This study was requested by Senator Joel Gougeon, Chairman of the Senate Appropriations Subcommittee on Community Health, based on rising claims and concern that the transition to Medicaid managed care was causing significant financial distress to Michigan hospitals and physicians. Questions concerning the study or its findings may be directed to its authors at the Senate Fiscal Agency.

Abstract

Background

Methodology

Results

Discussion

Endnotes

Addendum

There have been increasing strident claims from a number of sectors that the recent transition of the State's Medicaid program to a mandatory managed care system has resulted in significant financial losses to the actual providers of Medicaid medical services and is jeopardizing the entire medical care delivery system in this State. Despite funding increases and process changes adopted by the Legislature for fiscal year (FY) 1999-2000 (as well as additional increases for FY 2000-01) and further adjustments proposed by the Senate and the Engler Administration, these claims have not abated. Therefore, the Senate Fiscal Agency (SFA) was requested to undertake a study to assess the validity of these claims. The SFA developed an explanatory model to compare actual Medicaid expenditures with expected expenditures, based on a set of explicit variables and assumptions, as a means of exploring this issue. While it would be presumptuous of any study to state an absolute finding when the results are based on potentially contestable assumptions, the results of this study are sufficiently compelling to assert that they do not support the claim that the transition to Medicaid managed care, per se, has had a significant negative financial impact on Medicaid medical services providers.

Managed health care delivery systems, while relatively new to the heartland of this country, have a long history on the West Coast where, in effect, they have been the health care delivery system. Following a sustained period of double digit growth in health care expenditures during the 1980s and early 1990s, the remainder of the Nation began to turn its eyes toward the concept of managed health care systems.

As states found that their Medicaid programs were not immune to these expanding health cost pressures, they also began to explore alternatives to the existing fee-for-service system. The promise of managed care was that it would provide preventive health care services in the short run, thereby reducing the need for more costly services in the long run and rationalize (though some claim ration) the receipt of health care services such that the user population would get only those health services that they actually required. There was an additional allure for state Medicaid programs based on the theory that low-income persons would especially benefit by being placed in managed care systems. This was posited from the fact that poor individuals were less able to join access to and maneuver through the traditional system, and were much more prone to seek health care services on an episodic basis at an advanced stage of morbidity and generally in higher cost venues (hospital emergency rooms). Through the provision of a central source of care, it was felt that the health status of this population would be enhanced, with the added benefit of reducing the cost to states in providing Medicaid coverage. So began the transformation of the health care delivery system in both the private and public sectors.

It may come as a surprise to some that Michigan's Medicaid program has a two-decade history with managed care. In the 1980s the program had voluntary enrollments in health maintenance organizations (HMOs), capitated clinic plans, and a physician sponsor plan where recipients were assigned to a "gatekeeper" physician (though all medical services received under this plan were paid on a fee-for-service basis). As the 1990s unfolded, the State began to expand the breadth of the managed care system and placed a greater emphasis on enrolling recipients in these plans. Finally, an Executive decision was made to convert Medicaid to a mandatory managed care system and the Administration received a waiver from the Federal government to do just that.

Possibly, if the process had been that straightforward, the issues being raised today would not be as great of a concern as they are. There were complications, however even before the contracts with the QHPs were signed. The most notable was that DCH found itself at risk of ending up in court as some plans, not receiving contracts, contended that they were being unfairly shut out of the process. While the issues involved were not as simplistic as the previous statement would indicate, the final result was that QHP contracts were opened up to any entity that could meet established bid ranges.

The effect was that the estimated 750,000 lives eligible for coverage under managed care were to be spread across nearly 30 plans. It is axiomatic that the ability to manage risk under an insurance plan that is based on the "average" cost of a population, diminishes as the number of lives covered by any given plan decreases. In addition, some QHPs were probably overly optimistic as to their ability to operate in this new environment. As such, there were certain dislocations and consolidations that were and are necessary as the statewide system settles in. Finally, the Medicaid health care provider community, which was fairly accustomed just to providing a service, submitting a claim, and being reimbursed by a single entity with a well-established set of rules, found itself having to deal with a multitude of plans, many of which had different expectations and requirements.

It does need to be stated, in hindsight, that there are certain structural deficits relative to the Medicaid managed care system. As already noted, there are too many QHPs. Other problems include: a lack of adequate dispute resolution processes for conflicts between providers and QHPs; the absence of an inherent risk adjustor for the Aged, Blind and Disabled (ABAD) population; and no process for risk sharing if there is a sudden and unexpected occurrence related to a basic covered health service. In the instant case, this occurrence would be large increases in prescription costs. That these problems are important considerations cannot be overstated, as evidenced by the weight that the Senate has placed on them in its recent managed care reform proposal. It is also recognized that these items are having a significant financial impact on many QHPs and could produce systemic damage if left unattended.

At this time, however, the SFA cannot find that these factors have caused Medicaid provider revenue reductions in the magnitude so claimed. They have been mentioned here because they are real, they do need to be dealt with, and they have certainly helped produce an atmosphere so charged that speculation easily abounds. If a culprit does exist, however, it will have to be sought beyond these occurrences.

Without a doubt, the hardest part of this undertaking was defining the question that the SFA was trying to answer. This was primarily due to the fact that most of the claims are absent a context.¹ A statement that the Medicaid budget is . . . "under funded by $400 million to $800 million a year" leaves a fair amount of guesswork, namely "under funded" relative to what? Because the ongoing discourse has occurred during and after the last budget cycle, the SFA believes that the issue and question can be framed as follows: "Factors related to the transitioning of the State's Medicaid program from a fee-for-service system to a managed care system have resulted in financial losses to hospital and physician Medicaid providers in the hundreds of millions of dollars. Can this statement be verified?" Given this focus, the SFA was able to define a model to test this hypothesis.

After some consideration, it was decided that the best base for comparison was an examination of aggregate Medicaid payments made for hospital, physician, and pharmaceutical services, whether through fee-for-service or through managed care². This approach has the benefit of examining total dollars flowing from the Medicaid program to the providers directly affected by the shift to managed care, which, after all, is fundamental to the issue.

There were good data available from the "Medical Services Expenditures by Payroll" report published monthly by the Department of Community Health. One adjustment was made to the data; the psychiatric hospital portion of the Hospital Services expenditure was subtracted from the data. This funding is related to mental health services, not the physical health services which were covered in the managed care bid-out. From there the process was rather straightforward: Take the FY 1993-94 actual expenditures and update them through the subsequent years by making adjustments for those variables that could affect the level of expenditures. In this analysis the variables are caseload changes, caseload mix changes, and inflation. There were excellent data available from the Family Independence Agency (FIA) and DCH on caseload and caseload mix, while inflation data were obtainable from the Federal Bureau of Labor Statistics (BLS).

Caseload changes are relatively easy to deal with; as the overall Medicaid caseload changes, up or down, the resultant expenditures should change as well. Similarly, inflation numbers are fairly easy to apply with the assumption that as the price level rises so will expenditures³. Caseload mix was a more complicated issue to handle, but the detail in the data made it possible to make appropriate adjustments. The more expensive ABAD cases increased relative to the rest of the Medicaid caseload during the period in question; thus, due to the change in case mix, expenditures would be expected to rise between FY 1993-94 and FY 1999-2000. Finally, within the Family Independence Program (FIP) portion of the Medicaid caseload, there was an increase in the percentage of children eligible (and a corresponding decrease in the percentage of adults). This was largely due to expansions in Medicaid coverage for children. As children are less costly to cover than adults, this would result in a decline in expenditures for the change within the FIP portion of the Medicaid caseload.

After making these adjustments, expected expenditures from FY 1993-94 through FY 1999-2000 could be estimated by adjusting the FY 1993-94 base year using the actual changes that occurred in the caseload, caseload mix, and inflation through FY 1999-2000. The estimated expected expenditures for each fiscal year could then be compared with actual expenditure and the FY 1999-2000 appropriation for these lines and a judgment could be drawn as to whether there have been financial losses to providers in the range of hundreds of millions of dollars.

Table 1 displays the actual expenditures over the study period, as well as the number of individuals eligible for Medicaid ("eligibles") for September of each fiscal year. An "average" cost per eligible is computed from these data and as can be seen, rose and then flattened out over the last three fiscal years. However, it should be noted that the caseload declined about 12% from FY 1993-94 to FY 1999-2000; thus, the "base" number had to be adjusted to reflect caseloads in subsequent years.
 

As previously implied, a simple adjustment based on the nominal change in caseloads could understate or overstate its impact on expenditures. Therefore, these data were disaggregated and are displayed in Table 2.
 

TABLE 2 CHANGES IN THE COMPOSITION OF THE MEDICAID CASELOAD
	FY 1993-94	FY 1994-95	FY 1995-96	FY 1996-97	FY 1997-98	FY 1998-99	FY1999-2000
FIP	900,700	850,400	811,500	790,700	764,600	741,500	719,000
ABAD	244,300	254,800	263,900	264,800	267,100	269,700	272,000
OAA	50,600	54,700	56,300	57,400	58,100	61,000	64,000
Total Caseload	1,195,600	1,159,900	1,131,700	1,112,900	1,089,800	1,072,200	1,055,000
FIP %	75.3%	73.3%	71.7%	71.0%	70.2%	69.2%	68.2%
ABAD %	20.4%	22.0%	23.3%	23.8%	24.5%	25.2%	25.8%
OAA %	4.2%	4.7%	5.0%	5.2%	5.3%	5.7%	6.1%
Source: FIA

The relative increase in the ABAD population (from 20.4% to 25.8% of the caseload) was used to generate an adjustment to reflect the increase in higher-cost cases (i.e. ABAD cases receive capitation payments about three times higher than the payments for FIP and Old Age (OAA) cases). The net effect of this change, by FY 1999-2000, is about a 7.3% increase in costs due to this shift in case mix. The other case mix adjustment, related to the increase in children in the FIP eligibility group, is a decrease in the average cost of an FIP case of about $120 per year. This is due to the increase in the percentage of children in the FIP eligibility group from 67% to 77% and the fact that FIP eligibility group children cost less to cover than FIP eligibility group adults. The end result of all these adjustments was applied on the "base" FY 1993-94 expenditures.

As can be seen in Table 3 and Figure 1, expenditure need, adjusted for caseload, declined by $200 million between FY 1993-94 and FY 1999-2000. This indicates that caseload-related factors would account for a significant decline in spending relative to base year expenditures.

Federal Medicaid law does not require, per se, a systematic update in Medicaid provider payment rates for inflation. Nevertheless, inflation is a real factor that the Legislature does adjust for from time to time. We have already noted the difference between a medical inflation index based on the Producer Price Index (PPI) and the Consumer Price Index (CPI) in endnote 3. Because the issue at hand has an underlying aspect of "magnitude", the SFA includes the effect not only of inflation, but of inflation as measured both by the PPI and CPI. Table 4 lists the PPI and CPI values over the study period and the expected level of expenditures when the FY 1993-94 base is adjusted for both indexes. Finally, these values are compared with actual expenditures and the resultant differences can be seen in Figure 2.

After updating the FY 1993-94 expenditure number for caseload, case mix, and inflation, the end result for FY 1999-2000 is less than the actual FY 1999-2000 appropriation, no matter which inflation index is used. If the hypothesis that expenditures have been reduced by hundreds of millions of dollars were true, the updated FY 1993-94 expenditure base should have exceeded the FY 1999-2000 appropriation by that amount. Since this most clearly is not the case, the claim that Medicaid expenditures have been reduced by hundreds of millions of dollars is not supported by the data.

There are certainly many questions that may be raised about a study such as this. Any study will involve assumptions and this one is no exception. One of the most important assumptions is the choice of inflation index. As the Producer Price Index reflects actual increases in costs to providers, the SFA believes that it is a more valid inflation index for a study such as this one. Again, there is no requirement that provider rates be inflation adjusted but the study modeled how the numbers would look with either the PPI or the CPI applied to the data.

There are, as always, other concerns about the methods, not to mention issues that may not have been considered. Observations on some of these possible concerns follow.

a) "Prior to the increases provided in the FY 1999-2000 DCH budget, if one uses the CPI index, one sees that there was a $117 million deficit in FY 1998-99. Those sorts of costs can pile up over time; that is a debt that must be managed."

    This is a legitimate point; the cost of such a debt is the price of money; i.e. the       interest rate. In the case of a hypothetical $117 million deficit, the annual cost in subsequent years of managing that debt is roughly $10 million per year.

b) "The FY 1999-2000 numbers used are appropriated. Wouldn't it be more accurate to use the projected expenditures?"

    Yes. It does appear that actual expenditures will not rise to the level of FY 1999-2000 appropriations and, if significant, an adjustment downward (which would tilt the bottom line more toward the deficit side, dollar for dollar) would be appropriate. If, for instance, the surplus in Medicaid this year were $50 million Gross, then the net differences seen above would tilt toward the deficit side by $50 million more than in the estimates above.

c) "The use of QHP expenditure data does not take into account the fact that the money going to QHPs does not all go to providers; some stays with the QHP."

    This is also a valid point. Since the hypothesis related to the financial status of direct providers, not the financial status of QHPs, there should be an adjustment for money that does not flow through QHPs to providers. The share of expenditures paid through QHPs rose from 19% in FY 1993-94 to 52% in FY 1999-2000. If 10% of that increased QHP funding were not passed on to providers, that would equate to 3.3% of the total expenditures or about $80 million by FY 1999-2000.

Accepting all three of these points in their totality (while ignoring issues that point in the other direction) requires the addition of about $140 million to the FY 1999-2000 "difference" numbers above. Adding $140 million to those numbers still leads to a net surplus in the PPI-derived expenditure of around $70 million. In other words, even after conceding all of the reasonable points (and ignoring reasonable points that lean the other way), there still is no evidence of massive losses when the more appropriate inflation measure is used.

Adding $140 million to the CPI-derived expenditures finally leads to a FY 1999-2000 deficit, but one that totals $120 million (rather than hundreds of millions of dollars). Even in this case, after conceding several points for the sake of argument, the hypothesis that providers are facing annual losses of hundreds of millions of dollars is not supported by the data. Furthermore, even a $120 million deficit is about 5% of total spending, which is at or below the typical savings expected from a shift to managed care.

Again, even after addressing concerns of those who support the hypothesis of annual losses in the hundreds of millions of dollars, there is nothing in the data supporting the claim of losses of such a magnitude. In fact, if one looks at an exercise that accepts the concerns above, but updates the data for FY 2000-01, including the Executive's proposed 4% rate increase in the data (adjusting for a continued decline in caseload; continued changes in case mix; and a continuation of current inflation trends) the claims cannot be justified. The results are shown in Figure 3 below.

As the graph indicates, even after conceding the above three points (while not acknowledging any points leaning the other way) and even after using the less appropriate (but more "deficit-friendly") inflation index, there is no "deficit". The FY 2000-01 rate increases included in the FY 1999-2000 DCH budget would be sufficient to ensure that payments to providers remained at least as high, adjusted for caseload and inflation, as they were in FY 1993-94. The claim that there have been hundreds of millions in reductions to Medicaid providers, again, is simply not supported by the data.

2) By using these Medicaid expenditure lines, the SFA believes it is capturing nearly 100% of the costs related to these medical services that managed care enrollers would use. In addition, it also clarifies the overall amount of Medicaid revenues received by the physician and hospital provider community.

3) Regardless of the validity of providing any "inflation" update to Medicaid expenditures, the SFA believes that the most relevant factor to use is the PPI versus the CPI. In its simplest form the CPI reflects unadjusted provider charges to the consuming public in urban areas only. On the other hand, the PPI, in effect, measures the change in prices related to the movement of a person through an episode of care, however that may be defined.

Analysis of the Health Management Associates Study of Medicaid Expenditures

The Health Management Associates (HMA) study appears to claim that Medicaid payments in FY 2000-01 will be over $350 million less than they would have been under the previous fee-for-service system.

HMA arrives at this conclusion using the following assumptions:

First, HMA uses fee-for-service rate estimates from FY 1997-98 that were produced by the Medical Services Administration (MSA). HMA then updates the rates by 6.5% in FY 1998-99 and FY 1999-2000 and 9% in FY 2000-01 for caseload, utilization, pharmaceutical inflation, and the two years of 4% rate increase for Qualified Health Plans (QHPs).

Given those assumptions, HMA proceeds to estimate what the fee-for-service cost would be for the Medicaid population had fee-for-service costs increased by the assumed amounts. That leads to an estimated FY 2000-01 fee-for-service expenditure on the population currently covered by QHPs of about $1.71 billion. HMA estimates the FY 2000-01 cost under the current system at $1.36 billion, and thus concludes that the shift to managed care has led to a reduction in Medicaid expenditures of some $355 million.

The SFA has examined the HMA analysis and offer the following comments.

First, it is important to note that rates, inflation assumptions, and so forth are interesting, but what drives expenditures are expenditures. The HMA analysis starts with estimated fee-for-service equivalents to the expenditures made. This may closely approximate expenditures, but that is not absolutely clear.

The SFA looked at the inflation numbers that were used. The analysis separates out pharmaceutical costs, then possibly overstates the level of pharmaceutical inflation. HMA assumes pharmaceutical inflation between FY 1997-98 and FY 2000-01 at about 20% per year. While the Pharmaceutical Producer Price Index (PPI) for FY 1997-98 increased by about 20%, the Pharmaceutical PPI has been flat ever since, with an increase of about 0.5% per year. Even the prescription drug component Consumer Price Index (CPI) only broached 6% for the first time during December of 1999. Based on actual rates of inflation (using either the PPI or CPI) the HMA inflation adjustments are difficult to defend.

The HMA analysis also assumes a 2% increase for an "intensity factor" that apparently is based on caseload and a "sicker" Family Independence Program (FIP, the former Aid to Families with Dependent Children program) portion of the Medicaid caseload. HMA argues that, due to welfare reform, those who go to work are likely to be healthier than those who remain on welfare, thus the average cost per FIP Medicaid eligibility group client would increase.

The SFA would note that due to the expansion of Medicaid benefits and the low-wage jobs into which welfare recipients are generally placed, the vast majority of FIP clients (both adult and child) maintain Medicaid eligibility (the FIP Medicaid eligibility group includes both FIP clients and low-income adults and children whose income is too high to qualify for FIP but low enough to qualify for Medicaid). This means that drawing distinctions as to which portion of the population is healthier is not particularly important, as the vast majority of the population remains Medicaid-eligible.

Finally, the SFA does not believe that including the increases in fee-for-service rates is appropriate. In the theoretical world being explored in the HMA analysis, there was no shift to managed care. The decision to raise rates last year did not occur in a vacuum. One of the main reasons for the fee-for-service increases last year (as well as the QHP increase) was the claim by providers that they were losing money on managed care. If there is no managed care, there is no such claim and, likely, there is not an increase of 4% in the rates. Furthermore, providers lose money only if their revenue (i.e. Medicaid expenditures) grows more slowly than their costs. Cost increases are most appropriately modeled by using caseload changes and an inflation index, not by applying specific fee screen updates.

We believe that caseload shifts are best approximated with a 1.5% per year decrease in expenditure need and that cost increases are best approximated by the Medical PPI. We believe that those factors, when applied and compared to actual expenditures, adjusted for the amounts that cover QHP administration, result in an adjustment far smaller than the $350 million produced by the HMA analysis.