- Email: [email protected]
Guidelines for reporting morbidity and mortality after cardiac valvular operations
Volume 96,
September 1988
Number 3
THORACIC AND CARDIOVASCULAR SURGERY The Journal
J THORAC
CARDIOVASC SURG
of
1988;96:351-3
Original Communications
Guidelines for reporting morbidity and mortality after cardiac valvular operations L. Henry Edmunds, Jr., MD, Phi/adelphia, Pa., Richard E. Clark, MD, Bethesda, Md., Lawrence H. Cohn, MD, Boston, Mass., D. Craig Miller, MD, Stanford, Calif., and Richard D. Weisel, MD, Toronto, Ontario, Canada*
Purpose The purpose of these guidelines is to facilitate the analysis and reporting of results of operations on diseased cardiac valves. The definitions and recommendations that follow are guidelines, not standards, and are designed to facilitate comparisons between the experiences of different surgeons who treat different cohorts of patients at different times with different techniques and materials. Operative (hospital) mortality Death from any cause during or after operation within 30 days if the patient is discharged or within any Byagreement of the Editors, this article is being published concurrently in THE JOURNAL OF THORACIC AND CARDIOVASCULAR SURGERY, The Annals of Thoracic Surgery, and the European Journal of Cardia-Thoracic Surgery. Editor's note: This important publication describes guidelines, not standards or rules. All in the cardiac surgical community will benefit from careful study of these guidelines, and their incorporation into analyses and manuscripts. THE JOURNAL OF THORACIC AND CARDIOVASCULAR SURGERY endorses and respects these guidelines, but will maintain a flexible editorial policy. 'Members, Ad Hoc Liaison Committee for Standardizing Definitions of Prosthetic Heart Valve Morbidity of The American Association for Thoracic Surgery and The Society of Thoracic Surgeons.
interval if the patient is not discharged is designated operative mortality. Hospital to hospital transfer is not considered discharge. Definitions of morbidity Structural deterioration. Any change in valve function resulting from an intrinsic abnormality causing stenosis or regurgitation. Structural deterioration includes valve deterioration exclusiveof infected or thrombosed valves as determined by reoperation, autopsy, or clinical investigation. The term structural deterioration refers to changes intrinsic to the valve, such as wear, stress fracture, poppet escape, calcification, leaflet tear, stent creep, and disruption or stenosis of a reconstructed valve. Nonstructural dysfunction. Any abnormality resulting in stenosis or regurgitation at the valve that is not intrinsic to the valve itself. Nonstructural dysfunction refers to nonstructural problems that result in valve dysfunction exclusive of thromboembolism and infection diagnosed by reoperation, autopsy, or clinical investigation. Examples of nonstructural dysfunction include entrapment by pannus or suture, paravalvular leak, inappropriate sizing, and clinically important hemolytic anemia. 351
The Journal of Thoracic and Cardiovascular Surgery
3 5 2 Edmunds et al.
Thromboembolism. Any valve thrombosis or embolus exclusive of infection. Thromboembolism includes any new, permanent or transient, focal or global, neurologic deficit (exclusive of hemorrhage) and any peripheral arterial emboli unless proved to have resulted from another cause (e.g., atrial myxoma). Patients who do not awaken postoperatively or who awaken with a stroke or myocardial infarction are excluded. Acute myocardial infarction that occurs after operation is arbitrarily defined as a thromboembolic event in patients with known normal coronary arteries or those who are less than 40 years of age. Valve thrombosis may be proved by operation, autopsy, or clinical investigation (e.g., echocardiography, angiocardiography, or magnetic resonance imaging) and is listed as a separate subcategory of thromboembolism. Anticoagulant-related hemorrhage. Any episode of internal or external bleeding that causes death, stroke, operation, or hospitalization or requires transfusion. The category anticoagulant-related hemorrhage is restricted to patients who are receiving anticoagulants and/or antiplatelet drugs. Anticoagulant medications, doses, and therapeutic guidelines should be designated. Prosthetic valve endocarditis. Any infection involving a heart valve substitute or a reconstructed native valve. The diagnosis of prosthetic valve endocarditis is based on customary clinical criteria including an appropriate combination of positive blood cultures, clinical signs (fever, new or altered cardiac murmurs, splenomegaly, systemic emboli, or immunopathologic lesions), and/or histologic confirmation of endocarditis at reoperation or autopsy. Morbidity associated with active infection, such as valve thrombosis, embolus, or paravalvular leak, is included under this category and is not included in other categories of morbidity. Consequences of morbid events Reoperation. Any operation that repairs, alters, or replaces a previously placed prosthesis or repaired valve. The reasons for reoperation should be reported and may include reasons other than valve morbidity, such as recall, excessive noise, or prophylactic removal. Valve-related mortality. Death caused by structural deterioration, nonstructural dysfunction, thromboembolism, anticoagulant-related bleeding, prosthetic endocarditis, or death at reoperation. Sudden, unexplained, unexpected deaths of patients with prosthetic or repaired valves are included unless
autopsy proves that death is not valve-related. Deaths caused by heart failure in patients with advanced myocardial disease and satisfactorily functioning cardiac valves are not included. Causes of valve-related deaths should be designated and reported. Permanent valve-related impairment. Any permanent functional deficit caused by structural deterioration, nonstructural dysfunction, thromboembolism, anticoagulant-related bleeding, prosthetic endocarditis, or reoperation. Combinations of morbid consequences. Useful combinations for reporting include the following: 1. Valve-related mortality, including operative deaths, and reoperation 2. Valve-related mortality, including operative deaths, and permanent disability 3. All valve-related morbidity and mortality: This combination includes all operative deaths, all valve-related deaths including sudden unexplained deaths, and all nonfatal valve-related morbid events Data collection Stratification. Data collection should be stratified by valve location (aortic, mitral, aortic and mitral) and by valve manufacturer and model. Inclusion of results of different valve models and different locations in the same analysis is often not meaningful or instructive. Valve sizes. Valve sizes for each model should be stated. Additional pertinent material. In addition, each report should specify: I. The patient population from which the sample population was collected 2. The method employed for follow-up (e.g., examination, telephone, retrospective, prospective 3. The time interval (closing interval) required to complete the follow-up. (The closing interval should be as short as possible but will vary in proportion to the number of patients followed up.) 4. The completeness of follow-up during the closing interval. (This should be 95% or better.) 5. The incidence of autopsies and causes of death Data analysis and reporting The method of reporting data should facilitate comparison between reports and support the conclusions, inferences, and predictions made. At this time there is no consensus on the best way to do this. The incidence of some morbid events that occur within a short time frame (such as the interval between
Volume 96 Number 3 September 1988
operation and hospital discharge) may be reported as events divided by patients (e.g., percent operative mortality). Incident data may be compared by x' analysis or Fisher's exact test. Analysis of variance techniques and logistic regression analysis are available for comparisons between multiple variables. Most valve-related morbid events should be reported in a time-related manner, with operation designated as time zero. Kaplan-Meier or other life table techniques provide actuarial estimates of morbid events and should be reported with the standard error of the estimate or with 67% or 95% confidence limits. The number of patients remaining at risk should be indicated at appropriate intervals. Although comparisons between subsets of patients can be made, actuarial methods are not predictive beyond the time of the last actuarial estimate and cannot be adapted to multivariate analysis. The Cox proportional hazard model produces timedependent analysis of valve-related events and provides multivariate, stepwise regression methods to identify independent determinants associated with specificvalverelated morbid events. The Cox model calculates a semiparametric hazard function that necessitates the use of proportional hazard curves to indicate hazard changes with time. The Cox model is not as useful as a completely parametric method. A completely parametric method of calculating a hazard function of valve-related morbid events defines the instantaneous risk of an event at any time after operation. The method permits univariate and multivariate analysis, provides predictive information beyond the
Guidelines 3 5 3
time of the last event, indicates whether or not the risk is constant, and provides confidence limits. The method does not reflect repeated events in the same patient. Linearized rates (events per 100 patient-years or percent per patient-year) can be used to summarize the incidence of multiple events in individual patients. These rates should not be used unless the hazard function for the complication under study is proved to be constant during the entire time interval considered. Linearized rates should be reported with appropriate confidence limits (e.g., standard error) and compared by appropriate statistics (e.g., the maximum likelihood ratio test). Multiple-decrement analysis may be employed to evaluate composite indices summarizing combinations of multiple valve-related complications. The methods chosen to analyze the collected data will be influenced by the purpose of the report and the availability of various analytical techniques. As with cardiac surgery, statistical science is a dynamic discipline and the methods used may vary between statisticians. However, in all reports the conclusions, predictions, and inferences made should be supported by the collected data and appropriate analysis of the data. The committee members thank all who have contributed to the development of these guidelines, especially the following physicians: Cary W. Akins, Delos M, Cosgrove, Anthony R. C. Dobell, Paul A. Ebert, Bernard 1. Gersh, W. R. Eric Jamieson, John W. Kirklin, Nicholas T. Kouchoukos, Hillel Laks, Floyd D. Loop, Donald J. Magilligan, Jr., D. Glenn Pennington, Donald N. Ross, Hartzell V. Schaff, and Andrew S. Wechsler.
September 1988
Number 3
THORACIC AND CARDIOVASCULAR SURGERY The Journal
J THORAC
CARDIOVASC SURG
of
1988;96:351-3
Original Communications
Guidelines for reporting morbidity and mortality after cardiac valvular operations L. Henry Edmunds, Jr., MD, Phi/adelphia, Pa., Richard E. Clark, MD, Bethesda, Md., Lawrence H. Cohn, MD, Boston, Mass., D. Craig Miller, MD, Stanford, Calif., and Richard D. Weisel, MD, Toronto, Ontario, Canada*
Purpose The purpose of these guidelines is to facilitate the analysis and reporting of results of operations on diseased cardiac valves. The definitions and recommendations that follow are guidelines, not standards, and are designed to facilitate comparisons between the experiences of different surgeons who treat different cohorts of patients at different times with different techniques and materials. Operative (hospital) mortality Death from any cause during or after operation within 30 days if the patient is discharged or within any Byagreement of the Editors, this article is being published concurrently in THE JOURNAL OF THORACIC AND CARDIOVASCULAR SURGERY, The Annals of Thoracic Surgery, and the European Journal of Cardia-Thoracic Surgery. Editor's note: This important publication describes guidelines, not standards or rules. All in the cardiac surgical community will benefit from careful study of these guidelines, and their incorporation into analyses and manuscripts. THE JOURNAL OF THORACIC AND CARDIOVASCULAR SURGERY endorses and respects these guidelines, but will maintain a flexible editorial policy. 'Members, Ad Hoc Liaison Committee for Standardizing Definitions of Prosthetic Heart Valve Morbidity of The American Association for Thoracic Surgery and The Society of Thoracic Surgeons.
interval if the patient is not discharged is designated operative mortality. Hospital to hospital transfer is not considered discharge. Definitions of morbidity Structural deterioration. Any change in valve function resulting from an intrinsic abnormality causing stenosis or regurgitation. Structural deterioration includes valve deterioration exclusiveof infected or thrombosed valves as determined by reoperation, autopsy, or clinical investigation. The term structural deterioration refers to changes intrinsic to the valve, such as wear, stress fracture, poppet escape, calcification, leaflet tear, stent creep, and disruption or stenosis of a reconstructed valve. Nonstructural dysfunction. Any abnormality resulting in stenosis or regurgitation at the valve that is not intrinsic to the valve itself. Nonstructural dysfunction refers to nonstructural problems that result in valve dysfunction exclusive of thromboembolism and infection diagnosed by reoperation, autopsy, or clinical investigation. Examples of nonstructural dysfunction include entrapment by pannus or suture, paravalvular leak, inappropriate sizing, and clinically important hemolytic anemia. 351
The Journal of Thoracic and Cardiovascular Surgery
3 5 2 Edmunds et al.
Thromboembolism. Any valve thrombosis or embolus exclusive of infection. Thromboembolism includes any new, permanent or transient, focal or global, neurologic deficit (exclusive of hemorrhage) and any peripheral arterial emboli unless proved to have resulted from another cause (e.g., atrial myxoma). Patients who do not awaken postoperatively or who awaken with a stroke or myocardial infarction are excluded. Acute myocardial infarction that occurs after operation is arbitrarily defined as a thromboembolic event in patients with known normal coronary arteries or those who are less than 40 years of age. Valve thrombosis may be proved by operation, autopsy, or clinical investigation (e.g., echocardiography, angiocardiography, or magnetic resonance imaging) and is listed as a separate subcategory of thromboembolism. Anticoagulant-related hemorrhage. Any episode of internal or external bleeding that causes death, stroke, operation, or hospitalization or requires transfusion. The category anticoagulant-related hemorrhage is restricted to patients who are receiving anticoagulants and/or antiplatelet drugs. Anticoagulant medications, doses, and therapeutic guidelines should be designated. Prosthetic valve endocarditis. Any infection involving a heart valve substitute or a reconstructed native valve. The diagnosis of prosthetic valve endocarditis is based on customary clinical criteria including an appropriate combination of positive blood cultures, clinical signs (fever, new or altered cardiac murmurs, splenomegaly, systemic emboli, or immunopathologic lesions), and/or histologic confirmation of endocarditis at reoperation or autopsy. Morbidity associated with active infection, such as valve thrombosis, embolus, or paravalvular leak, is included under this category and is not included in other categories of morbidity. Consequences of morbid events Reoperation. Any operation that repairs, alters, or replaces a previously placed prosthesis or repaired valve. The reasons for reoperation should be reported and may include reasons other than valve morbidity, such as recall, excessive noise, or prophylactic removal. Valve-related mortality. Death caused by structural deterioration, nonstructural dysfunction, thromboembolism, anticoagulant-related bleeding, prosthetic endocarditis, or death at reoperation. Sudden, unexplained, unexpected deaths of patients with prosthetic or repaired valves are included unless
autopsy proves that death is not valve-related. Deaths caused by heart failure in patients with advanced myocardial disease and satisfactorily functioning cardiac valves are not included. Causes of valve-related deaths should be designated and reported. Permanent valve-related impairment. Any permanent functional deficit caused by structural deterioration, nonstructural dysfunction, thromboembolism, anticoagulant-related bleeding, prosthetic endocarditis, or reoperation. Combinations of morbid consequences. Useful combinations for reporting include the following: 1. Valve-related mortality, including operative deaths, and reoperation 2. Valve-related mortality, including operative deaths, and permanent disability 3. All valve-related morbidity and mortality: This combination includes all operative deaths, all valve-related deaths including sudden unexplained deaths, and all nonfatal valve-related morbid events Data collection Stratification. Data collection should be stratified by valve location (aortic, mitral, aortic and mitral) and by valve manufacturer and model. Inclusion of results of different valve models and different locations in the same analysis is often not meaningful or instructive. Valve sizes. Valve sizes for each model should be stated. Additional pertinent material. In addition, each report should specify: I. The patient population from which the sample population was collected 2. The method employed for follow-up (e.g., examination, telephone, retrospective, prospective 3. The time interval (closing interval) required to complete the follow-up. (The closing interval should be as short as possible but will vary in proportion to the number of patients followed up.) 4. The completeness of follow-up during the closing interval. (This should be 95% or better.) 5. The incidence of autopsies and causes of death Data analysis and reporting The method of reporting data should facilitate comparison between reports and support the conclusions, inferences, and predictions made. At this time there is no consensus on the best way to do this. The incidence of some morbid events that occur within a short time frame (such as the interval between
Volume 96 Number 3 September 1988
operation and hospital discharge) may be reported as events divided by patients (e.g., percent operative mortality). Incident data may be compared by x' analysis or Fisher's exact test. Analysis of variance techniques and logistic regression analysis are available for comparisons between multiple variables. Most valve-related morbid events should be reported in a time-related manner, with operation designated as time zero. Kaplan-Meier or other life table techniques provide actuarial estimates of morbid events and should be reported with the standard error of the estimate or with 67% or 95% confidence limits. The number of patients remaining at risk should be indicated at appropriate intervals. Although comparisons between subsets of patients can be made, actuarial methods are not predictive beyond the time of the last actuarial estimate and cannot be adapted to multivariate analysis. The Cox proportional hazard model produces timedependent analysis of valve-related events and provides multivariate, stepwise regression methods to identify independent determinants associated with specificvalverelated morbid events. The Cox model calculates a semiparametric hazard function that necessitates the use of proportional hazard curves to indicate hazard changes with time. The Cox model is not as useful as a completely parametric method. A completely parametric method of calculating a hazard function of valve-related morbid events defines the instantaneous risk of an event at any time after operation. The method permits univariate and multivariate analysis, provides predictive information beyond the
Guidelines 3 5 3
time of the last event, indicates whether or not the risk is constant, and provides confidence limits. The method does not reflect repeated events in the same patient. Linearized rates (events per 100 patient-years or percent per patient-year) can be used to summarize the incidence of multiple events in individual patients. These rates should not be used unless the hazard function for the complication under study is proved to be constant during the entire time interval considered. Linearized rates should be reported with appropriate confidence limits (e.g., standard error) and compared by appropriate statistics (e.g., the maximum likelihood ratio test). Multiple-decrement analysis may be employed to evaluate composite indices summarizing combinations of multiple valve-related complications. The methods chosen to analyze the collected data will be influenced by the purpose of the report and the availability of various analytical techniques. As with cardiac surgery, statistical science is a dynamic discipline and the methods used may vary between statisticians. However, in all reports the conclusions, predictions, and inferences made should be supported by the collected data and appropriate analysis of the data. The committee members thank all who have contributed to the development of these guidelines, especially the following physicians: Cary W. Akins, Delos M, Cosgrove, Anthony R. C. Dobell, Paul A. Ebert, Bernard 1. Gersh, W. R. Eric Jamieson, John W. Kirklin, Nicholas T. Kouchoukos, Hillel Laks, Floyd D. Loop, Donald J. Magilligan, Jr., D. Glenn Pennington, Donald N. Ross, Hartzell V. Schaff, and Andrew S. Wechsler.