Healthcare Coding Changes Yield New Insights On Injury Patterns

By changing how injuries are identified and categorized by health care providers, significant changes to the underlying healthcare coding system in 2015 have given injury epidemiologists new ways to track and understand patterns of injury across the United States.

For decades, injury epidemiologists in the U.S. and elsewhere relied on the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) to categorize hospital encounters. Yet in October 2015, the U.S. Department of Health and Human Services required all hospitals and health care providers in America to use the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM).

Faculty and staff from the Kentucky Injury Prevention and Research Center (KIPRC) participated in a four-state initiative to address both methodological and diagnosis-specific research using the ICD-10-CM coding system. Research from that initiative was recently featured in a special supplemental issue of the journal Injury Prevention. KIPRC faculty and staff were authors or co-authors on six of the featured seven original research manuscripts.

 “Most injury epidemiologists rely on secondary data sources, including hospital discharge records and other data created for purposes other than injury surveillance,” said KIPRC Associate Director Julia Costich, J.D., Ph.D., who is co-author on two reports in the issue. “These data sources often use coding systems to indicate the type and cause of injury. The adoption of a new coding system, ICD-10-CM, raised many questions about how injury cases would be defined to support accurate enumeration.”

The studies in the supplement cover a spectrum of injury topics including general injury, drug overdose, unintentional falls, intentional self-harm, violence, child abuse and neglect, traumatic brain injury, and work-related injury.

Costich said recommendations from the published studies have the potential to improve every state’s ability to reduce the burden of injury.

Peter Rock, who is with the University of Kentucky Center for Clinical and Translational Science (UKCCTS) and has an affiliation to KIPRC, said there were several major changes from ICD-9-CM to ICD-10-CM, including the number of injury codes. ICD-10-CM contains about 43,000 injury codes compared to ICD-9-CM’s roughly 2,600. Also, ICD-10-CM has about 7,500 external cause of injury codes compared to 1,300 for ICD-9-CM.

Rock was part of two reports in the supplemental issue—“Descriptive Exploration of Overdose Codes in Hospital and Emergency Department Discharge Data To Inform Development of Drug Overdose Morbidity Surveillance Indicator Definitions in ICD-10-CM” and “Interrupted Time Series Analysis To Evaluate the Performance of Drug Overdose Morbidity Indicators Shows Discontinuities Across the ICD-9-CM to ICD-10-CM Transition”—with both looking at the impact of code transition on overdose surveillance.

“These are critical components to understand the impacts of ICD-9/10-CM changes in measuring and reporting burden,” he said. “These definitions are important in informing public health programmatic work.”

In the exploration of drug overdose codes report, Rock said the authors’ research drove recommended changes to the consensus/standard definitions for drug overdose monitoring. Due to inherent differences in ICD-9/10-CM, the historic approach to classification of overdoses would likely miss many drug overdose-related cases.

For the “Interrupted Time Series” report, Rock said the analysis was an attempt to understand the impacts of the change in reporting. The primary result of the analysis is that extreme caution should be used when observing trends that cover the ICD9/10 threshold, as the stepwise changes that occur may likely be related to the change in the underlying measurement tool (i.e., ICD coding) or changes in coding guidelines that accompanied the ICD-10-CM coding system.

Costich and Svetla Slavova were co-authors on “Multisite Medical Record Review of Emergency Department Visits for Unspecified Injury of Head Following the ICD-10-CM Coding Transition” and “Multisite Medical Record Review of Emergency Department Visits for Traumatic Brain Injury” (TBI).

The two reports involved work done by injury investigators in four different states (Colorado, Kentucky, Massachusetts, Maryland) who looked at different ICD-10-CM codes included in the proposed CDC definition for identifying TBI cases in medical claims. The teams implemented medical record review studies to confirm if these ICD-10-CM codes included in the proposed TBI definition capture true TBI cases. The studies were carried out with involvement from CDC TBI experts and informed the CDC’s approach for TBI reporting.

Slavova, Anna Hansen, and Dana Quesinberry were co-authors on “Validation of ICD-10-CM Codes for Injuries Complicating Pregnancy, Childbirth, and the Puerperium: A Medical Record Review.” Slavova and Hansen said their study was the first to evaluate the quality of the added group of codes to correctly identify injuries experienced by pregnant and puerperal women.

“The grouping of ICD-10-CM codes proposed by the CDC to identify injuries did not include the codes specific for injuries complicating pregnancy, childbirth, and the puerperium. This shortcoming was corrected in 2019 when these codes were added to the list of diagnostic codes used for identification of injuries,” they said.

The goal of this study was to estimate the positive predictive value for the diagnostic codes identifying injuries complicating pregnancy and the post-partum period. The study found high (~80%) positive predictive value of the codes in a sample from UK Healthcare, which provides evidence that the inclusion of the codes in the updated CDC injury surveillance case definition is justified and will result in capturing true injury cases among pregnant and puerperal women.

“The study will hopefully trigger follow-up research in different healthcare settings and with larger cohorts to improve our understanding on the validity of the new surveillance definitions,” Slavova and Hansen said. “Additionally, this study may contribute to greater inclusion of pregnant and post-partum women in injury epidemiology research.”

Madison Liford, along with Terry Bunn and Ashley Bush, co-authored “Identification of Work-Related Injury Emergency Department Visits Using International Classification of Diseases, Tenth Revision, Clinical Modification Codes,” also included in the Injury Prevention supplemental issue

Liford said when researchers or public health workers want to identify work-related injuries in hospital data, they typically look for records that have an “expected payer” of workers’ compensation. She said the goal of their report was to improve the identification of work-related injuries in hospital (specifically emergency department) data by looking at the expected payer field and ICD-10-CM codes that imply work-relatedness.

“Often those individuals not captured by workers’ compensation payer approaches are those who do ‘gig’ work, contract work, or similar, and we hoped to enhance surveillance for these populations,” she said.

Liford said many emergency department visits do not have workers’ compensation as the expected payer, which means that there would be an underestimation of the number of work-related treated injuries. She said this report provides a large, reviewed list of work-related codes researchers can use to improve their identification of work-related emergency department visits.

To view the results of the reports and the Injury Prevention ICD-10-CM Injury Epidemiology and Surveillance Methods Supplement, visit https://injuryprevention.bmj.com/content/27/S1?current-issue=y.

KIPRC is a unique partnership between the Kentucky Department for Public Health (DPH) and the University of Kentucky’s College of Public Health. KIPRC serves both as an academic injury prevention research center and as the DPH’s designee or “bona fide agent” for statewide injury prevention and control.