Appendix B.-CALCULATION OF CONVERSION FACTOR This appendix specifies the data base, adjustments, and behavioral parameters to be used by the Secretary of Health and Human Services in calculating the conversion factor for 1990 under section 4001(a)(5) of the bill in a manner that will maintain budget neutrality after taking into account changes in the volume of services that might occur as a response to payment changes specified for 1990 in this section of the bill. The model differentiates between "first order" or prebehavioral estimates of the conversion factor and "second order" of postbehavioral calculations. First order estimates yield a conversion factor that takes into account the adjustments in prevailing charges that would be made in each of the services listed in appendix A. However, the first order estimate does not take into account changes in the mix, intensity, and volume (hereinafter simply referred to as volume) of services that might result from behavioral changes that occur as a result of these adjustments. The purpose of the second order adjustment is to account for such changes in volume as necessary to maintain budget neutrality. Behavioral changes could reflect altered demand for services by patients or changes in demand inducement by physicians in response to new Medicare payment amounts and new limits on actual charges. Estimated behavioral responses are not symmetric, however. Practices whose receipts would fall because of a payment change by Medicare would have a larger offsetting behavioral response than would practices whose receipts would increase. In particular, the offset to a reduction in payments is larger (55.5 percent) than the offset to an increase in payments (37.5 percent). In application, these offsets are adjustments to the underlying rate of growth in the volume of Medicare services per enrollee, so that volume growth would temporarily accelerate for practices that were "losers" under the payment change, while growth would decelerate for "winners." This appendix consists of three sections. The first section provides instructions on selecting and adjusting the data base to be used in simulating the payment change. The second section contains instructions on how to calculate the first-order conversion factor. The third section describes the iterative process used to adjust the first-order conversion factor to take account of behavioral responses so as to achieve budget neutrality. I. DATA SELECTION AND ADJUSTMENT The data base to be used is the part B Medicare Annual Data provider (BMAD III) file for calendar year 1987. This is a 5-percent sample of all Medicare providers. Two carriers-for Railroad Retirees and for Puerto Rico-are eliminated because of difficulties in defining appropriate cost indexes for the claims they process. In addition, all supplier specialties are eliminated because they are unaffected by the payment changes here. Although only the services listed in appendix A will be affected by the payment changes to be simulated here, all Medicare services provided by each practice must be retained in order to estimate behavioral responses appropriately. Entire physician practices may be eliminated, however, when they have no claims for the affected services. This means that radiologists and anesthesiologists may be eliminated, among others. Three kinds of adjustments to this data base are made. First, prevailing charges are added to all records. Second, the data are adjusted to reflect projected participation rates and assignment rates. Finally, prevailing charges, submitted charges, and allowed charges are adjusted to reflect values as projected for 1990. Each of these adjustments is explained in detail below. Imputing Prevailing Charges. The controlling prevailing charge during 1987 is added to each record. This can be completed in the sequence described below, but in each stage the process is constrained so that the imputed prevailing charge is greater than or equal to the allowed charge. Further, adjustment is made for the 4 percent prevailing charge differential that existed in 1987 between participating and nonparticipating physicians. For example, if an imputed prevailing charge for a nonparticipating practice is obtained from a participating practice, the value imputed is 96 percent of the value obtained from the participating practice. Finally, for assistants at surgery (type of service 8), the prevailing charge is set at 20 percent of the prevailing charge for surgery (type of service 2), separately by locality, specialty, service, and participating status. First, the prevailing charge for each record is set to the allowed charge when the record's payment indicator shows that payment was set by the prevailing charge. Second, prevailing charge values for records still without a prevailing charge are imputed from other records for the same service, locality, and specialty. Where this fails, prevailing charge values are imputed from other records for the same service and locality, using the specialty most likely to provide the service. Where this fails, matching prevailing charge values from the prevailing charge file (BMAD II) applicable during calendar year 1987 are imputed. If none of these methods succeeds, the prevailing charge is set equal to the allowed charge. Adjusting Participation and Assignment Rates. Practice participation rates are adjusted so that 60 percent of allowed charges are attributed to participating practices. To do this, records for practices with any participating claims are altered so that all claims submitted by those practices are identified as participating. If this is insufficient to reach the target, then nonparticipating practices are randomly reassigned as participating practices until those practices categorized as participating account for 60 percent of allowed charges. This can be done by assigning a 7-digit random number to each nonparticipating practice and using that random number to reclassify practices as participating ones until the target is reached. For all records reclassified as participating, the assignment indicator is changed appropriately. Also, the associated prevailing charge is increased by dividing the original prevailing charge by 0.96. For the remaining nonparticipating practices, another random number is assigned to each record and used to adjust assignment rates until 50 percent of all allowed charges for nonparticipating practices are assigned and 50 percent are unassigned. Because nonparticipating practices will (after the adjustment in the previous paragraph) account for 40 percent of all allowed charges, this means that 20 percent of all allowed charges will be unassigned in the adjusted data base to this point. The assignment indicator is changed appropriately on all affected records. Adjusting Charges. Appropriate adjustments to reflect projected 1990 charges differ for participating and nonparticipating practices, as redefined above. For participating practices: A. Prevailing charges are reduced for selected procedures (denoted hereinafter as OP codes) pursuant to Public Law 100-203 (OBRA-87, section 4045). The affected procedures, and the national average prevailing charges used to calculate the reductions, are listed in table 2 of section 5254 of the Medicare part B Carriers' Manual, part 3, Claims Process. The appropriate reductions are calculated for each record using the formulas below, where MEAN denotes the national average prevailing charge for the service and PC87 and PC88 denote the prevailing charge for 1987 and 1988, respectively. Else if PC87 > .85*MEAN and if .98*PC87 < = Else PC88.98*PC87*[1-(3/13)*(RATIO-0.85)]. = so that OP records can be included with all other nonprimary services in the updates made in the following steps. This is necessary because OP services did not receive the 1 percent prevailing charge update for 1988. B. The 1987 prevailing charges are updated to 1990 values by using the multiplicative product of the updates enacted for 1988 and 1989 in Public Law 100-203 (OBRA-87, section 4042), and enacted in the bill (section 4002) for 1990. For 1988 and 1989, the updates differ for primary care and all other services. Here, primary care services include HCPCS codes 90000-90080, 90100-90170, 90300-90370, 90400-90470, 90500-90570. C. The 1987 actual charges are increased to 1989 and 1990 values by using actual or projected values of the physician fee component of the Consumer Price Index, using the same values for primary care and other services: These are the values reported in the 1989 Trustees' Report for the Supplementary Medical Insurance fund. If different values are used to compute Medicare premiums for 1990 in the 1989 promulgation notice, those actual charge update factors will be used instead. D. Allowed charges for 1990 are calculated for each record as the minimum of the updated prevailing charge or the customary charge (defined as the updated actual charge for 1989). For nonparticipating practices: E. All 1987 prevailing charges on these records are first divided by 0.96, and then step A above is replicated. F. The new prevailing charges are multiplied by 0.95, to adjust for the 1 percent increase in the nonparticipating differential between 1987 and 1990. Then step B above is replicated to increase these adjusted prevailing charges to 1990 values. G. Actual charges are increased to 1989 and 1990 values by using actual or projected values of the physician fee component of the Consumer Price Index as given in step C above, subject to ceilings shown below by year. The net effect of maximum allowable actual charge ceilings and OP ceilings are shown below as MAAC amounts which are service and practice specific. These ceilings were enacted in Public Law 99-509 (OBRA86, section 9331(b)) and in Public Law 100-203 (OBRA-87, section 4045). = MAAC87 Mean actual charge for this service by this If MAAC87< 1.15*PC88 then = MAAC88 Max(MAAC87 +0.33*(1.15*PC88-MAAC87), Else MAAC88 = 1.01*MAAC87; = If HCPCS = OP code then MAAC88 Min(MAAC88, |