- Email: [email protected]
A Comparison of the Intracerebral Hemorrhage Score and the Acute Physiology and Chronic Health Evaluation II Score for 30-Day Mortality Prediction in Spontaneous Intracerebral Hemorrhage
ARTICLE IN PRESS
A Comparison of the Intracerebral Hemorrhage Score and the Acute Physiology and Chronic Health Evaluation II Score for 30-Day Mortality Prediction in Spontaneous Intracerebral Hemorrhage Koushik Pan,
MD,*
Ajay Panwar, MD, DM,† Ujjawal Roy, Bidyut K. Das, MD‡
MD,*
and
Background: The intracerebral hemorrhage (ICH) score is well established as a reliable prognostic score in ICH, whereas recently, Acute Physiology and Chronic Health Evaluation II (APACHE II) has been observed to have a better discrimination in predicting mortality in primary pontine hemorrhage. Further, physiological parameters of APACHE II have been associated with outcome in ICH. This study is the first to observe a direct comparison between APACHE II and ICH scores in predicting 30-day mortality in spontaneous intracerebral hemorrhage (SICH). Materials and Methods: This study was a prospective observational study where we compared the receiver operating characteristic (ROCs) of baseline ICH and APACHE II scores in patients with SICH for predicting 30-day mortality outcome. Results: We observed that both APACHE II and ICH scores were good for predicting 30-day mortality with both having an area under the ROC curve of more than .8 (.831 [95% confidence interval {CI}, .740-.922; P < .001] and .892 [95% CI, .757-.932; P < .001], respectively). However, the ICH score was better discriminative (area under the curve AUC, .892 versus .831; P = .040) and better calibrated (P = .037 versus P = .089, Hosmer–Lemeshow goodness-of-fit test for logistic regression) for the same. Both APACHE II and ICH scores had a sensitivity of 87% at cutoff values of 19 and 3, respectively; however, the ICH score had a better specificity (90% versus 76.5%). Conclusion: The ICH score was observed to have a better discrimination and calibration for predicting 30-day mortality in SICH. Key Words: ICH score—APACHE II score—intracerebral hemorrhage—receiver operating characteristic—mortality. © 2017 National Stroke Association. Published by Elsevier Inc. All rights reserved.
From the *Department of Neurology, Institute of Post Graduate Medical Education and Research (IPGMER), Kolkata, India; †Department of Neurology, Kakatiya Medical College and Mahatma Gandhi Memorial Hospital, Warangal, India; and ‡Department of General Medicine, Calcutta National Medical College and Hospital, Kolkata, India. Received March 4, 2017; revision received May 10, 2017; accepted June 1, 2017. Author contributions: Dr. Koushik Pan contributed in the literature search, study design, data collection, data analysis, and data interpretation. Dr. Ajay Panwar contributed in the data analysis, data interpretation, literature search, figures, writing of the manuscript, and study design. Dr. Ujjawal Roy contributed in the data analysis, literature search, writing of the manuscript, and study design. Dr. Bidyut K. Das contributed in the data analysis and literature search. Address correspondence to Ajay Panwar, MD, DM, Department of Neurology, Kakatiya Medical College and Mahatma Gandhi Memorial Hospital, Warangal, 506007, India. E-mail: [email protected]. 1052-3057/$ - see front matter © 2017 National Stroke Association. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2017.06.005
Journal of Stroke and Cerebrovascular Diseases, Vol. ■■, No. ■■ (■■), 2017: pp ■■–■■
1
ARTICLE IN PRESS K. PAN ET AL.
2
Introduction Scoring systems are considered vital for predicting outcomes in patients with critical illnesses. These scoring systems help in resource allocation, clinical decisions, and quality assurance of intensive care patients.1 Spontaneous intracerebral hemorrhage (SICH) is considered to be the stroke subtype with the worst prognosis. Until a decade ago, no standard early prognostic grading scale for acute hemorrhagic stroke existed. Thereafter, logistic regression was used to identify independent predictors of 30day mortality in case of acute intracerebral hemorrhage (ICH), and a risk stratification scale, named as the ICH score, was developed based on the strength of association of the predictors.2 Subsequently, several studies have proved the ICH score as a reliable and accurate predictor of 30-day mortality in SICH.3-5 On the other hand, the Acute Physiology and Chronic Health Evaluation II (APACHE II) score is a simple and probably the most widely used score to assess the severity of illness in patients admitted to the intensive care unit.6,7 The APACHE II score has been reported to be useful in predicting the outcomes in patients with intracranial infections, acute stroke, and traumatic brain injury, besides being useful in non-neurological conditions as well. In consideration of the fact that the APACHE II score is based on physiological parameters, whereas the ICH score is calculated primarily on the basis of volume, origin, and extent of ICH, both scores may have a bearing on the outcome of patients admitted with SICH. Niskanen et al studied 2 models of APACHE II score in patients with head injury and nontraumatic ICH and reported that “model A,” based on physiological parameters, performed better for mortality prediction in comparison with “model B” based on the Glasgow Coma Scale (GCS) alone. Huang et al compared the ICH score with the APACHE II sore in predicting 30-day mortality in primary pontine hemorrhage (PPH) and found APACHE II to be more discriminative.8 In another study, the APACHE II score was observed to be more discriminative than the simplified acute physiology score II, GCS, and the National Institutes of Health Stroke Scale (NIHSS) scores for mortality prediction in hemorrhagic stroke (ICH).9 There is no published data on the comparison of APACHE II and ICH scores in ICH. Based on the promising results of the above-mentioned studies, we aimed to compare APACHE II and ICH scores for predicting 30-day mortality outcome in SICH.
Materials and Methods The present study was a hospital-based prospective observational comparative study that started on February 2015 and continued up to July 2016. The study protocol was approved by the institutional ethical committee. The study population comprised patients with SICH admitted in the critical care unit and neurology ward of the
Institute of Post Graduate Medical Education and Research at Kolkata. ICH was diagnosed on the basis of a computerized tomography (CT) scan of the brain.
Study Group Patients aged 18 years or older who were admitted with symptoms of acute stroke and were subsequently diagnosed to have SICH were enrolled in the study after a written informed consent.
Exclusion Criteria All the patients having ICH secondary to a brain tumor, trauma, or vascular malformation were excluded from the study. Those having hemorrhagic transformation of a cerebral infarct also met the exclusion criteria. Further, patients with a past history of stroke and those having other comorbid fatal conditions like end-stage renal disease, ischemic heart disease, and sepsis were excluded from the study. The study group comprised only SICH patients undergoing medical management, and any patient who underwent surgery was excluded from the study.
Methodology and Patient Evaluation All patients with a suspected diagnosis of acute stroke were subjected to an urgent noncontrast CT scan of the brain within 1 hour of admission. Soon after, those diagnosed as SICH were enrolled and physically examined along with an evaluation for APACHE II and ICH scores. The scores used for 30-day mortality prediction were based on the first evaluation after the enrollment. The data of the study subjects were collected by oral questionnaire regarding age, sex, education level, and risk factors for SICH (hypertension, diabetes, alcoholism, smoking, and anticoagulant and antiplatelet medications). Pulse, blood pressure (BP), respiratory rate, temperature, and GCS score were noted on admission. The patients were classified into 3 groups based on the GCS score: (1) GCS 3-4, (2) GCS 5-11, and (3) GCS 12-15.10
Treatment Protocol All enrolled patients were initially assessed and treated for airway patency, breathing sufficiency, BP, and signs of raised intracranial pressure. Emergent measures for lowering intracranial pressure (head end elevation to 30°, intravenous 20% mannitol infusion at a dose of .25-1.0 g/kg, and hyperventilation to achieve a pCO2 of 30-35 mm Hg) were instituted in all those cases that presented with a comatose or stuporous state of consciousness and those having signs of brainstem herniation. Early intensive BP reduction was done in all patients with elevated systolic blood pressures (SBPs) to achieve a target of SBP of less than 140 mm Hg.11-15 BP monitoring was done every 15 minutes for an initial 2 hours and then on an hourly
ARTICLE IN PRESS COMPARISON OF ICH AND APACHE II SCORES FOR MORTALITY PREDICTION
basis for the next 24 hours. The mean systolic blood pressure (mSBP) of the initial 24 hours was used for statistical correlation and regression analysis. Intravenous labetalol was the primary antihypertensive agent used for intensive BP lowering as it was locally and easily available. Labetalol was used as an initial bolus dose of 20 mg followed by boluses of 20-80 mg and as an infusion in some cases, beginning at 1-2 mg/min titrated up to the desired SBP. Labetalol was used up to the maximum cumulative dose of 300 mg over 24 hours.16 Patients with a coagulation factor deficiency or thrombocytopenia were treated with appropriate factor replacements or platelet transfusions. The patients who were on vitamin K antagonist therapy had their vitamin K antagonist withheld along with fresh frozen plasma and intravenous vitamin K therapy aimed at a rapid correction of the international normalized ratio. Patients with seizures and those with a change in mental status, along with electrographic seizures, were treated with antiepileptic medicines. The water swallow test was done as a screening procedure for dysphagia before initiating oral intake in patients. Surgical Treatment Emergent hematoma evacuation was done in cerebellar hemorrhage patients who were deteriorating neurologically or those who were having brainstem compression and hydrocephalus from ventricular obstruction.17 All the cases with cerebellar hemorrhage were assessed by the neurosurgeon, soon after being diagnosed. Those who were stuporous or comatose and those who had signs of raised intracranial pressure, along with evidence of brainstem compression or hydrocephalus on brain CT, were operated on an emergent basis. Further, those who were clinically stable initially but developed a subsequent neurological deterioration with a decline of more than 2 points in the GCS score were made to undergo an urgent repeat CT scan of the brain and, if observed to have a worsening mass effect or hydrocephalus, were scheduled for immediate surgery. Those patients who underwent surgery were thereafter excluded from the study.
Outcome Assessment and Follow-Up Primary outcome was defined as mortality assessment at 30 days after SICH. Patients who survived and were discharged were followed up in the outpatient department and over telephone regarding their final outcome.
Statistical Analysis The data entry and analyses were done on software statistical package SPSS version 16.0 (IBM, Inc, Chicago, IL, USA). Data were summarized as mean ± standard deviation for numerical variables and counts and percentages
3
for categorical variables. Chi-square test was used for the assessment of statistical significance of demographic parameters. A P value less than .05 was considered significant. Discriminative power, cutoff values, sensitivity, and specificity for APACHE II and ICH scores were determined from the receiver operating characteristic (ROC) curve. Multivariate logistic regression analysis was used to determine the strength of association with 30-day mortality for the components of the APACHE II and ICH scores.
Results Baseline Characteristics A total of 109 patients were initially enrolled, out of which 5 patients underwent surgery and were thus excluded from the study. So, the final study group comprised 104 patients (Fig 1). The mean age of the study subjects was 61 years and the study group was composed mostly of men (66.3%) patients. Most of the study subjects were smokers, hypertensives, and residents of rural areas, and were in the lower income and lower education categories. Almost 67.30% of the study population died, and most of the deaths occurred among rural study populations who were smokers and belonged to low income and low education categories (P < .001). The frequencies of intraventricular hemorrhage (IVH) and hemorrhage with a volume of 30 cc or higher were significantly higher among the subjects who died (P < .001). Those who survived had a lower mSBP (136.6 ± 3.7 versus 140.4 ± 3.4) than those who died, the difference being statistically significant on an “independent t-test” (P < .001) (Table 1). The mean volume of ICH was 33.23. The mean APACHE II and ICH scores for the study population were observed to be 22.03 and 2.81, respectively. The majority of the hemorrhagic strokes occurred in the fifth to seventh decades. Among the individual components of the ICH score, a hemorrhage volume of 30 cc or higher was observed in 78% of the cases (81 out of 104), whereas 57% (59 out of 104) had a IVH. Supratentorial bleed was more common, being present in 88% of the individuals (92 out of 104).
Association between GCS Score and 30-Day Mortality The numbers of patients in the 3 GCS based groups were 33.6% (GCS 3-4), 67.7% (GCS 5-11), and 8.7% (GCS 12-15). The mortality rates at 30 days in the 3 groups were 88.6%, 61.7%, and 22.2%, respectively. This difference in the mortality rates was statistically significant (P < .001). Thus, mortality was significantly associated with a low GCS score on admission.
Multivariate Logistic Regression Analysis of the Predictive Mortality of Components of the APACHE II and ICH Scores Among the components of the APACHE II score, the GCS value was observed to be a significant independent
ARTICLE IN PRESS K. PAN ET AL.
4
Figure 1. Flow diagram showing the study design. Abbreviations: GCS, Glasgow Coma Scale; ICH, intracerebral hemorrhage; SBP, systolic blood pressure; SICH, spontaneous intracerebral hemorrhage.
predictor of mortality (P < .001). None of other parameters had a statistically significant association (Table 2). Concerning the components of the ICH score, the GCS score, the hematoma volume (≥30 cc), and the presence of IVH were significant independent predictors of mortality (P = .016, P = .026, and P = .004, respectively). The infratentorial origin of hemorrhage and age of 80 years or older were not observed to have a significant association with the 30-day mortality outcome; however, this
observation was limited by a very small number of patients in the respective categories (Table 3).
SBP as a Predictor of 30-Day Mortality and Its Association with the Significant Demographic Variables Among variables other than the components of the APACHE II and ICH scores, logistic regression analysis
Table 1. Association of 30-day mortality with demographic variables and risk factors
Sex Residence Education Diabetes Hypertension Smoking Hemorrhage volume (≥30 cc) Intraventricular hemorrhage Infratentorial origin mSBP (±SD)
Total no. = 104
Survival (N = 34)
Death (N = 70)
P Value
Male: 69 (66.3) Femal:e 35 (33.7) Rural: 56 (53.8) Urban: 48 (46.2) Lower: 67 (64.4) (
23 (33.3) 11 (31.4) 7 (12.5) 27 (56.25) 5 (7.46) 29 (78.38) 14 (35.9) 24 (31.2) 5 (7.1) 15 (18.5) 10 (16.9) 1 (8.3) 136.6 (±3.7)
46 (66.7) 24 (68.6) 49 (87.5) 21 (43.75) 62 (92.54) 8 (21.62) 25 (64.1) 53 (68.8) 65 (92.9) 66 (81.4) 49 (83.1) 11 (91.7) 140.4 (±3.4)
.84
Abbreviations: mSBP, mean systolic blood pressure; SD, standard deviation. Numbers in parentheses represent percentages. Boldfaced values indicate significant difference.
<.001 <.001 .91 .93 <.001 <.001 <.001 .09 <.001
ARTICLE IN PRESS COMPARISON OF ICH AND APACHE II SCORES FOR MORTALITY PREDICTION
Table 2. Multivariate logistic regression analysis of predictive mortality of components of the Acute Physiology and Chronic Health Evaluation II score
Components
P Value
Odds ratio
95% Confidence interval
Age Mean arterial pressure White blood cell PaO2 Potassium Heart rate Potential of hydrogen Creatinine Glasgow Coma Scale Temperature Respiratory rate Sodium Hematocrit
.647 .225 .104 .177 .095 .081 .523 .567 <.001 .671 .684 .242 .314
.988 1.017 1.000 .980 .659 .966 .075 1.179 .661 .802 .976 .965 1.046
.940-1.039 .990-1.045 .975-1.063 .953-1.009 .403-1.076 .930-1.004 .000-2.154 .672-2.069 .533-.820 .290-2.218 .870-1.096 .910-1.024 .958-1.141
The boldfaced value indicates significant difference.
also revealed mSBP as a strong independent predictor of 30-day mortality (P < .001, odds ratio 1.455 [1.164-1.817]). Further, on investigating the significant association of poor and low educated rural population with mortality as revealed on descriptive analysis, we performed a logistic regression analysis keeping “residence” as a response variable and logistically observed significant predictors of the 30-day mortality as independent variables. Consequently, we observed that rural residence was strongly associated with a higher achieved mSBP but none of the other predictors. “Lower education,” “lower income,” and “smoking” had a significant correlation with “rural residence” (Pearson correlation, P < .001) and hence were not entered in the logistic regression model. Thus, the higher mortality rate in the rural subgroup could be explained by its direct association with the inferior control of SBP (Supplementary Tables S1, S2).
5
Table 4. Receiver operating characteristics of APACHE II and ICH scores
Cutoff value AUC (95% CI) P Value Sensitivity (%) Specificity (%)
APACHE II
ICH
19 .831 (.740-.922) <.001 87 76.5
3 .892 (.757-.932) <.001 87 90
Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; AUC, area under the curve; CI, confidence interval; ICH, intracerebral hemorrhage.
ROC Curve and Calibration of APACHE II and ICH Scores The areas under ROC curves for the APACHE II and ICH scores were .831 (95% confidence interval, .740.922; P < .001) and .892 (95% confidence interval, .757.932; P < .001), respectively (Table 4). Thus, both scores having an area under the curve higher than .8 appeared good for 30-day mortality risk prediction; still, the discriminative power of the ICH score was higher (area under the curve, .892 versus .831; P = .040) Both ROCs suggested a sensitivity of 87% at cutoff values of 19 and 3, respectively, for the APACHE II and ICH scores. However, at the same cutoff values, the APACHE II and ICH scores had a specificity of 76.5% and 90%, respectively (Fig 2). So, the APACHE II and ICH scores both
Table 3. Multivariate logistic regression analysis of predictive mortality of components of the intracerebral hemorrhage score Odds 95% Confidence ratio interval
Components
P Value
Glasgow Coma Scale Hemorrhage volume (≥30 cc) Intraventricular hemorrhage Infratentorial origin Age
.016 .026
.781 6.750
.639-.955 1.250-36.455
.004
6.827
1.855-25.123
.081 .987
6.448 1.024
.797-52.156 .055-18.998
The boldfaced value indicates significant difference.
Figure 2. ROC curves for comparison of ICH and APACHE II scores in predicting 30-day mortality outcome. Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; ICH, intracerebral hemorrhage; ROC, receiver operating characteristic.
ARTICLE IN PRESS K. PAN ET AL.
6
had a similarly good sensitivity, albeit the ICH score was observed to be more specific. Further, the ICH score was observed to have a better calibration than the APACHE II (P = .037 versus P = .089) in the Hosmer–Lemeshow goodness-of-fit test for logistic regression analysis. Thus, the ICH score performed better in terms of discrimination as well as calibration.
Discussion The descriptive part of our data analysis revealed that the survival rate in ICH was better in urban individuals and in the higher education group. Earlier, Das et al. reported a better knowledge of stroke symptoms and risk factors among urban residents who were highly educated and those falling in the high-income category.18 However, considering that a better knowledge of risk factors may be associated with hemorrhagic stroke prevention but should not have a direct relationship with 30-day prognostic outcome after onset, we explored the same on logistic regression analysis. Consequently, residence, which was strongly correlated with income and educational status, had a strong association with mSBP, which, in turn, was observed to be an independent predictor of 30-day mortality. Urban residents had a better controlled mSBP, probably due to good affordability and prompt purchase of medications, and hence a lower 30day mortality. However, in our opinion, this is a preliminary observation that needs to be tested in further studies conducted in similar demographic conditions. We also observed a strong association between low GCS score and 30-day mortality, with death rate being highest in the patient group having a GCS score of 3-4. These results are in agreement with the earlier observations by Muengtaweepongsa and Seamhan, who observed a significantly high mortality rate in ICH among Thai patients having GCS scores of 3-4.10 On multivariate analysis of the predictive mortality of the components of the APACHE II score, only the GCS score was observed to be an independent predictor of 30-day mortality. In a previous study, Niskanen et al used logistic regression to study the association of various components of the APACHE II score with hospital mortality in patients with head injury or nontraumatic intracranial hemorrhage. Niskanen et al created 2 models from the APACHE II score and observed that model A, based on physiological parameters, was better than model B, based on GCS alone, to estimate the probability of death. It was thus suggested that the physiological parameters of APACHE II scores could contribute toward estimating the probability of death in neurological cases. In the same study, the GCS score, along with the mean arterial pressure and age, was observed to be the independent predictor of hospital death.19 In the present study, we did not observe any of the physiological parameters except GCS as the independent predictors of mortality. However,
our study group was different, consisting of SICH patients only, which could explain the disparity among the mortality predictors, in comparison with Niskanen et al’s study. In another study, Huang et al had shown that GCS alone had an excellent discrimination to predict 30-day mortality in PPH.8 Multivariate analysis of the components of ICH score in our study subjects revealed that GCS, hematoma volume (≥30 cc), and the presence of IVH were strong predictors of 30-day mortality. However, the infratentorial origin of hemorrhage, despite having high odds for death, was not significantly associated with the mortality prediction, which is in contradiction with the previous studies.2,10 This discordance could possibly be due to very small number of patients having infratentorial hemorrhage (12% infratentorial versus 88% supratentorial), which could not reach statistical significance. Age 80 years or older was an independent predictor of mortality in the original study of the ICH score.2 The lack of association between age of 80 years or older and mortality in the present study may be due to the small number of patients older than 80 years. On ROC analysis for mortality prediction, we observed that both the APACHE II and ICH scores had a good sensitivity of 87% at cutoff scores of 19 and 3, respectively. However, at the same sensitivity, the ICH score was observed to be far more specific (90%) than the APACHE II score (76.5%). The ICH score’s cutoff of 3 or higher for mortality prediction is in concordance with a study done by Cheung and Zou, where they compared 3 types of ICH scores (original, modified, and new) and observed a high mortality rate and a 0 rate of good outcome with a cutoff score of 3 or higher in all the 3 ICH scores. Among the 3 ICH scores, Cheung and Zou observed the “original ICH score” to be most specific, with a sensitivity of 78.6% and a specificity of 90.4%.20 Di Napoli et al also reported the original ICH score to have a good specificity of 84% with a sensitivity of 75% for predicting 30-day mortality outcome.21 Thus, our observations (87% sensitivity and 90% specificity) reinforce the already reported high specificity of ICH score in predicting 30-day mortality outcome albeit with a higher sensitivity. Unlike the ICH score, there are limited data on the use of the APACHE II as a prognostic score in ICH. Huang et al observed the APACHE II score to have a sensitivity and a specificity of 91% and 86.5% respectively, at a cutoff score of 16.5 in predicting 30-day mortality in PPH.8 However, we also observed a nearly high sensitivity of 87%, but with a lower specificity of 76.5%. Further, Huang et al also observed that the APACHE II score had a higher discriminative power as compared with the ICH score for predicting 30-day mortality in PPH. However, in our study, we observed the ICH score to have a higher discriminative power and a better calibration than the APACHE II score. The differences in results in our study could be due to a different study cohort as we enrolled all patients with ICH and not just PPH.
ARTICLE IN PRESS COMPARISON OF ICH AND APACHE II SCORES FOR MORTALITY PREDICTION
The present study had a few limitations. We probably could not estimate the true strength of association between “infratentorial origin of hemorrhage” and “age ≥ 80 years” while analyzing the predictive mortality of components of ICH score because of the small number of subjects in the respective categories. Further, ours was a single-center study and the results of the same need to be confirmed by a multicentric study with a larger sample size. In conclusion, we observed that both APACHE II and ICH scores on presentation can be used to have an approximate prediction of 30-day mortality in ICH. However, the ICH score has a better discriminative power and is better calibrated than the APACHE II score for the same.
9.
10.
11.
12.
13.
Appendix: Supplementary Material Supplementary data to this article can be found online at doi:10.1016/j.jstrokecerebrovasdis.2017.06.005.
References 1. Rapoport J, Teres D, Lemeshow S, et al. A method for assessing the clinical performance and cost-effectiveness of intensive care units: a multicenter inception cohort study. Crit Care Med 1994;22:1385-1391. 2. Hemphill JC, Bonovich DC, Besmertis L, et al. The ICH score. a simple, reliable grading scale for intracerebral hemorrhage. Stroke 2001;32:891-897. 3. Godoy DA, Piñero G, Di Napoli M. Predicting mortality in spontaneous intracerebral hemorrhage. Can modification to original score improve the prediction? Stroke 2006;37:1038-1044. 4. Aysenne AM, Albright KC, Mathias T, et al. 24-Hour ICH score is a better predictor of outcome than admission ICH score. ISRN Stroke 2013;2013:4. 5. Meyer DM, Begtrup K, Grotta JC, et al. Is the ICH score a valid predictor of mortality in intracerebral hemorrhage? J Am Assoc Nurse Pract 2015;27:351-355. 6. Vincent J-L, Moreno R. Clinical review: scoring systems in the critically ill. Crit Care 2010;14:207. 7. Bouch DC, Thompson JP. Severity scoring systems in the critically ill. CEACCP 2008;8:181-185. 8. Huang KB, Ji Z, Wu YM, et al. The prediction of 30-day mortality in patients with primary pontine hemorrhage:
14.
15.
16. 17.
18.
19.
20.
21.
7
a scoring system comparison. Eur J Neurol 2012;19:12451250. Moon BH, Park SK, Jang DK, et al. Use of APACHE II and SAPS II to predict mortality for hemorrhagic and ischemic stroke patients. J Clin Neurosci 2015;22:111-115. Muengtaweepongsa S, Seamhan B. Predicting mortality rate with ICH score in Thai intracerebral hemorrhage patients. Neurol Asia 2013;18:131-135. Sakamoto Y, Koga M, Yamagami H, et al. Systolic blood pressure after intravenous antihypertensive treatment and clinical outcomes in hyperacute intracerebral hemorrhage. The stroke acute management with urgent risk-factor assessment and improvement-intracerebral hemorrhage study. Stroke 2013;44:1846-1851. Butcher KS, Jeerakathil T, Hill M, et al. The intracerebral hemorrhage acutely decreasing arterial pressure trial. Stroke 2013;44:620-626. Arima H, Huang Y, Wang JG, et al. Earlier blood pressure-lowering and greater attenuation of hematoma growth in acute intracerebral hemorrhage: INTERACT pilot phase. Stroke 2012;43:2236-2238. Arima H, Anderson CS, Wang JG, et al. Lower treatment blood pressure is associated with greatest reduction in hematoma growth after acute intracerebral hemorrhage. Hypertension 2010;56:852-858. Anderson CS, Heeley E, Huang Y, et al. Rapid bloodpressure lowering in patients with acute intracerebral hemorrhage. N Engl J Med 2013;368:2355-2365. Varon J. The diagnosis and treatment of hypertensive crises. Postgrad Med 2009;121:5-13. Morgenstern LB, Hemphill JC 3rd, Anderson C, et al. Guidelines for the management of spontaneous intracerebral hemorrhage: a guideline for healthcare professionals from the American Heart Association/ American Stroke Association. Stroke 2010;41:2108-2129. Das BK, Karmakar PS, Santra G, et al. Awareness of stroke among elderly public in Eastern India. Neurol Asia 2011;16:119-126. Niskanen MM, Kari A, Hernesniemi JA, et al. Contribution of non-neurologic disturbances in acute physiology to the prediction of intensive care outcome after head injury or non-traumatic intracranial haemorrhage. Intensive Care Med 1994;20:562-566. Cheung RTF, Zou L-Y. Use of the original, modified, or new intracerebral hemorrhage score to predict mortality and morbidity after intracerebral hemorrhage. Stroke 2003;34:1717-1722. Di Napoli M, Godoy DA. Clinical grading scales in spontaneous intracerebral hemorrhage. Stroke 2007; 38:e133-e135.
A Comparison of the Intracerebral Hemorrhage Score and the Acute Physiology and Chronic Health Evaluation II Score for 30-Day Mortality Prediction in Spontaneous Intracerebral Hemorrhage Koushik Pan,
MD,*
Ajay Panwar, MD, DM,† Ujjawal Roy, Bidyut K. Das, MD‡
MD,*
and
Background: The intracerebral hemorrhage (ICH) score is well established as a reliable prognostic score in ICH, whereas recently, Acute Physiology and Chronic Health Evaluation II (APACHE II) has been observed to have a better discrimination in predicting mortality in primary pontine hemorrhage. Further, physiological parameters of APACHE II have been associated with outcome in ICH. This study is the first to observe a direct comparison between APACHE II and ICH scores in predicting 30-day mortality in spontaneous intracerebral hemorrhage (SICH). Materials and Methods: This study was a prospective observational study where we compared the receiver operating characteristic (ROCs) of baseline ICH and APACHE II scores in patients with SICH for predicting 30-day mortality outcome. Results: We observed that both APACHE II and ICH scores were good for predicting 30-day mortality with both having an area under the ROC curve of more than .8 (.831 [95% confidence interval {CI}, .740-.922; P < .001] and .892 [95% CI, .757-.932; P < .001], respectively). However, the ICH score was better discriminative (area under the curve AUC, .892 versus .831; P = .040) and better calibrated (P = .037 versus P = .089, Hosmer–Lemeshow goodness-of-fit test for logistic regression) for the same. Both APACHE II and ICH scores had a sensitivity of 87% at cutoff values of 19 and 3, respectively; however, the ICH score had a better specificity (90% versus 76.5%). Conclusion: The ICH score was observed to have a better discrimination and calibration for predicting 30-day mortality in SICH. Key Words: ICH score—APACHE II score—intracerebral hemorrhage—receiver operating characteristic—mortality. © 2017 National Stroke Association. Published by Elsevier Inc. All rights reserved.
From the *Department of Neurology, Institute of Post Graduate Medical Education and Research (IPGMER), Kolkata, India; †Department of Neurology, Kakatiya Medical College and Mahatma Gandhi Memorial Hospital, Warangal, India; and ‡Department of General Medicine, Calcutta National Medical College and Hospital, Kolkata, India. Received March 4, 2017; revision received May 10, 2017; accepted June 1, 2017. Author contributions: Dr. Koushik Pan contributed in the literature search, study design, data collection, data analysis, and data interpretation. Dr. Ajay Panwar contributed in the data analysis, data interpretation, literature search, figures, writing of the manuscript, and study design. Dr. Ujjawal Roy contributed in the data analysis, literature search, writing of the manuscript, and study design. Dr. Bidyut K. Das contributed in the data analysis and literature search. Address correspondence to Ajay Panwar, MD, DM, Department of Neurology, Kakatiya Medical College and Mahatma Gandhi Memorial Hospital, Warangal, 506007, India. E-mail: [email protected]. 1052-3057/$ - see front matter © 2017 National Stroke Association. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2017.06.005
Journal of Stroke and Cerebrovascular Diseases, Vol. ■■, No. ■■ (■■), 2017: pp ■■–■■
1
ARTICLE IN PRESS K. PAN ET AL.
2
Introduction Scoring systems are considered vital for predicting outcomes in patients with critical illnesses. These scoring systems help in resource allocation, clinical decisions, and quality assurance of intensive care patients.1 Spontaneous intracerebral hemorrhage (SICH) is considered to be the stroke subtype with the worst prognosis. Until a decade ago, no standard early prognostic grading scale for acute hemorrhagic stroke existed. Thereafter, logistic regression was used to identify independent predictors of 30day mortality in case of acute intracerebral hemorrhage (ICH), and a risk stratification scale, named as the ICH score, was developed based on the strength of association of the predictors.2 Subsequently, several studies have proved the ICH score as a reliable and accurate predictor of 30-day mortality in SICH.3-5 On the other hand, the Acute Physiology and Chronic Health Evaluation II (APACHE II) score is a simple and probably the most widely used score to assess the severity of illness in patients admitted to the intensive care unit.6,7 The APACHE II score has been reported to be useful in predicting the outcomes in patients with intracranial infections, acute stroke, and traumatic brain injury, besides being useful in non-neurological conditions as well. In consideration of the fact that the APACHE II score is based on physiological parameters, whereas the ICH score is calculated primarily on the basis of volume, origin, and extent of ICH, both scores may have a bearing on the outcome of patients admitted with SICH. Niskanen et al studied 2 models of APACHE II score in patients with head injury and nontraumatic ICH and reported that “model A,” based on physiological parameters, performed better for mortality prediction in comparison with “model B” based on the Glasgow Coma Scale (GCS) alone. Huang et al compared the ICH score with the APACHE II sore in predicting 30-day mortality in primary pontine hemorrhage (PPH) and found APACHE II to be more discriminative.8 In another study, the APACHE II score was observed to be more discriminative than the simplified acute physiology score II, GCS, and the National Institutes of Health Stroke Scale (NIHSS) scores for mortality prediction in hemorrhagic stroke (ICH).9 There is no published data on the comparison of APACHE II and ICH scores in ICH. Based on the promising results of the above-mentioned studies, we aimed to compare APACHE II and ICH scores for predicting 30-day mortality outcome in SICH.
Materials and Methods The present study was a hospital-based prospective observational comparative study that started on February 2015 and continued up to July 2016. The study protocol was approved by the institutional ethical committee. The study population comprised patients with SICH admitted in the critical care unit and neurology ward of the
Institute of Post Graduate Medical Education and Research at Kolkata. ICH was diagnosed on the basis of a computerized tomography (CT) scan of the brain.
Study Group Patients aged 18 years or older who were admitted with symptoms of acute stroke and were subsequently diagnosed to have SICH were enrolled in the study after a written informed consent.
Exclusion Criteria All the patients having ICH secondary to a brain tumor, trauma, or vascular malformation were excluded from the study. Those having hemorrhagic transformation of a cerebral infarct also met the exclusion criteria. Further, patients with a past history of stroke and those having other comorbid fatal conditions like end-stage renal disease, ischemic heart disease, and sepsis were excluded from the study. The study group comprised only SICH patients undergoing medical management, and any patient who underwent surgery was excluded from the study.
Methodology and Patient Evaluation All patients with a suspected diagnosis of acute stroke were subjected to an urgent noncontrast CT scan of the brain within 1 hour of admission. Soon after, those diagnosed as SICH were enrolled and physically examined along with an evaluation for APACHE II and ICH scores. The scores used for 30-day mortality prediction were based on the first evaluation after the enrollment. The data of the study subjects were collected by oral questionnaire regarding age, sex, education level, and risk factors for SICH (hypertension, diabetes, alcoholism, smoking, and anticoagulant and antiplatelet medications). Pulse, blood pressure (BP), respiratory rate, temperature, and GCS score were noted on admission. The patients were classified into 3 groups based on the GCS score: (1) GCS 3-4, (2) GCS 5-11, and (3) GCS 12-15.10
Treatment Protocol All enrolled patients were initially assessed and treated for airway patency, breathing sufficiency, BP, and signs of raised intracranial pressure. Emergent measures for lowering intracranial pressure (head end elevation to 30°, intravenous 20% mannitol infusion at a dose of .25-1.0 g/kg, and hyperventilation to achieve a pCO2 of 30-35 mm Hg) were instituted in all those cases that presented with a comatose or stuporous state of consciousness and those having signs of brainstem herniation. Early intensive BP reduction was done in all patients with elevated systolic blood pressures (SBPs) to achieve a target of SBP of less than 140 mm Hg.11-15 BP monitoring was done every 15 minutes for an initial 2 hours and then on an hourly
ARTICLE IN PRESS COMPARISON OF ICH AND APACHE II SCORES FOR MORTALITY PREDICTION
basis for the next 24 hours. The mean systolic blood pressure (mSBP) of the initial 24 hours was used for statistical correlation and regression analysis. Intravenous labetalol was the primary antihypertensive agent used for intensive BP lowering as it was locally and easily available. Labetalol was used as an initial bolus dose of 20 mg followed by boluses of 20-80 mg and as an infusion in some cases, beginning at 1-2 mg/min titrated up to the desired SBP. Labetalol was used up to the maximum cumulative dose of 300 mg over 24 hours.16 Patients with a coagulation factor deficiency or thrombocytopenia were treated with appropriate factor replacements or platelet transfusions. The patients who were on vitamin K antagonist therapy had their vitamin K antagonist withheld along with fresh frozen plasma and intravenous vitamin K therapy aimed at a rapid correction of the international normalized ratio. Patients with seizures and those with a change in mental status, along with electrographic seizures, were treated with antiepileptic medicines. The water swallow test was done as a screening procedure for dysphagia before initiating oral intake in patients. Surgical Treatment Emergent hematoma evacuation was done in cerebellar hemorrhage patients who were deteriorating neurologically or those who were having brainstem compression and hydrocephalus from ventricular obstruction.17 All the cases with cerebellar hemorrhage were assessed by the neurosurgeon, soon after being diagnosed. Those who were stuporous or comatose and those who had signs of raised intracranial pressure, along with evidence of brainstem compression or hydrocephalus on brain CT, were operated on an emergent basis. Further, those who were clinically stable initially but developed a subsequent neurological deterioration with a decline of more than 2 points in the GCS score were made to undergo an urgent repeat CT scan of the brain and, if observed to have a worsening mass effect or hydrocephalus, were scheduled for immediate surgery. Those patients who underwent surgery were thereafter excluded from the study.
Outcome Assessment and Follow-Up Primary outcome was defined as mortality assessment at 30 days after SICH. Patients who survived and were discharged were followed up in the outpatient department and over telephone regarding their final outcome.
Statistical Analysis The data entry and analyses were done on software statistical package SPSS version 16.0 (IBM, Inc, Chicago, IL, USA). Data were summarized as mean ± standard deviation for numerical variables and counts and percentages
3
for categorical variables. Chi-square test was used for the assessment of statistical significance of demographic parameters. A P value less than .05 was considered significant. Discriminative power, cutoff values, sensitivity, and specificity for APACHE II and ICH scores were determined from the receiver operating characteristic (ROC) curve. Multivariate logistic regression analysis was used to determine the strength of association with 30-day mortality for the components of the APACHE II and ICH scores.
Results Baseline Characteristics A total of 109 patients were initially enrolled, out of which 5 patients underwent surgery and were thus excluded from the study. So, the final study group comprised 104 patients (Fig 1). The mean age of the study subjects was 61 years and the study group was composed mostly of men (66.3%) patients. Most of the study subjects were smokers, hypertensives, and residents of rural areas, and were in the lower income and lower education categories. Almost 67.30% of the study population died, and most of the deaths occurred among rural study populations who were smokers and belonged to low income and low education categories (P < .001). The frequencies of intraventricular hemorrhage (IVH) and hemorrhage with a volume of 30 cc or higher were significantly higher among the subjects who died (P < .001). Those who survived had a lower mSBP (136.6 ± 3.7 versus 140.4 ± 3.4) than those who died, the difference being statistically significant on an “independent t-test” (P < .001) (Table 1). The mean volume of ICH was 33.23. The mean APACHE II and ICH scores for the study population were observed to be 22.03 and 2.81, respectively. The majority of the hemorrhagic strokes occurred in the fifth to seventh decades. Among the individual components of the ICH score, a hemorrhage volume of 30 cc or higher was observed in 78% of the cases (81 out of 104), whereas 57% (59 out of 104) had a IVH. Supratentorial bleed was more common, being present in 88% of the individuals (92 out of 104).
Association between GCS Score and 30-Day Mortality The numbers of patients in the 3 GCS based groups were 33.6% (GCS 3-4), 67.7% (GCS 5-11), and 8.7% (GCS 12-15). The mortality rates at 30 days in the 3 groups were 88.6%, 61.7%, and 22.2%, respectively. This difference in the mortality rates was statistically significant (P < .001). Thus, mortality was significantly associated with a low GCS score on admission.
Multivariate Logistic Regression Analysis of the Predictive Mortality of Components of the APACHE II and ICH Scores Among the components of the APACHE II score, the GCS value was observed to be a significant independent
ARTICLE IN PRESS K. PAN ET AL.
4
Figure 1. Flow diagram showing the study design. Abbreviations: GCS, Glasgow Coma Scale; ICH, intracerebral hemorrhage; SBP, systolic blood pressure; SICH, spontaneous intracerebral hemorrhage.
predictor of mortality (P < .001). None of other parameters had a statistically significant association (Table 2). Concerning the components of the ICH score, the GCS score, the hematoma volume (≥30 cc), and the presence of IVH were significant independent predictors of mortality (P = .016, P = .026, and P = .004, respectively). The infratentorial origin of hemorrhage and age of 80 years or older were not observed to have a significant association with the 30-day mortality outcome; however, this
observation was limited by a very small number of patients in the respective categories (Table 3).
SBP as a Predictor of 30-Day Mortality and Its Association with the Significant Demographic Variables Among variables other than the components of the APACHE II and ICH scores, logistic regression analysis
Table 1. Association of 30-day mortality with demographic variables and risk factors
Sex Residence Education Diabetes Hypertension Smoking Hemorrhage volume (≥30 cc) Intraventricular hemorrhage Infratentorial origin mSBP (±SD)
Total no. = 104
Survival (N = 34)
Death (N = 70)
P Value
Male: 69 (66.3) Femal:e 35 (33.7) Rural: 56 (53.8) Urban: 48 (46.2) Lower: 67 (64.4) (
23 (33.3) 11 (31.4) 7 (12.5) 27 (56.25) 5 (7.46) 29 (78.38) 14 (35.9) 24 (31.2) 5 (7.1) 15 (18.5) 10 (16.9) 1 (8.3) 136.6 (±3.7)
46 (66.7) 24 (68.6) 49 (87.5) 21 (43.75) 62 (92.54) 8 (21.62) 25 (64.1) 53 (68.8) 65 (92.9) 66 (81.4) 49 (83.1) 11 (91.7) 140.4 (±3.4)
.84
Abbreviations: mSBP, mean systolic blood pressure; SD, standard deviation. Numbers in parentheses represent percentages. Boldfaced values indicate significant difference.
<.001 <.001 .91 .93 <.001 <.001 <.001 .09 <.001
ARTICLE IN PRESS COMPARISON OF ICH AND APACHE II SCORES FOR MORTALITY PREDICTION
Table 2. Multivariate logistic regression analysis of predictive mortality of components of the Acute Physiology and Chronic Health Evaluation II score
Components
P Value
Odds ratio
95% Confidence interval
Age Mean arterial pressure White blood cell PaO2 Potassium Heart rate Potential of hydrogen Creatinine Glasgow Coma Scale Temperature Respiratory rate Sodium Hematocrit
.647 .225 .104 .177 .095 .081 .523 .567 <.001 .671 .684 .242 .314
.988 1.017 1.000 .980 .659 .966 .075 1.179 .661 .802 .976 .965 1.046
.940-1.039 .990-1.045 .975-1.063 .953-1.009 .403-1.076 .930-1.004 .000-2.154 .672-2.069 .533-.820 .290-2.218 .870-1.096 .910-1.024 .958-1.141
The boldfaced value indicates significant difference.
also revealed mSBP as a strong independent predictor of 30-day mortality (P < .001, odds ratio 1.455 [1.164-1.817]). Further, on investigating the significant association of poor and low educated rural population with mortality as revealed on descriptive analysis, we performed a logistic regression analysis keeping “residence” as a response variable and logistically observed significant predictors of the 30-day mortality as independent variables. Consequently, we observed that rural residence was strongly associated with a higher achieved mSBP but none of the other predictors. “Lower education,” “lower income,” and “smoking” had a significant correlation with “rural residence” (Pearson correlation, P < .001) and hence were not entered in the logistic regression model. Thus, the higher mortality rate in the rural subgroup could be explained by its direct association with the inferior control of SBP (Supplementary Tables S1, S2).
5
Table 4. Receiver operating characteristics of APACHE II and ICH scores
Cutoff value AUC (95% CI) P Value Sensitivity (%) Specificity (%)
APACHE II
ICH
19 .831 (.740-.922) <.001 87 76.5
3 .892 (.757-.932) <.001 87 90
Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; AUC, area under the curve; CI, confidence interval; ICH, intracerebral hemorrhage.
ROC Curve and Calibration of APACHE II and ICH Scores The areas under ROC curves for the APACHE II and ICH scores were .831 (95% confidence interval, .740.922; P < .001) and .892 (95% confidence interval, .757.932; P < .001), respectively (Table 4). Thus, both scores having an area under the curve higher than .8 appeared good for 30-day mortality risk prediction; still, the discriminative power of the ICH score was higher (area under the curve, .892 versus .831; P = .040) Both ROCs suggested a sensitivity of 87% at cutoff values of 19 and 3, respectively, for the APACHE II and ICH scores. However, at the same cutoff values, the APACHE II and ICH scores had a specificity of 76.5% and 90%, respectively (Fig 2). So, the APACHE II and ICH scores both
Table 3. Multivariate logistic regression analysis of predictive mortality of components of the intracerebral hemorrhage score Odds 95% Confidence ratio interval
Components
P Value
Glasgow Coma Scale Hemorrhage volume (≥30 cc) Intraventricular hemorrhage Infratentorial origin Age
.016 .026
.781 6.750
.639-.955 1.250-36.455
.004
6.827
1.855-25.123
.081 .987
6.448 1.024
.797-52.156 .055-18.998
The boldfaced value indicates significant difference.
Figure 2. ROC curves for comparison of ICH and APACHE II scores in predicting 30-day mortality outcome. Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; ICH, intracerebral hemorrhage; ROC, receiver operating characteristic.
ARTICLE IN PRESS K. PAN ET AL.
6
had a similarly good sensitivity, albeit the ICH score was observed to be more specific. Further, the ICH score was observed to have a better calibration than the APACHE II (P = .037 versus P = .089) in the Hosmer–Lemeshow goodness-of-fit test for logistic regression analysis. Thus, the ICH score performed better in terms of discrimination as well as calibration.
Discussion The descriptive part of our data analysis revealed that the survival rate in ICH was better in urban individuals and in the higher education group. Earlier, Das et al. reported a better knowledge of stroke symptoms and risk factors among urban residents who were highly educated and those falling in the high-income category.18 However, considering that a better knowledge of risk factors may be associated with hemorrhagic stroke prevention but should not have a direct relationship with 30-day prognostic outcome after onset, we explored the same on logistic regression analysis. Consequently, residence, which was strongly correlated with income and educational status, had a strong association with mSBP, which, in turn, was observed to be an independent predictor of 30-day mortality. Urban residents had a better controlled mSBP, probably due to good affordability and prompt purchase of medications, and hence a lower 30day mortality. However, in our opinion, this is a preliminary observation that needs to be tested in further studies conducted in similar demographic conditions. We also observed a strong association between low GCS score and 30-day mortality, with death rate being highest in the patient group having a GCS score of 3-4. These results are in agreement with the earlier observations by Muengtaweepongsa and Seamhan, who observed a significantly high mortality rate in ICH among Thai patients having GCS scores of 3-4.10 On multivariate analysis of the predictive mortality of the components of the APACHE II score, only the GCS score was observed to be an independent predictor of 30-day mortality. In a previous study, Niskanen et al used logistic regression to study the association of various components of the APACHE II score with hospital mortality in patients with head injury or nontraumatic intracranial hemorrhage. Niskanen et al created 2 models from the APACHE II score and observed that model A, based on physiological parameters, was better than model B, based on GCS alone, to estimate the probability of death. It was thus suggested that the physiological parameters of APACHE II scores could contribute toward estimating the probability of death in neurological cases. In the same study, the GCS score, along with the mean arterial pressure and age, was observed to be the independent predictor of hospital death.19 In the present study, we did not observe any of the physiological parameters except GCS as the independent predictors of mortality. However,
our study group was different, consisting of SICH patients only, which could explain the disparity among the mortality predictors, in comparison with Niskanen et al’s study. In another study, Huang et al had shown that GCS alone had an excellent discrimination to predict 30-day mortality in PPH.8 Multivariate analysis of the components of ICH score in our study subjects revealed that GCS, hematoma volume (≥30 cc), and the presence of IVH were strong predictors of 30-day mortality. However, the infratentorial origin of hemorrhage, despite having high odds for death, was not significantly associated with the mortality prediction, which is in contradiction with the previous studies.2,10 This discordance could possibly be due to very small number of patients having infratentorial hemorrhage (12% infratentorial versus 88% supratentorial), which could not reach statistical significance. Age 80 years or older was an independent predictor of mortality in the original study of the ICH score.2 The lack of association between age of 80 years or older and mortality in the present study may be due to the small number of patients older than 80 years. On ROC analysis for mortality prediction, we observed that both the APACHE II and ICH scores had a good sensitivity of 87% at cutoff scores of 19 and 3, respectively. However, at the same sensitivity, the ICH score was observed to be far more specific (90%) than the APACHE II score (76.5%). The ICH score’s cutoff of 3 or higher for mortality prediction is in concordance with a study done by Cheung and Zou, where they compared 3 types of ICH scores (original, modified, and new) and observed a high mortality rate and a 0 rate of good outcome with a cutoff score of 3 or higher in all the 3 ICH scores. Among the 3 ICH scores, Cheung and Zou observed the “original ICH score” to be most specific, with a sensitivity of 78.6% and a specificity of 90.4%.20 Di Napoli et al also reported the original ICH score to have a good specificity of 84% with a sensitivity of 75% for predicting 30-day mortality outcome.21 Thus, our observations (87% sensitivity and 90% specificity) reinforce the already reported high specificity of ICH score in predicting 30-day mortality outcome albeit with a higher sensitivity. Unlike the ICH score, there are limited data on the use of the APACHE II as a prognostic score in ICH. Huang et al observed the APACHE II score to have a sensitivity and a specificity of 91% and 86.5% respectively, at a cutoff score of 16.5 in predicting 30-day mortality in PPH.8 However, we also observed a nearly high sensitivity of 87%, but with a lower specificity of 76.5%. Further, Huang et al also observed that the APACHE II score had a higher discriminative power as compared with the ICH score for predicting 30-day mortality in PPH. However, in our study, we observed the ICH score to have a higher discriminative power and a better calibration than the APACHE II score. The differences in results in our study could be due to a different study cohort as we enrolled all patients with ICH and not just PPH.
ARTICLE IN PRESS COMPARISON OF ICH AND APACHE II SCORES FOR MORTALITY PREDICTION
The present study had a few limitations. We probably could not estimate the true strength of association between “infratentorial origin of hemorrhage” and “age ≥ 80 years” while analyzing the predictive mortality of components of ICH score because of the small number of subjects in the respective categories. Further, ours was a single-center study and the results of the same need to be confirmed by a multicentric study with a larger sample size. In conclusion, we observed that both APACHE II and ICH scores on presentation can be used to have an approximate prediction of 30-day mortality in ICH. However, the ICH score has a better discriminative power and is better calibrated than the APACHE II score for the same.
9.
10.
11.
12.
13.
Appendix: Supplementary Material Supplementary data to this article can be found online at doi:10.1016/j.jstrokecerebrovasdis.2017.06.005.
References 1. Rapoport J, Teres D, Lemeshow S, et al. A method for assessing the clinical performance and cost-effectiveness of intensive care units: a multicenter inception cohort study. Crit Care Med 1994;22:1385-1391. 2. Hemphill JC, Bonovich DC, Besmertis L, et al. The ICH score. a simple, reliable grading scale for intracerebral hemorrhage. Stroke 2001;32:891-897. 3. Godoy DA, Piñero G, Di Napoli M. Predicting mortality in spontaneous intracerebral hemorrhage. Can modification to original score improve the prediction? Stroke 2006;37:1038-1044. 4. Aysenne AM, Albright KC, Mathias T, et al. 24-Hour ICH score is a better predictor of outcome than admission ICH score. ISRN Stroke 2013;2013:4. 5. Meyer DM, Begtrup K, Grotta JC, et al. Is the ICH score a valid predictor of mortality in intracerebral hemorrhage? J Am Assoc Nurse Pract 2015;27:351-355. 6. Vincent J-L, Moreno R. Clinical review: scoring systems in the critically ill. Crit Care 2010;14:207. 7. Bouch DC, Thompson JP. Severity scoring systems in the critically ill. CEACCP 2008;8:181-185. 8. Huang KB, Ji Z, Wu YM, et al. The prediction of 30-day mortality in patients with primary pontine hemorrhage:
14.
15.
16. 17.
18.
19.
20.
21.
7
a scoring system comparison. Eur J Neurol 2012;19:12451250. Moon BH, Park SK, Jang DK, et al. Use of APACHE II and SAPS II to predict mortality for hemorrhagic and ischemic stroke patients. J Clin Neurosci 2015;22:111-115. Muengtaweepongsa S, Seamhan B. Predicting mortality rate with ICH score in Thai intracerebral hemorrhage patients. Neurol Asia 2013;18:131-135. Sakamoto Y, Koga M, Yamagami H, et al. Systolic blood pressure after intravenous antihypertensive treatment and clinical outcomes in hyperacute intracerebral hemorrhage. The stroke acute management with urgent risk-factor assessment and improvement-intracerebral hemorrhage study. Stroke 2013;44:1846-1851. Butcher KS, Jeerakathil T, Hill M, et al. The intracerebral hemorrhage acutely decreasing arterial pressure trial. Stroke 2013;44:620-626. Arima H, Huang Y, Wang JG, et al. Earlier blood pressure-lowering and greater attenuation of hematoma growth in acute intracerebral hemorrhage: INTERACT pilot phase. Stroke 2012;43:2236-2238. Arima H, Anderson CS, Wang JG, et al. Lower treatment blood pressure is associated with greatest reduction in hematoma growth after acute intracerebral hemorrhage. Hypertension 2010;56:852-858. Anderson CS, Heeley E, Huang Y, et al. Rapid bloodpressure lowering in patients with acute intracerebral hemorrhage. N Engl J Med 2013;368:2355-2365. Varon J. The diagnosis and treatment of hypertensive crises. Postgrad Med 2009;121:5-13. Morgenstern LB, Hemphill JC 3rd, Anderson C, et al. Guidelines for the management of spontaneous intracerebral hemorrhage: a guideline for healthcare professionals from the American Heart Association/ American Stroke Association. Stroke 2010;41:2108-2129. Das BK, Karmakar PS, Santra G, et al. Awareness of stroke among elderly public in Eastern India. Neurol Asia 2011;16:119-126. Niskanen MM, Kari A, Hernesniemi JA, et al. Contribution of non-neurologic disturbances in acute physiology to the prediction of intensive care outcome after head injury or non-traumatic intracranial haemorrhage. Intensive Care Med 1994;20:562-566. Cheung RTF, Zou L-Y. Use of the original, modified, or new intracerebral hemorrhage score to predict mortality and morbidity after intracerebral hemorrhage. Stroke 2003;34:1717-1722. Di Napoli M, Godoy DA. Clinical grading scales in spontaneous intracerebral hemorrhage. Stroke 2007; 38:e133-e135.