- Email: [email protected]
Radiologic consultation: Effect on inpatient diagnostic imaging evaluation in a teaching hospital
Departmental
Administration S t e v e n E. Seltzer, MD, Editor
Radiologic Consultation: Effect on Inpatient Diagnostic Imaging Evaluation in a Teaching Hospital R o n a l d H. Gottlieb, MD, G a r y M. Hollenberg, MD, Patrick J. Fultz, MD, D e b o r a h J. Rubens, MD
R a t i o n a l e a n d O b j e c t i v e s . The authors evaluated radiologic consultation affecting resident physician ordering of relatively expensive imaging studies. M e t h o d s . Requisitions (n -- 180) for expensive imaging studies from three general medicine floors (two consultation floors, one control floor) w e r e prospectively evaluated. Information on the requisitions was classified as appropriate, inappropriate, or undecided if insufficient information was provided. On the consultation floors, but not the control floor, the medical residents w e r e contacted about all studies initially considered inappropriate or undecided before they w e r e performed. Results. Nine of 119 requisitions (7.6%) from the consultation floors w e r e considered inappropriate. In three studies (2.5%) the imaging evaluation was modified as the result of the interaction with the radiologist. There was no significant difference in the total n u m b e r of radiologic studies or percentage of the total that w e r e expensive imaging studies w h e n comparing the consultation floors with the control floor.
C o n c l u s i o n . Radiologic consultation on expensive imaging studies through routine review of requisitions did not significantly change their use b y house staff. K e y W o r d s . Education; efficacy study; radiology and radiologists, departmental management.
n radiology, the overuse of relatively expensive imaging technology has b e e n cited as one of the leading c o m p o n e n t s of the increased cost of medical care [1]. In the early 1980s, Baker et al [2] reported a substantial reduction in the use of imaging studies (36% decrease in body c o m p u t e d t o m o g r a p h y [CT], 44% decrease in ultrasound [US], 57% decrease in contrast material-enhanced gastrointestinal [GI] studies, and 22% decrease in noncardiac nuclear medicine imaging) on inpatients through mandatory radiologic consultation with clinicians at a major university teaching hospital. Results of t w o other studies indicated that radiologic consultation results in significant alterations of the clinicians' imaging strategy [3,4]. These research efforts have generated support for radiologic consultation services. The p u r p o s e of this study was to evaluate w h e t h e r p r o m p t radiologic consultation after reviewing all requisitions for relatively expensive imaging studies would affect resident physician ordering of these examinations in a university teaching hospital. Resident physicians w e r e the focus of the interaction with the radiologist, since they are usually the first physicians to be contacted by the radiologist with concerns about the imaging strategy for a patient.
I
From the Department of Radiology, University of Rochester School of Medicine and Dentistry, Rochester, NY. Supported by funds provided by Preferred Care and Blue Cross and Blue Shield of the Rochester area. Address reprint requests to R. H. Gottlieb, MD, Department of Radiology, University of Rochester School of Medicine and Dentistry, 601 Elmwood Ave, Box 648, Rochester, NY 14642. Received May 21, 1996, and accepted for publication after revision October 8, 1996. Acad Radiol 1997;4:217-221
@1997, Association of University Radiologists
217
GOTTLIEB ET AL
MATERIALS AND METHODS Our study protocol was reviewed and a p p r o v e d by the H u m a n Subjects Review Board of our institution. The chairman of the Department of Medicine, the director of the Internal Medicine Residency Training Program, and all of the internal medicine residents w e r e informed about the nature of the project before its start and agreed to participate. We prospectively evaluated all requisitions for expensive imaging studies from three general medicine floors, on w h i c h patients had a similar mix of medical p r o b l e m s (no subspecialty bias), in a 722-bed university medical center over a 66-day period. The requisitions for after-hours and w e e k e n d studies w e r e reviewed within 12-48 hours during the w o r k w e e k (Monday through Friday, 8 AMtO 6 PM). Afterhours requests accounted for only 6% of the total. Expensive imaging studies included CT, magnetic resonance (MR) imaging, US, contrast-enhanced GI examinations, intravenous pyelography, and noncardiac nuclear medicine imaging. These studies constitute the majority of the relatively expensive noninvasive imaging examinations p e r f o r m e d in our department. Interventional procedures (eg, angiography, p e r c u t a n e o u s drainages, embolizations) w e r e excluded, as clinicians invariably consult with a radiologist before ordering these studies. Plain radiography was also excluded, as these studies are relatively inexpensive to p e r f o r m compared with the imaging studies w e evaluated. Requisitions for neuroradiology studies w e r e evaluated by one of two faculty neuroradiologists; the other requisitions w e r e evaluated by one of three faculty m e m b e r s in body imaging (R.H.G., P.J.F., D.J.R.). These imaging specialists evaluate the majority of the expensive imaging studies in our institution. The b o d y imaging faculty in our center review CT, MR, and US images of the chest, abdomen, and pelvis; for this study, however, they w e r e asked to evaluate the appropriateness of requisitions for musculoskeletal and noncardiac nuclear medicine studies. They sought advice, w h e n necessary, from specialists in these areas. Of the three patient care floors included in this study, two w e r e randomly designated consultation floors and one the control floor. Eight internal medicine residents (four senior, four junior) w e r e assigned to each floor and rotated b e t w e e n services every 3 weeks. They had no prior knowledge of w h e t h e r they w e r e covering a consultation or a control floor. Only one resident rotated on b o t h consultation and control floors
218
Vol.4, No.3, March 1997
during the study period. The study was p e r f o r m e d from March through May to negate resident experience level as a confounding factor. A variety of medical attending physicians admitted patients to each floor, but the residents w e r e responsible for ordering the vast majority of examinations after consulting with the respective attending physician. Approximately half of those admitted w e r e private patients (admitted to be cared for by a particular attending physician); the other haft w e r e admitted to the general medicine service on all three floors. In our hospital, there are no specifically scheduled radiology rounds during w h i c h consultation could potentially take place b e t w e e n radiologists and clinicians. During the study period, 52% of the patients w h o u n d e r w e n t expensive imaging studies w e r e m e n and 48% w e r e w o m e n (mean age, 53 years; range, 19-92 years; standard deviation, 16.7 years) on the consultation floors and 52% w e r e m e n and 48% w e r e w o m e n (mean age, 55 years; range, 21-89 years; standard deviation, 18.2 years) on the control floor. The appropriateness of the requested study was initially evaluated based on the clinical information provided on the requisitions. Supplemental information was obtained, w h e n necessary, through our radiology information system (IDXrad, Burlington, VT). The studies w e r e then classified as appropriate, inappropriate, or undecided if insufficient information was available to make this distinction. Each participating consultant radiologist made this evaluation based on previous experience and training. One b o d y imaging attending physician and one neuroradiologist w e r e available for the consultation service daily. On the consultation floors, but not the control floor, the medical residents w e r e contacted on all studies (before they w e r e performed) that w e r e initially considered inappropriate or undecided. A r e c o m m e n d a t i o n was made as to an alternative imaging strategy for studies still considered inappropriate after verbal interaction with the resident. A record was kept as to w h e t h e r this r e c o m m e n d a t i o n was followed and the length of the interaction. The total number of radiologic examinations, the n u m b e r of expensive imaging studies, the n u m b e r of patients, and the length of stay (Table 1) w e r e determined for each floor for the 66-day study, as well as for the 90 days before and 92 days after the study period. All requisitions (except for the neuroradiology cases) w e r e again independently reviewed several months after the study by the three b o d y imaging faculty to determine interobserver variation in classifying
Vol.4, No. 3, March 1997
R A D I O L O G I C CONSULTATION IN A T E A C H I N G HOSPITAL
TABLE 1 : Comparison of Consultation and Control Floors
act Test). On the consultation floors, two of the requisitions initially classified as inappropriate and two as undecided were ultimately classified as appropriate after the medical house staff was contacted and additional information was provided. Consequently, only nine of 119 studies, or 7.6% of expensive imaging studies ordered, were considered inappropriate for the specified clinical indication. Of these nine inappropriate studies, the radiologist reviewing the case would have canceled four and changed five to other types of imaging examinations. In only three cases (2.5% of the number of expensive imaging studies ordered on the consultation floors) did radiologic consultation result in a change in the imaging evaluation. One study was canceled and two abdominal CT examinations were ultimately performed instead of the initially ordered abdominal US. An average of 7 minutes was spent per consultation with the house staff (range, 3-10 minutes). During the study period, the distribution of expensive imaging studies on the consultation floors (38.7% CT, 31.1% US, 10.1% GI, 8.4% nuclear medicine, 10.1% MR imaging, and 1.7% intravenous pyelography) was similar to that on the control floor (40.0% CT, 28.9% US, 2.2% GI, 11.1% nuclear medicine, and 17.8% MR imaging) (no statistically significant difference, Fisher Exact Test). There was no significant difference in the total number of radiologic studies (normalized to patient number) or the percentage of tile total that were expensive imaging studies w h e n comparing the consultation floors with the control floor during the study period and over time (Table 1). The median length of stay was slightly shorter (5 days) on the consultation floors during the study period compared with the periods immediately preceding and immediately following the study (6 days each). However, this was not statistically significant, and the same trend was observed for the control floor (5 days for the study period and 6 days for the period immediately following). Interobserver agreement among the three body imaging faculty members on reevaluation of the requisitions (excluding the neuroradiology cases) was fair (n = 163, ~ = .37, P < .0001). There was at least one disagreement among the three body imaging attendings in 17.4% of the cases w h e n making the distinction between studies they considered appropriate and those they considered inappropriate or undecided. All available requisitions (n = 32) for expensive im-
Period Before study (90 d) Consultation floor A Consultation floor B Control floor During study (66 d) Consultation floor A Consultation floor B Control floor After study (92 d) Consultation floor A Consultation floor B Control floor
Total No. of Studies*
Relatively Expensive Imaging Studies (%)
122.7 115.2 95.5
24.2 23.3 26.3
102.2 107.0 107.0
25.4 27.6 27.8
116.9 108.5 120.4
30.8 27.2 30.4
*Normalized to 100 patients per floor.
the requisitions. Two categories of concordance were considered: (a) appropriate and (b) inappropriate or undecided. Studies considered inappropriate or undecided were classified into one group because both categories would have p r o m p t e d the consultant radiologist to contact the resident physician before performing the examination. The Fisher Exact Test (Instat II; GraphPad Software, San Diego, CA) was used to compare the percentage of studies considered inappropriate or undecided and the percentage distribution of expensive imaging studies between the consultation and control floors. The Zz test for independence (Instat ID was used to compare the percentage of radiologic examinations that were expensive imaging studies and the total n u m b e r of radiologic studies requested (normalized to patient number) between the consultation and control floors within the same time period and over time. The Kruskal-Wallis test was used to compare patient length of stay b e t w e e n floors for each time period and over time. The kappa statistic (True Epistat; Epistat Services, Richardson, TX) was used to evaluate interobserver variation b e t w e e n radiologists in classifying the requisitions.
RESULTS
Eleven studies (9.2%) were initially classified as inappropriate and two (1.7%) as undecided w h e n 119 requisitions were prospectively evaluated from the consultation floors compared with three (4.9%) inappropriate and two (3.3%) undecided out of 61 requisitions from the control floor (not statistically significant, Fisher Ex-
219
GOTTLIEB ET AL
aging studies (28% CT, 31% US, 6% GI, 19% nuclear medicine, 13% MR imaging, 3% intravenous pyelography) were reviewed from the consultation and control floors from the period immediately before the study. This was done to assess w h e t h e r the percentage of inappropriately ordered studies changed as a result of conducting the study. Only two of 32 expensive imaging studies (6.2%) were considered inappropriate for the specified clinical indication by consensus reading of two experienced body imagers (R.H.G., D.J.R.). This was not significantly different from what was found .during the study period on both the consultation and control floors.
DISCUSSION Radiologic consultation services are being explored as a mechanism to reduce misuse and overuse of expensive imaging studies and thus decrease the cost of medical care. This c o n c e p t is supported by previous work that has demonstrated either a significant reduction in the use of expensive imaging studies or a notable effect on the imaging strategy of clinicians as a result of radiologic consultation [2-4]. We found that radiologic consultation after routinely reviewing all requisitions did not substantially affect the n u m b e r of expensive imaging studies ordered. The medicine house staff, acting either independently or after conferring with their medicine attending physicians, were not responsive to radiologic consultation. In only three of nine cases considered inappropriate (2.5% of all studies ordered) did consultation result in a change in the imaging strategy for the patient (therefore, there was an inability to affect ordering practices without enforcement). Consultation did not result in a decrease in the total number of radiologic studies ordered (normalized to patient number) or the percentage of the total that were relatively expensive. In our study, radiologic consultation followed review of the requisition. This is quite different from the format for consultation b e t w e e n radiologist and clinician cited in previous work. In the study by Seltzer et al [3] the clinician had to contact the radiologist to order the study, and in the study by Baker et al [2] consultation occurred during rounds. One could argue that the chance to affect physician ordering practices is greater if the interaction between radiologist and clinician occurs early in the decision process and not late, once the requisition has been generated and the "gears have
22b
vol.4, No.3, March 1997
been set in motion." However, prospective review of requisitions by radiologists is more practical at our center, because participation in clinician rounds is inordinately time-consuming and because having radiologists required to approve studies is often viewed as confrontational by clinicians in our teaching hospital setting. We acknowledge, however, that at other centers radiologists may participate in clinician rounds and eventually managed care may give radiologists the authority to approve studies. The average time spent consulting with the house staff during our study was only 7 minutes per case. Another source of variation of our study from previous w o r k is that the information provided on the request forms may not have accurately reflected the true clinical situation in all cases. We could not control for misinformation that may have been provided on requisitions, w h i c h would have led to misclassification of some studies as appropriate. Misinformation on the requisitions w o u l d severely handicap the m e t h o d of consultation used in our study and reduce the potential effect of consultation. We did not validate the information on the requisition through chart review. In our center, however, it is our impression that pertinent clinical information is usually provided on the requisition. Supplemental information, as needed, was accessed through our radiology information system. Our radiology information system, however, provides a record of only previous radiologic studies and limited clinical information associated with these examinations. The frequency with which the radiology information system was used was not recorded, although it was available at all times to the participating consultant radiologists. Only 7.6% of the expensive imaging studies ordered on the consultation floors were considered inappropriate for the clinical indication given. Radiologists disagreed among themselves (17.4%) at a higher rate on reevaluation of the requisitions than they disagreed with what the clinicians ordered during the study period. This occurred even though all of the radiologists reviewing the requests were experienced in body imaging techniques and routinely interpreted body CT, MR imaging, and US studies. Use of standard criteria, such as the American College of Radiology appropriateness criteria [5], may result in improved agreement between radiologists. However, these criteria were not available at the time of our study. We could not control for the Hawthorne effect [6],
Vol. 4, No. 3, March 1997
as the house staff on both the consultation and control floors were aware they were being "observed" by the consultant radiologists. However, w e do not believe this had a substantial effect on their ordering practices because, on limited review of requisitions, the rate of inappropriately ordered expensive imaging studies during the study period did not differ significantly from the rate during the preceding 90 days. Our patient sample size was relatively small because we were able to c o n d u c t our study for a comparatively short period of time (66 days). We lacked the resources to continue our project beyond this point. It is possible that with a larger patient population we may have demonstrated a statistically significant difference in numbers of expensive imaging studies b e t w e e n the consultation and control groups. However, our results suggest that consultation, after requisitions are reviewed, does not significantly affect resident physician ordering practices in a teaching hospital setting, especially if the radiologist is not given the authority to alter the imaging strategy of the patient. Future prospective studies, with larger patient numbers, may demonstrate that radiology consultation has an effect on physician ordering practices if radiologists are permitted to change the request if the most appropriate imaging study has not been ordered. An electronic patient record, w h i c h provides accurate clinical information, may make active participation by radiologists in patient care more feasible. In summary, radiologic consultation with residents through routine review of requisitions in our teaching hospital inpatient setting had no significant effect on the n u m b e r of expensive imaging studies ordered. If a relatively small percentage of all expensive imaging
RADIOLOGIC
CONSULTATION
IN A T E A C H I N G
HOSPITAL
studies ordered at major university centers are considered inappropriate by radiologists (as we found), it may not be worthwhile for the radiologists in these centers to devote substantial amounts of time as part of a dedicated consultation service that screens all inpatient radiology requisitions in an effort to eliminate the occasionally ordered inappropriate study. This is certainly true if "gatekeeper" enforcement is not possible. Ongoing education of clinicians as to w h i c h (if any) radiologic studies are appropriate in specific clinical situations may currently be a more effective use of the radiologist's time. ACKNOWLEDGMENTS
The authors express their appreciation to Drs Robert Betts, Edgar Black, Joshua Chodosh, and Valerie Newman, and all of t h e residents in the Department of Internal Medicine for their valuable collaboration in making this project possible. REFERENCES 1. Hillman BJ. Outcomes research and cost-effectiveness analysis for diagnostic imaging. Radiology1994;193:307-310. 2. Baker SR, Rosenberg ZS, Ariel H. The operation of a ward-based radiology consultation service. Radiology 1984; 152:331-333. 3. Seltzer SE, Beard JO, Adams DF. Radiologist as consultant: direct contact between referring clinician and radiologist before CT examination. A JR 1985;144:661-665. 4. Khorasani R, Silverman SG, Meyer JE, Gibson M, Weissman BN, Seltzer SE. Design and implementation of a new radiology consultation service in a teaching hospital. AJR 1994;163:457-459. 5. American College of Radiology appropriateness criteria. Reston, VA: American College of Radiology, 1995. 6. Roethlisberger FJ. Management and the worker: an account of a research program conducted by the Western Electric Company, Hawthorne Works, Chicago. Cambridge, MA: Harvard University Press, 1947.
221
Administration S t e v e n E. Seltzer, MD, Editor
Radiologic Consultation: Effect on Inpatient Diagnostic Imaging Evaluation in a Teaching Hospital R o n a l d H. Gottlieb, MD, G a r y M. Hollenberg, MD, Patrick J. Fultz, MD, D e b o r a h J. Rubens, MD
R a t i o n a l e a n d O b j e c t i v e s . The authors evaluated radiologic consultation affecting resident physician ordering of relatively expensive imaging studies. M e t h o d s . Requisitions (n -- 180) for expensive imaging studies from three general medicine floors (two consultation floors, one control floor) w e r e prospectively evaluated. Information on the requisitions was classified as appropriate, inappropriate, or undecided if insufficient information was provided. On the consultation floors, but not the control floor, the medical residents w e r e contacted about all studies initially considered inappropriate or undecided before they w e r e performed. Results. Nine of 119 requisitions (7.6%) from the consultation floors w e r e considered inappropriate. In three studies (2.5%) the imaging evaluation was modified as the result of the interaction with the radiologist. There was no significant difference in the total n u m b e r of radiologic studies or percentage of the total that w e r e expensive imaging studies w h e n comparing the consultation floors with the control floor.
C o n c l u s i o n . Radiologic consultation on expensive imaging studies through routine review of requisitions did not significantly change their use b y house staff. K e y W o r d s . Education; efficacy study; radiology and radiologists, departmental management.
n radiology, the overuse of relatively expensive imaging technology has b e e n cited as one of the leading c o m p o n e n t s of the increased cost of medical care [1]. In the early 1980s, Baker et al [2] reported a substantial reduction in the use of imaging studies (36% decrease in body c o m p u t e d t o m o g r a p h y [CT], 44% decrease in ultrasound [US], 57% decrease in contrast material-enhanced gastrointestinal [GI] studies, and 22% decrease in noncardiac nuclear medicine imaging) on inpatients through mandatory radiologic consultation with clinicians at a major university teaching hospital. Results of t w o other studies indicated that radiologic consultation results in significant alterations of the clinicians' imaging strategy [3,4]. These research efforts have generated support for radiologic consultation services. The p u r p o s e of this study was to evaluate w h e t h e r p r o m p t radiologic consultation after reviewing all requisitions for relatively expensive imaging studies would affect resident physician ordering of these examinations in a university teaching hospital. Resident physicians w e r e the focus of the interaction with the radiologist, since they are usually the first physicians to be contacted by the radiologist with concerns about the imaging strategy for a patient.
I
From the Department of Radiology, University of Rochester School of Medicine and Dentistry, Rochester, NY. Supported by funds provided by Preferred Care and Blue Cross and Blue Shield of the Rochester area. Address reprint requests to R. H. Gottlieb, MD, Department of Radiology, University of Rochester School of Medicine and Dentistry, 601 Elmwood Ave, Box 648, Rochester, NY 14642. Received May 21, 1996, and accepted for publication after revision October 8, 1996. Acad Radiol 1997;4:217-221
@1997, Association of University Radiologists
217
GOTTLIEB ET AL
MATERIALS AND METHODS Our study protocol was reviewed and a p p r o v e d by the H u m a n Subjects Review Board of our institution. The chairman of the Department of Medicine, the director of the Internal Medicine Residency Training Program, and all of the internal medicine residents w e r e informed about the nature of the project before its start and agreed to participate. We prospectively evaluated all requisitions for expensive imaging studies from three general medicine floors, on w h i c h patients had a similar mix of medical p r o b l e m s (no subspecialty bias), in a 722-bed university medical center over a 66-day period. The requisitions for after-hours and w e e k e n d studies w e r e reviewed within 12-48 hours during the w o r k w e e k (Monday through Friday, 8 AMtO 6 PM). Afterhours requests accounted for only 6% of the total. Expensive imaging studies included CT, magnetic resonance (MR) imaging, US, contrast-enhanced GI examinations, intravenous pyelography, and noncardiac nuclear medicine imaging. These studies constitute the majority of the relatively expensive noninvasive imaging examinations p e r f o r m e d in our department. Interventional procedures (eg, angiography, p e r c u t a n e o u s drainages, embolizations) w e r e excluded, as clinicians invariably consult with a radiologist before ordering these studies. Plain radiography was also excluded, as these studies are relatively inexpensive to p e r f o r m compared with the imaging studies w e evaluated. Requisitions for neuroradiology studies w e r e evaluated by one of two faculty neuroradiologists; the other requisitions w e r e evaluated by one of three faculty m e m b e r s in body imaging (R.H.G., P.J.F., D.J.R.). These imaging specialists evaluate the majority of the expensive imaging studies in our institution. The b o d y imaging faculty in our center review CT, MR, and US images of the chest, abdomen, and pelvis; for this study, however, they w e r e asked to evaluate the appropriateness of requisitions for musculoskeletal and noncardiac nuclear medicine studies. They sought advice, w h e n necessary, from specialists in these areas. Of the three patient care floors included in this study, two w e r e randomly designated consultation floors and one the control floor. Eight internal medicine residents (four senior, four junior) w e r e assigned to each floor and rotated b e t w e e n services every 3 weeks. They had no prior knowledge of w h e t h e r they w e r e covering a consultation or a control floor. Only one resident rotated on b o t h consultation and control floors
218
Vol.4, No.3, March 1997
during the study period. The study was p e r f o r m e d from March through May to negate resident experience level as a confounding factor. A variety of medical attending physicians admitted patients to each floor, but the residents w e r e responsible for ordering the vast majority of examinations after consulting with the respective attending physician. Approximately half of those admitted w e r e private patients (admitted to be cared for by a particular attending physician); the other haft w e r e admitted to the general medicine service on all three floors. In our hospital, there are no specifically scheduled radiology rounds during w h i c h consultation could potentially take place b e t w e e n radiologists and clinicians. During the study period, 52% of the patients w h o u n d e r w e n t expensive imaging studies w e r e m e n and 48% w e r e w o m e n (mean age, 53 years; range, 19-92 years; standard deviation, 16.7 years) on the consultation floors and 52% w e r e m e n and 48% w e r e w o m e n (mean age, 55 years; range, 21-89 years; standard deviation, 18.2 years) on the control floor. The appropriateness of the requested study was initially evaluated based on the clinical information provided on the requisitions. Supplemental information was obtained, w h e n necessary, through our radiology information system (IDXrad, Burlington, VT). The studies w e r e then classified as appropriate, inappropriate, or undecided if insufficient information was available to make this distinction. Each participating consultant radiologist made this evaluation based on previous experience and training. One b o d y imaging attending physician and one neuroradiologist w e r e available for the consultation service daily. On the consultation floors, but not the control floor, the medical residents w e r e contacted on all studies (before they w e r e performed) that w e r e initially considered inappropriate or undecided. A r e c o m m e n d a t i o n was made as to an alternative imaging strategy for studies still considered inappropriate after verbal interaction with the resident. A record was kept as to w h e t h e r this r e c o m m e n d a t i o n was followed and the length of the interaction. The total number of radiologic examinations, the n u m b e r of expensive imaging studies, the n u m b e r of patients, and the length of stay (Table 1) w e r e determined for each floor for the 66-day study, as well as for the 90 days before and 92 days after the study period. All requisitions (except for the neuroradiology cases) w e r e again independently reviewed several months after the study by the three b o d y imaging faculty to determine interobserver variation in classifying
Vol.4, No. 3, March 1997
R A D I O L O G I C CONSULTATION IN A T E A C H I N G HOSPITAL
TABLE 1 : Comparison of Consultation and Control Floors
act Test). On the consultation floors, two of the requisitions initially classified as inappropriate and two as undecided were ultimately classified as appropriate after the medical house staff was contacted and additional information was provided. Consequently, only nine of 119 studies, or 7.6% of expensive imaging studies ordered, were considered inappropriate for the specified clinical indication. Of these nine inappropriate studies, the radiologist reviewing the case would have canceled four and changed five to other types of imaging examinations. In only three cases (2.5% of the number of expensive imaging studies ordered on the consultation floors) did radiologic consultation result in a change in the imaging evaluation. One study was canceled and two abdominal CT examinations were ultimately performed instead of the initially ordered abdominal US. An average of 7 minutes was spent per consultation with the house staff (range, 3-10 minutes). During the study period, the distribution of expensive imaging studies on the consultation floors (38.7% CT, 31.1% US, 10.1% GI, 8.4% nuclear medicine, 10.1% MR imaging, and 1.7% intravenous pyelography) was similar to that on the control floor (40.0% CT, 28.9% US, 2.2% GI, 11.1% nuclear medicine, and 17.8% MR imaging) (no statistically significant difference, Fisher Exact Test). There was no significant difference in the total number of radiologic studies (normalized to patient number) or the percentage of tile total that were expensive imaging studies w h e n comparing the consultation floors with the control floor during the study period and over time (Table 1). The median length of stay was slightly shorter (5 days) on the consultation floors during the study period compared with the periods immediately preceding and immediately following the study (6 days each). However, this was not statistically significant, and the same trend was observed for the control floor (5 days for the study period and 6 days for the period immediately following). Interobserver agreement among the three body imaging faculty members on reevaluation of the requisitions (excluding the neuroradiology cases) was fair (n = 163, ~ = .37, P < .0001). There was at least one disagreement among the three body imaging attendings in 17.4% of the cases w h e n making the distinction between studies they considered appropriate and those they considered inappropriate or undecided. All available requisitions (n = 32) for expensive im-
Period Before study (90 d) Consultation floor A Consultation floor B Control floor During study (66 d) Consultation floor A Consultation floor B Control floor After study (92 d) Consultation floor A Consultation floor B Control floor
Total No. of Studies*
Relatively Expensive Imaging Studies (%)
122.7 115.2 95.5
24.2 23.3 26.3
102.2 107.0 107.0
25.4 27.6 27.8
116.9 108.5 120.4
30.8 27.2 30.4
*Normalized to 100 patients per floor.
the requisitions. Two categories of concordance were considered: (a) appropriate and (b) inappropriate or undecided. Studies considered inappropriate or undecided were classified into one group because both categories would have p r o m p t e d the consultant radiologist to contact the resident physician before performing the examination. The Fisher Exact Test (Instat II; GraphPad Software, San Diego, CA) was used to compare the percentage of studies considered inappropriate or undecided and the percentage distribution of expensive imaging studies between the consultation and control floors. The Zz test for independence (Instat ID was used to compare the percentage of radiologic examinations that were expensive imaging studies and the total n u m b e r of radiologic studies requested (normalized to patient number) between the consultation and control floors within the same time period and over time. The Kruskal-Wallis test was used to compare patient length of stay b e t w e e n floors for each time period and over time. The kappa statistic (True Epistat; Epistat Services, Richardson, TX) was used to evaluate interobserver variation b e t w e e n radiologists in classifying the requisitions.
RESULTS
Eleven studies (9.2%) were initially classified as inappropriate and two (1.7%) as undecided w h e n 119 requisitions were prospectively evaluated from the consultation floors compared with three (4.9%) inappropriate and two (3.3%) undecided out of 61 requisitions from the control floor (not statistically significant, Fisher Ex-
219
GOTTLIEB ET AL
aging studies (28% CT, 31% US, 6% GI, 19% nuclear medicine, 13% MR imaging, 3% intravenous pyelography) were reviewed from the consultation and control floors from the period immediately before the study. This was done to assess w h e t h e r the percentage of inappropriately ordered studies changed as a result of conducting the study. Only two of 32 expensive imaging studies (6.2%) were considered inappropriate for the specified clinical indication by consensus reading of two experienced body imagers (R.H.G., D.J.R.). This was not significantly different from what was found .during the study period on both the consultation and control floors.
DISCUSSION Radiologic consultation services are being explored as a mechanism to reduce misuse and overuse of expensive imaging studies and thus decrease the cost of medical care. This c o n c e p t is supported by previous work that has demonstrated either a significant reduction in the use of expensive imaging studies or a notable effect on the imaging strategy of clinicians as a result of radiologic consultation [2-4]. We found that radiologic consultation after routinely reviewing all requisitions did not substantially affect the n u m b e r of expensive imaging studies ordered. The medicine house staff, acting either independently or after conferring with their medicine attending physicians, were not responsive to radiologic consultation. In only three of nine cases considered inappropriate (2.5% of all studies ordered) did consultation result in a change in the imaging strategy for the patient (therefore, there was an inability to affect ordering practices without enforcement). Consultation did not result in a decrease in the total number of radiologic studies ordered (normalized to patient number) or the percentage of the total that were relatively expensive. In our study, radiologic consultation followed review of the requisition. This is quite different from the format for consultation b e t w e e n radiologist and clinician cited in previous work. In the study by Seltzer et al [3] the clinician had to contact the radiologist to order the study, and in the study by Baker et al [2] consultation occurred during rounds. One could argue that the chance to affect physician ordering practices is greater if the interaction between radiologist and clinician occurs early in the decision process and not late, once the requisition has been generated and the "gears have
22b
vol.4, No.3, March 1997
been set in motion." However, prospective review of requisitions by radiologists is more practical at our center, because participation in clinician rounds is inordinately time-consuming and because having radiologists required to approve studies is often viewed as confrontational by clinicians in our teaching hospital setting. We acknowledge, however, that at other centers radiologists may participate in clinician rounds and eventually managed care may give radiologists the authority to approve studies. The average time spent consulting with the house staff during our study was only 7 minutes per case. Another source of variation of our study from previous w o r k is that the information provided on the request forms may not have accurately reflected the true clinical situation in all cases. We could not control for misinformation that may have been provided on requisitions, w h i c h would have led to misclassification of some studies as appropriate. Misinformation on the requisitions w o u l d severely handicap the m e t h o d of consultation used in our study and reduce the potential effect of consultation. We did not validate the information on the requisition through chart review. In our center, however, it is our impression that pertinent clinical information is usually provided on the requisition. Supplemental information, as needed, was accessed through our radiology information system. Our radiology information system, however, provides a record of only previous radiologic studies and limited clinical information associated with these examinations. The frequency with which the radiology information system was used was not recorded, although it was available at all times to the participating consultant radiologists. Only 7.6% of the expensive imaging studies ordered on the consultation floors were considered inappropriate for the clinical indication given. Radiologists disagreed among themselves (17.4%) at a higher rate on reevaluation of the requisitions than they disagreed with what the clinicians ordered during the study period. This occurred even though all of the radiologists reviewing the requests were experienced in body imaging techniques and routinely interpreted body CT, MR imaging, and US studies. Use of standard criteria, such as the American College of Radiology appropriateness criteria [5], may result in improved agreement between radiologists. However, these criteria were not available at the time of our study. We could not control for the Hawthorne effect [6],
Vol. 4, No. 3, March 1997
as the house staff on both the consultation and control floors were aware they were being "observed" by the consultant radiologists. However, w e do not believe this had a substantial effect on their ordering practices because, on limited review of requisitions, the rate of inappropriately ordered expensive imaging studies during the study period did not differ significantly from the rate during the preceding 90 days. Our patient sample size was relatively small because we were able to c o n d u c t our study for a comparatively short period of time (66 days). We lacked the resources to continue our project beyond this point. It is possible that with a larger patient population we may have demonstrated a statistically significant difference in numbers of expensive imaging studies b e t w e e n the consultation and control groups. However, our results suggest that consultation, after requisitions are reviewed, does not significantly affect resident physician ordering practices in a teaching hospital setting, especially if the radiologist is not given the authority to alter the imaging strategy of the patient. Future prospective studies, with larger patient numbers, may demonstrate that radiology consultation has an effect on physician ordering practices if radiologists are permitted to change the request if the most appropriate imaging study has not been ordered. An electronic patient record, w h i c h provides accurate clinical information, may make active participation by radiologists in patient care more feasible. In summary, radiologic consultation with residents through routine review of requisitions in our teaching hospital inpatient setting had no significant effect on the n u m b e r of expensive imaging studies ordered. If a relatively small percentage of all expensive imaging
RADIOLOGIC
CONSULTATION
IN A T E A C H I N G
HOSPITAL
studies ordered at major university centers are considered inappropriate by radiologists (as we found), it may not be worthwhile for the radiologists in these centers to devote substantial amounts of time as part of a dedicated consultation service that screens all inpatient radiology requisitions in an effort to eliminate the occasionally ordered inappropriate study. This is certainly true if "gatekeeper" enforcement is not possible. Ongoing education of clinicians as to w h i c h (if any) radiologic studies are appropriate in specific clinical situations may currently be a more effective use of the radiologist's time. ACKNOWLEDGMENTS
The authors express their appreciation to Drs Robert Betts, Edgar Black, Joshua Chodosh, and Valerie Newman, and all of t h e residents in the Department of Internal Medicine for their valuable collaboration in making this project possible. REFERENCES 1. Hillman BJ. Outcomes research and cost-effectiveness analysis for diagnostic imaging. Radiology1994;193:307-310. 2. Baker SR, Rosenberg ZS, Ariel H. The operation of a ward-based radiology consultation service. Radiology 1984; 152:331-333. 3. Seltzer SE, Beard JO, Adams DF. Radiologist as consultant: direct contact between referring clinician and radiologist before CT examination. A JR 1985;144:661-665. 4. Khorasani R, Silverman SG, Meyer JE, Gibson M, Weissman BN, Seltzer SE. Design and implementation of a new radiology consultation service in a teaching hospital. AJR 1994;163:457-459. 5. American College of Radiology appropriateness criteria. Reston, VA: American College of Radiology, 1995. 6. Roethlisberger FJ. Management and the worker: an account of a research program conducted by the Western Electric Company, Hawthorne Works, Chicago. Cambridge, MA: Harvard University Press, 1947.
221