Comparison of different scores as predictors of mechanical ventilation in drug overdose patients

Abstract

The decision of intubation and mechanical ventilation in poisoned patients with impaired consciousness can be a difficult task. The present study aimed to evaluate the power of Glasgow Coma Scale (GCS), acute physiology and chronic health evaluation (APACHE II), rapid acute physiology score (RAPS) and rapid emergency medicine score (REMS) at admission in predicting the need of intubation and mechanical ventilation in drug overdose patients with disturbed consciousness level (DCL). This prospective observational study was conducted on 104 poisoned patients who were admitted to Tanta Toxicological Unit with a DCL. Four scoring systems (GCS, APACHE II, RAPS and REMS) were recorded for all patients on admission. Discrimination was evaluated using receiver operating characteristics curve and calculating the area under the curve (AUC). Twenty-four cases required mechanical ventilation. The mechanically ventilated patients had significantly lower value of GCS and higher values of APACHE II, REMS and RAPS than other group (p < 0.001). Although the APACHE II score has the best AUC value (0.796) in predicting mechanical ventilation, there was no statistically significant difference between the four scores. GCS > 8 had 100% negative predictive value, while REMS > 8 had 100% positive predictive value.

Keywords

acute physiology and chronic health evaluation II score Glasgow Coma Scale rapid acute physiology score rapid emergency medicine score prognosis

Introduction

Disturbed consciousness level (DCL) may result from traumatic, toxic, metabolic or other pathologic causes.¹ Aspiration pneumonia and respiratory failure may occur as a complication in patients with DCL due to loss of protective airway reflexes.² These complications can be prevented by endotracheal intubation (ETI) and mechanical ventilation.

Numerous scoring scales have been proposed and used to assess DCL in prehospital setting, grade disease severity and predict patients’ outcome. The ideal scoring system should be easy and applicable to the greatest number of patients, recognize rapidly deteriorating patients and predict morbidity and mortality.³

Glasgow Coma Scale (GCS) has been used in intensive care units (ICUs) and emergency prehospital settings for assessment of consciousness level, patients’ outcomes and recovery in ICU, determination of the patients in need of intubation with antidepressant poisoning⁴ and prediction of acute and delayed poisoning outcome.^5

–10

The acute physiology and chronic health evaluation (APACHE II) score has been validated in general, surgical and ICU patients. However, APACHE II score includes several blood chemistry variables; therefore, it is not suitable for rapid scoring in the emergency department.^11,12 Rapid acute physiology score (RAPS) is a simplified version of the APACHE II score. It can be used to evaluate patients being transported between ICUs.¹² Further, it was modified to a more potent scoring system called the rapid emergency medicine score (REMS) for predicting mortality.¹¹

These scoring systems have been used to evaluate intoxicated patients.^13–15 However, the validity of these scoring systems to predict the need of ETI and mechanical ventilation is not well defined. Therefore, the aim of this study was to evaluate various scoring systems (GCS, APACHE II, REMS and RAPS) at admission in predicting the need of ETI and mechanical ventilation in drug overdose patients with DCL.

Patients and methods

This prospective study was conducted on admitted 104 acute drug overdose patients with DCL in Tanta toxicological unit, Tanta University Hospital, during the period from July 2014 to July 2015. Tanta University Hospital is a tertiary care centre that receives patients mainly from Gharbia governorate and other nearby regions in the Nile Delta. The study was approved by Research Ethics committee of Tanta Faculty of Medicine and conducted according to the World Medical Association Declaration of Helsinki. Written informed consent was obtained from relatives of the patients. Privacy of participants and confidentiality of the data were maintained by assigning code numbers to patients. Diagnosis of acute poisoning was based on history taking, clinical examination of the patients and exclusion of relevant differential diagnoses.

All patients of both sexes over 16 years old with GCS <15 caused by drug overdose (single or multiple drug ingestions) were included in this study.

Exclusion criteria were: (1) transferred patients from elsewhere or admitted after first 24 h, (2) patients with evidence of other causes of DCL and (3) patients poisoned with non-drug agents such as organophosphorus, phosphide and paraquat.

The data collection sheet included demographic data (patient code, age, sex and time of admission) and poisoning data (type, form, amount, route of exposure, mode of poisoning and delay between exposure and arrival to hospital ‘pre-hospitalization period’). Patient outcome (recovery, discharge upon request and complications or death) and hospitalization period were recorded in data sheet. Physical examination and calculation of the different scoring systems (GCS, APACHE II, RAPS and REMS) were performed to all patients at admission. Patients were divided into two groups: group I included patients who required ventilator support and group II those who did not need mechanical ventilation.

The GCS was determined based on three components: eyes (4 = opens, 3 = to verbal command, 2 = to pain and 1 = none), verbal (5 = oriented, 4 = disoriented, 3 = inappropriate words, 2 = incomprehensible sounds and 1 = none) and motor (6 = obeys, 5 = localizes pain, 4 = withdrawal, 3 = abnormal flexion, 2 = abnormal extension and 1 = none).¹⁶

To calculate the APACHE II score, age and 12 common physiological and laboratory values were employed including body temperature, mean arterial pressure, pulse rate, respiratory rate, oxygenation of arterial blood, arterial pH, serum sodium, serum potassium, serum creatinine, haematocrit, white blood cell count and GCS. The maximal APACHE II score is 71.¹⁷

The RAPS was developed by taking easily obtained elements of APACHE II in the hospital setting. It includes pulse rate, mean arterial pressure, respiratory rate and GCS.¹⁸ The maximum RAPS score is 16. The REMS is a recent modification of RAPS by adding peripheral oxygen saturation and chronological age to the four variables mentioned above. The scoring range for each variable was 0–4 except for age range from 0 to 6 as APACHE II score. The maximum score is 26.¹⁹

Statistical analysis

The collected data were statistically analysed using MedCalc Statistical Software version 15.8. For quantitative data, the median and interquartile range (IQR) (expressed as 25th percentile – 75th percentile) were calculated and Mann–Whitney U test was used to compare the two groups. Qualitative data were described as the frequency of occurrence (percentage) and compared using Pearson χ² for association.²⁰ The receiver operating characteristic (ROC) curve analysed the relation between true-positive results (number of predicted ventilation among those who actually were mechanically ventilated) and false-positive results (number of predicted ventilation among those who actually did not need) for each score. The area under ROC curve (AUC) is graded as follows: 0.90–1 = excellent, 0.80–0.90 = good, 0.70–0.80 = fair and 0.60–0.70 = poor. The ROC curve analysis was carried out according to DeLong et al.²¹ Significance of the tests was adopted at p < 0.05.

Results

One hundred and four patients were included in this study. The median age of the studied patients was 28 years with IQR 24–37. Most cases were females (57.69% of cases). The majority of cases (76.92%) alleged poisoning with one agent, while the remainder alleged ingestion of two or more drugs. Tramadol was the most frequently encountered drug (23.08%) followed by antipsychotics and multiple ingestions (19.23% each). The majority of toxic ingestions (96.15% of cases) were intentional. The median of pre-hospitalization period was 5 h (IQR = 3–6) (Table 1).

Table 1.

Patient characteristics in the two studied groups.

		Need for mechanical ventilation			Test statistic	p value
		Group I (n = 24)	Group II (n = 80)	Total (n = 104)	Test statistic	p value
Age^a (median (IQR)) (years)		29 (27–37)	27 (22–35)	28 (24–37)	792.00	0.194
Gender^b (N(%))	Male	12 (50.0%)	32 (40.0%)	44 (42.31%)	0.756	0.384
Gender^b (N(%))	Female	12 (50.0%)	48 (60.0%)	60 (57.69%)	0.756	0.384
Number of drugs^b (N(%))	Single	16 (66.7%)	64 (80.0%)	80 (76.92%)	1.849	0.174
Number of drugs^b (N(%))	Multiple	8 (33.3%)	16 (20.0%)	24 (23.08%)	1.849	0.174
Drug category^b (N(%))	Opioid	0 (0.0%)	24 (30.0%)	24 (23.08%)	14.498	0.008^c
	Benzodiazepine	4 (16.7%)	12 (15.0%)	16 (15.38%)
	Antipsychotic	8 (33.3%)	12 (15.0%)	20 (19.23%)
	Antidepressant	4 (16.7%)	8 (10.0%)	12 (11.54%)
	Antiepileptic	4 (16.7%)	8 (10.0%)	12 (11.54%)
	Multiple	4 (16.7%)	16 (20.0%)	20 (19.23%)
Manner^b (N(%))	Non-intentional	0 (0.0%)	4 (5.0%)	4 (3.85%)	25.718	<0.001^c
Manner^b (N(%))	Intentional	24 (100.0%)	76 (95.0%)	100 (96.15%)	25.718	<0.001^c
Outcome^b (N(%))	Recovery	12 (50.0%)	68 (85.0%)	76 (73.08%)	25.718	<0.001^c
	Complicated	8 (33.3%)	4 (5.0%)	12 (11.54%)
	Death	4 (16.7%)	0 (0.0%)	8 (7.69%)
	Discharge on request	0 (0.0%)	8 (10.0%)	8 (7.69%)
Prehospitalization period^a (median (IQR)) (hours)		6 (6–10)	4 (3–6)	5 (3–6)	632.00	0.009^c

IQR: interquartile range.

^aMann–Whitney U test.

^bχ² test of independence.

^cp is significant at <0.05.

Among the studied 104 patients, 24 cases required mechanical ventilation (group I). There was no statistically significant difference between the two studied groups as regards age, gender, number of ingested poison and manner. Group I had a significantly longer pre-hospitalization period (p = 0.009). As regards outcome of the studied patients, all cases that suffered from complications or died belonged to group I (Table 1). The most common complication was aspiration pneumonia in six cases followed by renal impairment, deep vein thrombosis and pneumothorax (three, two and one cases, respectively).

Table 2 illustrates that there was a statistically significant difference between group I and group II (p < 0.001) as regards the four evaluated scoring systems: GCS, APACHE II, RAPS and REMS. Group I had a significantly lower value of GCS and higher values of APACHE II, RAPS and REMS than group II (p < 0.001).

Table 2.

The median and IQR of evaluated scores in studied patients.

	Need for mechanical ventilation		Mann-Whitney U test
	Group I	Group II	Test statistic	p value
APACHE II median (IQR)	21 (14–23)	13 (8–16)	392.00	<0.001^a
GCS median (IQR)	5 (3–7)	9 (3–9)	568.00	0.002^a
RAPS median (IQR)	6 (4–8)	4 (2–4)	520.00	<0.001^a
REMS median (IQR)	9 (4–11)	4 (4-–7)	464.00	<0.001^a

IQR: interquartile range; APACHE II: acute physiology and chronic health evaluation; GCS: Glasgow Coma Scale; RAPS: rapid acute physiology score; REMS: rapid emergency medicine score.

^ap is significant at <0.05.

Table 3 and Figure 1 show the results of analysis of ROC curve in the studied scores. The optimal cut-off values of the four studied scores for predicting the need of mechanical ventilation were identified. The APACHE II score had the best AUC followed by REMS, RAPS then GCS. However, the differences between the AUC values of the four scores were not statistically significant. The APACHE II had 66.67% sensitivity and 90% specificity at a cut-off value >18. The GCS had a 100% sensitivity at a cut-off value ≤8, while REMS had a 100% specificity at cut-off value >8. The sensitivity and specificity of RAPS were 50% and 90%, respectively, at a cut-off value >6.

Table 3.

Comparison of the studied clinical scores in predicting the need for mechanical ventilation.

	APACHE II	GCS	RAPS	REMS
Cut-off value	>18	≤8	>6	>8
Sensitivity (%)	66.67	100.00	50.00	67.66
Specificity (%)	90.00	50.00	90.00	100.00
AUC	0.796	0.704	0.729	0.758
P (area = 0.5)	<0.001^a	<0.001^a	<0.001^a	<0.001^a
PPV	66.67%	37.5%	60%	100%
NPV	90%	100%	85.71%	90.90%
Pairwise comparison of ROC curves	z statistic	APACHE II vs. GCS: 1.748 APACHE II vs. RAPS = 1.818 APACHE II vs. REMS = 0.639 GCS vs. RAPS = 0.460 GCS vs. REMS = 0.675 RAPS vs. REMS = 0.609
Pairwise comparison of ROC curves	p value	APACHE II vs. GCS = 0.081 APACHE II vs. RAPS = 0.069 APACHE II vs. REMS = 0.523 GCS vs. RAPS = 0.645 GCS vs. REMS = 0.500 RAPS vs. REMS = 0.542

APACHE II: acute physiology and chronic health evaluation; GCS: Glasgow Coma Scale; RAPS: rapid acute physiology score; REMS: rapid emergency medicine score; PPV: positive predictive value; NPV: negative predictive value; ROC: receiver operating characteristic; AUC: area under the curve.

^ap is significant at <0.05.

Figure 1.

ROC curves for APACHE II score (a), GCS (b), RAPS (c) and REMS (d). APACHE II: acute physiology and chronic health evaluation; GCS: Glasgow Coma Scale; RAPS: rapid acute physiology score; REMS: rapid emergency medicine score.

Discussion

Multiple scoring systems have been studied in poisoned patients as prognostic factors for mortality including APACHE II,^12,22 modified APACHE II,²² GCS,^6,7,12,23,24 REMS,¹² RAPS¹² and other scoring systems.

However, few studies evaluated the use of clinical scores at hospital admission to determine the need of ICU admission or the need for mechanical ventilation. This study is one of the first studies to compare simple physiological scores in the emergency department as predictors of the need for mechanical ventilation in poisoned patients.

As regards GCS in this study, patients who required mechanical ventilation had significantly lower value of GCS. The GCS had a fair AUC, 100% sensitivity at a cut-off point ≤8 (it was able to identify all cases that were in need of mechanical ventilation) and it had a negative predictive value (NPV) 100% (the probability of a patient with GCS >8 not to need mechanical ventilation is 100%). This cut-off value is similar to the value indicated for intubation in head-injured patients.²⁵ Moreover, the previous studies that were conducted on poisoned patients showed similar results. Unverir et al.⁴ demonstrated that patients with antidepressant drug overdose and GCS scores of 8 or less were more likely to be intubated. Budhathoki et al.⁶ stated that GCS less than 8 was associated with mortality in poisoned children. In addition, Eizadi Mood et al.²² reported that patients with mixed drug poisoning who required intubation and ventilatory support had a GCS of 7.59 ± 0.52, while GCS in patients who were not intubated was 13.56 ± 0.17.

However, other values for GCS in ventilated patients were reported by some studies. Heyman et al.²⁶ concluded that intentionally poisoned patients with a GCS score of more than 6 may not be in need for ICU admission. Hamad et al.²⁷ illustrated that drug-overdosed patients with a GCS score of less than 13 were in need to be admitted to an ICU. Cattermole et al.²⁸ reported that the mean value of GCS in critically ill patients who either were admitted to ICU or died was 10.7 ± 4.9, while patients with good outcome had GCS of 14.2 ± 2.0. Eizadi Mood et al.²² suggested a cut-off value ≤10 for GCS (AUC was 0.933 ± 0.02 with sensitivity 90.91% and specificity 92.31%).

On the other hand, the predictive power of GCS was disputed by some studies. Donald et al.²⁹ stated that GCS alone is not a good predictor of ETI. Duncan and Thakore² suggested that GCS less than 8 does not mandate ETI.

This discrepancy in GCS score in different studies may be attributed to differences of the studied toxic agent or the poisoned patient population or the time at which GCS score was evaluated (as after using antidote).

As considers APACHE II score in this study, patients who underwent mechanical ventilation had significantly higher values of APACHE II at admission (21, 14–23 vs. 13, 8–16; p < 0.001) and it has the best AUC value (0.796) with cut-off value >18, 66.67% sensitivity and 90% specificity. In agreement to these results, Yoshida et al.³⁰ found that the mean value of APACHE II score in intubated patients with acute lung injury was significantly higher than those who were not intubated (20 ± 3 vs. 15 ± 5) and recommended a cut-off value >17 (AUC = 0.792, 86% sensitivity and 73% specificity). Additionally, Cattermole et al.²⁸ reported a significantly higher value of APACHE II score in critically ill patients who either died or were admitted to ICU as compared to those with good prognosis (18.8 ± 7.9 vs. 12.3 ± 6.8, respectively). Meanwhile, Safavi and Honarmand³¹ reported a lower AUC for APACHE II score (0.74) at cut-off value 12 with sensitivity 90.4% and lower specificity (32.3%) for ETI.

In this study, RAPS value was significantly higher in mechanically ventilated patients (6, 4–8 vs. 4, 2–4; p < 0.001) and its AUC showed a fair result. The RAPS had a lower sensitivity (50%) than the other three scores, but its specificity (90%) was comparable to APACHE II and higher than GCS. No other studies evaluated the use of RAPS as a predictor to the need of mechanical ventilation up to the best of our knowledge.

Regarding REMS in the current study, patients in need of mechanical ventilation had significantly higher score (9, 4–11 vs. 4, 4–7; p < 0.001) and AUC represented a fair result and the score had a 100% specificity at a cut-off level >8. This indicates that REMS was able to predict all patients who were not in need of mechanical ventilation. Moreover, REMS had a positive predictive value (PPV) of 100% (the probability that a patient with REMS >8 will need mechanical ventilation is 100%) and a NPV of 90.90%. In line with this result, Olsson et al.¹¹ reported that REMS was a good predictor of long-term mortality in patients attending non-surgical emergency department. Cattermole et al.²⁸ reported that REMS score was significantly higher in patients admitted to ICU or who died as compared to patients with good prognosis (9.1 ± 4.3 vs. 6.1 ± 4.0, respectively). In addition, Carrillo and Urrutia³² found that REMS was significantly associated with the need for hemodynamic and ventilator support in critically ill patients. Although REMS is derived from RAPS, the former score had better sensitivity, specificity and predictive values in this study. In accordance, Olsson and Lind³³ found that REMS was comparable to APACHE II, but superior to RAPS, for prediction of in-hospital mortality of non-surgical patients in the emergency department.

In the present study, although APACHE II score had the highest AUC compared to the other scores, there was no statistically significant difference between them. This result coincides with Grmec and Gasparovic³⁴ who found no statistically significant difference between APACHE II and GCS for prediction of mortality in patients with non traumatic coma and Olsson and Lind³³ who compared APACHE II and REMS. Moreover, Söyüncü and Bektaş¹² reported that despite APACHE II has the best AUC value, differences between it and GCS, RAPS and REMS were not significant as regards predicting mortality in poisoned patients admitted to ICU. The APACHE II score includes several blood chemistry variables and is therefore not suitable for quick scoring in the emergency room. On the other hand, GCS, RAPS and REMS neither require any laboratory parameters nor any staff training and extra skills. Therefore, simple clinical scores as GCS, RAPS and REMS, rather than complicated scores as APACHE II, may be used with reasonable results in the emergency room to predict the need of mechanical ventilation. Among the four evaluated scores in this study, REMS had the highest PPV and NPV.

Conclusions

In conclusion, admission REMS seems to be a valuable prognostic tool in acute poisoning for recognizing drug-overdosed patients with DCL in need of mechanical ventilation. The use of REMS is advocated as REMS >8 identified all cases in need of mechanical ventilation (PPV 100%). The combined use of GCS and REMS is recommended as GCS >8 had the added benefit of excluding patients not in need of ventilator support (NPV 100%).

Limitations

There are some limitations in our study.

It is a single-centre study, which may not be representative of all patients. The pattern of drug overdose encountered in this study represents the common pattern in the Nile Delta and may not be applicable to other localities.

The overall number of our patients was relatively small.

Therefore, we suggest performing a larger prospective study of poisoned patients, preferably multicentre study, to evaluate those clinical scores as predictors for the need of mechanical ventilation.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Stevens

Bhardwaj

. Approach to the comatose patient. Crit Care Med 2006; 34: 31–41.

Duncan

Thakore

. Decreased Glasgow Coma Scale score does not mandate endotracheal intubation in the emergency department. J Emerg Med 2009; 37: 451–455.

Kornbluth

Bhardwaj

. Evaluation of Coma: a critical appraisal of popular scoring systems. Neurocrit Care 2011; 14: 134–143.

Unverir

Atilla

Karcioglu

. A retrospective analysis of antidepressant poisonings in the emergency department: 11-year experience. Hum Exp Toxicol 2006; 25: 605–612.

Akkose

Turkmen

Bulut

. An analysis of carbon monoxide poisoning cases in Bursa, Turkey. East Mediterr Health J 2010; 16: 101–106.

Budhathoki

Poudel

Shah

. Clinical profile and outcome of children presenting with poisoning or intoxication: a hospital based study. Nepal Med Coll J 2009; 11: 170–175.

Christ

Arranto

Schindler

. Incidence, risk factors, and outcome of aspiration pneumonitis in ICU overdose patients. Intensive Care Med 2006; 32: 1423–1427.

Eizadi Mood

Sabzghabaee

Yadegarfar

. Glasgow coma scale and its components on admission: are they valuable prognostic tools in acute mixed drug poisoning? Crit Care Res Pract 2011; 2011: 952956.

Yang

Lee

. Predictors of carbon monoxide poisoning-induced delayed neuropsychological sequelae. Gen Hosp Psychiatry 2010; 32: 310–314.

10.

Teksam

Gumus

Bayrakci

. Acute cardiac effects of carbon monoxide poisoning in children. Eur J Emerg Med 2010; 17: 192–196.

11.

Olsson

Terent

Lind

. Rapid emergency medicine score can predict long-term mortality in nonsurgical emergency department patients. Acad Emerg Med 2004; 11: 1008–1013.

12.

Söyüncü

Bektaş

. Comparison of the scoring systems for predicting mortality in intoxicated patients hospitalized to the ICU: a prospective observational study. Erciyes Med J 2011; 33: 29–34.

13.

Chan

Gaudry

Grattan-Smith

. The use of Glasgow coma scale in poisoning. J Emerg Med 1993; 11: 579–582.

14.

Heard

Bebarta

. Reliability of the Glasgow coma scale for the emergency department evaluation of poisoned patients. Hum Exp Toxicol 2004; 23: 197–200.

15.

Kelly

Upex

Bateman

. Comparison of consciousness level assessment in the poisoned patient using the alert/verbal/painful/unresponsive scale and the Glasgow coma scale. Ann Emerg Med 2004; 44: 108–113.

16.

Teasdale

Jennett

. Assessment of coma and impaired consciousness: a practical scale. Lancet 1974; 304: 81–84.

17.

Knaus

Draper

Wagner

. APACHE II: a severity of disease classification system. Crit Care Med 1985; 13: 818–829.

18.

Rhee

Fisher

Willitis

. The rapid acute physiology score. Am J Emerg Med 1987; 5: 278–282.

19.

Olsson

Terént

Lind

. Rapid emergency medicine score: a new prognostic tool for in-hospital mortality in nonsurgical emergency department patients. J Intern Med 2004; 255: 579–587.

20.

Dawson

Trapp

. Basic and clinical biostatics. 3rd ed. New York: McGrow Hill Medical Publishing Division, 2001.

21.

DeLong

Clarke-Pearson

. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44: 837–845.

22.

Eizadi Mood

Sabzghabaee

Khalili-Dehkordi

. Applicability of different scoring systems in outcome prediction of patients with mixed drug poisoning-induced coma. Indian J Anaesth 2011; 55: 599–604.

23.

Bilgin

Camdeviren

Yapici

. The comparison of the efficacy of scoring systems in organophosphate poisoning. Toxicol Ind Health 2005; 21: 141–146.

24.

Eizadi-Mood

Saghaei

Jabalameli

. Predicting outcomes in organophosphate poisoning based on APACHE II and modified APACHE II scores. Hum Exp Toxicol 2007; 26: 573–578.

25.

Gentleman

Dearden

Midgley

. Guidelines for resuscitation and transfer of patients with serious head injury. BMJ 1993; 307: 547–552.

26.

Heyman

LoCastro

Gouse

. Intentional drug overdose: predictors of clinical course in the intensive care unit. Heart Lung 1996; 25: 246–252.

27.

Hamad

Al-Ghadban

Carvounis

. Predicting the need for medical intensive care monitoring in drug-overdosed patients. J Intensive Care Med 2000; 15: 321–328.

28.

Cattermole

Mak

SKP

Liow

CHE

. Derivation of a prognostic score for identifying critically ill patients in an emergency department resuscitation room. Resuscitation 2009; 80: 1000–1005.

29.

Donald

Duncan

Thakore

. Predictors of the need for rapid sequence intubation in the poisoned patient with reduced Glasgow coma score. Emerg Med J 2009; 26: 510–512.

30.

Yoshida

Takeda

Akada

. Factors predicting successful noninvasive ventilation in acute lung injury. J Anesth 2008; 22: 201–206.

31.

Safavi

Honarmand

. Comparison of infection probability score, APACHE II, and APACHE III scoring systems in predicting need for ventilator and ventilation duration in critically ill patients. Arch Iran Med 2007; 10: 354–360.

32.

Carrillo

Urrutia

. Risk assessment of critical patients requiring ambulance transfers. Rev Med Chil 2012; 140: 1297–1303.

33.

Olsson

Lind

. Comparison of the rapid emergency medicine score and APACHE II in nonsurgical emergency department patients. Acad Emerg Med 2003; 10: 1040–1048.

34.

Grmec

Gasparovic

. Comparison of APACHE II, MEES and Glasgow coma scale in patients with nontraumatic coma for prediction of mortality. Crit Care 2001; 5: 19–23.