Performance Audit

Medical Assistance Long-Term Care As Provided
In Nursing Facilities

September 30, 1997


Chapter IV - Reimbursement System

On January 1, 1996, Pennsylvania began using a case-mix system to reimburse nursing facilities participating in the Medical Assistance Long-Term Care Program. Therefore, our audit concentrated on the first six months of the reimbursement system's implementation.

The case-mix system changed the reimbursement approach from a cost settlement system to a preestablished payment rate system. Whereas the cost reimbursement system allowed for estimated interim payments followed by a cost settlement process, the case-mix system establishes a nursing facility's reimbursement rate prior to payment.

To calculate each facility's reimbursement rate, the case-mix system utilizes the Resource Utilization Groups III - an indexing scheme which classifies residents according to 44 categories using a combination of medical conditions and other dependency and service-level measures. As a result, long-term care facilities receive greater reimbursement for providing higher levels of care. Reimbursement rates are revised quarterly on established picture dates - dates on which a "snapshot" of updated resident information is collected for each facility.

Nursing facilities invoice the Department of Public Welfare monthly for the number of days they provide services for each resident. The reimbursements they receive are calculated from this data in conjunction with their applicable reimbursement rate, and are reduced by any amounts paid to the nursing facilities on behalf of the residents.

Objectives and Methodology

Our objectives included determining whether the Department of Public Welfare ensured services billed by long-term care facilities for Medical Assistance residents were actually provided. In addition, we elected to determine if the reimbursement rate was accurately calculated and used in determining the payment for the long-term care facilities in our sample.

To complete these objectives:

In addition, at 25 facilities we selected to visit (see Appendix E for sample methodology) -

Conclusion IV-A: The Case-Mix System is Accurately Calculating and Applying the Correct Long-Term Care Providers' Reimbursement Rates

Based upon our inquiries and testing, we concluded that the Nursing Home Information System (NIS) computer system is calculating applicable rate information accurately. Based upon our manual recalculation of the rates for a randomly selected peer group, we determined that the NIS database programs sufficiently apply the correct parameters for peer group and individual long-term care facility rate calculations.

In addition, we tested the application of the long-term care rate calculations for 25 long-term care facilities. We noted no exceptions between the rates calculated by the case-mix system and the rates used in the Medical Assistance Management Information System (MAMIS) payment system.

Conclusion IV-B: Long-Term Care Facilities Appropriately Bill for the Services They Provide

We selected ten invoices from each of 25 facilities for detailed testing of the billing system. The calculations and extensions were verified, the rates matched those used in the Medical Assistance Management Information System (MAMIS) payment system, and the dates of service matched the facilities' records. Our testing of the invoices and billing records submitted by the 25 facilities selected for detailed testing, and the supporting census data and other records of care at the 25 facilities, generally provided assurance that the long-term care facilities provided the services that were billed.

Our testing disclosed other errors that were not identified or corrected by the input controls in the Nursing Home Information System (NIS) or the MAMIS payment system. As a result, we recommend several improvements in the controls over these systems. In addition, our review of the internal control procedures of the Division of Long-Term Care Client Services revealed areas that could be strengthened or improved. The results of our audit are summarized in the following sections.

Conclusion IV-C: The Department of Public Welfare Needs to Strengthen the Controls Within the Nursing Home Information System

Part I - Auditor General Testwork

The Nursing Home Information System (NIS) is a computerized system that tracks all relevant statistics for calculating long-term care facilities' per diem reimbursement rates. The NIS calculates a new rate for each quarter of the year. Each facility's reimbursement rate is based upon the resident census data reported on the case-mix index (CMI) statements on a specific date, known as the picture date. A picture date is defined as "a day when a facility should take a picture of the residents in the facility and gather assessments for these residents." The picture dates are February 1, May 1, August 1, and November 1 of each year.

We tested the information transmitted to the NIS on a selected picture date from 25 long-term care facilities to determine the accuracy of this information. Out of the 25 CMI statements reviewed, 12 facilities' statements contained some type of error and 1 facility's records could not be reconciled to the underlying census records. Of the 12 facilities that contained errors, only 1 had a major impact on reimbursement. The remaining errors had no perceivable impact on the case-mix per diem reimbursement calculations.

The results of our testing revealed the following issues:

We concluded that control weaknesses existed in the NIS system since these discrepancies were not identified during the transmission process or subsequently corrected before the quarterly rate calculations.

It is important that the NIS contain valid and accurate information as required in Department of Public Welfare Regulations, Chapter 1187. The case-mix reimbursement system relies on the nursing facilities to provide accurate data for the calculation of per diem reimbursement rates. The CMI statement is an integral part of the case-mix process and its accuracy is the foundation for calculating the amount of payment a nursing facility receives on a quarterly basis. In order for this system to operate at an optimal level, accurate information is essential.


The Department of Public Welfare (Public Welfare) should implement safeguards in the NIS to prevent and detect input errors, such as:

Additionally, Public Welfare should recalculate the reimbursement rates for the facility detailed above for the picture dates we selected to recover any overpayments, including the estimated overpayment of $146,437.

Part II - Utilization Management Review Testwork

As part of our audit, we reviewed the Utilization Management Review (UMR) teams' responsibilities related to the case-mix reimbursement system. We examined the testwork the UMR teams performed at the 25 nursing facilities in our sample. A total of 28 picture date reconciliations were performed by the UMR teams since three of the facilities were visited twice by the UMR teams. Of the 28 statements, 10 contained some type of error. No recalculations of the CMI statements were performed by the UMR teams.

The UMR has chosen a proactive approach in treating errors made by the long-term care providers. It does not function in a cost settlement role. UMR management expressed its belief that its primary responsibility is educational. According to UMR management, by educating the personnel responsible for submitting accurate long-term care facility nursing information, the UMR teams will be able to prevent errors from occurring in the future. Since the nursing facility industry has a high turnover of personnel, training is an ongoing task and an important aspect to ensure accurate and complete information is submitted to and gathered by the Nursing Home Information System.

The UMR staff is the only oversight group for the quarterly case-mix index (CMI) statements. Since implementation of case-mix, the UMR staff has recommended only one gross adjustment in the amount of $127,000 to recoup excess reimbursement. Based on our review of the Chapter 1187 rules and regulations, the UMR is defined as:

An audit conducted by the Department's medical and other professional personnel to monitor the accuracy and appropriateness of payments to nursing facilities....
Chapter 1187.108 refers to gross adjustments to nursing facility payments. It further states that:
If a nursing facility's MA CMI changes as a result of UMR resident assessment audit adjustments, retrospective gross adjustments shall be made for the nursing facility involved.
In our view, because the case-mix rate calculation system is new, it needs to be more closely monitored. We agree that education is an important function throughout the case-mix process; however, the UMR is the only mechanism in place to assure that appropriate payments are made.

We understand the nature of the case-mix payment system is to eliminate the cost settlement process and to provide facilities with advance notice of their payment rate so that they can plan accordingly. However, the CMI statement's accuracy is vital to ensure proper quarterly per diem rates are calculated. The possibility exists that inaccurate picture date information can result in more or less reimbursement for a facility. The process of recalculating the impact on CMI statements is not a very time-consuming task considering that it would only be performed in those cases where warranted.


In cases where the UMR team identifies significant errors reported on the CMI statement, the team should determine the impact of those errors on the quarterly per diem rate reimbursement calculation, and adjust the facility's reimbursement amounts.

Conclusion IV-D: The Department of Public Welfare Needs to Strengthen the Controls Within the Medical Assistance Management Information System to Recover Payments for Services Claimed for Deceased Residents

As part of our testing of the billing and payment process, we reviewed 103 discharge invoices to ensure accuracy of reimbursement. Out of this amount, we identified one invoice that contained an error. The facility billed and received reimbursement for a resident beyond the date the resident died. The amount of overpayment in this specific case was minimal ($167.10). However, concern is raised by a weakness in the system that allows overpayments.

The Medical Assistance Management Information System (MAMIS) verifies a resident's eligibility by checking the Client Information System (CIS) database updated by the county assistance offices under the Office of Income Maintenance, Department of Public Welfare. If this database does not contain the current status of a resident, an inappropriate payment may be processed.

County assistance office caseworkers update the CIS database immediately upon notification by a nursing facility of a change in a resident's status. Inquiries of Office of Income Maintenance management revealed that the main reason for county assistance office delays in updating the CIS database is failure of nursing facilities to submit timely information.

This is not the first time that the Department of the Auditor General discovered this type of error. Testing performed in prior years by our Bureau of State-Aided Audits, responsible for auditing county-owned nursing facilities, disclosed other deceased residents appearing on the MAMIS runs. The major reason this occurred is related to the county nursing facility engaging the services of an outside agency to handle its billing process, which added to the chance that charges for a deceased resident were billed.

From our interviews, we learned that the Department of Public Welfare (Public Welfare) is aware of this problem and is coordinating with the Department of Health to cross-reference its files to detect and prevent the payment of billings for deceased residents from occurring in the future. Recently, Public Welfare conducted a pilot study using 1995 as the test year to determine if this type of control would be beneficial. The study compared Public Welfare's active social security numbers (over one million records) to the Department of Health's death records, resulting in approximately 1,000 matches. Approximately 800 (or 80 percent) of the matches indicated that Public Welfare records had not been closed when residents died. This outcome justifies implementation of controls to prevent payments for deceased residents. Within the next few months, Public Welfare plans to conduct a onetime five-year match of both active and inactive files. Once completed, monthly comparisons will be implemented.

From our inquiries of personnel in the Vital Records Division within the Department of Health, we also learned that the process of receiving death notices can take as long as four months. This delay is attributed to the time it takes death certificates to get through the entire process of updating death records in the files. However, because the process could take as long as four months, notification of death from Vital Records will not prevent payments for deceased residents. Nevertheless, by establishing a link with the Department of Health's files, Public Welfare will be able to check its database as well as the CIS database to identify cases of incorrect payments after-the-fact. Public Welfare can recover any excess reimbursement received by a nursing facility through the gross adjustment process.


The Department of Public Welfare should establish a regular procedure for matching the active records in the CIS database with the Department of Health's death records. In this manner, Public Welfare will be able to establish an independent check on the long-term care facilities for notification of residents who have died.

Conclusion IV-E: The Number of Residents Requiring Higher Levels of Care has Increased

The Medical Assistance Long-Term Care Program allocates Medical Assistance residents into 44 assessment categories, known as Resource Utilization Groups III (RUG-III) categories. Each category includes a description of the level of care or type of treatment for a resident in a long-term care facility. Consequently, each category is assigned a score, known as the PA Normalized Index. This index weights the level of care between 0.39 and 3.93 as of January 1, 1996. Low indexes generally equate to low levels or low costs of care and higher indexes equate to higher levels or higher costs of care.

The case-mix payment system was implemented with many goals for improving the quality and accessibility of nursing facility services to Medical Assistance recipients throughout the Commonwealth. The case-mix system adjusts the per diem rates received by facilities by their case-mix index which better reflects the resource uses of residents according to their health needs. This system encourages facilities to admit and provide appropriate quality services to more needy residents.

Under the previous payment system, facilities classified residents by three acuity levels: intermediate care, heavy intermediate care, and skilled care. They received reimbursement rates for only two levels: intermediate and skilled which included heavy intermediate care. The facilities received the same reimbursement rate for all levels of care within intermediate or skilled care regardless of the actual care needed. This payment system encouraged facilities to admit residents requiring lower levels of care within the two reimbursement categories since the cost of care was lower to the facility than the cost of higher levels of care. As a result, this system appeared to discourage the facilities from accepting more dependent or costly residents.

We performed a review of the classification of residents and noted relative shifts in classification over the three-year period from February 1, 1994 through February 1, 1997. Our review included both Medical Assistance and non-Medical Assistance residents in the 44 RUG-III categories. We observed a decrease of 18 percent in Medical Assistance residents and 22 percent in non-Medical Assistance residents, in RUG-III categories carrying a score of less than 1.00. Under the case-mix system, low RUG-III scores imply lower levels of care. During this same time frame, we noticed an increase of 26 percent in MA residents and 23 percent in non-Medical Assistance residents for RUG-III categories with case-mix index scores of 1.00 and above. While there has been an overall increase of 4 percent in both Medical Assistance and non-Medical Assistance residents over the time period of our review, these shifts in category totals seem to concur with the case-mix program intent described above. Without considerable detailed testing, however, we are unable to make definitive determinations about the reasons for the shifts in classification.


The Department of Public Welfare should continue to monitor the shifts in long-term care population to the higher scored categories in order to determine, if indeed, the case-mix system is providing the necessary incentive to long-term care facilities to admit Medical Assistance recipients that need higher levels of care.

Conclusion IV-F: Case-Mix Reimbursement Rates for Alzheimer's Residents may be Too Low

The Resource Utilization Groups III (RUG-III) categories described in the prior conclusion are also used by the case-mix reimbursement system to provide payment to the long-term care facilities. The index is used in the calculation of the facility's per diem reimbursement.

Based on our inquiries about the effectiveness of this system we learned that the long-term care facilities generally favored the 44 cost category portion of the case-mix system over the previous reimbursement system since it more accurately matches reimbursement with levels of care. However, our inquiries also revealed one exception to this belief. Reimbursement for residents suffering from Alzheimer's disease may be improperly classified.

The current indexing system places Alzheimer's residents in the behavioral problems categories. These categories carry a PA Normalized Index value below 1.00 (from 0.41 to 0.86). We found that residents in the behavioral problems categories required higher than normal levels of supervision. The case-mix indexing system was designed to provide reimbursements to nursing facilities for equipment and treatments (capital intensive), not increased staff (labor intensive), in weighting the costs of medical services. As a result, the cost of additional non-medical personnel who function as chaperones was not considered in the low index score. Long-term care facilities may not be adequately compensated by the low case-mix reimbursement rates and, as a result, the care of Alzheimer's residents may suffer.


The Department of Public Welfare should continue to monitor the adequacy of care to Alzheimer's residents and, if necessary, consider revising the indexing scores for these categories of care.


Chapter 3 Table of ContentsChapter 5