Measurement of Long-Term, Quality of Life Outcomes in Injury Databases

Abstract

Introduction:

Traumatic injuries affect patients’ long-term quality of life (QOL). In 2016, the National Academies highlighted the need to record long-term QOL outcomes within trauma databases. We studied national injury-specific databases to assess whether they capture QOL outcomes, and their key determinants, across the entire continuum of trauma care.

Methods:

We coded variables in the Trauma Quality Program (TQP), Burn Care Quality Platform, and the Burn, Spinal Cord Injury (SCI), Traumatic Brain Injury (TBI) Model Systems databases to identify QOL and social determinants of health (SDOH) measures. We also calculated the proportion of variables collected across distinct phases of care (e.g., prehospital care, hospital care, rehabilitation care).

Results:

The databases ranged from 127 variables for Burn Care Quality Platform to 995 variables for SCI Model System. TQP collected data on prehospital, emergency department, and hospital care but did not collect outcomes after hospital discharge or measure QOL outcomes. In contrast, Model Systems databases captured information about long-term outcomes, including QOL information (Burn: 289 variables [33% of total variables]; TBI: 159 variables [25%], SCI: 115 variables [12%]). All databases measured SDOH, ranging from 12 variables (3%) in TQP to 67 variables (8%) in Burn Model System. Economic stability was the most frequently measured SDOH domain.

Conclusions:

Despite the recognized need to measure and address QOL factors post-injury, there currently is no standardized system for capturing these long-term outcomes, except for patients enrolled in the SCI, TBI, and Burn Model System programs. The trauma community needs a national system to track long-term QOL outcomes in trauma survivors.

Introduction

As patients are surviving increasingly severe injuries due to high-quality trauma care,¹ the trauma community has started to consider the long-term outcomes of trauma survivors.² While survival to hospital discharge is the traditional metric of successful trauma care, it fails to capture a comprehensive understanding of trauma survivors’ long-term outcomes. Recent work has demonstrated that trauma survivors face numerous obstacles with their long-term, functional recovery, including physical, cognitive, and social-emotional deficits.^3-7 For example, 37% of injury survivors have a new, long-term functional limitation that prevents them from independently performing daily activities,³ and nearly 50% report daily chronic pain.⁴ These factors limit trauma survivors’ independence and quality of life in both home and community settings.

Quality of life (QOL) is an important long-term outcome for trauma survivors because it captures the patient’s perspective of their injury recovery, taking into consideration their personal and social context.^8,9 QOL is typically measured using patient-reported outcome measures (PROMs), which are standardized measures of health outcomes in which the patient is the direct reporter for all data collected. Through measuring QOL with PROMs, healthcare providers can better understand patients’ perspectives of their recovery, functional status, and physical and mental health. Furthermore, social determinants of health (SDOH), such as patients’ education level, income, access to housing, and social support, are known modifiers of long-term outcomes for trauma survivors and in some cases affect long-term outcomes more than the severity of the index injury.^3,10-12 Therefore, it is essential to measure SDOH alongside QOL. Understanding trauma survivors’ QOL outcomes and SDOH context provides a more complete view of trauma recovery, as it highlights areas that can be targeted for intervention to improve long-term outcomes. It is critical to integrate QOL and SDOH measures into national injury databases to promote tailored strategies to improve long-term outcomes for trauma survivors.

In 2016, the National Academies of Sciences, Engineering, and Medicine (NASEM) released a seminal report on trauma care in the United States (US) that emphasized the importance of measuring long-term QOL outcomes in trauma survivors.^13,14 Following this report, the National Trauma Research Action Plan (NTRAP) assembled a Delphi panel to identify a core group of PROMs to include in trauma registries and national injury databases in order to measure QOL outcomes.^15,16 However, 8 y after the publication of the NASEM report calling for routine measurement of QOL in trauma survivors, it is unclear as to what degree national injury databases are following these guidelines. We therefore reviewed national injury databases to quantify the degree to which the databases measure long-term QOL outcomes. To contextualize the measurement of QOL outcomes, we evaluated whether the databases collect data using PROMs and measure SDOH. We also assessed whether these databases were capturing data across the full continuum of trauma care, ranging from prehospital care through rehabilitation and long-term recovery. We hypothesized that currently available national injury databases do not consistently capture longterm QOL outcomes, and their key determinants, across the full continuum of trauma care.

Methods

Databases

We included five national injury databases in our study: the Trauma Quality Program (TQP) database (version 1.0, 2019),¹⁷ Burn Care Quality Platform (BCQP) database (version 4.1, January 2024),¹⁸ Burn Model System database (version 7, 2023),¹⁹ Spinal Cord Injury (SCI) Model System database (version 2021 AR),²⁰ and Traumatic Brain Injury (TBI) Model System database (2019).²¹ These databases were selected because they were previously identified as the core group of national injury databases,²² and they are commonly used by injury researchers. Although there are international trauma databases that measure QOL, we did not include international databases in this study because our goal was to assess the current state of injury databases in the US. TQP is administered by the American College of Surgeons’ Committee on Trauma and is based on the National Trauma Data Standard; it supersedes the National Trauma Data Bank (NTDB) and Trauma Quality Improvement Program (TQIP) databases. BCQP measures the quality of burn care and is administered by the American Burn Association. The Model Systems databases consist of data collected by specialized injury-specific treatment and rehabilitation centers across the US. These centers are funded by the National Institute on Disability, Independent Living, and Rehabilitation Research and provide coordinated systems of rehabilitation care and conduct research on recovery and long-term outcomes. There are 4 centers in the Burn Model System, 18 centers in the SCI Model System, and 16 centers in the TBI Model System. Because all data dictionaries are publicly available and do not contain any patient-specific information, this study was exempt from institutional board review.

Characterization of database variables

We reviewed each variable in each database and coded whether the variable measured QOL, used PROMs, or measured SDOH. These categories were coded independently and were not mutually exclusive (i.e., QOL variables could be measured using PROMs). QOL was coded using the World Health Organization (WHO) definition: “individuals’ perceptions of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards and concerns.”²³ Variables that were coded as QOL were additionally categorized into five domains based on the WHO QOL domain definitions: Physical, Psychological, Level of Independence, Social Relationships, and Environment.²³ If the variable captured information about more than one QOL domain, then the domain was coded as “Multiple.” While QOL is typically measured using PROMs, we did not require that variables use a validated PROM in order to be categorized as QOL as long as the variable captured the patient’s perspective of their QOL.

We coded variables as using a PROM if the variable contained information that was collected using a standardized, validated instrument that obtains information “directly from the patient without interpretation of the patient’s response by a clinician or anyone else.”²⁴ PROMs were categorized into domains using a categorization system defined by NTRAP: Mental Health, Physical Health, Social Health, Cognitive Health, and Quality of Life.¹⁵ If a PROM included information across multiple NTRAP domains, it was coded as “Quality of Life,” since this was the broadest NTRAP domain. Importantly, while PROMs are frequently used to measure QOL, the coding process for QOL and PROMs was performed separately, and it is possible that certain variables may be coded as a PROM variable but not a QOL variable (e.g., if the variable uses a PROM to quantify substance use but does not capture whether the patient’s QOL was impacted by the substance use).

SDOH was coded based on the WHO definition: “the non-medical factors that influence health outcomes,” which include “the conditions in which people are born, grow, work, live, and age, and the wider set of forces and systems shaping the conditions of daily life.”²⁵ Variables that contained SDOH information were also categorized in domains based on the US Department of Health and Human Services’ Healthy People 2030 project.²⁶ These domains were: Economic Stability, Education Access and Quality, Healthcare Access and Quality, Neighborhood and Built Environment, and Social and Community Context.

Since trauma care occurs across a continuum, ranging from prehospital care through post-injury rehabilitation and reentry into the community, we also characterized the phase of care of all variables. The prehospital phase of care included all information about care provided by emergency medical services (EMS) or lay responders prior to arrival to the hospital. The emergency department (ED) phase of care included all information about care provided in the emergency department prior to admission to the hospital. The hospital phase of care included all information about care in the hospital, including surgical procedures and care provided in an intensive care unit. The inpatient rehabilitation phase of care included information about care provided in an inpatient rehabilitation facility and outcomes measured while the patient was in an inpatient rehabilitation facility. Finally, we also assessed the measurement of long-term outcomes at three different timepoints: discharge to 30 d, 30 d to 1 y, and 1 y or more.

Coding process

We created a codebook to guide the process of categorizing the variables within each database. The codebook is provided as a Supplemental File. Seven researchers (AO, KG, SI, IA, KW, LS, RH) served as independent coders. Each researcher was trained on the codebook using 5% of variables for testing. Once 90% intra-rater reliability was achieved among each member of the research team, each variable in the databases was independently coded by two researchers. Any discrepancies were resolved in consensus meetings by reviewing the codebook definitions. A lead researcher (AO or KG), in consultation with a subject matter expert (JHE), served as the tiebreaker as needed.

Statistical analysis

The primary outcome was the measurement QOL within the included databases. Secondary outcomes included the use of PROMs, collection of SDOH measures, and collection of data across different phases of care. We reported the raw number of variables within each database that measured QOL, PROMs, and SDOH. We also calculated descriptive statistics by assessing the percentage of total variables in each database that captured data about QOL, PROMs, or SDOH. When describing the percent of variables related to specific domains of QOL, PROMs, and SDOH, we calculated descriptive statistics out of the total number of QOL, PROM, or SDOH variables, respectively, within each database. We compared the proportion of QOL, PROM, and SDOH variables between the databases using the Chi squared test if cell size was greater than 10 or Fisher’s exact test if the cell size was less than or equal to 10. We also described the proportion of variables in each database that related to the different phases of care and compared the proportion of variables relating to different phases of care between databases using similar methods. Statistical significance was determined based a threshold of P < 0.05. All statistical analyses were performed in STATA version 18 (StataCorp, College Station, TX).

Results

The included databases ranged in size from 127 variables for BCQP to 995 variables for SCI Model System. None of the databases collected data across the entire continuum of trauma care. TQP collected data at the following phases of care: prehospital (76 variables, 21% of all variables in the database), emergency department (97 variables, 27%), and hospital (113 variables, 32%) (Fig. & Supplemental Table 1). TQP did not collect information beyond the time of hospital discharge, such as rehabilitation care or long-term outcomes. In contrast, the Model Systems databases (i.e., Burn, SCI, TBI) collected information primarily 1 y or more from the time of injury and contained limited or no information during the prehospital, emergency department, or hospital phases of care. For example, the TBI Model System had 0 variables about prehospital care, 10 variables (2%) about emergency department care, and 11 variables (2%) about hospital care but had 302 variables (47%) about long-term outcomes occurring 1 y or more after the index injury.

Fig. –

Collection of variables across the continuum of trauma care in injury databases. Raw numbers indicate number of variables within each database that relate to the specific phase of care. Percentages are out of total number of variables in each database.

Measurement of QOL and use of PROMs

QOL was measured by 0 variables in TQP, 1 variable (0.8%) in BCQP, 289 variables (33%) in the Burn Model System database, 159 variables (25%) in the TBI Model System database, and 115 variables (12%) in the SCI Model System database (Table 1). The single variable in the BCQP that measured QOL assessed whether the patient had returned to their pre-burn injury activities. In the Model Systems databases, QOL was most commonly measured 1 y or more after injury, although the SCI Model System collected QOL information at the inpatient rehabilitation phase of care in addition to 1 y or more after injury (Table 2).

Table 1 –

Measurement of quality of life, patient-reported outcome measures, and social determinants of health across national injury databases.

	Trauma quality program	Burn care quality platform	Burn model system	TBI model system	SCI model system	P value
	N= 356 variables	N = 127 variables	N = 865 variables	N = 647 variables	N = 995 variables
Quality of life, n (%)	0 (0%)	1 (0.8%)	289 (33%)	159 (25%)	115 (12%)	<0.001
PROM, n (%)	0 (0%)	0 (0%)	315 (36%)	153 (24%)	143 (14%)	<0.001
Social determinant of health, n (%)	12 (3.4%)	8 (6.3%)	74 (8.6%)	45 (7.0%)	38 (3.8%)	<0.001

PROM = patient-reported outcome measure; TBI = Traumatic Brain Injury; SCI = Spinal Cord Injury.

Table 2 –

Measurement of quality of life across phases of care in national injury databases.

Phase of care*, n (%)†	Trauma quality program	Burn care quality platform	Burn model system	TBI model system	SCI model system
Baseline characteristics	0 (0%)	0 (0%)	34 (12%)	9 (5.7%)	0 (0%)
Prehospital	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Emergency department	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Hospital	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Inpatient rehabilitation	0 (0%)	0 (0%)	0 (0%)	0 (0%)	42 (37%)
Discharge to 30 d	0 (0%)	1 (100%)	19 (6.6%)	0 (0%)	0 (0%)
30 d to 1 y	0 (0%)	1 (100%)	236 (82%)	0 (0%)	0 (0%)
1 y or more	0 (0%)	0 (0%)	236 (82%)	150 (94%)	73 (64%)

^*Individual variables may relate to more than one phase of care.

^†Percentage is of total number of quality of life variables.

Most of the QOL instruments utilized by the Model Systems databases measured multiple domains of QOL within the same instrument (Table 3). The physical domain was assessed by the Burn and SCI Model Systems databases but not by the TBI Model System database. The psychological domain was measured by all Model Systems databases. The level of independence domain was measured in the TBI and SCI Model Systems but not in the Burn Model System. The environment and spirituality/religion domains were not assessed by any of the databases, and the social relationships domain was only assessed by the Burn Model System. Of the seven total domains of QOL, the TQP database measured zero domains, BCQP measured one domain, Burn Model System measured three domains, TBI Model System measured two domains, and SCI Model System measured three domains.

Table 3 –

Measurement of quality of life domains in national injury databases.

Domain, n (%)*	Trauma quality program	Burn care quality platform	Burn model system	TBI model system	SCI model system
Physical	0 (0%)	0 (0%)	34 (12%)	0 (0%)	3 (2.6%)
Psychological	0 (0%)	0 (0%)	54 (19%)	27 (17%)	48 (42%)
Level of independence	0 (0%)	1 (100%)	0 (0%)	51 (32%)	43 (37%)
Social relationships	0 (0%)	0 (0%)	8 (2.8%)	0 (0%)	0 (0%)
Environment	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Spirituality, religion, and personal beliefs	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Multiple domains	0 (0%)	0 (0%)	193 (67%)	81 (51%)	21 (18%)

TBI = Traumatic Brain Injury; SCI = Spinal Cord Injury.

^*Percentage is of total number of quality of life variables.

QOL was measured in the Model Systems databases using PROMs. Each Model Systems database used multiple PROMs, ranging from 29 unique PROMs in the Burn Model System database to 7 unique PROMs in the TBI Model System database (Supplemental Table 2). Many of the PROMs utilized by the Model Systems databases were specific to one type of injury. For example, the SCI Model System used the Spinal Cord Injury-Quality of Life instrument, the TBI Model System used the Glasgow Outcome Scale-Extended instrument, and the Burn Model System used the Children Burn Outcomes Questionnaire. The only PROM that was used across all the Model Systems databases was the Satisfaction with Life Scale. The TQP and BCQP databases did not use any PROMs.

Measurement of social determinants of health

SDOH were measured by all databases, ranging from 12 variables (3%) in TQP to 74 variables (9%) in Burn Model System. All databases assessed the economic stability and social and community context domains of SDOH (Table 4). Education access and quality was assessed in each of the Model Systems databases but was not assessed in the TQP or BCQP databases. Of the five total domains of SDOH, the TQP database measured two domains, BCQP measured three domains, Burn Model System measured four domains, TBI Model System measured four domains, and SCI Model System measured all five domains.

Table 4 –

Measurement of social determinants of health by domain in national injury databases.

Domain, n (%)*	Trauma quality program	Burn care quality platform	Burn model system	TBI model system	SCI model system
Economic stability	6 (50%)	4 (50%)	37 (50%)	18 (40%)	9 (24%)
Education access and quality	0 (0%)	0 (0%)	12 (16%)	8 (18%)	3 (7.9%)
Health care access and quality	0 (0%)	0 (0%)	0 (0%)	0 (0%)	9 (24%)
Neighborhood and built environment	0 (0%)	1 (13%)	13 (18%)	14 (31%)	13 (34%)
Social and community context	6 (50%)	3 (38%)	12 (16%)	5 (11%)	4 (11%)

TBI = Traumatic Brain Injury; SCI = Spinal Cord Injury.

^*Percentage is of total number of social determinants of health variables.

Discussion

The need to measure long-term QOL outcomes in trauma survivors has been well established since NASEM’s 2016 report on trauma care in the US.¹³ To determine the extent to which national injury databases capture long-term QOL outcomes, and their key determinants, across the continuum of trauma care, we performed a systematic assessment of these databases. We found that current national injury databases are across the full continuum of trauma care and do not consistently measure long-term QOL outcomes. Some databases like TQP and BCQP collect information about ED and hospital-level care but lack data on long-term outcomes. In contrast, the Model Systems databases capture extensive data on long-term QOL outcomes but lack detailed information about characteristics of prehospital, ED, and hospital-level care. Furthermore, while PROMs are utilized by the Model Systems databases, there is minimal overlap in the specific PROMs used across Model Systems databases, which makes it challenging to assess and compare outcomes across different types of injury. Our results highlight that long-term QOL outcomes are not consistently assessed for trauma patients across the care continuum.

While the need to measure long-term outcomes for trauma survivors has been well established for nearly a decade,^2,14,27,28 national injury databases do not consistently measure long-term outcomes. A previous study of US state trauma registries performed in 2016 found that although a third of state registries captured post-acute discharge data, none of the registries collected data beyond 30 d of hospital discharge.²⁸ Our study extends this work to national injuries databases and finds that although Model Systems databases collect robust data on long-term outcomes for patients with specific injury patterns, there is no national database measuring long-term outcomes for survivors of polytrauma. To address this shortcoming, regional databases have been created to assess the long-term functional recovery of trauma survivors. For example, the Functional Outcomes and Recovery After Trauma Emergencies (FORTE) database captures long-term functional and QOL outcomes for trauma survivors treated at three level I trauma centers in Boston.³ The FORTE database could serve as a model for a national trauma survivorship database.

Further complicating data collection is the fact that injured patients receive care in multiple settings, and it is important for injury databases to collect data on care provided and clinical outcomes across each of these phases of care. Initial care provided in the field by EMS clinicians, surgical procedures performed in the hospital, and rehabilitation care provided by occupational and physical therapists are likely to play important roles in shaping a patient’s long-term recovery, but no single database currently captures the full breadth of care provided to polytrauma patients. Therefore, it is impossible to assess the role of EMS or ED care on long-term outcomes using national injury databases. Similar fragmentation in data collection across the care continuum has been observed in burn-specific databases.²⁹

One way to ensure that data is captured across all phases of care is by linking databases. The National Emergency Medical Services Information System (NEMSIS) database, a national database that contains prehospital care records submitted by EMS agencies, is in the process of creating a standardized system to link prehospital data with TQP and state trauma registries using a uniform patient identifier.³⁰ Integration of these two databases would allow researchers to evaluate the impact of prehospital trauma care on hospital-level outcomes, which is not currently possible on a national level. Future work could link TQP data to databases assessing rehabilitation care and long-term outcomes like the Model Systems databases to fully assess the quality of all phases of trauma care. Another effort in this space is the National Trauma Research Repository (NTRR), which aims to collate trauma research data from a variety of different studies into a single repository, with a goal of having data available across the entire continuum of trauma care.³¹ The process of database linkage is not without challenges, including concerns about patient privacy given the need to identify patients in order to perform linkage and limited funding to link databases and collect long-term outcomes. However, the expected cost of database linkage is substantially lower than the cost involved with creating a new database, and prior work linking EMS data with trauma registries has demonstrated the feasibility of this approach.³² Linking national injury databases has the potential to transform trauma research by providing a holistic view of patients’ complete trajectory of trauma care, including long-term outcomes.

Creating common data elements across injury databases is another important approach to reducing fragmentation in data collection. Prior studies of national injury databases have identified that the collection of even routine demographic data is heterogenous between databases.²² For example, national injury databases use different data values to describe patient sex, race and ethnicity, and medical comorbidities,^22,29 which prevents individual databases from being compiled. While our study did not directly examine the collection of standardized data elements, we did identify substantial variation in the use of PROMs and SDOH measures. For example, there was only one PROM that was consistently used across all the Model Systems databases, and there was no consistency in SDOH measures across databases. While there is a national consensus regarding a core group of PROMs to assess in trauma survivors,^15,16 our study demonstrates that these core outcome measures have not been widely adopted by national injury databases. In addition, a similar consensus on core SDOH measures for trauma survivors has not been reached. The measurement of SDOH data is incredibly important for trauma survivors given the growing body of research demonstrating that, in many cases, SDOH affect long-term outcomes more than the severity of the index injury.^10-12 Furthermore, SDOH data provide important context that can help clinicians provide trauma informed care to patients. Creating a standard set of SDOH measures for the trauma community would help to homogenize data elements. The process of creating a core set of SDOH measures for trauma research would likely require a Delphi panel, similar to the process used to identify the core set of PROMs.

The biggest barrier to adopting a national injury database that captures long-term QOL outcomes is cost, and the funding source plays a role in shaping the breadth and content of current national injury databases. The Model Systems and TQP have distinct funding models. The Model Systems databases are supported through funding from the National Institute on Disability, Independent Living, and Rehabilitation Research, which allows the Model Systems to follow patients for years after the index injury but limits the total number of participating centers. In contrast, the collection of TQP data is funded by individual trauma centers, which support the salary of the trauma registrars responsible for abstracting the data. Because contributing data to the TQP database is a requirement for trauma center verification through the American College of Surgeons, a large number of trauma centers across the US contribute data to TQP, but data is not recorded after the time of hospital discharge. Collecting data using PROMs can be labor-intensive, often requiring researchers to speak with patients via the phone or in person. While automated computer-based methods may facilitate the collection of PROM data, lower response rates with this method of data collection remain a barrier.³³ The first step toward establishing a national database that collects longterm QOL data for survivors of polytrauma will be obtaining appropriate funding to allow for the robust collection of longterm outcomes.

Limitations

Our study has several limitations. There is inevitably some degree of subjectivity in the coding of database variables. To limit subjectivity, we created a codebook with detailed definitions for our outcomes of interest. We trained all coders on the codebook definitions and did not begin coding until all researchers had 90% intra-rater reliability. Percentages used to describe study outcomes for measurement of QOL, use of PROMs, and measurement of SDOH are out of the total number of variables in the data dictionary. However, there is no gold standard for what constitutes an “adequate” proportion or raw number of QOL, PROM, or SDOH variables in a database. Furthermore, there is variation in how databases are organized, which makes it difficult to draw direct comparisons between databases. Although there are established domains of QOL and SDOH, there is no gold standard about which specific domains need to be assessed. Finally, we did not include administrative databases, including those from Medicare or the Healthcare Cost & Utilization Project, in this study. Although these databases are used by the trauma research community, particularly to study trauma systems, we chose not to include them in our study because their data are obtained through insurance claims, and they were not part of the previously defined group of core injury databases.²²

Conclusions

In this systematic assessment of national injury databases, we identified that there is no standardized system for capturing long-term QOL outcomes for polytrauma patients. Current national injury databases do not capture information across the entire continuum of trauma care, and there is minimal overlap in specific PROMs and SDOH measures across databases. Our findings highlight the need to either add long-term QOL outcomes to the TQP database or create a new Model Systems database for polytrauma linked to the TQP database.

Supplementary Material

Supplementary data related to this article can be found at https://doi.org/10.1016/j.jss.2025.05.027.