- Email: [email protected]
Severe acidosis alone does not predict mortality in the trauma patient
Severe Acidosis Alone Does Not Predict Mortality in the TraurmaPatient Ron Robertson,
MD, John Eidt, MD, Lon Bitzer, MD, Bonny Wallace, MHSA, Terry Collins, RN, Claudia Parks-Miller, MNSc, John Cone, MD, Little Rock, Arkansas - -
BACKGROUND: Because severe acidosis is an indicator of poor prognosis in trauma patients, medical records of these patients were analyzed to determine whether aggressive resuscitation was appropriate. ~~nm-s AND METHODS: Data from a level 1 trauma center registry were reviewed retrospectively to identify patients with a pH 17.0. Thirty-seven patients were identified. Severely acidotic patients were compared to average trauma patients in terms of demographics, resuscitation, injury, and outcome. Surviving acidotic patients were also compared to nonsurviving acidotic patients. RESULTS: Half of the severely acidotic group survived initial resuscitation with approximately one third surviving to leave the hospital. There were no chronically disabled survivors. Nonsurviving acidotic patients were more unstable, more neurologically depressed, and more severely injured. Resuscitation efforts did not consume excessive hospital resources. CONCLUSION: Severe acidosis alone is not a sufficiently powerful predictor of outcome to justify withholding resuscitation; however, when combined with coma and shock, this condition had no survivors in this small series. Am J Surg. 1995;170:691-695.
R
esuscitation of trauma patients is often viewed as a straightforward, well-understood process; however, recent studies have challenged many of our preconceived notions.‘xl Some traditional practices, such as prehospita1 fluid resuscitation, may even be detrimental.2 Acidosis, whether measured by arterial pH or base deficit, is known to correlate with survival in injured patientsJs5 In the face of mounting pressures to avoid expending resources without benefit, such correlations may lead some trauma surgeons to withhold aggressive resuscitative efforts from the severely acidotic patient; however, the studies upon which the correlations are based include relatively few patients with extremely low pH
From the Division of Trauma, Critical Care, Department of Surgery, University of Arkansas for Medical Sciences, Lie Rock, Arkansas Requests for reprints should be addressed to Ron Robertson, MD, University of Arkansas for Medical Sciences, Division of Trauma, Critical Care, Department of Surgery, 4301 West Markham, #520, Little Rock, Arkansas 72205. Presented at the 47th Annual Meeting of the Southwestern Surgical Congress, San Antonio, Texas, April 23-26, 1995.
(pH 17.0). This review was undertaken to investigate the impact of severe acidosis on the management and outcome of trauma patients.
PATIENTS AND METHODS A review of the trauma registry at the University of Arkansas for Medical Sciences (a level 1 trauma center) was completed for all trauma admissions between January 1990 and June 1994 (n = 3,996). A subset of 37 patients with an initial arterial pH 17.0 was identified. The arterial blood gases were determined with a blood gas analyzer (Model AVL 995, AVL Scientific Corp., Roswell, Georgia or Model IL 13 12, Instrumentation Laboratory, Lexington, Maryland). Blood gases were obtained at the discretion of the senior surgical resident involved in the resuscitation; this decision was based upon the clinical condition of the patient. Arterial blood gases were more likely to be drawn on a critically injured patient judged salvageable than on a patient judged not salvageable, although no specific algorithm was used. These patients’ charts were reviewed and the following variables extracted: age; sex; race; mechanism of injury; Glasgow Coma Scale (GCS)$ Revised Trauma Scorei; Injury Severity Score (1SS)s; initial hematocrit; presence or absence of a coagulopathy (defined by a prothrombin time >14 minutes and/or a partial thromboplastin time >34 minutes); toxicology screen; alcohol (ETOH) level; time to initial treatment; volume of prehospital and emergency department resuscitation fluids; blood products used; bicarbonate administration; core temperature; use of a fluid warmer; and number of patients intubated prior to arrival and in the emergency department. The medical records were also reviewed for the clinical course and outcome of each patient to include intensive care unit length of stay (ICU-LOS) days, total length of hospital stay (LOS), operating room use, and discharge placement status (home versus rehabilitation). These data were analyzed using chi-square, Fisher’s exact test, Student’s t-test, Spearman’s regression, and stepwise logistic regression’ analysis by using a VAX 8530 mainframe computer (Digital Equipment Corp., Maynard, Massachusetts) and SAS software (Version 6.07; SAS Institute Inc., Cary, North Carolina).‘” Logistic regression was then used to develop an equation to predict survival.
RESULTS Severely Average Of the identified cidence
Acidotic Patients Versus Trauma Patients 3,996 patients reviewed for this study, only 37 were with an initial pH 17.0, for a severe acidosis inof 0.9%. These 37 severely acidotic patients were I
THE AMERICAN
JOURNAL
OF SURGERYa
VOLUME
170
DECEMBER
1995
691
iEVERE ACIDOSIS DOES NOT PREDICT MORTALITY/ROBERTSON ET AL
I
Severely Acidotic Patients: Survivors Versus Nonsurvivors Presenting Characterlstlcs of 37 Profoundly Acldotlc Trauma Patients The severely acidotic patients were subdiCompared to Those of 3,959 Average Trauma Patients vided into survivors and nonsurvivors for furAverage Acidotlc Average Trauma ther comparison. Survivors and nonsurvivors Patient Patient Characteristics did not significantly differ with respect to age, 37.7 f 14.5 36.0 i 43.3 Age (Y) sex, race, mechanism of injury, mean arterial Sex (%) pH, and mean base deficit Table II. Surviv81 76.5 Men ors, however, were found to be significantly 19 23.5 Women Mechanism of injury (%) less severely injured (lower ISS) and less neu63’ 40 Penetrating rologically depressed (higher GCS score) 32 57 Blunt than nonsurvivors; also, nonsurvivors were 5 3 Burn significantly more likely to have a low systolic 30.3 f 17.4+ 11 f 14.3 Injury Severity Score blood pressure and to be bradycardic, with a 3.07 * 2.80+ 7.8 i 1.79 Revised Trauma Score decreased respiratory rate (Table II). The sur6.3 f 4.90+ 13.5 f 3.50 Glasgow Coma Score vivors and nonsurvivors did not differ in 6.88 f 0.11+ 7.36 * 0.14 PH -20.8 zt 5.60+ -1.82 f 6.30 mean core temperature, initial hematocrit, Base deficit (mmol/L) 35.5 f 1.47 36.9 zt 1.82 Temperature (“C) incidence of IV access obtained in the field, 77 f 55.7+ 92.1 f 26.2 Heart rate fluid resuscitation in the field, blood products 70 zt 62.9+ 136.1 zt 36.9 Systolic blood pressure (mm Hg) used, blood warmer use, amount of bicarbon9.6 zt 13.9+ 20.5 f 7.79 Respiratory rate (bpm) ate administered, results of toxicology screen, 30.4 f 14.1 28.2 f 20.3 Prehospital time (min) ETOH level, coagulation status, ICU-LOS, 83 i 88.5 81.2 f 63.4 Emergency department time (min) or other variables examined in Table II. The 3.2 * 2.1 2.8 zt 2.41 Blood products (units) lowest arterial pH in the severely acidotic Jab/e data are reported as mean * standard deviation, unless otherwise indicated. group was 6.55 and the lowest arterial pH in P = 0.01 versus average trauma patient. a survivor was 6.74. Excluding the significant P = 0.001 versus average trauma patient. variables (ISS, GCS, SBP, HR, and RR), two important prehospital relationships were similar to the average trauma patients treated at this instiidentified. Survivors initially presented in the field less oftution in age, sex, race, and prehospital and emergency deten in shock (systolic blood pressure ~90 mm Hg) than nonpartment times. As expected, acidotic patients were more survivors (27% versus 88%); also, survivors were more likely severely injured, had a poorer neurologic status, and were to correct their hypotension by the time they reached the more unstable (lower systolic blood pressure [SBP], heart rate emergency department than were nonsurvivors (91% versus [HR], and respiratory rate [RR]) than the average trauma pa19%). Secondly, nonsurvivors were much more likely to pretient treated in this center (Table I). sent with hypercarbia (78% presented with a partial presPredictably, the severely acidotic patients did have a sure of arterial carbon dioxide [PaC02] ~50 mm Hg; in consignificantly higher mortality rate than the average trast, 60% of the survivors had a PaC02 ~40 mm Hg) than trauma patient (70% versus 7.4%; P
692
THE AMERICAN
JOURNAL
OF SURGERY@
VOLUME
170
DECEMBER
1995
1 SEVERE ACIDOSIS DOES NOT PREDICT MORTALITWROBERTSON ET Al
J
thoracotomies were performed in the TABLE II operating room with 1 patient survivPresenting Characteristics of the Profoundly Acidotic Trauma Patient: ing. Forty-six percent of the nonsurSurvivors Versus Nonsurvivors vivors and 82% of the survivors went to Characteristics Acidotic Survivors Acldotic Nonsurvivors the operating room after the initial re37.4 f 14.8 37.9 i 14.7 Age (Y) suscitation in the emergency departSex (%) ment. Only six (50%) of the nonsurMen 77 91 vivors lived through their operation and Women 23 9 Mechanism of injury (%) were subsequently transferred to the 69’ Penetrating 45 KU. Twenty-four of the 26 deaths oc27 Blunt 45 curred within 48 hours of admission 4 Burn 10 from uncontrollable hemorrhage and a 19.1 i 12.3 35 f 17.3+ Injury Severity Score persistent coagulopathy. Both survivors 11.2 * 4.80 4.3 f 3.40+ Glasgow Coma Score and nonsurvivors were coagulopathic; 6.9 f 0.1 6.9 zt 0.1 PH however, this did not significantly ef46.5 * 23.8 66.7 * 24.4+ PaCO, (mm Hg) fect survival (Table II). The 2 remain -20.8 f 5.77 Base deficit (mmol/L) -20 i 6.73 35.7 f 1.3 35.1 f 1.5 Temperature (“C) ing patients died later (4 days and 53 112 * 23.4 63.2 zt 54.39 Heart rate (bpm) days after admission) of adult respira126.5 i 55.8 46.8 f 50+ Systolic blood pressure (mm Hg) tory distress syndrome and renal failure, 21.6 f 13.7 4.5 f 10.5+ Respiratory rate (bpm) respectively. Nonsurvivors had an av29 f 12.1 33 + 16.6 Prehospital time (min) erage LOS of 2.4 days. 66 f 71.4 Emergency department time (min) 122 f 1,145 Forty-nine percent of the severely aci1.94 f 1.04 3.08 f 2.43 Prehospital and emergency dotic patients survived the initial resusdepartment fluids (L) citation, with 30% of the entire group 3.3 i 1.98 1.7 f 4.85 Blood products (units) 0.13 f 0.35 1.47 * 1.90 Bicarbonate administration (amp) of acidotic patients surviving to be dis84 73 Coagulopathy (%) charged. Eighty-two percent of the survivors were discharged home with inTable data are reported as mean f standard deviation, unless otherwise indicated dependent functional status. There P = 0.09 versus survivors. tp = 0.0001 versus survwors. were only 2 patients who required post*P = 0.047 versus survivors. discharge rehabilitation, with neither §P = 0.013 versus survivors. patient requiring inpatient status for >2 PaCO, = partial pressure of arterial carbon dioxide. weeks. It should also be noted that 80% of the survivors who suffered from a would be a contributing factor to the acidosis, but the severely blunt mechanism of injury had evidence of a closed head inacidotic patients did not have longer delays. Hypothermia did jury by computed tomography; yet, no patient required any not appear to play a significant role in these patients. These long-term hospitalization or had apparent sequelae secpatients did not differ in either drug or alcohol use from the ondary to this reported head injury. average trauma patient seen in this institution. The severely acidotic patients who survived received very Survival Analysis little in the way of alkalinizing agents such as bicarbonate. Using Student’s t-test, a significant mean difference was The survivors were able to rapidly correct their acid-base identified between severely acidotic survivors and nonsurbalance without assistance other than resuscitation; howvivors for the following variables: ISS, GCS, SBP, HR, and ever, the crystalloid and blood requirements for this group RR. Logistic regression was then used to develop a model to of severely acidotic patients were surprisingly modest, sugpredict survival. Two independent variables were identified as gesting that the cause of the acidosis may be more complex predictive for the model: Y = -3.899 + 0.2123 (GCS) + 0.0177 than simple blood loss. (SBP). The probability of survival (Ps) was calculated by: The current challenge to provide patient care without exPs = ek / 1 + ey. Concordance with this model was 91.3%. pending vast quantities of resources has led to the development of “outcome models” to predict complications or surCOMMENTS vival. Such models have been useful in the development of Acidosis in the trauma patient is a reliable, although nonquality assurance standards and resource allocation plans. specific, indicator of underlying metabolic dysfunction. The Arterial pH and base deficit have been identified as predicmagnitude of the acidosis reflects both the severity and durators of patient outcome for trauma patients.3e5 Although no tion of the problem. Thus, it is not surprising that the degree model is sufficiently reliable to predict the outcome of indiof acidosis, as measured by pH or base deficit, in trauma pavidual patients, such findings may lead some trauma surgeons tients correlates with outcome.3s5 This review of severely acior utilization reviewers to conclude that severely acidotic padotic patients treated at a single trauma center over a 4-year tients have little or no chance of survival and thus are not period identifies several expected correlations. The severely candidates for aggressive resuscitative efforts. Our data do not acidotic patients were more severely injured, more unstable, support such conclusions. If those patients without vital signs and had a much higher mortality. In a rural state such as on arrival are excluded, the overall survival approaches nearly Arkansas, it was anticipated that prolonged delays in rran~port THE AMERICAN
JOURNAL
OF SURGERYa
VOLUME
170
DECEMBER
1995
693
SEVERE ACIDOSIS
DOES NOT PREDICT MORTALITY/ROBERTSON
50%. Such statistics clearly indicate that acidosis alone is not a sufficiently powerful predictor of outcome to base withholding of care on this single variable; however, when combined with shock (systolic blood pressure ~90 mm Hg) and coma (GCS <8), outcome in this series was uniformly fatal. In most cases, death came quickly to the nonsurvivors. There were few in the nonsurviving group who lived to reach the surgical ICU, and only 1 lingered there for an extended period. Another concern often expressed over the resuscitation of such patients is the fear of producing chronically debilitated long-term survivors with no quality of life. This did not occur in this series. There were no patients in need of long-term rehabilitation or inpatient care. Consumption of health care resources is always difficult to calculate, but excessive patient days in the ICU, the hospital, or in rehabilitation facilities was not an issue in this series. This review raises several questions. Would a more ag gressive practice of intubation in the field have resulted in more of these patients in the surviving group? What is the role for artificially raising the pH with hyperventilation, bicarbonate, or other agents? Is the mechanism of metabolic acidosis in trauma patients more complex than simple hypovolemia followed by lactic acidosis, which can be corrected by volume and red cell infusions?
REFERENCES 1. Mattox KL, Bickell W, Pepe PE, et al. Prospective MAST study in 911 patients. _I Trauma. 198~;29:1104-4011. 2. Bickell W, Wall MJ Jr., Pepe PE, et al. Immediate versus delayed fluid resuscitation for hypotensive patients with penetrating torso injuries. NEJM. 1994;331:1105-1109. 3. Rutherford EJ, Morris JA, Reed GW, Hall KS. Base deficit stratifies mortality and determines therapy. J Trauma. 1993;33:417423. 4. Falcone RE, Santanello SA, Schulz MA, et al. Correlation of metabolic acidosis with outcome following injury and its value as a scoring too. world I surg. 1993;17:575-579. 5. Morel1 V, Lundgren E, Gillott A. Predicting severity of trauma by admission white blood cell count, serum potassium level, and arterial pH. South Med I. 1993;86:658-689. 6. Teasdale G, Jennett B. Assessment of coma and impaired consciousness, a practical scale. Lancet. 1974;2:81. 7. Champion HR, Sacco WJ, Copes WS, et al. A revision of the trauma score. J Trauma. 1989;29:623. 8. Baker SP, O’Neill B, Haddon W, et al. The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma. 1974;14:187. 9. Matthews DE, Farewell VT. Binary logistic regression. In: Using and Undersranding Medical St&&s. Basel, Switzerland: Karger; 1988: 141-150. 10. Cody RP, Smith JK. Applied Statistics and the SAS Progmmming Language. 3rd ed. New York: Elsevier Science; 1991.
DISCUSSION Jeffrey Saffle, MD (Salt Lake City, Utah): This retrospective study from a major trauma center has demonstrated that a certain percentage of patients, found to have very severe acidosis on admission, were nonetheless found salvable, and concludes with the caveat, which I agree with, that patients should not be denied aggressive support based on the finding of severe acidosis alone. That acidosis of this severity is a rare event is illustrated by the size, almost 4,000 patients, of Dr. Robertson’s denominator. It makes this publi694
THE AMERICAN
JOURNAL OF SURGERY@
VOLUME 170
ET AL 1
cation particularly important, because a sample like this one will be hard to gather again. Dr. Robertson is in agreement with several previous publications about the significance of severe acidosis, but casual readers should not misinterpret his conclusion. There clearly is an association between acidemia and outcome in trauma patients, though it is by no means absolute. Acidosis, particularly lactic acidosis, indicates an established oxygen debt, predicts increased fluid resuscitation requirements, and correlates with other ominous predictors of nonsurvival, including hypotension, coma, and hemorrhage. Since your paper calls into question the role of acidosis in traumatic shock, I have three questions along this line. First of all, you state that your control sample was made up of “average” trauma patients. What does this mean? Were these all of the other patients treated during this period? Were they all the patients with normal acid-base balances or some other subgroup? Since no trauma patient is truly average, I suggest that your sample be renamed. Secondly, your paper concludes that severe acidosis is not sufficient criteria for withholding support for trauma patients, but you go on to describe a group of nonsurvivors with acidosis that differs in several important aspects from your survivors, specifically in the incidence of coma and hy potension. Would you distill that experience into a set of patients in whom nonsurvival can be predicted reliably and from whom support should be withheld? Finally, the paper espouses cost-effective treatment, or is at any rate written in the name of it, but I wonder if you are taking your own advice in this regard. I note that your patients received very little fluid and almost no sodium bicarbonate, despite the presence of profound base deficits. A number of your nonsurvivors presented with no or low respirations, but there’s not much discussion of the ventilatory support given this group. At the same time, the manuscript makes the point that 14 of the 37 patients reviewed were subjected to emergent thoracotomy with predictably no survivors among those who arrived without vitals signs. In short, I guess I’m asking what your algorithm is for the treatment of the trauma patient who presents with severe acidosis? Should more attention have been paid to correcting the acidosis in these patients either with bicarbonate or ventilation, possible instead of a futile and expensive thoracotomy? The paper gives a mixed message in this regard, which I believe should be clarified. Juan Asensio, MD (Los Angeles, California): Dr. Robertson, thank you for sharing this very provocative work with us. In your group of acidotic patients, I noticed your transfusion requirements were rather low. As a matter of fact, your slide reveals 3.2 * 2.1 units of blood, which is a relatively low volume of transfusion for what I would expect to be very severely injured patients. My question to you is, could their acidosis be correlated or perhaps have been resolved by a more aggressive transfusion protocol? My other question is, did you measure lactic acid levels and perhaps share with us if you think that there is any correlation between lactic acidemia that clears after a certain amount of time with aggressive resuscitation and survival? David Hoyt, MD (San Diego, California): I would like to ask just two quick questions. One, in the group of patients DECEMBER
1995
SEVERE ACIDOSIS
that had a normal blood pressure, but an abnormal Glasgow Coma Scale who survived, did any of those have documented seizures in the immediate prehospital period that might explain the acidosis based on a different mechanism other than hypoperfusion? Along the same line, of the 37 patients, what percentage tested positive for amphetamines or amphetamine derivatives, another known source of “false-positive” creator of acidosis? CLOSING Ron Robertson, MD: The “average trauma patient” was defined as the average of all trauma admissions that occurred over the defined period of this study. This subset of trauma patients presented with severe acidosis, and yet survivors received very little in the way of bicarbonate or fluids. This raises many questions regarding the exact cause of their acidosis. Is it more complex than simple hypovolemia followed by lactic acidosis? Further studies of the severely acidotic patient need to be completed before this
THE AMERICAN
DOES NOT PREDICT MORTALITY/ROBERTSON
ET AL
question can be answered. We too were surprised hy the limited use of intubation in the field in these severely injured patients, and feel that early, more aggressive prehospita1 airway management is needed. Before a prospective study could be completed, we would need to stratify and develop an algorithm of consistent prehospital airway management. This review only analyzed the quantity of blood products administered in the emergency department. I am quite sure this number would be significantly higher if you included the operating room statistics. We do not routinely measure lactic acid levels in the workup of a trauma patient; therefore, I really can’t comment on that question. Lactic acid levels, however, may help to elucidate the cause of their extreme acidosis. Patients with a normal blood pressure and low Glasgow Coma Scale score made up approximately 15% of our acidotic population. We noted no seizures in any of these patients. Fourteen percent of the 37 severely acidotic patients had detectable levels of amphetamines present.
JOURNAL OF SURGERYa
VOLUME 170
DECEMBER
1995
695
MD, John Eidt, MD, Lon Bitzer, MD, Bonny Wallace, MHSA, Terry Collins, RN, Claudia Parks-Miller, MNSc, John Cone, MD, Little Rock, Arkansas - -
BACKGROUND: Because severe acidosis is an indicator of poor prognosis in trauma patients, medical records of these patients were analyzed to determine whether aggressive resuscitation was appropriate. ~~nm-s AND METHODS: Data from a level 1 trauma center registry were reviewed retrospectively to identify patients with a pH 17.0. Thirty-seven patients were identified. Severely acidotic patients were compared to average trauma patients in terms of demographics, resuscitation, injury, and outcome. Surviving acidotic patients were also compared to nonsurviving acidotic patients. RESULTS: Half of the severely acidotic group survived initial resuscitation with approximately one third surviving to leave the hospital. There were no chronically disabled survivors. Nonsurviving acidotic patients were more unstable, more neurologically depressed, and more severely injured. Resuscitation efforts did not consume excessive hospital resources. CONCLUSION: Severe acidosis alone is not a sufficiently powerful predictor of outcome to justify withholding resuscitation; however, when combined with coma and shock, this condition had no survivors in this small series. Am J Surg. 1995;170:691-695.
R
esuscitation of trauma patients is often viewed as a straightforward, well-understood process; however, recent studies have challenged many of our preconceived notions.‘xl Some traditional practices, such as prehospita1 fluid resuscitation, may even be detrimental.2 Acidosis, whether measured by arterial pH or base deficit, is known to correlate with survival in injured patientsJs5 In the face of mounting pressures to avoid expending resources without benefit, such correlations may lead some trauma surgeons to withhold aggressive resuscitative efforts from the severely acidotic patient; however, the studies upon which the correlations are based include relatively few patients with extremely low pH
From the Division of Trauma, Critical Care, Department of Surgery, University of Arkansas for Medical Sciences, Lie Rock, Arkansas Requests for reprints should be addressed to Ron Robertson, MD, University of Arkansas for Medical Sciences, Division of Trauma, Critical Care, Department of Surgery, 4301 West Markham, #520, Little Rock, Arkansas 72205. Presented at the 47th Annual Meeting of the Southwestern Surgical Congress, San Antonio, Texas, April 23-26, 1995.
(pH 17.0). This review was undertaken to investigate the impact of severe acidosis on the management and outcome of trauma patients.
PATIENTS AND METHODS A review of the trauma registry at the University of Arkansas for Medical Sciences (a level 1 trauma center) was completed for all trauma admissions between January 1990 and June 1994 (n = 3,996). A subset of 37 patients with an initial arterial pH 17.0 was identified. The arterial blood gases were determined with a blood gas analyzer (Model AVL 995, AVL Scientific Corp., Roswell, Georgia or Model IL 13 12, Instrumentation Laboratory, Lexington, Maryland). Blood gases were obtained at the discretion of the senior surgical resident involved in the resuscitation; this decision was based upon the clinical condition of the patient. Arterial blood gases were more likely to be drawn on a critically injured patient judged salvageable than on a patient judged not salvageable, although no specific algorithm was used. These patients’ charts were reviewed and the following variables extracted: age; sex; race; mechanism of injury; Glasgow Coma Scale (GCS)$ Revised Trauma Scorei; Injury Severity Score (1SS)s; initial hematocrit; presence or absence of a coagulopathy (defined by a prothrombin time >14 minutes and/or a partial thromboplastin time >34 minutes); toxicology screen; alcohol (ETOH) level; time to initial treatment; volume of prehospital and emergency department resuscitation fluids; blood products used; bicarbonate administration; core temperature; use of a fluid warmer; and number of patients intubated prior to arrival and in the emergency department. The medical records were also reviewed for the clinical course and outcome of each patient to include intensive care unit length of stay (ICU-LOS) days, total length of hospital stay (LOS), operating room use, and discharge placement status (home versus rehabilitation). These data were analyzed using chi-square, Fisher’s exact test, Student’s t-test, Spearman’s regression, and stepwise logistic regression’ analysis by using a VAX 8530 mainframe computer (Digital Equipment Corp., Maynard, Massachusetts) and SAS software (Version 6.07; SAS Institute Inc., Cary, North Carolina).‘” Logistic regression was then used to develop an equation to predict survival.
RESULTS Severely Average Of the identified cidence
Acidotic Patients Versus Trauma Patients 3,996 patients reviewed for this study, only 37 were with an initial pH 17.0, for a severe acidosis inof 0.9%. These 37 severely acidotic patients were I
THE AMERICAN
JOURNAL
OF SURGERYa
VOLUME
170
DECEMBER
1995
691
iEVERE ACIDOSIS DOES NOT PREDICT MORTALITY/ROBERTSON ET AL
I
Severely Acidotic Patients: Survivors Versus Nonsurvivors Presenting Characterlstlcs of 37 Profoundly Acldotlc Trauma Patients The severely acidotic patients were subdiCompared to Those of 3,959 Average Trauma Patients vided into survivors and nonsurvivors for furAverage Acidotlc Average Trauma ther comparison. Survivors and nonsurvivors Patient Patient Characteristics did not significantly differ with respect to age, 37.7 f 14.5 36.0 i 43.3 Age (Y) sex, race, mechanism of injury, mean arterial Sex (%) pH, and mean base deficit Table II. Surviv81 76.5 Men ors, however, were found to be significantly 19 23.5 Women Mechanism of injury (%) less severely injured (lower ISS) and less neu63’ 40 Penetrating rologically depressed (higher GCS score) 32 57 Blunt than nonsurvivors; also, nonsurvivors were 5 3 Burn significantly more likely to have a low systolic 30.3 f 17.4+ 11 f 14.3 Injury Severity Score blood pressure and to be bradycardic, with a 3.07 * 2.80+ 7.8 i 1.79 Revised Trauma Score decreased respiratory rate (Table II). The sur6.3 f 4.90+ 13.5 f 3.50 Glasgow Coma Score vivors and nonsurvivors did not differ in 6.88 f 0.11+ 7.36 * 0.14 PH -20.8 zt 5.60+ -1.82 f 6.30 mean core temperature, initial hematocrit, Base deficit (mmol/L) 35.5 f 1.47 36.9 zt 1.82 Temperature (“C) incidence of IV access obtained in the field, 77 f 55.7+ 92.1 f 26.2 Heart rate fluid resuscitation in the field, blood products 70 zt 62.9+ 136.1 zt 36.9 Systolic blood pressure (mm Hg) used, blood warmer use, amount of bicarbon9.6 zt 13.9+ 20.5 f 7.79 Respiratory rate (bpm) ate administered, results of toxicology screen, 30.4 f 14.1 28.2 f 20.3 Prehospital time (min) ETOH level, coagulation status, ICU-LOS, 83 i 88.5 81.2 f 63.4 Emergency department time (min) or other variables examined in Table II. The 3.2 * 2.1 2.8 zt 2.41 Blood products (units) lowest arterial pH in the severely acidotic Jab/e data are reported as mean * standard deviation, unless otherwise indicated. group was 6.55 and the lowest arterial pH in P = 0.01 versus average trauma patient. a survivor was 6.74. Excluding the significant P = 0.001 versus average trauma patient. variables (ISS, GCS, SBP, HR, and RR), two important prehospital relationships were similar to the average trauma patients treated at this instiidentified. Survivors initially presented in the field less oftution in age, sex, race, and prehospital and emergency deten in shock (systolic blood pressure ~90 mm Hg) than nonpartment times. As expected, acidotic patients were more survivors (27% versus 88%); also, survivors were more likely severely injured, had a poorer neurologic status, and were to correct their hypotension by the time they reached the more unstable (lower systolic blood pressure [SBP], heart rate emergency department than were nonsurvivors (91% versus [HR], and respiratory rate [RR]) than the average trauma pa19%). Secondly, nonsurvivors were much more likely to pretient treated in this center (Table I). sent with hypercarbia (78% presented with a partial presPredictably, the severely acidotic patients did have a sure of arterial carbon dioxide [PaC02] ~50 mm Hg; in consignificantly higher mortality rate than the average trast, 60% of the survivors had a PaC02 ~40 mm Hg) than trauma patient (70% versus 7.4%; P
692
THE AMERICAN
JOURNAL
OF SURGERY@
VOLUME
170
DECEMBER
1995
1 SEVERE ACIDOSIS DOES NOT PREDICT MORTALITWROBERTSON ET Al
J
thoracotomies were performed in the TABLE II operating room with 1 patient survivPresenting Characteristics of the Profoundly Acidotic Trauma Patient: ing. Forty-six percent of the nonsurSurvivors Versus Nonsurvivors vivors and 82% of the survivors went to Characteristics Acidotic Survivors Acldotic Nonsurvivors the operating room after the initial re37.4 f 14.8 37.9 i 14.7 Age (Y) suscitation in the emergency departSex (%) ment. Only six (50%) of the nonsurMen 77 91 vivors lived through their operation and Women 23 9 Mechanism of injury (%) were subsequently transferred to the 69’ Penetrating 45 KU. Twenty-four of the 26 deaths oc27 Blunt 45 curred within 48 hours of admission 4 Burn 10 from uncontrollable hemorrhage and a 19.1 i 12.3 35 f 17.3+ Injury Severity Score persistent coagulopathy. Both survivors 11.2 * 4.80 4.3 f 3.40+ Glasgow Coma Score and nonsurvivors were coagulopathic; 6.9 f 0.1 6.9 zt 0.1 PH however, this did not significantly ef46.5 * 23.8 66.7 * 24.4+ PaCO, (mm Hg) fect survival (Table II). The 2 remain -20.8 f 5.77 Base deficit (mmol/L) -20 i 6.73 35.7 f 1.3 35.1 f 1.5 Temperature (“C) ing patients died later (4 days and 53 112 * 23.4 63.2 zt 54.39 Heart rate (bpm) days after admission) of adult respira126.5 i 55.8 46.8 f 50+ Systolic blood pressure (mm Hg) tory distress syndrome and renal failure, 21.6 f 13.7 4.5 f 10.5+ Respiratory rate (bpm) respectively. Nonsurvivors had an av29 f 12.1 33 + 16.6 Prehospital time (min) erage LOS of 2.4 days. 66 f 71.4 Emergency department time (min) 122 f 1,145 Forty-nine percent of the severely aci1.94 f 1.04 3.08 f 2.43 Prehospital and emergency dotic patients survived the initial resusdepartment fluids (L) citation, with 30% of the entire group 3.3 i 1.98 1.7 f 4.85 Blood products (units) 0.13 f 0.35 1.47 * 1.90 Bicarbonate administration (amp) of acidotic patients surviving to be dis84 73 Coagulopathy (%) charged. Eighty-two percent of the survivors were discharged home with inTable data are reported as mean f standard deviation, unless otherwise indicated dependent functional status. There P = 0.09 versus survivors. tp = 0.0001 versus survwors. were only 2 patients who required post*P = 0.047 versus survivors. discharge rehabilitation, with neither §P = 0.013 versus survivors. patient requiring inpatient status for >2 PaCO, = partial pressure of arterial carbon dioxide. weeks. It should also be noted that 80% of the survivors who suffered from a would be a contributing factor to the acidosis, but the severely blunt mechanism of injury had evidence of a closed head inacidotic patients did not have longer delays. Hypothermia did jury by computed tomography; yet, no patient required any not appear to play a significant role in these patients. These long-term hospitalization or had apparent sequelae secpatients did not differ in either drug or alcohol use from the ondary to this reported head injury. average trauma patient seen in this institution. The severely acidotic patients who survived received very Survival Analysis little in the way of alkalinizing agents such as bicarbonate. Using Student’s t-test, a significant mean difference was The survivors were able to rapidly correct their acid-base identified between severely acidotic survivors and nonsurbalance without assistance other than resuscitation; howvivors for the following variables: ISS, GCS, SBP, HR, and ever, the crystalloid and blood requirements for this group RR. Logistic regression was then used to develop a model to of severely acidotic patients were surprisingly modest, sugpredict survival. Two independent variables were identified as gesting that the cause of the acidosis may be more complex predictive for the model: Y = -3.899 + 0.2123 (GCS) + 0.0177 than simple blood loss. (SBP). The probability of survival (Ps) was calculated by: The current challenge to provide patient care without exPs = ek / 1 + ey. Concordance with this model was 91.3%. pending vast quantities of resources has led to the development of “outcome models” to predict complications or surCOMMENTS vival. Such models have been useful in the development of Acidosis in the trauma patient is a reliable, although nonquality assurance standards and resource allocation plans. specific, indicator of underlying metabolic dysfunction. The Arterial pH and base deficit have been identified as predicmagnitude of the acidosis reflects both the severity and durators of patient outcome for trauma patients.3e5 Although no tion of the problem. Thus, it is not surprising that the degree model is sufficiently reliable to predict the outcome of indiof acidosis, as measured by pH or base deficit, in trauma pavidual patients, such findings may lead some trauma surgeons tients correlates with outcome.3s5 This review of severely acior utilization reviewers to conclude that severely acidotic padotic patients treated at a single trauma center over a 4-year tients have little or no chance of survival and thus are not period identifies several expected correlations. The severely candidates for aggressive resuscitative efforts. Our data do not acidotic patients were more severely injured, more unstable, support such conclusions. If those patients without vital signs and had a much higher mortality. In a rural state such as on arrival are excluded, the overall survival approaches nearly Arkansas, it was anticipated that prolonged delays in rran~port THE AMERICAN
JOURNAL
OF SURGERYa
VOLUME
170
DECEMBER
1995
693
SEVERE ACIDOSIS
DOES NOT PREDICT MORTALITY/ROBERTSON
50%. Such statistics clearly indicate that acidosis alone is not a sufficiently powerful predictor of outcome to base withholding of care on this single variable; however, when combined with shock (systolic blood pressure ~90 mm Hg) and coma (GCS <8), outcome in this series was uniformly fatal. In most cases, death came quickly to the nonsurvivors. There were few in the nonsurviving group who lived to reach the surgical ICU, and only 1 lingered there for an extended period. Another concern often expressed over the resuscitation of such patients is the fear of producing chronically debilitated long-term survivors with no quality of life. This did not occur in this series. There were no patients in need of long-term rehabilitation or inpatient care. Consumption of health care resources is always difficult to calculate, but excessive patient days in the ICU, the hospital, or in rehabilitation facilities was not an issue in this series. This review raises several questions. Would a more ag gressive practice of intubation in the field have resulted in more of these patients in the surviving group? What is the role for artificially raising the pH with hyperventilation, bicarbonate, or other agents? Is the mechanism of metabolic acidosis in trauma patients more complex than simple hypovolemia followed by lactic acidosis, which can be corrected by volume and red cell infusions?
REFERENCES 1. Mattox KL, Bickell W, Pepe PE, et al. Prospective MAST study in 911 patients. _I Trauma. 198~;29:1104-4011. 2. Bickell W, Wall MJ Jr., Pepe PE, et al. Immediate versus delayed fluid resuscitation for hypotensive patients with penetrating torso injuries. NEJM. 1994;331:1105-1109. 3. Rutherford EJ, Morris JA, Reed GW, Hall KS. Base deficit stratifies mortality and determines therapy. J Trauma. 1993;33:417423. 4. Falcone RE, Santanello SA, Schulz MA, et al. Correlation of metabolic acidosis with outcome following injury and its value as a scoring too. world I surg. 1993;17:575-579. 5. Morel1 V, Lundgren E, Gillott A. Predicting severity of trauma by admission white blood cell count, serum potassium level, and arterial pH. South Med I. 1993;86:658-689. 6. Teasdale G, Jennett B. Assessment of coma and impaired consciousness, a practical scale. Lancet. 1974;2:81. 7. Champion HR, Sacco WJ, Copes WS, et al. A revision of the trauma score. J Trauma. 1989;29:623. 8. Baker SP, O’Neill B, Haddon W, et al. The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma. 1974;14:187. 9. Matthews DE, Farewell VT. Binary logistic regression. In: Using and Undersranding Medical St&&s. Basel, Switzerland: Karger; 1988: 141-150. 10. Cody RP, Smith JK. Applied Statistics and the SAS Progmmming Language. 3rd ed. New York: Elsevier Science; 1991.
DISCUSSION Jeffrey Saffle, MD (Salt Lake City, Utah): This retrospective study from a major trauma center has demonstrated that a certain percentage of patients, found to have very severe acidosis on admission, were nonetheless found salvable, and concludes with the caveat, which I agree with, that patients should not be denied aggressive support based on the finding of severe acidosis alone. That acidosis of this severity is a rare event is illustrated by the size, almost 4,000 patients, of Dr. Robertson’s denominator. It makes this publi694
THE AMERICAN
JOURNAL OF SURGERY@
VOLUME 170
ET AL 1
cation particularly important, because a sample like this one will be hard to gather again. Dr. Robertson is in agreement with several previous publications about the significance of severe acidosis, but casual readers should not misinterpret his conclusion. There clearly is an association between acidemia and outcome in trauma patients, though it is by no means absolute. Acidosis, particularly lactic acidosis, indicates an established oxygen debt, predicts increased fluid resuscitation requirements, and correlates with other ominous predictors of nonsurvival, including hypotension, coma, and hemorrhage. Since your paper calls into question the role of acidosis in traumatic shock, I have three questions along this line. First of all, you state that your control sample was made up of “average” trauma patients. What does this mean? Were these all of the other patients treated during this period? Were they all the patients with normal acid-base balances or some other subgroup? Since no trauma patient is truly average, I suggest that your sample be renamed. Secondly, your paper concludes that severe acidosis is not sufficient criteria for withholding support for trauma patients, but you go on to describe a group of nonsurvivors with acidosis that differs in several important aspects from your survivors, specifically in the incidence of coma and hy potension. Would you distill that experience into a set of patients in whom nonsurvival can be predicted reliably and from whom support should be withheld? Finally, the paper espouses cost-effective treatment, or is at any rate written in the name of it, but I wonder if you are taking your own advice in this regard. I note that your patients received very little fluid and almost no sodium bicarbonate, despite the presence of profound base deficits. A number of your nonsurvivors presented with no or low respirations, but there’s not much discussion of the ventilatory support given this group. At the same time, the manuscript makes the point that 14 of the 37 patients reviewed were subjected to emergent thoracotomy with predictably no survivors among those who arrived without vitals signs. In short, I guess I’m asking what your algorithm is for the treatment of the trauma patient who presents with severe acidosis? Should more attention have been paid to correcting the acidosis in these patients either with bicarbonate or ventilation, possible instead of a futile and expensive thoracotomy? The paper gives a mixed message in this regard, which I believe should be clarified. Juan Asensio, MD (Los Angeles, California): Dr. Robertson, thank you for sharing this very provocative work with us. In your group of acidotic patients, I noticed your transfusion requirements were rather low. As a matter of fact, your slide reveals 3.2 * 2.1 units of blood, which is a relatively low volume of transfusion for what I would expect to be very severely injured patients. My question to you is, could their acidosis be correlated or perhaps have been resolved by a more aggressive transfusion protocol? My other question is, did you measure lactic acid levels and perhaps share with us if you think that there is any correlation between lactic acidemia that clears after a certain amount of time with aggressive resuscitation and survival? David Hoyt, MD (San Diego, California): I would like to ask just two quick questions. One, in the group of patients DECEMBER
1995
SEVERE ACIDOSIS
that had a normal blood pressure, but an abnormal Glasgow Coma Scale who survived, did any of those have documented seizures in the immediate prehospital period that might explain the acidosis based on a different mechanism other than hypoperfusion? Along the same line, of the 37 patients, what percentage tested positive for amphetamines or amphetamine derivatives, another known source of “false-positive” creator of acidosis? CLOSING Ron Robertson, MD: The “average trauma patient” was defined as the average of all trauma admissions that occurred over the defined period of this study. This subset of trauma patients presented with severe acidosis, and yet survivors received very little in the way of bicarbonate or fluids. This raises many questions regarding the exact cause of their acidosis. Is it more complex than simple hypovolemia followed by lactic acidosis? Further studies of the severely acidotic patient need to be completed before this
THE AMERICAN
DOES NOT PREDICT MORTALITY/ROBERTSON
ET AL
question can be answered. We too were surprised hy the limited use of intubation in the field in these severely injured patients, and feel that early, more aggressive prehospita1 airway management is needed. Before a prospective study could be completed, we would need to stratify and develop an algorithm of consistent prehospital airway management. This review only analyzed the quantity of blood products administered in the emergency department. I am quite sure this number would be significantly higher if you included the operating room statistics. We do not routinely measure lactic acid levels in the workup of a trauma patient; therefore, I really can’t comment on that question. Lactic acid levels, however, may help to elucidate the cause of their extreme acidosis. Patients with a normal blood pressure and low Glasgow Coma Scale score made up approximately 15% of our acidotic population. We noted no seizures in any of these patients. Fourteen percent of the 37 severely acidotic patients had detectable levels of amphetamines present.
JOURNAL OF SURGERYa
VOLUME 170
DECEMBER
1995
695