Chapter Four

Case Mix and Medicaid Reimbursement

The program review committee examined the relationship between Medicaid resident case mix, aggregated by facility, Medicaid reimbursement rates, and each facility's allowable direct care costs for the year ending September 30, 2000. The allowable direct care cost category is one of five components used in calculating Medicaid rates and includes salaries and related fringe benefits for nurses and nurse aides, and nursing pools.

Time measurement studies done by the federal Centers for Medicare and Medicaid Services (CMS) show that resident acuity (health and functional status) has a major impact on facility resource requirements, particularly in the varying amount of nurse and nurse aide time consumed by residents. Because of this relationship, Medicare, and Medicaid programs in 26 states, reimburse nursing homes based on some type of resident case-mix system. Most Medicaid case-mix reimbursement systems recognize higher costs that result from caring for residents with higher needs, typically by applying a case-mix index to a facility's direct care costs.

Uniform resident assessments. Federal law requires that nursing homes conduct a "comprehensive, accurate, standardized, reproducible assessment of each resident's functional capacity," within 14 days of admission, upon a significant change in health status, and annually. Facilities use a standardized resident assessment instrument, mandated by CMS, which includes the Minimum Data Set (MDS). The MDS is a core set of screening and assessment elements that forms the foundation of comprehensive assessments for all residents of long-term care facilities. It establishes common definitions and coding categories, and ensures uniformity in resident assessment across facilities.

Information from the MDS can be used to group residents into Resource Utilization Groups-Version III (RUGs-III), a resident classification system developed by CMS. The RUGs-III is based on three nursing staff time measurement studies conducted by CMS. Using the RUGs-III, residents can be first classified into one of seven broad categories (shown in descending order of their relative cost of nurse and nurse aide use). These categories are:

The resident's functional status or ability to perform activities of daily (ADLs) living further subdivides these groups into one of 34 categories for Medicaid residents.8

To evaluate facility case mix, each of the 34 RUG categories is assigned a case-mix index (CMI), also called a "weight". The weights quantify the differences among groups in the relative costliness of their care needs provided by direct care nursing staff. Overall, the CMI increases as more care is needed because of: poorer ADL functioning; need of nursing rehabilitation services; or signs of depression. The Medicaid weights are based on nursing staff times found in large multi-state research studies conducted in 1995 and 1997 and range from 0.59 for the lowest RUG classification to 2.10 for the highest group. (For a description of the seven broad categories identified above, as well as the associated CMI for each RUG category see Appendix D).

Methodology. Program review committee staff obtained authorization from CMS to access MDS information on all individuals who where in Connecticut nursing homes on September 30, 2000. There were 261 licensed nursing homes with 32,745 beds -- 29,949 chronic and convalescent nursing home (CCNH) beds and 2,796 rest homes with nursing supervision (RHNS) beds. Records were obtained on 31,476 nursing home residents from the Connecticut Department of Public Health. However, since the MDS does not capture payer source (i.e., Medicaid, Medicare, or private pay), or the level of nursing care (chronic and convalescent care or the lower care level provided in rest homes with nursing services), information on Medicaid-eligible residents and level of care had to be obtained from DSS and merged with the DPH data.


Payer source. A total of 31,476 nursing home residents were contained in the database analyzed by the program review committee staff. Of these, Medicaid paid for 19,719 residents of either CCNH or RHNS facilities. The payer sources for the remaining 11,757 residents included: Medicare; private pay; or other source. The committee staff further subdivided the Medicaid residents by the type of facility providing care -- 18,350 residents (93 percent) lived in a home licensed as a CCNH and 1,369 (7 percent) resided in an RHNS facility. The committee staff focused on the CCNH Medicaid resident population for most of its analysis, because this group accounts for the vast majority of Medicaid residents, and CCNH facilities receive higher Medicaid reimbursement.

RUG categories. Table IV-1 categorizes Medicaid residents and other residents (includes Medicare, private pay or other payment source) into one of the seven broad RUG categories. The majority (45 percent) of Medicaid residents fall into the "reduced physical function" category, followed by "clinically complex". The largest category for non-Medicaid residents is also "reduced physical function" (23 percent), followed by those in need of special rehabilitation. It is likely that the 2,563 residents in the "special rehabilitation" category are primarily Medicare residents, since Medicare requires a high level of rehabilitative services in order to be eligible for nursing home care.



Table IV-1. Nursing Home Residents by Major MDS Category.

Classification Category


Medicaid Residents

% of Total Medicaid Residents

Non-Medicaid Residents

% of Total Residents

Special Rehabilitation





Extensive Services





Special Care





Clinically Complex





Impaired Cognition





Behavior Problems





Reduced Physical Function










1,369 Medicaid residents in RHNS facilities are not included in the above analysis. The majority fell into the reduced physical function category, followed by the impaired cognition group.

Source: MDS DPH September 30, 2000.

Facility case mix. Using the MDS data, the committee staff calculated RUG scores for each nursing home resident, applied the Medicaid RUG weights established by CMS, and aggregated these by facility to determine each facility's case mix. It should be noted that in any population, 1.0 would not represent the average case-mix index. The reason for this is that the studies to develop the weights were biased to heavier care residents because of the greater resources used, relative to other categories. Therefore, groups that require little nursing time are weighted below 1.0.

Table IV-2 arrays the number of facilities by three measures - facilities' case-mix index, direct care costs, and Medicaid per diem rate - and shows the number of facilities falling within each quartile (shown on the right). For example, 60 facilities (the bottom 25 percent) have direct care costs at or below $75.65 per day, while the top 25 percent are at or above $93.11 - a difference of $17.46 per day.



Table IV-2. Number of Facilities within each Percentile.


Case Mix Index


Direct Costs


Medicaid Rate


25th Percentile

.926 / 59

$75.65 / 59

$143.39 / 60

50th Percentile

.957 / 118

$84.17 / 118

$157.04 / 120

75th Percentile

.996 / 177

$93.11 / 176

$173.97 / 180

100th Percentile

1.272 / 236

$128.08 / 235

$211.27 / 240

Source: LPR&IC Analysis.

Next, the committee staff correlated facilities' case-mix index with a variety of other nursing home measures. Possible correlation can range from -1.0 showing a strong negative correlation to 1.0 showing a strong positive correlation. A strong correlation (either negative or positive) means there is a close relationship between the two measures analyzed, but the cause of that relationship is not identified.

The committee staff selected five measures to correlate with facilities' case-mix index. Shown in Table IV-3 are the average, the minimum, and maximum ranges for each measure. Case-mix indices, based on Medicaid residents' RUG scores ranged from .77333 to 1.27214. On average, allowable direct care costs in CCNH facilities were $84.69. This is the cost component that is typically adjusted in Medicaid case-mix reimbursement systems because it includes nurse and aide costs, the category that research has found is most impacted by resident acuity.



Table IV-3. Analysis Variables used in Correlation.





Medicaid Rate




Direct Costs




Case Mix Index




Direct Care Staffing hours per Resident Day









Total Direct Care Staff 1




1Total direct care staff hours per day cannot be calculated by adding the columns because the numbers relate to different facilities.

Source: DSS, MDS, and Legislative Program Review and Investigations Committee analysis.

Based on correlation analysis conducted, the committee finds:

Furthermore, the committee finds not only is there no correlation between facilities' case mix and direct costs, but there is wide variation in direct costs, even when facilities have similar case-mix indices.

To examine the relationship between case mix and direct care costs more closely, the committee staff arrayed facilities' direct care costs and classified them into thirds. The bottom third was designated as low-cost facilities; the middle third as mid-cost; and the top third as high-cost. Facilities' case-mix indices were similarly arrayed and assigned to low, mid, and high case-mix categories. The analysis shows that of the 75 facilities with high costs, fully 25 percent had low case-mix indices, while another 40 percent had mid case-mix indices. Only about one-third had a high case-mix index to match the high costs. Conversely, of the 77 facilities classified as low-cost, a full 26 percent had a high case-mix index. Table IV-4 presents the full results of the analysis.


Table IV-4. Comparison of Case Mix and Direct Costs by Category.

Direct Costs

Case-Mix Index


Low Mix


High Mix

Low Cost










High Cost










Source: LPR&IC Analysis.

Based on the above analysis, the committee concludes there is no relationship between acuity and cost. The reason for this is Connecticut's Medicaid reimbursement system has never examined acuity as a factor in assigning costs. The only factor in evaluating costs has been a facility's past cost experience. Thus, high direct care costs are recognized by Connecticut's reimbursement system (up to 135 percent of median costs), but the basis for those costs are a result of: high historical costs that were built into the rate structure in 1992; interim rate approvals based on financial hardship but not on case mix since it is not a factor in the approval process, profit status; and union status. Another limitation of the system is that quality of care is not considered into the rate calculation, so there is no way to assess if facilities with high direct care costs, in reality, provide higher quality care.

The committee recognizes there are several impediments to adopting a case-mix reimbursement system. These include a number of different issues.

However, although both the union and industry oppose adoption of a case-mix system, the committee believes the extent of disconnect between resident acuity and Medicaid reimbursement poses unfairness and inequity that cannot be ignored. For example:

The committee concludes there are too many barriers to implement a complex, full-scale case-mix system using all 34 RUG-III categories to adjust direct care costs at this time. However, the committee recommends a simple case-mix reimbursement system be adopted that will begin to establish a link between allowable direct care costs, facilities' case mix, and the ultimate Medicaid per diem rate received.

Therefore, the committee recommends the following reimbursement approach:

A resident case-mix Medicaid reimbursement system shall be adopted by the Department of Social Services beginning in FY 04 for chronic and convalescent nursing homes and rest homes with nursing supervision. The case-mix system shall be implemented in the following manner:

First, facilities shall be separated into the peer groupings that currently exist - by license type, and by Fairfield county and the rest of the state.

Second, for years in which nursing home costs are rebased to set Medicaid rates, RUG scores shall be calculated by the Department of Social Services, in conjunction with the Department of Public Health, for each Medicaid resident in a nursing home. The RUG score shall be based on any full MDS assessments within the last cost-report period. The case-mix weights established by the Centers for Medicare and Medicaid Services appropriate for the 34-group RUG-III classification shall be applied to the calculated RUG to establish each facility's average Case Mix Index for the cost-report period used to rebase costs. If a Medicaid resident has more than one RUG group for the year, because of a significant change in health or functional status, the case-mix weights shall be applied to each group and weighted for the Medicaid days the resident was in each group.

For the purposes of determining allowable direct care costs under the Medicaid reimbursement system, three case-mix peer groups shall be established for each level of nursing care. All facilities' case-mix indices shall be arrayed and the case-mix peer groups shall be as follows:

Direct care costs shall be arrayed for each case-mix peer group and per diem maximum allowable direct care costs for each group shall be equal to:

As discussed in the Chapter Three, establishing cost ceilings on various categories of nursing home expenditures is one way the rate-setting system contains costs. Under the current system, allowable direct care costs are arrayed first by peer group (Fairfield county nursing homes are separated from facilities located in other counties) and cost ceilings are then established at 135 percent of median direct care costs for each licensure category. Costs a facility incurs above the ceilings are not included in the rate calculation.

The committee believes including case mix as a peer group to calculate allowable direct costs will begin to address some of the inequities present under the current system. In addition, allowing higher direct care cost ceilings for facilities that serve the neediest residents provides an incentive to care for those residents, and recognizes that higher direct care staffing ratios are needed.

In addition, the recommendation will ensure the reimbursement system bases a portion of a facility's rates on what research has shown to be more legitimate reasons for cost variations. The committee believes verification of case-mix groupings can be done through the current auditing function as recommended in Chapter Five, and therefore, should not add administrative costs to the system.

Furthermore, the program review committee believes tracking of resident acuity is necessary to develop the state's long-term care plan and formulate a methodology to determine bed need and evaluate the admissions assessment tools. This information will help gauge how well the state is meeting its objectives to promote community-based alternatives and service people in the least restrictive settings. Integrating cost and acuity data helps policymakers and regulators better understand the industry and the population being served, so that planning, oversight, and reimbursement methodologies can be improved.

8 Medicare uses a 44-group RUG-III version to account for the greater use of rehabilitation services.


Return to Table of Contents