MEDICOLEGAL DEATH INVESTIGATION, PUBLIC HEALTH, AND HEALTH CARE
THE USE OF MEDICAL EXAMINER AND CORONER DATA FOR PUBLIC HEALTH SURVEILLANCE
Medical examiner and coroner (ME/C) data hold great potential for public health surveillance and, ultimately, public health intervention. But barriers stand in the way of adopting a national surveillance system that uses common data elements from ME/C offices. Understanding the value of ME/C data for public health and surveillance and the barriers to developing a national system might be appropriate for an IOM study.
ME/C data have a proven ability to detect clusters and unusual deaths. They can be probed more deeply by using the detailed information collected during a death investigation. The data might be used to discern risk factors that are key to developing preventive interventions. If tissue is banked, it can be analyzed to characterize the natural history of a new and emerging illness, such as those caused by hantavirus or HIV. Finally, ME/C data can yield timely and specific information about an unfolding epidemic. The data provide considerable potentialin real timefor addressing terrorism and bioterrorism.
The value of a nationwide ME/C surveillance system was dramatically illustrated with respect to a previously unknown and preventable public health problem: children’s deaths from entrapment in car trunks. Over a 2- month period in 1998, 11 child deaths were identified through death investigations in three states: New Mexico, Utah, and Pennsylvania. The cluster of cases triggered a broader investigation, which uncovered a total of 21 deaths nationwide from 1987 to 1999. That investigation led to identification of the major risk factor: children’s inability to get out of a car trunk. An expert panel was convened by the National Highway Traffic Safety Administration (NHTSA), and it recommended an interior trunk release. NHTSA responded by setting a new safety standard for automobile manufacturers in September 2001. That example illustrates that over a 3-year period surveillance can be actively translated into a new public health intervention, once a risk factor is found. If the case cluster had not been identified by pooling data from many jurisdictions, an opportunity to prevent deaths of children would have been lost.
There are major impediments to the development of a national surveillance system with ME/C data: variability in data quality with respect to training and experience, investigation procedures, and reporting requirements (for example, common nomenclature); variability in technology and standards; and the lack of policies for data interchange. To determine the extent of the child-entrapment problem, for example, CDC searched media databases because there was no standardized means of tallying the cases across all ME/C offices.
An IOM study could help to spearhead standards for death investigation. Standards are the key to reducing the high variability in data quality. They could cover death investigation practice and data collection, storage, and exchange of data. IOM could assess the feasibility of an integrated information system, including how long it would take to establish and what types of resources and steps would be necessary. IOM could also provide an independent and systematic assessment of the benefits and costs of a national system to improve all death investigations. IOM’s credibility comes from its expertise and its independence. For almost a century, there has been widespread agreement that the current system is inadequate. An IOM study could
help to elucidate understand the barriers to and build momentum for systemwide transformation.
Medical Examiner and Coroner Data for Public Health: A Model Linked System
ME/C data serve as the cornerstone for a model surveillance system in the injury field— the Violence Fatality Reporting System (VFRS). That system was launched as a result of a recommendation in the IOM 1999 report, Reducing the Burden of Injury: Advancing Prevention and Treatment (1999a). After being piloted in Wisconsin, the system has now expanded to cover six states.
The VFRS is a partnership that links ME/C data, law enforcement data (such as, crime reports), and crime-laboratory data, including those on ballistics testing from the Bureau of Alcohol, Tobacco, and Firearms. The system was modeled after NHTSA’s Fatal Analysis Reporting System, which has been in operation since the 1960s to study trends in circumstances of fatal motor-vehicle crashes.
Wisconsin’s role in the VFRS dates back to 1994, when the speaker (an emergency-room physician) approached the Milwaukee medical examiner. The first meeting spawned a partnership among agencies. The system now contains nearly 20 data elements on the victim, the environment, and the agent (such as, a firearm). Data elements from the ME/C office include victim demographics, cause of death, alcohol and drug use, wound information, and circumstances of death. This linked dataset is a treasure trove for understanding trends and risk factors associated with firearm homicides and suicides and for evaluating the impact of different prevention strategies in curbing firearm deaths.
Quality of Death Certificate Data
Data from death certificates have well-recognized strengths and weaknesses. Their strengths are universal coverage, uniformity in content and format, and standardization in coding, processing, and
data presentation. Their weaknesses are of two types: classification and coding errors and certification errors. The former can be addressed by the National Center for Health Statistics (NCHS), which is the CDC center that compiles death-certificate data submitted by each state’s vital-statistics office. The most important classification and coding errors pertain to the Word Health Organization’s International Classification of Diseases (ICD) (now in its 10th revision) which sets criteria for selecting the underlying cause of death. Errors in ICD coding are handled by the Mortality Reference Group, an international committee that deals with proposals for changes in ICD mortality coding. Its recommendations are translated to the field through annual updates to the NCHS automated coding systems.
Certification errors cannot be readily addressed by NCHS. The foremost errors are inclusion of uninformative information on modes of death. For example, cardiac arrest is mentioned in 0.6% of deaths, but this is uninformative. Although NCHS discourages its use, the rate has not dropped over the last decade; this is one indication of continuing quality problems with death-certificate data. Other types of certification errors are an implausible sequence in the listed causes of death, late or unfiled amendments to original death certificates, and incorrect diagnoses.
Various studies have attempted to determine the frequency of incorrect diagnoses. Their findings have been highly variable, depending on the disease studied, and on the studies methodology, geographic coverage, and results. The problem is that there is no gold standard, such as, the actual cause of death with which to compare a reported cause of death. In the absence of a gold standard, studies can assess the reliability of death-certificate data by comparing them with autopsy records or medical records. Such studies have shown greater reliability for some causes of deathcancer and external causesthan for others, such as digestive diseases. Another way to assess the quality of data is to monitor trends in unknown causes of death and ill-defined conditions. Increasing rates would indicate a growing problem in data quality.
NCHS is interested in investigating whether the type of certifier—physician, medical examiner, coroner, or, in some jurisdictions a nurse practitioner—affects the reliability of the coded data. That kind of analysis is not yet possible, because NCHS does not
collect identifying information on the certifier. But it will be possible with the introduction of a new standard death certificate. The new certificates are already being used in some states in 2003, but they will not cover all US states and territories until 2006. NCHS will be working to improve the quality of the new certificates through better and clearer instructions, physician education and training, more timely filing of amendments, electronic death registration, and querying the states about cause of death. When a physician or other certifier writes "cardiac arrest" on a death certificate, for example, NCHS will electronically prompt the provision of more information or reexamine what has been coded.
Medical Examiner and Coroner Information Systems: Making the Case for Standardization & Integration
In 1986, CDC created its Medical Examiner and Coroner Information Sharing Program. The goal of the program has been to facilitate the utility and availability of ME/C data. Among other activities, the program published in 2000 an assessment of the ME/C information systems in six US jurisdictions, including two states (New Mexico and North Carolina). The assessment (unpublished) was designed to examine the relationship between death investigation and the information-collection activities used for surveillance.
The assessment found common processes and information content but wide variability in the vocabulary used to capture the same content and in the expertise and resources devoted to information technology. ME/C offices were found to have little capacity for information management and dissemination, despite the fact that at least 16 other agencies used their information. The "customers" of the data spanned local and state agencies, and national data systems of the Department of Justice, the Department of Health and Human Services (DHHS), the Department of Transportation, the Department of Labor, and the Consumer Product Safety Commission. In DHHS, five surveillance systems relied on ME/C data. ME/C jurisdictions used highly variable information systems, ranging from paper-based systems and combinations of paper and computer spreadsheets to
comprehensive, relational databases that were effective at capturing information.
Because of the lack of integration among ME/C systems, CDC has no capacity to act in the public interest in problems that fall outside the borders of current data systems. Existing systems, for example, missed childhood-entrapment deaths highlighted by Dr. Sosin. To determine the extent of the entrapment problem, CDC resorted to media database searches instead of the ME/C systems.
Wisconsin undertook a systematic approach to enhancing the utility of its ME/C data. Its Department of Public Health and the state’s ME/C association collaborated to determine the needs for data, software, an integrated data repository and data interchange. The assessment identified about 150 data elements that, if collected uniformly, would meet the multiple reporting requirements of its customers. In 2001, it also determined that fewer than two-thirds of medical examiner and coroner offices had dedicated office space. The collaborators leveraged funding from DHHS to develop an integrated data repository covering all 150 data elements.
Overall, the findings from the CDC program and Wisconsin suggest the importance of systematic activities to define data requirements, in recognition that there are different responsibilities across law enforcement, public safety, and public health; to standardize data quality through death-investigation methods and data standards; and to integrate and coordinate data through information management, data interchange, and information resources. The best and most efficient approach is to leverage money from different agencies that track many of the same data elements rather than to proceed health outcome by health outcome. The need for homeland security has generated overlapping concerns in many federal agencies and created incentives for states to standardize and coordinate their data-collection activities.
The Role of the Medicolegal Autopsy in Health Care
Autopsies are vitally important to medicine, especially as a tool for quality assurance— as a means of exposing medical errors and promoting quality and trust in medicine.
Since the 1960s, rates of nonforensic autopsies have dropped precipitously. From their peak in 1965when they were performed in 50% of hospital deathsautopsy rates declined by 1995 to less than 9%. (Lundberg, 1984). The downward trend came in spite of an extensive body of evidence that revealed a high rate of errors in clinical diagnosis uncovered by autopsy. Comparing antemortem clinical diagnoses with postmortem findings, researchers over several decades consistently documented 40-60% discordance (Agency for Healthcare Research and Quality, 2002). A substantial part of the discordance stems from cases of undiagnosed causes of death, two-thirds of them treatable (Nichols et al., 1998).
An autopsy study in the 1970s of 100 consecutive deaths found a 49% rate of error. In 1983, JAMA devoted a theme issue to the problem and called on the medical community to perform more autopsies as a service to the community and as teaching tools for medical students and residents (Lundberg, 1983).
Numerous social, economic, and legal factors are behind the declining trend in autopsies. The foremost are physician’s discomfort in facing a family that has lost a member under their care, hospital administrators' fear of liability, lack of reimbursement except for Medicare patients, and public ignorance about the value of autopsies and the availability of reimbursement under Medicare (Lundberg, 1984). Every Medicare patient is entitled to a prepaid autopsy, as long as the hospital accepts Medicare patients. The overwhelming majority of hospital administrators are either unfamiliar with the policy or disbelieve it, partly because the Centers for Medicaid & Medicare Services (formerly the Health Care Financing Administration) fails to publicize the policy.
Many attempts have been made to combat the decline in autopsies and to raise awareness about their benefits (Hanzlick and
Combs, 1998). The Joint Commission for the Accreditation of Health Care Organizations in 1983 set criteria for autopsies and underscored their value for quality assurance. But there are no policies on autopsies in organizations representing nursing homes, home health providers, and hospices. Furthermore, the organizations fail to collect data on death rates or autopsy rates. Each year, 3 million Americans are admitted to nursing homes each year and 600,000 die. Most are not autopsied, largely because the most frequent payer, Medicaid, does not reimburse for autopsies. About 4.5 million people enter home care, and 1.5 million enter hospices, but their death rates are completely unknown. Because the vast majority do not have third-party coverage, the only recourse for families is to pay the $2000-6000 charge for private autopsies plus the cost of transportation to one of the nation’s few private autopsy facilities.
Recent grounds for optimism have emerged from publication of an extensive meta-analysis by the Agency for Healthcare Research and Quality. The analysis found that the likelihood that an autopsy would reveal a missed diagnosis that affected the outcome was 10.2%. And the probability that an autopsy would detect a major error in a given case was about 26%. (Agency for Healthcare Research and Quality, 2002). The analysis concluded that autopsies have two key benefits for the health-care system: better quality of care and better quality of data (such as, vital statistics and epidemiology) for health policy.
Finally, health-care organizations representing hospitals and nursing homes appear to be recognizing that autopsies afford them legal protections. In the past, those organizations failed to support autopsies, because of concern that results would give ammunition to families in lawsuits against them; but they are now beginning to recognize that autopsies may offer a shield to protect them against liability claims.
Medicolegal Death Investigation and Medical Malpractice and Other Civil Suits
Medical examiners often find themselves ill-equipped for malpractice cases and other civil suits. The cases are different from
criminal cases with respect to what is at stake and the role of the medical examiner. In criminal cases, the stakes are extremely high, considering that a defendant faces a potentially long confinement or a judicial execution. In civil cases, money is at the heart of the issuemonetary damages for the plaintiff and contingency fees for the plaintiff's attorney. The pursuit of monetary damages as the driving force in civil cases leads to an adversarial process in which the medical examiner is likely to be vigorously challenged in ways that transcend his or her medical training.
The challenge often takes place outside the courtroom in depositions. Depositions are rare in criminal cases but highly common in civil cases, and they can be used in lieu of courtroom testimony. Attorneys often badger the medical examiner and ask aggressive questions that are rarely asked in front of a jury, because the attorney would lose the jury’s sympathies. Medical examiners are thus highly uncomfortable at the prospect of a deposition.
Common types of civil cases involve torts (such as, malpractice), contract disputes (such as, life-insurance claims), worker compensation, and equity actions (such as, requests for exhumations). Malpractice actions are the most common torts involving a hospital pathologist. They are negligence cases in which a plaintiff claims that a clinician has failed to conform to a standard of care. A negligence action has four key elements: duty to the plaintiff (which is usually assumed to be the case in a doctor-patient relationship), breach of that duty, damages, and a causal connection between the breach and the damages.
The question of a causal connection is most often addressed by the hospital pathologist and it is the most difficult to answer. Often, there are several disease processes, and it is difficult to determine the sequence and the consequences. A common question posed to the pathologist is whether a patient's death resulted from an error or a complication of therapy or diagnosis. The 1999 IOM report To Err Is Human highlighted the problem of 44,000-98,000 medical errors made each year (1999b). Identifying those errors in the context of a court case places the hospital pathologist in the position of feeling caught in the middlebetween the family, which is distrustful of the treating physician and the hospital, which is wary and uncooperative. The determination of errors is commonly made by hospital pathologists
because medical examiners often are not statutorily required to handle such cases. None of the parties is eager to pay for the services of the pathologist. The best guideline for attributing causation is known as the “Oh, My God!” rule. If those are the first words of a physician’s reaction to what happened in a malpractice case (for example, hooking up an anesthesia machine backwards), the causation is likely to be blatant.
Apart from questions of causation, commonly encountered issues are the contribution of natural disease, mental competence, pain and suffering, workers compensation, insurance issues surrounding motor-vehicle fatalities, and manner of death (such as, accident or suicide). Because the outcome of the case often hinges on how law and medicine intersect, the medical examiner, more than the hospital pathologist, is best suited to handle such a case, but the medical examiner often is not given sufficient resources, staffing, and training to play an effective role.
Medical errors increasingly fall under the purview of the medical examiner as a result of the forces working against hospital pathology, especially the broader cultural issue of death denial. The ME/C system is the last bastion of people who know how to do autopsies and know how to do them well. In England, coroners routinely have the option of coding 13 manners of death, one of which is therapeutic misadventure. In the United States, there are only five manners of death, and therapeutic misadventure is not one of them (George Lundberg). Medical examiners should play a greater role in malpractice cases as long as they are given the tools, the training, and the budget (Garry Peterson). Hospital pathologists are not allowed to charge families for autopsies of Medicare patients (George Lundberg).
Value of Death Certificates
Given its wide-ranging uses, ME/C data should be comprehensive enough to answer the full spectrum of data needs, and
it is in everyone's best interest to leverage resources to enable collection of information (Lois Fingerhut, Dan Sosin, Sam Groseclose, Kathleen Toomey).
Physicians are often unaware of the widespread use and value of death certificate data (Jonathan Arden, Lois Fingerhut). If their awareness were increased, they would have greater motivation to code certificates accurately; greater and more sustained training of medical students and residents in filling out death certificates correctly is needed (Robert Anderson, Kathleen Toomey, Steve Hargarten). Working through the Federation of State Medical Boards to mandate more training might be considered (Alan Trachtenberg, Garry Peterson).
The new death certificates issued by NCHS will reinstate coding to indicate whether an autopsy has been performed and whether it was used for the death investigation. Those data elements will be used to monitor trends in autopsy rates. The questions had been deleted in 1995, so there has been a data gap of about 8 years (Robert Anderson, Kurt Nolte).
Despite the good intentions behind nationally based datasets, it is unrealistic to expect coroners and medical examiners to obtain all the detailed information that is needed, because of limited resources (Vincent Di Maio). The first priority of coroners and medical examiners is to serve their jurisdiction, not to do research.
The quality of toxicologic testing, except for alcohol testing, is highly questionable in most jurisdictions (Vincent Di Maio).
Many rural jurisdictions do not perform any toxicologic testing during a death investigation, because they lack resources; that means that national datasets that rely on ME/C toxicologic data are skewed to urban deaths and so are not necessarily representative of the national picture (Steve Hargarten). Also needed is the actual concentration of the toxicant (such as alcohol), not just its presence or absence (Alan Trachtenberg).