Adding Infectious Insult to Traumatic Injury: The Impact of Infectious Complications in End-of-Life Decision Making

Abstract

Background:

Trauma increases the risk for infection, but it is unknown how infection affects goals-of-care (GOC) decision making. We sought to determine how infections impact transition to comfort measures only (CMO), hypothesizing that infectious complications would expedite withdrawal of life-sustaining treatment (WOLST).

Patients and Methods:

We performed a retrospective review at a level-one trauma center over two years for adult patients without pre-existing advance directives who were made CMO with length of stay longer than one day. Demographics, injuries, and hospital course including infections and the GOC timeline were collected. Patients were divided on the basis of infection development, defined as an infectious complication requiring antibiotics or more invasive intervention, with subgroup analysis comparing those with single versus multiple infections. The primary end point was time to death or discharge.

Results:

Two hundred thirty-two patients met inclusion criteria and 72 developed an infection. Pneumonia was the most common infection (53.8%). Although those in the infection group had no substantial difference in demographics or comorbidities, they had higher emergency department Glasgow Coma Scale (GCS; 14 vs. 13), lower rate of head injury (28.6 vs. 49%), and higher time to death or discharge (12 vs. 2 days). Goals-of-care discussions were initiated later based on time to first family meeting (7 vs. 1 days), most occurring after the first infection. Subsequent analysis showed that versus those with a single infection (n = 38), those with multiple infections (n = 34) had a higher time to death or discharge (16.5 vs. 10.5 days) despite no difference in demographics, comorbidities, or trauma severity. Time to first family meeting was longer (8.5 vs. 4.5 days) with most occurring after the first infection.

Conclusions:

We did not find that development of an infection shortens time to WOLST. The increased time to death or discharge in the setting of multiple infections and similar patient populations may be a marker of provider approach to GOC plus family beliefs. Infectious complications play an uncertain role in end-of-life discussions after trauma.

Patients who sustain traumatic injuries are at a high risk of developing infectious complications. However, it is unknown how these infectious complications affect goals-of-care (GOC) decision making among trauma patients. Although the early stages of trauma care focus on aggressive measures, the later stages often require a more nuanced approach [1]. Some injured patients will progress to comfort measures when deemed appropriate by the patient, their family, and the care team [2]. The trauma team must be able to discuss not only the injuries but the patient's likelihood of recovery to an acceptable quality of life in the context of the patient's baseline health status and physiologic reserve.

Infection occurrence is a marker of a patient's underlying frailty as well as the severity of the injury burden [3 –6]. Recent studies have found a relationship between infection development and outcomes including mortality [6 -8]. Injured patients who die after the first few days of admission are most likely to succumb to a combination of organ dysfunctions, organ failures, and infections [9,10]. Because treatment of infectious complications may require invasive procedures and increase the risk of drug-resistant infections, it makes sense to include the occurrence of infections in GOC discussions.

Trauma provides a unique and challenging setting for such conversations given the abrupt nature of the event and because patients are often unable to participate actively [11]. Prior research into this area has shown that a variety of factors play a role in these decisions including the beliefs of the patient, family, and providers as well as the underlying injuries and clinical course [12 –15]. We sought to determine if infectious complications would impact the transition to comfort measures only (CMO), hypothesizing that development of an infection would decrease the time to death or discharge to hospice (TDD).

Patients and Methods

We performed a retrospective analysis of the trauma registry for all adult (age ≥18) patients admitted to Rhode Island Trauma Center, an American College of Surgeons accredited level one trauma center, from April 2015 to June 2017. Those patients who were made CMO, had a length of stay (LOS) longer than one day, and did not have a pre-existing do-not-resuscitate (DNR) or do-not-intubate (DNI) order were included in the study population. The latter were identified based on inclusion of advance directive limiting care within their listed comorbidities in the registry. Registry data were also used to determine patient demographics, baseline health status including comorbidities, injury pattern, and overall hospital course, whereas individual chart reviews were performed to evaluate for any infectious complications and for the initiation of GOC discussions.

An infection was determined to have occurred for any patient who received a full course of antibiotic agents or underwent further intervention (e.g., percutaneous or open drainage) for a documented infectious complication. However, if the antibiotic agents were discontinued shortly after initiation because there was no longer any concern for a true infection, these patients were not considered to have an infection for the purposes of this study. For those who met this threshold, the type of infection and hospital day of development were recorded. The provider team's overall approach to GOC as well as their response to an infectious complication was hypothesized based on the time to first family meeting and whether these discussions were initiated before or after the first documented infection.

The patients were divided into two primary groups: those who developed any infection and those who did not develop an infection, referred to here as the no infection group. Further subgroup analysis was performed within the any infection group to compare those with a single infection to those with multiple infections. All infections were considered equal without any stratification based on complexity or need for intervention. The primary end point was time to death or discharge.

Statistical analysis

Continuous variables and categorical variables were each analyzed using independent two-sample t-tests and χ² tests, respectively. When appropriate, non-parametric tests were used for analysis. All analysis was performed using SPSS Statistics, version 26 (SPSS Inc., Chicago, IL).

Results

A total of 7,516 patients were admitted to Rhode Island Trauma Center during the study period, with 232 (3%) of those patients meeting inclusion criteria. Among this cohort of 232 patients, approximately one-third (72 patients) had a documented infection during their admission, representing a total of 104 infections. Overall, the most common infection was pneumonia (53.8%) with urinary tract infections (UTI) second (18.3%); this pattern remained true for each study group (Table 1). Patients were in the hospital for a median of five days (interquartile range [IQR], 3–8) prior to developing an infection.

Table 1.

Distribution of Infection Type in Trauma Patients Who Underwent Withdrawal of Life-Sustaining Treatment during Admission, by Study Group

Type of infection	Any infection	Single infection	Multiple infections
Type of infection	n = 72	n = 38	n = 34
Pneumonia	53.8%	73.7%	42.4%
Urinary tract infection	18.3%	15.8%	19.7%
Blood	11.5%	7.9%	13.6%
Surgical site infection	6.7%	0%	10.6%
Clostridium difficile infection	5.8%	2.6%	7.6%
Burn site infection	2.9%	0%	4.5%
Osteomyelitis	1%	0%	1.5%

There was no difference in baseline demographics between those in the any infection and no infection groups in terms of gender (72.2% vs. 65.6% male; p = 0.32), age (70 vs. 69.8 years; p = 0.94), or ethnicity (90.3 vs. 92.6% Caucasian; p = 0.35) (Table 2). There was no difference in the total number of comorbidities between groups (2.4 [standard deviation {SD} 1.06] vs. 2.2 [SD 1.14]; p = 0.28) and no association with specific comorbidities including dementia (18.1 vs. 10%; p = 0.09). Although the no infection group had a lower emergency department Glasgow Coma Scale (GCS; 13 [IQR, 3 –15] vs. 14 [IQR, 8 –15]; p < 0.05) and higher rate of head injury (49 vs. 28.6%; p < 0.05), both groups had a similar traumatic profile with predominantly blunt mechanisms of injury (90% vs. 83.4%; p = 0.19) and no significant difference in median injury severity score (ISS) (25 [IQR, 17–26] vs. 17 [IQR, 10–28]; p = 0.91). Despite these comparable patient populations and injury patterns, those in the any infection group had a longer time to first family meeting (7 [IQR, 2 –14] vs. 1 ([QR, 0–3] days; p < 0.05) and time to death or discharge (12 [IQR, 8–21] vs. 2 [IQR, 1–5] days; p < 0.05). Only 31% of family meetings were held before the patient developed their first infection.

Table 2.

Patient Characteristics, Trauma Severity, Hospital Course for No Infection versus Any Infection Groups

	No infection	Any infection	p
	n = 160	n = 72	p
Demographics and baseline health
Age in years, mean (SD)	69.8 (19.4)	70 (17.9)	0.94
Male gender, percent	65.6	72.2	0.32
Caucasian, percent	93.8	90.3	0.35
Comorbidities, percent
Any substance abuse	23.8	33.3	0.13
Dementia	10	18.1	0.09
Congestive heart failure	12.5	16.7	0.4
Hypertension	47.5	56.9	0.18
Diabetes	23.8	26.4	0.67
Chronic obstructive pulmonary disorder	8.1	15.3	0.1
Total number of comorbidities, mean (SD)	2.2 (1.1)	2.4 (1.1)	0.28
Nature of trauma
Mechanism of Injury, percent
Fall	69.4	65.3	0.19
Motor vehicle accident	20.6	18.1
Penetrating	3.8	1.4
Injury severity score, median (IQR)	25 (17–26)	17 (10–28)	0.91
Emergency department GCS, median (IQR)	13 (3–15)	14 (8–15)	< 0.05
Head injury, percent	49	28.6	< 0.05
Hospital course	3 (2–6)	13 (9–22)	< 0.05
Hospital LOS, median (IQR)	3 (2–6)	13 (9–22)	< 0.05
ICU LOS, median (IQR)	3 (2–6)	12 (7–21)	< 0.05
Time to first infection (d), median (IQR)		5 (3–8)
Time to first family meeting (d), median (IQR)	1 (0–3)	7 (2–14)	< 0.05
Family meeting prior to first infection, percent		31.4
Time to death or discharge (d), median (IQR)	2 (1–5)	12 (8–21)	< 0.05

SD = standard deviation; IQR = interquartile range; LOS = length of stay; ICU = intensive care unit; GSC = Glasgow Coma Scale.

A secondary analysis of the any infection group was performed based on those who developed a single infection and those who developed multiple infections prior to being made CMO. Those in the multiple infection group had a longer time to death or discharge (16.5 [IQR, 9.5–33.3] vs. 10.5 [IQR, 6.8–14.3] days; p < 0.05) without a difference in demographics or type of trauma experienced (Table 3). The two groups were of similar age (69.3 vs. 70.6 years; p = 0.76), gender (73.5% vs. 71.1% male; p = 0.82), and ethnicity (85.3% vs. 94.7% Caucasian; p = 0.18). Both groups had a similar baseline health status with no difference in average number of comorbidities (2.4 [SD 1.2] vs. 2.4 [SD 0.95]; p = 0.89) and no association with specific comorbidities including rate of dementia (8.8% vs. 26.3%; p = 0.06). The nature of the traumatic event was also similar in both groups as demonstrated by the ISS (21 [IQR, 15–28] vs. 17 [IQR, 10–28]), emergency department GCS (14 [IQR, 11 –15] vs. 14 [IQR, 8 –15]), and rate of head injury (27.3% vs. 29.7%; p = 0.82). Once again, there was a longer time before initiation of GOC discussions in the multiple infection group (8.5 [IQR, 3–30.3] vs. 4.5 [IQR, 2 –9] days; p < 0.05) with even fewer being initiated prior to diagnosis of an infection (23.5% vs. 38.9%; p = 0.17).

Table 3.

Patient Characteristics, Trauma Severity, Hospital Course for Single Infection versus Multiple Infection Groups

	Single infection	Multiple infections	p
	n = 38	n = 34	p
Demographics and baseline health
Age in years, mean (SD)	70.6 (18.5)	69.3 (17.4)	0.76
Male gender, percent	71.1	73.5	0.82
Caucasian, percent	94.7	85.3	0.18
Comorbidities, percent
Any substance abuse	26.3	41.2	0.18
Dementia	26.3	8.8	0.06
Congestive heart failure	13.2	20.6	0.4
Hypertension	57.9	55.9	0.86
Diabetes	28.9	23.5	0.6
Chronic obstructive pulmonary disease	13.2	17.6	0.6
Total number of comorbidities, mean (SD)	2.4 (0.95)	2.4 (1.2)	0.89
Nature of trauma
Mechanism of Injury, percent
Fall	71.1	58.8	0.45
Motor vehicle accident	18.4	17.6
Penetrating	0	2.9
Injury severity score, median (IQR)	17 (10–28)	21 (15–28)	0.81
Emergency department GCS, median (IQR)	14 (8–15)	14 (11–15)	0.82
Head Injury, percent	29.7	27.3	0.82
Hospital course	11.5 (7.8–15.3)	17.5 (10.5–34.3)	< 0.05
Hospital LOS, median (IQR)	11.5 (7.8–15.3)	17.5 (10.5–34.3)	< 0.05
ICU LOS, median (IQR)	11 (6–13.5)	17 (9–34)	< 0.05
Time to first infection (d), median (IQR)	4 (2.3–6)	5.5 (4–9.3)	0.16
Time to first family meeting (d), median (IQR)	4.5 (2–9)	8.5 (3–30.3)	< 0.05
Family meeting prior to first infection, percent	38.9	23.5	0.17
Time to death or discharge (days), median (IQR)	10.5 (6.8–14.3)	16.5 (9.5–33.3)	< 0.05

SD = standard deviation; IQR = interquartile range; GSC = Glasgow Coma Scale; LOS = length of stay; ICU = intensive care unit.

Discussion

Our results show that development of an infection after injury did not accelerate withdrawal of life-sustaining treatment and time to death or discharge. To our knowledge this is one of the first investigations assessing the role infection occurrence plays in end-of-life decision making and outcomes. Most patients sustained severe traumatic injuries requiring management in an intensive care unit and a mean age of almost 70 years further bolsters the value of the study as those of advanced age that suffer injury are more likely to develop infections that lead to significant morbidity and mortality. Our cohort also had substantial rates of dementia and cardiopulmonary comorbidities, which are known to be risk factors for infections after admission [16].

Contrary to our hypothesis, we identified that despite no substantial difference in patient characteristics including baseline health status and severity of traumatic injuries, those who developed infections during their hospitalization had a longer period of time before they ultimately died or were discharged to hospice. This same pattern was seen in the subgroup analysis evaluating those who developed multiple infections rather than just a single infection. Although there was a higher rate of head injury among those who did not develop an infection as well as a shorter time to death or discharge, this is likely because of how head injuries and their impact on functional outcomes are perceived by patients, families, and providers and subsequently affect end-of-life decision making. Otherwise, across the study groups, there was a direct relation between the number of infections developed and the time to death or discharge. Although there is no question that a longer hospital LOS independently increases the risk of developing an infection, the fact that the characteristics of the study groups were so similar supports the concept that additional, non-medical, factors are playing a substantial role in the decision making process.

Specifically, this delay could be representative of how the occurrence of an infection is perceived by the patient and family as well as the providers. Infections are often viewed as an easily treatable condition that represents a temporary setback in the hospital course. Given that the vast majority of family meetings were initiated shortly after patients had developed their first infection, it can be inferred that providers recognize this as a complication worthy of further discussion with the patient and family. However, waiting until after the occurrence of a complication or setback may not be adequate considering that the median time to first infection in this cohort was five days.

Current recommendations state that family meetings should be held within the first 72 hours after admission for patients who are at risk of significant long-term disability or death as a result of their injuries [17]. In addition to diagnoses and prognosis, these conversations should include a discussion of complications that may occur during the patient's treatment course [18]. This more proactive approach would have given providers the opportunity to provide the patient/family with some context for the development of an infection, for example, and discuss its clinical significance. Planting the seed early means that the patient/family and the providers will be in agreement if an infection does occur rather than initiating these difficult discussions in a delayed fashion [19]. This pattern appears to be true within our study population given that the time to first family meeting, rate of post-infection initiation of GOC discussions, and time to death or discharge all share a direct relation to the number of infections developed despite similar patient populations and trauma severity.

Alternatively, differences in the decision making timelines and outcomes may be because of fundamental beliefs held by a patient and their family with regard to end-of-life issues. Prior studies have demonstrated that there are clear differences in how end-of-life aspects of medical care are perceived, and these perceptions are often influenced by socioeconomic, racial/ethnic, spiritual, and gender-based factors [14,20 –23]. For those patients and families who are resistant to withdrawal of life-sustaining treatment because of this underlying philosophy, the development of one infection—or even multiple infections—may not be reason enough to consider a change in treatment approach [24,25]. This can be a significant source of discord within the patient's support structure as well as between the patient/family and the providers, potentially threatening the therapeutic relation. As a result, it is important for providers to attempt to understand the patient/family's pre-conceived notions, core values, and those factors that are most important to them (e.g., control, suffering, greater meaning) in this process. With a more tailored and personalized approach to GOC discussions, providers may be able to improve outcomes as well as the overall experience for all involved.

Conclusions

This study seeks to evaluate the role that infectious complications play in GOC decision making in severely injured and critically ill trauma patients. Infections remain an important, although still underappreciated, complication in the care of the trauma patient. Our findings demonstrate that providers need to approach the GOC process proactively, leading directed conversations about infections and other potential complications as well as frank discussions of what impact they may have on quality of life and survival.

Footnotes

Acknowledgment

The contents of this manuscript do not represent the views of the U.S. Department of Veterans Affairs or the United States Government.

Funding Information

This work was supported, in part, by the National Institutes of Health [K08-GM110495] (to D.S.H.).

Author Disclosure Statement

The authors have no conflicts of interest to disclose.

References

Vincent

, Velkoff

. The Next Four

Decades

. The Older Population in the United States: 2010 to 2050. Washington, D.C.: U.S. Census Bureau; 2010.

Banks

, Lewis

. Trauma in the elderly: Considerations for anesthetic management. Anesthesiol Clin, 2013; 31:127–139.

Rai

, Stephen

, Lu

, et al. Impact of multi-drug–resistant pneumonia on outcomes of critically ill trauma patients. Surg Infect, 2020; 21:422–427.

Stephen

, Shah

, Sim

, et al. Dismal outcomes following damage control laparotomy in injured older adults, a cohort study. Am J Surg, 2019; 218:82–86.

Weintrob

, Murray

, Xu

, et al. Early infections complicating the care of combat casualties from Iraq and Afghanistan. Surg Infect, 2018; 19:286–297.

Monaghan

, Heffernan

, Thakkar

, et al. The development of a urinary tract infection is associated with increased mortality in trauma patients. J Trauma, 2011; 71:1569–1574.

Joseph

, Pandit

, Zangbar

, et al. Superiority of frailty over age in predicting outcomes among geriatric trauma patients: A prospective analysis. JAMA Surg, 2014; 149:766–772.

Min

, Burruss

, Morley

, et al. A simple clinical risk nomogram to predict mortality-associated geriatric complications in severely-injured geriatric patients. J Trauma Acute Care Surg, 2013; 74:1125–1132.

Sobrino

, Shafi

. Timing and causes of death after injuries. Proc (Bayl Univ Med Cent), 2013; 26:120–123.

10.

Sauaia

, Moore

, et al. Epidemiology of trauma deaths: A reassessment. J Trauma, 1995; 38:185–193.

11.

Trunkey

, Cahn

, Lenfesty

, et al. Management of the geriatric trauma patient at risk of death. Arch Surg, 2000; 135:34–38.

12.

K-H

, An

G-J

, Han

. Family factors affecting on withdrawal of life-sustaining treatment in Korea. Int J Nurs Pract. 2012; 18:552–558.

13.

Wooster

, Stassi

, Hill

, et al. End-of-life decision-making for patients with geriatric trauma cared for in a trauma intensive care unit. Am J Hosp Palliat Med, 2018; 35:1063–1068.

14.

Lissauer

, Naranjo

, Kirchoffner

, et al. Patient characteristics associated with end-of-life decision making in critically ill surgical patients. J Am Coll Surg, 2011; 213:766–770.

15.

Lueckel

, Kosar

, Teno

, et al. Outcomes in nursing home patients with traumatic brain injury. Surgery, 2017; 164:150–154.

16.

Castle

, Uyemura

, Fulop

, et al. Host resistance and immune responses in advanced age. Clin Geriatr Med, 2007; 23:463–479.

17.

ACS TQIP Palliative Care Best Practice

Guidelines

. ACS TQIP: Trauma Quality Improvement

Program

, 2017. http://www.facs.org/-/media/files/quality-programs/trauma/tqip/palliative_guidelines.ashx?la=en (Last accessed June 29, 2021).

18.

Wentlandt

, Weiss

, O'Connor

, et al. Palliative and end of life care in solid organ transplantation. Am J Transplant, 2017; 17:3008–3019.

19.

Seaman

. Improving care at end of life in the ICU: A proposal for early discussion of goals of care. J Gerontol Nurs, 2013; 39:52–58.

20.

Berger

, Pereira

, Baker

, et al. A commentary: social and cultural determinants of end-of-life care for elderly persons. Gerontologist, 2002; 42(Spec No 3):49–53.

21.

Wales

, Kurahashi

, Husain

. The interaction of socioeconomic status with place of death: A qualitative analysis of physician experiences. BMC Palliat Care, 2018; 17:87.

22.

Muni

, Engelberg

, Treece

, et al. The influence of race/ethnicity and socioeconomic status on end-of-life care in the ICU. Chest, 2011; 139:1025–1033.

23.

Garrido

, Idler

, Leventhal

, et al. Pathways from religion to advance care planning: Beliefs about control over length of life and end-of-life values. Gerontologist, 2013; 53:801–816.

24.

Burnett

, Johnston

, Kearney

, et al. Understanding factors that impact on public and patient's risk perceptions and responses toward Clostridium difficile and other health care-associated infections: A structured literature review. Am J Infect Control, 2013; 41:542–548.

25.

Hart

, Cabalo

, Bess

, et al. Comparison of patient and surgeon perceptions of adverse events after adult spinal deformity surgery. Spine (Phila Pa 1976), 2013; 38:732–736.