Printer Friendly

Effects of the relative fee structure on the use of surgical operations.

Under the Omnibus Budget Reconciliation Act of 1989, Medicare's traditional charge-based method for determining physician fees will be replaced by a resource-based Medicare fee schedule. The fee schedule, which is being phased in over four years beginning in 1992, will increase fees for physician visits and consultations and reduce fees for most surgical and technical procedures (Physician Payment Review Commission 1990).

Changing Medicare's relative fee structure is likely to affect the use of physician services. The impact on surgical utilization is of particular practical and theoretical interest. Expenditures for surgery constitute a large portion of total Medicare spending on physician services. Moreover, there is evidence that inappropriate surgery may be commonplace (Chassin, Kosecoff, Park, et al. 1987; Winslow, Solomon, Chassin, et al. 1988; Greenspan, Kay, Berger, et al. 1988). Changes in surgical utilization under the new fee schedule could have important implications for Medicare costs and for the quality of care provided to Medicare enrollees.

In addition, some researchers believe that physicians have considerable control over decisions to perform surgery, and that surgeons "create demand" for their own financial gain (Fuchs 1978; Rice 1984; Cromwell and Mitchell 1986; Wedig, Mitchell, and Cromwell 1989). According to this theory, surgeons respond to increased competition for patients or to changes in fees by recommending and providing a different mix of services to their patients than if they acted solely in the patients' best interests. Other investigators question the importance of demand creation in physician services markets (Sloan and Feldman 1978). Changes in surgical utilization under the Medicare fee schedule also could shed light on the extent of demand creation in markets for surgical services.

To date, the effects on surgical utilization of changing the relative fee structure have received little attention from researchers. Based on a theoretical analysis, Pauly (1991) has argued that these effects are unpredictable and could differ among procedures. But only one empirical study has investigated the effect of changing relative fees on the use of surgical operations.(1) Using time-series data from Ontario, Canada, Hurley, Labelle, and Rice (1990) found that 10 of 16 operations selected for study exhibited a statistically significant utilization response to an isolated change in the fee for the operation. A fee reduction was associated with higher utilization in seven cases and lower utilization in three cases. However, the researchers did not describe their economic model or examine the effects of changes in the fees for other services on the use of the study operations.

This article presents a theoretical and empirical framework for studying the effects of changes in the relative fee structure on the utilization of surgical operations. The theoretical model considers how the use of an operation may be affected by two types of fees: the fee for the operation itself, and the fees for other services performed by surgeons in the same specialty. Analysis of the model suggests an empirical test of whether or not surgeons create demand for surgery. Econometric techniques are used to examine the effects of Medicare's relative fee structure on the use of selected, high-frequency operations by elderly Medicare enrollees.

A time-series design would be optimal for the empirical portion of this study. Until recently, however, the method used by Medicare to update physician fees precluded large longitudinal changes in relative fees, because fees for all physician services were increased yearly by a similar percentage.(2) On the other hand, recent evidence indicates that the geographic differences in Medicare's relative fee structure are considerable. Medicare fees for different services are not highly correlated across areas, and areas with high fees for one service often have low fees for other services (Pope et al. 1989; Escarce 1991). Therefore, this study uses a cross-sectional design.

The next section describes the conceptual framework for the study. The third section describes the empirical model, the estimation methods, and the data sources used in the study. The fourth section presents the results of the empirical analyses. The final section discusses the study's findings and conclusions.

THEORY

The theoretical framework for this study is based on an economic model of utilization of surgical operations performed by surgical specialists. Surgeons are regarded as multiproduct firms that maximize profits and whose products are different types of operations and physician visits and consultations.(3) The model examines how the use of an operation may be affected by two types of fees: the fee for the operation, and the fees for other services performed by surgeons in the same specialty.

To begin, consider the effect of a particular operation's fee on the use of the operation by elderly Medicare enrollees in a market area. Market demand for the operation can be described as a function of the fee for the operation. If patients do not have full insurance, a higher fee for the operation increases patients' out-of-pocket costs and is expected to reduce demand. However, recent estimates from the RAND Health Insurance Experiment indicate that the price elasticity of demand for medical care is less than 0.2 when the coinsurance rate is 25 percent and declines as the coinsurance rate decreases (Keeler and Rolph 1988). Since more than three-fourths of Medicare enrollees have either public (Medicaid) or private supplementary insurance that pays Medicare's deductibles and coinsurance (Nelson et al. 1989), the average coinsurance rate for Medicare enrollees is very low. In addition, many elders have chronic health problems, and there is evidence that the price elasticity of demand is lower for persons in poor health (Wedig 1988). For elderly Medicare enrollees, therefore, the own-price elasticity of demand for the operation is likely to be close to zero. This is shown by the vertical market demand curve, |D.sub.1~, in Figure 1.

Market supply for the operation also is a function of the fee for the operation. A higher fee for the operation makes the operation more profitable relative to other services provided by surgeons. Thus, a higher fee for the operation is expected to increase supply as surgeons devote more of their time to performing the operation and less to performing other services. The positive own-price elasticity of supply is shown by the upward sloping market supply curve, |S.sub.1~, in Figure 1.(4)

If the market for surgeons' services is in equilibrium, patients' welfare and surgeons' profits are maximized at the equilibrium fee for the operation, |P.sub.1~, and equilibrium quantity of the operation, |Q.sub.1~, shown in Figure 1.(5) Therefore, surgeons have no incentive to create demand. However, although not all observers agree (e.g., Schwartz and Mendelson 1990), some studies suggest that markets for surgical services may be not in equilibrium, but in excess supply (e.g., Graduate Medical Education National Advisory Committee 1980; Rutkow 1988).(6) The market is in excess supply whenever the fee for the operation is higher than the equilibrium fee, |P.sub.1~.

The quantity of a service actually provided in a neoclassical market in which there is excess supply is determined by the market demand curve. Thus, surgeons are unable to supply all of the quantity of the service that they wish to supply to maximize their profits. As a result, surgeons have a strong incentive to create demand. Demand creation can be understood as an attempt by surgeons to move the demand curve closer to the supply curve, as shown by the rotated market demand curve, |D.sub.1~|prime~, in Figure 1.

The preceding analysis indicates that the effect of a change in the fee for an operation on the actual use of the operation depends on whether the market for surgeons' services is in equilibrium and on whether demand creation occurs. If the market for surgeons' services is in equilibrium in every market area, or if the market is in excess supply but no demand creation occurs, then the use of the operation is determined by the neoclassical demand curve, |D.sub.1~, and a change in the fee for the operation does not affect use. This is the neoclassical prediction. On the other hand, if the market for surgeons' services is in excess supply and demand creation does occur, then the use of the operation is determined by the rotated demand curve, |D.sub.1~|prime~, and a higher fee for the operation results in higher use.

Next, consider the effect of the fees for other services provided by surgeons in the same specialty (i.e., other operations and visits and consultations) on the use of the operation of interest by elderly Medicare enrollees. For this part of the analysis, market demand for the operation is described as a function not of the fee for the operation, but of the fees for other services. For reasons similar to those discussed in connection with the own-price elasticity of demand, the cross-price elasticity of demand for the operation is likely to be close to zero. This is shown by the vertical market demand curve, |D.sub.2~, in Figure 2.

Market supply for the operation also is a function of the fees for other services provided by surgeons in the same specialty. Higher fees for other services make the operation less profitable relative to those services. Thus, higher fees for other services are expected to reduce the supply of the operation as surgeons devote less of their time to performing the operation and more to performing other services. The negative cross-price elasticity of supply is shown by the downward sloping market supply curve, |S.sub.2~, in Figure 2.(7)

If the market for surgeons' services is in equilibrium, patients' welfare and surgeons' profits are maximized at the equilibrium level of fees for other services, |P.sub.2~, and equilibrium quantity of the operation, |Q.sub.2~, shown in Figure 2 (in equilibrium, |Q.sub.1~ = |Q.sub.2~). Thus, surgeons have no incentive to create demand. However, surgeons have a strong incentive to create demand if there is excess supply, which occurs whenever the level of fees for other services is lower than the equilibrium level, |P.sub.2~. Demand creation is represented by the rotated market demand curve, |D.sub.2~|prime~.

This part of the analysis indicates that the effect on the use of an operation of changes in the fees for other services provided by surgeons in the same specialty also depends on whether or not the market for surgeons' services is in equilibrium and on whether demand creation does or does not occur. If the market for surgeons' services is in equilibrium in every market area, or if the market is in excess supply but there is no demand creation, then the use of the operation is determined by the neoclassical demand curve, |D.sub.2~, and the neoclassical prediction emerges: changes in the fees for other services do not affect the use of the operation. However, if the market for surgeons' services is in excess supply and demand creation occurs, then the use of the operation is determined by the rotated demand curve, |D.sub.2~|prime~, and higher fees for other services lead to lower use of the operation. The predictions of this model suggest a powerful empirical test of whether surgeons create demand for surgical operations.

METHODS

This study uses econometric techniques to examine the demand for surgical operations among elderly Medicare enrollees residing in the 316 Metropolitan Statistical Areas (MSAs) in the United States (excluding Alaska and Hawaii). Medicare enrollees less than 65 years old or with end-stage renal disease are excluded from the study.

STUDY OPERATIONS

The four surgical specialties that provide the most services to Medicare enrollees were chosen for study: ophthalmology, general surgery, orthopedic surgery, and urology.(8) From each specialty, two or three study operations were selected using the following criteria: high frequency of use by the Medicare population, performance almost exclusively by surgeons in one specialty, and no recent changes in the procedure codes used to identify the operations in Medicare claims data. |Study operations were defined in terms of Physicians' Current Procedural Terminology (CPT) codes (American Medical Association 1986).~ The 11 study operations were cataract extraction with intraocular lens insertion, retinal photocoagulation, laser trabeculoplasty, partial colectomy, cholecystectomy, inguinal hernia repair, internal fixation of hip fracture, total knee replacement, knee arthroscopy, transurethral prostatectomy, and transurethral fulguration or resection of bladder tumor.

EMPIRICAL MODEL

The empirical model estimated for each study operation was a structural demand equation, but modified to include the two types of fees considered in the theoretical model, as follows:

D = f(|P.sub.own~, |P.sub.cross~, A, S, X, Y, Z),

where D is the demand for the study operation, |P.sub.own~ is a measure of the Medicare fee for the operation (the operation's own-price), |P.sub.cross~ is an aggregate measure of the Medicare fees for other services performed by surgeons in the same specialty (the operation's cross-price), A is a vector of variables that affect enrollees' out-of-pocket costs for the operation, S is the supply of surgeons who perform the operation, X is a vector of health care market variables, Y is a vector of enrollee characteristics that may affect demand, and Z is a vector of additional control variables. All variables were measured at the level of MSAs.

Dependent Variable

Demand for the operation, D, was specified as a modified logit transformation of the utilization rate for the operation among elderly Medicare enrollees in the fee-for-service sector. The following transformation suggested by Cox (1970) for weighted least squares analysis was used:

D = log {|r - (1/2n)~/|1 - r - (1/2n)~}

where r is the proportion of enrollees who received the operation and n is the number of enrollees in the MSA. The transformation is undefined when r is zero, but such observations receive zero weight in the regression.

Explanatory Variables

The own-price for the study operation (|P.sub.own~) was measured as a geographic index of Medicare-allowed charges for the CPT codes that define the operation. The cross-price for the operation (|P.sub.cross~) was measured as a geographic index of Medicare-allowed charges for all of the surgical services that accounted for 95 percent of the specialty's total Medicare revenues from surgery in 1986 (excluding the study operation) plus office and hospital visits and consultations. (Services' national frequencies were used as weights in constructing the indexes to avoid confounding differences in allowed charges with differences in service mix.) Under the neoclassical model of utilization, the own-price and cross-price do not affect demand for the operation. Under the demand-creation model, a higher own-price leads to higher demand and a higher cross-price leads to lower demand.

The vector of variables that affect out-of-pocket costs, A, included the Medicare assignment rate for surgeons in the specialty, measured as the percentage of the specialty's total Medicare revenues derived from services provided on assignment, and the percentage of enrollees in the fee-for-service sector who had private or public (Medicaid) supplementary insurance coverage. Higher assignment rates and higher supplementary coverage result in lower out-of-pocket costs and are expected to increase demand.

The supply of surgeons who perform the operation, S, was measured as the ratio of surgeons in the specialty to population. Higher surgeon supply may reduce the time price of care. Surgeons in high-supply areas also may offer more amenities that patients value, or create more demand (Fuchs 1978; Cromwell and Mitchell 1986). Thus, higher surgeon supply is expected to increase demand.

The vector of health care market variables, X, included the supply of primary care physicians, measured as the ratio of general practitioners, family physicians, and general internists to population, and penetration of health maintenance organizations, measured as the percentage of elderly Medicare enrollees in HMOs. Primary care physicians represent an alternative source of care for many conditions, but more primary care physicians also could initiate more referrals. Therefore, the effect of primary care physician supply on demand is indeterminate. HMOs may tend to enroll healthier elders, leaving enrollees who are less healthy than average in the fee-for-service sector (Lichtenstein, Thomas, Adams-Watson, et al. 1991). Thus, higher HMO penetration is expected to increase demand among non-HMO enrollees.

The vector of enrollee characteristics, Y, included the percentage of enrollees in the fee-for-service sector who were 75 to 84 years old, the percentage who were more than 84 years old, the percentage female, the percentage nonwhite, and the percentage who died during the study year. These variables may be viewed as proxies for health status. Two additional control variables, Z, were included in the model: per capita income for the population and the percentage of the population more than 64 years old.

Definitions and descriptive statistics for all explanatory variables other than the fee variables are presented in Table 1. Because the fee variables are indexes, their means were all close to one. They are omitted in the interest of space.

ESTIMATION METHOD

For each study operation, the operation's own-fee, the operation's cross-fee, the Medicare assignment rate, and the supply of surgeons were considered to be endogenous. Ordinary least squares estimation results in biased coefficient estimates when there are endogenous explanatory variables. Therefore, the model for each operation was estimated using two-stage least squares, a method that corrects for endogeneity bias (Kmenta 1986).

The endogenous variables in each equation were first regressed on a set of instrumental variables consisting of all of the exogenous variables in the equation and additional variables suggested by prior studies of physician pricing, Medicare assignment, and physician location. TABULAR DATA OMITTED The additional instrumental variables included measures of physicians' practice costs (e.g., nonphysician employee wages), which are correlated with physician fees (Pope et al. 1989). They also included measures of Medicare carriers' administrative practices that affect physicians' costs of collecting from Medicare (e.g., claims denial rates), which are correlated with assignment rates (Mitchell and Cromwell 1982). Still further they included characteristics of the nonelderly population and measures of the professional and community amenities in an MSA (e.g., number of medical schools, climatic variables), which are correlated with surgeon supply (Ernst and Yett 1985). The predicted values of the endogenous variables were then used in estimating the coefficients of the demand equation.

All regressions were weighted using the weights suggested by Cox (1970).(9) Monetary variables in the regressions were deflated using a published cost-of-living index (American Chamber of Commerce Researchers Association 1986).(10) A p-value of .05 was chosen as the criterion for statistical significance.

DATA SOURCES

The principal database for the study was created by merging the Health Care Financing Administration's (HCFA's) 1986 Part B Medicare Annual Data (BMAD) Beneficiary file and the HCFA's 1986 Health Insurance Skeleton Eligibility Write-off (HISKEW) file. The BMAD Beneficiary file contains detailed information on all physician services received under fee-for-service payment by a 5 percent random sample of Medicare enrollees. (Services provided to enrollees in HMOs are not included.) The HISKEW file contains the age, sex, race, Medicaid eligibility status, HMO enrollment status, date of death, and county of residence of enrollees in the BMAD sample. The merged file was used to develop utilization rates for the study operations at the MSA level. The merged file also was the source of data on Medicare assignment rates, Medicaid coverage, and enrollee characteristics.

Several other data sources were used to obtain variables for the econometric analyses. Data to construct the fee variables were obtained from the 1986 BMAD Procedure file, which contains information on all physician services billed to Medicare aggregated to the level of charge locality and physician specialty. Data from the 1986 Current Population Survey were used to impute the percentage of elderly Medicare enrollees with private supplementary insurance coverage in each MSA. This percentage was added to the percentage of enrollees with Medicaid coverage to obtain the supplementary insurance variable. The HCFA's 1989 Adjusted Average Per Capita Cost master file was used to obtain information on the percentage of Medicare enrollees in HMOs. The Area Resource File (ARF) was used to obtain data on surgeon and primary care physician supply, per capita income, and community amenities. Data on physicians' practice costs were obtained from a published report (Welch, Zuckerman, and Pope 1989). Data on Medicare carriers' administrative practices were obtained from annual HCFA reports.

RESULTS

GEOGRAPHIC VARIATION IN UTILIZATION RATES

Means and coefficients of variation for study operations' rates of use are shown in Table 2. Using chi-square tests (Diehr et al. 1990), the null hypothesis of no variation in use rates across MSAs was rejected (p |is less than~ .05) for all of the study operations except partial colectomy.

GEOGRAPHIC DIFFERENCES IN OWN-PRICE TO CROSS-PRICE RATIOS

Geographic variation in the ratio of an operation's own-price to its cross-price can be regarded as a measure of geographic differences in the relative fee structure. Table 3 presents the lowest, mean, and highest own-price to cross-price ratios for each study operation, as well as the coefficients of variation for the ratio. The range in the own-price to cross-price ratio was nearly twofold or higher for most of the ophthalmology, orthopedic, and urology operations in the study, but was somewhat lower for the general surgery operations. Coefficients of variation confirmed that geographic differences in the own-price to cross-price ratio were smallest for the general surgery operations, although these coefficients also were low for the urology operations.

REGRESSION RESULTS

Results of weighted two-stage least squares estimation of the demand equation for each study operation are shown in Tables 4 and 5. Table 4 presents the results for the ophthalmology and general surgery operations; Table 5 presents the results for the orthopedic surgery and urology operations.

TABULAR DATA OMITTED

Fee Effects. Statistically significant (p |is less than~ .05) positive own-price effects were found for cataract extraction, knee replacement, and knee arthroscopy. In addition, significant negative cross-price effects were found for cataract extraction and knee replacement. These results are incompatible with the neoclassical model of utilization, but offer support for the demand-creation model. On the other hand, a significant positive cross-price effect was found for retinal photocoagulation, and this result is incompatible with both the neoclassical and demand-creation models. Fee effects were not significant for the other seven operations studied.

Effects of Other Variables. Consistent with theoretical expectations, statistically significant, positive effects of the assignment rate on demand were found for laser trabeculoplasty, retinal photocoagulation, and fulguration of bladder tumor. However, significant negative assignment-rate effects were found for fixation of hip fracture and transurethral prostatectomy. Similarly, a significant positive effect of TABULAR DATA OMITTED supplementary insurance coverage was found for transurethral prostatectomy, but a significant negative effect was found for partial colectomy.

A significant positive effect of surgeon supply on demand was found only for knee arthroscopy. The lack of significant surgeon-supply effects is not inconsistent with previous research. Although considerable evidence supports a positive association of total surgery rates with the supply of physicians (Fuchs 1978; Cromwell and Mitchell 1986), several studies have failed to find significant associations between physician supply and rates of use for individual operations (e.g., Cageorge, Roos, and Danzinger 1981; Roos 1984).

Theoretical predictions regarding the effect of primary care physician supply on demand were indeterminate. Therefore, the finding that significant negative effects occurred for five operations (cataract TABULAR DATA OMITTED extraction, laser trabeculoplasty, fixation of hip fracture, knee arthroscopy, and fulguration of bladder tumor) is of interest. These results suggest that the role of primary care physicians as an alternative source of care may be more important than their role as a source of referrals to surgeons. Consistent with theoretical expectations, a significant positive effect of HMO penetration on demand was found for cataract extraction and fixation of hip fracture. However, a significant negative effect of HMO penetration was found for partial colectomy.

TABULAR DATA OMITTED

Although there were few statistically significant coefficient estimates among the enrollee variables in the regression analyses, those that were significant generally were consistent in the direction of their effect. Thus, a higher percentage of enrollees 75 to 84 years old was associated with higher demand for two operations, a higher percentage of enrollees older than 84 years was associated with lower demand for two operations, and a higher percentage of female enrollees was associated with lower demand for three operations.

DISCUSSION

The findings of this study are inconclusive regarding the effect of variations in relative fees on the demand for surgical operations by elderly Medicare enrollees. Regression analyses offer support for a demand-creation model of surgical utilization in the case of cataract extraction, total knee replacement, and knee arthroscopy.(11) However, regression results are compatible with the neoclassical model of utilization in the case of laser trabeculoplasty, partial colectomy, cholecystectomy, inguinal hernia repair, fixation of hip fracture, laser trabeculoplasty, transurethral prostatectomy, and fulguration of bladder tumor. The result for retinal photocoagulation -- a statistically significant, positive cross-price effect -- cannot be explained using either the demand-creation or the neoclassical model. Of interest, all the operations that exhibited significant fee effects are ophthalmology or orthopedic operations. Several explanations for the inconsistent regression results should be considered.

A possible explanation is that fee effects truly are absent for all of the study operations, and that the statistically significant effects that were found are a consequence of conducting a large number of statistical tests. Under this scenario, the neoclassical model is actually the correct model of utilization for all the operations, and rejection of the neoclassical model in favor of the demand-creation model in the case of cataract extraction, knee replacement, and knee arthroscopy (as well as the positive cross-price effect for retinal photocoagulation) constitutes a type-I error.(12) There is no way to determine whether this is the correct explanation, but it is noteworthy that fee effects achieved significance at the 1 percent level (p |is less than~ .01) only in the case of cataract extraction.

Another possible explanation is that fee effects truly occur for most of the operations studied, but these effects are not statistically significant for many of the operations due to technical limitations of the regression analyses. In particular, own-price to cross-price ratios varied least for the general surgery and urology operations. Thus, for these operations, the geographic differences in the relative fee structure may have been too small for the efficient estimation of separate own-price and cross-price effects. Under this scenario, the demand-creation model is actually the correct model of utilization for most of the study operations, and failure to reject the neoclassical model in many cases constitutes a type-II error.(13)

A third possible explanation is that the appropriate model of surgical utilization differs among operations. Surgeons' ability or willingness to create demand in response to financial incentives may vary among operations, possibly related to the strictness of the clinical indications for the operation. For example, the clinical indications for laser trabeculoplasty, partial colectomy, inguinal hernia repair, fixation of hip fracture, and fulguration of bladder tumor, five of the operations that did not exhibit significant fee effects, are relatively strict. Small deviations from these indications could cause considerable harm to patients. Surgeons may be unable or unwilling to create demand for these operations irrespective of financial incentives, and the neoclassical model may be appropriate for describing their use.

By comparison, the clinical indications for cataract extraction, knee replacement, and knee arthroscopy, the three operations that exhibited results consistent with the demand-creation model, are much less strict. For these operations, there is a substantial gray area in which intervention may or may not be worthwhile. Surgeons may be more likely to create demand for these operations in response to financial incentives, and the neoclassical model may not be appropriate.

A fourth possible explanation for the inconsistent regression results is the cross-sectional design of the study. Incentives for demand creation in physician services markets are present only when there is excess supply. Therefore, the finding that neoclassical behavior is frequently observed in this study may simply be due to markets for surgical services that are in equilibrium -- or near equilibrium -- in most market areas. Alternatively, the extent of excess supply may differ among surgical specialties, which may help to explain the differences in results among the specialties studied. It is important to emphasize that lack of support for the demand-creation model of utilization in a cross-sectional analysis of geographic differences in relative fees does not preclude finding evidence of demand creation in a time-series study of utilization responses to exogenous changes in the relative fee structure.

Despite the lack of definitive results about the most appropriate economic model of surgical utilization, this study has several important implications for future research regarding the effect of changes in the relative fee structure on the use of surgical operations. First, the theoretical model used in the study highlights the importance of considering potential cross-price effects in addition to own-price effects. Previous studies of the utilization response to changes in relative fees have taken own-price effects into account, but not cross-price effects. Because the size and direction of the net impact on use may be determined by the complex interplay of both effects, ignoring cross-price effects could lead to wrong conclusions regarding the influence of changes in the own-price, which in turn would confuse interpretation of the underlying economic behavior. Overlooked cross-price effects could be partially responsible for the inconsistent own-price effects found by Hurley, Labelle, and Rice (1990).

Second, the model used in the study stresses the importance of the supply-demand balance in physician services markets. Knowing whether the market for surgeons' services is in equilibrium or in excess supply is crucial for interpreting the results of empirical analyses and inferring the appropriate economic model of utilization. Moreover, this is true not only in cross-sectional studies, such as the work presented in this article, but also in time-series studies, such as those that will be required to assess the impact of the Medicare fee schedule. The role of the supply-demand balance has been ignored in previous studies of changes in physician fees and demand creation (e.g., Rice 1984; Hurley, Labelle, and Rice 1990).

Third, the findings of the study suggest that the appropriate model of surgical utilization -- i.e., neoclassical versus demand-creation models -- may differ among operations. Investigating this question further is an important goal for future research, as is examining the characteristics of operations that may influence the choice of appropriate model. A barrier to drawing inferences about the relationship between characteristics of operations and models of utilization in this study is the small number of operations studied. Nonetheless, an intuitively appealing hypothesis that received limited support in the study is that demand creation is more likely when the clinical indications for an operation are less strict in the sense that the marginal benefits of the operation in different types of patients are similar or poorly defined. Studies to examine whether typologies of operations based on the strictness of their clinical indications predict the appropriate model of utilization could prove fruitful.

Fourth, the statistical concerns raised in the study suggest that examining the utilization responses of groups of operations rather than individual operations may be a useful modification in research design. Operations could be grouped according to the strictness of their clinical indications or other relevant criteria.

This study also has implications for efforts to monitor the effects of the Medicare fee schedule on surgical utilization among Medicare enrollees. In particular, the findings of the study suggest that, under the fee schedule, the rates of use of some operations may increase, use rates of others may decline, and use rates of still others may not change. Therefore, monitoring changes in access to surgical care and in the appropriateness of surgery may require assessing the utilization responses exhibited by individual operations or small groups of operations. Operations whose rates of use experience large declines under the fee schedule could raise concerns about impaired access, whereas operations whose use rates experience rapid growth could raise concerns about increasing levels of inappropriate surgery.

Motivated by the need to inform Medicare physician payment policy, the next several years are likely to see a growing interest in research to examine the effects of changes in the relative fee structure on the use of physician services, on access to these services, and on the quality and appropriateness of care. The theoretical and empirical framework presented in this article provides a foundation for this research.

NOTES

1. Several studies have examined the impact of across-the-board changes in physician fees -- that is, of changes that are similar, in percentage terms, for all physician services -- on aggregate measures of physician services utilization (e.g., Rice 1984; Wedig, Mitchell, and Cromwell 1989). The investigators in these studies generally have interpreted their findings as consistent with demand creation.

2. Prior to 1984, the annual increase in Medicare prevailing charges for physician services was limited by the Medicare Economic Index (Dutton and McMenamin 1981). The Medicare physician fee freeze was in effect from 1984 to 1986. Changes in Medicare's relative fee structure occurred as a result of provisions in the Omnibus Budget Reconciliation Acts of 1987, 1989, and 1990, which selectively reduced fees for surgical procedures identified as "overpriced" (Physician Payment Review Commission 1988, 1990, 1991). However, data to assess the effects of these provisions have only recently become available.

3. The assumption that surgeons maximize profits yields the simplest model, because changes in relative fees result only in substitution effects. An alternative approach is to assume that surgeons maximize utility. Utility maximization models are more complicated, because changes in relative fees lead to both substitution and income effects. However, differences in predictions between profit and utility maximization models are minor (see notes 4 and 7).

4. In utility maximization models, the substitution effect leads to a higher supply of the operation when the fee for the operation increases, but the income effect is in the opposite direction. The substitution effect generally dominates and the own-price elasticity of supply is positive over a range of the fee for the operation. However, unlike the result in profit maximization models, the income effect may dominate and the supply curve may become "backward-bending" under certain circumstances.

5. Although patients' welfare is maximized at the equilibrium fee and quantity, there is a societal welfare loss due to the moral hazard caused by full insurance.

6. Price rigidities introduced by charge-based methods of determining physician fees may prevent the adjustments in fees necessary to reestablish equilibrium in physician services markets after physician supply increases.

7. In utility maximization models, both the substitution and income effects lead to lower supply of an operation when the fees for other services increase. Thus, as in profit maximization models, the cross-price elasticity of supply is always negative.

8. In 1986, ophthalmology accounted for 13.3 percent of Medicare payments to physicians, second among all specialties; general surgery for 8.7 percent, fourth among all specialties; orthopedic surgery for 5.8 percent, sixth among all specialties; and urology for 4.1 percent, tenth among all specialties.

9. Weighting was necessary due to heteroscedasticity. The weight for each MSA was:

weight = {nr(1 - r)/|1 - (1/n)~}

where r is the proportion of enrollees who received the operation and n is the number of enrollees. Observations with r equal to zero receive zero weight.

10. Values of the cost-of-living index were available for 167 MSAs. Values were imputed for the remaining 149 MSAs using a separate regression equation developed for this purpose.

11. In theory, positive own-price effects and negative cross-price effects are also consistent with a neoclassical market that is in excess demand, since in such a market the quantity of the service actually provided is determined by the market supply curve. However, in practice this possibility can be dismissed. Most observers would agree that there is no shortage of surgeons in the United States, especially in urban areas.

12. A type-I error refers to rejection of the null hypothesis when the null hypothesis is true.

13. A type-II error refers to failure to reject the null hypothesis when the null hypothesis is false.

REFERENCES

American Chamber of Commerce Researchers Association. Inter-City Cost of Living Index. Indianapolis, IN: ACCRA, 1986.

American Medical Association. Physicians' Current Procedural Terminology, 4th Edition. Chicago: AMA, 1986.

Cageorge, S. M., L. L. Roos, and R. Danzinger. "Gallbladder Operations: A Population-based Analysis." Medical Care 19, no. 5 (May 1981): 510-25.

Chassin, M. R., J. Kosecoff, R. E. Park, C. M. Winslow, K. L. Kahn, N. J. Merrick, J. Kessey, A. Fink, D. H. Solomon, and R. H. Brook. "Does Inappropriate Use Explain Geographic Variations in the Use of Health Care Services? A Study of Three Procedures." Journal of the American Medical Association 258, no. 18 (13 November 1987): 2533-37.

Cox, D. R. Analysis of Binary Data. London: Methuen, 1970, pp. 78-79.

Cromwell, J., and J. B. Mitchell. "Physician-induced Demand for Surgery." Journal of Health Economics 5, no. 4 (December 1986): 293-313.

Diehr, P., K. Cain, F. Connell, and E. Volinn. "What is Too Much Variation? The Null Hypothesis in Small-Area Analysis." Health Services Research 24, no. 6 (February 1990): 741-71.

Dutton, B. L., and P. McMenamin. "The Medicare Economic Index: Its Background and Beginnings." Health Care Financing Review 3, no. 1 (September 1981): 137-40.

Ernst, R. L., and D. E. Yett. "Econometric and Statistical Studies of the Geographic Distribution of Physicians." In Physician Location and Specialty Choice. Ann Arbor, MI: Health Administration Press, 1985, pp. 179-226.

Escarce, J. J. "Geographic Variation in Relative Fees under Medicare." American Journal of Public Health 81, no. 11 (November 1991): 1491-93.

Fuchs, V. R. "The Supply of Surgeons and the Demand for Operations." Journal of Human Resources 13 (1978, Supplement): 35-56.

Graduate Medical Education National Advisory Committee (GMENAC). GMENAC Summary Report, Vol. 1. Washington, DC: Government Printing Office, 1980.

Greenspan, A. M., H. R. Kay, B. C. Berger, R. M. Greenberg, A. J. Greenspon, and M. J. S. Gaughan. "Incidence of Unwarranted Implantation of Permanent Cardiac Pacemakers in a Large Medical Population." New England Journal of Medicine 318, no. 3 (21 January 1988): 158-63.

Hurley, J., R. Labelle, and T. Rice. "The Relationship Between Physician Fees and the Utilization of Medical Services in Ontario." Advances in Health Economics and Health Services Research 11 (1990): 49-78.

Keeler, E. B., and J. E. Rolph. "The Demand for Episodes of Treatment in the Health Insurance Experiment." Journal of Health Economics 7, no. 4 (December 1988): 337-67.

Kmenta, J. Elements of Econometrics. New York: MacMillan, 1986, pp. 704-11.

Lichtenstein, R., W. Thomas, J. Adams-Watson, J. Lepkowski, and B. Simone. "Selection Bias in TEFRA At-Risk HMOs." Medical Care 29, no. 4 (April 1991): 318-31.

Mitchell, J. B., and J. Cromwell. "Physician Behavior under the Medicare Assignment Option." Journal of Health Economics 1, no. 3 (September 1982): 245-64.

Nelson, L., A. Ciemnecki, N. Carlton, and K. Langwell. "Assignment and the Participating Physician Program: An Analysis of Beneficiary Awareness, Understanding, and Experience." PPRC Background Paper No. 89-1. Washington, DC: Physician Payment Review Commission, 1989.

Pauly, M. V. "Fee Schedules and Utilization." In Regulating Doctors' Fees: Competition, Benefits, and Controls under Medicare. Edited by H. E. Frech. Washington, DC: American Enterprise Institute Press, 1991, pp. 288-305.

Physician Payment Review Commission. Annual Report to Congress. Washington, DC: PPRC, 1988.

-----. Annual Report to Congress. Washington, DC: PPRC, 1990.

-----. Annual Report to Congress. Washington, DC: PPRC, 1991.

Pope, G. C., W. P. Welch, S. Zuckerman, and M. G. Henderson. "Cost of Practice and Geographic Variation in Medicare Fees." Health Affairs 8, no. 3 (Fall 1989): 117-28.

Rice, T. "Physician-induced Demand for Medical Care: New Evidence from the Medicare Program." Advances in Health Economics and Health Services Research 5 (1984): 129-60.

Roos, N. P. "Hysterectomy: Variations in Rates across Small Areas and across Physicians' Practices." American Journal of Public Health 74, no. 4 (April 1984): 327-35.

Rutkow, I. M. "Surgical Operations and Manpower Considerations." In Surgical Care in the United States: A Policy Perspective. Edited by M. L. Finkel. Baltimore, MD: Johns Hopkins University Press, 1988, pp. 22-40.

Schwartz, W. B., and D. N. Mendelson. "No Evidence of An Emerging Physician Surplus: An Analysis of Change in Physicians' Work Load and Income." Journal of the American Medical Association 263, no. 4 (26 January 1990): 557-60.

Sloan, F. A., and R. Feldman. "Competition among Physicians." In Competition in the Health Care Sector: Past, Present, and Future. Edited by W. Greenberg. Germantown, MD: Aspen Systems, 1978, pp. 45-102.

Wedig, G. J. "Health Status and the Demand for Health: Results on Price Elasticities." Journal of Health Economics 7, no. 2 (June 1988): 151-63.

Wedig, G., J. B. Mitchell, and J. Cromwell. "Can Price Controls Induce Optimal Physician Behavior?" Journal of Health Politics, Policy and Law 14, no. 3 (Fall 1989): 601-20.

Welch, W. P., S. Zuckerman, and G. C. Pope. "The Geographic Medicare Economic Index: Alternative Approaches." Final Report to the Health Care Financing Administration. Baltimore, MD: Health Care Financing Administration, 1989.

Winslow, C. M., D. H. Solomon, M. R. Chassin, J. Kosecoff, N. J. Merrick, and R. H. Brook. "The Appropriateness of Carotid Endarterectomy." New England Journal of Medicine 318, no. 12 (24 March 1988): 721-27.
COPYRIGHT 1993 Health Research and Educational Trust
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Escarce, Jose J.
Publication:Health Services Research
Date:Oct 1, 1993
Words:6818
Previous Article:Interorganizational exchanges as performance markers in a community cancer network.
Next Article:Multiple-site physician practices and their effect on service distribution.
Topics:

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters