Chapter 2: Epidemiology

Injury is not an “accident” but rather a disease, much like malaria, tuberculosis and other public health problems, or cancer and heart disease. Injury, like other diseases, has variants such as blunt or penetrating. It has degrees of severity, rates of incidence, prevalence, and mortality that can differ by race and other sociodemographic factors. Injuries have a predictable pattern of occurrence related to age, sex, alcohol and other drugs, and again, sociodemographic factors, among others. They also have a predictable prognosis, based on age, sociodemographic factors, as well as injury severity.

This characterization of injury as a disease is an important one, and a matter of more than just semantics. It is only when public health concepts are applied to this disease of injury that it, like other public health diseases, can be controlled to a socially acceptable level. The first step after its recognition as a disease is to characterize the disease such that control strategies can be applied. Epidemiology is the study of patterns of disease occurrence in human populations and the factors that influence these patterns.¹ Thus, the majority of injury epidemiology relates to describing specific populations and the factors that influence injury occurrence in these populations.

Descriptive epidemiology refers to the distribution of disease over time, place, and within or across specific subgroups of the population. It is important for understanding the impact of injury in a population and identifying opportunities for intervention. The burden of injury can be described as the most common, most fatal, most debilitating, or most costly within a specific population.

Analytic epidemiology, in contrast, refers to the more detailed study of the determinants of observed distributions of disease in terms of causal factors. The epidemiological framework traditionally identifies these factors as related to the host (ie, characteristics intrinsic to the person), the agent (physical, chemical, nutritive, or infectious), and the environment (ie, characteristics extrinsic to the individual that influence exposure or susceptibility to the agent). The environment can be physical or sociocultural. The importance of this epidemiological approach is the direction it gives to injury prevention efforts as well as directing areas requiring further research.

Injuries can result from acute exposure to physical agents such as mechanical energy, heat, electricity, chemicals, and ionizing radiation in amounts or rates above or below the threshold of human tolerance.² The transfer of mechanical energy accounts for more than three quarters of all injuries.³ The extent and severity of injury is largely determined by the amount of energy outside the threshold of human tolerance. Both the exposure to energy and the consequences of that exposure are greatly influenced by a variety of factors both within and beyond individual or societal control.⁴

The concepts of the public health approach applied to injury control seek to modulate factors related to the host and agent and/or their interactions within the environment utilizing a number of strategies. These strategies encompass engineering, education, the enactment and enforcement of laws, and economic incentives and disincentives.

The public health approach as it applies to injury was first conceptualized by William Haddon in the late 1960s.² He developed and promulgated a phase-factor matrix that incorporated the classic epidemiological framework of host, agent, and environment in a time sequence that encompasses three phases: pre-event, event, and post-event. Factors related to the host, agent, or environment in the pre-event phase determine whether the event will occur (eg, motor vehicle crash). Factors in the event phase determine whether an injury will occur as a result of the event and the degree of injury severity. Factors in the post-event phase influence the outcome from, or consequences of, any injuries of any severity that do occur. An example of the Haddon Matrix applied to an actual injury event is depicted in Table 2-1.

Although the Haddon Matrix is the foundation of injury epidemiology, it is not enough to direct robust injury prevention and control efforts. The addition of potential control strategies to the matrix in a three-dimensional fashion results in an “injury control cube,” suggesting that injury prevention and control are not unidimensional or unifactorial and that the greater the number of sections of the “cube” that are addressed, the greater the control of the injury event. For example, gun control laws focus on only the agent, in the pre-event phase, using a legislative strategy (Fig. 2-1). However, there are many other counter measures that can be applied in other phases and to the host or environment.

Injuries rank fourth as a cause of death for all age groups in this country, and have consistently held that place for many years. It is the leading cause of death among children, adolescents, and young adults ages 1–44 (Table 2-2).³ In 2013, 192,945 people persons died in the United States as a result of an injury, up from approximately 150,000 in 2009 and resulting in an age-adjusted injury rate per 100,000 population of 58.53. The predominance of injury deaths among the young results in another measure of the burden of injury, years of productive life lost. This measure makes the assumption that individuals are most productive to society before the age of 65; given the ever-increasing length of productive life, this is an often incorrect assumption. Nevertheless, it does give some measure of comparison of the effect of various causes of mortality. In 2013, all causes of death contributed to over 11 million years of productive life lost. Deaths from injury were responsible for 31.2%, more than any other individual category. Unintentional injury was responsible for 19%, suicide for 7%, and homicide for 5% of total years of productive life lost.

From 2002 to 2010, the total trauma-related mortality decreased by 6%. However, mortality trends differed by mechanism. There was a 27% decrease in the motor vehicle-associated death rate associated with a 20% decrease in motor vehicle collisions, 19% decrease in the number of occupant injuries per collision, lower injury severity, and improved outcomes at trauma centers (Fig. 2-2). While firearm-related mortality remained relatively unchanged, mortality caused by firearm suicides increased, whereas homicide-associated mortality decreased. In contrast, fall-related mortality increased by 46%.⁵

The timing of trauma deaths, classically described as trimodal, has changed due to advances in resuscitation and ICU that have essentially eliminated the last peak of deaths from multisystem organ failure.^6,7,8 The majority of all deaths still occur within minutes of the injury, either at the scene prior to arrival of emergency medical service (EMS), en route to the hospital, or in the first hours of care. These immediate deaths are typically the result of massive hemorrhage or severe neurological injury. The second peak of the bimodal death distribution occurs within several hours to days of the event, and is due primarily to central nervous system (CNS) injury (Fig. 2-3).^7,8

Deaths represent only one small aspect of the injury disease burden. Each year, almost 31 million people suffer a nonfatal injury; the vast majority of these are seen in emergency departments or urgent care centers without requiring hospital admission, with almost 2.5 million people hospitalized and surviving to discharge. Many of these nonfatal injuries have far-reaching consequences with potential for reduced quality of life and high costs accrued to the health care system, employers, and society. In 2010, the estimated total lifetime costs associated with both fatal and nonfatal injuries occurring in any one year amount to over $420 billion (Tables 2-3 and 2-4).^3,9,10

The costs associated with injury deaths account for a disproportionate share of total injury costs. Estimates show that deaths account for less than 1% of all injuries but account for 31% of total injury costs. The remaining 69% of costs due to injury is associated with nonfatal injuries. These costs include direct expenditures for health care and other goods and services purchased as a result of the injury. Direct expenditures account for approximately 30% of the total cost of injury. The value of lost productivity due to temporary and permanent disabilities is also taken into account and represents 41% of the total costs. These are merely the financial costs and do not take account the pain and suffering to the patients, their families and associates that are the sequelae of nonfatal injuries.

Injury is a disease predominantly affecting young males. Seventy percent of injury deaths and over half of nonfatal injuries occur among males.^3,8 In every age group except ages 0–9, the rate of injury death for males is more than twice as high as the rate for females. For nonfatal injuries, males are only 1.3 times as likely as females to be affected. This gender-related risk reverses after the age of 65 with females 1.3 times as likely as males to suffer nonfatal injury in that age category. Therefore, injuries account for more premature deaths than cancer, heart disease, or HIV infection.³

The disease of injury has a bimodal age distribution of mortality that peaks for both genders in the 16- to 40-year-old age group and then again in those older than 65 years of age. Persons under the age of 45 account for 53% of all injury fatalities, just over 50% of hospitalizations, and nearly 80% of ED visits.^3,8 Hospitalizations and ED visits also follow this pattern of a bimodal peak related to age and a predominance among males.

The elderly, while being less likely to be injured, are more likely to be hospitalized or die from those injuries with a lesser degree of severity than their younger counterparts. Although representing only 3% of the US population, those over the age of 65 accounted for approximately 26% of all injury deaths and 30% of all injury-related hospitalizations. The proportion of citizens over the age of 65 is projected to increase to nearly 20% by the year 2030.¹¹ This has significant implications for the future of health care as over the next several decades it is expected that the elderly will account for approximately 40% of all injury deaths and hospitalizations. As a result, there are many suggestions that age should be taken into account in triage protocols, treatment protocols, and admission location within the hospital.

Injuries are typically categorized by their mechanism, intent, and place of occurrence. Mechanism refers to the external agent or particular activities that were associated with the injury (eg, motor vehicle related, falls, firearm related, etc). Intent of the injury is classified as either unintentional or intentional.

Injuries that are intentionally inflicted can be further subcategorized into interpersonal (eg, homicide) or intrapersonal (eg, suicide). Intent may not be always determinable. Injuries resulting from legal interventions and operations of war are typically classified separately as an “other intent” category.

The classification system most often used in describing the specific mechanism and intent of injury is the international classification of disease (ICD). This classification system was developed and promulgated by the World Health Organization and is now in its 10th edition.¹² The E-Code, or external cause of injury code, provides detailed information about the circumstances associated with injury-related ED visits and hospitalizations. These codes are essential to the epidemiology of injury and its accurate study. They provide critical public health information for monitoring health status, setting injury prevention priorities, and developing and evaluating injury prevention programs at the local, state, and national levels.

E-codes can also be useful for quality initiatives associated with injury-related claims (eg, motor vehicle crash-related injuries) that may assist CMS in making payment decisions. Hence, numerous professional organizations, including the American College of Surgeons Committee on Trauma, the American College of Emergency Physicians, the Emergency Nurses Association, and the National Safety Council, have published position statements to endorse the need for improving E-coding in state mortality and morbidity data systems. They have also urged that the capture of at least three fields as part of the E-code. Two of the E-code fields can be used for coding the precipitating and immediate causes (eg, the mechanism/intent of injury such as falls, motor vehicle traffic, fire/burn, cut/pierce, assault, self-harm, etc) and one other field to delineate place of occurrence (eg, home, street/highway, residential, institution, etc). Unfortunately, E-codes are not mandated by all payors or all states, and remain recommended but not mandated by CMS as a part of ICD-10.

The distribution of injuries by mechanism varies for deaths, hospitalizations, and ED visits. The two leading mechanisms of injury death are related to motor vehicles and firearms. Nearly 30% of all injury deaths are violence related. Emphasizing this point are statistics on intentionality of injury, which reveal that 93% of nonfatal injuries are unintentional whereas 68% of fatal injuries are unintentional.

Injury in the workplace is tracked by the Bureau of Labor Statistics within the United States Department of Labor. A total of 4679 fatal work injuries were recorded in the United States in 2014. This represents 3.3 fatal work injuries per 1,000,000 full-time equivalent workers.¹³ Overall, transportation-related incidents accounted for the majority (40%) of occupational injury deaths. Assaults and violent acts accounted for 16% of fatalities, contacts with objects and equipment 15%, and falls 17%. Nine percent of occupational-related deaths in 2014 were a result of homicide.

In addition to the fatalities associated with work-related activities, there were a total of 3.3 million nonfatal injuries recorded by the Bureau of Labor Statistics (3.3 cases per 100 workers).¹⁴ Seventy-one percent of these occurred in service providing industries and nearly half produced disability.

Leading Mechanisms of Injury

Motor vehicle-related injury

Traffic-related incidents involving motor vehicles are the leading cause of injury death and rank second as a cause of nonfatal injury in the United States. It is the leader of all causes of death in the 1–34 age group.

Adolescents and young adults are at the highest risk for both fatal and nonfatal injuries due to motor vehicles. Their rates of death, hospitalization, and ED visits are approximately twice the rate for all ages combined. White males age 15–24 are at particular risk. For black males in that same age group, traffic-related injury death rank second as a cause of death behind firearm-related injuries. The elderly, age 75 and older, are also at relatively high risk for dying from motor vehicle incident-related injury.

Males are more than twice as likely as females to die from motor vehicle crashes. Males under the age of 45 are also more likely to be hospitalized as a result of motor vehicle-related injuries, although the gender differential is not as great as for fatalities. Males and females age 45 and older, in contrast, are equally likely to be hospitalized.

Determinants of injury occurrence and severity in a motor vehicle-related incident relate to speed of impact, vehicle crash worthiness and the use of safety devices and restraints including safety belts, air bags, and helmets. This is an area where epidemiologic analyses have truly had an impact in reducing the burden of injury. When used, safety belts have been shown to reduce fatalities to front-seat occupants by 45% and the risk of moderate-to-critical injury by 50%. Currently, safety belt usage rates in the United States range from 69% in South Dakota to 98% in Oregon with a national average of 87% in 2013.¹⁵ This is an increase of 1% over 2012 and 6% over 2006. Thirty-two states have primary enforcement laws, with only one state achieving over 90% usage rate without a primary seatbelt law. Only two states with primary seatbelt laws have usage rates less than 80%. The additional presence of an air bag in belted drivers provides increased protection resulting in an estimated 51% reduction in fatality rate. Despite some success in reducing the role of alcohol in motor vehicle injuries, it remains a major factor in fatal crashes among adolescents and young adults. Approximately 50% of all traffic fatalities including the driver, occupant, bicyclist, or pedestrian have been found to have a blood alcohol concentration (BAC) of 0.08 g/dL or greater. The proportion of fatally injured drivers with elevated BAC varies with age. For all age groups, it has remained relatively constant since 1999 at approximately 26% for women and 43% for men. In contrast, the proportion of drivers with other drugs, including narcotics, depressants, and cannabinols has increased significantly.¹⁶

Firearms

In 2013, there were approximately 57,000 intentional and unintentional gun-related deaths in the United States. The overwhelming majority of these deaths are intentional (98%) and related to violence, with only 2% being unintentional.³ Seventy-two percent of all firearm deaths were suicide.

Firearm-related injuries disproportionately affect males and younger people. Approximately 87% involve males. In the 15–34 age group, firearm-related death rates for males are nearly seven times that for females. Firearm-related injuries are the leading cause of death in black males aged 15–34. From a global and cultural standpoint, firearm-related mortality is eight times higher in the United States than other high-income countries in the world.

The majority of firearm-related deaths among males aged 15–34 in the United States (67%) are homicides. Suicide accounts for an additional 33%. Suicide in the elderly is also a significant problem with 3895 firearm-related suicides among the elderly between ages 65 and 84 or greater. This represents 22% of all firearm-related injuries for both genders and all ages. Over 90% of the suicides in the elderly population were among males.

Data on nonfatal firearm-related injuries are not as complete; however in 2013, there were 67,394 reported nonfatal injuries caused by firearms, over three-quarters of which were intentional.³ One of the reasons for the lack of data similar to that which exist for motor vehicle collisions is the political and emotional issues that surround discussions around gun violence. It is clear, however, that both fatal and nonfatal firearm injuries account for just over $39 billion in 2010 in direct medical costs and productivity losses.¹⁰ It is estimated that firearms of some type are present in about 38% of all US households and carried by one in 12 students.^17,18 Some studies suggest that those who live in homes that harbor guns are more likely to die from homicide and suicide in the home than are residents of homes without firearms.^19,20

Falls

There were 23,443 fatal falls in 2007 and over 8.5 million nonfatal injuries that were a result of falls. Falls represent approximately 13% of all injury deaths. Falls account for over one third of all injury hospitalizations and one quarter of all injury- related visits to the ED.³

The greatest occurrence rate is witnessed in the younger and older age groups; however, the severity profile in the two groups is quite different. In children, falls are common but generally not severe or fatal. Falls are the leading cause of nonfatal injuries for all children aged 0–19. Approximately 8000 children are treated daily in US EDs for fall-related injury.²¹ This totals almost 2.8 million children each year. Less than 3% of these visits result in hospitalization. Approximately one half of all pediatric falls occur in the home and one quarter occur at school. Falls in children ages 0–4 years are most commonly from furniture or stairs. In older children, falls are commonly from standing and/or associated with recreational activities related to playground equipment, bicycling, or sports.

In adults of working age, most fatal falls are from buildings, ladders, and scaffolds. Falls on stairs increase in significance starting at age 45.³ ED visit rates for falls are consistently higher for men up to the age of 44. From age 45 and older, this trend reverses and by age 65, ED visit rates and hospitalizations for falls in women are nearly three times those in men. This finding is consistent with the increased fracture risk in women after menopause and, specifically, those with osteoporosis.

In the elderly, falls are a significant cause of mortality and morbidity being the cause of death in 23% of injury deaths for those 65 and over and 32% of injury deaths in those 85 years of age and older. The death rate from falls after age 85 is over three times that for people aged 75–84 years old. Falls are also the most common cause of nonfatal injury in the elderly, accounting for nearly 60% of injury-related ED visits and approximately 80% of injury-related hospitalizations for persons aged 65 years and older. In the United States, one in five people over the age of 65 will sustain a fall annually. Of these, about one quarter will be injured and another quarter will restrict their daily activities for fear of another fall.²² The economic impact of falls in the elderly is sizable and is estimated to reach nearly 55 billion dollars in 2020.²³

Major risk factors for falls among the elderly include those related to the host (advanced age, history of previous falls, hypotension, psychoactive medications, dementia, difficulties with postural stability and gait, visual disturbances, cognitive and neurological deficits, or other physical impairment) and environmental factors (loose rugs and loose objects on the floor, ice and slippery surfaces, uneven flooring, poor lighting, unstable furniture, absent handrails on staircases) to name a few. The risk of falling increases linearly with the number of risk factors present, and it has been suggested that falls and some other geriatric syndromes may share a set of predisposing factors. All of these factors are potentially modifiable with combinations of environmental, rehabilitative, psychological, medical, and/or surgical interventions.²⁴

Cataloging and analyzing the distribution of injuries by their nature and severity is important to efforts at establishing priorities for prevention as well as treatment, trauma system development, and research. Several systems for classifying the nature and severity of injury exist and a number of these are described elsewhere in this textbook.

The international collaborative effort on injury statistics has published an injury diagnosis matrix that provides a uniform framework for using the ICD codes in categorizing injury diagnosis by the body region involved and the specific nature of the injury.²⁵ The most comprehensive source of national data on the nature of injury death is death certificate data. However, these data have significant limitations due to variations and inaccuracy in the diagnosis listed as cause of death, terminology, and reporting practices for injury by geographic region and over different time periods.^26,27 The National Trauma Data Bank (NTDB)²⁸ also provides some insight as to the nature, severity, and types of injuries encountered utilizing a nonscientific sample of trauma centers that voluntarily contribute data to the data bank. Some of what is known about the overall nature of trauma deaths is based on a limited number of studies conducted in selected geographic regions using coroners’ reports and autopsy records.^29,30 These types of records, much like the death certificates upon which they are often based, are variable in completeness, accuracy, and utility.

CNS injuries are the most common cause of injury death, accounting for between 30% and 50% of the total number of injury deaths each year.³¹ Like other injuries, TBI can range from mild to severe, with many mild cases going undiagnosed. TBI-related injuries are the second-most common nonfatal injury, second only to extremity injuries. Even optimal treatment can result in permanent impairment and disability.^32,33,34

The second ranking cause of injury-related death is hemorrhage, accounting for an additional 30–35%. Hemorrhage may be related to either torso or extremity trauma.

The overall incidence and patterns of injury vary between urban and rural populations and across different regions of the country.³ Unintentional injury death rates are highest in rural areas, whereas homicide rates are several times higher in central cities compared to rural and suburban communities.

Injury death rates also vary by region of the country. Death rates for unintentional injury tend to be highest in the west and south, whereas suicide rates are highest in the west and homicide rates highest in the south. However, there is a substantial state-by-state and even county-by-county variation. To date, local data relating to nonfatal injuries are not uniformly available to examine trends by rurality or geographic region. The observed differences related to geographic location and population density may be a function of a number of confounding factors such as access to care, economic, or educational climate, to name a few. When these factors are controlled for, these geographic disparities can be less prominent, or nonexistent.

There are a number of confounding factors that may influence results and, more critically, the interpretation and conclusions drawn from epidemiological analyses of injury. These include race, ethnicity, culture, socioeconomic status, access to health care, mental health, alcohol and other drugs, as well as others. Due to their number and multiplicity, adequate control for any or all is difficult at best. Hence, forethought and caution should be exercised before making generalizations regarding epidemiological findings.

Although on the surface there may appear to be certain associations between race and injuries, particularly violent injuries, controlling for socioeconomic status, there is little disparity between races as perpetrators or victims of violence. For example, homicide rates have been shown to vary significantly by economic status.

Therefore, data that are stratified by race and/or Hispanic origin must be interpreted carefully. First, the number of people in the population that is used as the denominator in the calculation of the death rate generally comes from US Census Bureau estimates and the characteristics of those who died used in the numerator generally stems from either the funeral director or the medical examiner. As a result, to the extent that race and Hispanic origin are reported inconsistently by the different data sources generating the numerator and denominator, rates may be biased. Second, bias in estimates by race and ethnicity also can result from undercounting of specific populations in the census, thereby potentially producing an overestimation of death rates.

Differences in health status by race and Hispanic origin also are known to exist and may be explained by factors including socioeconomic status, health practices, psychosocial stress and resources, environmental exposures, discrimination, and access to health care.³⁵ As these factors are not routinely collected or controlled for, analysis of injury mortality and morbidity by race and ethnicity may lead to incorrect inferences. With specific regard to data on violence, estimates may be misinterpreted because attention may have been directed to the victim rather than to the perpetrator, for whom sufficient data are not routinely collected.

Although this chapter emphasizes the concept of injury being a disease entity in and of itself, data suggest that for a significant number of trauma patients, injuries may be an unrecognized symptom of an underlying alcohol or other drug use problem. Therefore, it may be that injury is actually a comorbidity of the disease that is alcohol and substance use disorder. Nearly 50% of injury deaths are alcohol related. Traumatic injury accounts for roughly the same number of alcohol-related deaths as cirrhosis, hepatitis, pancreatitis, and all other medical conditions associated with excessive alcohol use combined. A multicenter study that included data on more than 4000 patients admitted to six trauma centers demonstrated that 40% had some level of alcohol in their blood upon admission,³⁶ and up to 60% of patients test positive for one or more intoxicants.^36,37,38 Therefore, it is clear that alcohol and substance use must be considered in the epidemiology of injury as well as in the equation leading to effective injury control. Recognition of this fact, along with evidence that intervention to address alcohol use disorder in the acute setting was beneficial, helped drive the recommendation from the American College of Surgeons Committee on Trauma to provide screening and brief intervention for alcohol use disorder at Level I and Level II trauma centers.³⁹

There are a plethora of data available nationally⁴⁰ and locally to define and research injury epidemiology. A number of these data sources have been used in the production of this chapter and are referenced; the vast majority of the data come from the CDC and NCIPC, specifically the WISQARS (Web-Based Injury Statistics Query and Reporting System) program.³ This database resource is web-based and interactive, allowing the user to determine the population or subpopulation of interest as well as the specific epidemiologic factors of interest. Although data on nonfatal injuries are not as comprehensive or robust as those on fatal injuries, significant improvement has occurred in recent years relating to the scope and quality of data collection. This has enhanced the understanding of the magnitude and significance of injury as a major public health problem.

Several of these databases provide information on several types of work-related injuries with a number of others focusing on injuries and injury deaths related to other unintentional and intentional injuries. Many are ongoing surveillance systems. This collective group of databases varies in scope and the extent to which they provide information on mechanism and intent, nature and severity, risk factors, health services use, costs, and health outcomes. Some are population based and some are not. Comprehensive data on fatalities are available from vital statistics data, although these data do not provide detailed information about the extent and nature of injury sustained.

Standardized data on nonfatal injuries treated in the ED, including those treated and released, transferred, or hospitalized are available from the National Electronic Injury Surveillance System—All Injury Program (NEISS-AIP).⁴¹ The NEISS-AIP is a collaborative effort between the National Center for Injury Prevention and Control (NCIPC) and the US Consumer Products Safety Commission (CPSC). This database acquires information on over half a million injury-related ED visits secondary to a consumer product to a nationally representative sample of 66 hospitals on an annual basis. Products include things such as bicycles, treadmills, cars, or ladders. The NEISS-AIP is the most comprehensive database on all nonfatal injuries presenting to hospitals with EDs that is currently available. These data, together with injury mortality data, can be accessed through WISQARS.³

Injuries that result in hospitalization can also be obtained from both the National Hospital Discharge Survey (NHDS) and the Healthcare Costs and Utilization Project (HCUP-3).^42,43 Both sources can provide detailed information regarding the nature and severity of injuries, treatment, and discharge disposition, but they are limited in that codes for classifying the mechanism and intent of the injury are not routinely recorded. Although strategies exist for estimating distribution by mechanism and intent given incomplete data, the lack of uniform E-coding of hospital discharges as well as the exclusion of ED cases that are treated and released remains a significant impediment to the optimal use of these databases for studying the entire spectrum of injury epidemiology.

An additional confounder in the reliability and accuracy of these essentially administrative databases is the extent, prioritization, and accuracy of ICD coding. Also, it should be pointed out that these are only population based from the standpoint of the population of hospitalized patients. They do not capture all deaths and will not ever include patients with minor injuries not seeking treatment at hospitals.

More data on nonfatal injuries not resulting in hospital admission or death are available from the National Health Interview Survey (NHIS),⁴⁴ and the National Ambulatory Care Survey (NAMCS).⁴⁵ The NHIS relies on self-reports of injury events, whereas both the NAMCS and the NHIS rely on data abstracted from injury-related visits to hospital EDs, hospital outpatient departments, and/or physician offices. These databases do generally include E-codes.

In addition to these sources of comprehensive data across all types and severities of injury, several sources of national data exist that are specific to a particular mechanism or intent. Examples include the Fatal Accident Reporting System (FARS),⁴⁶ the National Automotive Sampling System—General Estimates System,⁴⁷ and the Crash Injury Research and Engineering Network (CIREN).⁴⁸ Also worthy of note are the National Occupant Protection Use Survey (NOPUS), National Fire Incident Reporting System (NFIRS), the National Traumatic Occupational Fatality Surveillance System and the Survey of Occupational Injuries and Illness for Occupational Injuries, the National Crime Victimization Survey (NCVS) and the Uniform Crime Reporting System for Intentional Injuries (which excludes suicides and self-inflicted injuries),⁴⁹ as well as the American Burn Association Burn Repository.⁵⁰ These databases are particularly useful for monitoring injury rates specific to certain mechanisms and for identifying risk factors associated with their occurrence.

Less developed are the data systems that deal with violence-related injuries overall and firearm-related injuries in particular. Created in 2002, the NVDRS is a surveillance system that pulls together data on violent deaths in 32 states, including information about homicides, such as homicides perpetrated by an intimate partner (eg, boyfriend, girlfriend, wife, husband), child maltreatment (or child abuse) homicides, suicides, and deaths where individuals are killed by law enforcement in the line of duty.^51,52 The system also collects data on unintentional firearm injury deaths and deaths of undetermined intent. Linking information about the “who, when, where, and how” from data on violent deaths provides insights about “why” they occurred. Frontline investigators, including homicide detectives, coroners, crime lab investigators and medical examiners, collect valuable information about violent deaths. But these data are often not combined in a systematic manner to provide a complete picture. NVDRS attempts to provide a more complete picture by collecting facts from four major sources about the same incident and pooling information into a usable, anonymous database. An incident can include one victim or multiple victims. The four major data sources are death certificates, coroner/medical examiner reports, law enforcement reports, and crime laboratories. The facts that are collected about violent deaths include circumstances related to suicide such as depression and major life stresses like relationship or financial problems, the relationship between the perpetrator and the victim—for example, if they know each other, other crimes, such as robbery, committed along with homicide, and multiple homicides, or homicide followed by suicide. As data are only provided from 32 states and are not nationally representative, and because of previously discussed issues with inaccurate data from some of the linked data sources, some estimates and conclusions that can be drawn from these data are limited. There has been some movement toward developing a data collection system similar to that developed for motor vehicle crashes, which would be an essential component to a nationwide effort at reducing the epidemic of violence currently being experienced in this country.

Of particular interest to trauma clinicians and clinical researchers are clinical databases. The most noteworthy of these is NTDB.²⁷ This database is the largest aggregation of US trauma registry clinical data ever assembled. Since its inception, more than 6 million records have been amassed emanating from more than 900 trauma centers of various levels. Data completeness, accuracy, and validity, have been continuous concerns, which have been increasingly ameliorated over time. Recent implementation of mandatory submission of from trauma centers accredited by the American College of Surgeons Committee on Trauma has increased the utility of this important resource. As a partial solution to these issues, the American College of Surgeons Committee on Trauma, which administrates the NTDB, has instituted the National Sample Program that specifically seeks more highly controlled data from a nationally representative sample of 100 Level I and Level II trauma centers in the United States. The American College of Surgeons Trauma Quality Improvement Program (ACS TQIP^®) collects data from over 300 trauma centers, ensures that they meet rigorous quality checks, and use them to provide feedback about each trauma center’s performance. The program uses risk-adjusted benchmarking to provide each trauma center with accurate national comparisons.⁵³

Two additional phases of trauma care where data have been lacking are the prehospital phase and the post–acute care phase or rehabilitation. The National Emergency Medical Services Information System (NEMSIS)⁵⁴ is the national repository that is being developed to store prehospital EMS data from every state in the nation. Since the 1970s, the need for EMS information systems and databases has been well established, and many statewide data systems have been created. However, these data- bases vary significantly in their ability to acquire patient and systems data and allow analysis at a local, state, and national level. Currently 37 states contribute to the NEMSIS database, which is being characterized as the National EMS Registry and utilizing the NEMSIS data dictionary. There is an increasing impetus due to initiatives from professional organizations and regulatory agencies to have prehospital run sheets and data systems be NEMSIS compliant, which will facilitate submission of consistent and valid data to the national database. Although a national database, it is a convenience sample and as a result does not give population-based estimates.

Information from the postacute phase of care is essential to long-term clinical and financial outcome studies. The Uniform Data System for Medical Rehabilitation (UDSMR)⁵⁵ catalogs data from rehabilitation hospitals nationwide for use in evaluating the effectiveness and efficiency of their rehabilitation programs. It provides the most comprehensive data available on rehabilitation patients across many diagnostic categories, including injuries. The database includes information on demographics, type of injury, length of stay, primary payor, and postinjury rehabilitation circumstances such as employment status, living situation and Functional Independence Measure (FIM), which is the most widely accepted functional assessment measure in use by the rehabilitation community.

National data can be used for drawing attention to the magnitude of the injury problem, for monitoring the impact of federal legislation, and for examining variations in injury rates by region of the country and by rural versus urban/suburban environments. They can also be useful in aggregating sufficient numbers of cases of a particular type of injury to analyze causal patterns and clinical or other outcomes on an individual or systems basis.⁵⁶ Often, however, these national data are not appropriate for developing and sustaining injury prevention programs at the state and local level. State and local data are more likely to reflect injury problems specific to the local area and therefore more useful in setting priorities and evaluating the impact of policies and programs in these more limited catchment areas. Additionally, local data are typically more persuasive than are national data in advocating to establish a policy or to achieve funding of injury control programs at the local level. Some of the previously described national databases do provide subsets of data at the state or even county level; however, many do not.

Availability, accuracy and completeness of local injury data varies substantially by state and county. Vital statistics and death certificate data are generally available for 100% of injury-related deaths. As previously discussed, however, these data are limited in the information they provide about the nature and circumstances of the injury, cause of death, and risk factors associated with the death. Medical examiner and coroner reports can be a useful adjunct to death certificate data, but once again, the completeness and quality of these data vary substantially from state to state. Autopsy rates are equally variable and are generally biased toward being performed in cases of suspected homicide.

State and local data on trauma hospitalizations are generally available from two principal sources, either uniform hospital discharge data and hospital or system trauma registries. Hospital discharge data are predominantly administrative in nature, whereas trauma registry data are primarily clinical. Both types of databases have limitations, which have been alluded to previously. Trauma registries suffer from both selection bias and inconsistent inclusion criteria as well as highly variable data integrity. In both types of databases, ICD coding is not uniform. Administrative databases in general are not useful in attempting to analyze clinical issues despite available methods to estimate injury severity using ICD codes.⁵⁷ Both hospital discharge databases and trauma registries do not include information on trauma deaths that occur at the scene, in transport, or in the ED nor do they routinely include patients treated and released.

Specifically, in comparison to hospital discharge data, trauma registries typically include more detailed information regarding the cause, nature, and severity of the injury. Some trauma registries will also include data on deaths occurring in the ED. Trauma registries, for the most part, may collect information only on “major trauma” patients. This leads to sample bias of a small subset of all injured patients in a population. It is important to reemphasize that caution should be exercised in using these databases for describing the epidemiology of trauma as neither is population based.

Uniform data on trauma patients treated and released from EDs, hospital clinics, and physicians’ offices are generally less accessible on a county or state level. Other data sources, often available at the state and local levels that can be useful in studying the epidemiology of injury, include routinely collected information from EMS, police, fire departments, poison control centers and child protective services, among others.

The utility of existing data at the state and local level can be significantly enhanced by linking data across multiple data sources. Single data sources are often limited in their content or scope of coverage, or both. Techniques have been developed and are continually being improved to facilitate linkage of these databases to avoid the high costs of gathering new data.⁵⁸ Several states have now linked hospital discharge data, vital statistics, police crash reports, and prehospital run sheet data.

Local, state, and national data are extremely expensive to collect and even more expensive to ensure they are reliable and accurate. As a result, funding for registry or database initiatives is often hard to come by, and those resources supported by federal dollars are ever at risk for budget cuts. However, these data are incredibly important, forming the cornerstone for injury prevention, identifying areas in which additional research is needed, and pointing out where advocacy efforts are necessary.

In summary, injury imposes a heavy burden on society in terms of both mortality and morbidity along with its sizable economic burden on the health care system and society. Largely unrecognized is the fact that many fatal and nonfatal injuries are preventable using specific strategies guided by the analysis of injury epidemiology. Hence there is no societal level of tolerance, or perhaps intolerance, and fear of incidence as there is for HIV or West Nile virus and H₁N₁ influenza. Yet, these diseases contribute much less to the burden of public health disease than do injuries.

Risks of injury death vary by age and gender. The majority of injury deaths are unintentional, with elderly people at a particularly high risk of death from unintentional injuries. Considering intentional injuries, overall, suicide greatly exceeds homicides, but rates again vary by age, gender, and urban or rural residence. Mechanisms of injury death also vary be age. The risk of injury death on the job varies by occupation. From a global perspective, the United States compares less than favorably with other countries in terms of fatal injury, particularly those related to firearms (Figs. 2-4 and 2-5).

In total, injury deaths declined slightly during the 2002–2010 period with significant variation by mechanism of injury. All causes are declining with the exception of fall-related deaths, which are increasing significantly. Injury morbidity rates have demonstrated declining trends among all age groups except the elderly. Although certain assumptions or “profiling” may arise from the association of injury and certain mechanisms thereof, a number of confounding factors unrelated to racial origin have been outlined, which should dissuade broad generalizations that are unfounded. Alcohol and other drugs continue to be intimately associated with all types and mechanisms of injury.

In conclusion, although significant strides have been made in reducing the rate at which injury occurs, trauma remains a major public health issue. More efficient ways of treating injuries as they occur, or tertiary prevention, should and will continue to be the major thrust of clinical care providers and researchers. However, it is equally and perhaps more important that efforts to develop appropriate programs and policies that will prevent them from occurring also be prioritized. Education of policy makers and the public that this public health epidemic can and must be controlled is an essential component of this effort. Accurate, easily obtainable and understandable data is a key first step in this process.

Integrated efforts at primary, secondary, and tertiary prevention, along with public information and education programs, are the only effective means to effect injury control and reduce the burden of injury on individuals, the health care system and society at large.⁵⁹ Understanding the importance of high-quality data as a building block for the study of injury science is important for all who participate in the care of the injured. Studying the epidemiology of injuries provides the opportunity for understanding how, when, and with whom to intervene.