Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Appendix A Background and Conduct of the Workshop STRUCTURE OF THE WORKSHOP This appendix describes the acute myocardial infarction (AMI) workshop project and the materials developed or used as background for committee discussions. The main elements of the project were background reading, information on data sets of the Health Care Financing Administration (HCFA) and analyses done for the workshop, and a committee homework exercise that yielded a first-round set of patient management topics. The workshop was conducted in three public sessions and an executive session, in addition to the previous evening's reception and background ses- sion. The first public session featured presentations by Michael McMullan and Henry J. Krakauer of HCFA, a brief review by Kenneth I. Shine (chair of the study committee) of the factors that the committee should keep in mind in recommending patient management topics, and a brief review of the homework exercise by Kathleen N. Lohr (of the Institute of Medicine LIOM] staff). The second session consisted of general discussion by the committee of the preliminary set of patient management topics, followed by a round of "voting" on the main patient management topics. The third ses- sion accomplished further refining of the patient management issues and a discussion of He primary research strategies related to those issues. The executive session focused on the final recommendations the committee wished to endorse. BACKGROUND READINGS The IOM staff compiled a large set of background materials, which was forwarded to the committee before the meeting, to establish the context of the Effectiveness Initiative and familiarize members win the issues to be discussed. These readings centered on Medicare data files; work in progress funded by the National Heart, Lung and Blood Institute of the National 49
so APPENDIX A Institutes of Health and the National Center for Health Services Research; and recent reports of the findings of studies relating to use of services and efficacy of various treatment regimens (see Bibliography). HCFA DATA Medicare/Medicaid Decision Support System The description of the Medicare decision support systems included a sim- plified version of the flow of data into the Medicare system. That flow starts with entitlement and demographic data for about 33 million Medicare beneficiaries, which are obtained initially by the Social Security Administration (SSA). Health care providers and contractors are the prima- ry source of Medicare utilization data. Providers (e.g., institutional providers, home care agencies, suppliers, and physicians) submit bills to fis- cal intermediaries (for Medicare Part A) and carriers (for Medicare Part B); they in turn adjudicate and then pay the bills and pass them are in the sys- tem. These utilization data are merged with the SSA demographic informa- tion, and from these main sources, several basic record groups are devel- oped. Basic Record Groups The first record group is the Health Insurance Master (HIM) Enrollment record, developed from the SSA file; these data, which are updated daily, include dates of birth and death, sex, race, residence, dates of entitlement, and dates of enrollment into health maintenance organizations. This is a rich source of data for identifying beneficiaries and drawing samples for fol- low-up research studies. The second file, the Provider of Service (POS) Record, contains considerable information on hospitals, skilled nursing facilities, home health agencies, independent laboratories, ambulatory surgi- cal centers, and similar providers for Medicare. The Gird and fourth files are He Utilization Records for Medicare Parts A and B billing information, including hospital days of care, diagnoses, surgical procedures, physician visits, charges, and payments. The fifth main record group is the Provider Cost Report Record, which has cost, accounting, and other data from partic- ipating institutional providers. For effectiveness research, other "derivative" files may be important sources of information: MEDPAR (Medicare Provider Analysis and Review file), MADRS (Medicare Automated Data Retrieval System), and BMAD (Medicare Annual Data System for Part B). The SSA-based HIM file provides the beneficiary identification number and demographic infor-
APPENDIX A 51 mation; that information can be used to enter these other files for more detailed utilization information. MEDPAR is a 100 percent file of Part A inpatient care (about 10 million admissions per year). Because it has person-level data with unique identi- fiers, it can be used to identify individuals who have received inpatient ser- vices related to the diagnosis of AMI. Among the information elements on this file are principal and secondary diagnoses and surgical procedures (ICD-9-CM [International Classification of Disease, ninth revision, clinical modification] codes), days of care, charges, and provider. This file is updat- ed quarterly. MADRS is a newer 100 percent file that links Part A and Part B data for all persons receiving inpatient hospital care; as of mid-l989, it covered 1986, 1987, and 1988 and is updated monthly. It allows the creation of episodes of care; Medicare-covered inpatient and outpatient care given to a beneficiary before and after a hospitalization can be identified. For this file, which contains about 250 million records per year, Part B (outpatient) data are in summary form only. The B MAD file is built on a 5 percent sample of beneficiaries and con- tains about 21 million records, which are updated annually. It provides somewhat more information than the MADRS file on all outpatient services for this sample, such as expenditures, place and type of service, visits, and procedures; the last are coded with the HCFA Common Procedure Coding System (HCPCS), which is based on CPT-4 (Current Procedural Terminology, fourth version) codes. An example was offered of how existing data sets might be used to con- duct analyses related to AMI (especially to monitor trends and examine variations in He use of services). First, researchers would select the ICD-9- CM code for acute myocardial infarction and Men enter an inpatient file to extract all records for individuals who had services with that code. Then, because of the presence of unique beneficiary identifiers, the researchers could enter a file that contains information, for each beneficiary, on all insti- tutional services and some summary data on outpatient care. Third, to obtain more detailed information on physician and supplier services, researchers could then examine a file that contains considerably more detailed data on a 5 percent sample of beneficiaries (i.e., BMAD). Acquiring Additional Clinical and Outcomes Data HCFA can obtain additional clinical information (such as data on treat- ments administered and physiologic aspects of the disease itself) from selected inpatient medical records. One mechanism may be through the Medicare PROs by means of He proposed Uniform Clinical Data Set.
52 APPENDIX A In 1987, HCFA's Health Standards and Quality Bureau (HSQB) began a complex project to develop a data set for use by He Medicare Peer Review Organizations (PROs) and the wider research community; it was intended to contain far more detailed clinical data Man heretofore available in HCFA data files. Known as the Uniform Clinical Data Set (UCDS), this project is part of a set of steps meant to expand and improve tile ability of the agency to assure the quality of care delivered to Medicare beneficiaries, by using He PROs as the principal mechanism. A second purpose of the UCDS is to permit the development of more and better information about what works in the practice of medicine, precisely the aim of effectiveness research. The availability of extensive clinical information collected in UCDS formats would support much more thorough and detailed analysis of patterns of interventions and of outcomes than is possible simply with billing data. Thus, for patients with particular medical conditions, such as acute myocar- dial infarction, a large body of information could be made available to He medical community and for intramural and extramural research. The basic operating premise of the UCDS is that relevant clinical data will be abstracted from medical records of all inpatient admissions Hat are reviewed by the PROs (This currently amounts to about 20 to 25 percent of all Medicare admission in a year, or about 2.0 to 2.5 million admissions; of these, about 3 percent are a truly random sample of admissions, and He remainder are cases mandated for review for venous reasons. However, the large denominator 2.0 to 2.5 million should reduce the potential impact of the nonrandom portion of this sample.) PRO personnel will abstract medical records either on-site or at a central office using desktop or laptop computers. The total number of data elements available on the UCDS is about 1600, although not every data element is needed or relevant for every case. The contents of the UCDS fall into 10 major categories: I. Patient Identifying Information IT. Patient History and Physical Examination and History and Physical Exam Findings m. Laboratory Findings IV. Imaging Findings and Other Diagnostic Test Findings V. Endoscopic Procedures VI. Operative Episodes VII. Treatment Interventions VIII. Medication Therapy in Hospital IX. Inhospital Course X. Patient Discharge Status and Discharge Planning
APPENDIX A 53 Detailed guidelines that describe precisely the data to be acquired have been developed; for an example relating to AMI, see Table A.1. As of April 1989, the project was in a pilot-test phase. Field testing of the whole approach was to be conducted in late 1989 and early 1990. An assessment and recommendation as to whether to go forward with this approach were expected late in 1990. HCFA is also working to develop mechanisms to collect measures of functional impairment and other patient outcome data more directly. Acute Myocardial Infarction Analyses Illustrating the Use of Medicare Data HCFA staff compiled an array of data tables from their analyses of Medicare files on myocardial infarction, exercise testing, revascularization, and other cardiovascular conditions and interventions illustrative of the types of analyses that might be done with AMI data. These materials were distributed the evening before the meeting and were further elaborated in presentations at the workshop. The data summarized the longitudinal expe- rience of patients with ischemic heart disease, including mortality rates, morbidity (as evidenced by rehospitalization, with dissection by time and cause, and ambulatory services), and disability (as evidenced by use of skilled nursing facility and home health agency services). The first set of analyses examined the incidence of exercise testing in the ambulatory setting and the events that follow it. In 1985, some 18,000 Medicare patients in the 5 percent Medicare sample reported in the B MAD file had electrocardiographic or radionuclide exercise testing, of which about one-~ird occurred in patients with ischemic heart disease identified in a prior hospitalization. This latter group, about 6,000 patients, was exclud- ed from the analyses. There was almost a ~ree-fold increase in the use of radionuclide testing between 1985 and 1986, although this increase was still less than 10 percent of the total undergoing stress testing. A marked increase in the risk of hospitalization occurs in the immediate period after exercise testing, for cardiovascular causes such as angina, AMI, congestive heart failure, arrhythmias, or interventions (e.g., bypass surgery or angioplasty). There is a three- to four-fold increase in the incidence of admission for an AMI in the 30 days after testing, but no apparent increase in the mortality rate for those admitted with an AMI. After 180 days the risk of admission is about four times greater for noncardiac than for cardiac reasons, implying significant other comorbidities in this population. However, interpretation of these data is subject to several problems: there is no equivalent time to event analysis, but in any year about 20 percent of
54 APPENDIX A TABLE A.~. Example of Data Elements Relevant to Acute Myocardial Infarction Recorded for He Uniform Clinical Data Set: Cardiac Ca~etenzation and Ventriculogram Formal report of the first cardiac catheterization and/or ventriculogram performed during the admission or up to 6 weeks before admission; if more than one test, use the one closest to admission. Catheterization takes precedence over ventriculogram Any procedure done in an operating room, minor treatment room, at the bedside, or in the radiology suite can be includ- ed. All appropriate categories of specified findings are checked. The general rules for recoding information for the UCDS are to change default values on the computer screen OF") to "T." The specific findings to be recorded for catheterization or ventriculogram are the following. The reviewer changes F to T unless a percentage is called for, in which case the worst percentage is recorded, or other information is specified. Normal AV shunt Ventricular/atrial septal defect Valvular defects: Aortic stenosis (~1 sq cm) Aortic regurgitation (moderate or severe) Mitral stenosis (<1 sq cm) Stenosis: Stenosis: Stenosis: Stenosis: left main (%) left anterior descending (%) circumflex (%) right (%) Cardiac output Qiters/minute) (%) LV ejection fraction (%) Abnormal chamber size/wall motion Ventricular aneurysm Congenital anomalies (patent ductus, ventricular septet defect) Acetic aneurysm Dissecting aortic aneurysm Other abnormal findings Coronary artery grafts number Number with >70~o stenosis Pressures Left ventricular systolic; diastolic Aorticsystolic; diastolic; mean Pulmonary artery (including Swan-Ganz) systolic; diastolic SOURCE: "Resource Manual for Uniform Clinical Data Set (UCDS)" prepared by Case Mix Research, Queens University, Department of Community Health and Epidemiology, Kingston, Ontario, Canada in association with Wisconsin Peer Review Organization (WIPRO), Madison, Wisconsin, 1988.
APPENDIX A 55 Medicare beneficiaries are hospitalized, compared with 25 percent of exer- cise-tested persons; there are no baseline data establishing the risk of ac~mis- sion for all Medicare patients from all causes dunug a similar time period; and the data cannot distinguish between the stress test as the incident dis- covery of ischemic disease in the patient and as a follow-up in a patient with known ischemic disease not previously recorded in Medicare data. These problems were discussed at length by the committee. The second set of analyses presented by HCFA staff concerned the fre- quency of admission for AMI and the relationship of admission rates to mortality after discharge and readmission at a later date. Over the period 1984-1987, there was a yearly decrease in the number of admissions for AMI, but a relative increase in the percentage of total admissions because of a greater decrease in admissions for all other causes. The aggregate mortali- ty one year after AMI was about 25 percent.~om aR causes; this trend was stable over the entire period. Following AMI there is a 70 percent risk of readmission for any cause In the first two years, and a 60 percent risk for cardiovascular reasons; 20 percent of these readmissions occur in the first 30 days after the initial hospitalization, most for cardiovascular reasons. During this 1984-1987 period there was approximately a two-fold increase in readmission for invasive procedures such as angioplasty, catheterization, or bypass surgery. Although the overall rate of readmission and duration remained relatively stable over the period, the associated expenditures rose substantially. The third set of analyses attempted to characterize the charges for health care services following hospitalization for AMI and extending for an aver- age of six months thereafter. Ambulatory utilization was posited as a surro- gate of morbidity, as was use of support services (skilled nursing facilities, home health) for disability. No HCFA data are available for pharmaceutical utilization in an ambulatory setting. During 1984-1987, there was a 50 per- cent yearly increase in charges for ambulatory services, with a concurrent ~ percent average increase in all Medicare charges, indicating a relative increase in ambulatory services. However, no conclusions can be drawn from this increase as to whether it represents an increase or decrease in overall morbidity, including both inpatient and outpatient care. Several issues raised by the committee members before the meeting were addressed during the ensuing discussion. First, the accuracy of the ICD-9- CM code for AMI has been questioned as He result of several published analyses. In 25 to 30 percent of hospitalizations with the principal diagnosis of AMI, there are no substantiating clinical objective findings (e.g., electro- cardiographic changes or enzyme elevations) in a clinical database abstract- ed by the PROs for different HCFA projects. If patients who died In the first day (often before objective changes) are separately classified in the
56 APPENDIX A analyses the mortality risk of those without objective findings drops to 16 percent at 30 days, as compared to 21 percent for those with objective f~nd- ings. One committee member cautioned that in the elderly population specifically, several other factors must be considered, including a higher percentage of older patients with no apparent change on repeated ECG and the possibility of a longer period of time between the incidence of infarction and admission. Second, because of the committee's interest in risk stratification, efforts by HCFA to model the predictive value of various clinical findings, test results, or interventions were reviewed. In general, age, the presence of an arrythmia (particularly atrioventricular dissociation), and mental disorienta- tion are associated with an excess risk of both short- and long-term mortali- ty. As for interventions, based on data in 1985 angioplasty and bypass were associated with a reduced risk of death, whereas thrombolytic agents showed a beneficial trend, but one lacking great statistical significance. HOMEWORK EXERCISE The homework exercise was conducted as the first part of a modified Delphi process, in which committee members completed questionnaires to nominate Tree major patient management topics and then to elaborate the research activities they would recommend for those specific topics. Table A.2 lists the topics nominated in the fust round. Of these 10 cate- gories, the issues mentioned most often dealt with risk stratification, use of thrombolytic agents, use of invasive interventions, and outcomes. The attention to outcomes other than deathhealth status and quality of life, as represented by functioning, emotional well-being, return to usual activities (e.g., works was striking. In addition, issues of costs, cost-effectiveness, and cost-eff~ciency were noted with some frequency. The research strategies that correspond to these study topics were quite varied. Tables A.3 through A.8 provide the full results of the homework exercise and recommended research activities. The tables cover the cate- gories listed in Table 1 of the text, and the entries in each table correspond, at least roughly, to the study topics. The information here;represents the news of skilled clinicians, some of whom specialize in the care of patients with cardiovascular disease, and experts in research and other fields relevant for effectiveness work. The workshop format did not allow for full explo- ration of all the issues raised by the AMI committee members, but they did believe that the breadth of topics concluded in these tables will provide guidance for a rich research agenda.
APPENDIX A TABLE A.2 Summary of Key Patient Management Topics Nominated by Acute Myocardial Infarction Committee in Round One of the Homework Exercise 57 A. Primary prevention and patient/public education strategies B. Out-of-hospital, prehospital, and emergency room treatment C. Use of thrombolytic therapies D. Use of invasive and surgical procedures E. Invasive/surgical versus noninvasive or drug options P. Use of other technologies G. Risk stratification for post-AMI management H. Other outcome research topics I. Rehabilitation TABLE A.3 Summary of Research Activities Recommended for Primary Prevention, Patien9Public Education, Physician Education, and General Management in Acute Myocardial Infarction PRIMARY PREVENTION AND PATENT/PUBLIC EDUCATION STRATEGIES I. Hypertension 1. Test usefulness of treating moderate hypertension with drug therapy in low-risk patients II. Patient education 1. Develop tools to educate patients about lifestyle and prognosis following AMI 2. Develop strategies to modify risk factors (dietary and blood pressure control) for indi- viduals win family history as a risk factor and encourage compliance III. Public education 1. Develop methods and models (e.g., use of media; strengthen 911 system; town meet- ings) to inform patients and public about risk factors, etc., and to reinforce behavior toward adherence to recommended practices and guidelines 2. Test mechanisms for decreasing delay in diagnosis and treatment of AMI (and reducing mortality), with specific attention to public education of early warning signs versus increasing the number or changing the location of health care facilities PHYSICIAN EDUCATION AND GENERAL MANAGEMENT ISSUES I. Priory care practitioners 1. Examine role of primary care provider (versus cardiologist) in routine management of cardiac conditions (including management and referral decision making) II. Decision trees 1. Develop decision trees or other aids for physicians to manage coronary/cardiac condi- tions in general and AMI in particular III. Physician education 1. Implement physician education programs centered on "best" uses of current treatments, especially those that may prevent complications of AMI
58 APPENDIX A TABLE A.4 Summary of the Research Activities Recommended for Out- of-Hospital, Prehospital, and Emergency Room Treatment, and for Rehabilitation in Acute Myocardial Infarction OUT-OF-HOSPITAL, PREHOSPITAL, AND EMERGENCY ROOM TREATMENT I. Helicopter transport 1. Determine the relationship of helicopter transport systems (especially for rural areas) to outcome II. Reorganization of emergency room 1. Test the reorganization of the emergency room for more efficient triage (diagnosis and treatment) of potential AMI patients (i.e., chest pain) III. Early intervention 1. Test systems for very early intervention versus triage of patients to tertiary care facili- ties, including benefits of prehospital electrocardiography and diagnosis and the value of initiating very early therapy in the home through emergency medical services (EMS) sys- tems REHABILITATION I. Rehabilitation services 1. Determine and evaluate range of rehabilitation and related services used to return car- diac patients to optimal quality of life and functional status II. Formal cardiac rehabilitation programs 1. Determine effectiveness of fonnal cardiac rehabilitation programs in terms of func- tional status, q ,rlity of life, return to work, etc., for asymptomatic and complicated post- AMI patients TABLE A.5 Summary of Research Activities Recommended for Use of Therapies and Procedures in Acute Myocardial l~farction USE OF THROMBOLY1IC TIERAPES I. Optimal timing, criteria, and variations in use 1. Determine the optimal timing and route for administration of thrombolytic therapies 2. Determine whether current criteria for use of intravenous thrombolytic agents can be extended to include more than 75 percent of AMI patients 3. Determine the variations in practice patterns of use or nonuse of thrombolytic therapy, learn whether patients receiving thrombolytic therapy use fewer resources or have better outcomes II. Comparative effectiveness and cost~ffectiveness 1. Determine the comparative effectiveness of thrombolytic therapy and other pharmaco- logic interventions (e.g., calcium channel blockers, beta blockers, prophylactic lidocaine) in pert-infarction period 2. Determine the cost-effectiveness of thrombolytic therapy in general, of aggressive thrombolytic therapy, and of the use of tPA versus less expensive agents such as urokinase and streptokinase
APPENDIX A TABLE A.5 (continued) 59 III. New thrombolytic agents 1. Study whether and how new thrombolytic agents influence patient management deci- sions and outcomes USE OF CATHETERIZATION I. Related to admission 1. Exannne the role of catheterization during and after an admission for AMI II. Optimalcatheterization rate 1. Define/detennine the optimal rate of catheterization in the year following uncompli- cated AMI (optimal being defined as the rate that will identify patients who will benefit from angioplasty or coronary artery bypass graft without incurring excessive cost or risk) USE OP SURGICAL INTERVENTION I. Outcomes 1. Determine the outcomes (mortality; morbidity; quality of life) of angioplasty and/or CABG in post-AMI patients in general, in post-AMI patients who have no symptoms and a normal exercise ECG/thallium scan, and as a function of age II. One vessel and more than one vessel disease 1. Determine whether angioplasty lowers the risk of AMI in patients with more than one vessel disease III. Alternative regimens 1. Examine alternative diagnostic and treatment regimens for cost-efficiency and cost- effectiveness, with attention specifically to multiple angioplasty versus surgery following one angioplasty 2. Determine the degree of small-area variation in repeat CABG, over time IV.Du~ing hospitalization 1. Examine the role of angioplasty (specifically percutaneous transluminal coronary angioplasty (PTCA)) during hospitalization for AMI USE OP INVASIVE/SURGICAL VERSUS NONINVASIVE OR DRUG OPTIONS I. Early treatment alternatives 1. Compare the diffusion and then the effectiveness of streptokinase, tPA, and PTCA in early treatment of AMI II. Primary angioplasty in different patient groups 1. Evaluate the value of primary angioplasty in three groups: those who would otherwise be candidates for thrombolytic therapy, those who are not candidates for thrombolytic therapy, and those who would otherwise receive no treatment whatsoever III. Immediate invasive/surgical treatment 1. Determine whether immediate invasive post-AMI evaluation and consequent therapy ~CA, or CABG, or both) confers a significant benefit or improvement in prognosis as compare with ``routine" thrombolytic care for all padents and for only high-risk patients
60 TABLE A.5 (continued) APPENDIX A IV. Suppression of post-AMI arrhythmias 1. Determine whether invasive (electrophysiologic) studies or noninvasive (treadmill; Halter) analyses provide more effective ways to suppress post-AMI arrhythmias in high- risk patients USE OF CORONARY CARE UNITS AND OTHER TECHNOLOGIES I. Automatic devices 1. Test whether use of automatic nonimplantable defibrillator and heart rate and blood pressure automatic system reduces CCU costs by allowing movement of CCU patients to regular beds 2. Determine what types of patients require, or do not require, CCU care when the main purpose of the CCU is to detect life-threatening arrhythmias TABLE A.6 Summary of Research Activities Recommended for Risk Stratification of Patients After Acute Myocardial Infarction I. Risk stratification procedures 1. Identify current procedures and examine their relation to modalities of care and out- come 2. Test use of "presence and severity of silent ST segment depression" as an element of stratification 3. Determine the effects on patient management and decision making II. Influence of risk stratification testing on outcomes 1. Determine whether and how data from tests used for stratification (e.g., echocardiogra- phy studies and ambulatory monitoring) influence patient management and outcomes 2. Determine what easily collected outcomes data could be added to inpatient records to improve prognostication III.Post-AMI testing and detection of high-risk patients 1. Examine the prognostic accuracy of post-AMI testing in various patient subgroups 2. Determine an appropriate strategy for detection of high-risk patients to implement before discharge IV. Standard prognostic battery and "optimal" stratification 1. Establish a standard prognostic battery and define an "optimal" risk stratification strat- egy for elderly patients with AMI V. Cost savings of early risk stratification 1. Test early risk stratification as a means of cost savings by distinguishing high-nsk patients needing PTCA or CABG from those warranting early discharge
APPENDIX A 61 TABLE A.7 Summary of Research Activities Recommended for Other Outcomes Topics Mentioned for Acute Myocardial infarction I. Health status outcomes of AMI 1. Determine the health status outcomes of AMI including disability, functional recov- ery, and quality of life for elderly AMI patients in general and for dose treated by differ- ent modalities, under different payment systems, in different facilities, and with different sociodemographic characteristics II. Outcomes and influence of age 1. Examine outcomes versus cost of resuscitation as a function of age 2. Determine influence of age on outcomes and as a determinant of treatment strategy III.Predictors of poor recovery 1. Determine predictors of poor psychosocial recovery or cardiac invalidism
