eAppendix. ICD 9-10 Codes for Injuries and Procedures
eTable. Demographic, Injury, and Discharge Characteristics of Matched Cohort (n=15456)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Ingram ME, Nagalla M, Shan Y, et al. Sex-Based Disparities in Timeliness of Trauma Care and Discharge Disposition. JAMA Surg. 2022;157(7):609–616. doi:10.1001/jamasurg.2022.1550
Are potential differences in timeliness to surgical care after severe trauma associated with disparities in discharge outcomes by sex?
In this cohort study of 28 332 patients from the 2013 to 2016 Trauma Quality Improvement Program databases, it was found that female patients were more likely to experience delays in undergoing pelvic fixation, delays to triage, and, overall, were more likely to be discharged to a skilled nursing facility after severe trauma than their cohort-matched male counterparts.
Results suggest that there are disparities in trauma care delivery by sex that may be excellent targets for quality improvement efforts in assessment, triage, and discharge planning for injured patients.
Differences in time to diagnostic and therapeutic measures can contribute to disparities in outcomes. However, whether there is an association of timeliness by sex for trauma patients is unknown.
To investigate whether sex-based differences in time to definitive interventions exist for trauma patients in the US and whether these differences are associated with outcomes.
Design, Setting, and Participants
This was a retrospective cohort study conducted from July 2020 to July 2021, using the 2013 to 2016 Trauma Quality Improvement Program (TQIP) databases from level I to III trauma centers in the US. Patients 18 years or older with an Injury Severity Score (ISS) greater than 15 and who carried diagnoses of traumatic brain injury, intra-abdominal injury, pelvic fracture, femur fracture, and spinal injury as a result of their trauma were included in the study. Data were analyzed from July 2020 to July 2021.
Main Outcomes and Measures
Primary outcomes assessed timeliness to interventions, using Wilcoxon signed rank and χ2 tests. Secondary outcomes included location of discharge after injury, using propensity score–matched generalized estimating equations modeling.
Of the 28 332 patients included, 20 002 (70.6%) were male patients (mean [SD] age, 43.3 [18.2] years) and 8330 (29.4%) were female patients (mean [SD] age, 48.5 [21.1] years), with significantly different distributions of ISS scores (ISS score 16-24: male patient, 10 622 [53.1%]; female patient, 4684 [56.2%]; ISS score 41-74: male patient, 2052 [10.3%]; female patient, 852 [10.2%]). Male patients more frequently had abdominal (4257 [21.3%] vs 1268 [15.2%]) and spinal cord (3989 [20.0%] vs 1274 [15.3%]) injuries, whereas female patients experienced greater proportions of femur (3670 [44.0%] vs 8422 [42.1%]) and pelvic (3970 [47.6%] vs 6963 [34.8%]) fractures. Female patients experienced significantly longer emergency department length of stay (median [IQR], 184 [92-314] minutes vs 172 [86-289] minutes; P < .001), longer time in pretriage (median [IQR], 52 [36-80] minutes vs 49 [34-77] minutes; P < .001), and increased likelihood of discharge to nursing or long-term care facilities instead of home after matching by age, ISS, mechanism, and injury type (male patient:female patient, odds ratio, 0.72; 95% CI, 0.67-0.78).
Conclusions and Relevance
Results of this cohort study suggest that female trauma patients experienced slightly longer delays in trauma care and had a higher likelihood of discharge to long-term care facilities than their male counterparts.
Sex differences in timeliness of diagnostic and therapeutic care occur in acute myocardial infarction care,1-3 stroke care,4-9 and intensive care unit (ICU) triage.10-12 Studies have shown that male patients receive higher levels of care and more timely admission to critical care units than female patients for the same disease acuity.13 Furthermore, severely injured female patients are more likely to be undertriaged by emergency medical services to nontrauma centers.14
Although sex differences in acute care are well documented, evaluations of sex differences in trauma are less well studied. Within both acute care and trauma resuscitation, timeliness of evaluation and intervention remain crucial to reducing mortality and preventing long-term morbidity.4,8,10 Identifying differences in efficiency measures that drive differences in outcomes is essential to understanding how to improve the quality of care delivered to trauma patients. For trauma care, the key efficiency measures associated with outcomes include prehospital time, captured as emergency medical services (EMS) time, emergency department (ED) length of stay (LOS), and time to initial stabilizing operative interventions for respective life-threatening injuries, including exploratory laparotomy, emergent angiography, cerebral monitor placement for severe traumatic brain injury, and femur or pelvic fixation.15-17 Accordingly, identifying sex disparities in timeliness of care, especially if associated with outcomes, would be excellent targets for interventions to improve quality of care delivered.
This cohort study sought to investigate whether sex-based differences exist in timeliness of trauma care. Our first aim was to describe sex differences in efficiency measures related to timeliness of trauma care, from the prehospital phase through to definitive surgical treatment phases, in injured patients across US trauma centers. Our second aim was to determine whether differences in timeliness of trauma care would be associated with discharge disposition differences between male and female patients.
This study was a retrospective observational cohort study, conducted from July 2020 to July 2021, of the 2013-2016 American College of Surgeons (ACS) Trauma Quality Improvement Project (TQIP) databases. The institutional review board of the Feinberg School of Medicine, Northwestern University, approved the project. This study followed Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
The TQIP database consists of patient encounters of patients who were admitted to predominantly ACS-verified level I or II trauma centers, with penetrating or blunt traumatic injury. Patients attended mostly ACS level I and II centers; however, more than 30% were cared for in non–ACS-verified trauma centers where state designation is extremely variable and may be equivalent to ACS verification or considerably less strenuous with less oversight.18 Many state-designated level II trauma centers may function as level III trauma centers.19 Patients cared for at state-designated centers were included to improve the generalizability of our findings, as were the 4 patients who were cared for at ACS level III trauma centers. TQIP-participating trauma centers have critical structural hospital resources necessary to provide uninterrupted high-quality trauma care while also adhering to stringent continuous quality improvement and standardized data collection.13 Patients 16 years or older with at least 1 severe injury (with notable exceptions of severe burns or isolated geriatric hip fractures), with an Abbreviated Injury Scale (AIS) score of 3 or more in at least 1 region of the body, who were not dead on arrival, and who do not have any preexisting advanced directives, were included in the TQIP data nationally. Hundreds of patient and hospital variables, including both efficiency and outcome measures of care, were abstracted by dedicated, certified surgical clinical reviewers using a validated system.20
For this study, we examined adult patients (≥18 years of age) with an Injury Severity Score (ISS) of 15 or more who presented to a participating TQIP hospital between January 1, 2013, and December 31, 2016, with time-sensitive injuries. Selected injuries included traumatic brain injury, intra-abdominal injury, pelvic fracture, spinal cord injury, and femur fracture. Injuries and associated surgical procedures were identified using International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes and procedure codes (eAppendix in the Supplement). Patients were excluded if they had an ISS of 15 or less or were missing any predictor variables of interest. Informed consent was not required because these were retrospective analyses of cohort data collected for quality improvement purposes that were completely de-identified. The Northwestern University institutional review board agreed to waive the need for informed consent because it would require obtaining protected health information data and to do so would increase risk of breach of privacy.
The primary predictor of interest was sex (male vs female). Additional covariates included age, race and ethnicity, ISS, Revised Trauma Score (calculated when able using the Glasgow Coma Scale, ED first systolic blood pressure, and respiratory rate),21 injury type (femur fracture, pelvic fracture, traumatic brain injury, intra-abdominal injury, spinal cord injury), injury intent (unintentional, self-inflicted, assault, undetermined), mechanism of injury, hospital trauma ICU bed size, hospital teaching status, ACS level verification, and measures related to timeliness of care (ie, efficiency measures). These 9 efficiency measures included prehospital emergency medical services (EMS) time (minutes), ED LOS (hours), time to venous thromboembolism (VTE) chemoprophylaxis (hours), time to invasive cerebral monitor placement (hours), time to angiography (hours), and time to operative procedure (days) (eg, time to pelvic fracture fixation, time to femur fracture fixation, time to anterior or posterior spinal fixation, and time to exploratory laparotomy) as indicated by injury type(s). All efficiency measures are discrete variables defined and captured by TQIP reviewers.
All data are derived from TQIP dataset. Race and ethnicity are classified in TQIP as defined by the Office of Management and Budget22 and the National Trauma Databank Data definitions.23 There are data in other medical disciplines that demonstrate racial and ethnic minority groups receive less timely care; as a result, race and ethnicity were included as a predictor of interest in this study. Patients included were of African American, American Indian, Asian, Native Hawaiian or Other Pacific Islander, White, and other race and ethnicity (other race and ethnicity is not further divided in the National Trauma Data Standards in order to protect patient privacy).
The primary outcome of interest was discharge disposition. This was a polytomous outcome with options of inpatient mortality, discharge to postacute care facility (PACF) (ie, home, inpatient psychiatry, to law enforcement, or to acute rehabilitation), or discharge to a long-term care facility (LTF) (ie, skilled nursing facility or long-term acute care hospital).
Predictor variables were compared by sex using the original cohort on univariate analysis with χ2 tests. Given the marked differences in demographic and injury characteristics by sex found in the cohort, propensity score matching was used to adjust for selection bias and match on the probability of being a male patient or female patient. A propensity score model predicting probability of being a female patient was created adjusting for race and ethnicity, age, ISS, mechanism, and injury type. Hospital identification was used as an additional exact-matching criterion. Propensity score matching was performed using the greedy nearest neighbor algorithm24 with a maximum specified caliper width of 0.5. The prematch and postmatch demographic and clinical patient variables were compared and demonstrated the postmatch cohorts to be well balanced (eTable in the Supplement).
After matching, categorical and continuous predictor variables were compared by sex on univariate analysis, demonstrating the cohort to be more balanced. Wilcoxon signed rank tests for paired differences were then performed on the matched cohort examining differences in EMS time, ED LOS, time to VTE chemoprophylaxis, time to invasive cerebral monitor placement, time to angiography, and time to definitive surgical treatment for injuries (eg, time to laparotomy, time to pelvic fixation, time to spinal fixation, time to femur fixation) by sex.
A multinomial logistic regression model using generalized logits was fit using the matched cohort to evaluate the association between final discharge disposition (mortality, PACF, or LTF) and sex after adjusting for ACS trauma level verification, ICU trauma bed size, hospital teaching status (community, nonteaching, teaching), injury intent (unintentional vs other) and Revised Trauma Score (≤4 or >4). For this model, observations were clustered by propensity score–matched pairs within hospital using an independent working correlation matrix. Interaction effects between race and sex were evaluated. All statistical evaluations were performed using SAS, version 9.4 (SAS Institute). Data were analyzed from July 2020 to July 2021.
The overall study cohort included a total of 28 332 patients with an ISS of 15 or more (ISS score 16-24: male patient, 10 622 [53.1%]; female patient, 4684 [56.2%]; ISS score 41-74: male patient, 2052 [10.3%]; female patient, 852 [10.2%]). Of the total population, 20 002 (70.6%) were male patients (mean [SD] age, 43.3 [18.2] years) and 8330 (29.4%) were female patients (mean [SD] age, 48.5 [21.1] years) (Table 1). Male patients included the following races and ethnicities: 3935 (20.4%) African American, 113 (0.6%) American Indian, 397 (2.1%) Asian, 43 (0.2%) Native Hawaiian or Other Pacific Islander, 12 908 (66.9%) White, and 1914 (9.9%) other race or ethnicity. Female patients included the following races and ethnicities: 1154 (14.2%) African American, 65 (0.8%) American Indian, 231 (2.9%) Asian, 19 (0.2%) Native Hawaiian or Other Pacific Islander, 5997 (74.0%) White, and 639 (7.9%) other race or ethnicity.
Female patients were more likely to be older (>54 years: female patients, 5681 [39.7%]; male patients, 3314 [28.5%]), have been injured in a motor vehicle collision (6217 [74.6%] vs 12 781 [63.9%]), and incur femur (3670 [44.0%] vs 8422 [42.1%]) and pelvic (3970 [47.6%] vs 6963 [34.8%]) fractures compared with male patients in the cohort. Male patients more frequently had abdominal (4257 [21.3%] vs 1268 [15.2%]) and spinal cord (3989 [20.0%] vs 1274 [15.3%]) injuries (Table 1).
After matching, there were 7728 total pairs. Wilcoxon rank sum tests of median differences by sex for 9 efficiency measures of care demonstrated significant differences only for ED LOS. Female patients were found to have longer ED LOS than male patients (median [IQR], 184 [92-314] minutes vs 172 [86-289] minutes; P < .001), longer time in pretriage (median [IQR], 52 [36-80] minutes vs 49 [34-77] minutes; P < .001), and increased likelihood of discharge to nursing or long-term care facilities over home after matching by age, ISS, mechanism, and injury type (Table 2). A significant P value for EMS time was found despite no calculated difference in median time by sex, indicating a skewed distribution for difference (EMS time for male patients − EMS time for female patients) and evidenced by a corresponding negative mean difference time. Similarly, for time to laparotomy, a marginally significant P value was found despite no difference in median time by sex. The corresponding positive mean difference time indicated a skewed distribution for difference (larger magnitude positive difference values vs negative difference values). Time to angiography, time to invasive cerebral monitor placement, time to femur fixation, time to pelvic fixation, and time to spinal fixation did not differ significantly by sex. Interaction associations between race and sex were also examined and are provided in Table 3. We found that Asian female patients had the longest ED LOS (median time, 245.0 [IQR, 140.0-376.0] minutes vs African American female patients, 217.0 [IQR, 103.0-364.0] minutes; White female patients, 204.0 [IQR, 112.0-335.0] minutes; and female patients of other race or ethnicity, 184.0 [IQR, 100.0-306.0] minutes) and also the greatest time to receiving VTE prophylaxis (median time, 63.5 [IQR, 32.8-102.0] hours vs African American female patients, 46.3 [IQR, 24.9-79.1] hours; White female patients, 46.5 [IQR, 25.7-86.0] hours; and female patients of other race or ethnicity, 45.7 [IQR, 25.7-76.5] hours). Female patients experienced delays in femur fixation across all races, but African American female patients experienced longest delays for pelvic fixation (median time, 2.0 [IQR, 1.0-4.0] days vs Asian female patients, 2.0 [IQR, 1.0-2.0] days; White female patients, 2.0 [IQR, 1.0-3.0] days; and female patients of other race or ethnicity, 1.0 [IQR, 1.0-3.0] days).
Multivariable regression modeling predicting odds of mortality vs discharge to PACF vs discharge to an LTF on the matched cohort showed sex differences in discharge disposition (Table 4). Male patients were more likely to die than be discharged to an LTF than female patients (male patient:female patient OR, 1.51; 95% CI, 1.36-1.69), but female patients were more likely to be discharged to an LTF than to a PACF (female patient:male patient, OR, 0.72; 95% CI, 0.67-0.78). Overall, in the matched cohort, odds of mortality were not found to be different by sex. All patients with a Revised Trauma Score less than 4 had greater odds of mortality (OR, 18.28; 95% CI, 14.33-23.21) compared with discharge to PACF, as well as being discharged to an LTF (OR, 2.3; 95% CI, 1.76-3.01) compared with discharge home. Patients whose injury was classified as intentional or other intent (eg, assault, undetermined, self-inflicted) presented with 2.74 times higher odds of mortality than discharge to an LTF compared with patients who experienced unintentional injury (OR, 2.74; 95% CI, 2.07-3.64). ACS verification level, number of ICU trauma beds, and teaching hospital status were not found to be significantly associated with discharge disposition.
This cohort study identified sex-based disparities in 1 of 9 efficiency measures associated with timeliness of care and in discharge disposition between male and female patients after severe injury. Results suggest that female patients experienced a significantly longer ED LOS, were more likely to have delays with femur or pelvic fracture repairs, and when matching for age, ISS, injury types, and mechanism, were more likely to be discharged to long-term care facility than to home compared with male patients.
There have been prior studies observing sex-based disparities of care in trauma care settings. There is literature to support that female injury burden is more frequently underestimated both in the field14 and in-hospital.25 It is well documented that female patients are more frequently undertriaged,26 which results in preventable deaths.27 Further, prehospital triage is one area where improvements in training and medical support can decrease early mortality after major trauma. A 2012 study examining the National Ambulatory Care Reporting System in Canada concluded that a smaller proportion of female patients received trauma care than male patients (49% vs 62%), that EMS services were less likely to transport female patients than male patients, and that physicians were less likely to transfer female patients to trauma centers than male patients.28 Previously published literature also found that male patients had a 2.75 higher odds of receiving the highest prehospital priority compared with female patients when evaluating EMS services specific to trauma response.29 A study of survivors of traumatic brain injury older than 65 years demonstrated that, after controlling for age, mechanism of injury, ISS, and LOS, female patients were 1.3 times more likely to be discharged to a long-term care facility than to home compared with male patients.30 In addition, discharge to postacute care facilities has been found to be an independent predictor for hospital readmission after controlling for age, mechanism, and severity of trauma among TQIP trauma centers.31
There is some evidence offering a suggestion as to why such discharge disposition disparities by sex exist. One possible reason is that, as traditional primary caretakers of households, female patients are susceptible to decreased social support or have family members who are more likely to request long-term care facility discharge than permitting continued recovery at home.32 Similarly, qualitative studies have observed that female patients who are single, widowed, or live alone may not be able to garner social support to enable home-based rehabilitation after trauma.33 Additional studies suggest a lack of appropriate psychological support and/or early referral to physical therapy services while admitted after trauma may contribute to decreased functional outcomes reported among female patients older than 16 years after major multiple traumas, compared with age- and ISS-comparable male counterparts.34-36 Without clinical notes available to provide context around the trends observed in our study, we are limited in our ability to determine where differences in decision-making may exist by sex. However, our findings align with other studies regarding increased ED LOS and disproportionate discharge to long-term care facilities rather than to home for female patients compared with male patients after trauma. The findings of this study suggest that future studies are needed to further examine potential local or systemic biases driving these differences. Additionally, some studies support using risk calculators and models to predict patients at risk of non–home discharge and aid institutions in mitigating these factors.37 In other domains of quality improvement to reduce sex or race disparities, implementation of clinical screening tools to heighten clinician awareness of possible disparities has been associated with a reduction in race and sex disparities.38 Further quality improvement efforts are needed to reduce the apparent disparities present in timeliness of trauma care from emergency department triage through to discharge planning.39
As a retrospective cohort study, this study had certain limitations. First, these data were not collected specifically for this study, and therefore, this was a secondary data analysis. However, the ACS-TQIP data collection processes are validated, standardized, and robust. Second, for this particular study, the database does not technically differentiate between sex and gender. The sex variable is defined as based on external anatomy, not self-reported identity-based differences. The National Trauma Data Standards before 2021 categorize male and female patients in a binary form only. Given increased understanding of the implications in differentiating between sex and gender associated with trauma, the findings of this study may overlook further disparities experienced by nonbinary individuals, and/or may be overestimating risk for individuals with varying sex.40 To address this, we would advocate for institutions to perform internal evaluations of their processes and include qualitative studies to ensure delivery of care across sexes is equitable. Third, the acuity of injury and illness encounters represented in this study still may be susceptible to selection bias. However, we controlled for this bias using propensity score matching, demonstrating balanced cohorts after matching. Additionally, these data are limited in reporting on insurance and the role it may play in discharge disposition. Finally, as this study disproportionally includes ACS-verified level I and II centers, the findings may not be generalizable to a broader context. However, it would be reasonable to expect that if the most quality-minded trauma centers have sex disparities in trauma care, these disparities may exist to similar, or even greater, extent when considering all hospitals receiving trauma patients.
Results of this cohort study suggest that female patients were found to have a significantly longer ED LOS after a traumatic event and were less likely to be discharged home or to a home equivalent than male patients. These findings suggest potential gaps of care that may be excellent targets for quality improvement of existing processes of assessment and triage and discharge planning.
Accepted for Publication: February 25, 2022.
Published Online: May 18, 2022. doi:10.1001/jamasurg.2022.1550
Corresponding Author: Martha-Conley E. Ingram, MD, MPH, Surgical Outcomes Quality Improvement Center, Feinberg School of Medicine, Northwestern University, 633 N St Clair, 20th Floor, Chicago, IL 60611 (firstname.lastname@example.org).
Author Contributions: Drs Ingram and Stey had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Ingram, Nagalla, Nasca, Stey.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Ingram, Nagalla, Shan, Stey.
Critical revision of the manuscript for important intellectual content: Ingram, Nagalla, Nasca, Thomas, Reddy, Bilimoria, Stey.
Statistical analysis: Ingram, Shan, Nasca, Thomas, Reddy, Stey.
Administrative, technical, or material support: Nagalla, Reddy, Stey.
Conflict of Interest Disclosures: Dr Thomas reported receiving the American College of Surgeons Firearm Clinical Scholar Fellowship, which was funded by the American Foundation for Firearm Injury Reduction in Medicine, the Eastern Association for the Surgery of Trauma, the American College of Surgeons Committee on Trauma, the Pediatric Trauma Society, the Western Trauma Association, and the American Association for the Surgery of Trauma. Dr Reddy reported receiving grants from the National Institutes of Health during the conduct of the study. Dr Stey reported receiving the American College of Surgeons James Carrico Faculty Research Fellowship and the American Association for the Surgery of Trauma Research and Education Fund Scholarship Award. No other disclosures were reported.
Meeting Presentation: This work was presented at the Academic Surgical Congress; February 4, 2021; virtual meeting.