HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to Personal Folders Download to citation manager Author home page(s): Christopher J. Riordan Milo Engoren Anoar Zacharias Thomas A. Schwann Gary L. Parenteau Samuel J. Durham Permission Requests Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Riordan, C. J. Articles by Habib, R. H. Search for Related Content PubMed PubMed Citation Articles by Riordan, C. J. Articles by Habib, R. H.

Ann Thorac Surg 2000;69:1092-1097
© 2000 The Society of Thoracic Surgeons

---

ORIGINAL ARTICLES: CARDIOVASCULAR

Resource utilization in coronary artery bypass operation: does surgical risk predict cost?

^a Department of Cardiovascular Surgery, St. Vincent Mercy Medical Center, Toledo, Ohio, USA
^b Department of and Anesthesia, St. Vincent Mercy Medical Center, Toledo, Ohio, USA
^c Medical College of Ohio, Toledo, Ohio, USA

Address reprint requests to Dr Habib, Cardiopulmonary Research, St. Vincent Mercy Medical Center, 2213 Cherry St, ACC Bldg, Suite 309, Toledo, OH, 43608
e-mail: robert_habib{at}mhsnr.org

	Abstract

Top Abstract Introduction Material and methods Results Comment References

 
Background. Current healthcare trends may render financial risk of cardiac operation a key component of clinical decision making. It has been suggested, based on large cohorts of patients stratified by clinical risk, that the cost of operation can be predicted from models of clinical risk since length of stay (LOS) is highly correlated to clinical risk, and LOS is correlated to hospital costs and charges. Direct correlation of actual surgical costs with surgical risk are lacking.

Methods. Variable direct costs, LOS, and The Society of Thoracic Surgeons predicted mortality risk [STS risk (%)] were collected and analyzed in 628 consecutive patients undergoing coronary artery bypass grafting (CABG) at our institution in 1997.

Results. Cost of CABG had a near-normal distribution, and cost in 21 outlier patients (cost > two standard deviations above the mean) was an average 5.3 times normal (median cost). For individual patients, cost was well correlated to LOS (R² = 0.48) but not with STS risk (R² = 0.12). LOS was also poorly predicted by STS risk (R² = 0.09). However, despite its poor prediction of cost, STS risk was an unbiased estimator over the entire population. A result manifested, when patients were grouped into similar risk (< 1%, 1–2%, 2⁺–3%, 3⁺–5%, 5⁺–10%, and >10%) cohorts, by high correlation between cost and STS risk (R² = 0.99), cost and LOS risk (R² = 0.99), and LOS and STS risk (R² = 0.97).

Conclusions. Our data demonstrated that, in large CABG cohorts, surgical risk models can accurately predict cost of CABG. However, despite a trend for increasing cost with increasing STS risk, surgical risk models based on preoperative data are poor predictors of cost in individual patients. Use of these models should be limited to analysis of cost trends in cardiac operation, but not for predicting financial risk in individual patients during clinical decision making.

	Introduction

Top Abstract Introduction Material and methods Results Comment References

 
Dramatic changes in the cost of medical care have put substantial pressures on healthcare providers to focus on efficient resource utilization. For especially high-cost, high-volume procedures such as cardiac operation, it may soon be necessary for cost-of-care analyses to become part of the clinical decision making process. Thus, the ability to estimate the financial cost of providing cardiac operation to an individual patient in an accurate manner may become crucial in efficient allocation of increasingly limited resources.

Presently, there are no algorithms specifically designed to predict costs for isolated patients undergoing cardiac operation. As a result, a number of authors have attempted to use some common models that predict clinical outcomes, such as morbidity and mortality, to predict financial risk. Williams and colleagues [1] recently showed that increases in the marginal cost of cardiac operation can be predicted for cohorts of patients, based on increases in their marginal surgical risk, as defined by their Parsonnet score [2]. Their derivation of a simple linear model, relating marginal risk to marginal cost, was based on the established strong correlation in patient cohorts between clinical risk and length of hospital stay (LOS) [1, 3–7] and LOS and cost [1, 8]. Whether a similar linear relationship, between cost and the more widely used The Society of Thoracic Surgeons surgical risk (STS risk) algorithm, exists has not been shown. More importantly, while these models appear to predict costs for patients grouped into fairly large cohorts, the ability of these models to predict financial risk in individual patients has not been tested.

While longer LOS can lead to greater resource utilization, costs can still vary widely among patients with similar LOS, depending on the requirements of care in each individual. Patients with similar predicted clinical risk may have significantly different LOS and postoperative course. For this reason, we hypothesized that the cost of coronary artery bypass grafting (CABG) can vary widely in individuals with similar predicted clinical risk, and so models derived solely from clinical outcomes, rather than actual cost data, may be inadequate predictors of financial risk for isolated patients. If this is the case, using clinical outcome based models during the course of actual day-to-day clinical decision making may lead to unexpected and potentially harmful financial and clinical decisions.

To examine this hypothesis, we evaluated the ability of the STS risk (%) to predict cost of CABG in individual patients, and contrasted these results to the corresponding prediction of CABG cost by STS risk in large patient cohorts grouped on a similar surgical risk basis.

	Material and methods

Top Abstract Introduction Material and methods Results Comment References

 
Detailed direct variable costs were prospectively collected in 628 consecutive patients undergoing isolated myocardial revascularization at St. Vincent Mercy Medical Center (SVMMC) during 1997. Isolated valve or combined CABG and valve operation patients were excluded. SVMMC is a university-affiliated, tertiary care hospital that acts as both a primary care and referral center for cardiac operation. Patients were under the primary care of their cardiovascular surgeon from the time of operation until discharge. All surgeons used standardized clinical pathways for intraoperative and postoperative care, thus minimizing intersurgeon variability. Cardiopulmonary bypass was performed using standard techniques, with normothermia used in most patients (95%). This study was performed with the approval of the Institutional Human Investigation Committee.

The variable direct costs were used as the measure of actual resource utilization (cost). These encompassed every care-related cost during the entire hospital admission.

Direct variable costs were obtained from the hospital’s internal accounting system. Each departmental manager, with aid from the finance department, calculated the labor and material costs of each item and service. For disposable items, the direct variable cost was close to the purchase price. Direct variable room costs were based on nursing labor costs (staffed at nurse to patient ratio of 1 to 1 in the cardiovascular intensive care unit, and 1 to 2 during days and 1 to 3 during nights in the cardiac stepdown ratio) and miscellaneous supplies, not elsewhere charged. Services, such as laboratory and radiographic tests had a direct variable cost based on the disposable supply used, such as test tubes or roentgenogram film, the labor involved in performing the service, and wear and tear of the equipment.

Univariate regression analysis (SigmaStat; Jandel Scientific, San Raphael, CA) was used to test the correlation between STS risk (%) and cost, STS risk (%) and LOS, as well as LOS (days) and cost. Cost was normalized by the median CABG cost for the entire group of patients, so that median cost equalled 1. Patients (n = 21) with cost greater than two standard deviations above the mean cost were considered as outliers. Analysis was done using both individual patient data and patients grouped into similar risk cohorts (Table 1). The bias and limits of agreement for predicting cost by LOS and STS risk were determined using the method of Bland and Altman [10].

	Results

Top Abstract Introduction Material and methods Results Comment References

View larger version (20K): [in this window] [in a new window]   Fig 1. Histogram of the frequency distribution of cost of coronary artery bypass grafting referenced to the median cost. Bell shaped curve indicates a near normal distribution of cost. Note, outlier patients (21 of 627) all with cost greater than 3 times median cost are lumped together.

View larger version (28K): [in this window] [in a new window]   Fig 2. (A) The relationship between cost and length of hospital stay (LOS) from 607 coronary artery bypass graft patients (excluding 21 outliers), indicating a linear correlation corresponding to R2 = 0.48. (B) Bland-Altman plot of the cost prediction error [error (%) = 100* (actual cost - estimated cost)/estimated cost] for each patient based on cost-LOS model depicted by regression line in panel A. These results indicate that LOS is an unbiased estimator of coronary artery bypass grafting cost (bias = 0.4%), but it can be quite inaccurate in individual patients as depicted by wide limits (~50%) of agreement defined by bias ± 2S of the error.

View larger version (28K): [in this window] [in a new window]   Fig 3. (A) The relationship between cost and The Society of Thoracic Surgeons Risk [STS risk (%)] from 607 coronary artery bypass graft patients (excluding 21 outliers) along with the linear regression results indicating a poor correlation (R2 = 0.12). (B) Bland-Altman plot of the cost prediction error [error (%) = 100* (actual cost - estimated cost)/estimated cost] for each patient based on cost-STS risk model depicted by regression line in panel A. These results indicate that STS risk is an unbiased estimator of coronary artery bypass grafting cost (bias = 0.4%), but it can be quite inaccurate in individual patients as depicted by wide limits (~63%) of agreement defined by bias ± 2S of the error.

View larger version (28K): [in this window] [in a new window]   Fig 4. (A) The relationship between length of stay (LOS) and The Society of Thoracic Surgeons Risk [STS risk (%)] from 607 coronary artery bypass graft patients (excluding 21 outliers) along with the linear regression results indicating a poor correlation (R2 = 0.09). (B) Bland-Altman plot of the LOS prediction error [error (%) = 100* (actual cost - estimated cost)/estimated cost] for each patient based on LOS-STS risk model depicted by regression line in panel A. These results indicate that STS risk is an unbiased estimator of LOS following coronary artery bypass grafting (bias = 0.4%), but it can be highly inaccurate in individual patients as depicted by the wide limits (~95%) of agreement defined by bias ± 2S of the error.

Patient cohorts defined by STS risk
When patients were arranged into cohorts of similar predicted STS risk (Table 1), correlation between mean CABG cost and mean total LOS was dramatically improved (R² = 0.99, Fig 5A). The mean cost (R² = 0.99; Fig 5B) and mean LOS (R² = 0.97; Fig 5C) were also highly correlated to the mean STS risk (%).

View larger version (20K): [in this window] [in a new window]   Fig 5. Prediction of coronary artery bypass graft cost from length of stay (LOS) (A) and STS risk (B), as well as LOS from STS risk (C) when patients were grouped into 6 patient cohorts, based on comparable risk as described in Table 1. Results indicate that cost-LOS, cost-STS risk and LOS-STS risk relationships, in sufficiently large patient cohorts, are all highly linear. These linear relationships provide simple models for predicting the impact of changes in clinical risk for the general coronary artery bypass graft population on expected costs.

	Comment

Top Abstract Introduction Material and methods Results Comment References

One way of allocating limited financial resources effectively is by attempting to control and predict financial risk associated with cardiac operation. While not currently a standard part of clinical practice, further resource constraints may soon make financial risk assessment an important element in patient selection. This potential has already led a number of groups to explore means of estimating the cost associated with CABG before operation.

Existing models of risk in cardiac operation have focused on predicting morbidity and mortality. These algorithms, derived from data on clinical outcomes, have been repeatedly tested and validated [2, 9]. Their accuracy in this setting, along with the familiarity most surgeons have with these models, made it logical to examine whether they would be useful in predicting costs. Williams and colleagues [1] have demonstrated that when patients are grouped into risk-based cohorts, there is a good correlation between predicted surgical risk, LOS, and hospital costs. Indeed, using actual cost data, our study confirms their conclusion that the marginal cost of cardiac operation can be predicted from the corresponding marginal surgical risk. For example, the derived cost-STS risk model [Fig 5B: cost = 0.98 + 0.044 x STS risk (%)] predicts a 4.4% increase in cost for every 1% increase in STS risk.

We hypothesized that while accurate for similar risk cohorts, models based on clinical risk might not be reliable predictors of cost in individual patients. This hypothesis was based on the observations that patients with similar LOS may have very different costs and patients with the highest LOS often are not those with the highest predicted risk.

We specifically examined the ability of the STS surgical risk algorithm to predict cost in individuals, as well as in patient cohorts grouped based on their surgical risk. Our results indicated that the calculated STS risk did not predict CABG costs effectively for single patients (Fig 3, Table 3), but was strongly linearly correlated for the risk-based cohorts (Fig 5B).

Our analysis of CABG cost data produced several interesting findings. A small fraction of outlier patients (3.3%) accounted for a disproportionate amount of the total surgical costs (~16%), and the STS risk model appeared particularly limited in its ability to prospectively identify such patients. In addition, individuals with similar LOS showed significant variability in costs, suggesting that costs may be driven more by the type of care needed during hospitalization than by overall duration of care. Also, the STS surgical risk algorithm appears to exclude a number of important preoperative and intraoperative variables that play a significant role in determining costs.

Despite the limitations seen in individual patients, there was excellent correlation between STS risk and both LOS and actual cost of CABG when patients were grouped into similar risk cohorts. The STS risk algorithm may thus be very useful in predicting the financial effects of a change in the risk make-up of a large population of patients. This has been demonstrated by other groups as well as ours, and suggests a continued role for clinical-outcomes based models in strategic and population based planning. In contrast, the inaccuracies seen in predicting CABG cost in our single patients from the STS risk model illustrates the potential pitfalls of using such a model as an aid in clinical decision making at the individual patient level. Indeed, our data suggests that it is almost as likely that operation in any single high-risk patient will have low costs as it is that operation in a low-risk patient will have very high costs.

To be clinically useful, a model of financial risk must predict costs on an individual, not group, basis. The inability of the STS risk model to predict individual costs is not surprising, since the algorithm was not designed to do so. Indeed, this model is quite reliable in predicting mortality in large groups of patients undergoing CABG [9], as it has been derived from such clinical outcomes. We speculate that a financial risk model derived from actual clinical costs would be more accurate in predicting costs in individual patients. A preliminary examination of our CABG cost data in terms of patient variables indicates that, to be accurate, future financial risk models should incorporate both a wider array of preoperative variables other than the STS risk algorithm, and variables that could predict key intraoperative parameters such as cardiopulmonary bypass and cross-clamp times. Until such models are available and validated, the findings reported in this study indicate that the success of clinical risk models in predicting cost trends in large patient cohorts may be useful for strategic and financial planning, but should not be extrapolated to individuals during clinical decision making.

	References

Top Abstract Introduction Material and methods Results Comment References

 

1. Williams T.E., Jr, Fanning W.J., Benton W.C., Jr, et al. What is the marginal cost for marginal risk in cardiac surgery?. Ann Thorac Surg 1998:1969-1971.
2. Parsonnet V., Dean D., Bernstein A.D. A method of uniform stratification of risk for evaluating the results of surgery in acquired adult heart disease. Circulation 1989;79(Suppl 1):13-112.
3. Williams T.E., Jr, Benton W.C., Jr, Fanning W.J., Hankins T.D., Kakos G.S. Quantitative quality descriptors for an open heart program. Qual Progr 1992:29-32.
4. Kirklin J.W., Akins C.W., Blackstone E.H., et al. ACC/AHA task force report. J Am Coll Cardiol 1991;17:543-589.[Medline]
5. Dickinson T.A. How to do it. J Extracorp Technol 1993;24:135-140.
6. Lahey S.J., Borlase B.C., Lavin P.T., Levitsky S. Preoperative risk factors that predict hospital length of stay in coronary artery bypass patients > 60 years old. Circulation 1992;86(5 Suppl):181-185.
7. Williams T.E., Fanning W.J., Link L., et al. Can we afford to do cardiac operations in 1996? A risk-reward curve for cardiac surgery. Ann Thorac Surg 1994;58:815-821.[Abstract]
8. Loop F.D. You are in charge of cost. Ann Thorac Surg 1995;60:1509-1512.[Abstract/Free Full Text]
9. The Society of Thoracic Surgeons. Data Analyses of The Society of Thoracic Surgeons National Cardiac Surgery Database, the Seventh Year—January, 1998.
10. Bland JM and Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1(8476):307–10.

This article has been cited by other articles:

				N. Noiseux, D. Bracco, I. Prieto, and T. M. Hemmerling Do patients after off-pump coronary artery bypass grafting need the intensive care unit? A prospective audit of 85 patients Interactive CardioVascular and Thoracic Surgery, February 1, 2008; 7(1): 32 - 36. [Abstract] [Full Text] [PDF]

				V. A. Ferraris, F. H. Edwards, D. M. Shahian, and S. P. Ferraris Risk Stratification and Comorbidity Card. Surg. Adult, January 1, 2008; 3(2008): 199 - 246. [Full Text]

				J. Nilsson, L. Algotsson, P. Hoglund, C. Luhrs, and J. Brandt EuroSCORE Predicts Intensive Care Unit Stay and Costs of Open Heart Surgery Ann. Thorac. Surg., November 1, 2004; 78(5): 1528 - 1534. [Abstract] [Full Text] [PDF]

				V. A. Ferraris INVITED COMMENTARY Ann. Thorac. Surg., November 1, 2004; 78(5): 1534 - 1535. [Full Text] [PDF]

				R. H. Habib, A. Zacharias, T. A. Schwann, C. J. Riordan, S. J. Durham, and A. Shah Adverse effects of low hematocrit during cardiopulmonary bypass in the adult: Should current practice be changed? J. Thorac. Cardiovasc. Surg., June 1, 2003; 125(6): 1438 - 1450. [Abstract] [Full Text] [PDF]

				V. A. Ferraris and S. P. Ferraris Risk Stratification and Comorbidity Card. Surg. Adult, January 1, 2003; 2(2003): 187 - 224. [Full Text]

				T. A. Schwann, R. H. Habib, A. Zacharias, G. L. Parenteau, C. J. Riordan, S. J. Durham, and M. Engoren Effects of body size on operative, intermediate, and long-term outcomes after coronary artery bypass operation Ann. Thorac. Surg., February 1, 2001; 71(2): 521 - 530. [Abstract] [Full Text] [PDF]

				J.-O. Defraigne, J. Pincemail, G. Dekoster, R. Larbuisson, M. Dujardin, F. Blaffart, J.-L. David, and R. Limet SMA circuits reduce platelet consumption and platelet factor release during cardiac surgery Ann. Thorac. Surg., December 1, 2000; 70(6): 2075 - 2081. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to Personal Folders Download to citation manager Author home page(s): Christopher J. Riordan Milo Engoren Anoar Zacharias Thomas A. Schwann Gary L. Parenteau Samuel J. Durham Permission Requests Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Riordan, C. J. Articles by Habib, R. H. Search for Related Content PubMed PubMed Citation Articles by Riordan, C. J. Articles by Habib, R. H.