In the wake of a large-scale disaster, from the initial devastation through the long tail of recovery, protecting the health and well-being of the affected individuals and communities is paramount. Accurate and timely information about mortality and significant morbidity related to the disaster are critical to supporting situational awareness for the disaster management enterprise and driving public health action to save lives and prevent further health impacts. Conversely, failure to capture mortality and morbidity data accurately and consistently undercuts the nation’s capacity to protect its population. Information about disaster-related mortality and morbidity adds value at all phases of the disaster management cycle. As a disaster unfolds, these data are crucial in guiding response and recovery priorities, ensuring a common operating picture and real-time situational awareness across stakeholders, and protecting vulnerable populations and settings at heightened risk. Public health messaging shaped by accurate morbidity and mortality data is critical for communicating vital information to the public, protecting them from ongoing and future hazards, and prompting protective actions from policy makers and other leaders. During the interim between disasters, these data provide the foundation for evaluation, prevention, mitigation, and preparedness activities designed to reduce morbidity and mortality when future events arise. In some instances, these data might also inform actions that can even prevent future disasters, such as through prompting actions to mitigate the risk of climate-related disasters. In sum,
1 This Summary does not include references. Citations for the discussion presented in this Summary appear in subsequent chapters.
accurate assessment of mortality and morbidity from disasters is not merely an academic or historical exercise: the timely and accurate counting and attribution of deaths and morbidities can improve disaster response and lead to a more accurate assessment of the extent, types, and causes of morbidity and mortality in disasters and drive changes in policy, practice, and behavior that will prevent suffering and save lives.
Extracting the maximum value from these data depends on having standard practices and systems in place for collecting and reporting accurate information, analyzing it appropriately, and translating the data into action to protect disaster-affected communities. However, it is challenging to coordinate these efforts effectively and uniformly across the disaster management enterprise, which comprises a vast and intricate network of federal and state, local, tribal, and territorial (SLTT) systems as well as a plethora of stakeholders ranging from emergency management, health care, government agencies, the general public, policy makers, and the public and private sectors. The practices and systems currently in place are inadequately robust, coordinated, or reliable to leverage mortality and morbidity data to their fullest potential. Instead, these systems and stakeholders are often splintered, siloed, and unable to rapidly share information with each other. As a result, the same incident can generate dramatically different estimates of mortality and morbidity depending on the policies and practices of the home jurisdiction. Data that are not captured uniformly across jurisdictions, federal and SLTT agencies, and professions are less conducive to comparative analysis and may exclude valuable information about deaths and morbidities that are indirectly or partially attributable to a disaster. In other cases, available data about mortality and morbidity are squandered because they are not or cannot be acted on to add value to disaster management or because existing systems are not deriving the optimal value from the data to guide action. Additionally, the consistent and accurate collection of disaster-related mortality and morbidity data is often not prioritized because the stakeholders involved may not be aware of its importance. Multiple approaches exist to assess disaster-related mortality and morbidity. Also of critical importance is the reality that stakeholders may have certain vested interests to selectively use data to depict the estimated impacts of a disaster in a certain manner. Whether explicit or inadvertent, this can lead to controversy and confusion. Avoiding these problems requires improving processes for vital statistics, public health, emergency management and emergency response data systems; integrating these data sources with other sources of health data, including indicators of the social determinants of health (especially where morbidity data are concerned); and, perhaps most important, standardizing the way these data collection processes are implemented in SLTT agencies throughout the United States.
The entire disaster management enterprise would benefit from changes in practices and policies to strengthen the systems and structures involved in the consistent assessment of mortality and morbidity data during and following large-scale disasters. The increasing frequency and severity of disasters worldwide underscore the need for improvement and standardization of these systems and for stakeholders to be more flexible, coordinated, and adept to meet the core mission of protecting the health and well-being of people impacted by disasters and bolstering community resilience.
BACKGROUND AND RECOMMENDATIONS
While it is impossible to know definitively the full impact of a large-scale disaster on the people directly and indirectly affected, this report attempts to highlight how the administrative, organizational, logistical, and analytical components associated with each of the major estimation approaches can be improved to produce more accurate and complete reflections of the disaster’s true effect. As such, the committee’s recommendations for improving the function and value of the nation’s systems and practices for assessing mortality and morbidity following large-scale disasters span the organizational, operational, analytical, and professional domains. To structure the process of developing their recommendations, the committee established a set of guiding precepts that synthesize the ethos and key characteristics of their vision for a highly effective system (see Box S-1).
Organizational Leadership and an Enterprise Approach
Successfully implementing the recommendations set forth in this report will hinge on strong organizational leadership to foster mutual commitment across stakeholders to achieve a coordinated, enterprise-wide approach—aligning the programs and priorities of multiple stakeholders under a shared mission of improving the assessment of mortality and morbidity following large-scale disasters. The full spectrum of systemic changes needed to achieve this improvement across multiple—and often siloed—stakeholders and systems will not be realized unless these agencies, entities, and their leadership rise to meet this collective mission. Leadership at the federal and SLTT levels must buy-in as champions to drive lasting changes across multiple systems and overcome fragmentation. Throughout the disaster management cycle, cross-agency responsibilities will need to be adapted to meet this mission for domestic action.
No federal entity or standardized system is singularly responsible for coordination or oversight of the practices for collection, reporting, and dissemination of disaster-related mortality and morbidity data at the national level in the United States. Rather, these responsibilities are primarily the domain of states, territories, and even counties, creating a patchwork of data systems, standards, and processes. Furthermore, the systems and incentive structures currently in place are inadequate to facilitate collaboration among those entities. But in the face of a disaster, these entities must be nimble enough to flex and adapt beyond their day-to-day functions to contribute their unique tools and capacities to a shared enterprise. The leadership of federal agencies is critical in championing for and investing in the necessary systems improvements as well as setting national standards, but federal leadership will not be enough in view of the underlying structures of public health systems in the United States. SLTT entities will need to promote the value and facilitate the operationalization of these activities in tandem with federal efforts.
Uniform Philosophy for Attribution
Across professions and jurisdictions, multiple terms are used to denote a death or injury as related to a disaster and, if so denoted, the degree to which it is related. These different terms are often conflated, which can lead to misunderstandings and mischaracterization of the estimated impact of a disaster. It also undercuts the ability to appropriately compare assessments of disaster-related mortality and morbidity over time and across disasters. As discussed in Chapter 2, there is widespread variation in what is being assessed and the context in which it is being assessed. The current terminology and case definitions used to describe disaster-related mortality and morbidity do not adequately capture the differences in assessment methods used
or the totality and temporality of disaster-related deaths and morbidity. The lack of a uniform framework for assessing disaster-related health impacts undermines the quality and usability of these data to inform disaster management, and it leads to confusion about or even misrepresentations of the impacts of disasters. The committee recommends the immediate adoption of a consistent framework, which includes approaches for assessment and case definitions, as well as a comprehensive understanding of how these approaches and definitions should be applied to the assessment of mortality and morbidity following large-scale disasters (see Recommendations 2-1 and 2-2).
Framework for Assessing Disaster-Related Mortalities and Morbidities
To contribute to improving the quality and utility of these data, the committee developed a framework to serve as an initial guide for (1) describing the two primary approaches for developing quantitative indicators of total mortality or morbidity and (2) parsing out individual case definitions that can be used to uniformly characterize the degree to which an individual death or morbidity is attributed to a disaster. First, quantitative indicators of an event’s impact can be generated using two basic approaches: counts of affected individuals or population estimates. One approach is not always better than the other and both produce estimates of a disaster’s impact at a distinct point of time, within a specific context, and based on particular assumptions. Each approach also encompasses a variety of methodologies and tools, each of which has its own unique timelines, challenges, and targets.
Estimating total mortality or morbidity by using counts of affected individuals uses data from record keeping systems such as death registration systems or case counts reported to public health agencies. While individual counting methods produce very precise estimates of mortality and morbidity (generally reported to the exact number), the accuracy of this approach is contingent on (1) a complete count of bodies or ill or injured individuals, and (2) accurate attribution of mortality or morbidity to the disaster using consistent case definitions. Because these preconditions are not always met, the committee recommends the adoption of the definitions provided in Box S-2 (see Recommendation 2-1 for the complete recommendation text) as a starting point.
The second basic approach to estimating mortality and morbidity is to use statistical methods, such as survey, sampling, or excess mortality methods. These methods typically generate more comprehensive data than the individual counting approach—because they inherently include both direct and indirect deaths, injuries, or illness—and produce results that are generalizable to the population level. They will typically also include confidence intervals to indicate the exact level of imprecision in the estimations.
Approaches for Assessment: Counts and Estimates
These two general approaches (individual counts from administrative records and population estimates based on statistical approaches) both provide essential information in the face of a disaster. Each approach has its own strengths, weaknesses, appropriate uses, and methodologies and each approach makes assumptions and is subject to bias. These approaches are similar in that they pertain to a time period and to a geographical area, and estimates from both can be refined over time. Estimates based on individual counts of deaths recorded in administrative systems are valuable for understanding the immediate impact of disasters, but their accuracy depends on the completeness with which individual cases are recorded and reported. The individual counting methods within this approach often fail to capture certain types of disaster-related deaths including individuals who would not have died but for the disaster (e.g., carbon monoxide poisoning from a poorly placed generator or a fatal heart attack during disaster cleanup). More about methods and considerations for conducting individual counts can be found in Chapter 3. Comparatively, population-based estimation methods are crucial for capturing a full understanding of the impacts of a disaster on health and mortality. These methods may incorporate less precise information on cause of death, and in some applications (e.g., estimates of “excess” deaths) cannot distinguish which individuals would have survived in the absence of the disaster from those who would have died during the period regardless. A review of methodologies and tools associated with population estimates can be found in Chapter 4. The complex nature of disasters and the multiple pathways through which they affect human health make it difficult to quantify the impact of a specific disaster on human health with precision. Together, however, the two approaches can
provide both timely information and a far more comprehensive picture of the health and mortality impacts of a disaster (see Recommendation 2-2).
Sometimes, the availability of different estimates for assessing mortality and morbidity can create confusion, and at worst, the impression that impact of the disaster is being minimized or exaggerated by certain stakeholders via the selective use of data. This was a major public concern following Hurricanes Katrina and Maria, and more recently during the coronavirus disease 2019 (COVID-19) pandemic. When this happens, it is important to realize that the variation in estimates may reflect different, but appropriate, methods and targets. For example: Does the count include deaths that are indirectly or partially attributed to the disaster? Over what time period and in what geographic area are deaths or morbidities being estimated?
However, while there are legitimate reasons for analysts to use different approaches and methods for assessing the impact of a disaster, a certain approach or method may be more appropriate than others in specific instances. This report describes the efforts needed to select the most appropriate approaches, methods, and tools according to the context.
Differences in estimates of mortality and morbidity often reflect the great variation in the public health and emergency response data systems
at the state and local levels in the United States. What is required for an individual death to count as being caused by a hurricane, for example, depends on the clarity and specificity of case definitions used within a jurisdiction and on the professional judgments of those making determinations. Whether an individual with respiratory illness is included in the COVID-19 case count depends on whether a test was available and reported. This variability provides an opportunity for intentional or inadvertent selective use of data to minimize or exaggerate the number of cases based on stakeholder interests. This report describes how this problem can be addressed by improving processes for vital statistics, public health, and emergency response data systems and standardizing the way they are implemented and the data are reported by SLTT agencies throughout the United States.
Operationalize Uniform Systems, Tools, and Practices for Assessing Individual-Level Mortality
In addition to widespread inconsistency in the practices and case definitions used to attribute deaths to a disaster, the quality and value of those data are threatened by variability in how deaths are registered and reported across the United States. Mortality data are currently collected and reported through a byzantine process that spans a disjointed network of federal and SLTT-level systems for death investigation and registration. No federal or nationally standardized system exists for death investigations; instead, these are carried out through a patchwork of different medicolegal death investigation systems that are governed at the SLTT level and thus vary by jurisdiction. For example, states may use a coroner system, medical examiner (ME) system, or a combination of both for certifying deaths—and there are notable differences in levels of training, appointment processes, and political standing for coroners and medical examiners, each of which can impact the quality of death investigations. In some states, these systems are centralized at the state level and in others these systems are decentralized at the county or district level. After a death is certified within the medicolegal death investigation system, it is registered at a state-level vital records office to create a death record. It is then shared with the National Center for Health Statistics (NCHS) via the National Vital Statistics System (NVSS). As described in Chapters 3 and 4, this heterogeneity across systems and lack of uniform standards for attributing mortality and morbidity to a disaster across different systems impedes meaningful analysis and undermines the value of these data in improving disaster management.
The committee identified multiple gaps in death investigation and registration systems that limit the accuracy and comparability of the individual-level data collected. For instance, different SLTT systems use incongruent definitions for what counts as a disaster-related death. Medical certifiers
of death play a crucial role on the front lines of this process, as they are responsible for assessing whether and the degree to which a death is attributable to the disaster and then inputting that information such that the appropriate information flows up through state-level death registration system and NCHS. Despite their essential role, training for medical certifiers is inconsistent and often inadequate. Because of the variation in the types of medicolegal death investigation systems that exist throughout the United States, the qualifications of these certifiers vary from being a registered voter with a high school diploma and free of a felony conviction in some coroner systems to forensic pathologists with 4–6 years of postgraduate medical training and board certification working in medical examiner systems as autopsy physicians and public health officers (see Chapter 3 for discussion of variation in the medicolegal death investigation system). Additionally, the current medicolegal death investigation and registration systems were not designed to capture information in a way that facilitates interoperability and timely data sharing within and across SLTT and federal levels. This adds to the administrative burden on medical examiners and coroners (ME/Cs) who are often required to enter data on a death into multiple systems. Electronic registration and coding of causes and contributors of death can streamline the process of death registration, but not all ME/C jurisdictions have electronic case management systems nor do all states have fully electronic death registration systems (see Recommendations 3-1 and 3-2).
The committee also identified opportunities to address these gaps through a series of immediate and future priorities for investment to promote the adoption of standard and uniform practices for collecting, recording, and reporting these data. For example, all jurisdictions would benefit from collecting a set of standard baseline data points about a death (e.g., geocoding of place of death and place of residence; practices for describing contributing factors of death). The NVSS is well positioned to drive many of the necessary systemic changes to improve vertical coordination across stakeholders and the interoperability of electronic systems between themselves, medical certifiers, and state vital records offices. Additional federal-level requirements could help ensure that these baseline data are collected uniformly and accurately by medical certifiers, so the information can be coded and entered into the national database.2 This will also require supporting those states without electronic death registration systems to put them in place.
2 This does not warrant changes to death certificates, which are legal documents governed at the state level.
Standards for Defining and Collecting Morbidity Data
Morbidity is a complex concept that covers a range of health outcomes from the physical to psychological, including cardiovascular, gastrointestinal, and respiratory diseases; physical injuries; and mental health outcomes. Collecting data about disaster-related morbidities presents its own set of challenges beyond those described for mortality data, given the broad definition of disaster-related morbidities, variation across disaster types, and logistical challenges associated with mining morbidity data from across broad network of unique federal, SLTT, and health care systems. Capturing disaster-related morbidity data is further complicated by several inextricably linked factors: for example, disasters can exacerbate pre-existing conditions and create additional health complications (e.g., the closing of dialysis centers in a disaster will create morbidity among people with kidney disease requiring dialysis); co-morbidities can predispose certain at-risk individuals to additional diseases in the post-disaster environment (e.g., individuals with diabetes are at additional risk for cardiac disease); and multiple social determinants of health (SDOH) can profoundly impact both mortality and morbidity in disasters. The appendixes to this report provide several case examples that explore these complications. Appendix D describes the impact of SDOH on disaster-related morbidity and mortality through two case examples. Appendix C provides a focused case study of the challenges in assessing mortality and morbidity during the early stages of the COVID-19 public health emergency. Both provide additional support for the report’s core recommendations.3
3 The Federal Emergency Management Agency, the study’s sponsor, provided permission for the inclusion of COVID-19 considerations in the report.
When acted on appropriately, morbidity data can help to reduce mortality (i.e., by preventing morbidities from becoming mortalities) and can be used to shape public health actions, including community health education, and medical preparedness (see Chapter 2 for additional uses of morbidity data). For end users in the field of disaster management, in particular, estimates of morbidity resulting from a disaster may actually be of more value than mortality data in informing life-saving recovery, mitigation, and preparedness activities; enhancing real-time response; and providing information on health care utilization. Therefore, exclusive focus on mortality data, the traditional outcome of interest, at the expense of morbidity data is tantamount to focusing only on the most severe cases.
Currently, there is no consensus on which morbidities or indicators should be consistently tracked in common types of disasters (e.g., floods, hurricanes, blizzards, tornadoes, pandemics/epidemics) to provide actionable information to end users such as public health and emergency planners, health care systems, and SLTT and federal agencies. As with all data, different types provide different opportunities for use. The collection of significant morbidity data, defined for the purposes of this report as serious morbidities with high risk of immediate hospitalization and death, could provide critical data on population needs regarding access to health care resources. However, exclusively focusing on significant morbidities would likely fail to capture population trends for less immediately critical but far more prevalent and costly morbidities related to the social determinants of health, such as respiratory impacts, mental health outcomes, and exacerbation of existing co-morbidities (see Appendix D). Defining what data to prioritize for collection is a critical first step to building the capability to collect and use these data. Therefore, investment in an ongoing process is needed to develop, validate, and disseminate national standards for data collection of key morbidities caused or exacerbated by specific types of disasters (see Recommendation 3-3).
Analytical Methods for Population Estimates
As indicated in the guiding precepts (see Box S-1), in assessing the impacts of disasters, the committee recognizes the importance of both counting individuals who die and estimating the morbidity and mortality impacts of disasters on the entire population at risk using statistical methods. Accurate counts of individual deaths can provide timely information on the scale of an event and is critical for assigning individual-level benefits, but individual counts rarely provide an adequate picture of a disaster’s impact on the mortality and morbidity of the population as a whole. For a number of reasons (described in detail in Chapter 3 and explored in regard to the COVID-19 case study in Appendix C), individual counting methods typically underestimate the total impact of a disaster on a population. Statistical estimation methods, by contrast, generate a more complete picture of the total impact of the disaster, but do not always yield an allocation of individual cases into mutually exclusive categories of death or illness resulting from the disaster versus from other causes. Survey and sampling methods and excess death methods are the primary methods encompassed by the population estimation approach. See Chapter 2 for a discussion of both individual count and population estimate approaches as they relate to the value and use of mortality and morbidity data and Chapter 4 for a discussion of population-level methods of estimation.
Given the variation in ways for attributing the cause of any death and morbidity, there can be more than one appropriate approach to answering the question: “How many total deaths and significant morbidities were caused by this disaster?” However, while there is no single standard method for generating mortality or morbidity estimates using population estimation, methodological best practices can be specified. A national research program is vital to further develop and validate these best practices and to indicate appropriate circumstances for the use of specific methods (see Recommendation 4-1). As in all areas of research, these practices should be characterized by clarity in the specification of study objectives and definition of terms, transparency in the statement of assumptions and the sourcing of data used in the study, and great caution in advancing any particular measure or method as a most perfect solution.
Developing an effective data and information structure for studying disaster impacts on mortality and morbidity should be a cornerstone of the nation’s operational disaster response function. Because the analytical
sophistication and high-quality fieldwork necessary to develop and implement data infrastructure nationally is generally beyond the capabilities and time availability of most SLTT health departments, the responsibility of building and sustaining the capacity of the nation’s existing research and survey infrastructure to support the collection of survey data on the health effects of disasters rests at the federal level (see Recommendations 4-1 and 4-2).
Access to and Use of Mortality and Morbidity Data
To maximize value, data on disaster-related mortality and morbidity should be available to all stakeholders and be provided in a manner that is actionable for informing response, recovery, and mitigation activities at all stages of the disaster management cycle. Ensuring equitable access to timely and actionable data is of particular importance in disaster-impacted localities that may have limited capacity to carry out their own data collection and analysis. Strategies to broaden access and improve the utility of these data are described in Chapters 3 and 4. For example, federal agencies (e.g., NCHS, Federal Emergency Management Agency, Office of the Assistant Secretary for Preparedness and Response, CDC) have the responsibility
to actively support partners at the SLTT levels by providing them with actionable, location-specific information from the federal database, including baseline comparison mortality data and up-to-date data on disaster deaths (see Recommendation 4-3). Access to data does not equate to the actionability of data. For data to be actionable, the stakeholder requires the tools, capacity, and expertise to analyze these data and act appropriately. Therefore, investment in the development and deployment of ready-to-use tools that are tailored for use by SLTT stakeholders is critical to ensuring action (see Recommendation 4-3). Leveraging federal data and expertise to address limitations of federal mortality and morbidity data, such as routine mortality and morbidity surveillance, continuous quality improvement of data collection and integration systems, and retrospective geocoding of death registry entries in certain disaster-impacted areas would also enhance the accuracy and actionability of these data (see Recommendation 4-3).
Professional Training and Support
A broad range of stakeholders across professions and jurisdictions is involved in capturing, recording, analyzing, and using disaster-related mortality and morbidity data. Major barriers identified by the committee include the ongoing siloing of death investigation and registration systems from the disaster management enterprise, disparate levels of professional training and expertise across medical certifiers, and poor continuing education regarding attribution and recording of disaster-related deaths. Cumulatively, these shortcomings hinder the ability of the nation to procure accurate and complete individual counts of reported deaths following a disaster. Supporting medical certifiers will require federal resources for ongoing guidance,
education, and training to enable their accurate input of data into death records and to appreciate the importance of their part in the enterprise. Additionally, structural changes in medicolegal death investigation systems may be warranted (e.g., transition to systems staffed with MEs and promulgation of standardized education and continuing training requirements) to ensure that this critically important system is staffed by professionals with medically oriented training and professional backgrounds. This will ensure that these professionals are equipped to perform essential system functions and support the consistent and unbiased assessment of individual deaths following disasters. Federal agencies and ME/C professional organizations are essential to engaging and supporting this workforce and making the essential changes (see Recommendations 3-4 and 3-5).
MEETING THE MISSION
In its review of the current landscape of practices, systems, and tools for assessing mortality and significant morbidity following large-scale disasters, the committee identified several persistent, systemic challenges as well as potential best practices that could be brought to scale. Chapter 5 offers a set of crosscutting recommendations intended to serve as a blueprint for moving forward. Box S-3 provides a high-level synthesis of these recommendations stratified into immediate actions and future priorities.