Outcome of Early versus Delayed Evacuation of Spontaneous Lobar Hematomas in Unconscious Adults

Patients

This is a retrospective analysis of data records of patients admitted to the Department of Neurosurgery with the diagnosis of spontaneous lobar ICH, during an interval of 18 years from January 1998 to December 2015. Patients included those admitted initially to the Neurosurgical Department in addition to those referred to it from the Neurology Department with the same diagnosis for surgical intervention. Patients included were those presented in a disturbed conscious status yet not comatose (GCS >8 and <14). The included cases were operated according to different timing strategies based on surgeons’ experience and patients’ suitability. Time to surgery was calculated for all patients using the earliest documented point of time available which was the time of medical presentation in the emergency form. Reference to the presentation time was used because data about time of onset of symptoms were missing in a significant proportion of patients to be reliably imputed by imputation methods. The results were thus expected to reflect the effects of the time window from medical presentation rather than from the ictus. The following patients were excluded: (1) those referred from the Department of Neurology in a poor neurological status (derived GCS score ≤8), (2) hematomas secondary to underlying lesions such as vascular malformation, tumor, and venous sinus thrombosis, (3) history of previous stroke, (4) sizable intraventricular clot occupying >1/3 of either lateral ventricle (allowing for partial or complete occlusion of third and fourth ventricles so long as no ventricular expansion detected), (5) uncorrected coagulopathy before surgery, (6) cases with successful conservative treatment and improvement of neurological status, and (7) associated hydrocephalus due to either; the hematoma mass effect or a blood clot obstructing cerebrospinal fluid pathways.

Data collection

The process of data collection was done in two parallel independent processes. It started by reviewing the outpatient clinic follow-up database to select cases with the diagnosis of lobar ICH, who underwent surgical evacuation in the Neurosurgical Department during the mentioned time interval. This was done by one of the authors who collected the serial numbers and follow-up data. The other authors blinded to the follow-up data used the serial numbers of the selected patients to review the department archive and extract the demographic, clinical, radiological, and management data to complete the process of selection by applying the remaining inclusion and exclusion criteria. After completion of data collection, authors provided the data to an independent biostatistician who assembled the data using the serial numbers to start the statistical analysis.

Demographic data included age, sex, marital status, residence, and smoking. Clinical data included initial and immediate preoperative GCS and National Institutes of Health Stroke Scale (NIHSS) scores, in addition to neurological deficits, systolic, diastolic, and mean blood pressures, associated medical conditions such as history of hypertension, diabetes mellitus, liver or renal disorders, and antiplatelet and/or anticoagulant medications. In addition, laboratory tests such as serum glucose and international normalized ratio (INR) were collected for analysis. The level of consciousness in this study was not based on “directly calculated” GCS scores because verbal responses were missing in a significant proportion of patients who presented with aphasia due to hematomas affecting their dominant hemispheres. Instead, we depended on the motor and eye scores to predict the missing verbal response and obtain “derived GCS scores” using a validated regression equation,[1718] then rounding the scores to the nearest integer.

Radiological data included side of hematoma, ICH volume in initial and follow-up computed tomography (CT) images using Broderick's formula (A × B × C/2),[4] minimum distance of ICH from cortical surface (in millimeters), associated intraventricular hemorrhage (IVH) and IVH score according to Hallevi et al. [Table 1], with estimation of the IVH volume and total volume (TV) (TV = ICH volume + IVH volume).[19] Rebleeding was defined as a hematoma size in a follow-up CT >33% of the ICH volume on the previously done CT.[202122] Data related to management including medical treatment, time of intervention, surgical technique, overall hospital stay, and postoperative mortality were also collected for analysis.

Table 1 IVH score according to Hallevi et al.[20]

Outcome assessment

Functional assessment was performed using the modified Rankin Scale (mRS). The mRS scores of patients were collected at all available points of time from discharge till data collection. The mRS score at the last visit was considered as the final outcome. The final scores were dichotomized into favorable versus poor outcomes taking into account the baseline severity. First, the prognostic scores of patients were calculated from the following equation: 10 × GCS − (age in years) − (0.64 × volume in mL),[15] followed by dividing the patients at the median score into good and poor prognosis groups. For the good prognosis group, favorable outcome was defined as good recovery or moderate disability (mRS ≤3), while for the poor prognosis group, it included also the moderately severe disability (mRS ≤4). Several researchers have proposed this prognosis-based methodology.[2324]

Ethical approval

Written informed consents were provided by relevant individuals before any surgical intervention. The study was assessed and approved by the local committee for medical research, MFM-IRB (code: R/16.03.30).

Missing data and statistical analysis

For statistical analysis, we used the Statistical Package for the Social Sciences (SPSS Statistics) for Microsoft Windows (Version 17.0, 2008; SPSS Inc., Chicago, IL, USA). There was an overall 4.9% missed values in collected data, on which multiple imputation procedure was conducted. The missing pattern was completely at random, the missed values were imputed multiply, the analyses were performed, and the results were pooled into one estimate. Descriptive analyses included mean ± standard deviation (SD) for parametric continuous data and mean and interquartile range (IQR) to describe nonparametric continuous data, while numbers and proportions were used for description of categorical data. For better statistical analysis, we used binary outcomes (dead vs. alive and favorable vs. poor). Comparing the outcome groups was done using Mann–Whitney U-test for continuous nonparametric variables and Chi-square test for categorical data. Significance was set at a P < 0.05.

Demographic data

Ages ranged from 55 to 79 years with a mean of 65.31 ± 5.95 years. Among the 82 patients, the majority were females (47 cases, 57.3%). Although history of hypertension was confirmed in only 38 patients, hypertension was diagnosed in 51 cases (62.2%). Diagnosis of hypertension was given for those already on antihypertensive drugs before admission, in addition to those with persistent elevation of systolic (above 165 mmHg) and/or diastolic (above 95 mmHg) pressures requiring continuation of antihypertensive therapy during the whole hospital stay and beyond. The mean was 157 ± 23 mmHg for systolic, 91 ± 10 mmHg for diastolic, and 107 ± 14 for mean blood pressures. Fifty-one cases (62.2%) were rural inhabitants and 68 patients (82.9%) were married.

Clinical presentation

Initial GCS scores ranged from 9 to 13, with a mean of 11.4 ± 1.5. The initial NIHSS ranged from 15 to 24 with a mean of 20.4 ± 2.1, indicating severe impairment. Preoperative GCS scores ranged from 9 to 13 with a mean score of 11.1 ± 1.4, while the corresponding NIHSS scores ranged between 16 and 24 with mean of 21.0 ± 1.9. Neurological deficits detected included hemiparesis and aphasia. Medical history and clinical presentations in patient population are reported in Table 2.

Table 2 Medical history and presentation data of study group

Blood glucose and international normalized ratio

Blood glucose level ranged from 90 to 390 mg/dL (5–22 mmol/L) with a median of 130 mg/dL and IQR of 64 (110–174). About 76% of patients had blood glucose levels below 180 mg/dL and needed no insulin therapy, while the remaining patients (with blood glucose level ≥180 mg/dL) followed a sliding insulin scale. The INR ranged from 1.00 to 2.20 with a mean of 1.24 and SD of 0.29. Those with INR >1.4 needed correction of their coagulation profile before surgical intervention [Table 2].

Hematoma location and size

The majority of hematomas (61%) were located on the left side. The most common location was the temporal lobe, followed by frontal, parietal, and then occipital lobes. This pattern of descending frequency was noticed in hematomas of both sides, as well as on each separate side [Table 3]. The size of hematomas ranged from 30 to 52 cc with a mean of 36.8 ± 5.6, median of 35.5, and IQR of 9. Half of hematomas were ≤35 cc. All hematomas were not <10 mm away from the cortical surface, ranging from 0 to 10 mm with a median of 5 mm and an IQR of 5 mm. Of all lobar hematomas, 35 cases (42.7%) were associated with IVH. The IVH scores ranged from 3 to 8 which corresponds to IVH volumes of 2–5 cc, giving TVs of hemorrhages (ICH volume + IVH volume) ranging between 30 and 57 cc.

Table 3 Sides and locations of lobar hematomas in patient population

Intervention time

The median time from presentation to skin incision was 13 h and 55 min, with IQR of 22 h and 40 min for the whole patient population. The time window ranged from 2 h and 40 min in the earliest operated case to 70 h and 30 min in the most delayed one. Among those patients, an initial decision of surgical evacuation was taken in only 33 cases, while in others, surgical evacuation was done after an initial conservative management. Cases with initial decision of conservative treatment included those in whom conservative management was considered promising by the treating neurologist (11 cases), as well as those in whom an associated comorbidity required an initial management to lessen surgical risk (38 cases). Those 49 cases received the best medical management till their transfer to surgical evacuation. Cases whose medical problems became resolved were directly operated, while those who underwent an initial trial of conservative treatment were referred to surgical intervention due to either a drop in their GCS score (1 point at least), a drop in their NIHSS scores (2 points at least), or no improvement.

Management of cases

All patients received the best medical treatment including osmotherapy, antiepileptic drugs, prophylaxis against deep venous thrombosis, H2 blockers, intravenous fluids, maintenance of normoglycemia, fever control, and early nutritional support. All patients were operated on using an open craniotomy (about 5–7 cm in diameter) through a C-shaped scalp incision centered over the most superficial point of ICH. After performing the cortical incision, effective debulking of the hematoma was carried on till the surrounding brain tissue became grossly lax and the cortical surface dropped away from the dura. Achievement of adequate decompression and hemostasis under illumination and magnification of the operating microscope was done using the usual microsurgical equipment and techniques. This was followed by proper closure of the dura with repositioning of the bone flap. Immediate postoperative CT scans revealed residuals ranging from traces to about 10 cc.

Mortality and outcome

The outcome was assessed using mRS scores. mRS score of six represented dead cases (6, 7.3%) while score 0 represented those with no residual symptoms (30, 36.6%). The last follow up at which outcomes were registered ranged from 2 to 14 months with a mean follow up of 7.5 ± 3 months. About two-thirds of cases were reported to have a favorable outcome at their last follow-up, while the other third of patients including cases of mortality were reported as poor outcome. The final dichotomized outcome was defined based on the initial basal status and predicted prognosis as previously mentioned [Table 4].

Table 4 Outcomes at final follow up in patient population

Exploring variables to intervention groups

Comparing time of intervention among various patient characteristics revealed no significant association related to age, sex, residence, marriage, hypertension, diabetes and associated chronic diseases, smoking, or use of antiplatelet therapy. The initial GCS and NIHSS scores, in addition to, immediate preoperative GCS scores, also showed no significant association with time of intervention. However, immediately preoperative NIHSS scores showed borderline significant (P = 0.056) weak association (ρ = 0.212) with time of intervention, indicating low probability of new deficits development in delayed surgery. Clinical presentation, glucose level, size and depth of the hematomas, as well as the presence of IVH also failed to show significant association with timing of surgery. Rebleeding was not reported before surgical interventions in both groups but detected after surgical evacuation in 12 cases (14.6%). Although the time of surgery was not found to be significantly different among those who bled and those who did not in the whole group, it was significantly shorter among rebleeding group in hematomas ≤35 cc but not in hematomas larger than 35 cc. The INR and presence of associated chest problems were significantly associated with longer time to surgery, indicating their important role in initial decisions of conservative management. Other variables were not affected by or played a role in the decision of timing of surgery [Table 5].

Table 5 Association between patient characteristics and time to surgery

Exploring variables to outcome

Demographic and clinical data

Bivariate analysis of various variables with final outcome (poor vs. favorable) revealed no statistical difference of age, sex, smoking, residence, marriage, associated conditions such as hypertension, diabetes, hepatic, renal and coagulopathy, or use of antiplatelet therapy among outcome groups. Moreover, clinical presentation such as the presenting GCS and NIHSS scores, in addition to neurological deficits, failed to show any significant differences between final outcome groups. On the other hand, serum glucose levels at presentation differed among outcome groups. Glucose levels above 180 mg/dL were significantly more associated with poor outcome [Table 6].

Table 6 Comparing variables according to the binary outcome

Intervention time

Comparing the time to surgery in different mRS score categories, as well as in both outcome groups, revealed no significant difference. Data were then split at the median size into two groups (≤35 cc and >35 cc). There were 41 cases with hematoma size ≤35 cc and another 41 hematomas >35 cc. In hematomas ≤35 cc, the time to surgery was shorter in the poor than favorable group; however, this difference failed to reach the statistical significance. On the other hand, in hematomas larger than 35 cc, the poor outcome group showed statistically significant longer time to surgery when compared to favorable group. Previous results suggest that delayed surgery is relatively a safer approach in hematomas ≤35 cc but a poor prognostic indicator in hematomas >35 cc. Receiver operating characteristic (ROC) curve analysis was done using the dichotomized outcome as the state variable and time to surgery as the test variable for the whole population, as well as for those with hematomas larger than 35 cc. Area under the ROC curve when the whole population was analyzed was 0.542, with standard error (SE) = 0.069 and 95% confidence interval (CI) = 0.407–0.678, indicating that the discriminative power of time to surgery in predicting outcome in the whole group was no better than chance. On the other hand, the ROC curve analysis on the subpopulation with hematomas >35 cc was of significant discriminative value (area = 0.732, SE = 0.079, and 95% CI = 0.577–0.888) [Figure 1]. Youden's J statistic was used to select the optimum cutoff point of time to surgery that can predict poor outcome (cutoff 11 h and 15 min with sensitivity of 88% and specificity of 56%).

Close

Figure 1

Results of receiver operating characteristic curve analysis of time to surgery in outcome groups; in the whole patient population “left” and in clots >35 cc “right.” The area under the receiver operating characteristic curve revealed that the time to surgery has a weak predictive power of outcome among the whole group (area under the receiver operating characteristic = 0.542) but with good predictive power (area under the receiver operating characteristic = 0.732) in clots >35 cc

Export to PPT

Exploring variables to mortality

Six cases died during the follow-up period. The difference in time to surgery between dead and alive was not significant both among the whole population [Table 7] and within each size-split group (Mann–Whitney test, P = 0.133 and 0.120 for clots ≤35 cc and >35 cc, respectively). When ICH volumes and TVs of hemorrhages were compared among dead and alive, we found ICH volumes to have lower mean rank among dead cases; while TV to have higher rank among them, however, both differences were not statistically significant (Mann–Whitney test, P = 0.816, 0.202, respectively). On the other hand, rebleeding after surgical evacuation occurred in 4 (66.7%) out of the six cases of mortality, which was found to be of statistical significance (Fisher's exact test, P = 0.004).

Table 7 Comparing time to surgery in mortality and outcome groups

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.