Comprehensive measurement of health-related quality of life is critical for assessing and communicating the health and economic impact of nonfatal injuries and violence. However, affected individuals, medical professionals, and the general public may have different perceptions of functional losses related to a health incident.^1–3 A direct comparison of health state valuations for a variety of short-term conditions with an explicit timeline as elicited from medical professionals versus general population respondents concluded there was consistency between those groups, while other studies based on health state valuations from injured individuals tend to indicate more impairment compared to valuations from the general public.^3–5 Affected individuals are an important source of health-related quality of life information—some researchers have cogently argued that affected individuals are the most accurate source for this information²—but data collection can be prohibitively complex, expensive, and have limited generalizability when affected individuals are questioned about their health states over time.^{2 3 6 7} A separate challenge is that combining published estimates of health-related quality of life from different groups of affected individuals can be challenging owing to a wide variety of data collection instruments—a recent systematic review identified 14 different generic instruments used in injury health state valuation studies in the previous decade.⁸

Discussions about from whom to elicit health state valuations are important and ongoing, but high-profile guidance in the United States on this topic is clear: In 1996, the Washington Panel on Cost-effectiveness in Health and Medicine recommended that “for purposes of resource allocation, the relevant preferences are those of the general public.”⁹ This was confirmed by the Second Panel in 2016 (“community preferences for health states are the most appropriate source of preferences for reference case analyses”).¹⁰ The Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) has facilitated likely the most comprehensive and accessible elicitation of health state valuations from general population respondents to date.^{11 12} GBD health state valuations are publicly available and based on regression modeling of pairwise ranking of multiple health state vignettes (i.e., “which description indicates better health?”) from tens of thousands of general population respondents from multiple countries, yielding scaled disability weights (0=no disability, 1=health loss equivalent to death) for 183 health conditions.

The most detailed US nonfatal injury surveillance data source, the National Electronic Injury Surveillance System - All Injury Program (NEISS-AIP), does not include data collection on health-related quality of life. Linking NEISS-AIP to an existing vetted data source for health state valuations can improve opportunities to measure and communicate the health and economic impact of injuries and violence. This study aimed to generate new health-related quality of life data for US nonfatal injury surveillance using disability weights calculated from general population surveys.

Methods used to generate GBD disability weights are detailed elsewhere.^{11 12} Briefly, surveys were fielded using paired comparisons to assess which health states were viewed as more debilitating. A lay description for each health state was developed by experts such that main outcomes, symptoms, and impact on daily activities were clearly explained using non-clinical terminology. Household surveys in 2010 and 2013 were conducted in nine countries (Bangladesh, Indonesia, Peru, Tanzania, United States, Hungary, Italy, Netherlands, Sweden), and a web-based survey included respondents recruited through media campaigns and scientific journal editorials, yielding 60,890 total respondents. Table 1 demonstrates GBD health state descriptions and disability weights for 47 injury types by nature of injury (e.g., dislocation of hip).

NEISS-AIP is a nationally representative probability sample of US hospitals that can provide weighted nationwide estimates of emergency department (ED) visits for injuries.¹³ In NEISS-AIP annual records of ED visits from 2015–2020 (unweighted n=2.9 million), there were 492 unique injury types by body part category, body part, and injury diagnosis (e.g., Arm/hand-Wrist-Fracture) based on 26 body parts and 30 diagnoses. NEISS-AIP has a higher number of injury types than GBD due to more specificity by both body part and diagnosis (e.g., thermal versus chemical burns). Some NEISS-AIP body part–diagnosis types were far more common than others—head internal injuries or concussion accounted for 11% of analyzed NEISS-AIP ED visit records and poisoning for 12% of records (data not shown).

Definitions underlying each injury diagnosis, including ICD-9-CM codes where available, from both NEISS-AIP and GBD were used to align the two data sources.^{14 15} There were 309 NEISS-AIP injury types with a direct match to a single GBD injury type. A further 183 NEISS-AIP injury types were matched to 2+ GBD injury types; a weighted average of the matching GBD disability weights was generated using each contributing injury type’s incidence from GBD 2021 US national estimates for 2015–2020 (Appendix, Table S1, Table S2).¹⁵ When the NEISS-AIP injured body part was “All parts,” the GBD disability weight for “Multiple fractures, dislocations, crashes, wounds, pains, and strains” was used (Table 1). When the NEISS-AIP injured body part was “Unknown/Not stated” but the diagnosis was provided (e.g., fracture), an average of GBD weights by injury diagnosis weighted by US incidence was applied. When the NEISS-AIP injury diagnosis was “Other,” an average of all GBD injury disability weights weighted by incidence was applied. Where possible the GBD disability weights corresponding to a short-term injury (duration less than one year, indicated in Table 1) were selected, to correspond most closely with the scope of injuries reported in NEISS-AIP, though this was not an option in all cases (appendix p 3). A supplemental table (S1) reports all assessed NEISS-AIP injury types, the mapped or calculated mean GBD disability weight, and corresponding 95% uncertainty interval (UI; reflecting the 2.5^th and 97.5^th percentile of the GBD standard simulation process to produce mean estimates from underlying data), and all GBD natures of injury (“n-codes”) used to generate the disability weight.

This study complies with the Guidelines on Accurate and Transparent Health Estimate Reporting (GATHER) recommendations (Table S3).¹⁶

Disability weights for NEISS-AIP injury types mapped from GBD ranged from 0.006 to 0.408 for all combinations of injured body part and diagnosis (Figure 1). The average disability value weighted by incidence for all NEISS-AIP types was 0.073 (95% uncertainty interval [UI]: 0.050–0.097) (S1). Only 28 NEISS-AIP injury types had a disability weight greater than 0.200. Diagnoses such as internal injury, crushing, burns, nerve damage, and some fractures had higher disability weights (e.g., Upper Trunk-Internal injury, 0.347 [0.235–0.471]), whereas injuries such as lacerations, avulsions, and contusions typically had lower weights (e.g., Finger-Puncture, 0.006 [0.002–0.012]). Injuries impacting the lower trunk, upper trunk, head, or neck also had higher disability weights (e.g., Lower Trunk-Hemorrhage, 0.324 [0.220–0.442]). Injuries that impacted multiple body parts had higher disability weights, such as burns covering 25–50% of the body or the entire body (0.314 [0.211–0.441]), or multiple fractured body parts (0.258 [0.172–0.356]). The most frequent injury reported in NEISS-AIP during 2015–2020 was poisoning (disability weight: 0.163 [0.109–0.227]), followed by internal injuries to the head (0.168 [0.112–0.232]) and lacerations of the face (0.018 [0.010–0.029]) (Figure 2). Of the top 20 most frequent injury types, 14 were sprains/strains, lacerations, or contusions (disability weight range: 0.006 to 0.025), and Lower Trunk-Fracture had the highest disability weight (0.182 [0.126–0.259]) (Figure 2).

General population valuations of injury health states for NEISS-AIP derived as described here can help to improve opportunities to measure and communicate the impact of nonfatal injuries. Disability weights can be used directly to compute years lived with disability to better understand the population-level health impact of nonfatal injuries, or standard monetary valuations of reduced health-related quality of life can be generated to accompany more common cost values such as medical care and lost employment productivity.^{17 18} The use of general population respondents to value health states is consistent with the premise that valuations should reflect a broad social understanding of health impairments.¹²

This study’s mapping of publicly available disability weights to a prominent nonfatal injury surveillance source can improve opportunities to quantify, monitor, and communicate the health and economic impact of injuries and violence.