Strategies to Improve External Cause-of-Injury Coding in State-Based Hospital Discharge and Emergency Department Data Systems

Recommendations of the CDC Workgroup for Improvement of External Cause-of-Injury Coding

Prepared by
Joseph L. Annest, PhD¹
Lois A. Fingerhut, MA²
Susan S. Gallagher, MPH³
David C. Grossman, MD⁴
Holly Hedegaard, MD⁵
Renee L. Johnson, MSPH¹
Mel Kohn, MD⁶
Donna Pickett, MPH²
Karen E. Thomas, MPH¹
Roger B. Trent, PhD^7
1National Center for Injury Prevention and Control, CDC
²National Center for Health Statistics, CDC
³Tufts University School of Medicine, Boston, Massachusetts
⁴Group Health Cooperative, Seattle, Washington
⁵Colorado Department of Public Health and Environment, Denver, Colorado
⁶Oregon Department of Human Services, Portland, Oregon
⁷California Department of Public Health, Sacramento, California

The material in this report originated in the National Center for Injury Prevention and Control, Ileana Arias, PhD, Director; and the Office of Statistics and Programming, Joseph L. Annest, PhD, Director; and the National Center for Health Statistics, Ed Sondik, PhD, Director; and the Office of Analysis and Epidemiology, Linda Bilheimer, PhD, Director.

Corresponding preparer: Joseph L. Annest, PhD, National Center for Injury Prevention and Control, 4770 Buford Highway, N.E., MS F-62, Atlanta, GA 30341-3717. Telephone: 770-488-4804; Fax: 770-488-1665; E-mail: jla1@cdc.gov.

Summary

Each year, an estimated 50 million persons in the United States experience injuries that require medical attention. A substantial number of these persons are treated in an emergency department (ED) or a hospital, which collects their health-care data for administrative purposes. State-based morbidity data systems permit analysis of information on the mechanism and intent of injury through the use of external cause-of-injury coding (Ecoding). Ecoded state morbidity data can be used to monitor temporal changes and patterns in causes of unintentional injuries, assaults, and self-harm injuries and to set priorities for planning, implementing, and evaluating the effectiveness of injury-prevention programs. However, the quality of Ecoding varies substantially from state to state, which limits the usefulness of these data in certain states.

This report discusses the value of using high-quality Ecoding to collect data in state-based morbidity data systems. Recommendations are provided to improve communication regarding Ecoding among stakeholders, enhance the completeness and accuracy of Ecoding, and make Ecoded data more useful for injury surveillance and prevention activities at the local, state, and federal levels. Implementing the recommendations outlined in this report should result in substantial improvements in the quality of external cause-of-injury data collected in hospital discharge and ED data systems in the United States and its territories.

Introduction

Each year, an estimated 50 million persons in the United States experience injuries that require medical attention, resulting in lifetime costs of more than $400 billion (1). In 2004, injuries in the United States resulted in an estimated 167,000 deaths, 1.9 million injury-related discharges from short-stay hospitals, and 31 million initial emergency department (ED) visits (2--4). These injuries, which represent 7% of deaths from all causes, 6% of hospitalizations, and 32% of ED visits, constitute a substantial public health burden. During 2004--2006, an estimated 48% of injured persons requiring medical attention received care in settings other than a hospital or an ED (e.g., outpatient clinics or physicians' offices) (CDC, unpublished data, 2007).

Population-based injury data assist public health authorities in identifying and tracking patterns and trends in the external causes of fatal and nonfatal injuries and in designing and implementing effective injury-prevention strategies (5). External cause-of-injury coding (Ecoding) in statewide mortality and morbidity data systems is the standard method used to classify injury incidents by intent (e.g., unintentional, homicide/assault, suicide/self-harm, or undetermined) and mechanism (e.g., motor vehicle, fall, struck by/against, firearm, or poisoning). For example, in the case of a hospitalization for which the principal diagnosis is a femur fracture, the Ecode identifies how the fracture was caused (e.g., the person was pushed intentionally or fell unintentionally from a ladder). Accurate information on the mechanism of injury is critical for informing prevention programs. Because trends and patterns of injury differ from state to state, state health department_based injury-prevention programs need state-specific data to understand and respond effectively to injury problems in their jurisdictions.

Injury mortality data are collected by the National Vital Statistics System, which is operated by CDC. Information on Ecodes is based on the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) (6,7). Mortality data are compiled each year by NCHS as a census of all deaths reported to state vital statistics departments (7). In the United States, Ecoding has been consistently more complete for mortality data than for morbidity data because states require that the external cause of death be listed as the underlying cause on a death certificate issued for an injury-related death. For example, if a person were in a fatal car crash and suffered a severe traumatic brain injury, the underlying cause-of-death would be the motor-vehicle crash rather than the brain injury. The Ecode for a fatal injury is assigned by state and federal vital records programs on the basis of cause-of-death information submitted on death certificates. If a death certificate is submitted without the information needed to assign a code, state vital records programs return the certificate to the certifier who submitted it and request more information. This quality-control process ensures that Ecodes are assigned for all deaths from injury. During 1999--2004, the underlying cause of death was coded as "unspecified cause" for only 6% of all injury-related deaths.

Although external cause-of-injury mortality data can be helpful for setting priorities and making policy decisions regarding injury prevention, these data are not a good surrogate for injury morbidity data (4,8). The leading causes of nonfatal injury-related ED visits, injury hospitalizations, and injury deaths differ substantially (Table 1 and Figure 1). Whereas motor-vehicle crashes are the leading cause of injury deaths, falls are the leading cause of injury hospitalizations and ED visits (Figure 1). In addition, the distributions of fatal and nonfatal injuries and the number of nonfatal injuries relative to deaths vary by external cause of injury (Table 1). For instance, the ratio of nonfatal ED visits to deaths differs for motor-vehicle traffic injuries (90 nonfatal ED visits per death) compared with poisonings (40 nonfatal ED visits per death).

For injury morbidity data, information on external causes of injury is coded using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) (9) and entered into the state's electronic hospital discharge data system (HDDS) or hospital ED data system (HEDDS). However, not all states have such data systems. For those with these data systems, the completeness and specificity of Ecodes for injury-related hospitalizations and ED visits vary substantially across states, limiting the usefulness of the Ecoded data available for certain states (10). Lack of Ecoding in state morbidity data also affects national injury statistics from federal data systems that derive their data from a sample of U.S. hospitals, such as the National Hospital Discharge Survey (i.e., 37% of injury hospital discharges with missing Ecodes) and the National Hospital Ambulatory Medical Care Survey (i.e., 10% of injury ED visits with missing Ecodes) (Table 1 and Figure 1). In contrast, certain states (e.g., Massachusetts and New York) have almost complete Ecoding for hospitalizations and ED visits, ensuring that useful data on both fatal and nonfatal injuries are available for prevention program planning and evaluation (Tables 2 and 3).

Although state HDDS and HEDDS data are used primarily for administrative and billing purposes, these systems provide the best available data sources on external cause of injury for measuring the impact of nonfatal injury on society. For example, falls are the leading cause of injury morbidity among persons aged >65 years (2,4). As the U.S. population continues to age, states can benefit from monitoring fall-related morbidity among older persons to assess changes in health-care use and cost and in the effectiveness of fall-prevention strategies (4,11--14).

This report discusses the role of Ecodes in injury prevention and practical strategies to improve Ecoding in statewide HDDS and HEDDS data systems. The CDC workgroup recommendations contained in this report outline feasible steps jurisdictions can take to improve the collection and availability of complete, specific, and high-quality Ecodes for use in injury surveillance and prevention efforts within all states, the District of Columbia, and U.S. territories. Implementing these recommendations can enhance the usefulness of state-level injury morbidity data for injury prevention program planning and implementation, priority setting, and policy setting for government and nongovernment organizations. Previous studies have demonstrated that the public health benefits will substantially outweigh the estimated economic costs of Ecoding (15,16).

Methods

Initial efforts to improve Ecoding in statewide hospital discharge data systems were made in the early to mid-1990s, but progress toward complete and accurate Ecoding in all states has been limited (10,17,18). In 2007, the Council of State and Territorial Epidemiologists (CSTE), the Injury Control and Emergency Health Services Section of the American Public Health Association (APHA-ICEHS), the State and Territorial Injury Prevention Directors Association (STIPDA), the Society for the Advancement of Violence and Injury Research (SAVIR), and the Association of State and Territorial Health Officers (ASTHO) issued position statements calling on CDC to develop strategies to improve Ecoding completeness and specificity in state hospital discharge databases (19--23).

CSTE asked CDC to take a leadership role in assembling an expert workgroup to recommend strategies to improve Ecoding in state hospital discharge databases. A workgroup of injury data experts from CDC, Tufts Medical School, state health departments, and other health professionals was established to provide recommendations for practical strategies for improving Ecoding. Members of the workgroup are associated with CSTE, SAVIR, and STIPDA. Because multiple states have indicated interest in establishing HEDDS in addition to HDDS, the workgroup decided to include in this report efforts to improve Ecoding in both statewide hospital discharge and ED data systems.

CDC has a history of working in collaboration with CSTE, SAVIR, and STIPDA on Ecoding and other projects to improve the usefulness of statewide HDDS and HEDDS data. STIPDA, in collaboration with CDC, CSTE, and SAVIR, has endorsed efforts to improve Ecoding by providing guidelines for injury surveillance in state health departments (24,25). The most recent STIPDA report provided a comprehensive guide that addresses key aspects of injury surveillance, including coding issues, data base management, quality assurance, data linkage, data reporting and dissemination, staffing, training, and partnerships (25). That report made recommendations regarding measures to establish and maintain ongoing state injury surveillance but did not discuss specific strategies for improving Ecoding and subsequent Ecoded data. The strategies outlined in this report will therefore extend and help in the implementation of STIPDA's recommendations.

The CDC workgroup used a consensus process to develop recommended strategies for improving Ecoding. First, the workgroup reviewed key recommendations in the most recent STIPDA report (25) and those in earlier position statements (19--23). Next, the workgroup developed a detailed outline of the report with draft strategies. The workgroup co-chairs wrote a first draft of the manuscript, which was sent to workgroup members for review and comment. The report sections and recommended strategies then were extensively discussed via e-mail and conference calls during August and September 2007 to arrive at a consensus. The final draft of the report was then reviewed by other state public health officials and other surveillance experts at CDC. Minor modifications were made as a result of these additional reviews before the report was finalized.

E-Coding in Morbidity Data Systems

ICD-9-CM is the standard classification system used for morbidity reporting in the United States (9). ICD-9-CM is used to classify all diseases, injuries, and their external causes in health-care records and surveys so they can be reported uniformly across institutions and jurisdictions. ICD-9-CM codes assigned in health-care records for diseases, injuries, and health conditions also are used as the basis for prospective payment to hospitals, other health-care facilities, and health-care providers. Nationally and in the majority of states, Ecodes are not used in determining reimbursement.

The U.S. Department of Health and Human Services (DHHS) has developed ICD-9-CM guidelines for assigning diagnosis and Ecodes (9). These guidelines are reviewed annually by the cooperating parties (the American Hospital Association, the American Health Information Management Association, the Centers for Medicare and Medicaid Services [CMS], and CDC). Certain states (e.g., California and New York) with mandated Ecoding have developed additional guidelines, which are not always consistent with the national guidelines.

In hospital settings, health information specialists are responsible for assigning Ecodes on the basis of the national guidelines or those mandated by their state. Certain states (e.g., California, Minnesota, and South Carolina) have ongoing quality-assurance practices, in certain cases tied to reimbursement or penalties, aimed at monitoring and maintaining the completeness and validity of Ecodes in their statewide HDDS and HEDDS systems. However, the majority of states lack policies or adequate resources to implement ongoing quality-assurance practices that would ensure high quality Ecoding.

Ecodes are structured to capture information on the intent and mechanism of injury. These circumstances usually are designated in the first three digits of the code (e.g., E884: unintentional fall from one level to another). For certain causes, the three-digit code is followed by a decimal and a fourth digit (e.g., E884.0: unintentional fall from playground equipment) that identifies more specific circumstances of the injury incident.

ICD-9-CM external cause-of-injury classification guidelines for an injury specify that health information specialists assign as many Ecodes as necessary to explain the cause, intent, and place of occurrence of the injury incident fully (26). However, in practice, Ecoding of hospital records often is incomplete, and the Ecodes that are assigned lack specificity because of insufficient documentation in the medical chart and lack of designated fields for recording Ecodes in electronic data systems. To ensure that the Ecodes assigned are as specific as allowed by the ICD-9-CM Ecode set and coding guidelines, the physician or other health-care provider must provide adequate documentation in the medical record of details regarding the incident (e.g., mechanism, intent, location, and activity at time of injury), the hospital health information specialist must view that information and assign specific Ecodes, and the Ecodes must be recorded accurately and appropriately in the database. Injury experts have recognized that this process could be facilitated by including designated fields for at least three Ecodes (for the immediate cause, the precipitating cause [i.e., the cause that started the chain of events that led to the injury, such as being struck by an object that precipitated a fall or vise versa], and the place of occurrence [e.g., at home or on the street or highway]) in the software used to capture medical information in these data systems (25). However, the majority of states currently have only one designated field in their HDDS for recording Ecodes; a few states have two designated fields; and a few states have no designated Ecode field and record E-codes in the existing diagnosis fields only (10). Experts also have recognized that because a trend exists in the United States toward uniform use of electronic health and patient records for administrative and billing purposes, creating a designated space on electronic forms to record a brief but informative narrative by physicians and other health-care providers regarding the circumstances of the injury incident could facilitate good E-coding (27).

E-Coding in States

Information on the status of Ecoding in state data systems has been published previously (10) and was updated in October 2007 (STIPDA, unpublished data, 2007). In 2007, only five U.S. states (Alabama, Idaho, Missouri, North Dakota, and South Dakota) did not have a statewide HDDS database in place. The District of Columbia (DC) and 41 states routinely collect some level of Ecodes; 26 (63%) states and DC mandated Ecoding in their statewide HDDS database (Figure 2), and 27 (54%) states and DC had an HEDDS database. DC and 25 (93%) states reported routine collection of some level of Ecodes, and 18 (72%) states mandated Ecoding in their statewide HEDDS (Figure 3).

Ecoding in the majority of state databases is incomplete. A survey conducted in 2004 reported that of 32 states that evaluated hospital records, 14 (44%) had Ecoded >90% of their injury-related hospitalizations; results varied for the other 18 states (range: 51%--89%) (10). Even among states with a high percentage of Ecoded hospital records, assigned Ecodes often lack specificity. If health-care providers do not understand why specific information needed for Ecoding is important, documentation in the medical record can be inadequate; this can lead to overuse of the unspecified Ecodes, such as those for unspecified fall (E888.9) or a motor-vehicle traffic accident of an unspecified nature (E819) (28). Unspecified Ecodes do not provide adequate detail needed for injury prevention.

Despite limitations associated with lack of completeness and specificity of Ecoding, CDC and certain states, in collaboration with CSTE, STIPDA and state partners, have published annual state injury indicator reports that contain injury mortality and morbidity data from participating states' data systems (29--31). States with injury program capacity-building grants from CDC are required to submit statewide HDDS data annually for inclusion in the annual report. All other states and territories are invited and encouraged to participate; 34 states participated in the most recent annual report (31). These reports have been used for comparison of injury patterns among states and have resulted in improved communication among states regarding injury-prevention efforts and efforts to improve statewide HDDS and HEDDS Ecoding (31). On the basis of data from annual state injury indicators reports (31), four states (Hawaii, Kansas, Oklahoma, and Oregon) increased completeness of external cause coding >20% during 1999--2004 (Table 4). Impetuses for improvement in these states included legislative initiatives, policy changes at hospitals, training of medical records coders, and increased awareness of health-care providers and health information specialists regarding the value of high-quality Ecoded data.

Certain states without Ecoding mandates from either state legislatures or hospital associations (e.g., Colorado, Minnesota, and Oregon) have relatively high rates of external cause coding (i.e., >84% completeness) (Table 2). Colorado serves as an example of how factors other than mandates can improve the completeness and accuracy of external cause coding. Colorado's HDDS is managed by the Colorado Health and Hospital Association (CHHA). In 1997, state health department staff began working with CHHA to encourage hospitals to assign Ecodes to appropriate hospital discharge records. CHHA approached medical records coders through their trade organization to outline how injury prevention activities conducted by state and local health departments and local groups could benefit by having access to Ecoded data. CHHA also pointed out to hospital chief executive officers that submitting Ecoded data would meet the hospital's requirement to report various conditions to the health department. For example, Colorado requires hospitals to report data to the statewide trauma registry. For certain hospitals, reporting Ecoded hospital discharge data to the state hospital association would meet the state's trauma registry reporting requirement, thereby saving the hospital from the need to develop an additional system for reporting injured patients to the trauma registry. During the next several years, CHHA staff, the state epidemiologist, and staff from the injury epidemiology program at the state health department met regularly with medical records coders to discuss the completeness and accuracy of Ecodes. The hospital association newsletter reported rates of Ecoded data by hospital and highlighted the hospitals that showed the most improvement. In 2003, when a comprehensive report on injury in Colorado was prepared, health department staff reviewed the results in advance with members of the hospital association and coders' trade organization. The contributions of the medical records coders and hospitals in providing complete and comprehensive Ecodes for data analysis were acknowledged in the report.

Challenges to Improving the Quality and Accessibility of E-Coded Data

Strategies are needed to improve quality assurance (QA) procedures used by states to evaluate Ecoding. QA practices vary by state; the majority of states do not evaluate Ecoded HDDS and HEDDS data routinely for completeness, specificity, and accuracy (10). Ongoing evaluation of the quality of Ecodes is needed to ensure consistency in completeness, specificity, and accuracy of Ecoding among all participating hospitals over time. States have increasing interest in using electronic health and patient records and in integrating medical care and billing data systems. This trend might encourage vendors to develop standard QA procedures and adapt them into database management software that would provide automated systems to assist coders in assigning Ecodes.

One problem with the quality of Ecoding in state databases is accuracy, as measured by percentage agreement in codes based on record reabstraction studies. Accuracy of ICD9CM Ecodes in hospital discharge records has been examined in the United States and internationally. The level of inaccurate coding for Ecodes (i.e., those with at least four digits) was 13% in the hospital discharge data system in Washington state in 1996 (28), 18% in the National Minimum Data Set in New Zealand during 1996--1998 (32), and 16% for the Victorian Inpatient Minimum Database in Australia during 1993--1994 (33). Although Ecodes were found to be reliable for reporting injury data by broad groupings (e.g., falls, motor vehicle traffic, poisoning, assault, and self-harm), a substantial lack of accuracy was reported in that part of the Ecode that identifies specific circumstances of an injury incident. The New Zealand study determined that the level of incorrect coding was similar for large and small hospitals (32). A recently published study of accuracy of Ecoding in an ED setting determined that coding was accurate for 65% of probable work-related injuries and 57% of nonwork-related injuries (34).

A study conducted in Oregon to assess methods for improving the quality of Ecoding and to identify potential barriers indicated that certain hospitals reported not having enough space on the electronic billing form to document circumstances of the injury event adequately. The majority of hospitals included in the study used software purchased from commercial vendors that was not modifiable to capture the information needed to assign specific Ecodes. Hospital reimbursement also was a potential deterrent to adequate Ecoding; not all hospitals understood the need for accurate Ecoding because Ecoding does not affect reimbursement. Hospital administrators also expressed concern that reporting Ecodes might alert an insurance company to the possibility that a third party was liable for the cost of care, possibly triggering an investigation that would delay payment to the hospital. Another issue that researchers initially thought could be a barrier was the cost of assigning Ecodes; however, the study indicated that the cost of Ecoding was minimal (an estimated $0.08 per Ecode assigned) (Oregon Department of Human Services, unpublished data, 2002).

As discussed earlier in this report, accurate and specific Ecoding requires that medical records contain sufficient detail regarding injury circumstances (e.g., "fell down stairs while working and hit head" rather than simply "hit head"). However, anecdotal evidence indicates that health-care providers might not record these details because they 1) see no reason to do so (care not being affected by these details), 2) have no financial incentive to do so (the information not being required for reimbursement), 3) are concerned about stigmatizing the patient (e.g., by documented suicidal behavior or drug use), or 4) do not understand the importance of documenting injury circumstance information in the chart for public health purposes. In addition, sometimes health-care providers simply are unable to determine the details of the circumstances. For example, a person who is discovered unconscious on the ground might not be able to provide any information as to how a head injury occurred.

In certain states, Ecoded data are not easily accessible. Certain HDDS and HEDDS databases are operated by nongovernment organizations that view these data as proprietary and do not necessarily wish to provide ready access to statewide administrative data by state health department officials or prevention program directors (10). In other states, these systems are operated by state health departments, which are willing to provide ready access to Ecoded injury data but might not have adequate resources to prepare and manage public-use data files and web-based query systems (10).

Uses of ECoded Data for Injury Prevention Decision Making

States with quality Ecoded HDDS data provide public health authorities with key data on the injury burden that help to facilitate better decision making for injury prevention (35). Collection and ready access to complete and reliable Ecoded injury data have assisted federal, state, and local public health authorities in making data-driven decisions regarding public health policy and in setting priorities (8). Public health authorities have used high-quality injury morbidity data on health-care use and costs associated with specific external causes of injury to estimate the impact of targeted, cause-specific prevention efforts on the health-care system and society (36,37).

Local hospital community health programs have used external cause-of-injury data to target the causes of specific injuries in their communities and make strategic decisions regarding where services should be offered. For example, in 1989, the Massachusetts Department of Health partnered with Cape Code Hospital, which had a high level of Ecoding, to summarize hospital discharge data by age. Falls among older adults were the primary cause of hospitalization in this area, which had a high percentage of retirees. Hospital staff met with staff of local organizations that provided services to older adults to integrate prevention of falls into the routine services of these agencies. In addition, on the basis of an analysis of the data, the state health department channelled funds from an Office of Disability Prevention grant to a local agency to conduct home visits focused on fall prevention among older adults (Holly Hackman, MD, Massachusetts Department of Health, personnel communication, 2007).

Health officials in local jurisdictions have used population-based injury morbidity data to evaluate injury prevention efforts (14). The California Department of Public Health has used Ecoded HDDS data to describe nonfatal drowning in swimming pools and spas among toddlers. Ecoded data were used in 1995 to describe nonfatal drowning rates, demographic risk factors, and hospital charges (38). A later study compared drowning deaths with hospitalized nonfatal drowning cases to calculate a fatality-to-case ratio of 1:3.5 (39). Partly as a result of these and other uses of Ecoded HDDS data, cities, counties, and the state began enacting pool safety requirements. For instance, in 1998, California enacted the Swimming Pool Safety Act to create a uniform statewide construction standard for safety devices for family pools and spas.

Beginning in 1998, the Colorado Department of Public Health and Environment (CDPHE) used Ecoded data on deaths and hospitalizations to analyze the burden of suicide and nonfatal self-harm injuries, leading to the establishment of an Office of Suicide Prevention (OSP) within CDPHE in 2000 (40). In 2006, OSP used information from Ecoded data in obtaining federal funding for suicide prevention work, including funds from the Garrett Lee Smith Memorial Act. The analysis of Ecoded data identified differences in the populations at risk for suicide compared with nonfatal self-harm hospitalizations. In Colorado, for example, the suicide rate was determined to be highest for men aged >65 years, whereas the rate of nonfatal self-harm hospitalization was determined to be highest for women aged 18--24 years (Colorado Department of Public Health and Environment, unpublished data, 2007) (2).

Quality Ecoded data in statewide HDDS and HEDDS also have been used by federal agencies that collect and use state injury data to monitor trends, set priorities for funding prevention programs, and assess program effectiveness in reducing nonfatal injuries. For example, CDC uses Ecoded injury data from statewide HDDSs in its annual state injury indicators report to measure improvements in injury surveillance capacity of state grantees (29--31). CDC also collects external cause-of-injury data in national morbidity data systems used widely by federal and state agencies for public health policy decisions, including the National Hospital Discharge Survey, the National Hospital Ambulatory Medical Care Survey, and the National Health Interview Survey (4). The Agency for Healthcare Research and Quality (AHRQ) has established a nationwide inpatient sample and makes available statewide hospital discharge and ED data sets as part of its Healthcare Cost and Utilization Project (HCUP) (41). These surveys and data sets include Ecodes (with the same limitations of accuracy, specificity and completeness as statewide HDDSs and HEDDSs) and are made available for analysis by public health researchers, medical researchers, economists, and others interested in health-care utilization, patient safety, and medical care cost issues (12,42). The Health Resources and Services Administration (HRSA) Maternal and Child Health Bureau uses available Ecoded data at the state and national levels to set program objectives targeting injury prevention among children (43). The National Highway Traffic Safety Administration (NHTSA) collects Ecoded data from ED and hospital discharge records as part of its state-based Crash Outcome Data Evaluation System (CODES) to help make policy decisions aimed at reducing motor-vehicle traffic-related injuries (44).

Professional organizations and other nonprofit entities (e.g., the American College of Emergency Physicians, the American College of Surgeons [ACS], the American Academy of Pediatrics, the American Medical Association, the National Safety Council, the Suicide Prevention Action Network, Safe Kids, the Home Safety Council, the American Trauma Society, CSTE, STIPDA, and SAVIR]) use Ecoded data in their injury prevention activities. For example, ACS has established a National Trauma Data Bank (NTDB) that contains trauma registry data from approximately 700 U.S. trauma center hospitals located across the country (45). Participating trauma centers routinely collect Ecodes in their trauma registries, and these external cause-of-injury data are required for submission to NTDB (45). CDC also has worked with the ACS Committee on Trauma to establish a national sample of trauma centers that submit data to NTDB (45). CDC data are made available to medical and injury researchers to assess the quality of trauma care and to characterize injured patients, injury circumstances, injury severity, and health outcomes. The uses of Ecoded data in statewide HDDS and HEDDS complements those from trauma registries of severe trauma patients treated in trauma centers by characterizing a broader representation of the injured patient population.

Ecodes also have been used by automobile insurance companies, health plans, health-care purchasers, and other private entities interested in injury prevention and safety issues (e.g., identifying causes associated with injuries to workers in the workplace and their families outside of the workplace and causes associated with injuries in motor-vehicle crashes). Ecoded statewide HDDS and HEDDS data have been used to help identify key causes of injuries requiring medical attention that could be addressed by implementing safety measures and policies, which has resulted in cost savings to these companies. For example, hospital discharge data from the Healthcare Cost and Utilization Project -- Nationwide Inpatient Sample (HCUP-NIS) were used to estimate the cost of fall-related hospitalizations in the United States (41).

Recommended Strategies for Improving ECoding

The workgroup made the following recommendations to improve Ecoding in statewide HDDS and HEDDS databases.

Improve Communication Among Stakeholders Regarding ECodes

CDC should facilitate a federal effort involving agencies with relevant research, programmatic, and regulatory activities in injury prevention (e.g., AHRQ, CMS, the National Institutes of Health, HRSA, NHTSA, the Bureau of Labor Statistics, the U.S. Fire Administration, the U.S. Department of Defense, and the Veteran's Administration) to

• discuss the need and uses of high-quality Ecoded nonfatal injury data for interagency collaborative efforts in injury prevention, and
• assess the inclusion of Ecodes in federal morbidity data systems, surveys, and data standards to facilitate routine collection of high-quality Ecodes in statewide HDDS and HEDDS databases.

In collaboration with CMS and state health departments, CDC should explore the possibility of linking Ecodes to uniform billing procedures used for reimbursement in government health insurance systems.

CDC should facilitate a meeting of state and federal injury surveillance and prevention experts with representatives from the health plan industry; medical, nursing, and hospital administrators' professional associations; and other health-care professional organizations to

• discuss how Ecoding and injury surveillance can be better used to drive injury prevention efforts in health-care settings, at work, and at home;
• solicit ideas and facilitate dialogue with these representatives regarding the efforts needed to improve Ecoding in HDDS and HEDDS;
• discuss efforts to make narrative documentation and coding of external cause of injury required data elements in electronic health and patient record systems and associated forms and software;
• discuss how to work with electronic health record vendors to facilitate integration of plain language Ecode dictionaries into software for easy point of care coding; and
• demonstrate the potential business case for Ecoding, from a health-care provider and purchaser perspective, such as using Ecoded data to assess health-care system demands and costs of care associated with specific causes of injury (e.g., falls among older adults in nursing homes);

In collaboration with STIPDA and CSTE, CDC should help facilitate implementation of the new STIPDA Injury Surveillance Workgroup recommendations for injury surveillance in all state and territorial health departments (25), including

• examining the use of financial incentives, enforcements, and mandates to improve the completeness and specificity of Ecoding, and
• developing methods to track improvement in the completeness, accuracy, and specificity of Ecodes in HDDS and HEDDS among states and territories.

CDC, through the International Collaborative Effort on Injury Statistics, a group of international data experts whose focus is on the standardization of injury data (46), should communicate and share ideas and methodologies on improving external cause coding in morbidity data system with international injury data experts and researchers, WHO injury prevention program representatives, and other interested groups.

Improve Collection of ECodes

In collaboration with STIPDA and CSTE, CDC should develop uniform methods to improve Ecoding through cost-effective quality assurance practices and evaluation, considering effective approaches already in place in certain states. Activities should include the following:

• CDC should develop uniform quality assurance practices (e.g., methods for ongoing evaluation to monitor completeness, accuracy and specificity of Ecodes) to ensure high-quality Ecodes and require demonstration of these practices in injury surveillance capacity-building cooperative agreements with states.
• State injury prevention programs should conduct ongoing evaluation to assess the completeness, accuracy, and specificity of Ecoding in hospitals within their jurisdiction.
• State injury prevention programs should provide feedback to data providers regarding the quality and usefulness of Ecoded data (e.g., reports of the completeness and accuracy of Ecoded data from their hospital; written reports to clinicians, coders, and hospital administrators showing how the data are being used).

CDC, in collaboration with STIPDA and CSTE, should develop training curricula for use in educational institutions (e.g., medical schools, nursing programs, health information specialist programs) and hospitals (e.g., continuing education) aimed at raising the awareness of physicians, nurses, health information specialists, and health plan and hospital administrators regarding their role in improving external cause-of-injury data, including training curricula for

• physicians and nurses on methods to document circumstances (i.e., who, what, when, where, and how) of injury incidents in the medical record,
• health information specialists with specific examples of how Ecodes are used and the need for accuracy and specificity in Ecoding, and
• hospital and health plan administrators regarding the importance of high quality Ecoded data for injury- and violence-related public health surveillance and prevention activities, and the need for hospital policies aimed at requiring high quality Ecoded data.

CDC, in collaboration with STIPDA and CSTE, should work with professional organizations of clinicians, nurses, medical records specialists, and health plan and hospital administrators to develop incentives and approaches to training their members on their role in collecting high quality external cause-of-injury data.

Improve the Usefulness of ECoded Data for Injury Prevention Efforts

CDC should engage in activities with state epidemiologists and state injury prevention directors to educate health-care workers, hospital association members, health plan staff, and the public regarding the uses of Ecoded data for prevention efforts.

State health departments should work with local health departments to develop and implement approaches, using local nonfatal injury data, to highlight injury as a public health concern and the importance of injury prevention in their community.

All state health departments with an existing statewide HDDS should participate in CDC's Injury Indicators Project to help improve communication among states on the use of Ecoded data for injury prevention efforts.

State health departments should develop and implement methods for timely and easy access to Ecoded data by policy makers, program planners, researchers, and the public through the internet (e.g., reports, slide sets, fact sheets, and web-based query systems) (25).

Conclusions

Ecodes can provide data to guide public health decisions to reduce injuries and health-care costs in the United States. Improving Ecoding in state-based hospital discharge and ED data systems is likely to help overcome current limitations of external cause-of-injury data in many states as a result of inadequate completeness and specificity. The strategies recommended in this report could facilitate communication among federal, state, and nongovernment stakeholders to determine collaborative approaches and methods for improving Ecoding in these administrative data systems. These strategies are designed to improve data collection, coding, QA practices, analysis, reporting, and dissemination of Ecoded data to policy makers, public health professionals, and the public. The goal is to have high-quality morbidity data by external cause of injury from all states and U.S. territories for use in monitoring trends, characterizing patterns, setting priorities for injury prevention programs, and assessing health-care costs to reduce the burden of injury in the United States.

Lessons learned in efforts to improve the quality and usefulness of Ecoded data from state-based morbidity data systems have pertinence to the World Health Organization's (WHO) Health Metrics Network (HMN) and its global initiatives to improve health data systems in developing countries (47). QA practices that have been demonstrated to be effective in collecting high-quality Ecoded data in statewide HDDS and HEDDS databases could be recommended for inclusion in the HMN global initiatives.

Acknowledgments

This report is based, in part, on data contributed by H Hackman, MD, Division of Violence and Injury Prevention, Massachusetts Dept of Public Health; B Hume, MPH, M McKenna, MPH, Injury Surveillance Program, Massachusetts Dept of Public Health; M Bauer, MS, S Hardman, Bur of Injury Prevention, New York State Dept of Health.

References

1. Corso P, Finkelstein E, Miller T, Fiebelkorn I, Zaloshnja E. Incidence and lifetime costs of injuries in the United States. Inj Prev 2006;12:212--8.
2. CDC. Web-based Injury Statistics Query and Reporting System (WISQARS^™). Atlanta, GA: US Department of Health and Human Services, CDC; 2007. Available at http://www.cdc.gov/ncipc/wisqars/default.htm.
3. Kozak LJ, DeFrances CJ, Hall MJ. National hospital discharge survey: 2004. Annual summary with detailed diagnosis and procedure data. Vital Health Stat 2006;13(162):38.
4. Bergen G, Chen L, Warner M, Fingerhut LA. Injury in the United States, 2007 chartbook. Hyattsville, MD: US Department of Health and Human Services, CDC, National Center for Health Statistics; 2008.
5. CDC. Surveillance for fatal and nonfatal injuries -- United States, 2001. In: Surveillance Summaries, September 3, 2004. MMWR 2004;53 (No. SS-7).
6. World Health Organization. International statistical classification of diseases and related health problems. Vol. 1. 10th Rev. Geneva, Switzerland: World Health Organization; 1992.
7. Miniño AM, Heron MP, Murphy SL, Kochanek KD. Deaths: final data for 2004. Natl Vital Stat Rep 2007;55:1--120.
8. US Department of Health and Human Services. Healthy people 2010 (conference ed. in 2 vols). Washington, DC: US Department of Health and Human Services; 2000. Available at http://www.healthypeople.gov.
9. CDC. Clinical modification, (ICD-9-CM). In: International classification of diseases. 6th ed. 9th rev. Hyattsville, MD: US Department of Health and Human Services, CDC; 2007. Available at http://www.cdc.gov/nchs/datawh/ftpserv/ftpicd9/ftpicd9.htm.
10. Abellera J, Annest JL, Conn JM, et al. How states are collecting and using cause of injury data: 2004 update to the 1997 report. Atlanta, GA: Council of State and Territorial Epidemiologists; 2004. Available at http://www.cste.org/pdffiles/newpdffiles/ECodeFinal3705.pdf.
11. CDC. Fatalities and injuries from falls among older adults -- United States, 1993--2003 and 2001--2005. MMWR, 2006;55:1221--4.
12. Stevens JA, Corso PS, Finkelstein EA, Miller TR. The costs of fatal and non-fatal falls among older adults. Inj Prev 2006;12:290--5.
13. CDC. Falls among older adults: figures and maps. Atlanta, GA: US Department of Health and Human Services, CDC; 2006. Available at http://www.cdc.gov/ncipc/duip/adultfallsfig-maps.htm.
14. Wadman MC, Muelleman RL, Coto JA, Kellermann AL. The pyramid of injury: using ecodes to accurately describe the burden of injury. Ann Emerg Med 2003;42:468--78.
15. Rivara FP, Morgan F, Bergman AB, Maier RV. Cost estimates for statewide reporting of injuries by E coding hospital discharge abstract data base systems. Public Health Rep 1990;105:635--8.
16. Bonnie RJ, Fulco CE, Liverman CT, eds. Reducing the burden of injury: advancing prevention and treatment. Washington, DC: National Academy Press; 1999.
17. Annest JL, Conn JC, McLoughlin E, Fingerhut LA, Pickett D, Gallagher S. How states are collecting and using cause of injury data. American Public Health Association; 1998.
18. Gallagher S. E codes: the missing link in injury prevention. Newton, MA: EDC; 1994.
19. Council of State and Territorial Epidemiologists. Position statement 07-INJ-01 Improving external cause coding in hospital discharge data. Available at http://www.cste.org/position%20statements/searchbyyear2007final.asp.
20. American Public Health Association. Position statement: improving external cause coding in hospital discharge data. Washington, DC: American Public Health Association. 2008. Available at http://www.apha.org/advocacy/policy/policysearch/default.htm?id=1352.
21. State and Territorial Injury Prevention Directors Association. Policy resolution; improving external cause coding in hospital discharge data. Marietta, GA: State and Territorial Injury Prevention Directors Association.; 2007. Available at http://www.stipda.org/associations/5805/files/External%20Cause%20Coding%202007.pdf.
22. Society for the Advancement of Violence and Injury Research. Policy resolution: improving external cause coding in hospital discharge data. Washington, DC: Society for the Advancement of Violence and Injury Research; 2008. Available at http://www.savirweb.org/content/index.php?pid=93.
23. Association of State and Territorial Health Officers. Position statement: improving external cause coding in hospital discharge data. Arlington, VA: Association of State and Territorial Health Officers. Available at http://www.astho.org/pubs/Ecodepositionstatement-FINAL10-07.pdf.
24. State and Territorial Injury Prevention Directors' Association. Consensus recommendations for injury surveillance in state health departments. Marietta, GA: State and Territorial Injury Prevention Directors' Association; 1999. Available at http://www.stipda.org/associations/5805/filesinjurysurveillance_recs.pdf.
25. State and Territorial Injury Prevention Directors' Association. Consensus recommendations for injury surveillance in state health departments. Marietta, GA: State and Territorial Injury Prevention Directors' Association; 2007. Available at http://www.stipda.org/displaycommon.cfm?an=8.
26. CDC. ICD-9-CM official guidelines for coding and reporting. Hyattsville, MD: US Department of Health and Human Services, CDC; 2007. Available at http://www.cdc.gov/nchs/datawh/ftpserv/ftpicd9/icdguide07.pdf.
27. Mikkelsen G, Aasly J. Narrative electronic patient records as source of discharge diagnoses. Comput Meth Programs Biomed 2003;71:261--8.
28. LeMier M, Cummings P, West TA. Accuracy of external cause of injury codes reported in Washington state hospital discharge records. Inj Prev 2001;7:334--8.
29. CDC. State injury indicators report. Atlanta, GA: US Department of Health and Human Services, CDC; 2001. Available at http://www.cdc.gov/ncipc/pub-res/stateindicators.htm.
30. CDC. State injury indicators report -- 1999 data. 2nd ed. Atlanta, GA: US Department of Health and Human Services, CDC; 2004. Available at http://www.cdc.gov/ncipc/pub-res/indicators/default.htm.
31. CDC. State injury indicators report 2004 data. 3rd ed. Atlanta, GA: US Department of Health and Human Services, CDC; 2007. Available at http://www.cdc.gov/ncipc/profiles/core_state/siir_2004.htm.
32. Langley J, Stephenson S, Thorpe C, Davie G. Accuracy of injury coding under ICD-9 for New Zealand public hospital discharges. Inj Prev 2006;12:58--61.
33. MacIntyre CR, Ackland MJ, Chandraraj EJ, Pilla JE. Accuracy of ICD-9-CM codes in hospital morbidity data, Victoria: implications for public health research. Aust N Z J Public Health 1997;21:477--82.
34. Hunt PR, Hackman H, Berenholz G, McKeown L, Davis L, Ozonoff V. Completeness and accuracy of International Classification of Disease (ICD) external cause of injury codes in emergency department electronic data. Inj Prev. 2007;13:422--5.
35. Minnesota Department of Health. Minnesota Data Access System (MIDAS). St. Paul, MN: Minnesota Department of Health; 2007. Available at http://www.health.state.mn.us/injury/midas/index.cfm.
36. Public Health Data Standards Consortium. Using external cause of injury codes: states' compelling stories. Baltimore, MD: Public Health Data Standards Consortium; 2006. Available at http://www.phdsc.org/default.asp.
37. Cohen L, Davis R, Cantor J, et al. Reducing health care costs through prevention. working document. Los Angeles, CA: Prevention Institute and the California Endowment with the Urban Institute; 2007. Available at http://www.preventioninstitute.org/healthcare.html.
38. Ellis AA, Trent RB. Hospitalizations for near drowning in California: incidence and costs. Am J Public Health 1995;85(8 Pt 1):1115--8.
39. Ellis AA, Trent RB. Swimming pool drownings and near-drownings among California preschoolers. Public Health Rep 1997;112:73--7.
40. Colorado State Public Law. Colorado's Office of Suicide Prevention. Colorado Statute Title 25 Health: Administration: Article 1.5 Powers and duties of the Department of Public Health and Environment: Part 1. General powers and duties: 25-1.5-101. Denver, CO: Colorado State Government, May 1, 2000.
41. Agency for Healthcare Research and Quality. Healthcare costs and utilization project (HCUP). Rockville, MD: Agency for Healthcare Research and Quality; 2007. Available at http://www.hcup-us.ahrq.gov.
42. Coben JH, Steiner CA, Barrrett M, Merrill CT, Adamson D. Completeness of cause of injury coding in healthcare administrative databases in the United States, 2001. Inj Prev 2006;12:199--201.
43. Health Resources and Services Administration. Child health USA 2006. Washington, DC: US Department of Health and Human Services, Health Resources and Services Administration; 2006. Available at http://www.mchb.hrsa.gov/chusa_06/healthstat/0300hs.htm.
44. National Highway Traffic Safety Administration. Crash outcome data evaluation system (CODES). Washington DC: National Highway Traffic Safety Administration; 2007. Available at http://www.nrd.nhtsa.dot.gov/departments/nrd-30/ncsa/CODES.html.
45. American College of Surgeons. National trauma data bank (NTDB). Chicago, IL: American College of Surgeons; 2007. Available at http://www.facs.org/trauma/ntdb.html.
46. CDC. International collaborative effort on injury statistics. Hyattsville, MD: US Department of Health and Human Services, CDC; 2007. Available at http://www.cdc.gov/nchs/advice.htm.
47. World Health Organization. Health Metrics Network: registering births and deaths: "the job that no one wants." Geneva, Switzerland: World Health Organization; 2007. Available at http://www.who.int/healthmetrics/en.

Acronyms Used in This Report

CDC Workgroup for Improvement of External Cause-of-Injury Coding

Co-chairs: Joseph L. Annest, PhD, National Center for Injury Prevention and Control, CDC; Lois A. Fingerhut,* MA, National Center for Health Statistics, CDC.

Members: Susan S. Gallagher,^† MPH, Tufts University School of Medicine, Boston, Massachusetts; David C. Grossman, MD, Group Health Cooperative, Seattle, Washington; Holly Hedegaard, MD, Colorado Department of Public Health and Environment, Denver, Colorado; Renee L. Johnson,^§ MSPH, National Center for Injury Prevention and Control, CDC; Mel Kohn,^§ MD, Oregon Department of Human Services, Portland, Oregon; Donna Pickett, MPH, National Center for Health Statistics, CDC; Karen E. Thomas, MPH, National Center for Injury Prevention and Control, CDC; Roger B. Trent,^¶ PhD, California Department of Public Health, Sacramento, California.

* Member, Council of State and Territorial Epidemiologists (CSTE), State and Territorial Injury Prevention Directors Association (STIPDA), and Society for the Advancement of Violence and Injury Research (SAVIR).

^† Member, STIPDA and SAVIR.

^§ Member, CSTE and STIPDA.

^¶ Member, STIPDA.

Disclaimer   All MMWR HTML versions of articles are electronic conversions from ASCII text into HTML. This conversion may have resulted in character translation or format errors in the HTML version. Users should not rely on this HTML document, but are referred to the electronic PDF version and/or the original MMWR paper copy for the official text, figures, and tables. An original paper copy of this issue can be obtained from the Superintendent of Documents, U.S. Government Printing Office (GPO), Washington, DC 20402-9371; telephone: (202) 512-1800. Contact GPO for current prices.

Date last reviewed: 3/17/2008