A comprehensive study of risk factors predicting hydrocephalus following decompressive craniectomy in traumatic brain injuries

Introduction Decompressive craniectomy [DC] is one of the leading armaments to lower refractory intracranial pressure. Post-DC hydrocephalus [PDCH] occurs in 11.9–36% of patients undergoing DCs for TBIs. Various theories have been given regarding pathophysiological mechanism of PDCH but remain dubious. Risk factors predicting PDCH still under research. Exact timeline regarding developmental process of PDCH remains undefined. Method This retrospective study was conducted on 422 patients who underwent DCs in our tertiary care trauma center over the period of one year. 60 patients out of 422 who developed PDCH were analyzed with respect to demographic variables and preoperative and postoperative risk factors. A total of 20 randomly selected patients, who underwent DCs but did not develop hydrocephalus, were selected and compared with patients who developed PDCH. Outcome analysis was done by dichotomizing the groups into independent and dependent groups. Results Among 422 patients undergoing DC, 14.21%[ n = 60] developed PDCH. Younger [34.2 y vs 43.3 y, p = 0.0004] male age group was predominant in our study. Age [ p = 0.021, multivariate analysis] and midline shift [ p = 0.008, multivariate analysis] were significant preoperative predicting risk factors for PDCH. Interhemispheric hygroma [ p = 0.031], brain bulge [ p = 0.008], and blood in postoperative scan [ p = 0.029] were significant postoperative risk factors. Lower GCS score at admission [ p = 0.0003], postoperative day 10 and at the time of establishment of PDCH were significantly predicted surgery to hydrocephalus time. Midline shift [ p = 0.007] and thickness of interhemispheric hygroma [ p = 0.021] were associated with poor outcome in patients with PDCH. Conclusion Younger age group and presence of midline shift are significant preoperative predictors of PDCH. Blood in postoperative scan, interhemispheric hygroma and brain bulge in postoperative period are significant predictors for PDCH. Deterioration in GCS score in postoperative period following DC should be taken as high index of suspicion for developing PDCH.


Introduction
Decompressive craniectomy (DC) is one of the leading armaments to lower refractory intracranial pressure (ICP) in the continuum of severe traumatic brain injuries (TBIs) in present era [1].DC is associated with numerous complications which may arise in sequential manner at specific time span [2].Post-DC Hydrocephalus (PDCH) is a dreaded complication which may be associated with poor outcomes.PDCH can occur in 11.9-36% of patients undergoing DC for TBIs [3].Various theories have been given regarding pathophysiological mechanism of PDCH but remain dubious [4].Cerebrospinal fluid (CSF) malabsorption at arachnoid granulations due to deposition of blood or blood products and alteration in CSF flow due to pressure differences across DC site can lead to development of PDCH [4][5][6].Sometimes associated intraventricular haemorrhage (IVH) may also disrupt CSF circulation by obstructing ventricular pathway leading to PDCH [7].Although no exact timeline has been defined but in most of the literature, PDCH reportedly occurs between few week to months [8].There are various risk factors which have been studied to predict the risk of development of PDCH [8,9].The risk factors identified are age, radiological findings like subarachnoid haemorrhage (SAH), subdural hygroma (SHY), interhemispheric hygroma (IHH), distance of craniectomy margins from midline (MMD) and lower Glasgow Coma Scale (GCS) scores, etc. [8][9][10].
The aim of this study is to describe demographical characteristics of patients developing PDCH and to identify risk factors predicting development of PDCH.Additional objectives achieved in this study were to define timeline for onset of PDCH and analyze outcome of the patients with PDCH.

Study design
This retrospective study was conducted at our tertiary care trauma center in Department of Neurosurgery over the period of one year from December 2021 to November 2022.A total of 564 DCs were performed during this period.Indications of surgery were diffuse cerebral edema with or without midline shift along with subdural or parenchymal lesion.Surgical procedures involved unilateral or bilateral frontotemporoparietal DC.Decision of augmented duraplasty was taken on assessment of intraoperative cerebral edema.Unfortunately 16 patients, in which bilateral DCs were performed, died within 15 days, so were excluded.Apart from those, a total of 106 patients who died within 15 days of surgery, the period we believe is not long enough for development of PDCH, were also excluded from the study.After surgery patients were observed in postoperative ward or critical care unit.Serial computed tomography (CT) scans of head were taken in all patients.Initial demographical variables were recorded like age, gender, mode of injury, consciousness level by GCS at the time of admission.

Radiological evaluation
Radiological variables were recorded as lesions in initial CT scan, i.e. parenchymal contusion/hematoma or subdural hematoma, SAH, IVH, and midline shift (MLS).Findings in postoperative day 1 scans were recorded.The shortest axial distance from medial craniectomy margin to midline [MMD] measured and patients divided into two groups: MMD less than 2.5 cm and more than 2.5 cm.Presence of blood in subdural/subgaleal space was also recorded.Thickness of hematoma more than 5 mm was regarded as 'blood in post-op scan.' Presence of IHH and SHY was noted in serial CT scans, and maximum thickness was recorded.Brain bulge was noted at 7th day; brain bulge was defined as more than 1.5 cm of brain herniation from outer table of cranium on DC side.PDCH was defined as modified frontal horn index (MFHI) score more than 33%, enlargement of temporal horns and periventricular ooze.MFHI calculated by dividing the greatest width of frontal horns by bicortical distance in same plane.On the basis of imaging, PDCH was divided into communicating or obstructive type, i.e. all ventricles seen dilated in communicating type hydrocephalus.

Timeline of PDCH
Interval between DC to establishment of PDCH was noted in days and considered as 'surgery to hydrocephalus time.' Patients were divided into two groups on the basis of surgery to hydrocephalus time:(1) early-patients developing PDCH in < 42 days, and (2)late-patients developing PDCH in > 42 days.
Other additional variables recorded were GCS score at 10th day and at the time of detection of PDCH, side of DC (right sided or left sided), intraoperative augmented duraplasty during DC, hospital stay, ventilator stay, and any form of postoperative intracranial infection.
To compare the patients with PDCH to patients not developing hydrocephalus post-DC, we randomly selected 20 patients who did not develop hydrocephalus at the end of 6 months after DC.Variables recorded in these patients were age, GCS score at admission, side of DC, augmented duraplasty during DC, hospital stay, ventilator stay and GOS at 6 months.Radiological parameters recorded were IVH, SAH, and MLS in initial CT scans, blood in post-operative CT, brain bulge at 7th postoperative day, MMD, IHH and SHY in serial CT scans, and any form of intracranial infection.
Presentation of the categorical variables was done in the form of number and percentage (%).On the other hand, the quantitative data were presented as the means ± standard deviation (SD).
Following statistical tests were applied for the results: Comparison of the variables which were quantitative in nature was analyzed using Independent t test.Comparison of the variables which were qualitative in nature was analyzed using Chi-square test.If any cell had an expected value of less than 5, then Fisher's exact test was used.We divided above-mentioned variables into preoperative and postoperative risk factors that were significant in univariate analysis predicting PDCH and separate multivariate logistic regression model was applied to establish significance for both.Kaplan-Meier curve was plotted to analyze outcome on GOS scale with respect to surgery to hydrocephalus time by dividing into dependent (GOS 1,2,3) and independent ( GOS 4,5) groups.
For statistical significance, p value of less than 0.05 was considered statistically significant.

Results and observations
In a total of 422 patients who were included in the study, 14.21% (n = 60) patients developed PDCH.The baseline demographic, clinical and radiological description is given in Table 1.Mean age of patients developing PDCH was 34.2 ± 11.1 years with male gender preponderance (66.7%,n = 40).Road traffic accidents were commonest mode (66.7%) of injuries.88.53% (n = 53) patients had communicating type hydrocephalus.

Association between PDCH and preoperative demographical, clinical and radiological variables
In our study, age was significant preoperative risk factor for PDCH in both univariate (p = 0.004, independent t test) and multivariate analysis (p = 0.021).GCS score at admission was not a predictor for PDCH (p = 0.167, Independent t test).Intraparenchymal lesion was significantly associated with development of PDCH in univariate analysis (p = 0.013, chi-square test).IVH and SAH did not predict development of PDCH in our study.MLS was a significant predictor of PDCH in our study in both univariate (p = 0.001, Independent t test) and multivariate analysis (p = 0.008).(Tables 2 and 3).MMD (p = 0.028, chi-square test) was significantly associated with PDCH in univariate analysis but not in multivariate analysis.Side of DC (p = 0.014, chi-square test) and intraoperative duraplasty (p = 0.0007, chi-square test) were significantly associated with PDCH in univariate analysis.(Tables 2 and 3).  2 and 4).1).

Discussion
Incidence of hydrocephalus following decompressive craniectomy varies in the range of 11.9%-36% in most recent literatures [3,11].This variation is due to studies with fewer patients, differences in volume of patients requiring DCs, different evaluation criteria for diagnosing PDCH, and lack of resources to access medical facility leading to underdiagnosis in low-to middle-income countries.In our high volume trauma center incidence of PDCH was 14.21%.

Risk factors predicting PDCH
PDCH is more common in younger age group.High energy trauma, higher and compact brain volume attributable to higher incidence of PDCH in younger age group [3,12].Poor GCS score in preoperative period is highly associated with PDCH [13][14][15].Although we did not find any predictive value of preoperative GCS score for PDCH.Several preoperative and postoperative radiological criteria used to predict the risk of PDCH.Preoperative risk factors include presence of parenchymal lesion,  IVH, SAH, and extent of MLS [3,8,14,16,17].Among these MLS, indicator of severity of brain injury was a significant risk factor for PDCH in our study.Superior sagittal sinus is usually deviated to right side of midline [18].There are higher chances of injury to draining venous sinuses in right sided DC which may be a causative factor in failure of CSF circulation [19].MMD and its relation with PDCH are controversial [17,20,21].Extending craniectomy margins to medially close to sinuses can lead to disruption of draining venous channels, which may be responsible for CSF malabsorption ( Fig. 2).In our study, right sided DCs and MMD were statistically insignificant on multivariate analysis.Among postoperative criteria, brain bulge, IHH and SHY are risk factors under research for predicting PDCH.IHH found in 32% -85% of patients developing PDCH.External brain bulge is seen in first week of DC which usually precedes IHH [22].After DC, pressure on brain parenchyma reverses and creates suctioning effect over interhemispheric fissure which leads to expansion of fissure and makes it amenable to fluid collection (Fig. 3).Brain bulge and IHH both are significant predictor of PDCH [3,9,22,23].Role of augmented duraplasty in DCs is controversial.In some literatures, intraoperative augmented duraplasty has shown   favorable results in preventing complications like brain herniation, hydrocephalus, subdural collections, etc. [24,25].SHY is associated with 28.4%-49% of patients undergoing DCs [9,26].Arachnoid rupture and blood brain barrier failure due to brain injury lead to accumulation of fluid in subdural space [26] (Fig. 4).It is role in predicting PDCH is dubious [9,26,27].In our study, it was statistically insignificant in multivariate analysis.
We included a new criterion labeled as blood in postop scan in our study which was significant predictor of PDCH in multivariate analysis.Blood and blood products present in subdural and subgaleal space travel to draining channels and may block absorption of CSF [28,29] which may lead to PDCH(Fig.5).

Timeline for establishment of PDCH
PDCH usually occurs between three weeks to months post-DCs.[2,30,31].In our study mean surgery to hydrocephalus time was 47.65 days.Factors predicting surgery to hydrocephalus time are not much elaborated in literature.Although in our study patients with poor GCS at presentation, 10th day and at the time of establishment of PDCH had lower surgery to hydrocephalus time.

Outcome analysis of patients with PDCH
Patient with poor GCS score at arrival usually has unfavorable long-term outcome in TBI.In our study, GCS score on arrival was not contributory to poor outcome in patients of PDCH.Development of process of PDCH can lead to deterioration of GCS score in postoperative period which may have significant impact on outcome.In our study, GCS score at 10th day and at the establishment of PDCH was significantly associated with outcome of the patients.Midline shift, a significant risk factor in predicting PDCH, is also invariably associated with outcome of the patients [32].

Limitations
This study is not without limitations.Retrospective nature of study with a small sample size prone to bias.It is very much possible that prospective study with a larger sample size could have produced more significant results.Differences in surgical techniques between surgeons with respect to DCs may confound the results.Nonsignificant correlation of preoperative GCS score with PDCH was statistical limitation unlike other studies.Various other factors like size of DCs, thickness of subdural hematoma, etc., which may affect the process of PDCH, were not taken into consideration.

Conclusion
PDCH is a common complication following DCs.Younger age group and presence of midline shift are significant preoperative predictors of PDCH.IHH and brain bulge in postoperative period are significant postoperative predictors of PDCH.Presence of blood in subdural or subgaleal space in postoperative scan is significant risk factor for PDCH.Deterioration in GCS score in postoperative period should be taken as high index of suspicion for developing PDCH.Results from this study should add valuable information in literature regarding knowledge on PDCH.

Fig. 1 Fig. 2 (Fig. 3 (
Fig. 1 Kaplan-Meier survival analysis curve of Glasgow Outcome Scale (GOS) at 6 months with respect to surgery to hydrocephalus time

Fig. 4 (
Fig. 4 (A) Preoperative axial CT scan of head of a patient with traumatic brain injury showing diffuse cerebral edema with left sided parietal contusions( thin arrow) with Subarachnoid Haemorrhagee(thick arrow).(B)At postoperative day 16, axial CT scan of head of same patient showing Interhemispheric hygroma (thin arrow) with left sided Subdural hygroma (thick arrow).(C) At postoperative day 46, axial CT scan of head of same patient showing PDCH (thin arrow) (CT: Computed Tomography, PDCH: Post-decompressive craniectomy hydrocephalus)

Table 2
Risk factors predicting development of PDCH ‡ Independent t test, * Fisher's exact test, † Chi-square test GCS: Glasgow Coma Scale, GOS: Glasgow Outcome Scale respectively, independent t test)( Table6).Surgery to hydrocephalus time was plotted in relation to GOS (dependent and independent) in Kaplan-Meier curve.Patients with longer surgery to hydrocephalus time had good outcome but statistically insignificant on long rank test (p = 0.188).(Table

Table 3
Preoperative risk factors predicting development of Post-decompressive craniectomy hydrocephalus (Multivariate analysis)

Table 4
Post-operative risk factors predicting development of Post-decompressive craniectomy hydrocephalus (Multivariate analysis)

Table 5
Association of parameters with surgery to hydrocephalus time ‡ Independent t test, * Fisher's exact test, † Chi-square test

Table 6
Outcome analysis ‡ Independent t test, * Fisher's exact test, † Chi-square test

Table 7
Kaplan-Meier survival analysis of GOS at 6 months for with respect to surgery to hydrocephalus time