Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Trauma Acute Care Surg. Author manuscript; available in PMC 2014 July 1.
Published in final edited form as:
PMCID: PMC3744055

Do Not Resuscitate Orders in Trauma Patients May Bias Mortality-Based Effect Estimates: an evaluation utilizing the PROMMTT study

Charles E Wade, PhD,1 Deborah J. del Junco, PhD,1 Erin E Fox, PhD,2 Bryan A Cotton, MD, MPH,1 Mitchell J Cohen, MD,3 Peter Muskat, MD,4 Martin A Schreiber, MD,5 Mohammad H. Rahbar, PhD,2,6 R. Michelle Sauer, PhD, ELS,1 Karen J Brasel, MD, MPH,7 Eileen M Bulger, MD,8 John G. Myers, MD,9 Herb A Phelan, MD, MSCS,10 Louis H Alarcon, MD,11 John B Holcomb, MD,1 and on behalf of the PROMMTT Study Group



The impact of do not resuscitate (DNR) orders have not been systematically evaluated in acute trauma research. We determined the frequency, timing and impact on mortality-based effect estimates for patients with DNR orders in the the PRospective Observational Multicenter Major Trauma Transfusion (PROMMTT) study.


Trauma patients surviving at least 30 minutes and transfused ≥1 RBC unit within 6 hours of admission (n=1245) from ten Level 1 centers were enrolled. We report descriptive statistics and results from survival analysis to compare the association of blood product transfusion ratios with outcome defined as mortality and as a composite, DNR and death.


DNRs were reported for 95 patients (7.6%) with 94 in-hospital deaths. There were 172 deaths without DNRs. Of 90 known DNR order times, the median was 53 hours (IQR=9-186 hours) after admission; the median DNR-to-death time was 10 (2-32) hours. DNRs were for comfort measures only (43%), no CPR (40%), and no intubation or CPR (16%). Compared with the 116 non-DNR deaths that occurred after the earliest DNR order (2-hours), the DNR decedents were significantly older with a less severe base deficit, fewer RBC and plasma transfusions, and a later median time of death: 98 (21-230) vs. 17 (4-91) hours. In multivariable Cox models that accounted for time-varying blood product ratios, the associations were consistent regardless of whether outcome was defined as mortality or the composite.


DNR orders were instituted after the 24-hour period of highest mortality risk and more often in older patients not in severe shock. Findings from the primary PROMMTT analyses of the impact of blood product ratios on survival did not materially change when the original mortality outcome was redefined as a composite of DNR or death. DNR orders are potentially an important mediating variable that should be systematically evaluated in trauma research.

Keywords: survival, resuscitation, withdrawal of care


Efficacy studies of initial interventions in the care of the patient with traumatic injuries use 28-30 day survival as the primary endpoint.1, 2 Historically the impetus for this endpoint was the tri-modal distribution of mortality when a significant number of deaths occurred beyond seven days after injury.3-5 Though this pattern is no longer observed the 28-30 day mortality endpoint is still employed.6-8 Recent work has shown the majority of deaths due to traumatic injuries occur within three days after hospital admission. There is a period of early deaths within the first 6-24 hours and a second population at 24 to 72 hours. In fact, less than eight percent of the deaths associated with traumatic injury now occur after seven days.6-8

A potential factor impacting deaths occurring beyond 24 hours is the implementation of do not resuscitate (DNR) orders. Among trauma patients, these orders are initiated in 5 to 7% percent of the population.9, 10 In contrast, the proportion of medical patients with a DNR order is 15 to 18%.11, 12 The average time of implementation of a DNR order is 24-48 hours after admission in trauma patients.9, 10, 13 Furthermore the mortality risk in trauma patients with these orders in place is 43% to 94%. 9, 10, 13, 14 Whereas the mortality risk with a DNR in place is 23 to 37% for general surgery patients, but in some cases may be as high as 82%. 15-17 Franklin et al. suggested that because of the presence of DNR orders after traumatic injury that medical care could not have been expected to impact 62% of trauma deaths.18

It is clear that DNR orders could be an important mediating variable in the 30-day survival paradigm utilized by regulatory bodies. In light of these findings we proposed to determine the impact of DNR orders on the hazard ratios associating mortality with high vs. low plasma:RBC and platelet:RBC ratios the primary outcome analysis of the PRospective Observational Multicenter Major Trauma Transfusion (PROMMTT) study. Of specific interest is the validity of the 30-day survival endpoint when DNR orders are considered.


The PROMMTT study was a prospective, multicenter observational cohort study conducted at ten Level 1 trauma centers in the US. At each study site and the Data Coordinating Center, the local institutional review board approved the study as did the US Army Human Research Protections Office.19, 20

Patients with severe traumatic injuries were enrolled in PROMMTT and data collection was initiated upon hospital arrival. Patients were eligible if they required the highest level of trauma activation at the center, were age 16 or older, and were transfused at least one unit of RBCs in the first six hours after admission. Patients were excluded if: 1) transferred from other facilities; 2) declared dead within 30 minutes of admission; 3) received more than 5 minutes of CPR prior to or within 30 minutes of admission; 4) prisoners; 5) burn injury > 20% of total body surface area; 6) inhalation injury diagnosed by bronchoscopy; or 7) pregnant. No changes in clinical practice were implemented in this observational study. A massive transfusion protocol was in place at all centers.

Data collection and management procedures have previously been described in detail.19, 20 In the data collection forms it was recorded if there was a do not resuscitate (DNR) order in place. DNR orders were documented in the data collection forms. If the response was “yes” the nature of the order was further broken down into the following categories: 1) comfort measures only, 2) no CPR, and 3) no intubation or CPR. In addition, the time of initiation of the DNR order was documented.

Statistical Analysis

To examine the extent to which DNR orders influenced mortality-based effect estimates in the entire PROMMTT study population (n=1,245), survival, regardless of DNR, was compared with DNR-free survival using Kaplan Meier survival curves, in which outcome was defined as a composite of either the time of a DNR order or the time of death. Of the patients declared DNR after ED admission, all but one later died in-hospital, and the group of decedent DNR patients was compared with the group of decedent non-DNR patients who died in-hospital without receiving a DNR order. Between-group differences were assessed using the Pearson chi square test for categorical variables and the nonparametric Wilcoxon rank sum two-sample tests for continuous variables. To assess whether DNR orders influenced the magnitude of association between a life-saving intervention i.e., blood component transfusions, and outcome, we compared results from Cox proportional hazards models defining outcome first as in-hospital mortality (published previously)20 and then as a composite outcome (time of DNR order or time of death). The rationale for the original analysis has been published previously. 20 All analyses were performed using SAS/STAT and Stata/MP software packages.21, 22


Trauma patients surviving at least 30 minutes and transfused ≥1 RBC unit within 6 hours of admission from ten Level 1 centers were enrolled (n=1,245). There were 266 (21.4%) deaths of which 172 (64.7%) were not preceded by a DNR order. DNRs were reported for 95 patients (7.6%) with 94 in-hospital deaths, thus 99% of the patients with a DNR died. The implementation of DNR orders varied between centers from 0 to 16.5% of the patient population of the center. DNR order times were known on 90 patients with a median of 53 hours (IQR=9-186 hours) after admission; the median DNR-to-death time was 10 (2-32) hours. DNRs were for comfort measures only (43%), no CPR (40%), and no intubation or CPR (16%).

Compared with the 116 non-DNR patients whose deaths occurred sometime after the earliest DNR order (2-hours post admission), DNR decedents were significantly older (Table 1) with less severe base deficit, and fewer RBC and plasma transfusions (Table 2). Patients with a DNR in place had a later median time of death: 99 (21-230) vs. 17 (4-91) hours (p<0.001).

Table 1
PROMMTT Demographic and Clinical Factors among DNR and Non-DNR Patients Who Died In-Hospital.
Table 2
PROMMTT Continuous Demographic and Clinical Factors Among DNR and Non-DNR Patients Who Died In-Hospital (-6 hour deaths).

Clinical characteristics for the total study population of 1,245 PROMMTT patients and for the analysis sub-cohort of 905 patients who received 3 or more total blood product transfusions within 24 hours have been reported.20 DNR orders did not materially influence the overall Kaplan Meier survival curves in PROMMTT patients as illustrated by the similarity of the curves and the overlap in their 95% confidence intervals (Figure 1). The graphs reflect the PROMMTT eligibility criterion that patients had to survive the first 30 minutes post ED admission.

Figure 1
Kaplan-Meier Survival estimated for 30 days for Death as Outcome (right) and Declared DNR or Death as Outcome (left).

In the multivariable Cox models that accounted for time-varying blood product ratios among the 905 patients receiving three or more total blood product transfusions within 24 hours, the associations were consistent regardless of whether outcome was defined as mortality or the composite, DNR/death (Tables 3, ,44 and and5).5). While overall outcome analysis was not significantly altered there were specific observations of interest. For example the contribution of platelet:RBCs ratio became significant with the analysis of DNR/death within 6 hours (Table 3). The analyses of deaths from 6-24 hours and 24 hours to 30 days were not significantly impacted by employing the composite of the time of death and the time of DNR.

Table 3
For the first 6 hours Post ED admission a comparison of Death as Outcome and Declared DNR or Death as Outcome using a Multivariable Cox Time-dependent Analysis.
Table 4
At 6-24 hours Post ED admission a comparison of Death as Outcome and Declared DNR or Death as Outcome using a Multivariable Cox Time-dependent Analysis.
Table 5
At 24 hours-30 days Post ED admission a comparison of Death as Outcome and Declared DNR or Death as Outcome using a Multivariable Cox Time-dependent Analysis.


In patients with traumatic injuries the use of DNR orders is highly variable between centers with an overall average of about 7%.9, 10 The average rate of the presence of a DNR order in the PROMMTT study, that was comprised of 10 level I trauma centers, was 7.6%. The reported mortality rates of trauma patients with a DNR order ranges from 40 to 94%.9, 10, 13, 14 In the PROMMTT study the mortality was 99% for patients with a DNR order. Of the in-hospital deaths associated with traumatic injuries 48-61% have a DNR order in place.9, 13, 23 In the present study 36% of the deaths were preceded by a DNR order. This difference is, in part, a result of the study population, those patients requiring a transfusion and presumed to be hemorrhaging and thus at risk for an early death. In the present study 50% of the deaths occurring in the first 24 hours were predominantly associated with hemorrhage.20 The institution of a DNR order usually occurs 24 to 48 hours after admission similar to that noted in the present study (median of 53 hours).9, 10, 24 Thus those patients enrolled in PROMMTT with early deaths were not amenable to having DNR order instated in part accounting for the lower percentage of overall deaths with a DNR. This is further supported by the admission physiological differences between patient who died prior to institution of a DNR order and those with a DNR order (Table 2). Most importantly, the profile of frequency and timing of DNR orders in PROMMTT provides a unique opportunity to assess the impact, if any, on the evaluation of outcomes, specifically in the present study the ratios of blood product ratios on survival.

In the clinical evaluation of various resuscitation procedures, fluids and blood products, long-term survival has been readily accepted as it is an outcome that is understood by everyone, an important clinical endpoint, a clear objective criterion, and one that is accepted by regulatory agencies.1, 2 Though a clear endpoint, improved survival has been difficult to demonstrate with 30 day mortality and its utility questioned.1 One of the issues is the timing of deaths. In the case of early deaths, the alteration of resuscitation procedures may be readily apparent, but later impacted by the complexity of care and variations in treatment days if not weeks beyond the initial intervention. Important to this time-dependent variable is the effect of competing risks of mortality, such that patients surviving a near-fatal hemorrhage may later die of their head injuries. An additional confounder is the initiation of DNR orders. Franklin et al assessed the impact of withdrawal of care and futile care on mortality in trauma patients.18 In their study there were 347 deaths of which 43% had DNR orders or were provided futile care. The overall mortality rate was 6.4% however mortality was only 2.4% in those patients they were allowed to treat. For 62% of the deaths it was deemed “that medical care could not have been reasonably expected to impact survival.” A similar finding was reported by Scarborough et al in general surgery patients.17 These studies demonstrated that there are a high percentage of deaths that present a limited chance of survival, especially when the decision is made to withdraw or limit care. This is clearly demonstrated in the PROMMTT study in that death occurred within hours of initiation of a DNR order (median 10 [2-32] hours).

The impact of DNR orders has not been systematically evaluated in acute trauma research. This is important because most trauma studies (by design) strive to enroll patients that will suffer high mortality rates. Small differences in rates, via institution of DNR orders, may alter outcomes of these studies. In the PROMMTT study, the use and timing of DNR orders appeared to be similar to those in the overall trauma population.9, 10, 13 To assess the impact of initiation of DNR orders on analysis of the effect of increased ratios of plasma and platelets to red cells on survival we determined the frequency and timing for patients with DNR orders in the PROMMTT study. While the initiation of DNR orders occurred later in the course of care, death happened shortly thereafter. Thus, while a significant portion of deaths were associated with DNR orders, especially late deaths, in the present analysis there was not a significant impact in the overall analysis of survival.

While no significant effect was noted on the present analysis, initiation of DNR orders should be considered in future studies. This is of particular importance in those studies of interventions applied later in the course of care when patients who die acutely are eliminated from the study population. In the present study a significant number of deaths (36%) occurred in the first six hours after admission.20 If a study was assessing outcomes of a population that excludes this group of patients the deaths associated with DNR orders would increase from 36% to 55%.

In the present study DNR orders were classified as comfort measures only, no CPR, and no intubation or CPR. These definitions were inadequate to clearly define what care the patient received. At present end of life care is poorly delineated.9, 25, 26 Finally, the sustainment of life support for the harvest of tissues may also affect survival time. In the formulation of studies these endpoints should be defined in order to assure uniformity across institutions to adequately account for care decisions in the analysis of survival.

The PROMMTT study was a prospective observational study with the inherent limitations with this type of study design. The present analysis was limited by possible under-ascertainment of DNR orders in patients who survived to discharge or the thirtieth hospital day, the absence of a uniform definition of DNR, as noted above, and variance of implementation of DNR orders between centers (data not presented). In addition, there was not a systematic evaluation of life saving interventions throughout hospitalization to adequately evaluate the impact of care on survival, which should be considered in future studies.

In conclusion, a significant number of deaths are associated with presence of DNR orders in patients with traumatic injuries. The initiation of DNR orders occurs well after the period of initial resuscitation, the period of highest mortality risk, and is followed shortly thereafter by death in a high percentage of patients. In PROMMTT the presence of a DNR order did not materially change the outcome of the survival analysis when the original mortality outcome was redefined as a composite of DNR order or death. We conclude that DNR orders are an important outcome that should be systematically evaluated in trauma research.


Disclosures of funding: This project was funded by the U.S. Army Medical Research and Materiel Command subcontract W81XWH-08-C-0712. Infrastructure for the Data Coordinating Center was supported by CTSA funds from NIH grant UL1 RR024148.

Role of the Sponsor: The sponsors did not have any role in the design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review or approval of the manuscript; or the decision to submit this manuscript for publication.


Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr. Holcomb reported serving on the board for Tenaxis, the Regional Advisory Council for Trauma, and the National Trauma Institute; providing expert testimony for the Department of Justice; grants funded by the Haemonetics Corporation, and KCI USA, Inc. and consultant fees from the Winkenwerder Company. Dr Wade reported serving on the Science Board for Resuscitation Products, Inc. and the Advisory Board for Astrazeneca. No other disclosures were reported.

Previous Presentation of the Information Reported in the Manuscript: Portions of these data were presented at the PROMMTT Symposium held at the 71st Annual Meeting of the American Association for the Surgery of Trauma (AAST) on September 11-15, 2012 in Kauai, Hawaii.

Disclaimer: The views and opinions expressed in this manuscript are those of the authors and do not reflect the official policy or position of the Army Medical Department, Department of the Army, the Department of Defense, or the United States Government.

Author Contributions:

Study concept and design: Wade, Holcomb, del Junco, Rahbar, Fox

Acquisition of data: Brasel, Bulger, Cohen, Cotton, Holcomb, Muskat, Schreiber, Myers, Phelan, Alarcon

Analysis and interpretation of data: Wade, del Junco, Fox, Cotton, Holcomb

Drafting of the manuscript: Wade, Cotton, Holcomb, Fox, del Junco, Sauer

Critical revision of the manuscript for important intellectual content: Wade, del Junco, Holcomb, Fox, Rahbar, Bulger, Cotton, Sauer, Schreiber, Cohen, Muskat, Brasel, Myers, Phelan, Alarcon

Statistical analysis: Wade, del Junco, Fox, Rahbar

Obtained funding: Rahbar

Administrative, technical, or material support: Rahbar, Holcomb, Fox, del Junco, Bulger, Cotton, Schreiber, Wade, Myers, Phelan, Alarcon

Study supervision: Rahbar, Wade, Cotton, Holcomb

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


1. Holcomb JB, Weiskopf R, Champion H, Gould SA, Sauer RM, Brasel K, Bochicchio G, Bulger E, Cotton BA, Davis D, et al. Challenges to effective research in acute trauma resuscitation: consent and endpoints. Shock. 2011;35:107–113. [PubMed]
2. Wade CE, Holcomb JB. Endpoints in clinical trials of fluid resuscitation of patients with traumatic injuries. Transfusion. 2005;45:4S–8S. [PubMed]
3. Baker CC, Oppenheimer L, Stephens B, Lewis FR, Trunkey DD. Epidemiology of trauma deaths. Am J Surg. 1980;140:144–150. [PubMed]
4. Trunkey DD. Trauma. Accidental and intentional injuries account for more years of life lost in the U.S. than cancer and heart diesease. Among the prescribed remedies are improved preventive efforts speedier surgery and further research. Sci Am. 1983;249:28–35. [PubMed]
5. Sauaia A, Moore FA, Moore EE, Moser KS, Brennan R, Read RA, Pons PT. Epidemiology of trauma deaths: a reassessment. J Trauma. 1995;38:185–193. [PubMed]
6. Clark DE, Qian J, Sihler KC, Hallagan LD, Betensky RA. The distribution of survival times after injury. World journal of surgery. 2012;36:1562–1570. [PMC free article] [PubMed]
7. Demetriades D, Kimbrell B, Salim A, Velmahos G, Rhee P, Preston C, Gruzinski G, Chan L. Trauma deaths in a mature urban trauma system: is “trimodal” distribution a valid concept? Journal of the American College of Surgeons. 2005;201:343–348. [PubMed]
8. McGwin G, Nunn A, Mann J, Griffin R, Davis G, MacLennan P, Kerby J, Acker J, Rue L. Reassessment of the tri-modal mortality distribution in the presence of a regional trauma system. J Trauma. 2009;66:526–530. [PubMed]
9. Cooper Z, Rivara FP, Wang J, MacKenzie EJ, Jurkovich GJ. Withdrawal of life-sustaining therapy in injured patients: variations between trauma centers and nontrauma centers. J Trauma. 2009;66:1327–1335. [PubMed]
10. Nathens AB, Rivara FP, Wang J, Mackenzie EJ, Jurkovich GJ. Variation in the rates of do not resuscitate orders after major trauma and the impact of intensive care unit environment. J Trauma. 2008;64:81–88. discussion 88-91. [PubMed]
11. Boyd K, Teres D, Rapoport J, Lemeshow S. The relationship between age and the use of DNR orders in critical care patients. Evidence for age discrimination. Arch Intern Med. 1996;156:1821–1826. [PubMed]
12. Cardenas-Turanzas M, Gaeta S, Ashoori A, Price KJ, Nates JL. Demographic and clinical determinants of having do not resuscitate orders in the intensive care unit of a comprehensive cancer center. J Palliat Med. 2011;14:45–50. [PubMed]
13. Watch LS, Saxton-Daniels S, Schermer CR. Who has life-sustaining therapy withdrawn after injury? J Trauma. 2005;59:1320–1326. discussion 1326-1327. [PubMed]
14. Mosenthal AC, Murphy PA, Barker LK, Lavery R, Retano A, Livingston DH. Changing the culture around end-of-life care in the trauma intensive care unit. J Trauma. 2008;64:1587–1593. [PubMed]
15. Kazaure H, Roman S, Sosa JA. High mortality in surgical patients with do-not-resuscitate orders: analysis of 8256 patients. Arch Surg. 2011;146:922–928. [PubMed]
16. Meissner A, Genga KR, Studart FS, Settmacher U, Hofmann G, Reinhart K, Sakr Y. Epidemiology of and factors associated with end-of-life decisions in a surgical intensive care unit. Crit Care Med. 2010;38:1060–1068. [PubMed]
17. Scarborough JE, Pappas TN, Bennett KM, Lagoo-Deenadayalan S. Failure-to-Pursue Rescue: Explaining Excess Mortality in Elderly Emergency General Surgical Patients with Preexisting “Do-Not-Resuscitate” Orders. Ann Surg. 2012;256:453–461. [PubMed]
18. Franklin GA, Cannon RW, Smith JW, Harbrecht BG, Miller FB, Richardson JD. Impact of withdrawal of care and futile care on trauma mortality. Surgery. 2011;150:854–860. [PubMed]
19. Rahbar MH, Fox EE, del Junco DJ, Cotton BA, Podbielski JM, Matijevic N, Cohen MJ, Schreiber MA, Zhang J, Mirhaji P, et al. Coordination and management of multicenter clinical studies in trauma: Experience from the PRospective Observational Multicenter Major Trauma Transfusion (PROMMTT) Study. Resuscitation. 2012;83:459–464. [PMC free article] [PubMed]
20. Holcomb JB, del Junco DJ, Fox EE, Wade CE, Cohen MJ, Schreiber MA, Alarcon LH, Bai Y, Brasel KJ, Bulger EM, et al. The Prospective, Observational Multicenter, Major Trauma Transfusion (PROMMTT) Study: Comparative effectiveness of a time-varying treatment with competing risks. Arch Surg. 2012 In Press, ePub October 15. [PMC free article] [PubMed]
21. Stata Statistical Software Release 11. StataCorp LP; College Station, TX: 2009.
22. SAS/STAT Software for Windows Version 9.2. SAS Institute, Inc.; Cary, NC: 2008.
23. Sise MJ, Sise CB, Thorndike JF, Kahl JE, Calvo RY, Shackford SR. Withdrawal of care: a 10-year perspective at a Level I trauma center. J Trauma Acute Care Surg. 2012;72:1186–1193. [PubMed]
24. Keenan SP, Busche KD, Chen LM, McCarthy L, Inman KJ, Sibbald WJ. A retrospective review of a large cohort of patients undergoing the process of withholding or withdrawal of life support. Crit Care Med. 1997;25:1324–1331. [PubMed]
25. Hakim RB, Teno JM, Harrell FE, Jr., Knaus WA, Wenger N, Phillips RS, Layde P, Califf R, Connors AF, Jr., Lynn J. Factors associated with do-not-resuscitate orders: patients’ preferences, prognoses, and physicians’ judgments. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment. Annals of internal medicine. 1996;125:284–293. [PubMed]
26. Mosenthal AC, Weissman DE, Curtis JR, Hays RM, Lustbader DR, Mulkerin C, Puntillo KA, Ray DE, Bassett R, Boss RD, et al. Integrating palliative care in the surgical and trauma intensive care unit: a report from the Improving Palliative Care in the Intensive Care Unit (IPAL-ICU) Project Advisory Board and the Center to Advance Palliative Care. Crit Care Med. 2012;40:1199–1206. [PMC free article] [PubMed]