Beyond In-Hospital Mortality: Use of Post-Discharge Quality-Metrics Provides A More Complete Picture of Older Adult Trauma Care

Abstract

Objective:

To identify the distributions of and extent of variability among three new sets of post-discharge quality-metrics measured within 30/90/365 days designed to better account for the unique health needs of older trauma patients: mortality (expansion of the current in-hospital standard), readmission (marker of health-system performance and care coordination), and patients’ average number of healthy days at home (HDAH; marker of patient functional status).

Summary and background data:

Traumatic injuries are a leading cause of death and loss of independence for the increasing number of older adults living in the United States. Ongoing efforts seek to expand quality evaluation for this population.

Methods:

Using 100% Medicare claims, we calculated hospital-specific reliability-adjusted post-discharge quality-metrics for: older adults aged ≥65 years admitted with a primary diagnosis of trauma, older adults with hip fracture, and older adults with severe traumatic brain injury (TBI). Distributions for each quality-metric within each population were assessed and compared to results for in-hospital mortality, the current benchmarking standard.

Results:

A total of 785,867 index admissions (305,186 hip fracture, 92,331 severe TBI) from 3,692 hospitals were included. Within each population, use of post-discharge quality-metrics yielded a broader range of outcomes compared to reliance on in-hospital mortality alone. None of the post-discharge quality-metrics consistently correlated with in-hospital mortality, including death within one year (r=0.581[95%CI:0.554–0.608]). Differences in quintile-rank revealed that when accounting for readmissions (8.4%, κ=0.029) and patients’ average number of HDAH (7.1%, κ=0.020), as many as 1-in-14 hospitals changed from the best/worst performance under in-hospital mortality to the completely opposite quintile rank.

Conclusions:

Use of new post-discharge quality-metrics provides a more complete picture of older adult trauma care: one with greater room for improvement and better reflection of multiple aspects of quality important to the health and recovery of older trauma patients when compared to reliance on quality benchmarking based on in-hospital mortality alone.

Mini-Abstract

Ongoing efforts seek to expand quality evaluation of older adult trauma. This study presents an initial assessment of three new sets of proposed post-discharge quality-metrics measured within 30/90/365 days designed to better account for the unique health needs of older trauma patients. They include: mortality, readmission, and patients’ average number of healthy days at home.

Introduction

Traumatic injuries are a leading cause of death and loss of independence for the >54 million older adults aged ≥65 years living in the United States (US). In 2019, 4.8 million non-fatal traumatic injuries were reported among US older adults, of which 1.2 million required hospitalization.¹ An additional 68,000 older adults died during hospitalization as a result of their injuries, costing hospitals and payers a combined total >$68.0 billion per year.¹ By 2050, the number of US older adults is projected to rise to 89 million (22.1% of the total population),² with it is expected to come a parallel increase in the number of older adult trauma patients. Among older adult trauma patients, hip fracture and severe traumatic brain injury (TBI) are two of the most common and debilitating forms, often resulting in long-term functional impairment, nursing-home admission, and shortened life-expectancy. 1-in-5 older adults with hip fracture dies within one year of injury.³ 1-in-3 who lived independently before a hip fracture remains in a nursing-home for at least one year after injury,³ and 1-in-10 will require readmission within 30 days.⁴ For severe TBI, as many as 1-in-2 to 1-in-3 injured older adults dies within one year of injury.^5,6

Given these statistics, improving trauma outcomes for older adults is an important national priority. Meaningful approaches to evaluate older adult trauma are needed in order to gauge hospital quality and establish targets capable of best allocating limited resources and improving future care for the increasing number of older trauma patients. While external benchmarking (the use of agreed upon quality-metrics, typically derived from hierarchical regression models)^7–11 has evolved as the preeminent methodology used to compare hospital outcomes when measuring the quality of older adult care, its use within trauma is currently limited by a lack of available post-discharge data¹² and historical focus on in-hospital mortality as the singular measure of a hospital’s overall quality of trauma care.^13,14 Recognizing the limitations of this situation, in 2016, the National Academies of Sciences, Engineering, and Medicine (NASEM) called for efforts to expand external benchmarking within trauma to include additional post-discharge outcomes.¹⁵ Suggested post-discharge outcomes for older adults include the development of quality-metrics capable of assessing: patients’ extent of functional recovery and preference for increased time at home,¹⁶ the importance of care-coordination and successful care-transfers,^17,18 and patients’ risk of death post-discharge.

To accomplish this goal, one of the first steps needed is to understand the distributions of suggested quality-metrics. The ED.TRAUMA Study (Evaluating the Discordance of Trauma Readmission And Unanticipated Mortality in the Assessment of Hospital Quality) was designed to explore the implications of their potential use. The results presented in this paper represent the first part of the ED.TRAUMA Study’s work, looking at the distributions of and extent of variability among three new sets of proposed older adult trauma post-discharge quality-metrics measured within 30, 90, and 365 days. They include: mortality (expansion of the current in-hospital standard), readmission (marker of health-system performance and care-coordination), and patients’ average number of healthy days at home (HDAH; new administrative claims-based marker of patient functional status). Performance of the proposed quality-metrics among (1) all older adult trauma, (2) older adults with hip fracture, and (3) older adults with severe TBI was considered as were comparisons of each new quality-metric to results for in-hospital mortality.

Methods

Data source and study population

Records of index trauma admissions for older adult Medicare patients aged ≥65 years were abstracted using January 2014-December 2015 Centers for Medicare & Medicaid Services (CMS) 100% fee-for-service hospital inpatient claims. Inpatient, outpatient, and health-services utilization data from 12 months prior (January 2013-December 2015) were used to establish previous health-services utilization and existing comorbidities. Data through 12 months after (January 2014-December 2016) were used to determine subsequent utilization and time at home. Death certificate-linked fatalities were tracked through 2018.

To be included, patients needed to present with a principal diagnosis of trauma. Following the definition of trauma developed by the American College of Surgeons’ (ACS) National Trauma Data Standard,¹⁹ we excluded patients with diagnoses of late-effects of injury or poisoning, superficial injuries, foreign bodies, and burns. Patients without an underlying mechanism of injury reported as blunt or penetrating were also excluded. Presence of severe TBI was defined in accordance with the Centers for Disease Control and Prevention.²⁰ Traumatic hip fractures were defined in accordance with CMS.²¹ Index admissions were defined as those without a prior trauma hospitalization in the preceding 30 days. Included patients were limited to their first recorded index admission during the two-year study period. To ensure estimate stability, included hospitals were restricted to those with annual trauma volumes ≥10 older adult trauma cases.

Potential confounders

Potential confounders utilized in the calculation of reliability-adjusted (risk-adjusted) quality-metrics included: age in years on index admission, gender (categorized as male versus female), presence of severe head injury (maximum head Abbreviated Injury Scale [AIS] <3 versus ≥3), overall Injury Severity Score (ISS; categorized as minor ≤8, severe 9–15, and major ≥16), extent of multimorbidity (number of pre-existing conditions based on the Elixhauser Comorbidity Index), Hospital Frailty Risk Score (categorized as low risk ≤4, intermediate risk 5–15, and high risk ≥16), and individual binary comorbidity indicators contained within the Elixhauser Comorbidity Index reported on index admission or up to 12 months prior. ISS, AIS, Hospital Frailty Risk Score, and multimorbidity/comorbidities were calculated using ICD-9/10-CM primary/secondary diagnosis and external cause of injury codes. The icdpicr (ICD Programs for Injury Categorization in R) program in R was used to calculate ISS and AIS.^22,23

Reliability-adjusted (risk-adjusted) quality-metrics

Mortality was defined from the date of index admission. It included deaths occurring during index hospitalization, readmission, and at home.

Readmission was defined from the date of discharge among patients discharged alive. Patients discharged as transfers or who were discharged and readmitted within the same day (presumed transfers) were not counted as readmissions.

HDAH was defined as a hospital’s average number of days at home among patients discharged alive. It was measured from the date of discharge and calculated as the sum-total of patients’ time during that period less any time spent in the hospital or ED, step-down/rehabilitation/nursing-care, home-health, or after death.²⁴ Patients living in skilled nursing-facilities (SNFs) prior to index admission were excluded as were those discharged on hospice. For the purposes of comparison to readmission and mortality measured on a 0–100% scale, HDAH were reported as both hospitals’ average number of healthy days and the average percent of time within a 30/90/365-day period that patients spent healthy at home.

Using an approach defined by CMS to calculate risk-standardized readmission/mortality rates^7–11 and more recent recognition of the need to incorporate volume-based reliability-adjustment of risk-standardized rates,^25,26 hospital performance on included quality-metrics was calculated using multivariable hierarchical regression with hospital-level random intercepts and patient-level fixed-effects followed by reliability-adjustment. Analyzing data in this way accounted for clustering of patients within hospitals (hierarchal models), unstable estimates due to small sample-size (reliability-adjustment), and risk-adjustment for known confounders.

After calculating hierarchical models for each quality-metric (mortality, readmission, HDAH) at each timepoint (30, 90, 365 days) in each population (all trauma, hip fracture, severe TBI), pooled random-effects intercepts were used to determine hospitals’ expected rates.^7–11 Predicted rates were taken from random intercepts. Risk-standardized rates were then calculated as the ratio of predicted/expected rates multiplied by hospitals’ unadjusted rates.^7–11 To incorporate reliability-adjustment, we modeled a risk-standardization algorithm using an empirical-Bayes estimator.²⁷ To further allow for shrinkage of estimates closer to hospitals with similar older adult trauma volumes, observed volumes were included as an additional hospital-level fixed-effect.²⁵

Correlations with in-hospital mortality

To assess for potential correlations between in-hospital mortality (including patients who died during index hospitalization; excluding deaths on transfer and discharge to hospice) and each proposed post-discharge quality-metric, we calculated volume-weighted Pearson correlation coefficients. Estimates of linear correlations were volume-weighted in order to account for the amount of weight that should be given to each hospital based on the number of older adult trauma patients that it treated.²⁸ To identify potential nonlinear relationships, we fit volume-weighted generalized-additive models using reliability-adjusted in-hospital mortality as the dependent variable and a cubic-spline smoother of the other quality-metric as the independent variable.²⁸ To further determine if there was a difference in the quintile of hospital quality for each hospital, we categorized hospitals into five equal-sized groups based on their quintile of performance on each quality-metric and measured the magnitude of change in quintile (Cohen’s κ) when hospital quality was compared between reliability-adjusted in-hospital mortality and each other potential new quality-metric.¹⁸ The proportion of hospitals assigned to similar (|0|) or different (|1| to |4|) quintiles in each pair was calculated and reported.¹⁸

Initial data analysis and cleaning were conducted using Statistical Analysis Software: Version 9.4. Statistical analyses were conducted using Stata Statistical Software: Version 17.0. Graphs were plotted in R. Missing data were minimal (<1.0% across all considered confounders) and, when present, were addressed using multiple imputation. The study was approved by the institutional review boards of Brigham & Women’s Hospital and Yale University.

Results

Population characteristics and unadjusted outcomes

A total of 785,867 index trauma admissions (n=305,186 hip fractures, n=92,331 severe TBI) from 3,692 hospitals were included (Figure 1), representing all states and both large and small trauma hospitals across the US. Demographic characteristics are presented in eTable 1. The majority of included patients (mean age 82.1±8.5 years) were female (68.0%). Most were admitted with major injuries (ISS ≥16: 53.3%). Many had a history of multiple comorbidities (mean number 2.7±0.3; e.g., 20.9% diabetes, 18.1% chronic pulmonary disease); 11.9% had a documented severe head injury. Forty-six percent of patients (46.4%) presented with moderate frailty, while 22.5% presented with severe frailty.

Figure 1.

Geographic distribution of hospitals classified by annual trauma volume (size) and travel time by road to the nearest Level 1 (L1) or 2 (L2) Trauma Center (color)

Within 30 days of admission (eTable 1), 7.6% (n=59,962) of older adult trauma patients died (2.8% in-hospital mortality). By 90 days, mortality increased to 13.2% (n=104,050), and, by 365 days, mortality increased to 24.7% (n=193,797). Readmissions rose from 11.2% within 30 days of discharge (n=88,254) to 20.0% within 90 days (n=157,332) and 36.2% within 365 days (n=284,644). Following discharge, older adult trauma patients spent an average of 31.3% of their first 30 days healthy at home (mean 9.4±0.5 days). Within 90 days, they spent an average of 42.4% of their time healthy at home (mean 38.2±1.8 days). Within 365 days, they spent an average of 45.1% of their time healthy at home (mean 164.5±7.6 days).

Reliability-adjusted (risk-adjusted) mortality

Distributions of hospital-specific, reliability-adjusted (risk-adjusted) mortality are presented in Figure 2A. Within 30 days, reliability-adjusted mortality for trauma ranged from 4.6% to 10.1% (mean: 7.3±0.6%). Results for hip fracture were similar (mean: 7.6±0.4%), while 30-day mortality for severe TBI was higher with a mean of 17.0±0.8%. By 90 and 365 days, reliability-adjusted mortality increased (e.g., results for trauma: 90-day mean 15.3±1.0%; 365-day mean 25.9±1.9%), and quality-metric distributions became wider/more dispersed (90-day trauma range 10.2%–21.6%; 365-day trauma range 16.0%–37.0%). A total of 34 to 97 hospitals (representing 0.5% to 5.7% of index admissions) were more than 2 standard deviations above the mean, indicating clear outliers with adverse quality (i.e., higher mortality).^7–11

Figure 2.

Smoothed histogram (density plots) showing the distribution and variability of reliability-adjusted (A) mortality, (B) readmission, and (C) patients’ average number of healthy days at home (HDAH) within 30, 90, and 365 days among older adults admitted with trauma, hip fracture, and severe traumatic brain injury (TBI)

When compared to reliability-adjusted in-hospital mortality in an effort to assess for potential differences between longer-term follow-up and the current in-hospital mortality standard, reliability-adjusted 30-day mortality exhibited a strong positive linear correlation (Figure 3A, volume-weighted linear correlation coefficient [r, 95%CI]: 0.968[0.966–0.970]). Three-fourths of hospitals (74.7%) were assigned to the same quintile (Cohen’s κ=0.684; complete quintile results are presented in the Appendix). 90-day mortality exhibited a slightly weaker positive linear correlation (Figure 3B, r=0.907[0.899–0.915]) with 62.4% of hospitals assigned to the same quintile (κ=0.530; 34.2% differed by one quintile). By 365 days, the positive linear correlation began to fade, dropping to r=0.581 (Figure 3C, 95%CI: 0.554–0.608). 42.6% of hospitals were assigned to the same quintile (κ=0.283). 39.5% differed by 1 quintile, while 1-in-5 hospitals (17.9%) differed by ≥2 quintiles.

Figure 3.

Comparison of reliability-adjusted in-hospital mortality and mortality within (B) 30, (C) 90, and (D) 365 days.. Solid lines indicate mean values. Insets show non-linear volume-weighted generalized-additive models. Bars show the extent of quintile change (|0| to |4|).

Reliability-adjusted (risk-adjusted) readmission

Distributions of hospital-specific, reliability-adjusted (risk-adjusted) readmission are presented in Figure 2B. Within 30 days, reliability-adjusted readmission for trauma ranged from 8.6% to 18.2% (mean: 11.7±1.0%). Results for hip fracture (mean: 10.9±0.9%) and severe TBI (mean: 12.9±0.7%) were similar. By 90 and 365 days, reliability-adjusted readmission increased (e.g., results for all trauma: 90-day mean 21.1±1.8%; 365-day mean 33.0±2.2%), and quality-metric distributions became more dispersed (90-day trauma range 15.0%–22.0%; 365-day trauma range 24.8%–47.8%). A total of 40 to 93 hospitals (representing 0.3% to 1.3% of index admissions) were more than 2 standard deviations above the mean, indicating clear adverse quality (i.e., higher readmissions).^7–11

When compared to reliability-adjusted in-hospital mortality in an effort to assess for potential differences between readmission as a marker of health-system performance and care-coordination and the current in-hospital mortality standard, reliability-adjusted 30-day readmission showed no linear correlation (Figure 4A). One-fourth of hospitals (24.7%) were assigned to the same quintile, while 5.2% of hospitals were assigned to completely opposite quintiles (Cohen’s κ=0.059; complete quintile results are presented in the Appendix). 90-day readmission exhibited a weakly negative linear correlation (Figure 4B, volume-weighted r[95%CI]: −0.049[−0.010 to −0.088]) with 24.0% of hospitals assigned to the same quintile (κ=0.050). 32.9% of hospitals differed by one quintile, while 6.2% were assigned to completely opposite quintiles. Within 365 days, a more substantial negative linear correlation emerged (Figure 4C, volume-weighted r[95%CI] −0.221[−0.182 to −0.260]). 22.3% of hospitals were assigned to the same quintile (κ=0.029). 31.5% of hospitals differed by one quintile, while 8.4% were assigned to completely opposite quintiles.

Figure 4.

Comparison of reliability-adjusted in-hospital mortality and readmission within (B) 30, (C) 90, and (D) 365 days. Solid lines indicate mean values. Insets show non-linear volume-weighted generalized-additive models. Bars show the extent of quintile change (|0| to |4|).

Reliability-adjusted (risk-adjusted) average number of HDAH

Distributions of hospital-specific, reliability-adjusted (risk-adjusted) HDAH are presented in Figure 2C. Within 30 days, patients’ percentage of time spent healthy at home after trauma ranged from 26.4% to 36.2% (mean 30.2±1.0%, corresponding to 9.1±0.3 days). Results for severe TBI were similar (mean 30.8±0.9%, corresponding to 9.2±0.3 days), while results for hip fracture reflected patients’ increased use of post-discharge rehabilitation/SNF care within 30 days (mean 26.7±0.1%, corresponding to 8.0±0.2 days). By 90 and 365 days, patients’ percent of time spent healthy at home increased (e.g., results for all trauma within 90 days, mean 39.2±1.5%; within 365 days, mean 43.6±1.6%), and quality-metric distributions became more dispersed (90-day trauma range 32.3%–48.7%; 365-day trauma range 35.0%–53.1%). Higher post-discharge mortality among severe TBI patients (Figure 2C) yielded less time at home within 90 and 365 days. A total of 34 to 73 hospitals (representing 0.2% to 1.2% of index admissions) were more than 2 standard deviations below the mean, indicating clear adverse quality (i.e., less HDAH).^7–11

When compared to reliability-adjusted in-hospital mortality in an effort to assess for potential differences between patients’ average number of HDAH as a marker of patient functional status and the current in-hospital mortality standard, differences in reliability-adjusted HDAH within 30 days showed no linear correlation (Figure 5A). One-fifth of hospitals (21.6%) were assigned to the same quintile, while 7.1% of hospitals were assigned to completely opposite quintiles (Cohen’s κ=0.020; HDAH were reported as revered quintile rank so that Q5 represented the worst performance in both quality-metrics; complete quintile results are presented in the Appendix). Patients’ average number of HDAH within 90 days exhibited a moderately strong negative linear correlation (Figure 5B, volume-weighted r[95%CI]: −0.704[−0.686 to −0.722]), suggesting that at hospitals with lower in-hospital mortality (better quality), older adult trauma patients were more likely to spend more post-discharge time at home (better quality). One-half of hospitals (48.2%) were assigned to the same quintile (κ=0.353). 38.9% of hospitals differed by one quintile, while 0.3% were assigned to completely opposite quintiles. Within 365 days, a weaker negative linear correlation emerged (Figure 5C, volume-weighted r[95%CI] −0.329[−0.388 to −0.290]). 36.5% of hospitals were assigned to the same quintile (κ=0. 207). 37.5% of hospitals differed by one quintile, while 1.8% were assigned to completely opposite quintiles.

Figure 5.

Comparison of reliability-adjusted in-hospital mortality and patients’ average number of healthy days at home (HDAH) within (B) 30, (C) 90, and (D) 365 days.. Solid lines indicate mean values. Insets show non-linear volume-weighted generalized-additive models. Bars show the extent of quintile change (|0| to |4|).

Discussion

Additional post-discharge trauma quality-metrics for older adults are needed. As a first step in answering calls from NASEM¹⁵ and others^29,30 to move beyond in-hospital mortality, the results demonstrate that within a large national cohort of Medicare patients, assessment of post-discharge trauma quality is possible. Use of proposed reliability-adjusted trauma quality-metrics designed to account for differences in mortality (both before and after discharge), readmissions as a marker of health-system performance and care-coordination, and patients’ average number of HDAH as a marker of patient functional status within 30, 90, and 365 days provided a more complete picture of older adult trauma care. Initial assessment of their distributions showed that use of such quality-metrics critically: (1) remained consistent across all older adult trauma patients and two of older adult trauma’s most prevalent and devastating forms, (2) allowed for room for improvement among hospitals falling at the worst quality extremes and opportunities to learn from hospitals performing exceptionally well (wide range of outcomes for each quality-metric), and (3) captured different aspects of quality from those currently reflected by the use of in-hospital mortality alone (non-linear or weak correlations with reliability-adjusted in-hospital mortality).

Understanding the distributions of these proposed post-discharge quality-metrics and the extent to which they associate with in-hospital mortality will be critical in designing future efforts to expand external benchmarking in trauma. Future research, planned as an ongoing part of the ED.TRAUMA Study, is needed to next determine how the proposed quality-metrics should best be used and to ascertain the extent to which they associate with differences in hospital-level factors.

In the US, national efforts to benchmark trauma first emerged in 2009 following the introduction of the ACS Trauma Quality Improvement Program (TQIP)¹⁴ as an extension of the national cross-sectional trauma-registry, the National Trauma Data Bank (NTDB).¹³ Since that time, individual hospitals have made efforts to expand TQIP data to include “missing information” for outcomes as far out as 30 days.³¹ However, national quality-metrics remain limited to in-hospital data. The situation has created a dependence on in-hospital mortality and yielded mounting concern that in relying on in-hospital mortality, existing benchmarking efforts are insufficient. They fail to capture the full experience of older adult trauma patients during their extended course of recovery and care, leading to calls for change^15,29,30 and ongoing development of a National Trauma Research Action Plan.²⁹ In looking at a broader array of outcomes at additional timepoints throughout older adult trauma patients’ first year of recovery and examining the variable range of results, this study adds to that work, underscoring the discrepancy and lack of strong correlation between the, e.g., 2.8% of older adults who died during index hospitalization and 24.7% who died within 365 days.

In an ideal world, specialized trauma centers designed around the need to reduce adult in-hospital mortality (stemming from the work of the NSCOT trial, which looked at in-hospital mortality for patients aged 18–84 years)³² would also be able to optimize outcomes for older adults (average age in our study 82.1±8.5 years). However, emerging literature suggests that this might not always be true with current trauma systems potentially resulting in conversely higher mortality for older adults at larger level 1 and 2 trauma centers.³³ Whether such centers are able to provide better post-discharge outcomes in terms of readmission and HDAH among older adult trauma patients remains to be seen, but, it is possible, that failure to account for older adults’ longer-term recovery needs could be driving many of the differences observed between in-hospital mortality and other proposed forms of longer-term post-discharge quality-metrics.

Existing literature on the topic is limited; however, among older adults hospitalized for heart failure and acute myocardial infarction, ^28,34–36 lack of an association and/or a potential negative association between 30-day mortality and readmission is in keeping with expectations. Comparisons between in-hospital mortality and readmission in our study suggest that in addition to differences in mortality over time, variations in the type of outcome could also reflect important differences in quality. In 2014, an NTDB study of 248 trauma centers treating 450,000 adult trauma patients aged ≥16 years found that when classified by quintile of in-hospital mortality and major morbidity, only 21.0% of trauma centers were assigned to the same quintile.¹⁸ Among older adult trauma patients, the results of our study suggest evidence of a weak negative correlation between in-hospital mortality and readmission and marked differences in quality when hospitals were evaluated based on changes in quintile (≤24.7% of hospitals were assigned to the same quintile). Lack of a strong inverse association suggests that low in-hospital mortality is unlikely to be driving high post-discharge readmission rates.^28,34–36

Anticipated correlations between in-hospital mortality and patients’ average number of HDAH are largely unknown. HDAH was recently adapted for use as a quality-metric in Medicare in an effort to better account for patient-centered outcomes in a way that can be readily measured using existing billing claims.^24,37 It is thought to reflect patients’ extent of functional recovery and preference for a return to independence and increased time at home¹⁶ but has yet to be rigorously studied among trauma patients. Existing literature from emergency general surgery^38,39 and early reports in trauma⁴⁰ using various definitions for assessing “time at home” suggest that HDAH are likely to be associated with both patient experience and quality-of-life. Versions have been used in several related older adult populations, including recovery from stroke⁴¹ and elective (oncology) procedures.^42,43 While somewhat associated with mortality given how HDAH are defined (i.e. patients who die after discharge spend less time at home), the results of our study suggest that HDAH still appears to capture a novel aspect of quality—one that uniquely addresses patient experiences and realities encountered by diverse members of the trauma care team while avoiding many of the pitfalls faced when alternatively collecting information on functional status through the use of patient-reported outcome measures (e.g. poor response rates, substantial required investments in personnel/resources/time, reliance on caregiver report).³⁰ HDAH is, ultimately, an assessment of not just in-hospital quality and patient-status at discharge but of also how well hospitals help patients navigate the complex post-discharge continuum of care, address barriers to recovery, and coordinate follow-up to best optimize functional outcomes that matters most to patients¹⁶ while simultaneously reducing adverse outcomes traditionally focused on by payers, like the need for readmission⁴⁴ and related risk of repeat falls.

Used individually or in combination through the development of some form of composite measure, the proposed set of additional older adult trauma quality-metrics has the potential to address needed calls for the expansion of external benchmarking in trauma. Each was designed to capture different aspects of older adult trauma quality that differ from potentially troublesome reports based on the use of mortality alone.³³ Whether added to existing registries or adopted more broadly by public reporting for government organizations like the Veterans Health Administration or CMS, widely-available quantitative follow-up of older adult trauma patients offers a simple and needed means of expanding efforts to evaluate the quality of older adult trauma care.

For hospitals in integrated health systems, a more nuanced understanding of older adult trauma outcomes offers a direct measure of the success of the system in providing coordinated management critical in the recovery of older injured patients. Areas of deficit can be targeted and improved. For more isolated hospitals without ready access to an integrated health system and with correspondingly less control over what happens after discharge, an understanding of longer-term outcomes becomes even more critical as a means to: (1) empower hospitals to take responsibility for the coordinated care needed in the management of older adult trauma patients, (2) help identify areas in which such hospitals can improve post-discharge planning and care transitions, and (3) establish collaborative regional networks capable of acting as centers of excellence in the care of older trauma patients, many of whom do not present to larger trauma centers.⁴⁵

The study has limitations. The most important reflect its reliance on administrative claims. Use of Medicare data allowed for a large national assessment of older adult trauma patients, including the majority of older adults and hospitals across the US. However, in relying on 100% Medicare fee-for-service claims, we were not able to capture the outcomes of older adults with other forms of insurance or those enrolled in managed care. Quality-metrics used in this study followed established methodology for the calculation of reliability-adjusted outcomes.^7–11,25,26

As efforts to understand trauma quality begin to expand toward the use of additional post-discharge quality-metrics, meaningful approaches are needed in order to gauge hospital quality and establish targets capable of best directing limited resources and improving future care for the increasing number of older trauma patients. The results of our study show that when evaluating older adult trauma quality based on reliability-adjusted mortality, readmission, and patients’ average number of HDAH, each type of outcome and timepoint captured different aspects of patients’ recovery from that reflected based on the use of in-hospital mortality alone. As a first step in answering calls from NASEM¹⁵ and others^29,30 to move beyond in-hospital mortality, the results point toward important opportunities to increase ways of looking at trauma and better capture the experience of older trauma patients. The proposed set of additional post-discharge quality-metrics provided a more complete picture of older adult trauma care: one that reflected ongoing changes during patients’ first year of recovery, accounted for care-transitions and multiple provider perspectives, and included patient-centered outcomes all in a way that could be easily measured using existing billing claims. Future research is needed to determine how they should best be used.