Evaluation of physician practice using risk-adjusted commercial databases.

In light of fiscal constraints currently being imposed upon all components of the health care system within the US, it is efficacious to assess the efficiency of physician practice patterns within any component of the health care system. It has been noted elsewhere that physicians control more than 80 percent of the decisions affecting health costs (Chilingerian & Sherman, 1990). Consequently, management of physician practice patterns presents an important opportunity for reducing health care costs.

Assessment of physician practice patterns can be undertaken to: (1) determine whether there exists any room for improvement in the efficiency of these practice patterns, and if so; (2) estimate how much savings in hospital costs could reasonably be expected were physician practice patterns to change in a specific manner.

While the literature is replete with examples of strategies employed to reduce hospital resource utilization (Billi et al. 1987; Gortmaker et al. 1998; Hart & Balas 1994; Johnson & Martin 1996; Johnson et al. 1993; Kramolowsky et al. 1995; Kay & Wadsworth 1995; Pugh et al. 1989), less has been written concerning methodologies that can be employed to target inefficient practice behaviors of the individual physician that contribute significantly to unnecessary use of hospital resources. Much of the literature concerns itself with the impact of a strategy implemented to reduce the time required to complete a specific process within the continuum of patient care, or to reduce the amount of material consumed during a such a process (Davis et al. 1995; Eisenberg 1986; Oxman et al. 1995). Very often, the typical article dealing with hospital resource utilization begins with the assertion that examination has revealed an over-utilization of a specific diagnostic test, or that a review of wait times has demonstrated a process inefficiency.

The purpose of this paper is to explain a methodology, which can be utilized to identify inefficient physician practice behaviors that significantly contributed to unnecessary hospital resource utilization. The authors employed a large, commercially available, risk and severity adjusted national database owned and maintained by HCIA, a health care information company specializing in the development and marketing of clinical and financial decision support systems. Such databases furnish estimates of expected resource use, such as hospital costs, hospital charges, lengths of hospital stay, complication rates and mortality rates for any patient, taking into account that patient's severity of illness and risk of mortality. Each patient is assigned to a severity of illness class and a risk of mortality class on the basis of principal diagnosis, secondary diagnoses, procedures performed, age, sex and discharge disposition of the patient. Both risk of mortality and severity of illness are categorical variables. Each category is tied to an expected level of resource consumption based upon national, regional or local data. The authors used deviations from these expected values (either positive or negative) to assess physician practice pattern efficiency within a regional medical center setting. Specifically, information regarding length of hospital stays and hospital charge information was analyzed within several diagnostic related groups (DRGs), selected because of their large clinical volumes within the hospital.

ANALYTIC OVERVIEW

There are two phases utilizing these types of databases, differing mainly on the level of analysis. A macro phase compares the charges of the target hospital to the charges of peer and best practice or "benchmark" hospitals for the same clinical procedure. This approach is useful for identifying specific areas of clinical practice that may be problematic for the target hospital in terms of over- or inappropriate use of resources. The second phase focuses on each individual physician in comparison to "benchmark" hospital physicians and colleagues within the target hospital setting. This micro phase is most useful for identifying which physicians may be candidates for some sort of strategy designed to increase practice efficiency, and which physicians have already achieved an efficient level of practice.

The macro phase compares the "manageable" portion of the target hospital's charges with peer and benchmark hospitals to identify the savings opportunity available to the target hospital. The manageable portion is defined as the portion of a hospital's charges that is directly attributable to the hospital's pricing structure (price) and physician's practice patterns, after being adjusted for illness severity and cost of living variations. Peer hospitals are a group of hospitals in HCIA's database selected by HCIA to represent the national average. Benchmark hospitals consist of several hospitals chosen by HCIA on the basis of their efficiency and quality outcomes for a specific DRG. Expected values for hospital length of stay and hospital costs and charges are generated by HCIA's SoleSource[R] Hospital/Physician Profiler software, and are derived for both peer and benchmark hospitals. The analysis proceeds to identify the target hospital's savings opportunity for a DRG, based on comparisons with both peer and benchmark hospitals (HCIA 1997).

Savings opportunity is partitioned into four categories: (1) price, (2) frequency, (3) distribution, and (4) other resources. Price is the dollar amount the hospital charges for a unit of service (how much the hospital charges for an aspirin). Frequency is the number of units of service item per case (how many times a patient receives an aspirin), and distribution is the percentage of total patients receiving a service (the percentage of patients who receive an aspirin). Other resources are those not in laboratory, imaging or pharmacy, and include medical and surgical supplies, specialty clinical services (respiratory therapy, physical therapy, catheter laboratory and others), as well as various non-clinical services. It is important to note that when analyzing charge data, we are not analyzing the variance between actual and expected charges due to hospital pricing structure, but are separating this portion of the variance from the analysis.

Because each element of the database is generated from the patient's bill, savings opportunity can be broken down by various revenue centers within the target hospital, allowing the analyst to pinpoint areas of opportunity with extreme precision. For example, within the pharmacy revenue center, it is possible to identify the amount of savings opportunity associated with the administration of a specific drug. Further, it is possible to determine how much of that savings opportunity is due to: (1) the price the target hospital charges for the drug, (2) how many patients receive the drug, and (3) how many times each patient receives the drug, when compared to both peer and benchmark hospitals. Thus it is easy to identify, and concentrate strategies for creating change, in areas that will yield the greatest impact on utilization of the target hospital's resource.

After identifying areas and dollar amounts of savings opportunities, the micro phase of the analysis begins. Individual physicians are graphed on an efficiency grid, with the x-axis co-ordinate representing deviations from expected length of hospital stay in days, and the y-axis coordinate representing deviations from expected average hospital charge per case. An example of this type of grid appears in Figure 1. This approach affords a relatively quick and accurate depiction of one physician's performance relative to his/her peers. Physicians with patients having lower than expected average hospital lengths of stay and hospital charges (i.e., high efficiency) appear in the lower left-hand quadrant of the efficiency grid. Physicians whose patients exhibit higher than expected average lengths of hospital stay and hospital charges appear in the upper right hand quadrant of the efficiency grid (i.e., low efficiency).

[FIGURE 1 OMITTED]

For each of the selected DRGs, the authors were able to identify physicians whose practice patterns were efficient, as well as those physicians who evidenced higher than expected lengths of stay, hospital charges, or both. This information can be used to determine how efficiently a disease can be managed at an institution, and identify "best practice" physicians as opposed to less efficient practitioners. Once these physicians have been classified, standards for best practice can be established and the behavior of "outlier" physicians can be modified through individualized strategies designed to increase practice efficiency.

The focuses for a physician-specific program to increase practice efficiency can be determined from the database as well. Again, because the database is derived from the patient's hospital bill, it is possible to break down the patient's total hospital charges into specific charges, such as pharmacy charges, laboratory charges, radiology charges, etc. These specific charges for each physician can then be compared to national best practice or "benchmark" hospital charges, and overuse or inappropriate use of pharmaceuticals, radiographic or laboratory tests can be identified for each physician, enabling him/her to focus attention on various aspects of his/her practice pattern.

A CASE EXAMPLE

To illustrate how a risk-adjusted commercially available database can be used to identify physician practice inefficiency within the hospital setting, the authors present the following case example based upon actual empirical results. HCIA's SoleSource[R] Hospital/Physician Profiler software package was used to analyze data from a 300-bed regional medical center located in Maryland. This example demonstrates how such databases can supply the information necessary to design a program to promote change in physician practice patterns. The case example focuses on DRG 209: Major Joint Replacement and Reattachment Procedures, and more specifically, on a subset of that DRG, total hip replacements due to osteoarthritis.

Table 1 offers a summary report of the database for the target hospital, the group of peer hospitals, and the group of benchmark hospitals. These data used in the case example were for the period July 1, 1996 through June 30, 1997. The average total charge for all hospitals is broken down into the average routine charge (room, food and services associated with each) and the average ancillary charge (pharmacy, imaging, laboratory, prosthesis and other charges). While the target hospital compares favorably to the group of peer hospital (the national average), the benchmark hospitals have a shorter average length of hospital stay, and a smaller average total charge compared to the target hospital.

Table 2 presents the savings opportunity for the total hip replacement procedure, broken down by the components of the manageable portion. The total manageable portion is that portion of the charge directly attributable to hospital pricing and physician practice patterns. Basically, the table demonstrates that if the target hospital pricing and physician practice efficiency were strictly comparable to those of the benchmark hospitals', the target hospital could save approximately $248,000. Approximately $34,300 of this total would be due to hospital price reductions for its imaging, laboratory, and pharmacy services, and $55,000 would be due to a reduction in the frequency of the use of these services by physicians. The remainder, about $169,000, would be gained from a combination of reduction in price and frequency of use in other resources, such as operating room charges, anesthesia charges, medical, surgical and nursing supplies, etc. Table 2 also highlights the fact that it is possible to have a negative savings opportunity, which occurs when the target hospital or its affiliated physicians are outperforming the peer and/or benchmark hospitals or their affiliated physicians.

Table 3 summarizes the savings opportunity by revenue center within the target hospital. It is immediately evident that a large percentage of the total savings opportunity for the target hospital when compared to the benchmark hospital can be realized by shortening the average length of hospital stay, as well as by reducing the hospital charges associated with room and board. Other areas where the target hospital could realize savings is a reduction in the utilization of laboratory services, a more circumspect use of hospital imaging facilities, and less frequent use of hospital pharmacy services. A reduction in both pricing and utilization of services in the "other resources" category is another large component of the potential savings that could be realized by the target hospital. The current inability to break down these other services into their constituent parts is a limitation of the Hospital/Physician Profiler software that will be addressed by HCIA in the near future.

The summary report provided by Table 4 is especially interesting, since it uses a simple index to show how much target hospital parameters differ from the peer and benchmark hospitals. For example, the average length of stay in the target hospital is 13% longer in the target hospital than in the peer hospital group, and 27% longer than in the benchmark group. On the other hand, the average pharmacy charge of the target hospital is 77% less than the peer group of hospitals, and 72% less than the benchmark group of hospitals. This summary serves to quickly focus attention on areas that require attention to conserve target hospital resources. In this example, length of hospital stay, routine charges, ancillary charges, prosthetics charges and laboratory charges are all areas that require some sort of strategic plan to reduce resource consumption.

Why are the target hospital's laboratory charges so high compared to the benchmark hospitals? Table 5, Laboratory Summary Report, identifies several laboratory services that are being utilized more by physicians affiliated with the target hospital than by physicians affiliated with the benchmark hospitals. The total average laboratory charge for the target hospital is 91% greater than the total average laboratory charge for benchmark hospitals. The average blood bank charge for the target hospital is 345% higher than in the benchmark hospitals and 71% more patients receive blood bank services in the target hospital than in the benchmark hospitals. It also appears that the average charge for coagulation testing is 64% greater in the target hospital than in the benchmark hospitals, and that 64% more patients in the target hospital receive coagulation testing than in the benchmark hospitals. These trends are highlighted in Figures 2 and 3. Figure 2 graphically highlights the differences between physicians associated with the target hospital, and physicians associated with the benchmark hospitals. In the target hospital, physicians order packed red blood cells for 76% of patients, compared with 24% of patients at the benchmark hospital. Blood processing is ordered for 80% of patients at the target hospital, but for only 23% of patients at the benchmark hospitals. Finally, 98% of patients at the target hospital receive crossmatch services, ABO RH typing and AB screening, as compared with 51%, 15% and 36% of patients at the benchmark hospitals, respectively. Figure 3 highlights a difference in the average number of units of blood that are processed for patients in the target hospital (3.4 units) compared with patients in the benchmark hospitals (1.4 units). Is this a common practice among all physicians in the target hospital, or are certain physicians responsible for the apparent over-utilization of these laboratory services? At this point, phase two of the analysis begins.

The efficiency grid in Figure 4 was derived based upon data for the eight orthopedic surgeons at the target hospital who accounted for more than 87% of the target hospital's charges for this procedure. It shows that three of the eight orthopedic surgeons (B, A, and G) have patients with a higher than expected average length of hospital stay and higher than expected average hospital charges, when controlling for the risk and severity of their patients. Orthopedic surgeons A, F and D have patient populations with higher than expected average hospital charges, but a lower than expected average length of hospital stay. A pattern such as this could indicate these surgeons may be over-utilizing pharmacy, radiology, laboratory or other hospital services, since their higher than expected average hospital charges are not associated with a higher than expected average length of hospital stay. Orthopedic surgeons C, E and H have patients with lower than expected average lengths of hospital stay, and lower than expected average hospital costs, and are high efficiency practitioners.

[FIGURE 4 OMITTED]

The efficiency grid highlights several things. First, the practice patterns of a majority of orthopedic surgeons at our institution need to be modified in some way to increase practice efficiency. Second, orthopedic surgeons located in the upper fight-hand quadrant of the grid should consider discharging patients from the hospital sooner. Third, surgeons located in the upper left-hand quadrant of the grid need to examine their utilization of hospital services with an eye to discovering unnecessary and/or inappropriate use of laboratory or radiological tests, pharmaceutical use and prosthesis costs. And fourth, surgeons in the upper fight-hand quadrant of the grid may also need to examine the possibility that they over-utilize hospital services.

A singular advantage of this type of analysis that may not be immediately apparent is that any hospital pricing issues are controlled in the surgeon-to-surgeon comparison, since the data for all surgeons pertain to only one institution. Therefore, it is impossible to argue that differences in hospital pricing account for differences in practice efficiency between surgeons, assuming of course that each surgeon's cases are more or less randomly dispersed over the time period being examined. Clearly, this would not obtain if surgeons in hospital A were being compared to surgeons in hospital B; in such an instance, differences in hospital pricing would have to be factored into the equation.

[FIGURE 4 OMITTED]

It is important to note that an efficiency grid of this type ignores each surgeon's patient volume, and thus does not portray an accurate picture of how each surgeon's practice affects the hospital's total costs for this procedure. Clearly, if surgeon A has a caseload of 25 patients in the above example, while surgeon B has a patient caseload of five, surgeon A's practice accounts for a much larger share of the hospital's total costs for this procedure. Therefore, when allocating hospital resources dedicated to increasing practice efficiency, surgeon practice volume as well as the degree of surgeon practice inefficiency (i.e., how much a surgeon's average hospital charge deviates from the expected average hospital charge) must be taken into account.

Because databases of this sort are based on the hospital charges generated by the surgeon's treatment of his/her patients, it is possible to disaggregate total hospital charges for each patient into discrete charges linked to specific services. Thus, it is possible to compare surgical practices for a particular surgical procedure at an item-specific level; the number and types of imaging studies routinely administered to a patient by one surgeon can be compared to those routinely administered by another surgeon, for example. The data presented in Table 6 has been partially disaggregated into discrete hospital "revenue" centers.

The table displays data for each surgeon, and, in the rightmost column labeled "Bench," displays data for surgeons in "benchmark" hospitals. These benchmark hospitals are selected from the national database due to the efficient practice patterns and quality outcomes of their surgeons for the particular DRG being examined. This provides an assessment of how well surgeons within the institution under study are doing compared to surgeons who practice in several of hospitals that have exemplary efficiency and quality outcomes. In this instance, not only has risk and severity been adjusted for each discrete patient population being compared, but charge figures have been adjusted for the benchmark hospitals using the Health Care Financing Administration's wage index, in order to account for geographic cost-of-living differences.

In addition to the average length of stay for each physician, the hospital charges are further disaggregated by hospital revenue centers. Routine charges are charges associated with room and board. Ancillary charges consist of recovery room charges, and ICU/CCU charges. The "other charges" category is composed of such things as anesthesia and operating room charges, orthopedic supplies, medical, surgical and nursing supplies, charges associated with respiratory and pulmonary services, physical therapy and rehabilitative services, and miscellaneous surgical services.

From Table 6, it is apparent that there is a wide variation in practice patterns, indicated by the variation in hospital lengths of stay, prosthetics charges, as well as laboratory, imaging and pharmacy charges. Surgeon H has the lowest average total charge per patient, due primarily to having the lowest average prosthetics charge, the lowest average ancillary charges, and the lowest average pharmacy charges. Surgeon D has the lowest routine charge per patient, since his/her patients also have the lowest average length of hospital stay. Surgeon E has the lowest average imaging charge, and also the lowest average laboratory charge; he/she is the surgeon utilizing the lowest number of laboratory tests per patient. At this point, four surgeons have been identified as being efficient in their use of various hospital resources, each in different areas. This information is useful for interviewing each surgeon regarding how his/her practice patterns differ from other surgeons in each of the areas where he/she is the most efficient. These surgeons may also be used as a "core" group of experts when designing or modifying the clinical pathway for this particular procedure. If each surgeon's individual areas of efficiency could be combined, the result would be a practice pattern that would be more efficient than any one of the individual surgeons.

To illustrate the strength of these types of databases in examining the differences in practice patterns among a group of surgeons, and the level of detail that can be obtained, Table 7 presents a further disaggregation of laboratory charges. Consistent with Table 6, Table 7 identifies Surgeon E as the most efficient user of hospital laboratory services, with an average charge of $413 per patient. This surgeon utilized only 15 "units" of laboratory services per patient. From Table 7, Surgeon E's average blood bank charge per patient is $242, or less than half of any other surgeon. Compared to his/her colleagues, Surgeon E utilizes blood bank services for 80% of his/her patients, while other surgeons utilize these same services for 100% of their patients. Although almost trivial, Surgeon E's average chemistry charge per patient is $6, again less than half that of almost all other surgeons, since this service is used for only 20% of his/her patients.

Disaggregating the data still further, Table 8 examines the average number of units of blood bank services utilized by each surgeon (frequency), and the proportion of his/her cases receiving such services (distribution). Table 8 demonstrates that for each type of blood bank service, Surgeon E utilizes the fewest average number of services per patient, and utilizes such services on a lower proportion of his/her patients. For example, Surgeon E uses 1.5 tests per patient for crossmatching, and crossmatches 80% of his patients, while his/her colleagues use from 2.2 to 3.1 tests per patient for crossmatching, and crossmatch virtually 100% of their patients. So to answer the question that began the micro phase of the analysis, the practice behaviors documented in Figures 2 and 3 are, almost without exception, universal among the target hospital's surgeons, with the exception of Surgeon E.

CONCLUSION

The efficacy of using a risk-adjusted commercially available database to evaluate physician practice patterns is evident from the case example presented above. It is important to note that the case example above focused on a single aspect of physician practice behavior that appeared to be at odds with the practice behavior of physicians affiliated with the benchmark hospitals for this surgical procedure, namely, the utilization of blood bank services. There are many other practice behaviors of physicians affiliated with the target hospital that differed significantly from those affiliated with the benchmark hospitals that might have been chosen; the blood bank services example was selected purely for heuristic purposes. It is readily apparent that there is a significant amount of additional knowledge to be gained about the practice patterns of physicians affiliated with the target hospital by pursuing further analysis of the database.

The utility of analyzing this type of data is manifold. First, the results can be used as a basis for the development of a strategic plan to change physician practice patterns. Second, the analysis readily identifies the types of physician practice behaviors that require modification to properly utilize hospital resources. Third, the analysis also identifies those physicians whose practice patterns require modification, allowing change agents and resources to be focused on the specific practice behaviors of specific physicians that require modification. Fourth, the analysis detects physicians who already exhibit efficient practice patterns, thereby identifying potential candidates to act as informal change agents or local opinion leaders. And lastly, the analysis provides a focus for the development of a standardized protocol or clinical pathway, providing a nucleus for changing physician practice patterns no matter what vehicle is chosen to promote revision of them.

FIGURE 2

BLOOD BANK SERVICES--DISTRIBUTION

              Percentage of Patients Receiving Service

                   Hospital    Peer    Bench

Red Cells            76%       42%      24%
Processing           80%       18%      23%
Crossmatch           98%       51%      44%
ABO_Rh_Type          98%       17%      15%
Ab_Scm               98%       36%       8%
Type_Screen           0%       17%      11%
Rh Phentype           0%       33%      11%
ABO Type              0%       27%      11%

Note: Table made from bar graph.
FIGURE 3

BLOOD BANK SERVICES--FREQUENCY

                    Average Units Patient

                  Hospital    Peer    Bench

Red Cells           2.4       2.0      2.2
Processing          3.4       2.0      1.4
Crossmatch          2.7       2.2      2.4
ABO-Rh_Type         1.1       1.1      1.0
Ab-Scm              1.1       1.1      1.0
Type_Screen         0.0       0.9      1.3
Rh_Phentype         0.0       1.2      1.2
ABO_Type            0.0       1.1      1.2

Note: Table made from bar graph.
TABLE 1

SUMMARY REPORT

                           Hospital     Peer        Bench

Cases                          90         1598          187
Avg. Length of Stay             5.1          4.5          4.0
Avg. Total Charge         $15,709      $19,008      $12,954
Avg. Routine Charge        $2,921       $2,226       $1,583
Avg. Ancillary Charge     $12,788      $16,782      $11,371
TABLE 2

SAVINGS OPPORTUNITY BY COMPONENT OF MANAGEABLE PORTION

                                          PEER
                                        VARIANCE

     Component of                   Avg.
  Manageable Portion          Opportunity/Case        Total

Price                               ($974)           ($87,660)
Frequency                            $316             $28,440
Distribution/Treatment Mix           $149             $13,410
Other Resources                   ($2,790)          ($251,100)
Total Manageable Portion          ($3,300)          ($296,931)

                                         BENCHMARK
                                         VARIANCE

     Component of                   Avg.
  Manageable Portion           Opportunity/Case       Total

Price                                $381             $34,290
Frequency                            $613             $55,170
Distribution/Treatment Mix          ($115)           ($10,350)
Other Resources                    $1,876            $168,840
Total Manageable Portion           $2,755            $247,950
TABLE 3

SAVINGS OPPORTUNITY BY REVENUE CENTER

                                  PEER VARIANCE

  Variance Component          Avg. Case        Total

Length of Stay
  Total                           $695        $62,556
  Price                           $353        $31,757
  Frequency                       $342        $30,799
Laboratory
  Total                           ($61)       ($5,512)
  Price                          ($301)      ($27,096)
  Frequency                        $22         $1,962
  Distribution                    $218        $19,622
Imaging
  Price                          ($106)       ($9,495)
  Frequency                       ($20)       ($1,771)
  Distribution                     $35         $3,171
  Total                           ($91)       ($8,095)
Pharmacy
  Total                        ($1,053)      ($94,741)
  Price                          ($921)      ($82,847)
  Frequency                       ($28)       ($2,534)
  Distribution                   ($104)       ($9,360)
Other Resources
  Total                        ($2,790)     ($251,139)
Total Manageable Portion       ($3,300)     ($296,931)

                                    BENCH VARIANCE

  Variance Component           Avg. Case      Total

Length of Stay
  Total                         $1,338       $120,441
  Price                           $780        $70,191
  Frequency                       $558        $50,250
Laboratory
  Total                           $360        $32,421
  Price                           ($53)       ($4,743)
  Frequency                        $49         $4,397
  Distribution                    $364        $32,767
Imaging
  Price                           ($45)       ($4,035)
  Frequency                       ($37)       ($3,309)
  Distribution                     $45         $4,042
  Total                           ($37)       ($3,302)
Pharmacy
  Total                          ($782)      ($70,457)
  Price                          ($301)      ($27,135)
  Frequency                        $43         $3,828
  Distribution                   ($524)      ($47,150)
Other Resources
  Total                         $1,876       $168,856
Total Manageable Portion        $2,755       $247,959
TABLE 4

TOTAL HIP REPLACEMENT SUMMARY REPORT

Parameter                   Hospital       Peer        Bench

Number of Cases                   90         1598         187
Avg. Length of Stay                5.1          4.5         4.0
Avg. Total Charge            $15,709      $19,008     $12,954
Avg. Routine Charge           $2,921       $2,226      $1,583
Avg. Ancillary Charge        $12,788      $16,782     $11,371
Avg. Prosthetics Charge       $6,854       $6,509      $4,583
Avg. Pharmacy Charge            $309       $1,355      $1,092
Avg. Pharmacy Units               94           92         106
Avg. Laboratory Charge          $754         $816        $394
Avg. Laboratory Units             22           21          18
Avg. Imaging Charge             $170         $260        $207
Avg. Imaging Units                 2            2           3
Avg. Other Charge             $4,699       $7,841      $5,095

Parameter                     Peer        Bench
                              Index       Index

Number of Cases                 --          --
Avg. Length of Stay            1.13        1.27
Avg. Total Charge              0.83        1.21
Avg. Routine Charge            1.31        1.85
Avg. Ancillary Charge          0.76        1.12
Avg. Prosthetics Charge        1.05        1.50
Avg. Pharmacy Charge           0.23        0.28
Avg. Pharmacy Units            1.02        0.89
Avg. Laboratory Charge         0.92        1.91
Avg. Laboratory Units          1.05        1.22
Avg. Imaging Charge            0.65        0.82
Avg. Imaging Units             1.00        0.67
Avg. Other Charge              0.60        0.92
TABLE 5

LABORATORY SUMMARY REPORT

                              AVG. CHARGE/CASE

                                     Peer     Bench
Lab Sub-Department       Hospital    Index    Index

Lab Total                   $754      0.92    1.91
Blood Bank                  $563      2.01    4.45
Chemistry                    $26      0.23    0.61
Coagulation                  $52      0.41    1.65
Hematology/Urinalysis        $47      0.32    0.57
Immunology                    $0      0.28    0.19
Microbiology                 $35      1.05    1.04
Pathology/Cytology           $31      0.28    0.41
Laboratory Misc.              $0        --      --

                          DISTRIBUTION % OF CASES

                                     Peer     Bench
Lab Sub-Department       Hospital    Index    Index

Lab Total                  100.0%     1.00     1.00
Blood Bank                  97.8%     1.22     1.71
Chemistry                   43.3%     0.82     1.30
Coagulation                 71.1%     0.58     1.64
Hematology/Urinalysis      100.0%     1.01     1.00
Immunology                   1.1%     1.00     0.69
Microbiology                60.0%     2.24     2.01
Pathology/Cytology          64.4%     0.76     0.66
Laboratory Misc.             0.0%       --       --
TABLE 6

PHYSICIAN SUMMARY BY HOSPITAL REVENUE CENTER

                                        SURGEON

                               A            B           C

Cases                             9           19           15
Avg. Length of Stay               5.7          7.9          3.9
Avg. Routine Charge          $3,296       $4,763       $2,132
Avg. Ancillary Charge       $14,654      $14,390      $11,873
Avg. Prosthetics Charge      $8,643       $7,821       $5,920
Avg. Pharmacy Charge           $213         $254         $287
Avg. Pharmacy Units              91          130           71
Avg. Laboratory Charge         $677         $857         $798
Avg. Laboratory Units            18           30           22
Avg. Imaging Charge             $88         $238         $161
Avg. Imaging Units                2            3            2
Avg. Other Charge            $5,033       $5,219       $4,707
Avg. Total Charge           $17,950      $19,153      $14,005

                                        SURGEON

                               D           E           F

Cases                             3            5           14
Avg. Length of Stay               3.0          3.8          4.7
Avg. Routine Charge          $1,627       $2,056       $2,598
Avg. Ancillary Charge       $13,933      $11,712      $13,309
Avg. Prosthetics Charge      $7,813       $6,531       $7,348
Avg. Pharmacy Charge           $498         $218         $517
Avg. Pharmacy Units              69           85           85
Avg. Laboratory Charge         $850         $413         $870
Avg. Laboratory Units            22           15           20
Avg. Imaging Charge            $123          $83         $295
Avg. Imaging Units                3            2            2
Avg. Other Charge            $4,649       $4,466       $4,279
Avg. Total Charge           $15,560      $13,768      $15,907

                                        SURGEON

                               G            H          Bench

Cases                             5            8          187
Avg. Length of Stay               5.2          4.3          4.0
Avg. Routine Charge          $2,984       $2,343       $1,583
Avg. Ancillary Charge       $14,187      $10,875      $11,371
Avg. Prosthetics Charge      $8,228       $5,216       $4,583
Avg. Pharmacy Charge           $291         $199       $1,092
Avg. Pharmacy Units             117           73          106
Avg. Laboratory Charge         $769         $642         $394
Avg. Laboratory Units            26           20           18
Avg. Imaging Charge            $118         $117         $207
Avg. Imaging Units                2            3            3
Avg. Other Charge            $4,782       $4,701       $5,095
Avg. Total Charge           $17,171      $13,218      $12,954
TABLE 7

SURGEON SUMMARY BY LABORATORY DEPARTMENT

                                         AVERAGE CHARGE/CASE
                                              SURGEON

                            A         B         C         D         E

Blood Bank                 $542      $554      $620      $645      $242
Chemistry                   $24       $53       $10        --        $6
Coagulation                 $18       $79       $61       $20       $46
Hematology/Urinalysis       $49       $77       $35       $55       $30
Microbiology                $44       $45       $21      $100       $35
Pathology/Cytology           --       $48       $51       $29       $55
Total                      $677      $857      $798      $850      $413

                                   AVERAGE CHARGE/CASE
                                         SURGEON

                             F        G         H       Bench

Blood Bank                 $748      $520      $474      $127
Chemistry                   $35       $19       $19       $42
Coagulation                  $6      $134       $51       $32
Hematology/Urinalysis       $43       $38       $36       $83
Microbiology                $28       $25       $51       $34
Pathology/Cytology           $9       $34       $10       $75
Total                      $870      $769      $642      $394

                                      DISTRIBUTION (% OF CASES)
                                              SURGEON

                            A         B         C         D         E

Blood Bank               100.0%    100.0%     93.3%    100.0%     80.0%
Chemistry                 55.6%     52.6%     33.3%     --        20.0%
Coagulation               33.3%     89.5%     93.3%     66.7%    100.0%
Hematology/Urinalysis    100.0%    100.0%    100.0%    100.0%    100.0%
Microbiology              66.7%     63.2%     53.3%     66.7%     80.0%
Pathology/Cytology        --        94.7%    100.0%     66.7%    100.0%

                               DISTRIBUTION (% OF CASES)
                                        SURGEON

                            F        G         H       Bench

Blood Bank               100.0%    100.0%    100.0%     57.2%
Chemistry                 50.0%     40.0%     37.5%     33.2%
Coagulation               14.3%    100.0%     87.5%     43.3%
Hematology/Urinalysis    100.0%    100.0%    100.0%    100.0%
Microbiology              64.3%     60.0%     50.0%     29.9%
Pathology/Cytology        14.3%     80.0%     25.0%     97.9%
TABLE 8

SURGEON SUMMARY BY BLOOD BANK SERVICES

                        FREQUENCY (AVERAGE UNITS)
                                 SURGEON

                 A         B         C         D         E

Red Cells       2.8       1.9       2.9       2.3       1.5
Processing      2.3       3.1       4.3       3.0       2.7
Crossmatch      2.6       2.6       3.1       2.3       1.5
ABO Rh Type     1.1       1.2       1.0       1.0       1.0
Ab Scrn         1.1       1.2       1.0       1.0       1.0
Bbank Other      --       6.3        --        --        --

                     FREQUENCY (AVERAGE UNITS)
                             SURGEON

                 F         G         H       Bench

Red Cells       2.5       2.2       2.2       2.2
Processin$      4.5       3.3       2.8       1.4
Crossmatch      3.1       2.2       2.5       2.4
ABO Rh Type     1.2       1.0       1.1       1.0
Ab Scrn         1.2       1.0       1.1       1.0
Bbank Other      --        --       3.5       5.0

                        DISTRIBUTION (% OF CASES)
                                SURGEON

                 A         B         C         D         E

Red Cells       66.7%     78.9%    80.0%     100.0%    40.0%
Processing      88.9%     89.5%    80.0%      66.7%    60.0%
Crossmatch     100.0%    100.0%    93.3%     100.0%    80.0%
ABO Rh Type    100.0%    100.0%    93.3%     100.0%    80.0%
Ab Scrn        100.0%    100.0%    93.3%     100.0%    80.0%
Bbank Other       --      15.8%      --         --       --

                      DISTRIBUTION (% OF CASES)
                              SURGEON

                  F        G         H        Bench

Red Cells      100.0%    100.0%     62.5%     24.1%
Processing      78.6%     80.0%     62.5%     23.0%
Crossmatch     100.0%    100.0%    100.0%     43.9%
ABO Rh Type    100.0%    100.0%    100.0%     15.0%
Ab Scrn        100.0%    100.0%    100.0%      7.5%
Bbank Other       --        --      25.0%      1.6%

REFERENCES

Billi, J.E., G.F. Hejna, F.M. Wolf, L.R. Shapiro, & J.K. Stross (1987) "The Effects of a Cost-Educating Program on Hospital Charges," Journal of General Internal Medicine 2(5):306-311.

Burns, L.R., J.A. Chilingerian, & D.R. Wholey (1994) "The Effect of Physician Practice Organization on Efficient Utilization of Hospital Resources," Health Services Research 29(5):583-603.

Chilingerian, J.A. & H.D. Sherman (1990) "Managing Physician Efficiency and Effectiveness in Providing Hospital Services," Health Services Management Research 3(1):3-15.

Davis, D.A., M.A. Thomson, A.D. Oxman & R.B. Haynes (1995) "Changing Physician Performance: A Systematic Review of the Effect of Continuing Medical Education Strategies," The Journal of the American Medical Association 274(9):700-705.

Eisenberg, J.M. (1986) Doctors' Decisions and the Cost of Medical Care, Ann Arbor, Michigan: Health Administration Press.

Gortmaker, S.L., A.F. Bickford, H.O. Mathewson, K. Dumbaugh & P.C. Tirrell (1998) "A Successful Experiment to Reduce Unnecessary Laboratory Use in a Community Hospital," Medical Care 26(6):631-642.

Hart, D.S. & A. Balas (1994) "Managing Physician Practice Patterns: Providing Information Feedback to Improve Quality Care & Reduce Cost," Missouri Medicine 91(3):138-139.

HCIA (19971) User's Guide: HCIA SoleSource[R] Hospital/Physician Profiler, Version 3.5, Baltimore: HCIA, Inc.

Johnson, C.C. & M. Martin (1996) "Effectiveness of a Physician Education Program in Reducing Consumption of Hospital Resources in Elective Total Hip Replacement," Southern Medical Journal 89(3):282-289.

Johnson, C.C., M. Martin, S.M. Epstein & J.D. Lee (1993) "The Effect of a Physician Education Program on Hospital Length of Stay and Total Patient Charges," The Journal of the South Carolina Medical Association 89(6):293-301.

Kay, R. & T. Wadworth (1995) "Utilizing Physician Leadership to Reduce Length of Stay," Medical Group Management Journal 42(5):66-70, 72-74.

Kramolowsky, E.V., N.L. Wood, K.L. Rollins, W.P. Glasheen & C.M. Nelson (1995) "Impact of Physician Awareness on Hospital Charges for Radical Prostatectomy," The Journal of Urology 154(1):139-142.

Oxman, A.D., M.A. Thompson, D.A. Davis & R.B. Haynes (1995) "No Magic Bullets: A Systematic Review of 102 Trials of Interventions to Improve Professional Practice," Canadian Medical Association Journal 153(10):1423-1431.

Pugh, J.A., L.M. Frazier, E. DeLong, A.G. Wallace, P. Ellenbogen & E. Linfors (1989) "Effect of daily charge feedback on inpatient charges and physician knowledge and behavior," Archives of Internal Medicine 149(2):426-9.

Dale Buchbinder
Greater Baltimore Medical Center (USA)

Clifford F. Melick
Greater Baltimore Medical Center (USA)

David P. Coil
Greater Baltimore Medical Center (USA)

Sylvia Moore
Saint Mary's Hospital (USA)

Address for correspondence: Dale Buchbinder, Department of Surgery, Greater Baltimore Medical Center, 656.9 N. Charles St., Suite 701, Baltimore, MD 21204, USA, [e-mail].