Anthony Nespole
Medical Director
CALIFORNIA MEDICAL REVIEW, INC.
INTRODUCTION
Thank you Mr. Chairman and Members of the Committee for this opportunity to state CMRI's views of important issues raised by the standardization of health care data. My name is Anthony Nespole and I am the Medical Director for California Medical Review, Inc..
CMRI is a private, not-for-profit, physician membership organization, and has been under contract with the federal Health Care Financing Administration (HCFA) to perform the Medicare Peer Review Organization (PRO) work in California since the beginning of the PRO program in 1984. CMRI also provides health care review, analysis, and intervention strategies for other government entities, university and hospital research projects, and managed care organizations.
While the devil is always in the details, yet to be written, CMRI believes Subtitle F of the Health Insurance Portability and Accountability Act of 1996 (HIPAA) is a long overdue, fundamental prerequisite to understanding the delivery of health care services in our complex system. The lack of standardized, validated, comparable health care data is a public policy failure that has cost this nation dearly in both treasure and suffering. CMRI is hopeful that Subtitle F will launch a national, uniform system of health care data that will make reliable comparisons of patients, practitioners, providers, and plans a routine dimension of the system.
Focus of CMRI's Comments
Given the complexity of the questions raised by the Committee and our brief time together today, my comments will focus on two crucial dimensions of standardizing the data: Πimproving the richness and the reliability of the data available, and the confidentiality of sensitive information.
CMRI sees a need to expand somewhat the list of mandatory data elements abstracted from the medical record. From a public health perspective, it is important to collect accurate patient data on race and ethnicity. Such data will enhance understanding of the health care needs of particular communities, a big issue in a state as diverse as California. While the problems inherent in schemes to catagorize such data are daunting, the ability to identify patients by race and ethnicity on the UB-92 will improve the health care system. CMRI urges the Committee to recommend collection of this information.
"Garbage in, garbage out"
The cornerstone of any national data system is the accuracy, validity and reliability of the data. All of us know the saying: "garbage in, garbage out." There is no reason to have a system if the data are not accurate, valid, and reliable. The need for routine checks on the validity of Medicare data is demonstrated by the results of research conducted by CMRI on the reliability of Medicare Part A claims data.
In 1994 CMRI concluded a data reliability study to determine the degree of agreement between MEDPRO data and the medical records from which it originally came. Before sharing specific details about the study itself, it is useful to understand the basic process by which Medicare data are turned into MEDPRO data, the data HCFA sends to the PROs.
The HCFA data trail shows the basic steps Medicare claims data go through on the way to becoming MEDPRO data for use by the PROs. The process begins with the data being abstracted manually by hospitals from the medical record to the UB-92 claim form. The hospital forwards the UB-92 form to the Medicare fiscal intermediary (FI) for further processing. The FI then forwards the claims information to HCFA for inclusion in the Common Working File (CWF). From there the information is processed for inclusion in the National Claims History data base from which the MEDPAR file (Medicare Provider Analysis and Review file) is created. Finally, the MEDPAR file is augmented with beneficiary identifiers and death information to become the MEDPRO file.
Recognizing that this process generates opportunities for human errors, CMRI sought to quantify the degree of concordance between the MEDPRO file and the medical records from which it came. A random sample of 267 records was selected for comparison against the MEDPRO data based on those records. CMRI developed a medical record abstraction tool containing 19 categories of information with 48 data variables, including: length of stay, types of service, and the number of procedures. The medical records were abstracted by experienced personnel, trained in how to use the abstracting tool.
After chart abstraction, the data were entered into a PC database for analysis. As a check on data abstraction, each completed abstraction form was checked against the corresponding medical record by a physician or a review nurse. As a check on data entry, the abstraction forms were entered twice by two different people.
The results of the study show varying and potentially important difference between the MEDPRO file and the data in the medical records. Overall, two-thirds of the medical records had two or more differences with the MEDPRO file. Nearly 40 percent had three or more differences with the MEDPRO file. The highest number of errors was for type (emergency, elective, newborn, etc.) of admission (21 percent of the records) and source (emergency room, transfer, HMO, etc.) of admission (17 percent of records). Errors for discharge destination, and indicators for intensive care, coronary care, blood transfusion and number of blood units ranged from 7-11 percent of the medical records.
Work with the data shows that some of the errors in the MEDPRO file can be overcome and others will compromise the validity and reliability of study findings. MEDPRO users must be cautious in the use of variables for which rates of difference are relatively high. Use of such variables as proxies for other measures further diminish their reliability. For example, there is a 10 percent chance of error when using coronary and intensive care indicators to measure coronary and intensive care units (CCU and ICU) utilization. As a measure of severity of illness, the 10 percent error is compounded by the imprecision of the measurement by proxy. The severity of illness measure is further complicated if source and type of admission variables are included in the proxy. CMRI's study demonstrated that the use of MEDPRO CCU and ICU data as a proxy for severity of illness is tenuous at best.
The study concluded that careful consideration must be given to selecting the specific variable to be used in the analysis of MEDPRO data in order to maximize the confidence level in the reliability of detected patterns of care, and in the interpretation of outcomes following interventions in the processes of care.
CMRI's data reliability study demonstrates that routine checks of the validity of the data are vital to the integrity of the system. There is no doubt similar problems plague non-Medicare systems too. CMRI believes that routine, unannounced audits of health care data practices and procedures are a vital part of establishing a national health care data system. There is gold in the data only if it is valid and reliable data.
The most sensitive dimension to developing the health care information superhighway is the privacy of health care information. Strong privacy protections are vital to achieving public support for a more comprehensive health care information system. Moreover, such protections must be in place before computerized patient specific information is available for analysis.
The complexity of modern health care, the increase in the number of practitioners needing access to the medical record, and the expanding use of clinical data all point to the need for strong patient privacy protections. According to a survey reported recently in the newsletter Medicine and Health, however, the public has little confidence that the privacy of patient specific information will be maintained in a reformed health care system. Nevertheless, there is general recognition that patient specific health care information is highly sensitive and that improper disclosure could be harmful . . .
The realization of administrative simplification will require new ways of maintaining the confidentiality of private health care information. The existing state and federal laws protecting patient privacy have major weaknesses. The most common weakness is that privacy protections are aimed at controlling the holder of the information and the type of information held, rather than the use of the information to be protected. Such laws do not protect how the information will be handled, thus leaving patients unprotected in a variety of common situations.
Often overlooked in consideration of privacy and confidentiality issues are the rights of the physician whose records are the source of health care data. Physicians have a legitimate interest in knowing how their records will be used and in protections against the inappropriate use of data. These concerns go beyond the problems created by sophisticated hackers "surfing" through medical databases, although such concerns are well founded. Physicians' concerns about the confidentiality of medical records goes to issues like malpractice and "economic credentialling."
PROs have first-hand experience with physicians' fears that data may cause them malpractice problems. Medicare program rules give physicians a veto over whether the PRO can report findings from reviews initiated by beneficiary complaints. In California, nearly all physicians refuse CMRI permission to release review findings to Medicare beneficiaries whether the findings are adverse to the physician or not. Physicians fear that the PRO's review findings will be the basis for a malpractice claim. While physicians' fear of Medicare patients becoming malpractice claimants is not supported by fact, it is nonetheless a major source anxiety for physicians.
Similarly, physicians are concerned about their data being used for "economic credentialling," where plans select and de-select participating physicians based on how expensive their practice style is. Physicians fear that such decisions will be based on incomplete information and without any opportunity for the aggrieved physician to challenge the decision or the process by which it is made. If not addressed by the system, physicians fear of such practices could taint the medical record. CMRI urges the Committee to focus on these fears and provide due process for physicians before such decisions are final.
Finding the balance between the legitimate use of confidential data and the privacy rights of patients, practitioners, and providers will require new rights of action for individuals aggrieved by improper disclosures and a tough prosecutorial attitude by government officials charged with protecting the integrity of the national health care data resource.
CMRI is optimistic that a balance can be struck and that patients' rights to privacy can be protected in an environment enriched by accurate clinical data. CMRI urges this distinguished Committee to be bold in your recommendations to secure a reliable, valid, uniform and comparable health care data system for our country.
Thank you again Mr. Chairman for this opportunity to testify on this important new law. I'll be pleased to answer any questions the Committee may have of me.