Background and Purpose Stroke is a common disease with a yearly cost in the United States of approximately $30 billion. The increasing prevalence of managed care and cost-containment measures may affect the delivery of stroke care now and in the future. This study was performed to determine (1) hospital charges and test utilization for stroke patients and (2) the effectiveness of educational efforts in modifying test utilization and related hospital charges.
Methods Patients with a diagnosis of stroke who were discharged from either the neurology service or another service of the Department of Medicine (DOM) were identified. Data on test utilization and hospital charges were collected and analyzed. Following this analysis, educational sessions were held in an effort to reduce the use of specific diagnostic tests. The effectiveness of these methods was studied in a second group of stroke patients.
Results In the baseline period there were 303 stroke patients, of which 262 (86%) were discharged from the neurology service and 41 (14%) were discharged from other services of the DOM. Patients on the neurology service had a lower mean length of stay than patients on the other services of the DOM (9.2 days versus 10.5 days) and lower mean total charges per case ($13 149 versus $15 727), although the respective differences were not statistically significant. Patients on the neurology service were more likely to have both brain CT and MRI performed (82 of 262 patients, 31.3%) than patients on the other services of the DOM (4 of 41, 9.8%, P=.005). In addition, patients on the neurology service were more likely to undergo a transthoracic echocardiogram than patients on the other services of the DOM (71.8% versus 53.7%, P=.025). After educational sessions, the percentage of stroke patients on the neurology service having both CT and MRI fell from 31.3% to 17.7% (P=.005), and the number of stroke patients having a transthoracic echocardiogram fell from 71.8% to 60.3% (P=.025). However, the overall charges for stroke patients on the neurology service did not decrease.
Conclusions Education can be successful in reducing the utilization of and associated charges for specific diagnostic tests for some stroke patients. A multidisciplinary approach to case management, using tools such as care maps, may be necessary to realize significant cost savings in certain groups of stroke patients.
Stroke affects approximately 500 000 individuals in the United States each year and is the third leading cause of death in the United States.1 It is the most common admission diagnosis in patients with an acute neurological disorder, and it has a yearly cost of approximately $30 billion.2 Considering this profile, managed care initiatives could potentially have a significant impact on the costs and charges of caring for patients with stroke. This study was performed to explore two main issues: (1) hospital charges and other care parameters for stroke patients at a tertiary care hospital and (2) the impact of directed educational efforts on modifying specific hospital charges incurred by stroke patients.
Subjects and Methods
All patients with a DRG of 014 (specific cerebrovascular disorders except transient ischemic attack) who had been discharged from the neurology service and from other services of the DOM were identified. Mean and median charge and LOS data were reviewed for two time periods. Charge data, including data related to specific tests, did not include professional fees.
Two time epochs were used for comparison: a baseline period (July 1993 through May 1994) and a posteducational period (July 1994 through December 1994). For all patients, we analyzed the utilization of tests and procedures and determined that radiological procedures and echocardiographic procedures accounted for a significant amount of the test expenditures. Data were then collected on utilization and charges for head CTs, brain MRIs, TTEs, and TEEs on a per-patient basis.
Comparisons of median total charges and median LOS between cases managed by the neurology service and cases managed by the other services of the DOM were done using the Wilcoxon rank sum test. Comparisons between the two time periods on the use of various tests were made using the χ2 test. Trimmed means were calculated by omitting the lowest 5% and highest 5% of the charges.
In the baseline period, there were 303 stroke patients, of which 262 (86%) were discharged from the neurology service and 41 (14%) were discharged from other services of the DOM. Total charges per case are summarized in Table 1⇓. Many of the stroke patients on the neurology service were cared for in the stroke acute care unit during the initial phases of their hospitalization. Total charges per case were higher for patients managed by other services of the DOM than for patients managed by the neurology service, with a mean difference of $2578, a trimmed mean difference of $1958, and a median difference of $2746 (P=.17) (Table 1⇓). The mean LOS for patients managed by other services of the DOM was 10.5 days compared with 9.2 days for patients managed by the neurology service. The median LOSs were 6.0 and 7.0 for the respective groups, for a median difference of only 1 day (P=.22) (Table 1⇓).
Stroke patients on the neurology service had a mortality of 8.0% compared with a 29.3% mortality for stroke patients on other services of the DOM (P<.001). This did not appear to be explained by the mix of stroke types, since ischemic stroke and cerebral hemorrhages were evenly distributed between the two services (Table 2⇓). Among the patients who expired, there were no obvious differences between the two services in terms of concomitant diseases or stroke type. The patients expiring on the DOM services did have a shorter average LOS compared with those on the neurology service (Table 2⇓).
For the baseline group, 234 of the 303 patients (77.2%) had a head CT, 129 of 303 (42.6%) had a brain MRI, and 86 of 303 (28.4%) had both. Stroke patients on the neurology service were more likely to have both tests (82 of 262, 31.3%) than patients managed by other services of the DOM (4 of 41, 9.8%, P=.005; see Table 3⇓). The charges for the 234 CTs and the 129 MRIs totaled $232 845. The majority of patients (210 of 303, 69.3%) underwent a TTE, but only 8.3% (25 of 303) had a TEE. Stroke patients on the neurology service were more likely to have a TTE (71.8% versus 53.7%, P=.025) and tended to have a TEE done more frequently (9.2% versus 2.4%, P>.10) than stroke patients managed by other services of the DOM. The charges for the 210 TTEs and 25 TEEs totaled $71 575.
The data showed that nearly one third of stroke patients on the neurology service had dual imaging with head CT and MRI, which may be excessive. In particular, a pattern was identified whereby some acute stroke patients evaluated with a head CT late at night subsequently had an MRI the next morning. On the basis of these results, we initiated educational meetings (ie, lectures, stroke rounds, stroke conference, and grand rounds) with the faculty and house staff of the neurology service, medical services, and the emergency department, with a focus on the optimal utilization of the above tests.
We encouraged the house staff to forego the CT at night unless one of the following conditions was met: (1) an intracerebral or subarachnoid hemorrhage was suspected; (2) a hemorrhage had to be ruled out before beginning heparin therapy; (3) the patient was very ill or had a progressive clinical deficit; or (4) the diagnosis of stroke was in doubt. In addition, over 70% of stroke patients on the neurology service had a TTE done, which may have been an overutilization of this procedure.3 Similar educational efforts were begun to encourage the proper use of TTE in the workup of these patients.
In the posteducation period, 158 patients were analyzed (141 on the neurology service, 17 on the other services of the DOM). For the patients on the neurology service, the mean total charges per case increased from $13 149 to $14 564. However, the mean LOS fell from 9.2 to 9.0 days. The mortality rate for stroke patients on the neurology service was 8.5% in the follow-up study compared with 8.0% in the baseline study (see Table 1⇑). The small number of stroke patients on the other services of the DOM made comparisons less meaningful for this group.
The use of dual imaging (CTs and MRIs) for stroke patients managed by the neurology service fell from 31.3% to 17.7% (relative reduction of 43%, P=.005; see Table 3⇑). The utilization of TTE for patients managed by the neurology service also fell from 71.8% to 60.3% (relative reduction of 17%, P=.025; see Table 4⇓). The estimated savings for elimination of dual imaging of the brain for the posteducational period was $11 438 (a 43% decrease in charges for head CTs for neurology patients). This reduction was not accompanied by an increase in the utilization of MRI (see Table 3⇑). The savings would have been $13 547 if the same number of MRI studies had been eliminated. The reduction in utilization of TTE for neurology service patients resulted in a $10 532 savings (a 17% decrease in charges). These figures tend to underestimate the total savings associated with a decrease in the use of these tests, since professional fees were not included in the calculations.
A review of stroke cases in the follow-up period identified one instance where the changes in practice did affect patient care. The case was a male patient admitted after midnight with a history and physical examination consistent with a lacunar stroke producing a pure motor hemiparesis. Because the patient was clinically stable, it was decided that treatment with anticoagulants (ie, intravenous heparin) was not necessary at the time of admission. Because a brain MRI could not be obtained at the time of admission and it was highly unlikely that a head CT would demonstrate the lacunar infarction, it was decided to delay performing a brain imaging study until later that morning, when an MRI would be performed.
He remained clinically stable after admission. At approximately 8:00 am, a brain MRI was performed, and it showed a large subdural hematoma with significant mass effect. The MRI did not show a lacunar infarction. The patient underwent evacuation of the hematoma and returned to his premorbid condition after surgery.
This study was undertaken to examine some of the major hospital charges for stroke patients and to determine which charges were amenable to alterations. We found that (1) stroke patients on the neurology service had lower total charges and mortality than such patients on other services of the DOM, (2) the utilization of neuroimaging and echocardiography was significantly higher for patients managed by the neurology service, and (3) educational efforts were successful in altering patterns of usage and associated charges for these tests. Although the differences in total charges were not statistically significant, they were financially significant in that over $2500 was saved per case on the neurology service during the baseline period, and over $8000 per case in the posteducational period, when compared with the non-neurology services.
Direct comparisons between stroke patients on the neurology service with stroke patients on other services of the DOM should be made with caution, since this was not a randomized, prospective study. We cannot rule out a systematic bias in terms of admission triage, whereby selected patients were admitted to other services of the DOM instead of neurology. For example, some of the stroke patients on other services of the DOM might have been admitted with another initial diagnosis and then had an in-hospital stroke that further complicated and lengthened their hospitalization. Patients with more severe strokes or other complicating illnesses might have been admitted to the non-neurology services, further biasing our results. However, on examination of the baseline data by one of the authors (M.J.A.), the per-case frequency of secondary diagnoses in both groups was similar, with the exception that more patients with atrial fibrillation were managed by services of the DOM other than neurology. We recognize that more exact measures of stroke severity are available and should be used in future studies of the issues raised in this article. However, our overall findings are consistent with several other prospective, randomized studies, demonstrating that stroke patients receive more cost-effective care when they are cared for on a stroke unit or by a multidisciplinary stroke team.4 5 6 7
The relative number of diagnostic tests for patients managed by the neurology service was somewhat unexpected. It may reflect more complicated illnesses on this service, although our data do not address this issue. Physician response to the educational efforts was also somewhat surprising. Although the attending physicians were generally supportive, the most resistance came from the neurology house staff. Their resistance might reflect their relative dependence on high-technology diagnostic tools as part of their training. This is an important observation because it is the house staff that orders the majority of tests at most tertiary care institutions. Therefore, educational efforts at tertiary care hospitals may have to focus on house staff practices if they are to be successful. Consideration might also be given to reducing the authority of the house staff in ordering some high-cost tests and procedures. On the basis of these observations, some educational efforts may be more successful at nonacademic institutions without house staff.
Despite our educational efforts, the overall charges for neurology service cases increased a nonsignificant amount from the baseline to the follow-up period, although the LOSs were essentially unchanged. This finding might be explained by an increased severity of illness for patients admitted during the posteducational period, as well as an increase in hospital charges for certain items. These findings might also indicate that to achieve a global reduction in hospital charges, a coordinated multidisciplinary approach may be needed to streamline the continuum of management of stroke patients. Care maps (clinical pathways) for stroke patients are being developed and may be useful for guiding the use and timeliness of tests and procedures.8 Applying care maps in this patient population will be challenging because of the marked heterogeneity in stroke types, etiology, presentation, and treatments.3
The issue of quality of care versus cost is of obvious concern. In this pilot study, there was no evidence of a reduction in the quality of care, at least as reflected in overall mortality and LOS. These two gross measures may have missed changes in quality of care, however. Future studies should include more comprehensive measures of quality of care. One case (cited above) did result in a delay in the proper diagnosis and treatment of a patient with a subdural hematoma. Fortunately, no permanent neurological damage resulted from this delay. As systematic changes in patient care become more widespread, healthcare providers must ensure that quality of care is not adversely affected by cost-containment measures.9
This pilot study was limited because it analyzed patients at one academic institution for relatively short periods. In addition, the retrospective nature of the study and lack of randomization for patients on neurology versus other services of the DOM limit some of the conclusions. However, we did study more than 450 patients, which is more than has been reported in prior similar studies. In addition, by performing a prospective follow-up study with clearly delineated benchmarks, we were able to target specific areas and outcomes. Similar programs at other hospitals may show different patterns of test utilization and outcomes. In addition, our study focused on short-term aspects of patient care. Future studies should investigate both long- and short-term outcomes.
In that regard, it should be noted that a majority of our cases were Medicare patients (63.4%). Our mean LOS was 9.2 days for Medicare cases in DRG 014 from July 1993 through May 1994, which was similar to the mean LOS of 10.0 days for Medicare cases in DRG 014 at major teaching hospitals in 1993. Likewise, our mean total charges were $13 057 for Medicare cases in DRG 014 during the same time period and were similar to the mean total charges of $13 560 for Medicare cases in DRG 014 reported at teaching hospitals in 1993.10 These figures appear quite consistent with a recent article on costs associated with stroke at five academic medical centers.11 These findings suggest that our study population and results may be applicable to other comparable institutions.
The findings of this study offer some areas of focus for reducing test use in some acute stroke patients. The challenge for managing and standardizing stroke care is that stroke is a heterogeneous condition, which makes a uniform approach quite difficult. However, much of the heterogeneity in stroke care appears to be in the treatment aspects as opposed to diagnostic areas. This study focused on diagnostic tests, which may be a promising area for further improvements.
Selected Abbreviations and Acronyms
|DOM||=||Department of Medicine|
|DRG||=||diagnosis related group|
|LOS||=||length of stay|
- Received April 8, 1996.
- Revision received June 28, 1996.
- Accepted June 28, 1996.
- Copyright © 1996 by American Heart Association
American Heart Association. 1995 Heart and Stroke Facts and Statistics. Dallas, Tex: American Heart Association; 1994.
National Stroke Association. Cost of stroke. Stroke Clinical Updates. 1994:5:9-12.
Mohr J, Sacco R. Classification of stroke. In: Barnett H, Mohr J, Stein B, Yatsu F, eds. Stroke: Pathophysiology, Diagnosis, and Management. New York, NY: Churchill Livingstone; 1992:271-283.
Wentworth D, Atkinson R. Implementation of an acute stroke program decreases hospitalization costs and length of stay. Stroke. 1995;26:161. Abstract.
Jorgensen H, Nakayama H, Raaschou H, Larsen K, Hubbe P, Olsen T. The effect of a stroke unit: reductions in mortality, discharge rate to nursing home, length of hospital stay, and cost. Stroke. 1995;26:1178-1182.
Indredavik B, Bakke F, Solberg R, Rokseth R, Haaheim L, Holme I. Benefit of a stroke unit: a randomized controlled trial. Stroke. 1991;22:1026-1031.
Spath P. Clinical Paths: Tools for Outcomes Management. American Hospital Publishing Inc; 1994.
Medicare Summary Information for All US Hospitals. HCIA Inc and Ernst & Young LLP; 1995.