1337

Productivity of Radiologists:Estimates Based on Analysis of RelativeValue Units

Patrick M. Conoley1 Analysis of relative value units (RVUs) was used to quantify patient-care productivitySally W. Vernon2 of radiologists in 19 multispecialty group practices and to determine how productivity is

affected by certain characteristics of the practices. The RVUs used in this study are theprofessional component RVUs developed by the American College of Radiology and theHealth Care Financing Administration and published as the Radiology Relative ValueScale. An RVU workload was calculated by multiplying the number of times eachprocedure was performed by the procedure’s corresponding RVU; the sum of theseproducts gave the overall professional RVU workload. Five productivity Indexes werecalculated. The physician index denotes the ratio of the total number of physicians inthe clinics to the total number of radiologists. The availability index denotes the fractionof radiologists who are available to perform clinical work after deductions are made fortime away from clinical work. The difficulty index measures, in RVUs per examination,

the level of complexity of the overall examination mix. The examination index measuresexaminations per available radiologist, and the RVU index measures RVUs per availableradiologist. Altogether, the 19 clinics reported 3,234,451 examinations performed by 299radiologists. The computed overall indexes were as follows: physician index = 20

physicians per radiologist; availability index = 0.77; difficulty index = 2.27 RVUs per

examination; examination index = 14,098 examinations per year per available radiolo-gist; RVU index = 32,065 RVUs per year per available radiologist. When the clinics weregrouped according to characteristics of the practices, the RVU index was higher forsingle-site practices, high-prepaid practices, outpatient-only practices, and practiceswithout radiology training programs. Fifty-two percent of the RVUs were in generalradiology, 37% in sectional imaging, and 10% in special procedures.

The concept of RVU workload is timely because it undoubtedly will be used to compareworkloads across medical subspecialties, and these workloads are likely to be relatedby third-party payers to compensation.

AJR 157:1337-1340, December 1991

The American College of Radiology (ACR) and Health Care Financing Adminis-tration have quantified radiology services for purposes of reimbursement by usingrelative value units (RVUs) [1]. The service of interpreting a posteroanterior chestradiograph was assigned the value of 1 .00 RVU. All other procedures in theradiology section of the Current Procedural Terminology (CPT) coding system [2]were assessed in relation to the service of reading a posteroanterior chest radio-

Received March 21 , 1991 ; accepted after revi- graph, and each procedure was assigned a value in RVUs. The list of these valuession July 24, 1991 . is the Relative Value Scale. The scale rates both technical and professional

1 Kelsey-Seybold Clinic, P. A., 6624 Fannin St., components of each procedure. In the assignment of the professional component,Ste. 1800, Houston, TX 77030. Address reprint consideration was given to such factors as the training, knowledge, skill, stress,requests to P. M. Conoley. . . .

and time required of the radiologist to perform the procedure. Thus, the scale

Houston,School � measures the overall reimbursable service provided by the radiologist during the77225. procedure. The Relative Value Scale is analogous to the Resource-Based Relative0361-803X/91/1576-1337 Value Scale, which will be applied by Medicare to nonradiology physician reimburse-0 American Roentgen Ray Society ment in 1992 [3].

Dow

nloa

ded

from

ww

w.a

jron

line.

org

by 2

02.6

2.17

.27

on 1

0/12

/15

from

IP

addr

ess

202.

62.1

7.27

. Cop

yrig

ht A

RR

S. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved

In this study, we used the RVU concept to quantify theproductivity of radiologists in multispecialty group practices,to determine how productivity is affected by certain charac-teristics of the practices, and to provide a basis for analyzingsecular trends in the distribution of radiologists’ productivityamong the various imaging techniques. The study acceptsthe Relative Value Scale as the measure of productivity. Itshould be noted, however, that only work devoted to the careof patients is included in the CPI and Relative Value Scalesystems. Many other activities of radiologists do not haveRVUs assigned to them; these include continuing education,research, didactic teaching, clinical instruction of residents,and administrative duties. In this paper, the radiologist re-sources allocated to these non-RVU efforts are removed fromthe staffing analysis through the concept of “available radiol-ogists,” defined in the Methods section.

The study of total professional RVU workload of radiolo-gists is timely and relevant because the RVU workload un-doubtedly will be compared with the total Resource-BasedRelative Value Scale workload of other specialists, andthese workloads will be related by third-party payers tocompensation.

Methods

Twenty-two large, multispecialty clinics, scattered across theUnited States, that are members of the American Society of ClinicRadiologists were asked to provide a comprehensive list of the

volume of procedures performed during a recent 1 2-month period for

each of the CPT codes in the Diagnostic Radiology section of theRelative Value Scale [1 , 2]. Nuclear imaging and radiotherapy pro-cedures were not included in the study. New diagnostic radiologyCPT codes for 1990 were included because some of the clinics wereusing these codes, even though RVUs have not yet been assignedto the procedures by ACA and the Health Care Financing Administra-tion. We assigned the new codes AVUs equivalent to AVUs of similartechniques so that all reported examinations would be included in theAVU analysis. These new codes accounted for approximately 0.7%of the examinations and approximately 2% of the total RVUs in thestudy. No audit was performed to determine if the clinics used theCPT-coding system accurately or to ascertain whether examinations

were underreported or overreported. This study assumes that differ-ences between clinics in the use of the CPT-coding system haveintroduced only random error into the data.

Of the 22 clinics, 21 responded to the questionnaire, but only 19provided a comprehensive list of CPT-code volumes. These 1 9 wereincluded in the RVU analysis. Nine of the clinics were located in the

Midwest, four in the South, three in the Northeast, and three in theWest/Southwest. Data were submitted for 12-month periods withending dates from as early as December 31 , 1 989, to as recent asSeptember 30, 1 990. Two practices estimated volumes for calendar-year 1990 on the basis of two thirds of the year’s work.

An AVU workload was calculated by multiplying the number oftimes each procedure was performed by its corresponding profes-sional-component RVU. The sum of these individual CPT workloadsgave the entire professional RVU workload. The total AVU workloadand the total number of examinations provided numerators for theproductivity calculations.

Practice styles have considerable variation both in the amount oftime off and in types of work that cannot be expressed in RVUs. Indetermining denominators for the calculations, an attempt was madeto adjust for these differences by reducing the total number ofradiologists in the practice by a self-reported amount of “radiologists

not available for clinical work.” “Not available” deductions were madefor continuing medical education, vacations, academic activities, and

administrative duties. Residents were not included in the quantifica-tion of radiology staffing. Practices also were asked to report the

total number of physicians in the multispecialty practice.Five productivity indexes were calculated, as defined in Table 1 . A

physician index was calculated as the ratio of the total number of

physicians (“headcount”) in the multispecialty clinics to the total

number of radiologists (headcount) in the clinics. An availability index

was calculated as the ratio of the number of available radiologists tothe total number of radiologists, indicating the fraction of time theradiologists were available to do clinical work. A difficulty index was

calculated by dividing the total AVU workload by the total number ofexaminations to express in AVUs per examination the weighted levelof “difficulty” of the procedures performed. Finally, two workload

indexes were calculated: an examination index (the total number ofexaminations per year per available radiologist) and an RVU index(the total number of RVUs per year per available radiologist).

Other data were collected in the survey in order to categorize theclinics according to selected variables that might affect workload.

The criteria and resultant clinic groupings are presented in Table 2.Productivity indexes were calculated on the combined data submitted

by all the clinics and by clinic categories on the combined data of all

the clinics within each category. The indexes also were calculated foreach clinic to obtain mean values, standard deviations, and standarderrors of the indexes among the clinics.

Each CPT-code was also categorized to permit analysis of theworkload distribution among examination techniques. The categorieswere (1) general radiography and fluoroscopy (head and neck, chest,extremity, and spinal plain film radiography; genitourinary, gastroin-

testinal, musculoskeletal, and intrathecal contrast studies; and mam-

mography), (2) sectional imaging (sonography, MA, and CT of thehead and body), (3) specials (angiography, neuroangiography, andinterventional), and (4) other (unlisted procedures, review of outsidefilms). In a previous report, Johnson and Abernathy [4] provided a

breakdown by imaging technique of projected national radiologicprocedures in the United States for 1 980 that we compared with ourdata. Nuclear procedures reported in their study were excluded in

our analysis because that technique was not included in our survey.Although the Relative Value Scale did not exist in 1980, a weighteddifficulty index calculated from the current data for each technique

category was used to estimate the RVU workload that would havebeen attributed to that technique in 1 980. Thus, the distribution of

the 1 980 AVU workload among techniques could be estimated.

Results

Altogether, these 1 9 practices reported 3,234,451 exami-nations, corresponding to 7,356,462 RVUs. These were per-formed by 299 radiologists among the total of 6055 physiciansin the clinics. The productivity indexes are presented in Table2. Based on the combined data of all clinics, the examinationindex is 14,098 examinations per year per available radiolo-gist, and the RVU index is 32,065 RVUs per year per availableradiologist. If the availability correction is omitted, there were1 0,81 8 examinations and 24,604 RVUs per year per radiolo-gist by “headcount.”

The questionnaire did not ask for volumes of “injectioncodes,” which should accompany procedures billed with “su-pervision and interpretation” CPT codes. Overall, 14,1 80 su-pervision and interpretation procedures were reported. Al-though RVUs have not yet been approved for injection codes,it is possible to estimate that approximately 90,000 RVUshave not been counted. Thus, in order to obtain a moreaccurate total RVU workload, the RVU index should be in-creased to 32,344 to correct for these lost RVUs.

As shown in Table 2, the physician index was high in thesingle-site practices, in the high-prepaid practices, and in the

Dow

nloa

ded

from

ww

w.a

jron

line.

org

by 2

02.6

2.17

.27

on 1

0/12

/15

from

IP

addr

ess

202.

62.1

7.27

. Cop

yrig

ht A

RR

S. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved

outpatient-only practices. The availability index was higher foroutpatient practices and practices without training programs.The difficulty index was quite low in the outpatient practicesand slightly high in multiple-site practices, low-prepaid prac-tices, inpatient practices, and very large practices, as well asin those with radiology training programs. The examinationand RVU indexes were higher for single-site practices, high-prepaid practices, outpatient-only practices, the large as op-posed to very large practices, and the practices withoutradiology training programs.

Table 3 presents the breakdown of the reported examina-tions and the distribution of RVUs by technique category.Also, the 1980 data from Johnson and Abernathy [4] areincluded in the table for comparison. Their examination cate-gones have been grouped to correspond with our currentgroups. Sectional imaging represented 1 4% of all examina-tions in the clinics in 1 989 compared with 6% of all nationalexaminations in 1980, and general radiology represented 83%

of all examinations in 1989 compared with 93% in 1980. Theweighted difficulty index of sectional imaging of 5.86 RVUsper examination was higher than the difficulty index for gen-eral radiology of 1 .44. This fourfold difference in difficulty

TABLE 1: Definitions of Productivity Indexes

Index Definition

Physician

Availability

Difficulty

Total physicians

Total radiologists

Available radiologists

Total radiologists

Relative value units (RVUs)

Number of examinations

Examination

RVU

Examinations per yearAvailable radiologist

RVUS per year

Available radiologist

indexes accentuates the drop in general radiology from 74%of the national RVU workload in 1980 to 52% of the clinics’RVU workload in 1989. By percentage of examinations andRVUS, special procedures remained stable. The overallweighted difficulty index for all national procedures in 1980was 1 .82, whereas the overall weighted difficulty index in theclinic procedures in 1989 was 2.28, a 25% increase.

Discussion

The ACR research department was consulted to providean external reference to compare with our data. According toACR data (Sunshine J, Mabry M, personal communication),the examinations in this study represent approximately 1% ofthe total number of examinations performed by radiologistsin the United States in 1 989, and the radiologists in the studyconstitute approximately 1 % of the radiologists estimated bythe ACR to have been working in the United States in thatyear. The ACR reported a national mean of approximately11 ,700 examinations per radiologist per year, but this esti-mate does not take availability into account. From the 1987Medicare Part B Reimbursement data (Sunshine J, Mabry M,personal communication) consisting of 1 10,000,000 exami-nations, mostly diagnostic radiology, the ACR calculated anationally weighted difficulty index of 2.31 RVU per exami-nation for diagnostic radiology procedures performed on Medi-care patients. By using the 1 989 ACR examination index, the1987 Medicare difficulty index as an estimate of the nationaldifficulty index, and the 1989-1 990 overall availability indexfrom this survey as an estimate of national availability, a roughestimate of a national RVU index can be made: 35,1 00 RVUSper year per available radiologist. This calculated nationalfigure is probably an overestimate because the difficulty indexfor the Medicare population undoubtedly is higher than theindex for the general population, because of greater use ofsectional imaging and special procedures in older patients. In

TABLE 2: Productivity Indexes for Individual Clinics, All Clinics, and by Characteristics of Clinics

Type of Clinic or Characteristic No.Productivity Indexes

P1 Al DI El Rl�

Individual clinicsMean 19 23 0.77 2.25 15,231 33,705Standard deviation 7 0.07 0.24 4,577 7,855Standard error 1 .6 0.02 0.06 1 ,050 1,802

All clinics 19 20 0.77 2.27 14,098 32,065Number of sites

Single 6 24 0.77 2.15 16,410 35,268Multiple 13 19 0.77 2.31 13,551 31,306

Prepaid fraction�5% 7 17 0.76 2.31 13,440 31,017>5% 12 24 0.77 2.24 14,750 33,104

Inpatient fraction0% 3 30 0.80 2.01 18,666 37,464

>0% 16 19 0.76 2.30 13,737 31,639Radiology residenc?

�1 Resident 11 23 0.79 2.22 15,374 34,114>5 Residents 8 19 0.76 2.31 13,347 30,858

Size

�300,000 RVU (large) 10 22 0.75 2.19 14,843 32,564>300,000 RVU (very large) 9 19 0.77 2.31 1 3,762 31,841

Note.-Pl = physician index, Al = availability index, Dl = difficulty index, El = examination index, RI = RVU index.a Theoretically, Dl x El = RI. Discrepancies in the products and in the standard errors are due to rounding error.b No practice had a radiology training program with two to four residents.

Dow

nloa

ded

from

ww

w.a

jron

line.

org

by 2

02.6

2.17

.27

on 1

0/12

/15

from

IP

addr

ess

202.

62.1

7.27

. Cop

yrig

ht A

RR

S. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved

TABLE 3: Percentage Distribution of Examinations and of Professional Relative Value Units Among Examination Techniques

Technique

1989

(%)

1980

(%)

Examinations RVUS Examinations RVUS

General (radiography and fluoroscopy) 83 52 93 74Sectional imaging (CT, sonography, MA) 14 37 6 16Specials (vascular, interventional) 1 1 0 1 10

Other (outside films, miscellaneous) 2 1 - -

Note.-RVUs = relative value units. 1980 data from Johnson and Abemathy [4] are included for reference.

addition, almost two thirds of the radiologists in this study areinvolved in practices with radiology training programs, whichis probably a higher proportion than in the general populationof radiologists in the United States. Because as a group,practices with training programs have a lower RVU index thanpractices without training programs, the mean RVU index forour study is probably an underestimate of the national RVUindex. When these two points are kept in mind and when thestandard error of 23% is considered, this figure extrapolatedfrom ACR and Medicare data is fairly close to the RVU indexthat we report.

In the analysis by clinic characteristics, the high examinationand RVU indexes in the single-site clinics may indicate thatthese practices are more efficient, possibly because of theease of scheduling work assignments, immediate availabilityof cross coverage, or less time spent in commuting. Likewise,the high physician, examination, and RVU indexes of high-prepaid practices corroborate the managed-care philosophyof high-volume, low-cost care. Of interest, the difficulty indexis lower in the prepaid practices, possibly because utilization-review processes control access to the more advanced im-aging procedures.

The higher difficulty index of practices with an inpatientcomponent is due to the admixture of high RVU proceduressuch as interventional and angiographic work in the inpatientsetting. However, the higher availability, physician, and ex-amination indexes in the outpatient-only practices compen-sated for low difficulty, so that the workload as measured byRVU index was 18% greater in the outpatient-only settingcompared with the inpatient setting.

The practices with radiology training programs had a higherdifficulty index but lower examination and RVU indexes thanpractices without training programs. The lower productivitymay be due to an intrinsic inefficiency in teaching residentsduring film interpretation; however, this teaching is a validactivity within the mission of the sponsoring organizations. Itsimply is not measured by RVUs.

Seven of the nine very large practices (>300,000 RVUS/year) have training programs, whereas only one among the10 large practices (s300,000 RVU5/year) does. Thus, thecategorization by size largely duplicates the trends in thetraining/nontraining category, with difficulty indexes higherand workload indexes lower in the very large (training) clinicsthan in the large (nontraining) clinics.

The 1% of injection code RVUS not included in the surveydata were spread fairly evenly among the practices (0.01-4.05%) and probably do not represent a systematic error.These “lost RVUs,” however, may also represent lost reve-nues to the practices if corresponding injection codes werenot billed along with supervision and interpretation proce-dures.

Finally, Johnson and Abernathy’s study [4] was designedto project overall national volumes of examinations for 1980and the distribution of examinations among techniques. Ourdata show actual examinations performed by a subset ofpractitioners in multispecialty clinics in 1989. Although extrap-olation of our data to give an estimate of the 1 989 nationaldistribution is beyond the scope of this study, it is plausibleto suggest that the dramatic shift of RVU workload fromgeneral radiology to advanced imaging techniques (Table 3)reflects trends in the national data.

Conclusions

This study attempted to quantify work devoted to the careof patients that is included in the CPI and Relative ValueScale systems. The results of this RVU analysis are similar tothe results of extrapolations based on ACR and Medicaredata. Many other non-RVU activities of radiologists do nothave assigned RVUS but must be recognized as productivework. As third-party payers shift to RVU-based methods ofreimbursement, accurate and complete coding will be essen-tial to include all the billable RVUs; injection codes, in partic-ular, must not be neglected. The RVU method allows analysisof the distribution of workload among the various imagingtechniques, which can be used to study trends in utilization.Although this method was developed for comparing radiologypractices within multispecialty groups, it can be used toevaluate other radiology practice settings, and it can serve asa model for making workload comparisons among specialties.

ACKNOWLEDGMENTS

We express appreciation to Michael Lenker of Kelsey-SeyboldClinic for helpful editorial suggestions, Virginia Heckel of Kelsey-Seybold Foundation Crump Cancer Center for statistical assistance,

Mike Nelson of Park Nicollet Medical Center, Christopher Merritt ofOchsner Clinic, Reilly Kidd of Mason Clinic, Timothy Parker of Love-lace Medical Center for suggestions regarding the survey question-

naire and methods, and all respondents to the questionnaire.

REFERENCES

1 . Medicare programs: fee schedules for radiologists’ services. Federal Reg-

ister. March 2, 1989;54:8994-90232. American Medical Association. CPT: Physician’s Current Procedural Ter-

minology. Chicago: American Medical Association, 19903. Hsiao WC, Braun P, Becker ER, et al. A national study of resource-based

relative value scales for physician services: final report to the Health CareFinancing Administration (Publication 17-C-98795/1-03). Cambridge, MA:Harvard School of Public Health. September 1988.

4. Johnson JL, Abernathy DL. Diagnostic imaging procedure volume in the

lhiited States. Radiology 1983;146:851-853

Dow

nloa

ded

from

ww

w.a

jron

line.

org

by 2

02.6

2.17

.27

on 1

0/12

/15

from

IP

addr

ess

202.

62.1

7.27

. Cop

yrig

ht A

RR

S. F

or p

erso

nal u

se o

nly;

all

righ

ts r

eser

ved