APPENDIX B
Medicare Clinical Laboratory Payments: The National Limitation Amount and Its Relationship to Payment Amounts
Katie Merrell
Center for Health Administration Studies, University of Chicago
Under current policy, Medicare payments for ambulatory clinical laboratory services are based on 56 regional fee schedules, subject to a service-specific national limit. Each regional fee schedule is based on charges in 1984, subject to a series of annual reductions and updates since then. The median of these updated base payment rates is calculated for each service to establish the National Limitation Amount (NLA), which is currently set at 74 percent of the median. Actual payment for a particular service in a particular area is then equal to the lesser of the regional base rate and the NLA. This appendix explores the relationships among regional fee schedules, the NLA, and actual payment amounts. By comparing current laboratory payments to the NLA, it provides estimates of the financial implications of an NLA-based national fee schedule.
In 2000, nearly 84 percent of payment amounts (at the region service level) were set at the NLA.1 This suggests that there is effectively a national fee schedule, where relative service payments are determined by the relationship between median charges in 1984 across the carriers, coupled with gap-filled and cross-walked values for codes established since then.2
Analysis of 2000 payment rates reveals that the high prevalence of the NLA in determining payments suppresses the variation in payment amounts across carriers. More than 16 percent of updated base amounts are at least 25 percent less than the median base amount, while more than 21 percent exceed the
1 |
The 2000 fee schedules were downloaded from the Health Care Financing Administration’s (HCFA) Web page (http://www.hcfa.gov/stats/pufiles.htm) in February 2000. |
2 |
Gap-filling and cross-walking are two techniques used by HCFA and its carriers to develop fee schedule values for new services. |
median by at least 25 percent (Figure B.1). In other words, nearly 40 percent of the values in carrier fee schedules differ by more than 25 percent from the relevant service median value. Those that fall well below the median will be paid at rates below the NLA, while all of those above 74 percent of the median will be paid at the NLA, reducing the effective geographic differences in payments. The NLA is set at 74 percent of the median, so only 16 percent of service payment amounts are less than the NLA. The base rates, however, exhibit geographic variation not reflected in actual payments. For some, such as non-automated urinalysis (CPT 81000), base amounts are tightly clustered across the regional fee schedules (coefficient of variation = 16 percent), while, for others, such as HIV-1 (CPT 87536), there is considerable spread in base rates (coefficient of variation = 67 percent) (Table B.1). With the exception of the HIV-1 test listed, payment for 20 services studied, which include high-volume Medicare services and others of particular policy interest, is set at the NLA in at least 80 percent of carriers; for three services (digoxin assay, parathormone assay, and Pap cytopathology thin layer preparation), all payments are at the NLA.
The NLA is based on an unweighted median of regional fee schedule amounts. As a result, it is not the median value of actual payments for each service since service volumes vary across regions. The median value of actual lab payments for a particular service may be much higher (or lower) than the median used to set the NLA. A simple three-region, three-service example illustrates this (Table B.2).
In this example, all fee schedule values in Regions 2 and 3 exceed the NLA, so all payments in these two regions would be set at the NLA. In Region 1, payments for Services A and B would be set at the regional fee schedule amount while payment for Service C would be capped at the NLA. Overall, 77 percent of payment amounts at the region service level in this hypothetical system would be at the NLA.
TABLE B.1 National Limits and Updated Carrier Base Rates, Selected Laboratory Services
TABLE B.2 Illustrative Regional Fee Schedules and National Limitation Amounts
TABLE B.3 Service Volumes
|
Region 1 |
Region 2 |
Region 3 |
Service A |
100 |
1,500 |
1,000 |
Service B |
500 |
4,500 |
4,000 |
Service C |
1,000 |
5,000 |
4,000 |
This is not a particularly helpful number, however, because service volumes vary across services and regions. Service C is at the NLA in all three areas, so if it is a high-volume service, the actual number of payments at the NLA would be greater than 77 percent. Similarly, payments in Regions 2 and 3 are all at the NLA, so if they account for more than two-thirds of service volume, the actual number of services paid at the NLA would exceed 77 percent.
Given service volumes, in fact, it would be possible to figure out exactly what percentage of services are paid at the NLA, what share of total spending is at the NLA, and how much more money would be required to pay for all services at the NLA. Based on a set of volumes for the hypothetical system described above (Table B.3), these various measures can be calculated. The resulting estimates indicate the financial implications of an NLA-based fee schedule.
These service volumes imply that, in the hypothetical system, more than 97 percent of service payments would be at the NLA, compared to only 77 percent of regional payment amounts. The share of spending at the NLA would be almost 1 percent lower, because Services A and B, which are not at the NLA in Region 1, account for relatively more spending than volume. In this example, total spending is about 99.4 percent of what it would be if all services were paid at the NLA.3
In the case of actual Medicare payments, the simple fact that 84 percent of carrier payment amounts are set at Medicare’s NLA does not provide a very accurate estimate of how close current payments are to an NLA-based fee schedule. First, even those services not paid at the NLA may be paid very close to the
3 |
Multiplying the payment amounts (the lesser of the fee schedule amount or NLA) in Table B.2 by the volumes in Table B.3 suggests that total spending in this system was $162,440. Multiplying the volumes in Table B.3 by the NLA in Table B.2 shows that if all services were paid at the NLA, then total payments would be $163,392. |
NLA. In fact, payment levels not at the NLA average 75 percent of the NLA, across services and carriers. Second, since volumes vary across services and regions, payments for the actual mix of services used by beneficiaries may be closer to (or farther from) the NLA. The ideal way to get a more accurate estimate of how actual payments relate to the NLA would be to use data on service volumes within each carrier to repeat the calculations illustrated above. Unfortunately, reliable service volume data are unavailable at the code-carrier level.
Reasonable estimates can be made, however, based on available data, namely, the distribution of beneficiaries across carriers and of total payments for each lab code. The number of fee-for-service beneficiaries in each carrier area can be used to get some sense of the distribution of service volumes for each code across carriers, although this approach ignores any important regional variation in per capita service use across services.4 With this distribution, a weighted mean of regional payment amounts can be calculated, which can then be compared with the NLA. The resulting “NLA ratio” tells, in essence, how the mean payment for that service relates to the NLA. In turn, the mean of this service-level NLA ratio, weighted by total spending for each service, would reveal how close total spending is to what would occur under the NLA.
The modified calculation is perhaps most easily illustrated by revisiting the calculations in the above example. Instead of service volumes (Table B.3), imagine that only the distribution of beneficiaries across the three areas and the total spending for each service are available (Table B.4).
A slightly different approach is required to calculate measures such as those reported earlier in the absence of service volumes. First, the distribution of beneficiaries can be used to calculate a service-specific, population-weighted payment amount across the three regions (Table B.5).
The ratio of this amount to the NLA can then be calculated (Table B.5). Finally, the mean of this ratio, weighted by the distribution of total spending (Table B.4), can be calculated. The estimate of 0.992 corresponds well to the earlier estimate that spending was about 99.4 percent of what would occur under the NLA. This alternative approach, although somewhat complicated, appears to allow for developing reasonable estimates without the benefit of service volumes but with the data that are available.
This second approach—population-weighted service-level payment amounts, NLA ratio, and service-level spending-weighted mean of the NLA ratio—can be used with available Medicare data. The distribution of fee-for-service Medicare
beneficiaries across counties is available from the Health Care Financing Administration (HCFA) and can be mapped into the 56 regions underlying the lab fee schedules. Total charge data are available for the 100 codes that accounted for the largest majority of Medicare outpatient lab spending in 1998.5 Combined, these codes accounted for more than 83 percent of spending, so an analysis of them should be fairly suggestive.6
For these top codes, the unweighted service-level NLA ratio across the 56 fee schedules is about 0.98 (compared to about 0.96 for all codes), suggesting that among high-cost or high-volume services, the fee schedule amount is closer to the NLA than for other services. The charge-weighted mean NLA ratio for these services is 0.985, so that across all payments, service payments are about 98.5 percent of the NLA.7
TABLE B.4 Population by Region and Spending by Service
|
Region 1 |
Region 2 |
Region 3 |
Total Spending |
Service A |
|
|
|
$34,600 |
Service B |
|
|
|
$105,640 |
Service C |
|
|
|
$22,200 |
% beneficiaries |
7 |
51 |
42 |
|
TABLE B.5 Service Payments and NLA Ratio
Population-Weighted |
Payment Amount |
NLA |
Ratio: Weighted Payment Amount to NLA |
|
Service A |
$13.30 |
$13.32 |
0.998 |
|
Service B |
$11.70 |
$11.84 |
0.988 |
|
Service C |
$2.22 |
$2.22 |
1.000 |
|
NOTE: The population-weighted payment amount is calculated from the payment amounts in Table B.2 (minimum of fee schedule amount and NLA) and the population distribution in Table B.4. |
5 |
This analysis is based on 1998 fee schedules, as downloaded from HCFA’s Web page (http://www.hcfa.gov/stats/pufiles.htm) in June 2000, to match the year for which spending data are available. |
6 |
Several carriers were omitted from this analysis because of problems matching with data on beneficiary counts used for weighting. Similarly, coding discrepancies between data sources limited analyses of the 100 codes with the highest allowed charges to 92 codes. These omissions are unlikely to have important implications for the simple analyses described here. |
7 |
There is a slight error in this estimate because the total spending weight reflects actual payment levels rather than the NLA. The high correlation between the NLA and the calculated service payment amounts suggests that this error is inconsequential, particularly because spending is used as a weight and not as a measure of absolute levels. |
One way to interpret these results is to use them to anticipate the implications of an NLA-based fee schedule. Based on these estimates, an across-the-board NLA reduction of about 1.5 percent would be necessary to create a budget-neutral lab fee schedule based on current NLAs. Conversely, Medicare outpatient lab spending would increase about 1.5 percent if all payments were raised to the NLA.
This estimate is based only on those high-cost or high-volume services that account for most of Medicare spending. The unweighted estimate for all services implies that a reduction of about 4 percent would be necessary, but this is an upper bound that is likely to dramatically overstate the correct amount. Although 1.5 percent is a lower bound, the correct adjustment will lie much closer to 1.5 than to 4 percent, because of the large share of spending accounted for by the services studied. More accurate estimates could be calculated easily from data on total spending for all codes or, better still, service volumes for each code in each region. The present estimates, however, provide fairly strong evidence that Medicare’s present payment policy is, in effect, an NLA-based fee schedule.