Section II: Medicaid Study
Focus of the Texas Health Care Claims Study Medicaid Review
The purpose of this study is to measure the incidence of potential overpayments, which could include occurrences of fraud and abuse in the Texas Medicaid program. Section 403.028 of the Texas Government Code requires the Comptroller to perform this study biennially in consultation with the State Auditor’s Office.
An overpayment is a payment for a service that is not in accordance with the policies of the Medicaid program. The Medicaid definitions of fraud and abuse are: 
• Fraud means an intentional deception or misrepresentation made by a person with the knowledge that the deception could result in some unauthorized benefit to himself or some other person. It includes any act that constitutes fraud under applicable federal or state law.
• Abuse means provider practices that are inconsistent with sound fiscal, business or medical practices and result in an unnecessary cost to the Medicaid program, or in reimbursement for services that are not medically necessary or that fail to meet professionally recognized standards for health care. It also includes beneficiary practices that result in unnecessary cost to the Medicaid program.
With few exceptions, the study randomly sampled claims from Texas Medicaid clients receiving covered services from all the actively enrolled Medicaid providers. The sample was selected from paid claims with dates of service from September 1, 1999 through November 30, 1999. The providers were stratified into seven categories based on the type of services provided and their specialties within that service type. Those categories are:
Category Size Ancillary/OutpatientHome HealthHospitalsMental HealthOther (includes Dental)PhysiciansSupplies/Durable Medical Equipment (DME) 100100100100100100100 Total 700
(Detailed information on the methodology and sampling criteria are discussed in the Methodology section on page 16 and in Appendix J.)
In addition to hospitals (inpatient and outpatient) and physicians services, the study focused on some specific Medicaid-covered services that were considered areas of potential overpayments, fraud or abuse. Specifically, these services were the medical supplies, durable medical equipment, home health, and mental health care. The study’s intended goals were to measure an overall overpayment error rate, determine if the same type of errors found in the December 1998 study still prevailed and identify any new types of Medicaid overpayments.
The study reviewed a sample from three months of paid services totaling $632,687,878.03. Expenditures for 1999 totaled $3,063,080,008.38, which is 27 percent of the $11.3 billion overall Medicaid spending for fiscal 2000.
Client and Provider Exclusions
Before selecting the sample, some Medicaid eligible groups and provider types were excluded. Clients receiving services exclusively from Managed Care HMOs, the Long-Term Care program and the Children with Special Health Care Needs (CSHCN) program were excluded.
The Texas Commission for Alcohol and Drug Abuse (TCADA) clinics were excluded, because HHSC performs a 100 percent review on all the services they provide. The other provider exclusions were due to the client eligible groups they served, such as the long-term care and military hospitals and providers who supply medical equipment for only Medicare clients.
Potential Overpayment Measurement Calculation
The basic components of the overpayment measurement are the potential overpayments for each category of service and the total dollars paid in the sample for each category. The total dollars paid in the sample were adjusted by subtracting the dollars paid to providers under investigation.
The potential overpayment measurement was determined using a weighted average calculation that adjusted the sampled services to reflect their proportion in the three-month universe. The universe was also adjusted to exclude the same proportion of paid services delivered by providers under investigation that occurred in the sampled services. The dollar amount of the weighted average of potential overpayments was divided by the dollar amount of the weighted average of the total sampled dollars to derive a 7.24 percent potential overpayment measurement for the Medicaid acute medical care fee-for-services program. The margin of error is plus or minus three percent. A comprehensive explanation of the calculation is on page 34 and in Appendix L.
December 1998 Study Overview
The first study published in December 1998, consisted of 600 randomly selected sample days stratified into five categories of 100 days each and five categories of 20 days each. A Vendor Drug sample of 100 was deleted from the original work plan because of time and system constraints.
The review methodology included a client telephone interview and a contextual data analysis of the claim data in the Medicaid Management Information System (MMIS) at NHIC. A medical record was requested for review only if a medical necessity problem was suspected. The sample was pulled from the month of November 1997. The sample consisted of 1,043 claims with total payments amounting to $1,129,713. Table 1 lists the service areas sampled by category.
Table 1: December 1998 Texas Health Care Claims Study Sample Categories
Category Sample Days Category Sample Days Ancillary/Outpatient 100 Medical Supplies/DME 20 Home Health 100 Dental 20 Hospitals 100 Ambulance 20 Other/Other Health Professionals 100 Chiropractors 20 Physicians 100 Podiatrist 20
Source: Texas Comptroller of Public Accounts, Final Staff Draft Report on Health Care Claims Study and Comment from Affected State Agencies, December 1998.
January 2001 Study Medicaid Sample Improvements
Early in the 2001 study project, the Comptroller and State Auditor staff met with representatives from TDH and HHSC/OIE to evaluate and improve the study methodology used in the December 1998 study. One main concern was maintaining the integrity and validity of the study sample to ensure the accuracy of the statistical results. After consulting with professional statisticians in all these agencies, the following improvements were added.
• The Medicaid sample size was increased from 600 to 700 sample days.
• The sample time period was increased from one month to three months.
• The categories with a low volume of services and/or low incidence of overpayments were moved to the Other category.
• The Mental Health Category was pulled out of the Other category, and the quantity of services selected was increased from 20 to 100.
• A medical record review was added to the study review methodology. All the medical records for the sample days were requested from the providers paid for the services.
• The Vendor Drug services (Medicaid prescription services) are paid through the Vendor Drug Program at TDH. These were analyzed separately from the Medicaid study.
Medicaid Sample Selection
The TDH Claims Administrator contractor, NHIC, performed the sample selection. NHIC systems engineer staff specifically coded a program to randomly select patient days, by date of service, either from the services submitted by professional health care providers or from hospitals according to the length of stay during a specified range of dates. The services included 700 core services, plus 1,216 associated services. A core service is defined as the primary service delivered to a client.
NHIC selected the data for the seven service categories mentioned above from the state’s Medicaid Management Information System (MMIS) data. (Detailed information on the sample selection is in Appendix J.)
The final Medicaid Study sample selected consisted of 700 sample days with a total of 1,916 core and associated health care services. The total paid for the sample services, $181,483.48, represents 0.03 percent of the three month sample universe.
Providers under Investigation Exclusions
Once the sample was selected and before initiating the study reviews, any services delivered by Medicaid providers who were being investigated by the HHSC/OIE were excluded. Clients were only excluded from the sample if all the services on their sample date-of-service were performed by a provider(s) under investigation. A total of 307 claims for $51,421 were excluded from this study for 78 providers under investigation. This exclusion represents a 39 percent increase over the 56 providers excluded in the December 1998 study.
As noted earlier in this report, this study was performed in consultation with the State Auditor’s Office’s (SAO). The agency’s consultation services included reviewing the procedures used to generate the study’s sample services. This review entailed examining NHIC’s program code and supporting documentation and reconciling the monthly totals within the sample.
On March 27, 2000, the State Auditor’s Office said there “was no indication that the population used for the study sample did not include all claims paid to date for the sample months.”
(A copy of the agency’s letter is included in Appendix M, Exhibit 1.)
Medical Information Confidentiality
Both the Comptroller’s office and the SAO along with TDH and HHSC were concerned about protecting the confidentiality of information, include medical records. Some of the measures taken to ensure confidentiality were:
• all persons working with the information, consultants and study project team members, signed confidentiality statements (copy included in Appendix K, Exhibit 6),
• all medical records were requested and processed by HHSC/OIE staff,
• online information was reviewed at the TDH/NHIC building, and
• the medical records were secured in a locked file cabinet.
A Memorandum of Understanding (MOU) between the Comptroller’s Office and HHSC/OIE detailed all the guidelines and precautions that both agencies agreed to follow. (See Appendix J, Exhibit 4, for a copy of this MOU.)
At the study’s end, the review documentation, including the medical records, were returned to HHSC/OIE, which will use the documentation for further investigation and action.
Medicaid Review Methodology
The study used three different review analyses. Two methods, contextual data analysis (CDA) and client telephone interviews, were carried over from the December 1998 study. The other, medical record review was one of the improvements listed above. A summary of each is described in Table 2.
Table 2: Texas Health Care Claims Study – Medicaid Review Methods:
Type of Review Description Client Telephone Interview The Medicaid clients were contacted by phone to validate that they received the sample services on the sample date-of-service from the providers who billed the claim. Potential overpayments were only counted if the client stated the service did not occur. Contextual Data Analysis Review of claims data submitted by the provider for the client. This information was reviewed online using the Medicaid Management Information System (MMIS). Analysts reviewed each claim in relation to other claims paid 2 months prior and 1 month after the sample date-of-service to identify any potential billing trends that are known to result in overpayments, such as unbundling lab services (billing laboratory panel test components separately to increase their reimbursement.) Medical Record Review Medical records for the sample date-of-service were requested from the billing providers. Nurses reviewed these documents to validate that the sample service occurred on the sample date-of-service and was delivered by the provider who submitted the claim and was paid by National Heritage Insurance Company.
Client Telephone Interview Methodology
As noted above, the study used three types of review methods to validate the health care services paid by the Medicaid program. A summary of each follows. (Detailed information on the methods is in Appendix J and examples of the tools used for each method is in Appendix K.)
Client Telephone Interviews
The client telephone interview questionnaire consisted of 13 sections related to either the health care claim or client information. Each individual was asked questions from a minimum of two sections: the demographic information section and the specific section that corresponded to the type of service provided. If there were multiple services, the individual was asked questions from each section that applied. The interview sections were:
• Demographic information
• Physician visit
• Dental visit
• Visit to other health care professional
• Inpatient hospital stay
• Outpatient hospital visit/Clinic visit
• Emergency room visit
• Prescription drugs
• Laboratory and/or Radiology
• Medical supply/Durable medical equipment
• Home health
• Outpatient mental health
Two versions of the Medicaid interview tool were developed: one for adult patients and one for children under age 18. In cases where the patient was a child, the interviewer asked to speak to the parent or guardian, identified from the TDHS eligibility system (SAVERR).
Up to 10 attempts were made for each contact. The interviews were conducted in English, Spanish and Vietnamese as necessary. Clients were not required to participate in the interviews, and they were allowed to end the interview at any point. If clients said they could not remember a service, the interview was continued with the specific section appropriate for the health care service billed by the provider.
The interviewers used the client’s Social Security Number (SSN) to verify the client’s identity. The interview tool provided three opportunities for the client (or parent/guardian) to remember receiving the service:
• the client was asked if they went to a doctor (hospital, etc.) in the sample month,
• then the client was asked if they visited a doctor or received a service on the sample date, and,
• finally the client was asked if they saw the provider (name specified) on the sample date.
Contextual Data Analysis Review Methodology (CDA)
Research analysts conducted contextual data analyses of all services associated with the sample clients using the Medicaid Management Information System. They reviewed each client’s claim in relation to other claims paid during the time that spanned two months prior to and one month after the sample day, regardless of the provider.
The objective of this analysis was to note:
• any unusual provider billing or client utilization trends that might not be observed when reviewing only the sample claim
• any services that did not appear related to the diagnoses or procedure billed on the claim, and
• any services billed and paid that were not in accordance with the policies and/or procedures stated in the 1999 Texas Medicaid Provider Procedure Manual and/or the 1999 Texas Medicaid Bulletins.
The 1999 version was used because all the sample services occurred in 1999, and the providers were required to follow the policy and procedures in this manual and any Medicaid Bulletin received before the sample dates of service.
Sample services were identified as potential overpayments whenever the services did not appear related to the diagnosis or procedure billed, or the services did not comply with the policies and procedures in effect for that date-of-service. Any billing or client utilization patterns considered aberrant were referred to both TDH and HHSC/OIE for further investigation.
Medical Record Review Methodology
As noted earlier, medical record reviews were added as an improvement from the December 1998 study. In the prior study, one of the issues that evolved was confirming a CDA observation that could not definitively be considered a potential overpayment without reviewing medical record documentation. In most of the cases where the medical record was requested to clarify a potential discrepancy, the record was received after the project’s deadline for final conclusions.
At the beginning of the January 2001 study, the Comptroller’s office, SAO, TDH and HHSC/OIE developed a process for requesting medical records under the authority of federal and state regulations as well as the provider’s Medicaid agreement with TDH. This process included a three-phase request letter sent certified mail, a tracking system between the Comptroller’s study team and HHSC/OIE Study staff and follow-up telephone calls. The HHSC/OIE staff’s diligence in identifying correct physical addresses and contacting providers repeatedly to either remind them of the request or respond to questions, resulted in an extraordinary 96 percent return of records.
Medicaid Medical Record Request Letter
An initial certified letter was sent to each provider with an attachment identifying the client by name and date of birth, the sample date of service, the client’s Medicaid identification number and the number of the claim as it appeared on their Medicaid payment statement. In the case of multiple claims for different clients for the same provider, all of the claims were listed on one attachment. Also included, was an affidavit form for the provider to affirm that the record sent was the complete documentation for the client and date-of-service requested.
If the provider failed to return the appropriate documentation, a second certified letter was sent. If there was no response to the second certified letter, a third and final certified letter was sent, followed by a telephone call explaining what had been requested and the consequence of non-compliance.
All three of the letters had specified deadlines ranging from 14 to 7 calendar days, a list of the information requested and an explanation of the consequences for non-compliance. The consequence issued by HHSC/OIE was total recoupment of the claim amount from the provider because the Medicaid regulations require the providers to comply with medical record requests for paid services.
Medical Record Review Criteria
The study used the following guidelines to conduct the medical record reviews:
• The sample client’s name was documented on the record.
• The sample service billed and paid was documented on the record.
• The sample date of service (month, day, and year) was documented on the record.
• The performing provider’s signature was on the record.
• For lab and x-ray services, the results of the test were proof the service was performed.
• Durable medical equipment (DME) required prior authorization and a delivery receipt signed by the client or parent/guardian.
• Disposable supplies required a prior authorization.
General medical record documentation requirements by TDH/NHIC were published in the 1999 August-September Texas Medicaid Bulletin. This publication was available to all Texas Medicaid providers by September 6, 1999. None of the documentation errors found were for services prior to September 6, 1999.
Medicaid Review Results
Of the three review types, the medical record reviews proved to be the most significant in uncovering errors in payments and potential fraud. The contextual data analysis could identify lab unbundling and some potential upcoding of office visits.
The study team acquired 320 out of 607 valid telephone numbers, 53 percent of the total eligible clients to be contacted. An overview of the results from each study is summarized below. Detailed results are discussed in Appendix L. Included in the client telephone interview results below is a discussion of issues related to the Medicaid eligibility system.
Client Telephone Interview Review Results
Two factors, locating valid client telephone numbers and the time period from the date of service to the telephone interview (up to 11 months in some cases), diminished the value of this review method.
Of the 516 sample clients with valid telephone numbers, 12 refused to be interviewed and 154 could not be reached due to disconnected telephone numbers, or the client was not at the number identified for them. Seventeen of the clients did not answer their phones, and the family members of 12 clients stated the client was unavailable. Tables 3 and 4 illustrate the distribution of the contact results of the entire sample by the study categories.
Table 3: Results of the Client Telephone Contacts
Results of Client Telephone Interviews Number of Client Contacts Percent of TotalSample Clients Successfully completed interviews 320 53% Client deceased 1 0% Client refused to be interviewed 11 2% Client refused to complete the interview 1 0% Disconnected telephone number 96 16% Client not at the address or telephone number supplied for that client 58 9% No answer after repeated attempts 17 3% Unable to contact (most common reason was out-of-town) 12 2% Client had no telephone or no telephone number available for the client 91 15% Subtotal 607
Client not contacted – all services by provider under investigation 93
Table 4: Results of the Client Telephone Contacts for the Sample Categories
Interview Results Ancillary/Outpatient HomeHealth Hospitals MentalHealth Other(Dental) Physicians Supplies/DurableMedicalEquipment Totals Completed 50 40 41 36 50 46 57 320 Deceased 0 1 0 0 0 0 0 1 Refusals 0 1 4 3 1 0 3 12 Disconnected numbers 14 5 18 14 14 14 17 96 Not at number 12 4 11 8 12 8 3 58 No answer 2 2 3 1 0 4 5 17 Unable to contact 6 1 1 0 1 2 1 12 No phone number 6 19 14 25 11 9 7 91 Provider under investigation 10 27 8 13 11 17 7 93 Totals 100 100 100 100 100 100 100 700
The overall results of the telephone interviews were:
• A total of 15 clients denied having a total of 41 services. When the results of the client telephone interview error findings (denials) were compared to the medical record review errors, two of the services for two clients were found to be medial record review errors: a pair of bifocals for which no documentation was received and an oxygen tank for which no delivery receipt was sent by the supplier.
A potential pattern was identified within the Supplies/Durable Medical Equipment category. Three clients denied receiving eyeglasses. The only documentation received on these services was a purchase order. None of these documents actually had a client or guardian signature of receipt. Since these items are supplies and signed receipts were not required by the Texas Medicaid program until July 2000, these three services were not counted as errors in the medical record review.
Sufficient medical record documentation was received on the services for the remaining ten clients that proved the services were actually provided.
Besides the clients who stated they could not recall services outright, the recall issue was apparent when one of these ten clients with nine lab services stated the lab services were not done. The client denied the lab tests when asked the questions during the lab section of the interview tool, however, stated that blood tests and a urinalysis were done when responding to questions in the physician office visit section. Lab results were found in the medical record.
• Thirty-three clients stated they could not recall if they received a total of 82 sample services.
Study Categories Number of ClientsDenying Services Number of Serviceswith Errors (Denials) Number of Clients“Do Not Recall” Services Number of Services Client’s “Do Not Recall” Ancillary/Outpatient 2 5 7 14 Home Health 1 1 4 5 Hospitals 5 25 4 26 Mental Health 0 0 1 4 Others (Dental) 0 0 7 18 Physicians 2 3 2 4 Supplies/Durable Medical Equipment (DME) 5 7 8 11 Totals 15 41 33 82
Client Eligibility Determination
Texas Medicaid serves eligible populations with insured medical services. A portion of the 64 various programs within the Texas Medicaid program are federally required. Each of these programs may have different eligibility requirements. To receive Medicaid services in Texas, a client must be eligible for those services based on income and the current value of selected assets. The Texas Department of Human Services (TDHS) determines Medicaid eligibility for acute care clients such as children and pregnant women as well as the acute care eligibility for long-term care clients. The U.S. Social Security Administration, with the assistance of the Texas Rehabilitation Commission’s Disability Determination Services, determines the eligibility for Supplement Security Income (SSI) for disabled persons. Eligibility for Medicaid is automatic for clients receiving SSI payments.
Potential Medicaid clients, other than SSI clients, must complete application forms, including documentation of financial status. Clients other than long-term care clients must submit to a face-to-face interview with TDHS staff. TDHS re-determines eligibility every six months for these acute care clients who must again must submit to a face-to-face interview. Such interviews for the eligibility of long-term care clients are encouraged but not required. Annual reviews for long-term care clients are usually processed by mail.
Clients are not required to have a telephone or provide a telephone number to TDHS staff. However, according to TDHS, clients other than SSI and long-term care clients are asked for telephone information during the initial eligibility determination and every six months. Clients are required to provide TDHS a change of address within 10 days, but are not required to provide changes in phone numbers. TDHS staff has had the ability to enter telephone numbers into its electronic database since October 1996. This database is commonly referred to as the SAVERR system or the System for Application, Verification, Eligibility Referrals and Reporting
Any telephone information for SSI clients depends upon the data provided to TDHS by the U.S. Social Security Administration.
The lack of correct and complete telephone numbers hampered the successful completion of the telephone interview portion of this study. There were similar problems with complete and accurate telephone numbers during the first study.
(See Recent Events in Client Eligibility Determination on page 27 for a discussion of other current issues related to the Medicaid eligibility determination process.)
Repeated attempts were made to contact 607 of the 700 Medicaid clients in the sample. Ninety-three, 13 percent, of the clients were excluded from the telephone interview process because their provider(s) of Medicaid services were under investigation by the HHSC. To enhance the likelihood that more telephone interviews would be completed during this study, the following actions were undertaken:
• The TDHS’s SAVERR system was queried for telephone numbers of the study sample clients in February 2000.
• The Texas Health and Human Services Commission mailed informational postcards with a reply form to sampled clients in March 2000. (See Exhibit 2 in Appendix K for a copy of the form.)
• Telephone numbers were culled from the medical records received for the sampled clients from May through August 2000.
• The Texas Department of Protective and Regulatory Services caseworkers were contacted for sampled clients participating in foster care programs in July 2000.
• The Texas Comptroller’s office of Criminal Investigation Division searched proprietary databases and phone directories to locate telephone numbers for the sampled clients in July 2000. Databases with criminal histories were not accessed.
• The TDHS’s SAVERR was again queried for telephone numbers of the sampled clients in July 2000.
Each one of these actions produced an incremental number of correct phone numbers. For instance, of 587* postcards sent, 77 (13 percent) were returned with telephone numbers. Even more significantly, only 29 (5 percent) of the cards were returned as undeliverable. A review of the medical records found a telephone number for at least 132 (22 percent) where no telephone number was listed in the SAVERR system. The second SAVERR query in July produced 57 (9 percent) new telephone numbers. The phone calls were halted in August 2000 in order to compile the study.
* Twenty clients (out of 607) were excluded from the postcard mail-out. These clients received some of their services from providers under investigation and at the time of the mail-out, a final strategy for contacting these clients had not been finalized with HHSC/OIE.
Even with this extensive effort to locate correct telephone numbers, a total of 206 (34 percent) non-SSI clients had not yet been contacted as of August 2000. This list did include clients with a new telephone number from the July SAVERR query. A list of these clients was sent to TDHS so that the staff in the local offices could review their records and provide more information on why these clients could not be located by telephone. Neither the total number of new telephone, nor the location of new telephone numbers, such as paper files, could be determined from the information provided by TDHS. However, some reasons were given about why telephone numbers were not available such as: the death of the client; the telephone number to contact was that of a landlord or employer; the client is no longer eligible for Medicaid; the client has no phone; the client has moved to another state; or the client resides in a nursing facility.
Recent Events in Client Eligibility Determination
The following is a summary of other recent events that are directly related to the current and future Medicaid eligibility determination process at TDHS.
1. The Texas State Auditor’s Office (SAO) concluded in a report published in July 2000 that the TDHS has “processes in place to ensure the integrity of client data” in SAVERR. The SAO also determined that TDHS regularly performs data matches with other electronic data bases such at the Texas Department of Health Bureau of Vital Statistics to identify Medicaid recipients that may be receiving benefits fraudulently or inappropriately. A review of telephone numbers was not included in the scope of the SAO review.
2. The Texas Department of Health’s Bureau of Managed Care also has need for reliable and current phone numbers for Medicaid clients. The bureau uses Medicaid client information from SAVERR to enroll Medicaid clients in managed care in those areas of Texas that require Medicaid client participation in managed care. This may be in either a Health Maintenance Organization (HMO) or one of the state’s health network plans, such as the Primary Care Case Management or the Prepaid Health Plan. The Medicaid client information is forwarded to a contractor of TDH called the enrollment broker. It is the enrollment broker’s responsibility to inform Medicaid clients of their choices in selecting a managed care program. A Medicaid client must be enrolled within 30 days of their eligibility determination. Accurate and up-to-date telephone numbers in SAVERR are important to completing this process.
3. In response to changes in state law, TDH and TDHS have initiated a new process to expedite enrollment in Medicaid HMOs for eligible pregnant women. This new program makes the need for TDHS to collect and maintain current telephone numbers even more crucial since pregnant women who have applied for Medicaid should be scheduled for an appointment with a provider within 30 days from the date of application. Obviously, any client telephone number is invaluable in order to complete this process of enrollment in an HMO and selection of a provider within the 30-day time period. In lieu of telephone contact, the entire process must be completed by mail.
4. Legislators have made proposals for changing the eligibility policies for Medicaid during the 2001Texas Legislative Session. The Senate Committee on Human Services has recommended eliminating face-to-face interview requirements, allowing mail-in eligibility or telephone re-certification, and eliminating the assets test for children’s Medicaid eligibility. The committee also recommend extending the eligibility period from six months to 12 months for children in the Medicaid program. Moreover, the committee recommended creating a work group to improve the application and review process for Medicaid and other programs at TDHS. The suggested composition of the work group includes legislative staff, advocates, service providers and stakeholders.
5. Beginning in December 2000, TDHS will begin to schedule telephone interviews for the re-determination of Medicaid eligibility for those clients who only receive Medicaid services such as children. A face-to-face re-determination interview is still required for those clients who receive both Medicaid and cash assistance under the Temporary Assistance to Needy Families program.
6. SAVERR, the automated system that TDHS uses for eligibility determination for Medicaid and other programs for low-income persons, is scheduled for updates and revisions through a process known as the Texas Integrated Eligibility Re-determination System (TIERS). House Bill 1 of the 1999 Legislature authorized this redesign. The goals of the project are to replace the current automated systems with a single integrated system to support eligibility determination processes, ensure effective and efficient business processes and establish a foundation for a comprehensive integrated eligibility process. The final stage of TIERS is expected to be completely implemented in 2004.
Contextual Data Analysis Review Results
Analysts found potential errors with 89 sample services for 19 clients out of the total 1,609 services reviewed. The distribution of these errors are illustrated in Table 5. Of the 89 CDA errors, 30 were confirmed as errors during the review of the medical record documentation. These errors were:
Type of Error Number of Errors Clerical errors 2 Documentation errors 11 Lab services billed incorrectly 15 Procedure coding errors 2 Total 30
These errors are discussed in the medical record review results section. The other 59 potential service errors were proven to be appropriate through the medical record documentation.
Table 5: CDA Errors by Ranking of the Number of Errors in the Sample Categories
Study Categories Number of CDA Errors Ranking Ancillary/Outpatient 32 1 Other (Dental) 23 2 Hospitals 16 3 Physicians 12 4 Home Health 4 5 Supplies/Durable Medical Equipment 2 6 Mental Health 0 7 Total 89
Medical Record Review Results
The medical record (MR) review proved to be the most revealing on potential overpayments of all three review methods. Of the 1,609 sample services reviewed, 269 were found to have errors. Before finalizing the results of the medical record review, the study project team from the Comptroller’s office and SAO met with staff from the HHSC/OIE, TDH and NHIC. During these meetings, each sample service considered a discrepancy by the study team was discussed and researched. Both the medical review and the interagency meetings used the 1999 Texas Medicaid policies and procedures in place at the time when these sample services were performed as a reference. The HHSC/OIE, TDH and NHIC staff was given an opportunity to agree or disagree with the findings. Some of the claims with questionable service errors went under additional review by HHSC/OIE specialized department staff, and a physician and dentist, to provide a final opinion on the discrepancy.
The agencies confirmed 214 sample service errors presented by the study team, leaving 47 reclassified as no-errors. Of those 47 sample service errors, the team agreed to change the classification of 36 services for disposable supplies for which no delivery receipt or a delivery receipt without a client or parent/guardian signature was included with the medical record documentation. This provision has been a requirement for the durable medical equipment (DME) since June 1, 1998. The addition of the Home Health and DME supplies did not occur until after the study. Texas Medicaid providers were reminded in the July/August 2000 Texas Medicaid Bulletin that all Home Health and DME supplies require a delivery receipt signed by the client or parent/guardian as documentation for these services.
The remaining 11 service errors out of the 47 were reclassified as non-errors based on the interpretation and adjudication of the Medicaid policies by TDH/NHIC and agreed to by HHSC/OIE. For example, the study team counted an injection service as an error because it appeared to be part of the injection medication charge. Medicaid policy allows for the reimbursement of the injection service, thus this service was not counted as an error.
As noted in the medical record methodology discussion, services for which the providers sent in no medical record documentation were all considered payment errors. This accounted for 55 sample services, which brought the total discrepancy count from 214 to 269. The total overpaid dollars for the 269 services was $10,458.30.
What Type of Errors Were Found?
Table 6 illustrates the distribution of the errors found in the January 2001 study. Explanations of the categories are provided below Table 6.
Table 6: Type of Medical Record Review Errors by Study Categories
Type and Number of Errors Found with the Sample Services
Study Categories by Rank NoMedicalRecordReceived NoDocumentfor theSampleDayReceived NoDocumentof theSampleService IncompleteDocumentationof the Service ClericalErrors LabUnbundled Upcoding PotentialFraud orAbuse Totals Ancillary/Outpatient 4 1 5 0 2 17 0 0 29 Home Health 4 2 2 0 0 0 0 0 8 Hospitals 14 10 44 0 1 2 1 0 72 Mental Health 2 4 1 1 3 0 1 2 14 Other 3 9 21 2 32 0 3 10 80 (Dental) (0) (4) (18) (2) (27) (0) (3) (5) (59) Physicians 22 7 4 1 5 0 6 1 46 Supplies/DME 6 1 6 0 6 0 0 1 20 Totals 55 34 83 4 49 19 11 14 269
Various documentation errors were classified into four types based on the existing problem:
• Medical records not sent after receipt of certified request letters;
• Medical records received without any documentation for the sample date-of service requested;
• Medical records received for the sample date-of-service requested without documentation of one of the services billed and paid;
• Medical records received for the sample date-of-service with incomplete information based on the Medicaid requirements, such as the client’s name.
Lack of documentation was the most prevailing problem found with the medical records sent by the providers. For example, multiple lab tests would be documented as ordered, however, no documentation was included to prove the tests had been performed, for example, the results of the tests. This error accounted for 83 of the 269 errors.
Medical records were not received from providers for 55 sample services.
Disregarding the request for complete records for the specific claim and sample date, a few of the providers sent volumes of pages documenting services up to 5 years prior to the sample date or the most recent services that had not even been billed to the Medicaid program. These providers were frequently contacted by HHSC to explain specifically what was being requested. HHSC staff was often told by the provider’s office staff that the records sent “were all we have on that patient.” There were 34 sample services for which no documentation for the sample date-of-service was in the medical records received.
Also found, though not as frequently, were records with documentation for the sample date-of-service that were missing evidence for one or more of the services paid. For instance, a well-child check-up billed with immunizations lacked one or more of the immunizations billed. Or specific information required by TDH/NHIC for a Medicaid claim, such as the patient or provider name. This discrepancy was found with 4 sample services.
In total, there were documentation errors on 176 sample services in this study.
As noted in the methodology section, providers were informed of the Texas Medicaid documentation requirements in September 1999. In all the occurrences in which records were received without documentation of the sample services paid, the HHSC/OIE will investigate the providers and potentially recoup the monies paid for those services. On all claims for which no medical record was received, the HHSC/OIE intends to recoup the entire claim amount paid and investigate the provider.
Errors that were considered potentially clerical (not intentional) were found with 47 sample services. The most common examples of these were billing an accounting posting date instead of the actual service date and billing a wrong procedure or diagnosis code than was documented in the medical record. These were referred to HHSC/OIE for further investigation and potential adjustment.
During the agency review meeting, NHIC staff stated that the provider had already adjusted some of these errors. If the adjustment occurred after the provider received the first request letter, the service was still counted as a discrepancy, assuming the state’s interest in the documentation prompted the provider to review how the claim was billed and consequently adjusted the error found to avoid further state action.
Laboratory Tests Unbundling
Billing lab tests individually instead of in panels (groupings) as directed by the Texas Medicaid Provider Manual was found in 19 sampled services. NHIC will adjusts these tests and follow up with provider education.
Abusive practices such as coding office visits for a more complex level of care than was documented in the medical records received was found with 11 sampled services. This is also referred to as “upcoding” and is considered abusive because it is primarily done to maximize the reimbursement to the provider. A medical necessity evaluation by a professional physician was performed on all these records prior to confirming them as errors.
Potential Fraud or Abuse
Potential fraudulent actions were noted with 14 sample services billed by and paid to six providers. Examples were billing services using another provider’s Medicaid enrollment number and changing the date-of-service billed to meet the 90 day filing date when the actual date-of-service occurred 2 months prior to the billed date. All of these services will be investigated by HHSC/OIE with potential recoupment or appropriate sanctions.
Which Provider Types Made The Most Errors?
Within the stratified study categories were associated services and core services that, in some instances, were performed by types of providers that belonged to one of the other study categories. To accurately represent which provider types were responsible for the errors identified, the selected sample was resorted by the provider’s types and specialties. This resorting placed the appropriate providers into the category of service with their specific type and specialty designated for that category of service. For example, following the resorting, all services performed by laboratory and x-ray facilities were grouped together in the Ancillary/Outpatient category, regardless of the category in which they were selected and assigned in the primary sample. Also, grouping the individual services by appropriate health care service areas was required to identify the potential overpaid dollars and calculate the overpayment measure for this study. (A list of the type of providers grouped for each of the study’s categories of health care services is in Appendix J, Exhibit 2.)
When all the sample services determined as errors were resorted by the provider types and specialties, the following information was found about which type of providers within this study sample rendered the errors noted above. The resorting process was performed after removing the excluded providers under investigation. Table 7 illustrates how these category of services ranked for the number of errors discovered in the study.
Table 7: Number of Services with Errors by Rank per Re-sorted Provider Types in Each Stratified Category
Type and Number of Errors Found with the Sample Services
Study Categoriesby Rank NoMedicalRecordReceived NoDocumentfor theSampleDayReceived NoDocumentof theSampleService IncompleteDocumen-tationof theService ClericalErrors LabUnbundled Upcoding PotentialFraud orAbuse Totals Rank Other 3 9 21 2 32 0 3 10 80 1 (Dental) (0) (4) (18) (2) (27) (0) (3) (5) (59) 1 Hospitals 14 10 44 0 1 2 1 0 72 2 Physicians 22 7 4 1 5 0 6 1 46 3 Ancillary/Outpatient 4 1 5 0 2 17 0 0 29 4 Supplies/DME 6 1 6 0 6 0 0 1 20 5 MentalHealth 2 4 1 1 3 0 1 2 14 6 HomeHealth 4 2 2 0 0 0 0 0 8 7 Totals 55 34 83 4 49 19 11 14 269
The majority of the errors were for sample services billed in the Other category. Eighty errors out of the 267 total errors on sample services were found in this category. Most of these errors were due to the dental providers included in that category. After resorting the sample by the provider’s type and specialty, there were a total of 59 dental providers included in either the random core sample service selection or associated services. The two most common errors noted with the dental services were clerical (32 sample services), followed by various documentation errors (24 sample services). One dental provider billed five of the potential fraudulent services.
The next category in rank of errors was the Hospitals. The hospital providers made seventy-two of the 269 errors. The majority of these were documentation errors. Specifically, these hospitals did not send any record after three request letters, did not include documentation for any of the services on the sample date, and did not include documentation on the sample date-of-service for specific services billed.
The most common error made by the Physicians was not sending the medical record requested. This was followed by errors in which the physician did not send documentation for the sample date-of-service, clerical errors and upcoding their office visits.
Within the Ancillary/Outpatient category, the majority of errors were unbundling lab tests made by the independent laboratory providers. There were 17 sample laboratory services out of the 269 errors billed individually that should have been billed as part of a panel (group) according to the 1999 Texas Medicaid Providers Procedures Manual.
No delivery receipts or receipts without the client or parent/guardian’s signature for durable medical equipment was the most significant error with the Supplies/Durable Medical Equipment category.
Home Health providers had the least number of errors, eight out of the 269. All of the errors were with documentation.
How Many Providers Were Under Investigation?
Noted in the methodology section was the exclusion of sample services from the study, delivered by providers who HHSC/OIE identified as under open investigations. These investigations were due to a variety of reasons:
• Referrals from TDH, providers or clients,
• Referrals from the NHIC SURS, and
• Medicaid Fraud and Abuse Detection System research processes.
To prevent interruption and complication of these investigations, no contact was made with these providers by phone or mail.
Since the services for these providers under investigation were not reviewed and the potential outcome of an investigation is determined by the research and decision of the HHSC/OIE staff, none of the dollars associated with these services were counted towards the overall overpayment measure.
Potential Overpayment Measurement
Initially, the calculation of the potential overpayment measurement requires the assignment of an overpayment amount to those services that were paid in error.
The amount of overpayment was either the entire dollar amount of the service, an adjusted dollar amount or zero. For example, if the service was classified as an error because the service was not documented in the medical record, the entire dollar amount of the service was counted as an error. However, if a service was documented in the medical records but not coded correctly on the claim form, the amount of the error would be the difference in the amount paid and the amount that should have been paid. The dollar amount of the error was zero if the service was documented in the medical record and the error was an incorrect date or some other clerical error.
The total dollars calculated as potentially overpaid from the sample services with errors is $10,458.30 out of $130,062.89 paid for these services.
Next, the determination of the percentage of potential overpayments requires the following series of calculations performed for each category of service, such as ancillary/outpatient:
1. Category providers under investigation* dollars = Percentage of dollars under investigation in the sample for each category
Category sampled dollars Example: $1,910.67 = 0.23 $8,264.26
2. Category universe* dollars x Percentage of dollars under investigation = Category universe dollars assumed to be under investigation Example:
$28,597,072.797 x 0.23 = $6,611,550.11
3. Category universe dollars – Category universe dollars assumed to be under investigation = Category universe dollars without dollars under investigation Example:
$28,597,072.797 – 6,611,550.11 = $21,985,522.69
4. Category universe dollars = Percentage of the category in the universe
Sum of Universe Dollars Example: $21,985,522.69 = 0.05 $464,052,754.83 5. Sum the following calculation for each category to determine the total weighted average of potentially overpaid dollars.
a. Potential overpayment dollars x Percentage of category in study universe = Weighted overpayment dollars Example:
$508.25 x 0.05 = $25.41
b. Category sampled dollars – Category sampled dollars under investigation = Category sampled dollars without dollars under investigation Example:
$8, 264.26 – $1,910.67 = $6,353.59
c. Category sampled dollars without investigation dollars x Percentage of category in study universe = Weighted sampled dollars Example:
$6, 353.59 x 0.05 = $317.68
6. Total potentially overpaid dollars weighted average = Study weighted average percentage of potentially overpaid dollars
Total study sampled dollars weighted average Example: $2,218.07 = 7.24 $30, 627.25
Providers under Investigation – are the providers excluded from the study because they were under current investigation by the Texas Health and Human Services Commission’s Office of Investigation and Enforcement.
Universe – is the three month study period of September 1, 1999 through November 30, 1999. The universe dollars do not include the dollars for the provider types excluded from the study.
See Table 8 below for a summary of the calculations and the tables in Appendix L, Exhibit 1 for all the calculations.
The result of this series of calculations is the determination of the weighted average overpayment measurement of 7.24 percent. The margin of error is plus or minus three percent.
Table 8: Medicaid Summary Data for the Potential Overpayment Calculation
Study Categories Total Dollars Paid for Reviewed Sample Services Overpaid Dollars for Errors Proportion of Sample Universe Service Dollars Overpayment Proportion Percentage Ancillary/Outpatient $ 6,353.59 $ 508.25 0.05 8.00% Home Health $ 13,580.34 $ 1,549.03 0.03 11.41% Hospitals $ 54,419.64 $ 3,793.78 0.42 6.97% Mental Health $ 11,664.79 $ 499.74 0.05 4.28% Other (Dental) $ 21,946.01 $ 783.25 0.15 3.57% Physicians $ 11,242.69 $ 1,291.45 0.27 11.49% Supplies/DME $ 10,855.83 $ 2,032.80 0.04 18.73% Un-weighted Totals $130,062.89 $10,458.30 1.00 8.04% Weighted Average Totals $ 30,627.25 $ 2,218.07
These numbers exclude services for provider’s under investigation by HHSC/OIE
Dollars at Risk
The overpayment measurement can be applied to the annual Medicaid expenditures to determine the “dollars at risk” in the Medicaid program. The term “dollars at risk” is used because the amount that is computed when applying the overpayment measurement to the annual expenditures is not recoverable unless all of those individual services that are questionable are identified through a complete medical record review of all services submitted for payment, which is not possible given the size of the Texas Medicaid program. However, HHSC can use the findings in this study to target specific areas of investigation.
The “dollars at risk” for 1999 is $217 million. This figure was computed by multiplying the overpayment measurement rate of 7.24 percent times $3 billion of the Medicaid fee-for services expenditures for calendar year 1999. This figure assumes that providers who were under investigation by HHSC/OIE within the fee-for service realm but were not included in the sample reviewed by this study will be also be responsible for 7.24 percent of overpayments.
Texas Department of Health
A. TDH should reinforce its provider education in both workshops and publications in the following areas:
• Providers’ responsibilities when medical records for services billed and paid for Medicaid clients are requested by an authorized state agency.
• Required documentation on a medical record for a Medicaid client as part of their provider agreement to follow Medicaid policies and procedures when billing and accepting reimbursement.
• Training clerical staff on appropriate diagnosis and procedure coding.
B. TDH should require NHIC to enhance MMIS edits to deny lab tests billed separately that are required to be billed as panels to avoid reimbursing unbundled laboratory tests.
Texas Health and Human Services Commission
A. HHSC should consider the following areas of potential billing abuse to include in its Medicaid Fraud and Abuse Detection System (MFADS) model and targeted query development:
• Areas of health care delivery in which different types of providers may be billing higher procedure codes to maximize their reimbursement. These areas should not only identify physician office visits, but also the medical supplies and durable medical equipment area of health care.
• Increased targeted queries on medical supplies and durable medical equipment with follow-up on delivery and client receipt by requesting the documentation.
• Periodic requests for medical record documentation of a targeted sample of providers to ensure the required documentation procedures published by TDH are being performed by the Medicaid providers. This requirement not only identifies appropriate payments for services billed but also should help improve the quality of care Medicaid client receive.
Legislative—Medical Record Confidentiality
A. State law should be changed to ensure the confidentiality of medical records in the custody of the Comptroller.
Since the Comptroller does not typically have custody of medical records through its other functions, it would be prudent to change state law to exempt these medical records from disclosure through the Texas Open Records Act. The exemption should cover the release of client data and medical provider information.
B. The Legislature should include the Comptroller as a participant in any work group that is charged with reviewing or improving TDHS’s systems for determining Medicaid eligibility.
The statute requiring the Texas Health Care Claims Study also directs the Comptroller to determine “the need for changes to the eligibility determination system used under the state Medicaid program.” The Comptroller’s participation in this work group would enable the Comptroller to fulfill its responsibilities under this statute.
These recommendations can be implemented within these agency’s current appropriations.
 US Department of Health and Human Services, Health Care Financing Administration, Guidelines for Addressing Fraud and Abuse in Medicaid Managed Care: A Product of the National Medicaid Fraud & Abuse Initiative (August 2000), p. 12 (http://www.hcfa.gov/medicaid/smd80700.htm). (Internet document.)
 National Heritage Insurance Company, Medicaid Management Information System paid claims history database, Plano, Texas, October 2000.
 Texas Department of Human Services, “State of Texas – Medicaid Expenditure Information,” Austin, Texas, November 15, 2000.
 Texas Comptroller of Public Accounts, Final Staff Draft Report on Health Care Claims Study and Comments from Affected State Agencies (Austin, Texas, December 1998), p. 68.
 Letter from Jon Nelson, project manager, Texas State Auditor’s Office, March 27, 2000.
 Interview with Jessica Shahin, Texas Department of Human Services, Austin, Texas, September 18, 2000; and e-mail communications from Dee Church, Texas Department of Human Services, Austin, Texas, October 19 and 25, 2000.
 National Heritage Insurance Company. “Home Health/DME/Medical Supplies Providers, Billing Reminder,” Texas Medicaid Bulletin, No. 149 (July/August 2000), p. 23.
 National Heritage Insurance Company, “General Medical Record Documentation Requirements.” Texas Medicaid Bulletin, No. 142 (September/October 1999), p. 10.
 Texas Department of Health, 2000 Texas Medicaid Provider Procedures Manual, pp. 25-7 to 25-9.
 V.T.C.A., Government Code §403.028 (Texas Legislative Council text online at http://www.capitol.state.tx.us/statutes/go.go040300.html#go037.403.028).