- Email: [email protected]
The Rapid Acute Physiology Score
The Rapid Acute PhysiologyScore KENNETH J. RHEE, MD,* CHARLES J. FISHER, JR., MD,t NEIL H. WILLITIS, PhD*The Rapid Acute Physiology Score (RAPS) was developed and tested for use as a severity scale in critical care transports. RAPS is an abbrevii version of the Acute physiology and Chronic Health Evaluation (APACHE-II) using only parameters routinely available on all transported patients (I.e. pulse, blood pressure, respiratory rate, and Glasgow Coma Scale). RAPS has a range from 0 (normal) to 16. Two hundred eighty-three patients were transported by helicopter; 62 died. Pretransporl RAPS was available on 262 of 263 patients (mean, 3.85; mediin, 3). Seceuse of death, dii, or tmnsfer, 227 compkte APACHEII scores using least physiologic values for the first 24 hours after transfer were collected (mean, 14.98; median, 13). Stepwfse logistic regression showed that when all APACHE-II and RAPS values were available, the best single predictor of mortality was worst value APACHE-5 (X: = 57.09, P < .Ol). When pretransport RAPS was considered as a single explanatory variable, it too had significant predictive power for moltallty (X$ = 62.53, P < .Ol). Correlation analysis comparing RAPS with APACHE-II values at similar points in time revealed a significant relationship in all cases, wfth the highest correlation between RAPS worst values and APACHE-II worst values (r = 6472, P < .Ol). lt was concluded that RAPS can be applied usefully in complement with APACHEII and may have limited utility whsn used alone. (Am J Emerg Med 1987; 5:276-266)
Severity scales have been used for both clinical evaluation of patients and system-wide assessment. Two areas in which clinical scoring methods have been applied are cardiopulmonary resuscitation, where significant factors have been identified that permit stratification of patients into risk groups, lb3 and trauma care, where severity scales have been used widely for system evaluation and triage.4.5 The care of critically ill patients who must be transported either to
From the *Division of Emergency Medicine and Clinical Toxicology, Department of Internal Medicine, the *Department of Statistics, University of California, Davis, California; and the tDivision of Critical Care Medicine, Department of Internal Medicine, University Hospitals of Cleveland and Case Western Reserve University, Cleveland, Ohio. Manuscript received May 23, 1986; revision accepted November 24, 1986. Presented at the University Association of Emergency Medicine Meeting,
Key Words: Acute Physiology and Chronic Health Evaluation, critical
care transport, helicopter, Rapid Acute Physiology Score, severity scale. 0735-8757187 $0.00 + .25
278
or between hospitals might be similarly improved by the use of severity scoring. Because there is no widely accepted severity score that can be applied rapidly to almost all critical care transports, a simplified version of the Acute Physiology and Chronic Health Evaluation (APACHE-W6 score was developed. The usefulness of this score, the Rapid Acute Physiology Score (RAPS), was evaluated using a group of helicopter-transported patients.
METHODS The Rapid Acute Physiology Score was developed by taking those elements of APACHE-II that can be obtained reliably on all patients in the field or in a hospital emergency department. These elements consist of pulse, blood pressure, respiratory rate, and Glasgow Coma Scale (GCS).’ Point values for these factors were identical to APACHEII except for GCS points, which were scored as follows: GCS = 14-lS/RAPS = 0, GCS = 1 I-13/RAPS = 1, GCS = &IO/RAPS = 2. GCS = 5-7lRAPS = 3, GCS = 3--4/RAPS = 4 (Tables 1 and 2). Two thirds of the points for APACHE-II are laboratory generated and therefore not included. Because RAPS has only about one third the potential sum of APACHE-II, the GCS contribution to RAPS was decreased by two thirds to keep its proportional contribution similar for the two scores. The possible RAPS range is 0 (normal) to 16. Adult patients (older than 10 years) returning to the University Medical Center (UMC) by helicopter were assigned RAPS using information obtained before transfer, on arrival at UMC, following one day of hospitalization, and using worst values (least physiologic values) obtained during the first 24 hours at UMC. APACHE-II scores also were calculated at these times when possible. If the patient had any missing values he or she was excluded from analysis with the following exceptions: 1) If creatinine was not available but BUN was normal, creatinine was assumed to be normal. 2) If arterial blood gas testing was not performed because the patient’s clinical condition did not indicate the need for this measurement, arterial blood gases were assumed to be normal. The power of APACHE-II and RAPS to predict mortality was assessed using stepwise logistic regression. This model assumes that ln(Pr[survival]/Pr[death]) is linear in the explanatory variables. The model fit the data reasonably well and allowed a predicted survival curve to be generated. Because we were interested also in the relationship between
RHEE ET AL n THE RAPID ACUTE PHYSIOLOGY SCORE
TABLE1. Rapid Acute Physiology Score (RAPS)* Points
MAPT HR RespS GCS
+4
+3
+2
3160 a160 a50
130-l 59 140-l 79 3549
110-129 110-139
+1
25-34
0
70-l 09 70-l 09 12-24 a14
+1
t2
10-11 11-13
50-69 55-69 6-9 10-8
+3
40-54 5-7
t4
s49 s39 G5 <4
TOTAL ABBREVIATIONS: MAP, mean arterial pressure; HR, heart rate; Resp, respirations; GCS, Glasgow Coma Scale.
‘To determine RAPS, score patient in each of four subgroups and total. Score of 0 is normal. T(2 x diastolic = systolic) + 3. *Spontaneous or, if no spontaneous, ventilated rate.
TABLE2. Acute Physiology and Chronic Health Evaluation” Points
MAPt Ha Resplmin# Temp Cxygen§ 250% <50% PH” or HCOJ Sodium Potassium Creatinine” Hematocrit WBC” GCS
t4
+3
+2
2160 2160 a50 241
130-l 59 140-l 79 35-49 39--40.9
110-129 110-139
3500
350499
200-349
27.7 7.6-7.69 252 41-51.9 160-l 79 2160 a7 6.0-6.9 2.0-3.4 23.5 260 240 15 - GCS = points
155-l 59 1.5-1.9 5059.9 20-39.9
+1
0
25-34 38.5-38.9
70-l 09 70-l 09 12-24 36.0-38.4
Cl
t2
50-69 55-69
10-11 34-35.9
+3
s49
4054
6-9 32.0-33.9
+4
30.0-31.9
s39 s5 629.9
6200
46.0-49.9 15.10-19.9
>70 7.33-7.49 22.031.9 130-l 49 3.5-5.4 0.6-l .4 30-45.9 3-l 4.9
55-64 = 3
64-75 = 5
7.5-7.59 32.040.9 150-l 54 5.5-5.9
61-70
3.0-3.4
7.25-7.32 19.0-21.9 120-l 29 2.5-2.9 CO.6 20.0-29.9 1.0-2.9
55-60 7.15-7.24 15.0-l 7.9 111-119
<55 c7.15 <15 a110 <2.5 <20
TOTAL -
Age points 644 = 0 Chronic health pointstt
45-54 = 2 0 2 5
475 = 6
ABBREVIATIONS: APS,
Acute Physiology Score; GCS, Glasgow Coma Scale; HC03, bicarbonate radical; HR, heart rate; MAP, mean arterial pressure; Resp, respirations; Temp, temperature; WBC, white blood cell count. ‘Italic print denotes elements of Rapid Acute Physiology Score (RAPS). T(2 x diastolic = systolic) + 3. *Spontaneous or, if no spontaneous, ventilated rate. §lf O2 2 50%, calculate by the formula F102 (713) - PaC02 - PaOn; if O2 < 50%, use the PO*. #Arterial pH is preferred; if not available, use venous HC03. lllf renal failure is acute, double points. **Multiply by 1,000. ttlf the patient has a history of severe organ system insufficiency or is immunocompromised, assign 5 points for nonoperative or emergency postoperative patients and 2 points for elective postoperative patients. NOTE: Organ insufficiency must be present prior to hospitalization. See full definitions for questions.
279
AMERICAN JOURNAL OF EMERGENCY MEDICINE n Volume 5, Number 4 n July 1987
283
Patients
Transferred
to UCD
1 NJ-S
TABLE3. Patient Diagnostic Groups
{‘:zty APACHE-II
6 With worst APACHE-II
t 220
Patients
in Hospital
1Without
worst value APACHE-II worst
v&e
I
1 Worst value and 24-hour APACHE-II
at 24 hrs
1 Without
FIGURE
value scores
Diagnostic Group
No.
Percentage
Multiple trauma Neurologic trauma Orthopedics Burn Spinal cord Surgery-other Pulmonary Cardiac Medical--other Obstetrics/Gynecology Other
121
42.8 20.1 4.2 1.a 1.8 3.6 4.2 6.0 9.2 4.9 1.4
RESULTS There were 283 patients transported by helicopter to UMC during the study period. Two hundred twenty patients survived and were in the hospital at 24 hours. Thirty-six died, and 27 were discharged or transferred. There were 227 complete APACHE-II scores using least physiologic values, 220 APACHE-II scores using values at 24 hours, and 282 pretransport RAPS (Fig. 1). Of the 283 patient transfers, 125 were from the accident scene (440/o), 65 were inpatient hospital transfers (23%), and 93 were from emergency departments (33%). Most of the patients were victims of trauma (70.7%), while the other
‘I-=---\ 60-
E 8 $ P
40-
I
5
IO
15
20
25
30
35
40
APACHE-II
FIGURE 2
(lefr). Expected survival curve for the Acute Physiology
FIGURE 3
(right).
280
5 5 10 12 17 26 14 4
patients represented a variety of assorted medical and nontraumatic surgical conditions (Table 3). The average pretransport RAPS was 3.85, with a median score of 3. The average APACHE-II worst-value score was 14.98. with a median score of 13. Thirty-six of the 62 patients who died did so within 24 hours of transfer. The stepwise logistical regression showed that when all APACHE-II and RAPS values were available, the best single predictor of mortality was the worst-value APACHEII. It had significant predictive power (XT = 57.09, P < .Ol). and the incorporation of other values including diagnostic group and the location of the patient before transfer did not significantly improve predictive power. Figure 2 is the predicted mortality curve for worst-value APACHE-II. When pretransport RAPS was considered as a single explanatory variable, it too had significant predictive power for mortality (XT = 92.53, P < .Ol). Figure 3 is the predicted mortality curve for pretransport RAPS. Correlation analysis comparing RAPS with APACHE-II values at similar points in time revealed a significant rela-
100
2
12
or 24 hour APACHE-II
Patient status 24 hours after transfer.
RAPS and APACHE-II, a correlation analysis was performed to determine the strength of that relationship at similar points in time.
ii .? ?
57
and Chronic Health Evaluation
Expected survival curve for the Rapid Acute Physiology
Score.
RHEE ET AL n THE RAPID ACUTE PHYSIOLOGY SCORE
TABLE4. Correlations Between Rapid Acute Physiology Score and Acute Physiology and Chronic Health Evaluation Time of Collection
r
Number of Patients
Pretransfer First UMC values 24 hours at UMC Worst values
.6140 .7254’ .6969* .8472*
26 244 219 227
ABBREVIATION:UMC,
University Medical Center.
‘P < .Ol.
tionship in all cases (Table 4), with the highest correlation between worst-value RAPS and worst-value APACHE-II (r = .8472, P < .Ol). The correlation between worst-value APACHE-II and pretransport RAPS was lower but significant (r = S904, P < .Ol). DISCUSSION
A severity scale for critically ill patients requiring transport ideally would encompass all diagnostic groups, be simple, and be predictive of outcomes. Rather than using a single score to meet this challenge, two complementary but related scores might be used. The Acute Physiology and Chronic Health Evaluation Score6,8-‘o is a more complete physiologic profile of 12 variables plus chronic disease status and age. This extensive physiologic profile permits APACHE-II to be an accurate predictor of mortality.6 Our data demonstrate that using the worst physiologic APACHEII values during the first 24 hours of care is the most accurate predictor of mortality. The addition of the Rapid Acute Physiology Score and APACHE-II at single points in time do not increase predictive accuracy. However, worst-value APACHE-II is not always available. RAPS is simple to score and documents the initial status of all patients including those who died during transport or during the first few hours of care. In a patient’s disease or injury course, when he is examined has importance clinically and may also have a significant effect on the severity score. Time in the disease course was indirectly examined by using the location of the patient before transfer. Accident scene pick-up usually occurs shortly after the acute event, emergency department transfer usually occurs within hours of an acute event, and interhospital transfer occurs hours if not days later. The sample size and design of this study did not allow more meaningful conclusions to be drawn regarding the relationship between mortality and the temporal changes of APACHEII or RAPS. Consideration of diagnostic groups also did not add significantly to the predictive power of APACHE-II or RAPS.
Previous literature6 suggests that although APACHE-II is a reasonably accurate predictor of mortality across diagnostic groups, some subgroups (diabetic ketoacidosis, coronary artery bypass) behave quite differently. Our results may have been affected by the fact that the patient sample was fairly small, did not contain significant numbers of diabetics or postoperative coronary artery bypass patients, and had a large number of patients with trauma-related diagnoses. There are a number of ways in which RAPS/APACHEII might be applied to critical care transport systems. The average worst-value APACHE-II and average pretransport RAPS might be used to give a balanced measure of the severity of illness of transported patients and permit comparison between systems. Such comparisons could document the severity of patient transports for regulatory and planning agencies interested in monitoring use. Besides severity of illness, patient outcomes might be compared. Using a worst-value APACHE-II score, a patient’s chance of survival could be determined accurately by using the estimated probability of survival multiplied by the number of patients with that APACHE-II score. For patients with only pretransport RAPS, a similar process could be used based on the data reported here. The RAPS and APACHE-II groups could be added together, and the predicted number of survivors then could be compared to the actual number of survivors. A similar method using the Trauma Score and the Injury Severity Score has been used to study the impact of helicopters on trauma mortality. “*‘2 RAPS/APACHE-II might also be used in the screening and selection of patients for quality of care studies. The most seriously ill patients during transport could be selected by finding patients with high RAPS; the most seriously ill hospitalized patients could be selected by finding patients with high APACHE-II scores. Potentially preventable deaths during transport could be selected by finding patients in whom pretransport RAPS scores suggested little physiologic derangement but who died early in the course of hospitalization. Potentially preventable deaths during hospitalization could be selected by finding patients in whom worstvalue APACHE-II scores were fairly low but who died later in the hospital course. It must be remembered that because treatment may influence a severity scale collected later, use of RAPS/APACHE-II is most suitable for screening. Finally, RAPS might be used alone in situations in which full APACHE-II scoring is not possible. The correlation between RAPS and APACHE-II at similar points in time and particularly using worst values suggests that RAPS might serve as a surrogate measure in some instances. SUMMARY
The Rapid Acute Physiology Score was developed and applied to patients transported by helicopter to illustrate its use as a simple measure of physiologic status early in the 281
AMERICAN JOURNAL OF EMERGENCY MEDICINE n Volume 5, Number 4 n July 1987
course of patient care. It can be applied usefully in complement with APACHE-II and may have limited utility when used alone. REFERENCES 1. Cummings R, Eisenberg M: Prehospital cardiopulmonary resuscitation: Is it effective? JAMA 1985;253:2408-2412 2. Eisenberg M, Bergner L, Hearne T: Out of hospital cardiac arrest: A review of major studies and a proposed uniform reporting system. Am J Public Health 1980;70:236-240 3. Thompson B, Stueven H, Mateer J, et al: Comparison of clinical CPR studies in Milwaukee and elsewhere in the United States. Ann Emerg Med 1985;14:750-754 4. Champion H, Sacco W, Carnazzo A, et al: Trauma Score. Crit Care Med 1981;9:672-676 5. Baker S, O’Neill B: The Injury Severity Score: An update. J Trauma 1976;18:882-685
282
6. Knaus W, Draper E, Wagner D, et al: APACHE-II: A severity of disease classification system. Crft Care Med 1985;13:818829 7. Teasdale G, Jennett 6: Assessment of coma and impaired consciousness. A practical scale. Lancet 1974;2:81-84 8. Knaus W, Draper E, Wagner D, et al: Evaluating outcome from intensive care: A preliminary multihospital comparison. Crit Care Med 1982;10:491-496 9. Wagner D, Knaus W, Draper E: Statistical validation of a severity of illness measure. Am J Public Health 1983;73:878-884 10. Bion J, Edlin S, Ramsay G, et al: Validation of a prognostic score in critically ill patients undergoing transport. Br Med J 1985;291:432-434 11. Baxt W, Moody P: The impact of a rotocraft aeromedical emergency care service on trauma mortality. JAMA 1983;249:3047-3051 12. Baxt W, Moody P, Cleveland HC, et al: Hospital-based rotorcraft aeromedical emergency care services and trauma mortality: A multicenter study. Ann Emerg Med 1985;14:859-864
Severity scales have been used for both clinical evaluation of patients and system-wide assessment. Two areas in which clinical scoring methods have been applied are cardiopulmonary resuscitation, where significant factors have been identified that permit stratification of patients into risk groups, lb3 and trauma care, where severity scales have been used widely for system evaluation and triage.4.5 The care of critically ill patients who must be transported either to
From the *Division of Emergency Medicine and Clinical Toxicology, Department of Internal Medicine, the *Department of Statistics, University of California, Davis, California; and the tDivision of Critical Care Medicine, Department of Internal Medicine, University Hospitals of Cleveland and Case Western Reserve University, Cleveland, Ohio. Manuscript received May 23, 1986; revision accepted November 24, 1986. Presented at the University Association of Emergency Medicine Meeting,
Key Words: Acute Physiology and Chronic Health Evaluation, critical
care transport, helicopter, Rapid Acute Physiology Score, severity scale. 0735-8757187 $0.00 + .25
278
or between hospitals might be similarly improved by the use of severity scoring. Because there is no widely accepted severity score that can be applied rapidly to almost all critical care transports, a simplified version of the Acute Physiology and Chronic Health Evaluation (APACHE-W6 score was developed. The usefulness of this score, the Rapid Acute Physiology Score (RAPS), was evaluated using a group of helicopter-transported patients.
METHODS The Rapid Acute Physiology Score was developed by taking those elements of APACHE-II that can be obtained reliably on all patients in the field or in a hospital emergency department. These elements consist of pulse, blood pressure, respiratory rate, and Glasgow Coma Scale (GCS).’ Point values for these factors were identical to APACHEII except for GCS points, which were scored as follows: GCS = 14-lS/RAPS = 0, GCS = 1 I-13/RAPS = 1, GCS = &IO/RAPS = 2. GCS = 5-7lRAPS = 3, GCS = 3--4/RAPS = 4 (Tables 1 and 2). Two thirds of the points for APACHE-II are laboratory generated and therefore not included. Because RAPS has only about one third the potential sum of APACHE-II, the GCS contribution to RAPS was decreased by two thirds to keep its proportional contribution similar for the two scores. The possible RAPS range is 0 (normal) to 16. Adult patients (older than 10 years) returning to the University Medical Center (UMC) by helicopter were assigned RAPS using information obtained before transfer, on arrival at UMC, following one day of hospitalization, and using worst values (least physiologic values) obtained during the first 24 hours at UMC. APACHE-II scores also were calculated at these times when possible. If the patient had any missing values he or she was excluded from analysis with the following exceptions: 1) If creatinine was not available but BUN was normal, creatinine was assumed to be normal. 2) If arterial blood gas testing was not performed because the patient’s clinical condition did not indicate the need for this measurement, arterial blood gases were assumed to be normal. The power of APACHE-II and RAPS to predict mortality was assessed using stepwise logistic regression. This model assumes that ln(Pr[survival]/Pr[death]) is linear in the explanatory variables. The model fit the data reasonably well and allowed a predicted survival curve to be generated. Because we were interested also in the relationship between
RHEE ET AL n THE RAPID ACUTE PHYSIOLOGY SCORE
TABLE1. Rapid Acute Physiology Score (RAPS)* Points
MAPT HR RespS GCS
+4
+3
+2
3160 a160 a50
130-l 59 140-l 79 3549
110-129 110-139
+1
25-34
0
70-l 09 70-l 09 12-24 a14
+1
t2
10-11 11-13
50-69 55-69 6-9 10-8
+3
40-54 5-7
t4
s49 s39 G5 <4
TOTAL ABBREVIATIONS: MAP, mean arterial pressure; HR, heart rate; Resp, respirations; GCS, Glasgow Coma Scale.
‘To determine RAPS, score patient in each of four subgroups and total. Score of 0 is normal. T(2 x diastolic = systolic) + 3. *Spontaneous or, if no spontaneous, ventilated rate.
TABLE2. Acute Physiology and Chronic Health Evaluation” Points
MAPt Ha Resplmin# Temp Cxygen§ 250% <50% PH” or HCOJ Sodium Potassium Creatinine” Hematocrit WBC” GCS
t4
+3
+2
2160 2160 a50 241
130-l 59 140-l 79 35-49 39--40.9
110-129 110-139
3500
350499
200-349
27.7 7.6-7.69 252 41-51.9 160-l 79 2160 a7 6.0-6.9 2.0-3.4 23.5 260 240 15 - GCS = points
155-l 59 1.5-1.9 5059.9 20-39.9
+1
0
25-34 38.5-38.9
70-l 09 70-l 09 12-24 36.0-38.4
Cl
t2
50-69 55-69
10-11 34-35.9
+3
s49
4054
6-9 32.0-33.9
+4
30.0-31.9
s39 s5 629.9
6200
46.0-49.9 15.10-19.9
>70 7.33-7.49 22.031.9 130-l 49 3.5-5.4 0.6-l .4 30-45.9 3-l 4.9
55-64 = 3
64-75 = 5
7.5-7.59 32.040.9 150-l 54 5.5-5.9
61-70
3.0-3.4
7.25-7.32 19.0-21.9 120-l 29 2.5-2.9 CO.6 20.0-29.9 1.0-2.9
55-60 7.15-7.24 15.0-l 7.9 111-119
<55 c7.15 <15 a110 <2.5 <20
TOTAL -
Age points 644 = 0 Chronic health pointstt
45-54 = 2 0 2 5
475 = 6
ABBREVIATIONS: APS,
Acute Physiology Score; GCS, Glasgow Coma Scale; HC03, bicarbonate radical; HR, heart rate; MAP, mean arterial pressure; Resp, respirations; Temp, temperature; WBC, white blood cell count. ‘Italic print denotes elements of Rapid Acute Physiology Score (RAPS). T(2 x diastolic = systolic) + 3. *Spontaneous or, if no spontaneous, ventilated rate. §lf O2 2 50%, calculate by the formula F102 (713) - PaC02 - PaOn; if O2 < 50%, use the PO*. #Arterial pH is preferred; if not available, use venous HC03. lllf renal failure is acute, double points. **Multiply by 1,000. ttlf the patient has a history of severe organ system insufficiency or is immunocompromised, assign 5 points for nonoperative or emergency postoperative patients and 2 points for elective postoperative patients. NOTE: Organ insufficiency must be present prior to hospitalization. See full definitions for questions.
279
AMERICAN JOURNAL OF EMERGENCY MEDICINE n Volume 5, Number 4 n July 1987
283
Patients
Transferred
to UCD
1 NJ-S
TABLE3. Patient Diagnostic Groups
{‘:zty APACHE-II
6 With worst APACHE-II
t 220
Patients
in Hospital
1Without
worst value APACHE-II worst
v&e
I
1 Worst value and 24-hour APACHE-II
at 24 hrs
1 Without
FIGURE
value scores
Diagnostic Group
No.
Percentage
Multiple trauma Neurologic trauma Orthopedics Burn Spinal cord Surgery-other Pulmonary Cardiac Medical--other Obstetrics/Gynecology Other
121
42.8 20.1 4.2 1.a 1.8 3.6 4.2 6.0 9.2 4.9 1.4
RESULTS There were 283 patients transported by helicopter to UMC during the study period. Two hundred twenty patients survived and were in the hospital at 24 hours. Thirty-six died, and 27 were discharged or transferred. There were 227 complete APACHE-II scores using least physiologic values, 220 APACHE-II scores using values at 24 hours, and 282 pretransport RAPS (Fig. 1). Of the 283 patient transfers, 125 were from the accident scene (440/o), 65 were inpatient hospital transfers (23%), and 93 were from emergency departments (33%). Most of the patients were victims of trauma (70.7%), while the other
‘I-=---\ 60-
E 8 $ P
40-
I
5
IO
15
20
25
30
35
40
APACHE-II
FIGURE 2
(lefr). Expected survival curve for the Acute Physiology
FIGURE 3
(right).
280
5 5 10 12 17 26 14 4
patients represented a variety of assorted medical and nontraumatic surgical conditions (Table 3). The average pretransport RAPS was 3.85, with a median score of 3. The average APACHE-II worst-value score was 14.98. with a median score of 13. Thirty-six of the 62 patients who died did so within 24 hours of transfer. The stepwise logistical regression showed that when all APACHE-II and RAPS values were available, the best single predictor of mortality was the worst-value APACHEII. It had significant predictive power (XT = 57.09, P < .Ol). and the incorporation of other values including diagnostic group and the location of the patient before transfer did not significantly improve predictive power. Figure 2 is the predicted mortality curve for worst-value APACHE-II. When pretransport RAPS was considered as a single explanatory variable, it too had significant predictive power for mortality (XT = 92.53, P < .Ol). Figure 3 is the predicted mortality curve for pretransport RAPS. Correlation analysis comparing RAPS with APACHE-II values at similar points in time revealed a significant rela-
100
2
12
or 24 hour APACHE-II
Patient status 24 hours after transfer.
RAPS and APACHE-II, a correlation analysis was performed to determine the strength of that relationship at similar points in time.
ii .? ?
57
and Chronic Health Evaluation
Expected survival curve for the Rapid Acute Physiology
Score.
RHEE ET AL n THE RAPID ACUTE PHYSIOLOGY SCORE
TABLE4. Correlations Between Rapid Acute Physiology Score and Acute Physiology and Chronic Health Evaluation Time of Collection
r
Number of Patients
Pretransfer First UMC values 24 hours at UMC Worst values
.6140 .7254’ .6969* .8472*
26 244 219 227
ABBREVIATION:UMC,
University Medical Center.
‘P < .Ol.
tionship in all cases (Table 4), with the highest correlation between worst-value RAPS and worst-value APACHE-II (r = .8472, P < .Ol). The correlation between worst-value APACHE-II and pretransport RAPS was lower but significant (r = S904, P < .Ol). DISCUSSION
A severity scale for critically ill patients requiring transport ideally would encompass all diagnostic groups, be simple, and be predictive of outcomes. Rather than using a single score to meet this challenge, two complementary but related scores might be used. The Acute Physiology and Chronic Health Evaluation Score6,8-‘o is a more complete physiologic profile of 12 variables plus chronic disease status and age. This extensive physiologic profile permits APACHE-II to be an accurate predictor of mortality.6 Our data demonstrate that using the worst physiologic APACHEII values during the first 24 hours of care is the most accurate predictor of mortality. The addition of the Rapid Acute Physiology Score and APACHE-II at single points in time do not increase predictive accuracy. However, worst-value APACHE-II is not always available. RAPS is simple to score and documents the initial status of all patients including those who died during transport or during the first few hours of care. In a patient’s disease or injury course, when he is examined has importance clinically and may also have a significant effect on the severity score. Time in the disease course was indirectly examined by using the location of the patient before transfer. Accident scene pick-up usually occurs shortly after the acute event, emergency department transfer usually occurs within hours of an acute event, and interhospital transfer occurs hours if not days later. The sample size and design of this study did not allow more meaningful conclusions to be drawn regarding the relationship between mortality and the temporal changes of APACHEII or RAPS. Consideration of diagnostic groups also did not add significantly to the predictive power of APACHE-II or RAPS.
Previous literature6 suggests that although APACHE-II is a reasonably accurate predictor of mortality across diagnostic groups, some subgroups (diabetic ketoacidosis, coronary artery bypass) behave quite differently. Our results may have been affected by the fact that the patient sample was fairly small, did not contain significant numbers of diabetics or postoperative coronary artery bypass patients, and had a large number of patients with trauma-related diagnoses. There are a number of ways in which RAPS/APACHEII might be applied to critical care transport systems. The average worst-value APACHE-II and average pretransport RAPS might be used to give a balanced measure of the severity of illness of transported patients and permit comparison between systems. Such comparisons could document the severity of patient transports for regulatory and planning agencies interested in monitoring use. Besides severity of illness, patient outcomes might be compared. Using a worst-value APACHE-II score, a patient’s chance of survival could be determined accurately by using the estimated probability of survival multiplied by the number of patients with that APACHE-II score. For patients with only pretransport RAPS, a similar process could be used based on the data reported here. The RAPS and APACHE-II groups could be added together, and the predicted number of survivors then could be compared to the actual number of survivors. A similar method using the Trauma Score and the Injury Severity Score has been used to study the impact of helicopters on trauma mortality. “*‘2 RAPS/APACHE-II might also be used in the screening and selection of patients for quality of care studies. The most seriously ill patients during transport could be selected by finding patients with high RAPS; the most seriously ill hospitalized patients could be selected by finding patients with high APACHE-II scores. Potentially preventable deaths during transport could be selected by finding patients in whom pretransport RAPS scores suggested little physiologic derangement but who died early in the course of hospitalization. Potentially preventable deaths during hospitalization could be selected by finding patients in whom worstvalue APACHE-II scores were fairly low but who died later in the hospital course. It must be remembered that because treatment may influence a severity scale collected later, use of RAPS/APACHE-II is most suitable for screening. Finally, RAPS might be used alone in situations in which full APACHE-II scoring is not possible. The correlation between RAPS and APACHE-II at similar points in time and particularly using worst values suggests that RAPS might serve as a surrogate measure in some instances. SUMMARY
The Rapid Acute Physiology Score was developed and applied to patients transported by helicopter to illustrate its use as a simple measure of physiologic status early in the 281
AMERICAN JOURNAL OF EMERGENCY MEDICINE n Volume 5, Number 4 n July 1987
course of patient care. It can be applied usefully in complement with APACHE-II and may have limited utility when used alone. REFERENCES 1. Cummings R, Eisenberg M: Prehospital cardiopulmonary resuscitation: Is it effective? JAMA 1985;253:2408-2412 2. Eisenberg M, Bergner L, Hearne T: Out of hospital cardiac arrest: A review of major studies and a proposed uniform reporting system. Am J Public Health 1980;70:236-240 3. Thompson B, Stueven H, Mateer J, et al: Comparison of clinical CPR studies in Milwaukee and elsewhere in the United States. Ann Emerg Med 1985;14:750-754 4. Champion H, Sacco W, Carnazzo A, et al: Trauma Score. Crit Care Med 1981;9:672-676 5. Baker S, O’Neill B: The Injury Severity Score: An update. J Trauma 1976;18:882-685
282
6. Knaus W, Draper E, Wagner D, et al: APACHE-II: A severity of disease classification system. Crft Care Med 1985;13:818829 7. Teasdale G, Jennett 6: Assessment of coma and impaired consciousness. A practical scale. Lancet 1974;2:81-84 8. Knaus W, Draper E, Wagner D, et al: Evaluating outcome from intensive care: A preliminary multihospital comparison. Crit Care Med 1982;10:491-496 9. Wagner D, Knaus W, Draper E: Statistical validation of a severity of illness measure. Am J Public Health 1983;73:878-884 10. Bion J, Edlin S, Ramsay G, et al: Validation of a prognostic score in critically ill patients undergoing transport. Br Med J 1985;291:432-434 11. Baxt W, Moody P: The impact of a rotocraft aeromedical emergency care service on trauma mortality. JAMA 1983;249:3047-3051 12. Baxt W, Moody P, Cleveland HC, et al: Hospital-based rotorcraft aeromedical emergency care services and trauma mortality: A multicenter study. Ann Emerg Med 1985;14:859-864