PEDIATRICS Vol. 108 No. 4 October 2001, pp. 827-834

What If Pediatric Residents Could Bill for Their Outpatient Services?

Manny Ng, MD^* and Stephen T. Lawless, MD, MBA^{, §}

From the Departments of ^* Pediatrics and  Anesthesia/Critical Care, Thomas Jefferson University, Philadelphia, Pennsylvania; and ^§ Nemours Office of Operational Assessment, Nemours Foundation, and the Alfred I. duPont Hospital for Children, Wilmington, Delaware.

	ABSTRACT

Top Abstract Methods Results Discussion Conclusion References

Objective.  We prospectively studied the potential of billing and coding practices of pediatric residents in outpatient clinics and extrapolated our results to assess the financial implications of billing inaccuracies. Using Medicare as a common measure of "currency," we also used the relative value unit (RVU) and ambulatory payment class methodologies as means of assessing the productivity and financial value of resident-staffed pediatric clinics.

Methods.  Residents were asked to submit voluntarily shadow billing forms and documentation of outpatient clinic visits. Documentation of work was assessed by a blinded reviewer, and current procedure terminology evaluation and management codes were assigned. Comparisons between resident codes and calculated codes were made. Financial implications of physician productivity were calculated in terms of dollar amounts and RVUs. Resource intensity was measured using the ambulatory payment class methodology.

Results.  A total of 344 charts were reviewed. Coding agreement for health maintenance visits was 86%, whereas agreement for acute care visits was 38%. Eighty-three percent of coding disagreement in the latter group was resulting from undercoding by residents. Errors accounted for a 4.79% difference in potential reimbursement for all visit types and a 19.10% difference for acute care visits. No significant differences in shadow billing discrepancies were found between different levels of training. Residents were predicted to generate $67 230, $87 593, and $96 072 in Medicare revenue in the outpatient clinic setting during each successive year of training. On average, residents generated 1.17 ± 0.01 and 0.81 ± 0.02 work RVUs for each health maintenance visit and office visit, respectively. Annual productivity from outpatient clinic settings was estimated at 548, 735, and 893 work RVUs in the postgraduate levels 1, 2, and 3, respectively.

Conclusion.  When pediatric residents are not trained adequately in proper coding practices, the potential for billing discrepancies is high and potential reimbursement differences may be substantial. Discussion of financial issues should be considered in curriculum development.  Key words:  ambulatory payment classes, graduate medical education, pediatric residency, physician billing and coding, prospective payment system, relative value units.

Health maintenance organizations and the Health Care Financing Administration (HCFA) redefined the payment systems for medical care during the 1990s. The traditional fee-for-service system has evolved to include various forms of discounted fee-for-service and capitation. Reimbursement issues are critical in any practice because they are related directly to financial performance. In a private practice, without access to endowments or subsidies for education or research, personal income is dependent on reimbursement. In larger health systems, even with added nonclinical subsidies, productivity increasingly is being measured in terms of physician productivity, as defined by billing activity, and resource consumption.

Third-party payers rely on billing codes to determine the proper reimbursement for services. These codes are based on the American Medical Association's Current Procedure Terminology (CPT)¹ system. This system measures the value of a patient encounter on the basis of the extent of history taking, physical examination, and medical decision making that is required. It follows that the accurate assignment of billing codes is important for reimbursement. Inaccuracies in billing could result in reduced revenues and, if perceived as intentional, penalties for fraud.²

No published studies have evaluated resident knowledge of billing issues. A few studies^3-6 examined resident productivity in the context of cost analysis to estimate the financial benefits of the outpatient portion of a residency program. In this article, we examine the potential of a pediatric continuity clinic as a billing entity if shadow-billed by residents and discuss the financial implications of our findings in terms of application to a clinical practice as well as in the scope of clinical supervision of a resident-staffed clinic.

	METHODS

Top Abstract Methods Results Discussion Conclusion References

Participants

The Thomas Jefferson University/Alfred I. duPont Hospital for Children pediatric residency program trains 54 residents in a 3-year program. Residents from this program staff 3 clinical sites as part of their continuity of care requirement for the internal Residency Review Committee as established by the Accreditation Council for Graduate Medical Education. Each resident attends continuity clinic for 1 afternoon a week. In addition, each resident spends 6 weeks of each year providing care in the general outpatient clinic setting. All sites are supervised by attending physicians, who review every patient visit. Education regarding billing and practice management is not included in the residency curriculum. Participation in the study was voluntary. The reviewer and residents were blinded to any actual attending billing documentation. Attending supervision, patient contact, and billing were done as per normal guidelines and independent of the resident shadow billing study. This study was approved by the institutional review board of the Alfred I. duPont Hospital for Children and the affiliated study sites.

Instrument

All residents were given shadow billing forms identical to those used by attendings at each site. Residents were asked to record CPT codes, procedures, and diagnoses that they believed were relevant to each visit to mimic a practice setting. A blinded copy of the resident note for each visit was made and attached to the corresponding resident shadow billing form for the purpose of coding review by a single reviewer. Each note was analyzed for documentation of specific aspects of history taking, physical examination, and decision-making complexity, as described in the 2000 Current Procedure Terminology Manual.¹ An algorithm was constructed on the basis of the CPT manual to calculate an appropriate evaluation and management (E/M) code for each record. Health maintenance visits (HMVs; well visits) refer to visits that are assignable to CPT codes 99381 to 99395, and acute care visits refer to all other clinic visits that would have received CPT codes from 99201 to 99215.

Validity Testing

A computer-generated random sample (n = 16) of office visit notes was selected for review by a hospital billing liaison who specializes in outpatient billing and is a certified procedural coder tested by the American Academy of Professional Coders. The liaison assigned E/M codes for each chart on the basis of the given documentation, and comparisons were made to the reviewer's and residents' shadow coding assignments.

Data Collection and Financial Analysis

Data were collected during a 3-month period. Announcements were made, and electronic mail messages were sent to each resident to encourage participation. All patients seen by residents at the 3 sites were eligible for inclusion. Each chart was analyzed as described above for documentation and assignment of an E/M code. Data from residents' study sheets were entered into a database. Discrepancies between the E/M shadow codes assigned by the reviewer and the resident were recorded. The procedures commonly performed were routine immunizations and standard laboratory tests. Procedure codes were not analyzed for accuracy in this study, because the majority of procedures performed in our clinics were not expected to involve interpretation errors.

The National Physician Fee Schedule Relative Value File⁷ for calendar year 2000 was used for relative value unit (RVU) assignment. Differences in work RVUs (RVUws) and total facility RVUs based on the computed and resident-assigned E/M codes were calculated. Potential reimbursement values were calculated by matching E/M codes with 1998 to 1999 Medicare and Medicaid reimbursement rates available from the American Academy of Pediatrics Division of Health Policy Research.⁸ This information provided state-specific reimbursement rates for CPT codes, which were compiled from surveys of Medicaid directors of each individual state. We used the national Medicaid average in our computations. In addition, Medicare reimbursement rates based on RVUs were included, using the year 2000 nongeographically adjusted rate of $36.61 per RVU. Resource intensity was measured by using the methodology of the Medicare Outpatient Prospective Payment System.⁹ In this methodology, CPT codes are classified into ambulatory payment classes (APCs). Each APC is designated a service weight that is used to determine bundled reimbursement for services. The year 2000 nonwage adjusted rate for reimbursement is $48.49 per APC unit.

A model was constructed using Microsoft Excel (Redmond, WA) spreadsheet software to perform sensitivity analyses of potential clinical scenarios of operation. The adjustable variables for this analysis were proportion of acute care visits, number of patients seen per clinic session, number of clinic sessions per year, resident coding error rate, attending salary, and resident to faculty as a supervisory ratio. We used the average number of RVUs and APCs that were billable based on documentation to calculate annual revenues. The outcome variables in this analysis were annual revenues based on RVU and APC units.

Attending costs were calculated using the median salary of a general pediatrician in a group practice of $120 000 and a 20% benefit amount. Attending productivity was calculated on the basis of the number of RVUs that were generated by residents. We assumed that attendings would be able to collect 100% of Medicare net revenue (charge minus contractual adjustment) generated under their supervision. Salary shortfall was defined as the difference in attending costs between having only 1 attending supervise each clinic hour and the number of attendings required based on resident-to-attending ratio.

Statistical Analysis

Sample size 95% confidence interval was estimated by the following: Data were entered into SPSS software for Windows (version 10.0; SPSS, Chicago, IL) for analysis. Analysis of variance was used to determine differences between groups, and Cronbach's test was used to test for reliability. Categorical variables were compared by ² analysis, whereas analysis of variance was used to compare continuous variables. Data are presented as mean ± standard error of the mean. P < .05 was considered significant.

	RESULTS

Top Abstract Methods Results Discussion Conclusion References

Demographics

Data from 344 patient visits were collected, reflecting approximately 13% of all visits during the collection period. Thirty-four (57%) residents submitted an average of 10.1 patient encounters (median: 6.5; range: 1-43). First-year residents contributed 45% of all charts during this period; second- and third-year residents contributed 29% and 26%, respectively. The number of residents in each class that contributed to this study was similar (range: 10-12).

Fifty-four percent of patients were seen in a continuity clinic setting, and the remainder were seen in the general clinics. Sixty-three percent of visits were HMVs. The median age of patients was 27.5 months (range: 6 days-20 years). Approximately half of the HMVs were for children during the first year of life, whereas most acute care visits were for children ages 5 to 11 (Table 1).

The majority of acute care visits were shadowed as 99212 (30%) or 99213 (60%), whereas the documentation supported a higher level of service in most cases. On the basis of chart review, 55% of all acute care visits could have been shadow-billed as a level 3 visit, and 32% met criteria for a level 4 visit (P < .05; Table 2).

Reliability

An independent billing liaison was asked to review a random sample of charts to test for the reliability of investigator-assigned E/M codes. Of 16 charts (5%), agreement was reached on 15 cases (Cronbach's  = 0.96). The one discrepancy was an office visit for attention deficit, for which the investigator's E/M code was 99213 and the liaison's assigned code was 99212.

Coding Accuracy

Resident coding accuracy was assessed by comparing the assigned E/M shadow code with the code calculated on the basis of chart review. Overall agreement was 68% (233 of 344). There were no statistically significant differences in coding agreement between different classes of residents (Table 3). Eleven office visit records (3%) were assigned an E/M code for an HMV; 2 charts were not assigned E/M codes. These 13 charts were considered gross errors and were excluded from data analysis. Coding agreement for well and acute care visits was 86% and 38%, respectively. The frequency of error types is summarized in Table 4.

Financial Impact

Mean RVU and APC values for each visit are listed in Table 5. Estimated Medicaid and Medicare reimbursements for all visits are displayed in Table 6. Differences between resident-assigned and reviewer-assigned E/M codes accounted for a 4.79% discrepancy in the total number of RVUs and a 6.22% difference in RVUws. A 19.1% discrepancy in RVUs was found in the analysis of acute care visits, with a 23.4% difference in RVUws (see Table 3). No statistically significant differences were noted between residents with different levels of training. Analysis of RVU differences based on coding discrepancy is listed in Table 4. Seventy-two percent of RVU differences were accounted for in cases in which acute care visits were assigned a lower level of service by residents than by the investigator. Based on data from the clinics in this residency program, an additional $43 676 in Medicaid-equivalent funds may have been collected during a 1-year period for 15 200 patients.

Productivity

We entered into our sensitivity analysis model the average RVUs and APCs per visit. Using the distribution of visits and documented work that we found in this study and study site-specific general staffing of clinics by attending (3 residents per attending), we estimated that the average resident would produce 547.87, 734.88, and 893.39 RVUws in each respective year of training. Total RVUs, which include reimbursement for practice expenses and malpractice costs, produced by each resident was estimated to be 964.87, 1251.77, and 1493.29 annually. Resource intensity, as measured using the APC methodology, was predicted to be 658.00, 861.33, and 853.85 units annually. The dollar values of these projections are listed in Table 7. On the basis of the number of patients that residents are expected to see in the outpatient clinic setting in 1 year, we estimated the total Medicare revenue for the outpatient clinics in this program to be $4 516 123. The average estimated annual revenue generated by a resident in each year of residency was $67 230, $87 593, and $96 072 (Table 8).

We manipulated the adjustable variables in the sensitivity analysis model to estimate changes in productivity that would result from changes to the structure of the clinics. The results are illustrated in Fig 1. A change in resident-to-attending ratio from 3:1 to 4:1 accounted for a 33% increase in attending productivity as measured by RVU production. A 1 percentage point change in coding error or distribution of acute care visits resulted in 0.9% and 0.4% changes in overall productivity, respectively. An increase in the number of RVUws per visit by 0.1 would account for a 10.1% increase in productivity. These findings reflect that the greater number of patient visits that would be attributable to each attending is the major variable associated with attending clinic RVUw productivity.

View larger version (7K): [in this window] [in a new window]   Fig. 1.   Marginal impact of changes in clinic factors on attending productivity. R:A Ratio, resident to attending ratio change from 3 to 4; RVUw, relative value units to offset physician salary change by 0.1 per visit; Acute Care Visits, change in proportion of acute care visits by 1 percentage point; Error Rate, change in error rate by 1 percentage point.

We used our model to calculate the optimal resident-to-attending ratio that would balance productivity with attending salary and administrative time requirements and impact an academic practice. On the basis of the total number of clinical hours, we calculated that, at a minimum, 10.7 full-time clinic-based attendings would be required to staff the clinics. We found that a resident-to-attending ratio of 3.34:1 and 4.83:1 would be required to match the 20th and 50th percentile benchmarks for attending productivity, as measured by RVUw (Fig 2). To meet the 50th percentile for productivity, 12 attendings would have to supervise 4.5 residents, spend 89% of their hours in the clinic, and require $1.7 million in salary and benefits, with a salary shortfall of $187 754. To meet the 20th percentile for productivity, 17 attendings would supervise 3.2 residents and spend 63% of their hours in clinic but with a cost of $2.4 million and a shortfall of $907 754. This shortfall would be worsened to $1 361 631 if the clinic were entirely a Medicaid population.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Projected attending productivity and comparison with American Medical Group Association17 data for pediatricians in a multispecialty group practice. The salary shortfall (right side y axis) is calculated as a function of the resident:attending ratio chosen assuming a "fixed" total productivity per resident. Similarly, using the same resident productivity, the productivity credited per attending (left side y axis) is correlated to the chosen resident:attending ratio. The horizontal lines represent the AMGA General Pediatric RVUw/attending percentiles.

We also used the model to calculate the resident-to-attending ratio at which revenue generated by a resident-staffed clinic would match the attending salary costs. Assuming that all resident services are billable by their attendings and 100% Medicare reimbursement of RVUs, a clinic could break even with a resident-to-attending ratio of 3.5:1. In this scenario, 15.5 full-time attendings would work 69% of their hours in clinic with an average salary of $120 438 and a 20% benefit package. Additional clinic costs, which include facility costs and salaries for residents and ancillary staff, were not included in this calculation.

A more accurate measure of physician reimbursement would be based on RVUws, which are used to offset physician salaries. A 5.4 resident-to-attending ratio would be required to match physician salary costs with reimbursement. However, 10 attendings would have to work 107% of a full-time week at a salary of $119 501. Because this is not a practical option, additional funding would be required.

	DISCUSSION

Top Abstract Methods Results Discussion Conclusion References

Few published studies have formally examined the billing practices of physicians. Chao et al¹⁰ observed 138 family practice physicians and found 55% accuracy in CPT coding, with discrepancies divided evenly between overcoding and undercoding. Horner et al¹¹ reviewed progress notes from 1253 visits and concluded that billing forms did not accurately reflect the level of service provided or the diagnoses that were treated. HCFA found an 8.0% overpayment rate for adult Medicare claims in 1999.²

We found agreement between resident shadow billing and case review in two thirds of all cases. However, analysis of the discrepancies reveals some simple errors that could be corrected easily before formal bill submission. Health maintenance visits are billed on the basis of only 2 variables: whether a patient has been seen by the practice in the last 3 years and patient age. In this study, 14% of these visits were shadow miscoded. Half of these errors were attributable to coding for the wrong age. Whereas the financial implications of these errors are minimal in terms of penalties, they do represent a simple error that is easily avoidable.

Agreement for acute care visits (CPT 99201-99215) was 38% in this study. This is lower than the 55% found in Chao's study.¹⁰ We found that shadow downcoding was much more common than shadow upcoding in our study, in comparison with an even distribution in Chao's study. No significant differences were found between residents at different levels of training in terms of coding disagreement. This was expected, because >50% of residents receive significant training in practice management and billing issues.¹²

The coding scheme that is used by CPT categorizes services into several categories. For the purpose of the primary care provider, the relevant categories are divided by site of service (inpatient vs outpatient), type of patient (new vs established), and level of visit complexity. Services are billed according to 5 levels of visit complexity, as defined in the CPT manual,¹ and adherence to those guidelines is monitored by third-party payers. The level of accuracy that is discovered on chart review influences the decisions on payment and penalties along with intent. Because none of the residents had any impact on the actual billing process, intent should not have been a driver during this study.

We measured the costs of shadow billing discrepancies in terms of RVUs. A significant proportion (82%) of RVU differences in this study was attributed to acute care visits, which accounted for 37.5% of all visits in our sample. This can be explained by the fact that the difference between the number of RVUs assignable to different acute care visits is much greater than with HMVs. Although the incidence of errors was 62% for acute care visits, the cost in terms of RVUs was a 19.1% net decrease, based on reimbursement for the incorrect E/M code. Because we are assuming that reimbursement will be provided for both correctly coded and miscoded charts, the total RVU difference is less than the error rate (Fig 3). Actual penalties also would include a fine per occurrence plus interest. Payers conceivably could penalize for undercoding. In addition to the reimbursement lost in undercoding, undercoding could be perceived as a discount to entice patients.

View larger version (11K): [in this window] [in a new window]   Fig. 3.   Marginal impact of coding errors on RVU assignment.

In 2000, Medicare implemented the use of APCs to pay hospitals prospectively for outpatient services.⁹ This system was designed by HCFA to control costs by making bundled payments based on the costs that hospitals may incur to provide specific outpatient services. Specific services, as defined by their CPT codes, are categorized APC groups, each of which carries a specific weight for reimbursement. For each visit, Medicare will offer a predetermined amount of funds as a bundled payment for nonphysician services, such as drugs and supplies. A complete list of services that are covered by the APC system is listed in Table 9. The APC is mandated only for Medicare but can be used as a benchmark comparison. The APC system does not provide reimbursement for the physician service portion of visits; the RVU system is the system of physician reimbursement. Acute care visits are categorized into 3 groups, with payments based on severity of illness. A hospital can collect $80.49 for a high-level visit (CPT 99214 or 99215), $48.49 for a mid-level visit (CPT 99213), and $47.52 for a low-level visit (CPT 99211 or 99212). The substantial difference in payments between high- and mid-level visits accounts for the larger relative difference in reimbursement based on reviewer-assigned and resident-assigned codes found in Table 6.

We used the APC and RVU data to calculate the cost ratios of physician services. This type of information is used in physician profiling for the purposes of cost containment, benchmarking, and potential billing quality assurance. The volume performance standards originally were used to compare the volume of services provided per admission between different hospitals.¹³ Hospitals that provided an excessive ratio of services were subject to withholding of reimbursement as an incentive to control costs. We found that our residents provided mean APC to RVU ratios of 0.53 ± 0.08 and 1.08 ± 0.02 for HMVs and acute care visits, respectively (Table 5). No reference data are available for comparison.

In this sample of residents, the financial implications of shadow billing errors accounted for a 4.3% to 6.7% difference in net revenue, depending on the method of reimbursement. This is lower than the estimated 8.0% overpayment rate for Medicare claims in 1999.² If one were to look at acute care visits exclusively, then the shadow errors account for a 15% to 20% difference in net revenue. To use this information to estimate the total financial implications of shadow billing errors, one would have to assume that this sample is representative of all clinic visits in a given time period, especially in respect to the ratio of well and acute care visits. Our model allows for adjustment of this ratio. Also, capitated systems would be affected by billing errors in a more indirect manner, because the distribution of charges and services provided are important in negotiating contracts and capitation amounts with third-party payers.

Although shadow undercoding was common in this study, the estimated financial losses are less than the proportion of miscoded charts. Although this number may seem small, the practical implications may be considerable. In a busy practice, a 5% increase in net revenue may support an additional staff member. We did not calculate penalties into our analysis. If a third party were to penalize providers for improperly coded bills, then the significance of billing errors would be more pronounced. The current HCFA penalty for overcoding is 3 times the value of the visit.¹⁴

In comparison with data collected in 1995 by the American Academy of Pediatrics (AAP),¹⁵ our residents' documentation reflects a greater proportion of high-level visits than pediatricians in practice (Fig 4). The distribution of assigned sick visit codes in our study is similar to those reported by the AAP in 1995, which compiled data from pediatric practices nationwide. In our sample, 31% of acute care visits are shadow-billed a 99212, and 61% received a 99213 code. The AAP sample of attending pediatricians assigned a 99212 to 23% of acute care visits and a 99213 to 46% of acute care visits. However, E/M codes as based on chart review in our study show a distribution toward higher levels of service. Also, national data from the Medical Group Management Association databases¹⁶ show that pediatricians tend to bill a greater proportion of 99212 and a lesser proportion of 99214 than do primary care practitioners in other specialties. Does this reflect a tendency to undercode, or do pediatricians simply perform less-complex services? It may be of interest to examine this question further.

View larger version (11K): [in this window] [in a new window]   Fig. 4.   Distribution of E/M codes for acute care visits.

Although there have been no published studies to evaluate resident knowledge of billing issues, a few studies^3-6 have examined resident productivity in the context of cost analysis to estimate the financial benefits of a residency program. Furthermore, no published studies have used the RVU and APC mechanisms to calculate reimbursements. Jones et al⁴ found that family medicine residents generated $8447, $25 529, and $35 525 in charges annually in each respective year of residency. Although our study did not use actual financial data, we estimate that our residents would generate more system-wide net revenue based on Medicare payments (Table 8).

The information in Table 5 shows that the average productivity per visit, as measured in RVUws, exceeds the 80th percentile benchmark (0.82 RVUw) for general pediatricians in a multispecialty practice.¹⁷ This reflects the higher level of service that our residents documented for acute care visits, as illustrated in Fig 4. Another factor that would affect productivity is the proportion of well to acute care visits. Because HMVs are eligible for higher levels of reimbursement under Medicare, a practice that sees more well visits would be more productive, assuming that these visits are covered by insurance. Total annual productivity values are not comparable, because our model accounts only for outpatient clinical services, which compose approximately 20% of resident scheduled time.

A few studies have examined the effect of supervising residents on hospital and attending productivity. Albritton et al¹⁸ found that internal medicine attendings earned more RVUs when they supervised residents in a continuity clinic setting than they did in a private practice setting or while supervising medical students using preliminary RVU values. The Urban Institute¹⁹ estimated graduate medical education costs to be 2055 RVUs per resident per year in their analysis of inpatient hospital expenditures. Our model, based on year 2000 RVUs, predicted that each resident would generate 1236 RVUs annually in the outpatient clinic setting. A total of 725 of those RVUs theoretically would be used to offset physician salary costs. Considering that approximately 20% of annual resident hours, excluding on-call time, are scheduled for clinic in our program, the number of potential RVUs that may be generated by our residents should equal if not exceed the costs of education.

Other factors that need to be considered in an actual clinic setting include no-show rates and variations in reimbursement. The pediatric clinics in most residency programs care for a large proportion of patients with Medicaid, which tends to reimburse at a lower level than Medicare. Thus, additional funding would be required to support the outpatient clinics, because the Medicare numbers in our calculations probably reflect an overestimation of revenue.

Our model allows for adjustment of several variables that will help predict productivity in an outpatient clinic. The variable that accounts for the most pronounced change in productivity is the resident-to-attending ratio. We found that under current economic conditions, the reimbursement for physician services, as measured by RVUw, does not meet the salary costs of the physicians. Either the number of patients would have to be increased or salary requirements would have to be reduced. The addition of funds to finance the salary shortfall may help to balance a clinic's budget.

Our methodology has several limitations. Because participation by residents was voluntary, we are assuming that the reviewed records are representative of all clinic visits, both in patient characteristics and in resident billing practices. Random chart audits with extrapolation to more charts is a common methodology of auditors. Although some selection bias may exist, we estimated a 95% confidence interval of ±5% for our results.

We deliberately did not obtain any information regarding attending billing to obviate potential problems with payers. Attending physicians have all bills submitted by qualified coders and ongoing compliance audits.

We used many assumptions in our model. The actual volume of visits in the resident-staffed clinics may vary, depending on factors such as the no-show rate, resident-to-attending ratio, and the use of telephone triage. We also used Medicare-based values for measuring productivity and reimbursement. Although the majority of reimbursement in our clinics comes from Medicaid, we believe that the Medicare-derived RVU and APC systems will serve as a model for other payers for determining reimbursement and resource utilization. Changes and costs are too sensitive to local practice and hospital financial and operational issues and cost shifts. Applying Medicaid systems as a "common valuation currency" allows easier adaptation of standardized productivity units by individual hospitals with subsequent variability in net revenue projections. Information to assist with these calculations has been made available by the AAP.²⁰ The actual difference in reimbursement between Medicaid and Medicare will vary from state to state, as partly reflected in Table 6.

	CONCLUSION

Top Abstract Methods Results Discussion Conclusion References

This study found that shadow-billing discrepancies were common among pediatric residents in this program. This was not surprising, because these residents do not receive any training regarding financial and business issues associated with practicing medicine. The emphasis of residency is to teach doctors to refine their patient care skills and their medical knowledge for a specific area of medicine. However, many physicians in practice have to struggle with the business issues of practicing medicine when their residency is complete. We found that our residents documented more intensive services than the average pediatrician but tended to undercode, which would result in decreased reimbursement in an actual billing situation. Under current economic conditions, a resident-staffed pediatric clinic cannot be financially self-sufficient. Because the reimbursement for medical services is evolving constantly, pediatricians in training need more preparation for the financial issues of practice management that they must face in the real world.

	FOOTNOTES

Received for publication Oct 18, 2000; accepted Feb 12, 2001.

Reprint requests to (S.T.L.) Nemours Office of Operational Assessment, Nemours Foundation, 1600 Rockland Rd, Wilmington, DE 19899. E-mail: slawless{at}nemours.org

	ABBREVIATIONS

HCFA, Health Care Financing Administration; CPT, current procedure terminology; E/M, evaluation and management; HMV, health maintenance visit; RVU, relative value unit; RVUw, work relative value unit; APC, ambulatory payment class; AAP, American Academy of Pediatrics.

	REFERENCES

Top Abstract Methods Results Discussion Conclusion References

1. American Medical Association. Current Procedural Terminology 2000 Professional Edition. Washington, DC: American Medical Association; 1999
2. Inspector General. Improper Fiscal Year 1999 Medicare Fee-for-Service Payments. Washington, DC: Department of Health and Human Services; 2000 Memorandum No: A-17-99-01999
3. Flanagan T, Mitchner B, Weyl-Feyling D, Laros RK Jr The cost of teaching residents outpatient obstetrics and gynecology in a university medical center. Obstet Gynecol 1995; 86:1014-1017 [CrossRef][Medline]
4. Jones TF, Culpepper L, Shea C Analysis of the cost of training residents in a community health center. Acad Med 1995; 70:523-531 [Medline]
5. Franzini L, Berry JM A cost-construction model to assess the total cost of an anesthesiology residency program. Anesthesiology 1999; 90:257-268 [CrossRef][Medline]
6. Kearl GW, Mainous AG, 3d Physicians' productivity and teaching responsibilities. Acad Med 1993; 68:166-167 [Medline]
7. Health Care Financing Administration. National Physician Fee Schedule Relative Value File Calendar Year 2000. Available at: http://www.hcfa.gov/medicare/pfsmain.htm. Accessed December 2000
8. American Academy of Pediatrics. Medicaid Reimbursement Survey Fixed Fee Schedule 1998/1999. Available: http://www.aap.org/research/medreimPDF/fixedfee. PDF
9. Prospective Payment System for Hospital Outpatient Services; Final Rule. 42 CFR Parts 409, 410, 411, 412, 413, 419, 424, 489, 498, and 1003 (April 7, 2000)
10. Chao J, Gillanders WG, Flocke SA, Goodwin MA, Kikano GE, Stange KC Billing for physician services: a comparison of actual billing with CPT codes assigned by direct observation. J Fam Pract 1998; 47:28-32 [Medline]
11. Horner RD, Paris JA, Purvis JR, Lawler FH Accuracy of patient encounter and billing information in ambulatory care. J Fam Pract 1991; 33:593-598 [Medline]
12. Mulvey H, Ogle-Senett E, Cheng T, Johnson R Pediatric residency education. Pediatrics 2000; 106:323-329 [Abstract/Free Full Text]
13. Welch WP, Miller ME Proposals to control high-cost hospital medical staffs. Health Aff (Millwood) 1994; 13:42-57 [Abstract]
14. Inspector General. Medicaid Post Payment Safeguards. Washington, DC: Department of Health and Human Services; 2000 Memorandum No: OEI-05-99-00072
15. American Academy of Pediatrics. Utilization and Charges for Pediatric Services in 1995. Washington, DC: American Medical Association; 1996
16. Physician's Services Practice Analysis Data. Englewood, CO: Medical Group Management Association Center for Research. Available at: http://www.mgma.com/g/crahca/emcodes. Accessed December 2000
17. American Medical Group Association. 1999 Medical Group Compensation and Productivity Survey. Alexandria, VA: American Medical Group Association; 1999
18. Albritton TA, Miller MD, Johnson MH, Rahn DW Using relative value units to measure faculty clinical productivity. J Gen Intern Med. 1997; 12:715-717 [CrossRef][Medline]
19. Miller ME, Welch WP. Analysis of Hospital Medical Staff Volume Performance Standards. Technical Report. Washington, DC: The Urban Institute; 1993 Report No. 6210-01
20. American Academy of Pediatrics. 2001 RBRVS: What It Is and How Does It Affect Pediatrics? Elk Grove Village, IL: American Academy of Pediatrics; 2001