equivalent employees on staff. Funding for the staff should move from their respective agencies to HHSC on September 1, 1997. Combining OIG and the Policy and Analysis group with the Sanctions Division would bring more resources to the task of fraud detection, investigation, and enforcement. More efficient collection of AFDC and food stamp overpayments would free investigative resources for Medicaid provider cases involving significantly more state funds per case.
HHSC should prioritize the work of the Investigations and Enforcement Office to maximize recovery of funds and opportunities for referrals to OAG. This transfer of resources requires leadership committed to making the best use of resources and staff to ensure effective enforcement.
Should HHSC fail to meet the implementation date in this recommendation, state law should authorize the Comptroller's office to contract for or conduct Medicaid investigations with comptroller or other designated staff. In addition, the law should require that HHSC and Comptroller execute a memorandum of understanding to ensure the Comptroller have access to all necessary health and human services staff for the purpose of conducting such investigations
D. HHSC should develop clear objectives for the Investigations and Enforcement Office that emphasize aggressive recovery of funds and sanctions for fraudulent and abusive providers.
HHSC should establish and apply strong performance standards to the coordinated investigative efforts to maximize recoveries. Emphasis should be placed on teamwork and on working cases with the strongest supportive evidence and the greatest potential for recoveries.
E. State law should direct HHSC to provide TDH's Vendor Drug Bureau, Medical Appeals Section, and any contractor processing Medicaid claims with fraud and abuse detection training.
Training should be provided at least annually. Individuals in these divisions should be encouraged to refer suspicious activities to HHSC's Investigations and Enforcement Office.
F. HHSC should cross-train Investigations and Enforcement Office investigators to enable them to pursue priority Medicaid and welfare fraud and abuse cases as needed.
Moving OIG staff to HHSC would give HHSC experienced field staff. By cross-training staff, HHSC should be better able to prioritize its workload to ensure that the most promising cases are worked.
G. State law should be amended to allow part of the recoveries made by the Investigations and Enforcement Office to be used in HHSC's method of finance to fund fraud and abuse control activities.
This should serve as another incentive for the Investigations and Enforcement Office to increase recoveries.
Improved quality of cases and better coordinated referrals would dramatically increase the amount of money recovered from fraudulent and other inappropriate claims payments. These recommendations would increase investigative resources needed to develop the cases for the recovery of funds.
This estimate of savings has been reduced to reflect the increased funding required to implement the neural network MFD system statewide. During fiscal 1998-99, $4.6 million in funding is recommended to pay for the equipment and support costs associated with operating the system. This estimate assumes that the commitment of 90 percent federal funding for program implementation will be received in fiscal 1998 and 75 percent federal funding will be received in fiscal 1999.
An additional $2 million per year is recommended to fund additional costs for investigation and administrative processing of the neural network cases. This estimate assumes that 50 percent of these costs will be paid with federal funds. The additional $2 million per year would be available to pay for the automated and investigative tools, such as criminal background checks, recommended elsewhere in this report, and the use of technological tools to identify fraud and abuse.
The fiscal 1997 estimate is based on the cases identified from a sample of providers used to "train" the neural network system. This sample represents about 30 percent of all providers. The fiscal 1998 estimate is based on applying the neural network system to all Medicaid medical care services under TDH's jurisdiction. In fiscal 1999, Medicaid long-term care services would be added to the neural network program. All Medicaid services would be added to the program in future years; however, the impact of adding all other programs is not included in this estimate.
To achieve certifiable savings identified in this issue paper, recommendations A and C must be implemented by September 1, 1997 and the state's Medicaid appropriations should be reduced by these amounts in the fiscal 1998-1999 biennium.
Savings to the
General Revenue Fund
1997 $1,034,000 $1,688,000 1998 18,347,000 28,871,000 1999 38,911,000 63,330,000 2000 61,747,000 101,376,000 2001 71,509,000 117,304,000 2002 81,948,000 134,335,000
 Jim Morris, "Hearing to Probe Medicaid Dentists," Houston Chronicle (October 6, 1995).
 Fax communications from Beth Taylor, director, Medicaid Fraud Control Unit, Office of the Attorney General, Austin, Texas, December 21, 1995 and January 15, 1996.
 Texas Health and Human Services Commission, "Medicaid Expenditure Information FFY 1970-1997" (Austin, Texas, May 15, 1996); and Texas Department of Health, "Average Monthly Medicaid Eligibles" (Austin, Texas, May 29, 1996). (Computer printouts.)
 Interview with Pauline Starr, R.N., manager, Health Services, National Heritage Insurance Company, Austin, Texas, January 17, 1996; and National Heritage Insurance Company, "Total Number of Providers Per Provider Type," Austin, Texas, July 26, 1996. (Computer printout.)
 Senate Special Committee on Aging, Gaming the Health Care System, Investigative Staff Report, by Senator William S. Cohen, Ranking Minority Member, Senate Special Committee on Aging (Washington, D.C., July 7, 1994), p. 1.
 Texas Department of Human Services, "Breakout by Quarter of HCFA-37 by Agency, Total Agencies," Austin, Texas, May 15, 1996. (Computer printout.)
 Texas Department of Human Services, Annual Report, FY 1995 (Austin, Texas), p. 73.
 Texas Department of Human Services, "Texas Food Stamp Corrective Action Plan" (Austin, Texas, May 1, 1996); and "AFDC 1995 State Corrective Action Plan" (Austin, Texas, November 9, 1995).
 V.T.C.A., Government Code SS531.
 Texas Department of Health, "Sanctions Division: Fraud and Abuse" (Austin, Texas, February 27, 1996).
 Fax communication from Sharon Thompson, director, Sanctions Division, Texas Health and Human Services Commission, Austin, Texas, August 9, 1996.
 Interview with Steve Scarborough, chief, Bureau of Statistics and Analysis, Texas Department of Health, Austin, Texas, September 30, 1996.
 Interview with Ron Borg, director of governmental affairs, and Bill Whalen, deputy inspector general, Texas Department of Human Services, Austin, Texas, September 18, 1996.
 Interview with Bill Whalen, deputy inspector general for investigations, Office of Inspector General, Texas Department of Human Services, Austin, Texas, July 8, 1996.
 Texas Performance Review, DHS Office of the Inspector General: Performance Review (Austin, Texas, February 8, 1993); Texas Department of Human Services, Audit Report Performance Review Office of the Inspector General Management Policies and Practices (Austin, Texas, February 1995); and Texas Department of Human Services, Internal Audit Implementation by the Office of Inspector General, by Ray Associates, Inc. (Austin, Texas, June 1996) (consultant's report).
 Texas Department of Human Services, Internal Audit Implementation by the Office of Inspector General.
 Texas Department of Health, "Statistics and Analysis" (Austin, Texas, February 27, 1996). (Computer printout.)
 Interview with Allan Trosclair, vice president, Fraud Control, Visa U.S.A., San Mateo, California, October 3, 1996.
 "Tax Refund Fraud Could Cost the U.S. Billions, Special IRS Report Indicates," Wall Street Journal (February 11, 1994).
 Baseline test on the performance on the MFD system conducted by Intelligent Technologies Corporation for the Texas Comptroller of Public Accounts, Austin, Texas, August 1996.
 Interview with John Hunsucker, director, Office of Fraud and Abuse, Georgia Department of Human Resources, Atlanta, Georgia, September 24, 1996.
|Texas Comptroller of Public Accounts|| Window on State Government|
Privacy and Security Policy