How to Do It: Utilizing Risk Stratification to Evaluate Outcomes in Adult Open-Heart Operations

Transcription

1 THE JouRNAl OF ExTRA-CoRPOREAl TECHNOlOGY Original Article Presented at the AmSECT 30th International Conference March 13-16, 199, Washington, D.C. How to Do It: Utilizing Risk Stratification to Evaluate Outcomes in Adult Open-Heart Operations Timothy A. Dickinson, CCP, MHA Desert Springs Hospital, Las Vegas, Nevada Keywords: database, risk stratification, open heart surgery Abstract The purpose of this study is to demonstrate that by using a proven method of stratifying open-heart operations into levels of predicted mortality, hospitals can closely monitor trends of their open-heart programs and possibly improve the health planning decisions for their institution. A proven method of uniform risk stratification utilizing objective and readily available preoperative patient data was implemented at our institution for a 1 month period (September I, 1991 through August 31, 199). A total of 367 patients were included in this study. The patients were categorized into four risk ranges (0 to 4% good risk, to 9% fair risk, to 14% poor risk, and :?:1% high risk) indicating a predictive percent probability of operative mortality. The number patients categorized as either 0 to Introduction The Wall Street J oumal reported in December of 1991 that health expenditures are expected to climb to a record 14% of our nation's total gross national product (GNP) for 199. The article also stated that overall health expenditures will continue to increase at an average annual rate of 1% to 13% during the next five years (I). Unarguably, open-heart surgery is expensive and represents a substantial percentage in total Address correspondence to: Timothy A. Dickinson, CCP, MHA Department of Perfusion Services Desert Springs Hospital 07 East Flamingo Road Las Vegas, NV % good risk, to 9% fair risk, to 14% poor risk, and :?:1% high risk were46, 74, 84and 163, respectively. The patient's average post -operative length of stay in each risk category was 7.6 days, 8. days, days and 1 days, respectively. The patient's average total hospital charges in each risk category were $48,41,$3,31,$60,416 and $7,, respectively. This information has helped our hospital administration make relevant and objective decisions concerning our open-heart program. Uniform risk stratification( outcomes research) should be incorporated into all adult open-heart surgery programs because it is simple, inexpensive, and can evaluate the outcomes and cost of open-heart surgery. health care expenditures. The 1990 average hospital charge per admission for coronary bypass surgery in the four openheart programs in Las Vegas, Nevada ranged from $7,634 to $76,30. These figures represent an average increase of 1.8% over 1989 levels (). Factors that influence the cost of openheart surgery are the duration of hospitalization, the severity of illness, and the complexity and intensity of care (3). It is no wonder that those who pay for health care, such as insurance companies and Medicare, are deeply interested in the costs and benefits of open-heart procedures. In 1986, the Health Care Financing Administration (HCFA) began publicly reporting mortality results in the Medicare population ( 4). Today these reports list hospitals by region and specifically show data representing the results of open-heart surgery in the Medicare population (). Until recently, an inherent difficulty in comparing the results of open-heart surgery 13 Vol. 4 No. 4, 7 993

2 THE JouRNAL OF ExTRA-CoRPOREAL TECHNOLOGY Legend: for Table 1 BP=blood pressure. IABP=intra-aortic balloon pump. PTCA=percutaneous transluminal coronary angioplasty. PD=peritoneal dialysis. Hemo=hemodialysis. PA=pulmonary artery. AV=aortic valve. HD=heart disease. PA=pulmonary artery. CABG=coronary artery bypass graft. Vent=ventri cu lar. T ach=tachycardi a. Fi b=fi bri II ation. UO=urine output. Ml=myocardial infarction. CHF=congestive heart failure. CR=Creatinine. COPD=chronic obstructive pulmonary disease. PAP=pulmonary artery pressure. CVA=cerebral vascular accident. MS=muscular dystrophy. HDL=high-density lipoprotein. HIV=human immunodeficiency virus. between surgeons and institutions is the lack of a simple and widely accepted method of quantifying patient risk that defines the severity of disease and identifies the many variables that contribute to a predictable outcome. The HCFAreport, for example, represents raw mortality data for Medicare patients undergoing coronary artery bypass grafting procedures while omitting many of the risk factors associated with predictably poor outcomes. There is a simple and effective way of quantifying risk thus allowing a fair comparison of different institutions' results. Dr. Parsonnet et al described a method of stratifying openheart procedures into levels of predicted operative mortality (3). This methodology was based upon preoperative, objective risk factors that are readily available in any hospital. The risk factors and their assigned weight were generated from various statistical methods including univariate and multivariate logistic-regression analyses. The criteria for including a risk variable included (3): 1) Predictive value must be demonstrated by univariate analysis; ) Data must be available for every patient; 3) Data must be available at every institution; 4) The variable must be as free as possible from subjectivity or bias; and ) The variable must be simple and direct (not derived from other information).* The predictive operative mortality is computed by summing all the individual risk factors. The predicted probability of operative mortality can be stratified into ranges reflecting a patient's expected mortality rate: good risk (0-4% ), fair risk (-9%), poor risk (-14%), high risk (~1%). The before mentioned ranges with their respective title (i.e. good risk, fair risk, etc.) were contrived by Dr. Parsonnet, et al. A comparison of predicted and observed outcomes for each of the risk ranges reveals a close correlation using various statistical methods. * Reprinted with permission. Circulation 1989 Supplement 1; 3: Table 1 Revised risk factors and their respective weights Risk Factor Female Gender Morbid obesity (;:::1. x ideal weight) Diabetes (unspecified type) Hypertension (systolic BP>l40mmHg) Ejection Fraction(%): Good (;:::SO) Fair (30-49) Poor (<30) Age (yr): ;:::so Reoperation First Second Preoperative IABP Left ventricular aneurysm resected Emergency surgery following PTCA or catheterization complications Dialysis dependency (PD or Hemo) Valve surgery: Mitral (PA systolic <9mmHg) Mitral (PA pressure ;:::60mmHg) Aortic (A V gradient< mmhg) Aortic (Pressure gradient ;:::I0mmHg) Tricuspid CABG at the time of valve surgery Cardiac conditions: Left main disease; unstable angina Vent. Tach.Nent. Fib. aborted sudden death Assigned Weight I Cardiogenic shock (UO <IOcc/hr) Transmural MI (within 48 hours) 7 Chronic CHF (peripheral edema, pleural effusion) Pacemaker dependent Aortic regurgitation, acute Mitral regurgitation, acute Ventricular septal defect, acute 0 Constrictive pericarditis Congenital heart disease in adult, cyanotic I 0 Hepato-Renal conditions: Renal Failure, chronic (CR>, without dialysis) Renal Failure, acute Cirrhosis of liver (serum bilirubin >) Pulmonary Conditions: COPD, severe Pulmonary hypertension (mean PAP >30mmHg) Idiopathic thrombocytopenic purpura Endotracheal tube, pre-operation Asthma (peak expiratory flow rate <1 00) 0 Asthma (peak expiratory flow rate <00) Vascular Conditions: Peripheral vascular disease, severe Carotid disease, unilateral occlusion Carotid disease, bilateral Abdominal aortic aneurysm, asymptomatic Dissecting Thoracic Aneurysm Miscellaneous conditions Severe neurological disorder (healed CV A, paraplegia, M.S., hemiparesis) Diabetes, Juvenile Hyperlipidemia (cholesterol >300, HDL <30) 3 Jehovah's Witness Cold Agglutinins Aspirin prescription Substance abuse (alcohol, drugs), severe 3 AIDS, Active disease (HIV positive excluded) Active neoplasm (leukemia, lymphoma, etc.) High dose steroids, active TOTAL POINTS Vol. 4 No. 4,

3 THE JouRNAL of ExTRA-CoRPOREAL TECHNOLOGY (Personal Communication with Dr. Parsonnet, et al Letter; January, 199). The risk stratification method first described by Dr. Parsonnet has since been extended to include more preoperative variables that can be used in predicting risk(table 1 ). This modification eliminated potential interpretational variances. (Personal Communication with Dr. Parsonnet, M.D. Letter; July 30, 1991). Material and Methods: For the purposes of the present study further information rele~ant to our program (i.e. surgeon's name, anesthesiologist name, cardiologist's name, perfusionist's name, hospital number, procedure, and date of surgery), were included with the variables shown in Table 1. On September 1, 1991, Dr. Parsonnet's risk stratification method was implemented into the open-heart surgery program at Desert Springs Hospital, Las Vegas, Nevada. All members of the perfusion staff were instructed to record on the perfusion record all the patient's preoperative risk factors shown in Table 1. At the end of each month, the data from the perfusion records were entered into a database program. As of August 31, 199, (1 month period) 367 open-heart procedures were recorded in this database. The material used to conduct this risk stratification study included the following items: I. Hardware A) Apple Macintosh SE II. Software A) FileMaker Prob database program B) Microsoft Excelc spreadsheet program Results: The database was queried in order to provide information concerning the relationship between preoperative patient risk factors and postoperative patient outcome. Figure 1 shows the distribution of open-heart surgery patients versus their preoperative risk classification and predicted percent probability of operative mortality. For the 1 month period ending August 31, 199; 46, 7 4, 84, and 163 patients were categorized as either 0 to 4% good risk, to 9% fair risk, to 14% poor risk, and ~1% high risk, respectively. High risk patients, defined as those patients with a ~1% predicted mortality, accounted for 44% (163/367) of the total cardiac surgery population. Figure compares the duration of postoperative hospitalization to the preoperative risk classification and predicted a Apple Computer Inc., Cuppertino, CA 9014 b Claris Corporation, Santa Clara, CA 90 c Microsoft Corporation, Redmond, CA 980 Figure 1 Distribution by Operative Risk Classification of Cardiac Surgery Patients (n=367). Number of Patients Distribution by Operative Risk Oassification of Cardaic Surgery Patients (n-367) [ ~---- -r 0 to 4% Good to 9% Fair to 14% Poor 1 %+ H1gh Pr00pe,ative Risk Classification by Predicted Percent Probability of Operative Mortality percent probability of operative mortality. The patients categorized as either 0 to 4% good risk, to 9% fair risk, to 14% poor risk, and~ 1% high risk had average postoperative length of stays of 7.6 days. 8. days, days and 1 days, respectively. The average postoperative length of stay for all patients was. days Figure 3 contrasts the average hospital charges to the preoperative risk classification and predicted percent probability of operative mortality. The patients categorized as eithero to4% goodrisk, to9% fair risk, loto 14% poor risk, and ~1% high risk had average hospital charges of$48,41, $3,31,$60,416,$7,, respectively. The average hospital charge for all patients was $64,. Discussion: The collected data has benefited our hospital in many different ways. This information has helped our hospital administration make relevant and objective decisions concerning our open-heart program. Our hospital is involved with capitated open-heart surgery contracts (i.e. contracts that pay the physician and hospital a flat fee regardless of severity of illness). The database was further queried in order to provide outcome data for individual physicians. It is vitally important to any open-heart program that enters into a capitated health care plan to choose cardiac surgeons and cardiologists that have proven skills to provide quality and cost-effective care for patients in all levels of risk. The collected data has helped 137 Vol. 4 No. 4, 1993

4 THE journal OF ExTRA-CORPOREAl TECHNOlOGY Figure Preoperative Risk Classification Versus Duration of Postoperative Hospitilization (n=367). Preoperative Risk Classification Versus Duration of Postoperative Hospitalization (n=367) 1 8 length of Stay (Days) j i 0 to 4% Good to 9% Fair 1 0 to 1 4% Poor 1%+ High Preoperative Risk Classification by Predicted Percent Probability of Operative Mortality Figure 3 Preoperative Risk Classification Versus Total Hospital Charges (n=367). Preoperative Risk Classification Versus Total Hospital Charges (n=367) 80,000,- 70,000 60,000 0,000 Hospital Charge (in dollars) ' 30,000 0,000, to 4% Good to 9% Fair to 14% Poor 1%+ High Preoperative Risk Classification by Predicted Percent Probability of Operative Morta6ty Vol. 4 No. 4,

5 THE journal OF EXTRA-CORPOREAL TECHNOLOGY our hospital administration select cardiac surgeons and cardiologists for our capitated health care plans. Another advantage of using Dr. Parsonnet's model has been the heightened awareness by both the cardiac surgeon and cardiologist that they are being closely monitored in regards to their medical practice. The risk stratification (outcomes) data is periodically presented at monthly cardiovascular services committee meetings which are attended by staff cardiologists, staff cardiac surgeons and one or more representatives from our hospital administration staff. This level of accountability, with additional pressure from our hospital administration, has led some of the cardiologists to send some of their elective high-risk open-heart patients to tertiary care centers, rather than referring them for surgery in our community hospital. Furthermore, some of the cardiac surgeons are refusing to operate on extremely high-risk patients referred for elective open-heart surgery. The end result of this analysis at our institution is an openheart program that includes improved physician involvement. It is an open-heart program that will almost assuredly be profitable and provide high quality care with acceptable mortality and morbidity statistics. On a national level we are in agreement with Dr. Clark who states that "It is time for a national cardiothoracic surgical database" (6). Some of the reasons given for the generation of this database include: * 1) The Liability Committee of the Society of Thoracic Surgeons (STS) has revealed that misrepresentation and misconception of operative risks are leading causes of malpractice actions. ) The Standards and Ethics Committee of the STS has shown that in some cases the hospitals, the cardiologists, and the surgeons have not recognized the changing patterns of patient populations receiving coronary artery bypass grafting procedures. This has been especially true with development of newer catheter intervention techniques, thrombolysis, and revascularization in the setting of acute myocardial ischemia. These circumstances have helped to perpetuate the prevalent professional and public perception that coronary artery operations carry little or no risk. 3) The Joint Commission on Accreditation of Health Care Organizations (JCAHO) has made it a requirement to have quality assurance programs for every segment of the health care system. Recredentialing on an annual basis is based on Reprinted with permission from the Society of Thoracic Surgeons (The Annals of Thoracic Surgery, 1989;48: the surgeon's results from the previous year. * As mentioned earlier HCFA's evaluations of open-heart programs fall short of being reasonable. The HCFA report predicts operative mortality without using many of the risk factors associated with predictably poor outcomes. For this reason, a system of uniform risk stratification( outcomes research) needs to be initiated now by all open-heart programs. Dr. Parsonnet and colleagues have gone a long way in providing a straightforward method for categorizing heart surgery patients preoperatively into various risk groups. His data shows that a prospective score is found to be highly predictive of operative mortality and closely relates to the overall complication rate and duration of postoperative hospitalization. If such a system was applied more universally it might be possible to compare outcomes of open-heart surgery at any institution. Comparisons between individual cardiac surgeons would also be possible. This type of analysis would also provide a basis for a meaningful informed consent or reveal an operative risk that is unacceptably high. As hospitals and cardiac surgeons enter into capitated health care contracts, the need to analyze the open-heart surgery program's patient selection and patient outcome will be increased. The widespread presence of coronary artery disease, the rapid growth of cardiac surgical programs across this country(?), the development of alternative interventional therapies, and an aging population have encouraged an increasing number of cardiac surgeons to treat high-risk patients even though manpower and hospital-based resources may be limited. Identifying preoperatively, those patients that fall into high-risk groups may provide useful information to the attending cardiac surgeon, the patient and family members. Acknowledgments: The author wishes to express his deep appreciation to his brother Chris J. Dickinson, MD, for his constructive criticisms and encouragement. The author also wishes to thank Jack Prichard, CCP and Faye Rieckens, CCP for their assistance in collection of the data and for their review of this article. References 1. Bangdanich W. Health expenditures expected to climb to record 14% of U.S. output in 199. Wall Street Journal. 1991; Dec 30: A.. Bureau of Health Planning. Personal Health Choices State of Nevada Department of Human Resources Division of Health. 4th Edition, November Parsonnet V, Dean D, Bernstein AD. A method of uniform stratification of risk for evaluating the results of surgery in acquired heart disease. Circulation Supplement 1 ;79: Vol. 4 No. 4, 1993

6 THE JouRNAL of ExTRA-CORPOREAL TEcHNOLOGY 4. Rosen HM, Green BA. The HCFA excess mortality lists: A methodological critique. Hosp Health Ser Admin. 1987;3: Medicare Hospital Mortality Information 1987, 1988, Volume 1 U.S. Department of Health and Human Services. Health Care Financing Administration. HCFA Pub. No Clark, R. It is time for a national cardiothoracic surgical data base. Ann Thorac Surg. 1989;48: Souhrada, L. Hospitals pursue heart programs-despite pitfalls. Hospitals. Oct. 0, 1989: Vol. 4 No. 4,