Incorporating a Modified Graeb Score to the Modified Fisher Scale for Improved Risk Prediction of Delayed Cerebral Ischemia Following Aneurysmal Subarachnoid Hemorrhage

Incorporating a Modified Graeb Score to the Modified Fisher Scale for Improved Risk Prediction of... Abstract BACKGROUND Delayed cerebral ischemia (DCI) is a cause of poor outcome after aneurysmal subarachnoid hemorrhage. Risk scales to predict DCI have scarcely been evaluated for predictive accuracy. Accounting for volume of intraventricular hemorrhage (IVH) in the modified Fischer scale (mFS) may improve its predictive accuracy. OBJECTIVE To compare the modified Graeb score (mGS) to the mFS for risk prediction of DCI, and to investigate whether incorporating an mGS cut-point into the mFS could improve predictive accuracy. METHODS This retrospective analysis was based on the Clazosentan to Overcome Neurological Ischemia and Infarction Occurring after Subarachnoid Hemorrhage (CONSCIOUS-1) trial cohort. IVH volume was quantified with the mGS. The relation of the mGS to DCI was evaluated using logistic regression and the area under the receiver operator characteristics curve (AUC). An optimized mGS cut-point was identified using the Youden index, and was incorporated into the mFS to dichotomize grades 2 and 4. The AUC was used to compare the predictive performance of the mGS with that of the mFS, and to assess whether there was an improvement in DCI prediction after creation of the dichotomized scale. RESULTS The mFS and the mGS had similar discrimination for DCI (AUC: 0.608 vs 0.618; P = .79). A new scale including both the mFS and mGS significantly improved the AUC compared to the mFS (AUC: 0.647 vs 0.608; P = .022). CONCLUSION The mFS and the mGS have similar prognostic utility. Accounting for IVH volume improved prediction of DCI by the mFS. Delayed cerebral ischemia, Vasospasm, Subarachnoid hemorrhage, Intraventricular hemorrhage, Fisher scale, Graeb score ABBREVIATIONS ABBREVIATIONS aSAH aneurysmal subarachnoid hemorrhage AUC area under the receiver operator characteristics curve CI confidence interval CONSCIOUS-1 Clazosentan to Overcome Neurological Ischemia and Infarction Occurring after Subarachnoid Hemorrhage DCI delayed cerebral ischemial FS Fisher scale GS Graeb Score ICC interclass correlation coefficient IVH intraventricular hemorrhage mFS modified Fischer scale mGS modified Graeb score Delayed cerebral ischemia (DCI) has been reported in 20% to 30% of patients who suffer aneurysmal subarachnoid hemorrhage (aSAH).1,2 It is one of the strongest predictors of poor outcome following aSAH.3 Furthermore, DCI is also associated with substantially higher costs to the healthcare system, and slower return to employment in affected patients.4 The risk of DCI may be quantified using a number of grading scales that are based primarily on neuroimaging;5-17 however, the properties of these scales have scarcely been studied nor compared. Fisher et al18 created a scale that has, historically, been widely used. Although the Fisher scale (FS) was among the first to demonstrate that patients with thick cisternal clots are at highest risk of developing DCI,7,19 it did not adequately address patients who may have thick cisternal clots with concomitant intraventricular hemorrhage (IVH).20 IVH has been found to be associated with poor outcome in patients who suffer aSAH.21 Some studies have found that IVH is an independent predictor of DCI,9,22 and that the risk of developing DCI may increase with greater volumes of IVH.5,14,23 Graeb et al24 proposed a semiquantitative measure of IVH,24 and this scale (GS) has been shown to correlate with poor outcomes following aSAH,25 yet little is known regarding its value in predicting DCI. In order to improve predictive performance of the FS, Frontera et al9 modified it to account for patients with both thick cisternal clots and IVH.9 This modification of the FS (mFS) is a better predictor of DCI than the original FS;9,26 however, the authors did not assess whether prediction could be further improved by accounting for IVH volume. The objective of this study was to investigate whether a modification of the GS (mGS) is valuable for DCI risk assessment in patients who suffer aSAH. The study also assessed whether altering the mFS by incorporating a measure of IVH volume, such as the mGS, would improve DCI risk prediction. Given the results of previous work,3,5,14,23 we hypothesized that this modification would improve the ability of the mFS to predict patients who will suffer DCI. In addition, it was hypothesized that data correlating IVH with DCI would support several different pathophysiological mechanisms that may cause DCI. METHODS The methodology for this post hoc analysis was designed to comply with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. Patient Population The study was based on the 413 patients who were enrolled in the Clazosentan to Overcome Neurological ischemia and Infarction Occurring after Subarachnoid Hemorrhage (CONSCIOUS-1) study (clinical trial registration no.: NCT00111085 [clinicaltrials.gov]) between January 2005 and March 2006. The methods and results of this trial have been published elsewhere.27 The institutional review board at St. Michael's Hospital granted approval for this study. As this was a post hoc analysis, renewed patient consent was not required for study approval. Variables All patients had baseline cranial CT scans that were analyzed by 2 independent, blinded reviewers. IVH was quantified using the mGS, which assigned a value of 0 (no blood), 1 (less than 25% filled), 2 (moderately filled), or 3 (completely filled) to each ventricle for a total score out of 12.28,29 For purposes of the analysis, the average of the reviewers’ scores was obtained. The mFS was computed for each patient according to Frontera et al.9 These scores were based on the reviewers’ interpretation of clot thickness, and whether they identified any IVH in their mGS. Where disagreement occurred between the 2 reviewers regarding clot thickness, an independent arbitrator made the final decision, which was considered for the analysis. Three patients did not have their clot thickness reported, and they were excluded from any analysis that related to the mFS. The primary outcome variable was the development of DCI, which was defined as angiographic vasospasm associated with neurological worsening (a decrease of at least 2 points on the modified Glasgow Coma Scale, or an increase of at least 2 points on the National Institutes of Health Stroke Scale) that lasted at least 2 h. When patients were not able to undergo a neurological examination, DCI was defined as angiographic vasospasm associated with clinical signs of DCI (unexplained fever or new neurological deficit), or a new infarction on a postprocedure CT scan not attributable to any other cause.27 Statistics Baseline characteristics of the cohort were compared according to DCI status using frequency tables and bar charts. Interobserver reliability for the mGS was calculated using the interclass correlation coefficient (ICC). We used binary logistic regression models to assess the relation of the mFS and the mGS to DCI. In order to assess for potential confounding between clot thickness and IVH volume, a multivariate regression model was created with clot thickness and mGS as predictors of DCI. The 2 scales were compared for predictive accuracy by calculating the area under the receiver operator characteristics curve (AUC). The Youden index was then used to identify an optimal cut-point for the mGS, in terms of maximal sensitivity and specificity, for predicting DCI. A new scale was created, which dichotomized grades 2 and 4 of the mFS based on the optimal mGS cut-point. This new scale was then compared to the mFS to examine if there was any increase in risk prediction of DCI. Improvement in prediction was evaluated by comparing the AUC of the 3 models. Significance level was set at P < .05. Statistical analysis was carried out using Stata 13.1 (Statacorp, College Station, Texas). RESULTS Of the 413 patients in the CONSCIOUS-1 dataset, 78 (19%) developed DCI. Patients who developed DCI tended to present with higher World Federation of Neurosurgical Societies grades, were more likely to have their aneurysm clipped, have an external ventricular drain placed, have thick cisternal clots, and tended to have greater IVH volume as assessed by the mGS. Baseline characteristics of the cohort are shown in Table 1. Median mGS was 3 with interquartile range of 1 to 4.5. A 2-way, mixed effects ICC analysis of the mGS demonstrated excellent interobserver agreement (ICC 0.84; 95% confidence interval [CI] 0.81-0.86). Both the mFS and the mGS were significant predictors of DCI on univariate regression (Table 2), and the significance of the mGS was maintained when combined into a multivariate regression with clot thickness to account for potential confounding (odds ratio 1.13; 95% CI 1.02-1.25). The optimal mGS cut-point, as determined by the Youden index, was 3.25. Because the mGS is a discrete scale, we used a score of greater than 3 as our cut-off value. For patients with an mGS greater than 3, logistic regression demonstrated no significant difference in the risk of DCI for patients with a single ventricle score of 3 (P = .63), with IVH in all 4 ventricles (P = .85), with bilateral IVH (P = .89), or with 2 ventricles having mGSs of 2 to 3 (P = .89). The mFS was related to DCI in a gradient manner (Figure 1). Incorporating the optimized mGS cut-point altered the risk of DCI compared to the mFS (Figure 2). FIGURE 1. View largeDownload slide Bar chart showing the graded relation between mFS and DCI. FIGURE 1. View largeDownload slide Bar chart showing the graded relation between mFS and DCI. FIGURE 2. View largeDownload slide Bar chart showing the relation between the dichotomized scale and DCI. FIGURE 2. View largeDownload slide Bar chart showing the relation between the dichotomized scale and DCI. TABLE 1. Baseline Characteristics of the Study Cohort Compared According to DCI Status   Total  DCI  No DCI  N  413  78  335  Age (median and IQR)  51 (44, 60)  51 (43, 60)  51 (44, 59)  Female sex  292 (70.1%)  60 (76.9%)  232 (69.3%)  Race     White  368 (89.1%)  73 (93.6%)  295 (88.1%)   Black  27 (6.5%)  4 black (5.1%)  23 (6.9%)   Other  18 (4.4%)  1 other (1.3%)  17 (5.1%)  WFNS grade     1  181 (43.8%)  20 (25.6%)  161 (48.1%)   2  117 (28.3%)  22 (28.2%)  95 (28.4%)   3  12 (2.9%)  2 (2.6%)  10 (3.0%)   4  97 (23.5%)  33 (42.3%)  64 (19.1%)   5  3 (0.7%)  1 (1.3%)  2 (0.6%)  History of hypertension  175 (42.4%)  39 (50.0%)  136 (40.1%)  Nicotine use  217 (52.5%)  38 (48.7%)  179 (53.4%)  Surgical clipping  186 (45.0%)  43 (55.1%)  143 (42.7%)  EVD placement  149 (36.1%)  48 (61.5%)  101 (30.1%)  Shunt placement  37 (9.0%)  9 (11.5%)  28 (8.4%)  Hydrocephalus  392 (94.9%)  77 (98.7%)  315 (94.0%)  Thick cisternal clot  302 (73.1%)  69 (88.5%)  233 (69.6%)  Modified Graeb score (median and IQR)  3 (1, 4.5)  4 (2, 5)  2.5 (1, 4)  Modified Fisher grade (N = 410)     1  22 (5.4%)  1 (1.3%)  21 (6.3%)   2  86 (21.0%)  8 (10.3%)  78 (23.5%)   3  42 (10.2%)  7 (9.0%)  35 (10.5%)   4  260 (63.4%)  62 (79.5%)  198 (59.6%)  Unfavorable outcome (GOSE 1-4)  125 (30.3%)  46 (60.0%)  79 (23.6%)    Total  DCI  No DCI  N  413  78  335  Age (median and IQR)  51 (44, 60)  51 (43, 60)  51 (44, 59)  Female sex  292 (70.1%)  60 (76.9%)  232 (69.3%)  Race     White  368 (89.1%)  73 (93.6%)  295 (88.1%)   Black  27 (6.5%)  4 black (5.1%)  23 (6.9%)   Other  18 (4.4%)  1 other (1.3%)  17 (5.1%)  WFNS grade     1  181 (43.8%)  20 (25.6%)  161 (48.1%)   2  117 (28.3%)  22 (28.2%)  95 (28.4%)   3  12 (2.9%)  2 (2.6%)  10 (3.0%)   4  97 (23.5%)  33 (42.3%)  64 (19.1%)   5  3 (0.7%)  1 (1.3%)  2 (0.6%)  History of hypertension  175 (42.4%)  39 (50.0%)  136 (40.1%)  Nicotine use  217 (52.5%)  38 (48.7%)  179 (53.4%)  Surgical clipping  186 (45.0%)  43 (55.1%)  143 (42.7%)  EVD placement  149 (36.1%)  48 (61.5%)  101 (30.1%)  Shunt placement  37 (9.0%)  9 (11.5%)  28 (8.4%)  Hydrocephalus  392 (94.9%)  77 (98.7%)  315 (94.0%)  Thick cisternal clot  302 (73.1%)  69 (88.5%)  233 (69.6%)  Modified Graeb score (median and IQR)  3 (1, 4.5)  4 (2, 5)  2.5 (1, 4)  Modified Fisher grade (N = 410)     1  22 (5.4%)  1 (1.3%)  21 (6.3%)   2  86 (21.0%)  8 (10.3%)  78 (23.5%)   3  42 (10.2%)  7 (9.0%)  35 (10.5%)   4  260 (63.4%)  62 (79.5%)  198 (59.6%)  Unfavorable outcome (GOSE 1-4)  125 (30.3%)  46 (60.0%)  79 (23.6%)  DCI, delayed cerebral ischemia; IQR, interquartile range; WFNS, World Federation of Neurological Societies; GOSE, Glasgow Outcome Scale Extended; EVD, external ventricular drain. View Large TABLE 2. Univariate Regression Analysis of mFS, mGS, and New Dichotomized Scale Scale  Odds ratio  95% CI  mFS  1.76  1.27 - 2.44  mGS  1.15  1.04 - 1.28  Scale  Odds ratio  95% CI  mFS  1.76  1.27 - 2.44  mGS  1.15  1.04 - 1.28  mFS, modified Fisher Scale; mGS modified Graeb Score. View Large We found that the mGS had a slightly higher AUC (0.618; 95% CI: 0.549-0.686) than the mFS (0.608; 95% CI: 0.558-0.659); nonetheless, the difference was not statistically significant (P = .79). This suggests that both scales are essentially equivalent for predicting risk of DCI after aSAH. The dichotomized scale, as described in Table 3, incorporated both the mGS cut-point and the mFS. It had a significantly higher AUC (0.647; 95% CI: 0.587-0.707; P = .022) than the mFS, indicating that accounting for IVH volume improved the predictive accuracy of the mFS in our sample. TABLE 3. Comparison of Modified Fisher Scale According to Frontera et al (2006) and New Dichotomized Scale Modified Fisher grade according to Frontera et al (2006)9  Cisternal clot thickness  IVH  New dichotomized scale grade  Cisternal clot thickness  IVH  1  Thin  Absent  1  Thin  Absent  2  Thin  Present  2a  Thin  mGS ≤ 3        2b  Thin  mGS > 3  3  Thick  Absent  3  Thick  Absent  4  Thick  Present  4a  Thick  mGS ≤ 3        4b  Thick  mGS > 3  Modified Fisher grade according to Frontera et al (2006)9  Cisternal clot thickness  IVH  New dichotomized scale grade  Cisternal clot thickness  IVH  1  Thin  Absent  1  Thin  Absent  2  Thin  Present  2a  Thin  mGS ≤ 3        2b  Thin  mGS > 3  3  Thick  Absent  3  Thick  Absent  4  Thick  Present  4a  Thick  mGS ≤ 3        4b  Thick  mGS > 3  IVH, intraventricular hemorrhage; mGS, modified Graeb score. View Large DISCUSSION Our study demonstrated that the mGS and the mFS are essentially equivalent scales for DCI risk prediction. While there was no significant difference in the predictive value of the 2 scales, Ibrahim et al30 previously investigated the interobserver variability in both the mFS and mGS. They found that there was more variability in qualitative scales (eg, mFS) than in semiquantitative scores (eg, mGS). This led them to infer that the use of quantitative scores, such as the mGS, may be more consistent when assessing the risk of DCI.30 Our analysis also found excellent interobserver agreement for the mGS. Nonetheless, our most significant finding is that accounting for IVH volume in the mFS significantly improves the risk prediction of DCI in patients who suffer aSAH. Past radiographic scales to predict DCI after aSAH have varied in their parameters and performance (Table 4).5-17 These scales have scarcely been compared for predictive accuracy. Of note, a 2008 study by Kramer et al26 used logistic regression analysis to suggest that both the mFS and the scale proposed by Claassen et al5 were superior to the original FS, but the authors did not evaluate risk discrimination using the AUC.26 TABLE 4. Scales for Radiographic Prediction of DCI Authors and year  Patients  Type of Study  Parameters  Performance  Fisher et al (1980)7  47  Prospective  Grades 1 to 4 based on clot thickness, presence/absence of IVH or ICH  Grade 3 SAH—92% sensitive and 95% specific for DCI  Mizukami et al (1980)15  177  Prospective  Presence clot 1 to 4 d after rupture  High-density clot on CT within 4 d—100% sensitive and 67% specific for DCI  Davis et al (1980)6  50  Prospective  Class 1 to 4 based on increasing extent of SAH  Class 4 SAH—55% sensitive and 75% specific for severe angiographic vasospasm  Mohsen et al (1984)16  100  Prospective  Grades 1 to 3 based on clot thickness and extent  Grade 3 SAH—71% sensitive and 64% specific for DCI  Fujita (1985)10  36  Retrospective  Grades 1 to 5 based on Hounsfield numbers  Grade 5 SAH—75% sensitive and 86% specific for severe vasospasm or death  Knuckey et al (1985)13  46  Prospective  Grades 1 to 4 increasing with more diffuse SAH  Grade 4 SAH—47% sensitive and 87% specific for DCI  Hijdra et al (1990)11  182  Prospective  Extent of SAH scored 0 to 30; extent of IVH scored 0 to 12  Not reported  Claassen et al (2001)5  276  Prospective  Grades 1 to 4 based on clot thickness and presence/absence of bilateral IVH or ICH  Grade 4 SAH—35% sensitive and 40% specific for DCI or infarction  Friedman et al (2002)8  40  Prospective  Software-based volumetric quantification  SAH volume > 20 cm—46% sensitive and 100% specific for DCI  Klimo et al (2004)12  266  Retrospective  Created Fisher 3+4 grade to include patients with thick clot and IVH (>5 mL)  Grade 3+4 SAH—50% sensitive and 57% specific for vasospasm  Frontera et al (2006)9  1355  Retrospective  Grades 1 to 4 to incorporate patients with thick clot and IVH  Not reported  Ko et al (2011)14  160  Retrospective  CIHV stratified into quintiles  Fifth quintile—35% sensitive and 35% specific for DCI  Wilson et al (2012)17  250  Retrospective  BNI grades 1 to 5 based on maximal clot thickness in 5 mm increments  BNI grade 5 SAH—15% sensitive and 50% specific for symptomatic vasospasm  New Dichotomized Scale  410  Retrospective  Grades 1 to 4b account for clot thickness and mGS cut-off of >3  Grade 4b SAH—53% sensitive and 28% specific for DCI  Authors and year  Patients  Type of Study  Parameters  Performance  Fisher et al (1980)7  47  Prospective  Grades 1 to 4 based on clot thickness, presence/absence of IVH or ICH  Grade 3 SAH—92% sensitive and 95% specific for DCI  Mizukami et al (1980)15  177  Prospective  Presence clot 1 to 4 d after rupture  High-density clot on CT within 4 d—100% sensitive and 67% specific for DCI  Davis et al (1980)6  50  Prospective  Class 1 to 4 based on increasing extent of SAH  Class 4 SAH—55% sensitive and 75% specific for severe angiographic vasospasm  Mohsen et al (1984)16  100  Prospective  Grades 1 to 3 based on clot thickness and extent  Grade 3 SAH—71% sensitive and 64% specific for DCI  Fujita (1985)10  36  Retrospective  Grades 1 to 5 based on Hounsfield numbers  Grade 5 SAH—75% sensitive and 86% specific for severe vasospasm or death  Knuckey et al (1985)13  46  Prospective  Grades 1 to 4 increasing with more diffuse SAH  Grade 4 SAH—47% sensitive and 87% specific for DCI  Hijdra et al (1990)11  182  Prospective  Extent of SAH scored 0 to 30; extent of IVH scored 0 to 12  Not reported  Claassen et al (2001)5  276  Prospective  Grades 1 to 4 based on clot thickness and presence/absence of bilateral IVH or ICH  Grade 4 SAH—35% sensitive and 40% specific for DCI or infarction  Friedman et al (2002)8  40  Prospective  Software-based volumetric quantification  SAH volume > 20 cm—46% sensitive and 100% specific for DCI  Klimo et al (2004)12  266  Retrospective  Created Fisher 3+4 grade to include patients with thick clot and IVH (>5 mL)  Grade 3+4 SAH—50% sensitive and 57% specific for vasospasm  Frontera et al (2006)9  1355  Retrospective  Grades 1 to 4 to incorporate patients with thick clot and IVH  Not reported  Ko et al (2011)14  160  Retrospective  CIHV stratified into quintiles  Fifth quintile—35% sensitive and 35% specific for DCI  Wilson et al (2012)17  250  Retrospective  BNI grades 1 to 5 based on maximal clot thickness in 5 mm increments  BNI grade 5 SAH—15% sensitive and 50% specific for symptomatic vasospasm  New Dichotomized Scale  410  Retrospective  Grades 1 to 4b account for clot thickness and mGS cut-off of >3  Grade 4b SAH—53% sensitive and 28% specific for DCI  DCI, delayed cerebral ischemia; IVH, intraventricular hemorrhage; ICH, intracerebral hemorrhage; SAH subarachnoid hemorrhage; CIHV, cisternal and intraventricular hemorrhage volume; BNI, Barrow Neurological Institute; mGS, modified Graeb Score. View Large Based on the findings of previous work suggesting a strong association between total hemorrhage volume with DCI14,31 and the negative impact of IVH on patient outcomes,3 we had anticipated that accounting for IVH volume would improve the predictive performance of the mFS. A comparison of the AUC for the mFS and the dichotomized scale validated this prediction by demonstrating improved predictive accuracy when the mGS cut-point was incorporated into the mFS. It is possible that IVH volume is correlated with the other risk items in the mFS (eg, clot thickness or localization), which may have overstated the predictive information contributable by IVH volume. A study from Wilson et al17 found that after stratifying clot thickness into 5 mm increments, IVH was no longer associated with the development of DCI, indicating that any risk increases associated with IVH may be secondary to the relationship between total hemorrhage volume and IVH.17 However, the mGS remained a significant predictor of DCI when combined with SAH clot thickness in a multivariate regression model, suggesting that the mGS was independently associated with DCI in our cohort. A number of reasons may help explain our findings and contribute to hypothesis generation regarding the pathophysiology of DCI. It is possible that a greater amount of blood in the ventricular system is more likely to trigger angiographic vasospasm, microcirculatory constriction, microthrombosis, cortical spreading ischemia, and delayed cell apoptosis, all of which have been implicated in the pathogenesis of DCI.1,7,32-35 Another possibility is that the distribution of the IVH, rather than the quantity, is the mediating factor. Claassen et al5 found that bilateral IVH was a stronger predictor of DCI than the presence or absence of IVH alone. This led them to propose a modification to the FS that accounted for bilateral IVH.5 More widespread IVH would also be represented in the mGS, as each ventricle with blood in it would contribute to the score. However, our study is the first to our knowledge that incorporated an optimized cut-point for IVH volume (as determined by the Youden index) to the mFS, and subsequently evaluated the predictive accuracy of the scales by comparing the AUC. A study examining whether bilateral IVH or the optimized mGS cut-point is a better predictor of DCI is warranted. In their review, Klimo et al18 cited ease of use as one of the main advantages of the original FS. The mGS is more complex to compute and incorporating it into the mFS may result in an unwieldy scale with only marginally improved predictive performance for DCI. More complex grading schemes such as the one proposed by Ko et al,14 which attempted to quantify cisternal plus IVH volume, have failed to gain the same widespread use as the FS, likely because of the time and expertise required to calculate them. However, of note is the drastic difference in DCI risk between a grade 2a aSAH and a grade 2b aSAH according to our dichotomized scale. The implication of this finding is that patients with a thin clot and an mGS greater than 3 should not be classified as having the same risk of DCI as patients with a thin clot and an mGS less than 3, which was the case in the mFS. Limitations The strengths of this study are that blinded reviewers conducted CT scan measurements, and the definition of DCI was pre-specified. Possible limitations in this study also deserve mention. This is a retrospective analysis based on cohorts that were not primary assembled for the purpose of the study. One implication was that we had to resort to deriving the mFS from variables indicating clot thickness, location, and presence of IVH in the CONSCIOUS-1 dataset, since the mFS was not recorded a priori. Another issue is that an exclusion criterion in the CONSCIOUS-1 trail was IVH or intracerebral hemorrhage in the absence of localized thick or diffuse aSAH, which may have over-represented patients with mFS grades 3 and 4.27 Lastly, our study may not have been sufficiently powered given the small number of patients who went on to develop DCI. CONCLUSION The objective of this study was to examine whether the mGS is a valuable tool in predicting risk of DCI in patients who suffer aSAH, and to assess whether the mFS could be improved by incorporating an optimized mGS cut-point to dichotomize grades 2 and 4. We found that for purposes of risk classification of DCI, the mGS and mFS are essentially equivalent. The mGS remained a significant predictor of DCI when combined with clot thickness in a multivariate regression model. Altering the mFS to account for IVH volume according to an optimized mGS cut-point improved predictive accuracy. Future work should look to validate these findings in a prospective study that does not rely on the use of preassembled data. Disclosures Matt Eagles received funding for this work from St. Michael's Hospital as part of the Keenan Research Summer Studentship. Dr Macdonald reports ownership of Edge Therapeutics, Inc. and receiving non-study-related clinical or research effort support from the Brain Aneurysm Foundation, Heart and Stroke Foundation of Canada, Canadian Institutes for Health Research, and Physicians Services Incorporated Foundation. The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article. A portion of the work was accepted as an electronic poster at the 2016 AANS Annual Scientific Meeting in Chicago, Illinois from May 29 to June 4, 2016. REFERENCES 1. Macdonald RL. Delayed neurological deterioration after subarachnoid haemorrhage. Nat Rev Neurol . 2014; 10( 1): 44- 58. Google Scholar CrossRef Search ADS PubMed  2. Dorhout Mees SM, Kerr RS, Rinkel GJ, Algra A, Molyneux AJ. Occurrence and impact of delayed cerebral ischemia after coiling and after clipping in the international subarachnoid aneurysm trial (ISAT). J Neurol . 2012; 259( 4): 679- 683. Google Scholar CrossRef Search ADS PubMed  3. Rosengart AJ, Schultheiss KE, Tolentino J, Macdonald RL. Prognostic factors for outcome in patients with aneurysmal subarachnoid hemorrhage. Stroke . 2007; 38( 8): 2315- 2321. Google Scholar CrossRef Search ADS PubMed  4. Rivero-Arias O, Wolstenholme J, Gray A et al.   The costs and prognostic characteristics of ischaemic neurological deficit due to subarachnoid haemorrhage in the united kingdom. evidence from the MRC international subarachnoid aneurysm trial. J Neurol . 2009; 256( 3): 364- 373. Google Scholar CrossRef Search ADS PubMed  5. Claassen J, Bernardini GL, Kreiter K et al.   Effect of cisternal and ventricular blood on risk of delayed cerebral ischemia after subarachnoid hemorrhage: The fisher scale revisited. Stroke . 2001; 32( 9): 2012- 2020. Google Scholar CrossRef Search ADS PubMed  6. Davis JM, Davis KR, Crowell RM. Subarachnoid hemorrhage secondary to ruptured intracranial aneurysm: prognostic significance of cranial CT. AJR Am J Roentgenol . 1980; 134( 4): 711- 715. Google Scholar CrossRef Search ADS PubMed  7. Fisher CM, Kistler JP, Davis JM. Relation of cerebral vasospasm to subarachnoid hemorrhage visualized by computerized tomographic scanning. Neurosurgery . 1980; 6( 1): 1- 9. Google Scholar CrossRef Search ADS PubMed  8. Friedman JA, Goerss SJ, Meyer FB et al.   Volumetric quantification of fisher grade 3 aneurysmal subarachnoid hemorrhage: a novel method to predict symptomatic vasospasm on admission computerized tomography scans. J Neurosurg . 2002; 97( 2): 401- 407. Google Scholar CrossRef Search ADS PubMed  9. Frontera JA, Claassen J, Schmidt JM et al.   Prediction of symptomatic vasospasm after subarachnoid hemorrhage: the modified fisher scale. Neurosurgery . 2006; 59( 1): 21- 27; discussion 21-27. Google Scholar CrossRef Search ADS PubMed  10. Fujita S. Computed tomographic grading with hounsfield number related to delayed vasospasm in cases of ruptured cerebral aneurysm. Neurosurgery . 1985; 17( 4): 609- 612. Google Scholar CrossRef Search ADS PubMed  11. Hijdra A, Brouwers PJ, Vermeulen M, van Gijn J. Grading the amount of blood on computed tomograms after subarachnoid hemorrhage. Stroke . 1990; 21( 8): 1156- 1161. Google Scholar CrossRef Search ADS PubMed  12. Klimo P Jr, Kestle JR, MacDonald JD, Schmidt RH. Marked reduction of cerebral vasospasm with lumbar drainage of cerebrospinal fluid after subarachnoid hemorrhage. J Neurosurg . 2004; 100( 2): 215- 224. Google Scholar CrossRef Search ADS PubMed  13. Knuckey NW, Fox RA, Surveyor I, Stokes BA. Early cerebral blood flow and computerized tomography in predicting ischemia after cerebral aneurysm rupture. J Neurosurg . 1985; 62( 6): 850- 855. Google Scholar CrossRef Search ADS PubMed  14. Ko SB, Choi HA, Carpenter AM et al.   Quantitative analysis of hemorrhage volume for predicting delayed cerebral ischemia after subarachnoid hemorrhage. Stroke . 2011; 42( 3): 669- 674. Google Scholar CrossRef Search ADS PubMed  15. Mizukami M, Takemae T, Tazawa T, Kawase T, Matsuzaki T. Value of computed tomography in the prediction of cerebral vasospasm after aneurysm rupture. Neurosurgery . 1980; 7( 6): 583- 586. Google Scholar CrossRef Search ADS PubMed  16. Mohsen F, Pomonis S, Illingworth R. Prediction of delayed cerebral ischaemia after subarachnoid haemorrhage by computed tomography. J Neurol Neurosurg Psychiatry . 1984; 47( 11): 1197- 1202. Google Scholar CrossRef Search ADS PubMed  17. Wilson DA, Nakaji P, Abla AA et al.   A simple and quantitative method to predict symptomatic vasospasm after subarachnoid hemorrhage based on computed tomography: Beyond the fisher scale. Neurosurgery . 2012; 71( 4): 869- 875. Google Scholar CrossRef Search ADS PubMed  18. Klimo P Jr, Schmidt RH. Computed tomography grading schemes used to predict cerebral vasospasm after aneurysmal subarachnoid hemorrhage: a historical review. Neurosurg Focus . 2006; 21( 3): E5, 1- 8. Google Scholar CrossRef Search ADS   19. Takemae T, Mizukami M, Kin H, Kawase T, Araki G. Computed tomography of ruptured intracranial aneurysms in acute stage–relationship between vasospasm and high density on CT scan (author's transl). No To Shinkei. 1978; 30( 8): 861- 866. 20. Rosen DS, Macdonald RL. Subarachnoid hemorrhage grading scales: a systematic review. Neurocrit Care . 2005; 2( 2): 110- 118. Google Scholar CrossRef Search ADS PubMed  21. Rosen DS, Macdonald RL, Huo D et al.   Intraventricular hemorrhage from ruptured aneurysm: Clinical characteristics, complications, and outcomes in a large, prospective, multicenter study population. J Neurosurg . 2007; 107( 2): 261- 265. Google Scholar CrossRef Search ADS PubMed  22. Macdonald RL, Rosengart A, Huo D, Karrison T. Factors associated with the development of vasospasm after planned surgical treatment of aneurysmal subarachnoid hemorrhage. J Neurosurg . 2003; 99( 4): 644- 652. Google Scholar CrossRef Search ADS PubMed  23. Wilson TJ, Stetler WR Jr, Davis MC et al.   Intraventricular hemorrhage is associated with early hydrocephalus, symptomatic vasospasm, and poor outcome in aneurysmal subarachnoid hemorrhage. J Neurol Surg A Cent Eur Neurosurg . 2015; 76( 2): 126- 132. Google Scholar PubMed  24. Graeb DA, Robertson WD, Lapointe JS, Nugent RA, Harrison PB. Computed tomographic diagnosis of intraventricular hemorrhage. etiology and prognosis. Radiology . 1982; 143( 1): 91- 96. Google Scholar CrossRef Search ADS PubMed  25. Czorlich P, Mende KC, Vettorazzi E, Regelsberger J, Westphal M, Schmidt NO. Validation of the modified graeb score in aneurysmal subarachnoid hemorrhage. Acta Neurochir (Wien) . 2015; 157( 11): 1867- 1872. Google Scholar CrossRef Search ADS PubMed  26. Kramer AH, Hehir M, Nathan B et al.   A comparison of 3 radiographic scales for the prediction of delayed ischemia and prognosis following subarachnoid hemorrhage. J Neurosurg . 2008; 109( 2): 199- 207. Google Scholar CrossRef Search ADS PubMed  27. Macdonald RL, Kassell NF, Mayer S et al.   Clazosentan to overcome neurological ischemia and infarction occurring after subarachnoid hemorrhage (CONSCIOUS-1): randomized, double-blind, placebo-controlled phase 2 dose-finding trial. Stroke . 2008; 39( 11): 3015- 3021. Google Scholar CrossRef Search ADS PubMed  28. LeRoux PD, Haglund MM, Newell DW, Grady MS, Winn HR. Intraventricular hemorrhage in blunt head trauma: an analysis of 43 cases. Neurosurgery . 1992; 31( 4): 678- 684; discussion 684-685. Google Scholar PubMed  29. Ibrahim GM, Vachhrajani S, Ilodigwe D et al.   Method of aneurysm treatment does not affect clot clearance after aneurysmal subarachnoid hemorrhage. Neurosurgery . 2012; 70( 1): 102- 109; discussion 109. Google Scholar CrossRef Search ADS PubMed  30. Ibrahim GM, Weidauer S, Macdonald RL. Interobserver variability in the interpretation of computed tomography following aneurysmal subarachnoid hemorrhage. J Neurosurg . 2011; 115( 6): 1191- 1196. Google Scholar CrossRef Search ADS PubMed  31. Dupont SA, Wijdicks EF, Manno EM, Lanzino G, Rabinstein AA. Prediction of angiographic vasospasm after aneurysmal subarachnoid hemorrhage: value of the hijdra sum scoring system. Neurocrit Care . 2009; 11( 2): 172- 176. Google Scholar CrossRef Search ADS PubMed  32. Crowley RW, Medel R, Dumont AS et al.   Angiographic vasospasm is strongly correlated with cerebral infarction after subarachnoid hemorrhage. Stroke . 2011; 42( 4): 919- 923. Google Scholar CrossRef Search ADS PubMed  33. Friedrich B, Muller F, Feiler S, Scholler K, Plesnila N. Experimental subarachnoid hemorrhage causes early and long-lasting microarterial constriction and microthrombosis: an in-vivo microscopy study. J Cereb Blood Flow Metab . 2012; 32( 3): 447- 455. Google Scholar CrossRef Search ADS PubMed  34. Stein SC, Browne KD, Chen XH, Smith DH, Graham DI. Thromboembolism and delayed cerebral ischemia after subarachnoid hemorrhage: an autopsy study. Neurosurgery . 2006; 59( 4): 781- 787; discussion 787-788. Google Scholar CrossRef Search ADS PubMed  35. Dreier JP. The role of spreading depression, spreading depolarization and spreading ischemia in neurological disease. Nat Med . 2011; 17( 4): 439- 447. Google Scholar CrossRef Search ADS PubMed  COMMENTS Scales may serve 2 important functions. One is as a clinical tool, and as a rule these sorts of scales should be easy to remember and calculate and composed of elements with high inter-observer reliability. The second function is as a research tool, as a finite set of elements that shed light on associations or causations that may not otherwise be readily apparent. Many are some combination of the 2. The scale that the authors introduce in this paper is arguably more of the latter sort. It is too cumbersome to be readily used as a clinical tool, and not so simple conceptually as the regular Fisher scale or even the Barrow subarachnoid hemorrhage scale. From the standpoint of parsimony, it is most elegant when the grading scale has a direct relationship between grade and severity or risk which rises on an even slope (see Figure 1 in the paper, showing the Modified Fisher Scale Grade). For this scale, the disparity between the 2a and 2b groups disrupts this. Again, it does not detract from the conclusions it allows one to draw, but the lack of symmetry may impact the enthusiasm for adoption as a clinical tool. We can dream of a time when the extra data element involved is seamlessly incorporated into intelligent medical records by artificial intelligences who are not put off by such trivial complexities, so that this and other insights bubble up of their own accord – I may not be the only neurosurgeon to doubt whether we are actually getting closer to that goal. In the meantime, I applaud the authors for their insight that ventricular clot has a stronger positive association with vasospasm than some have previously appreciated. Now the observation needs to be tested against others’ data retrospectively, tracked prospectively, and hypotheses for the mechanism explored. Peter Nakaji Phoenix, Arizona The authors present an interesting analysis of the use of the modified Fisher scale combined with a modified Graeb score to predict risk of vasospasm in patients with aSAH. This analysis was carried out in the CONSCIOUS trial cohort. They find that a modified Graeb score of > 3 increases risk of vasospasm, even if the cisternal clot is not thick. This is an important observation. Specifically, intraventricular clot is predictive of vasospasm. While most clinicians likely will not find it necessary to calculate this new score for each new SAH patient, it will be helpful to note that patients with more intraventricular blood are at higher risk for delayed cerebral ischemia. Michael Froehler Nashville, Tennessee Copyright © 2017 by the Congress of Neurological Surgeons http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Neurosurgery Oxford University Press

Incorporating a Modified Graeb Score to the Modified Fisher Scale for Improved Risk Prediction of Delayed Cerebral Ischemia Following Aneurysmal Subarachnoid Hemorrhage

Loading next page...
 
/lp/ou_press/incorporating-a-modified-graeb-score-to-the-modified-fisher-scale-for-vxtFNaFK2g
Publisher
Oxford University Press
Copyright
Copyright © 2017 by the Congress of Neurological Surgeons
ISSN
0148-396X
eISSN
1524-4040
D.O.I.
10.1093/neuros/nyx165
Publisher site
See Article on Publisher Site

Abstract

Abstract BACKGROUND Delayed cerebral ischemia (DCI) is a cause of poor outcome after aneurysmal subarachnoid hemorrhage. Risk scales to predict DCI have scarcely been evaluated for predictive accuracy. Accounting for volume of intraventricular hemorrhage (IVH) in the modified Fischer scale (mFS) may improve its predictive accuracy. OBJECTIVE To compare the modified Graeb score (mGS) to the mFS for risk prediction of DCI, and to investigate whether incorporating an mGS cut-point into the mFS could improve predictive accuracy. METHODS This retrospective analysis was based on the Clazosentan to Overcome Neurological Ischemia and Infarction Occurring after Subarachnoid Hemorrhage (CONSCIOUS-1) trial cohort. IVH volume was quantified with the mGS. The relation of the mGS to DCI was evaluated using logistic regression and the area under the receiver operator characteristics curve (AUC). An optimized mGS cut-point was identified using the Youden index, and was incorporated into the mFS to dichotomize grades 2 and 4. The AUC was used to compare the predictive performance of the mGS with that of the mFS, and to assess whether there was an improvement in DCI prediction after creation of the dichotomized scale. RESULTS The mFS and the mGS had similar discrimination for DCI (AUC: 0.608 vs 0.618; P = .79). A new scale including both the mFS and mGS significantly improved the AUC compared to the mFS (AUC: 0.647 vs 0.608; P = .022). CONCLUSION The mFS and the mGS have similar prognostic utility. Accounting for IVH volume improved prediction of DCI by the mFS. Delayed cerebral ischemia, Vasospasm, Subarachnoid hemorrhage, Intraventricular hemorrhage, Fisher scale, Graeb score ABBREVIATIONS ABBREVIATIONS aSAH aneurysmal subarachnoid hemorrhage AUC area under the receiver operator characteristics curve CI confidence interval CONSCIOUS-1 Clazosentan to Overcome Neurological Ischemia and Infarction Occurring after Subarachnoid Hemorrhage DCI delayed cerebral ischemial FS Fisher scale GS Graeb Score ICC interclass correlation coefficient IVH intraventricular hemorrhage mFS modified Fischer scale mGS modified Graeb score Delayed cerebral ischemia (DCI) has been reported in 20% to 30% of patients who suffer aneurysmal subarachnoid hemorrhage (aSAH).1,2 It is one of the strongest predictors of poor outcome following aSAH.3 Furthermore, DCI is also associated with substantially higher costs to the healthcare system, and slower return to employment in affected patients.4 The risk of DCI may be quantified using a number of grading scales that are based primarily on neuroimaging;5-17 however, the properties of these scales have scarcely been studied nor compared. Fisher et al18 created a scale that has, historically, been widely used. Although the Fisher scale (FS) was among the first to demonstrate that patients with thick cisternal clots are at highest risk of developing DCI,7,19 it did not adequately address patients who may have thick cisternal clots with concomitant intraventricular hemorrhage (IVH).20 IVH has been found to be associated with poor outcome in patients who suffer aSAH.21 Some studies have found that IVH is an independent predictor of DCI,9,22 and that the risk of developing DCI may increase with greater volumes of IVH.5,14,23 Graeb et al24 proposed a semiquantitative measure of IVH,24 and this scale (GS) has been shown to correlate with poor outcomes following aSAH,25 yet little is known regarding its value in predicting DCI. In order to improve predictive performance of the FS, Frontera et al9 modified it to account for patients with both thick cisternal clots and IVH.9 This modification of the FS (mFS) is a better predictor of DCI than the original FS;9,26 however, the authors did not assess whether prediction could be further improved by accounting for IVH volume. The objective of this study was to investigate whether a modification of the GS (mGS) is valuable for DCI risk assessment in patients who suffer aSAH. The study also assessed whether altering the mFS by incorporating a measure of IVH volume, such as the mGS, would improve DCI risk prediction. Given the results of previous work,3,5,14,23 we hypothesized that this modification would improve the ability of the mFS to predict patients who will suffer DCI. In addition, it was hypothesized that data correlating IVH with DCI would support several different pathophysiological mechanisms that may cause DCI. METHODS The methodology for this post hoc analysis was designed to comply with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. Patient Population The study was based on the 413 patients who were enrolled in the Clazosentan to Overcome Neurological ischemia and Infarction Occurring after Subarachnoid Hemorrhage (CONSCIOUS-1) study (clinical trial registration no.: NCT00111085 [clinicaltrials.gov]) between January 2005 and March 2006. The methods and results of this trial have been published elsewhere.27 The institutional review board at St. Michael's Hospital granted approval for this study. As this was a post hoc analysis, renewed patient consent was not required for study approval. Variables All patients had baseline cranial CT scans that were analyzed by 2 independent, blinded reviewers. IVH was quantified using the mGS, which assigned a value of 0 (no blood), 1 (less than 25% filled), 2 (moderately filled), or 3 (completely filled) to each ventricle for a total score out of 12.28,29 For purposes of the analysis, the average of the reviewers’ scores was obtained. The mFS was computed for each patient according to Frontera et al.9 These scores were based on the reviewers’ interpretation of clot thickness, and whether they identified any IVH in their mGS. Where disagreement occurred between the 2 reviewers regarding clot thickness, an independent arbitrator made the final decision, which was considered for the analysis. Three patients did not have their clot thickness reported, and they were excluded from any analysis that related to the mFS. The primary outcome variable was the development of DCI, which was defined as angiographic vasospasm associated with neurological worsening (a decrease of at least 2 points on the modified Glasgow Coma Scale, or an increase of at least 2 points on the National Institutes of Health Stroke Scale) that lasted at least 2 h. When patients were not able to undergo a neurological examination, DCI was defined as angiographic vasospasm associated with clinical signs of DCI (unexplained fever or new neurological deficit), or a new infarction on a postprocedure CT scan not attributable to any other cause.27 Statistics Baseline characteristics of the cohort were compared according to DCI status using frequency tables and bar charts. Interobserver reliability for the mGS was calculated using the interclass correlation coefficient (ICC). We used binary logistic regression models to assess the relation of the mFS and the mGS to DCI. In order to assess for potential confounding between clot thickness and IVH volume, a multivariate regression model was created with clot thickness and mGS as predictors of DCI. The 2 scales were compared for predictive accuracy by calculating the area under the receiver operator characteristics curve (AUC). The Youden index was then used to identify an optimal cut-point for the mGS, in terms of maximal sensitivity and specificity, for predicting DCI. A new scale was created, which dichotomized grades 2 and 4 of the mFS based on the optimal mGS cut-point. This new scale was then compared to the mFS to examine if there was any increase in risk prediction of DCI. Improvement in prediction was evaluated by comparing the AUC of the 3 models. Significance level was set at P < .05. Statistical analysis was carried out using Stata 13.1 (Statacorp, College Station, Texas). RESULTS Of the 413 patients in the CONSCIOUS-1 dataset, 78 (19%) developed DCI. Patients who developed DCI tended to present with higher World Federation of Neurosurgical Societies grades, were more likely to have their aneurysm clipped, have an external ventricular drain placed, have thick cisternal clots, and tended to have greater IVH volume as assessed by the mGS. Baseline characteristics of the cohort are shown in Table 1. Median mGS was 3 with interquartile range of 1 to 4.5. A 2-way, mixed effects ICC analysis of the mGS demonstrated excellent interobserver agreement (ICC 0.84; 95% confidence interval [CI] 0.81-0.86). Both the mFS and the mGS were significant predictors of DCI on univariate regression (Table 2), and the significance of the mGS was maintained when combined into a multivariate regression with clot thickness to account for potential confounding (odds ratio 1.13; 95% CI 1.02-1.25). The optimal mGS cut-point, as determined by the Youden index, was 3.25. Because the mGS is a discrete scale, we used a score of greater than 3 as our cut-off value. For patients with an mGS greater than 3, logistic regression demonstrated no significant difference in the risk of DCI for patients with a single ventricle score of 3 (P = .63), with IVH in all 4 ventricles (P = .85), with bilateral IVH (P = .89), or with 2 ventricles having mGSs of 2 to 3 (P = .89). The mFS was related to DCI in a gradient manner (Figure 1). Incorporating the optimized mGS cut-point altered the risk of DCI compared to the mFS (Figure 2). FIGURE 1. View largeDownload slide Bar chart showing the graded relation between mFS and DCI. FIGURE 1. View largeDownload slide Bar chart showing the graded relation between mFS and DCI. FIGURE 2. View largeDownload slide Bar chart showing the relation between the dichotomized scale and DCI. FIGURE 2. View largeDownload slide Bar chart showing the relation between the dichotomized scale and DCI. TABLE 1. Baseline Characteristics of the Study Cohort Compared According to DCI Status   Total  DCI  No DCI  N  413  78  335  Age (median and IQR)  51 (44, 60)  51 (43, 60)  51 (44, 59)  Female sex  292 (70.1%)  60 (76.9%)  232 (69.3%)  Race     White  368 (89.1%)  73 (93.6%)  295 (88.1%)   Black  27 (6.5%)  4 black (5.1%)  23 (6.9%)   Other  18 (4.4%)  1 other (1.3%)  17 (5.1%)  WFNS grade     1  181 (43.8%)  20 (25.6%)  161 (48.1%)   2  117 (28.3%)  22 (28.2%)  95 (28.4%)   3  12 (2.9%)  2 (2.6%)  10 (3.0%)   4  97 (23.5%)  33 (42.3%)  64 (19.1%)   5  3 (0.7%)  1 (1.3%)  2 (0.6%)  History of hypertension  175 (42.4%)  39 (50.0%)  136 (40.1%)  Nicotine use  217 (52.5%)  38 (48.7%)  179 (53.4%)  Surgical clipping  186 (45.0%)  43 (55.1%)  143 (42.7%)  EVD placement  149 (36.1%)  48 (61.5%)  101 (30.1%)  Shunt placement  37 (9.0%)  9 (11.5%)  28 (8.4%)  Hydrocephalus  392 (94.9%)  77 (98.7%)  315 (94.0%)  Thick cisternal clot  302 (73.1%)  69 (88.5%)  233 (69.6%)  Modified Graeb score (median and IQR)  3 (1, 4.5)  4 (2, 5)  2.5 (1, 4)  Modified Fisher grade (N = 410)     1  22 (5.4%)  1 (1.3%)  21 (6.3%)   2  86 (21.0%)  8 (10.3%)  78 (23.5%)   3  42 (10.2%)  7 (9.0%)  35 (10.5%)   4  260 (63.4%)  62 (79.5%)  198 (59.6%)  Unfavorable outcome (GOSE 1-4)  125 (30.3%)  46 (60.0%)  79 (23.6%)    Total  DCI  No DCI  N  413  78  335  Age (median and IQR)  51 (44, 60)  51 (43, 60)  51 (44, 59)  Female sex  292 (70.1%)  60 (76.9%)  232 (69.3%)  Race     White  368 (89.1%)  73 (93.6%)  295 (88.1%)   Black  27 (6.5%)  4 black (5.1%)  23 (6.9%)   Other  18 (4.4%)  1 other (1.3%)  17 (5.1%)  WFNS grade     1  181 (43.8%)  20 (25.6%)  161 (48.1%)   2  117 (28.3%)  22 (28.2%)  95 (28.4%)   3  12 (2.9%)  2 (2.6%)  10 (3.0%)   4  97 (23.5%)  33 (42.3%)  64 (19.1%)   5  3 (0.7%)  1 (1.3%)  2 (0.6%)  History of hypertension  175 (42.4%)  39 (50.0%)  136 (40.1%)  Nicotine use  217 (52.5%)  38 (48.7%)  179 (53.4%)  Surgical clipping  186 (45.0%)  43 (55.1%)  143 (42.7%)  EVD placement  149 (36.1%)  48 (61.5%)  101 (30.1%)  Shunt placement  37 (9.0%)  9 (11.5%)  28 (8.4%)  Hydrocephalus  392 (94.9%)  77 (98.7%)  315 (94.0%)  Thick cisternal clot  302 (73.1%)  69 (88.5%)  233 (69.6%)  Modified Graeb score (median and IQR)  3 (1, 4.5)  4 (2, 5)  2.5 (1, 4)  Modified Fisher grade (N = 410)     1  22 (5.4%)  1 (1.3%)  21 (6.3%)   2  86 (21.0%)  8 (10.3%)  78 (23.5%)   3  42 (10.2%)  7 (9.0%)  35 (10.5%)   4  260 (63.4%)  62 (79.5%)  198 (59.6%)  Unfavorable outcome (GOSE 1-4)  125 (30.3%)  46 (60.0%)  79 (23.6%)  DCI, delayed cerebral ischemia; IQR, interquartile range; WFNS, World Federation of Neurological Societies; GOSE, Glasgow Outcome Scale Extended; EVD, external ventricular drain. View Large TABLE 2. Univariate Regression Analysis of mFS, mGS, and New Dichotomized Scale Scale  Odds ratio  95% CI  mFS  1.76  1.27 - 2.44  mGS  1.15  1.04 - 1.28  Scale  Odds ratio  95% CI  mFS  1.76  1.27 - 2.44  mGS  1.15  1.04 - 1.28  mFS, modified Fisher Scale; mGS modified Graeb Score. View Large We found that the mGS had a slightly higher AUC (0.618; 95% CI: 0.549-0.686) than the mFS (0.608; 95% CI: 0.558-0.659); nonetheless, the difference was not statistically significant (P = .79). This suggests that both scales are essentially equivalent for predicting risk of DCI after aSAH. The dichotomized scale, as described in Table 3, incorporated both the mGS cut-point and the mFS. It had a significantly higher AUC (0.647; 95% CI: 0.587-0.707; P = .022) than the mFS, indicating that accounting for IVH volume improved the predictive accuracy of the mFS in our sample. TABLE 3. Comparison of Modified Fisher Scale According to Frontera et al (2006) and New Dichotomized Scale Modified Fisher grade according to Frontera et al (2006)9  Cisternal clot thickness  IVH  New dichotomized scale grade  Cisternal clot thickness  IVH  1  Thin  Absent  1  Thin  Absent  2  Thin  Present  2a  Thin  mGS ≤ 3        2b  Thin  mGS > 3  3  Thick  Absent  3  Thick  Absent  4  Thick  Present  4a  Thick  mGS ≤ 3        4b  Thick  mGS > 3  Modified Fisher grade according to Frontera et al (2006)9  Cisternal clot thickness  IVH  New dichotomized scale grade  Cisternal clot thickness  IVH  1  Thin  Absent  1  Thin  Absent  2  Thin  Present  2a  Thin  mGS ≤ 3        2b  Thin  mGS > 3  3  Thick  Absent  3  Thick  Absent  4  Thick  Present  4a  Thick  mGS ≤ 3        4b  Thick  mGS > 3  IVH, intraventricular hemorrhage; mGS, modified Graeb score. View Large DISCUSSION Our study demonstrated that the mGS and the mFS are essentially equivalent scales for DCI risk prediction. While there was no significant difference in the predictive value of the 2 scales, Ibrahim et al30 previously investigated the interobserver variability in both the mFS and mGS. They found that there was more variability in qualitative scales (eg, mFS) than in semiquantitative scores (eg, mGS). This led them to infer that the use of quantitative scores, such as the mGS, may be more consistent when assessing the risk of DCI.30 Our analysis also found excellent interobserver agreement for the mGS. Nonetheless, our most significant finding is that accounting for IVH volume in the mFS significantly improves the risk prediction of DCI in patients who suffer aSAH. Past radiographic scales to predict DCI after aSAH have varied in their parameters and performance (Table 4).5-17 These scales have scarcely been compared for predictive accuracy. Of note, a 2008 study by Kramer et al26 used logistic regression analysis to suggest that both the mFS and the scale proposed by Claassen et al5 were superior to the original FS, but the authors did not evaluate risk discrimination using the AUC.26 TABLE 4. Scales for Radiographic Prediction of DCI Authors and year  Patients  Type of Study  Parameters  Performance  Fisher et al (1980)7  47  Prospective  Grades 1 to 4 based on clot thickness, presence/absence of IVH or ICH  Grade 3 SAH—92% sensitive and 95% specific for DCI  Mizukami et al (1980)15  177  Prospective  Presence clot 1 to 4 d after rupture  High-density clot on CT within 4 d—100% sensitive and 67% specific for DCI  Davis et al (1980)6  50  Prospective  Class 1 to 4 based on increasing extent of SAH  Class 4 SAH—55% sensitive and 75% specific for severe angiographic vasospasm  Mohsen et al (1984)16  100  Prospective  Grades 1 to 3 based on clot thickness and extent  Grade 3 SAH—71% sensitive and 64% specific for DCI  Fujita (1985)10  36  Retrospective  Grades 1 to 5 based on Hounsfield numbers  Grade 5 SAH—75% sensitive and 86% specific for severe vasospasm or death  Knuckey et al (1985)13  46  Prospective  Grades 1 to 4 increasing with more diffuse SAH  Grade 4 SAH—47% sensitive and 87% specific for DCI  Hijdra et al (1990)11  182  Prospective  Extent of SAH scored 0 to 30; extent of IVH scored 0 to 12  Not reported  Claassen et al (2001)5  276  Prospective  Grades 1 to 4 based on clot thickness and presence/absence of bilateral IVH or ICH  Grade 4 SAH—35% sensitive and 40% specific for DCI or infarction  Friedman et al (2002)8  40  Prospective  Software-based volumetric quantification  SAH volume > 20 cm—46% sensitive and 100% specific for DCI  Klimo et al (2004)12  266  Retrospective  Created Fisher 3+4 grade to include patients with thick clot and IVH (>5 mL)  Grade 3+4 SAH—50% sensitive and 57% specific for vasospasm  Frontera et al (2006)9  1355  Retrospective  Grades 1 to 4 to incorporate patients with thick clot and IVH  Not reported  Ko et al (2011)14  160  Retrospective  CIHV stratified into quintiles  Fifth quintile—35% sensitive and 35% specific for DCI  Wilson et al (2012)17  250  Retrospective  BNI grades 1 to 5 based on maximal clot thickness in 5 mm increments  BNI grade 5 SAH—15% sensitive and 50% specific for symptomatic vasospasm  New Dichotomized Scale  410  Retrospective  Grades 1 to 4b account for clot thickness and mGS cut-off of >3  Grade 4b SAH—53% sensitive and 28% specific for DCI  Authors and year  Patients  Type of Study  Parameters  Performance  Fisher et al (1980)7  47  Prospective  Grades 1 to 4 based on clot thickness, presence/absence of IVH or ICH  Grade 3 SAH—92% sensitive and 95% specific for DCI  Mizukami et al (1980)15  177  Prospective  Presence clot 1 to 4 d after rupture  High-density clot on CT within 4 d—100% sensitive and 67% specific for DCI  Davis et al (1980)6  50  Prospective  Class 1 to 4 based on increasing extent of SAH  Class 4 SAH—55% sensitive and 75% specific for severe angiographic vasospasm  Mohsen et al (1984)16  100  Prospective  Grades 1 to 3 based on clot thickness and extent  Grade 3 SAH—71% sensitive and 64% specific for DCI  Fujita (1985)10  36  Retrospective  Grades 1 to 5 based on Hounsfield numbers  Grade 5 SAH—75% sensitive and 86% specific for severe vasospasm or death  Knuckey et al (1985)13  46  Prospective  Grades 1 to 4 increasing with more diffuse SAH  Grade 4 SAH—47% sensitive and 87% specific for DCI  Hijdra et al (1990)11  182  Prospective  Extent of SAH scored 0 to 30; extent of IVH scored 0 to 12  Not reported  Claassen et al (2001)5  276  Prospective  Grades 1 to 4 based on clot thickness and presence/absence of bilateral IVH or ICH  Grade 4 SAH—35% sensitive and 40% specific for DCI or infarction  Friedman et al (2002)8  40  Prospective  Software-based volumetric quantification  SAH volume > 20 cm—46% sensitive and 100% specific for DCI  Klimo et al (2004)12  266  Retrospective  Created Fisher 3+4 grade to include patients with thick clot and IVH (>5 mL)  Grade 3+4 SAH—50% sensitive and 57% specific for vasospasm  Frontera et al (2006)9  1355  Retrospective  Grades 1 to 4 to incorporate patients with thick clot and IVH  Not reported  Ko et al (2011)14  160  Retrospective  CIHV stratified into quintiles  Fifth quintile—35% sensitive and 35% specific for DCI  Wilson et al (2012)17  250  Retrospective  BNI grades 1 to 5 based on maximal clot thickness in 5 mm increments  BNI grade 5 SAH—15% sensitive and 50% specific for symptomatic vasospasm  New Dichotomized Scale  410  Retrospective  Grades 1 to 4b account for clot thickness and mGS cut-off of >3  Grade 4b SAH—53% sensitive and 28% specific for DCI  DCI, delayed cerebral ischemia; IVH, intraventricular hemorrhage; ICH, intracerebral hemorrhage; SAH subarachnoid hemorrhage; CIHV, cisternal and intraventricular hemorrhage volume; BNI, Barrow Neurological Institute; mGS, modified Graeb Score. View Large Based on the findings of previous work suggesting a strong association between total hemorrhage volume with DCI14,31 and the negative impact of IVH on patient outcomes,3 we had anticipated that accounting for IVH volume would improve the predictive performance of the mFS. A comparison of the AUC for the mFS and the dichotomized scale validated this prediction by demonstrating improved predictive accuracy when the mGS cut-point was incorporated into the mFS. It is possible that IVH volume is correlated with the other risk items in the mFS (eg, clot thickness or localization), which may have overstated the predictive information contributable by IVH volume. A study from Wilson et al17 found that after stratifying clot thickness into 5 mm increments, IVH was no longer associated with the development of DCI, indicating that any risk increases associated with IVH may be secondary to the relationship between total hemorrhage volume and IVH.17 However, the mGS remained a significant predictor of DCI when combined with SAH clot thickness in a multivariate regression model, suggesting that the mGS was independently associated with DCI in our cohort. A number of reasons may help explain our findings and contribute to hypothesis generation regarding the pathophysiology of DCI. It is possible that a greater amount of blood in the ventricular system is more likely to trigger angiographic vasospasm, microcirculatory constriction, microthrombosis, cortical spreading ischemia, and delayed cell apoptosis, all of which have been implicated in the pathogenesis of DCI.1,7,32-35 Another possibility is that the distribution of the IVH, rather than the quantity, is the mediating factor. Claassen et al5 found that bilateral IVH was a stronger predictor of DCI than the presence or absence of IVH alone. This led them to propose a modification to the FS that accounted for bilateral IVH.5 More widespread IVH would also be represented in the mGS, as each ventricle with blood in it would contribute to the score. However, our study is the first to our knowledge that incorporated an optimized cut-point for IVH volume (as determined by the Youden index) to the mFS, and subsequently evaluated the predictive accuracy of the scales by comparing the AUC. A study examining whether bilateral IVH or the optimized mGS cut-point is a better predictor of DCI is warranted. In their review, Klimo et al18 cited ease of use as one of the main advantages of the original FS. The mGS is more complex to compute and incorporating it into the mFS may result in an unwieldy scale with only marginally improved predictive performance for DCI. More complex grading schemes such as the one proposed by Ko et al,14 which attempted to quantify cisternal plus IVH volume, have failed to gain the same widespread use as the FS, likely because of the time and expertise required to calculate them. However, of note is the drastic difference in DCI risk between a grade 2a aSAH and a grade 2b aSAH according to our dichotomized scale. The implication of this finding is that patients with a thin clot and an mGS greater than 3 should not be classified as having the same risk of DCI as patients with a thin clot and an mGS less than 3, which was the case in the mFS. Limitations The strengths of this study are that blinded reviewers conducted CT scan measurements, and the definition of DCI was pre-specified. Possible limitations in this study also deserve mention. This is a retrospective analysis based on cohorts that were not primary assembled for the purpose of the study. One implication was that we had to resort to deriving the mFS from variables indicating clot thickness, location, and presence of IVH in the CONSCIOUS-1 dataset, since the mFS was not recorded a priori. Another issue is that an exclusion criterion in the CONSCIOUS-1 trail was IVH or intracerebral hemorrhage in the absence of localized thick or diffuse aSAH, which may have over-represented patients with mFS grades 3 and 4.27 Lastly, our study may not have been sufficiently powered given the small number of patients who went on to develop DCI. CONCLUSION The objective of this study was to examine whether the mGS is a valuable tool in predicting risk of DCI in patients who suffer aSAH, and to assess whether the mFS could be improved by incorporating an optimized mGS cut-point to dichotomize grades 2 and 4. We found that for purposes of risk classification of DCI, the mGS and mFS are essentially equivalent. The mGS remained a significant predictor of DCI when combined with clot thickness in a multivariate regression model. Altering the mFS to account for IVH volume according to an optimized mGS cut-point improved predictive accuracy. Future work should look to validate these findings in a prospective study that does not rely on the use of preassembled data. Disclosures Matt Eagles received funding for this work from St. Michael's Hospital as part of the Keenan Research Summer Studentship. Dr Macdonald reports ownership of Edge Therapeutics, Inc. and receiving non-study-related clinical or research effort support from the Brain Aneurysm Foundation, Heart and Stroke Foundation of Canada, Canadian Institutes for Health Research, and Physicians Services Incorporated Foundation. The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article. A portion of the work was accepted as an electronic poster at the 2016 AANS Annual Scientific Meeting in Chicago, Illinois from May 29 to June 4, 2016. REFERENCES 1. Macdonald RL. Delayed neurological deterioration after subarachnoid haemorrhage. Nat Rev Neurol . 2014; 10( 1): 44- 58. Google Scholar CrossRef Search ADS PubMed  2. Dorhout Mees SM, Kerr RS, Rinkel GJ, Algra A, Molyneux AJ. Occurrence and impact of delayed cerebral ischemia after coiling and after clipping in the international subarachnoid aneurysm trial (ISAT). J Neurol . 2012; 259( 4): 679- 683. Google Scholar CrossRef Search ADS PubMed  3. Rosengart AJ, Schultheiss KE, Tolentino J, Macdonald RL. Prognostic factors for outcome in patients with aneurysmal subarachnoid hemorrhage. Stroke . 2007; 38( 8): 2315- 2321. Google Scholar CrossRef Search ADS PubMed  4. Rivero-Arias O, Wolstenholme J, Gray A et al.   The costs and prognostic characteristics of ischaemic neurological deficit due to subarachnoid haemorrhage in the united kingdom. evidence from the MRC international subarachnoid aneurysm trial. J Neurol . 2009; 256( 3): 364- 373. Google Scholar CrossRef Search ADS PubMed  5. Claassen J, Bernardini GL, Kreiter K et al.   Effect of cisternal and ventricular blood on risk of delayed cerebral ischemia after subarachnoid hemorrhage: The fisher scale revisited. Stroke . 2001; 32( 9): 2012- 2020. Google Scholar CrossRef Search ADS PubMed  6. Davis JM, Davis KR, Crowell RM. Subarachnoid hemorrhage secondary to ruptured intracranial aneurysm: prognostic significance of cranial CT. AJR Am J Roentgenol . 1980; 134( 4): 711- 715. Google Scholar CrossRef Search ADS PubMed  7. Fisher CM, Kistler JP, Davis JM. Relation of cerebral vasospasm to subarachnoid hemorrhage visualized by computerized tomographic scanning. Neurosurgery . 1980; 6( 1): 1- 9. Google Scholar CrossRef Search ADS PubMed  8. Friedman JA, Goerss SJ, Meyer FB et al.   Volumetric quantification of fisher grade 3 aneurysmal subarachnoid hemorrhage: a novel method to predict symptomatic vasospasm on admission computerized tomography scans. J Neurosurg . 2002; 97( 2): 401- 407. Google Scholar CrossRef Search ADS PubMed  9. Frontera JA, Claassen J, Schmidt JM et al.   Prediction of symptomatic vasospasm after subarachnoid hemorrhage: the modified fisher scale. Neurosurgery . 2006; 59( 1): 21- 27; discussion 21-27. Google Scholar CrossRef Search ADS PubMed  10. Fujita S. Computed tomographic grading with hounsfield number related to delayed vasospasm in cases of ruptured cerebral aneurysm. Neurosurgery . 1985; 17( 4): 609- 612. Google Scholar CrossRef Search ADS PubMed  11. Hijdra A, Brouwers PJ, Vermeulen M, van Gijn J. Grading the amount of blood on computed tomograms after subarachnoid hemorrhage. Stroke . 1990; 21( 8): 1156- 1161. Google Scholar CrossRef Search ADS PubMed  12. Klimo P Jr, Kestle JR, MacDonald JD, Schmidt RH. Marked reduction of cerebral vasospasm with lumbar drainage of cerebrospinal fluid after subarachnoid hemorrhage. J Neurosurg . 2004; 100( 2): 215- 224. Google Scholar CrossRef Search ADS PubMed  13. Knuckey NW, Fox RA, Surveyor I, Stokes BA. Early cerebral blood flow and computerized tomography in predicting ischemia after cerebral aneurysm rupture. J Neurosurg . 1985; 62( 6): 850- 855. Google Scholar CrossRef Search ADS PubMed  14. Ko SB, Choi HA, Carpenter AM et al.   Quantitative analysis of hemorrhage volume for predicting delayed cerebral ischemia after subarachnoid hemorrhage. Stroke . 2011; 42( 3): 669- 674. Google Scholar CrossRef Search ADS PubMed  15. Mizukami M, Takemae T, Tazawa T, Kawase T, Matsuzaki T. Value of computed tomography in the prediction of cerebral vasospasm after aneurysm rupture. Neurosurgery . 1980; 7( 6): 583- 586. Google Scholar CrossRef Search ADS PubMed  16. Mohsen F, Pomonis S, Illingworth R. Prediction of delayed cerebral ischaemia after subarachnoid haemorrhage by computed tomography. J Neurol Neurosurg Psychiatry . 1984; 47( 11): 1197- 1202. Google Scholar CrossRef Search ADS PubMed  17. Wilson DA, Nakaji P, Abla AA et al.   A simple and quantitative method to predict symptomatic vasospasm after subarachnoid hemorrhage based on computed tomography: Beyond the fisher scale. Neurosurgery . 2012; 71( 4): 869- 875. Google Scholar CrossRef Search ADS PubMed  18. Klimo P Jr, Schmidt RH. Computed tomography grading schemes used to predict cerebral vasospasm after aneurysmal subarachnoid hemorrhage: a historical review. Neurosurg Focus . 2006; 21( 3): E5, 1- 8. Google Scholar CrossRef Search ADS   19. Takemae T, Mizukami M, Kin H, Kawase T, Araki G. Computed tomography of ruptured intracranial aneurysms in acute stage–relationship between vasospasm and high density on CT scan (author's transl). No To Shinkei. 1978; 30( 8): 861- 866. 20. Rosen DS, Macdonald RL. Subarachnoid hemorrhage grading scales: a systematic review. Neurocrit Care . 2005; 2( 2): 110- 118. Google Scholar CrossRef Search ADS PubMed  21. Rosen DS, Macdonald RL, Huo D et al.   Intraventricular hemorrhage from ruptured aneurysm: Clinical characteristics, complications, and outcomes in a large, prospective, multicenter study population. J Neurosurg . 2007; 107( 2): 261- 265. Google Scholar CrossRef Search ADS PubMed  22. Macdonald RL, Rosengart A, Huo D, Karrison T. Factors associated with the development of vasospasm after planned surgical treatment of aneurysmal subarachnoid hemorrhage. J Neurosurg . 2003; 99( 4): 644- 652. Google Scholar CrossRef Search ADS PubMed  23. Wilson TJ, Stetler WR Jr, Davis MC et al.   Intraventricular hemorrhage is associated with early hydrocephalus, symptomatic vasospasm, and poor outcome in aneurysmal subarachnoid hemorrhage. J Neurol Surg A Cent Eur Neurosurg . 2015; 76( 2): 126- 132. Google Scholar PubMed  24. Graeb DA, Robertson WD, Lapointe JS, Nugent RA, Harrison PB. Computed tomographic diagnosis of intraventricular hemorrhage. etiology and prognosis. Radiology . 1982; 143( 1): 91- 96. Google Scholar CrossRef Search ADS PubMed  25. Czorlich P, Mende KC, Vettorazzi E, Regelsberger J, Westphal M, Schmidt NO. Validation of the modified graeb score in aneurysmal subarachnoid hemorrhage. Acta Neurochir (Wien) . 2015; 157( 11): 1867- 1872. Google Scholar CrossRef Search ADS PubMed  26. Kramer AH, Hehir M, Nathan B et al.   A comparison of 3 radiographic scales for the prediction of delayed ischemia and prognosis following subarachnoid hemorrhage. J Neurosurg . 2008; 109( 2): 199- 207. Google Scholar CrossRef Search ADS PubMed  27. Macdonald RL, Kassell NF, Mayer S et al.   Clazosentan to overcome neurological ischemia and infarction occurring after subarachnoid hemorrhage (CONSCIOUS-1): randomized, double-blind, placebo-controlled phase 2 dose-finding trial. Stroke . 2008; 39( 11): 3015- 3021. Google Scholar CrossRef Search ADS PubMed  28. LeRoux PD, Haglund MM, Newell DW, Grady MS, Winn HR. Intraventricular hemorrhage in blunt head trauma: an analysis of 43 cases. Neurosurgery . 1992; 31( 4): 678- 684; discussion 684-685. Google Scholar PubMed  29. Ibrahim GM, Vachhrajani S, Ilodigwe D et al.   Method of aneurysm treatment does not affect clot clearance after aneurysmal subarachnoid hemorrhage. Neurosurgery . 2012; 70( 1): 102- 109; discussion 109. Google Scholar CrossRef Search ADS PubMed  30. Ibrahim GM, Weidauer S, Macdonald RL. Interobserver variability in the interpretation of computed tomography following aneurysmal subarachnoid hemorrhage. J Neurosurg . 2011; 115( 6): 1191- 1196. Google Scholar CrossRef Search ADS PubMed  31. Dupont SA, Wijdicks EF, Manno EM, Lanzino G, Rabinstein AA. Prediction of angiographic vasospasm after aneurysmal subarachnoid hemorrhage: value of the hijdra sum scoring system. Neurocrit Care . 2009; 11( 2): 172- 176. Google Scholar CrossRef Search ADS PubMed  32. Crowley RW, Medel R, Dumont AS et al.   Angiographic vasospasm is strongly correlated with cerebral infarction after subarachnoid hemorrhage. Stroke . 2011; 42( 4): 919- 923. Google Scholar CrossRef Search ADS PubMed  33. Friedrich B, Muller F, Feiler S, Scholler K, Plesnila N. Experimental subarachnoid hemorrhage causes early and long-lasting microarterial constriction and microthrombosis: an in-vivo microscopy study. J Cereb Blood Flow Metab . 2012; 32( 3): 447- 455. Google Scholar CrossRef Search ADS PubMed  34. Stein SC, Browne KD, Chen XH, Smith DH, Graham DI. Thromboembolism and delayed cerebral ischemia after subarachnoid hemorrhage: an autopsy study. Neurosurgery . 2006; 59( 4): 781- 787; discussion 787-788. Google Scholar CrossRef Search ADS PubMed  35. Dreier JP. The role of spreading depression, spreading depolarization and spreading ischemia in neurological disease. Nat Med . 2011; 17( 4): 439- 447. Google Scholar CrossRef Search ADS PubMed  COMMENTS Scales may serve 2 important functions. One is as a clinical tool, and as a rule these sorts of scales should be easy to remember and calculate and composed of elements with high inter-observer reliability. The second function is as a research tool, as a finite set of elements that shed light on associations or causations that may not otherwise be readily apparent. Many are some combination of the 2. The scale that the authors introduce in this paper is arguably more of the latter sort. It is too cumbersome to be readily used as a clinical tool, and not so simple conceptually as the regular Fisher scale or even the Barrow subarachnoid hemorrhage scale. From the standpoint of parsimony, it is most elegant when the grading scale has a direct relationship between grade and severity or risk which rises on an even slope (see Figure 1 in the paper, showing the Modified Fisher Scale Grade). For this scale, the disparity between the 2a and 2b groups disrupts this. Again, it does not detract from the conclusions it allows one to draw, but the lack of symmetry may impact the enthusiasm for adoption as a clinical tool. We can dream of a time when the extra data element involved is seamlessly incorporated into intelligent medical records by artificial intelligences who are not put off by such trivial complexities, so that this and other insights bubble up of their own accord – I may not be the only neurosurgeon to doubt whether we are actually getting closer to that goal. In the meantime, I applaud the authors for their insight that ventricular clot has a stronger positive association with vasospasm than some have previously appreciated. Now the observation needs to be tested against others’ data retrospectively, tracked prospectively, and hypotheses for the mechanism explored. Peter Nakaji Phoenix, Arizona The authors present an interesting analysis of the use of the modified Fisher scale combined with a modified Graeb score to predict risk of vasospasm in patients with aSAH. This analysis was carried out in the CONSCIOUS trial cohort. They find that a modified Graeb score of > 3 increases risk of vasospasm, even if the cisternal clot is not thick. This is an important observation. Specifically, intraventricular clot is predictive of vasospasm. While most clinicians likely will not find it necessary to calculate this new score for each new SAH patient, it will be helpful to note that patients with more intraventricular blood are at higher risk for delayed cerebral ischemia. Michael Froehler Nashville, Tennessee Copyright © 2017 by the Congress of Neurological Surgeons

Journal

NeurosurgeryOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off