Post-traumatic hydrocephalus – incidence, risk factors, treatment, and clinical outcome

Abstract

Objectives

Post-traumatic hydrocephalus (PTH) is well-known after traumatic brain injury (TBI), but there is limited evidence regarding patient selection for ventriculo-peritoneal (VP)-shunt treatment. In this study, we investigated the incidence and risk factors for PTH and the indication for and outcome after shunt treatment.

Materials and methods

In this retrospective study, 836 TBI patients, treated at our neurointensive care (NIC) unit at Uppsala university hospital, Sweden, between 2008 and 2018, were included. Demography, admission status, radiology, treatments, and outcome variables were evaluated.

Results

Post-traumatic ventriculomegaly occurred in 46% of all patients at NIC discharge. Twenty-nine (3.5%) patients received a VP-shunt. Lower GCS M at admission, greater amount of subarachnoid hemorrhage, meningitis, decompressive craniectomy (DC), and ventriculomegaly at NIC discharge were risk factors for receiving a VP-shunt. Fourteen of the PTH patients showed impeded recovery or low-pressure hydrocephalus symptoms, of whom 13 experienced subjective clinical improvement after shunt treatment. Five PTH patients showed deterioration in consciousness, of whom four improved following shunt treatment. Five DC patients received a shunt due to subdural hygromas (n =2) or external brain herniation (n = 3), of whom two patients improved following treatment. Five patients were vegetative with concurrent ventriculomegaly and these patients did not have any positive shunt response. Altogether, 19 (66%) PTH patients improved after shunt surgery.

Conclusion

Post-traumatic ventriculomegaly was common, but few developed symptomatic PTH and received a VP-shunt. Patients with low-pressure hydrocephalus symptoms had the best shunt response, whereas patients with suspected vegetative state exhibited a minimal shunt response.

Keywords:

• Neurointensive care
• post-traumatic hydrocephalus
• traumatic brain injury
• ventriculo-peritoneal shunt

Introduction

Traumatic brain injury is a leading cause of mortality and morbidity in young adults.¹ Post-traumatic hydrocephalus (PTH) after the acute phase is estimated to occur at an incidence between 0.7–45%, depending on the diagnostic criteria and patient population.^2–7 The etiology of PTH is not fully understood, but previous studies have highlighted several risk factors including higher age, lower Glasgow Coma Scale (GCS) score at admission, intraventricular/subarachnoid hemorrhage (IVH/SAH), decompressive craniectomy (DC), and meningitis with positive cerebrospinal fluid (CSF) culture.^3–11 PTH may have different clinical presentations such as impeded neurological recovery, low-pressure hydrocephalus-symptoms (gait disturbance, urinary incontinence, and cognitive decline), or high-pressure hydrocephalus-symptoms (reduction in consciousness and papilledema).² The treatment of choice for PTH is CSF diversion, usually by means of a ventriculo-peritoneal (VP)-shunt. In some of these patients, shunt surgery is associated with substantial clinical improvement,^6,7,12,13 particularly if done early (the first months after TBI).⁶

There is no gold standard methodology for the PTH diagnosis, but it is often based on a combination of the symptoms/clinical states outlined above in association with venticulomegaly. However, it may both be difficult to identify PTH symptoms in poor-grade patients and to differentiate post-traumatic ventriculomegaly due to PTH from brain atrophy (hydrocephalus “ex vacuo”).

In order to improve the diagnosis of PTH and the selection of patients for shunt surgery, 836 TBI patients treated at our NIC unit between 2008 and 2018 were studied, The aim of the study was to better characterize the incidence of PTH and shunt surgery, to identify explanatory variables, and to study clinical outcome and prognostic factors for those patients who received a VP-shunt due to PTH.

Materials and methods

Patients

The Department of Neurosurgery at the University Hospital in Uppsala provides neurosurgical care for a central part of Sweden. There were 926 TBI patients aged 15 or older who were treated at our NIC unit between 2008 and 2018 and who were eligible for inclusion in this study. Seventy-five patients were excluded because they had been treated at another NIC in Sweden or abroad before admission to Uppsala or were discharged to a different catchment area/country with another neurosurgery department responsible for the follow-up. Seven patients were treated twice for TBI at our NIC and were therefore registered twice, but we only included data from their first NIC visit in this study. Six patients who had a VP-shunt due to hydrocephalus prior to the TBI and admission to our NIC were also excluded. Two patients were excluded due to chronic hydrocephalus that was discovered on the computed tomography (CT) scan at admission and was not considered to be a consequence of TBI (one with normal-pressure hydrocephalus and one with aqueductal stenosis). Hence, the final study population included 836 TBI patients.

Neurointensive care management protocol

During NIC, patients were treated in accordance with our standardized intracranial pressure (ICP)- and cerebral perfusion pressure (CPP)-oriented treatment protocol.^14,15 Treatment goals were ICP ≤ 20 mm Hg, CPP ≥ 60 mm Hg, systolic blood pressure > 100 mm Hg, central venous pressure 0–5 mm Hg, pO₂ > 12 kPa, arterial glucose 5–10 mmol/L (mM), hemoglobin (Hgb) > 100 g/L, electrolytes within normal ranges, normovolemia, and body temperature < 38 °C. Patients were initially mildly hyperventilated (4.0–4.5 kPa) and normoventilated as soon as ICP allowed.

All unconscious (GCS M 1–5) patients were intubated and sedated with propofol infusion and received morphine for analgesia. ICP was monitored in all unconscious patients. An external ventricular catheter (EVD; HanniSet, Xtrans, Smith Medical GmbH, Glasbrunn, Germany) was the first choice of ICP monitor, but an intraparenchymal sensor device (Codman ICP Micro-Sensor, Codman & Shurtleff, Raynham, MA) was used in case of compressed ventricles. If the Codman monitor was used, the EVD was added if there were ICP problems and the ventricles had enlarged. CSF was then drained if ICP was high and there was no mass lesion. VP-shunts were not inserted in the acute phase. Intracranial lesions with significant mass effect were surgically evacuated. In case of high ICP with simultaneous high blood pressure and tachycardia, stress was treated with increased sedation, pain relief, β₁-antagonist infusion, and repeated injections of α₂-agonists. If ICP still remained elevated, a thiopental infusion was started, and finally, if high intracranial hypertension was still refractory, a decompressive craniectomy (DC) was performed.

Ventriculo-peritoneal-shunt practice

After the acute phase in the NIC unit, the level of care was gradually stepped down to care in general surgical-medical wards and rehabilitation units. In case the patients exhibited any symptoms and signs of PTH, including impeded improvement, secondary deterioration, gait disturbance, urinary incontinence, cognitive decline, or external brain herniation after DC, a CT scan of the brain was performed, and the Department of Neurosurgery was consulted regarding VP-shunt treatment. The Hydrocephalus section of the Department of Neurosurgery then determined if the patient was considered to have symptomatic PTH and would require a VP-shunt. It was avoided to insert a shunt early after the trauma if possible to avoid shunt infections and with the hope of spontaneous improvement and to avoid overtreatment with shunts. In cases with less severe symptoms, the clinical course was therefore followed together with evaluation of repeated CT scans. In case of more severe neurological symptoms and substantial ventriculomegaly and/or rapidly increasing hydrocephalus, a shunt was inserted without further delay. Evaluation of CSF pressure dynamics prior to shunt-treatment was not mandatory (occurred in 2/29 cases). The type of shunt valve varied over the study period among Codman-Hakim, Codman Certa, proGAV Miethke, and Strata, depending on the time period and the individual decision by the neurosurgeon.

Demography, admission status, treatment variables, and complications

Demographic data, admission status, treatment variables, and complications were extracted from the Uppsala TBI register¹⁶ and medical records.

Radiological measures

The first CT scan immediately after the trauma was analysed according to the Marshall score¹⁷ and Fisher scale.¹⁸ We used the latter scale to determine the association between the amount of traumatic SAH and the development of PTH.³ The degree of ventricular dilation was estimated with the Evans’ index¹⁹ and the modified frontal horn index (mFHI)²⁰ at NIC discharge and before/after shunt surgery. Evans’ index > 0.3 and mFHI > 0.33 were considered as significant ventriculomegaly, respectively.

Outcome

Clinical outcome was assessed by specially trained personnel with structured telephone interviews at 6 months post-injury using the Extended Glasgow Outcome Scale (GOS-E), containing eight categories of global outcome, from death to upper good recovery.^21,22 The interviews were held with the patients if they had recovered sufficiently, otherwise with their next of kin. The modified Rankin Scale (mRS)²³ was estimated based on notes on medical records from the admission before VP-shunt surgery and the follow-up visit that usually took place 3–12 months after surgery. Subjective improvement of symptoms (yes/no) were also evaluated based on medical records from the follow-up after shunt surgery.

Statistical analysis

The statistical analyses were performed in SPSS version 25 (IBM Corp, Armonk, NY, USA). Descriptive data were presented as medians (interquartile range (IQR)) or as number (n) and proportion (%) of patients. Potential risk factors for PTH/VP-shunt based on demography, admission status, radiology, and, treatments, were evaluated with univariate analyses, including Mann-Whitney U-test or Fisher’s exact test. We were unable to proceed with multiple logistic regression analysis regarding risk factors for PTH/VP-shunt, due to the limited number of shunt patients (dependent variable). The association between PTH and VP-shunt in relation to mortality and favourable clinical outcome (GOS-E 5–8) at 6 months was evaluated with both univariate analyses (Mann-Whitney U-test or Fisher’s exact test) and multiple logistic regression analysis. The multiple logistic regression analyses included VP-shunt due to PTH (yes/no) in addition to the IMPACT core variables age, GCS M, and pupillary status as the explanatory variables.²⁴ Those with missing values were excluded from the statistical analyses. A p-value < 0.05 was considered statistically significant.

Results

Demography, admission status, treatments, ventricular size at neurointensive care discharge, and clinical outcome in the entire population

Descriptive data of the entire TBI patient population are presented in Table 1. Median age for all patients was 54 (IQR 34–67) years and the male/female ratio was 640/196 (77%/23%). Median GCS M at admission was 5 (IQR 5–6) and 128 (15%) had abnormal pupillary findings. ICP was monitored in 512 (61%) patients of which 210 (25%) patients were monitored with an EVD, 373 (45%) patients were operated with a craniotomy, 65 (8%) patients were treated with thiopental, and 57 (7%) patients with DC. CT at discharge from NIC was available in 805/836 patients. Twenty-three percent of the entire TBI population had an Evans’ index above 0.30 and 46% an mFHI above 0.33 at NIC discharge. At 6 months, the rate of favourable outcome (GOS-E 5–8) was 60% and the mortality rate 17%.

Table 1. Demography, admission status, radiology, treatments, and clinical outcome – relation to shunt vs no shunt.

Variables	All patients	VP-shunt	No VP-shunt	p Value
Patients, n (%)	836 (100)	29 (3.5)	807 (96.5)	NA
Age (years), median (IQR)	54 (34–67)	55 (45–63)	54 (33–67)	0.97
Sex (male), n (%)	640 (77%)	25 (86%)	615 (76%)	0.27
GCS M at admission, median (IQR)	5 (5–6)	5 (3–5)	5 (5–6)	0.005
Pupillary abnormality (yes), n (%)	128 (15%)	4 (14%)	124 (15%)	1.0
Marshall score, median (IQR)	2 (2–5)	2 (2–5)	2 (2–5)	0.97
CT Fisher, median (IQR)	2 (1–2)	2 (2–3)	2 (1–2)	0.04
EVD (yes), n (%)	210 (25%)	16 (55%)	194 (24%)	0.001
Meningitis yes, n (%)	17 (2%)	5 (17%)	12 (1%)	0.0001
Craniotomy (yes), n (%)	373 (45%)	15 (52%)	358 (44%)	0.45
Thiopental (yes), n (%)	64 (8%)	7 (32%)	57 (7%)	0.004
DC (yes), n (%)	57 (7%)	8 (28%)	49 (6%)	0.0001
Evans’ index at discharge, median (IQR)	0.27 (0.25–0.30)	0.30 (0.27–0.35)	0.27 (0.25–0.30)	0.001
Modified frontal horn index at discharge, median (IQR)	0.32 (0.30–0.36)	0.35 (0.32–0.40)	0.32 (0.30–0.36)	0.003
Mortality at 6 months, n (%)	131 (17)	2 (7)	129 (17)	0.15
GOS-E at 6 months, median (IQR)	6 (3–7)	3 (3–3)	6 (3–8)	0.001
Favourable/unfavourable outcome (GOS-E 5–8/1-4), n (%)	461/310 (60/40)	5/24 (17/83)	456/286 (61/39)	0.0001

IQR: interquartile range; GCS M: Glasgow Coma Scale Motor score; CT: computed tomography; EVD: external ventricular drain; DC: decompressive craniectomy; GOS-E: Glasgow Outcome Scale-extended.

Ventriculo-peritoneal-shunt surgery

In total, 29 (3.5%) out of 836 TBI patients received a VP-shunt due to PTH. Time to shunt treatment was in median 5 (range 1–85) months post-TBI and 80% received the shunt within the first year. Eight (28%) patients required at least one shunt revision.

Factors associated to ventriculo-peritoneal-shunt surgery due to post-traumatic hydrocephalus

There was no difference in age or sex between those with PTH/VP-shunt and the other TBI patients (Table 1). VP-shunt patients had lower GCS M (median 5 IQR 3–5 vs. 5 IQR 5–6, p-value = 0.005) and higher CT Fisher grade (median 2 IQR 2–3 vs. 2 IQR 1–2, p-value 0.04) at admission. The VP-shunt group was more often treated with EVD (16 (55%) vs. 194 (24%), p-value = 0.001), thiopental (7 (32%) vs. 57 (7%), p-value = 0.004), and DC (8 (28%) vs. 49 (6%), p-value = 0.0001). Bacterial meningitis during NIC was more common for patients with PTH treated with VP-shunt (5 (17%) vs. 12 (1%), p-value = 0.0001). Those with PTH treated with VP-shunt had significantly higher Evans’ index (0.30 IQR 0.27–0.35 vs. 0.27 IQR 0.25–0.30, p-value = 0.001) and mFHI (0.35 IQR 0.32–0.40 vs. 0.32 IQR 0.30–0.36, p-value = 0.003) at NIC discharge .

Chariacteristics of patients treated with ventriculo-peritoneal-shunt

All patients treated with a VP-shunt had ventriculomegaly (Evans’ index > 0.30 or mFHI > 0.33), but the symptoms that led to surgery differed among the patients. Eight (28%) patients received a VP-shunt due to impeded neurological recovery (after initial improvement), six (21%) patients because of development of gait disturbance, urinary incontinence, or cognitive decline, five (17%) patients deteriorated in consciousness due to PTH, two (7%) patients had symptomatic subdural hygromas, three (10%) patients due to external brain herniation after DC, and five (17%) patients who were vegetative with some concurrent ventriculomegaly.

Decompressive craniectomy had been done in 8 (28%) out of the 29 VP-shunt patients due to intracranial hypertension. Out of these 8 DC patients, three patients received VP-shunt prior to cranioplasty, due to symptomatic SDG (n = 2) or deterioration in consciousness (n = 1) due to PTH. Three patients were operated with VP-shunt and cranioplasty at the same occasion, of which all three had significant external brain herniation out of the craniectomy defect. Two patients were first operated with cranioplasty and only later with VP-shunt, because of poor neurological recovery/vegetative with some concurrent ventriculomegaly.

Ventriculo-peritoneal-shunt response

In general, the ventricular indexes did not change following shunt surgery – Median Evans’ index went from 0.37 (IQR 0.34–0.43) to 0.38 (IQR 0.34–0.42; p-value 0.19) and the mFHI went from 0.46 (IQR 0.39–0.50) to 0.45 (IQR 0.39–0.49; p-value = 0.56). Figure 1 illustrates the CT scan in one patient case before and after shunt surgery.

Figure 1. Computed tomography scan before and after ventriculo-peritoneal-shunt surgery in one patient with post-traumatic hydrocephalus. The figure demonstrates the computed tomography before and after ventriculo-peritoneal shunt surgery in one patient with post-traumatic hydrocephalus. The width of the lateral ventricles and temporal horns decreased after shunt surgery. Similarly, the transependymal edema became less pronounced and the convexity sulci more open. The patient experienced subjective clinical improvement following surgery.

The mRS before shunt treatment was in median 5 (IQR 4–5) and showed a slight non-significant improvement at the follow-up visit after shunt-treatment with a median mRS at 4 (IQR 3–5; p-value 0.10). Eight (26%) patients improved in mRS, 3 (10%) deteriorated and 18 (64%) remained at the same level.

Subjective improvement (based on description in medical records at follow-up) was reported in 19/29 (66%) after shunt surgery. Seven out of the eight (88%) patients operated due to impeded neurological recovery had subjective improvement of symptoms (Figure 2), but only two patients improved in mRS (1 and 2 p, respectively), whereas one deteriorated with 1 p. All 6 (100%) patients with low-pressure hydrocephalus symptoms had subjective improvement after VP-shunt, of which three improved in mRS (two with 1 point and one with 2 points). Four out of five (80%) patients with deterioration in consciousness had some subjective improvement; one improved in mRS (with 4 points) and one died within a year. One out of three (33%) patients with external brain herniation out of the craniectomy had subjective improvement after VP-shunt insertion, but none improved in mRS. One out of two (50%) patients with SDGs improved both subjectively and with 1 point in mRS. None of the five (0%) vegetative patients improved subjectively or in mRS; one died within one year after VP-shunt surgery. Patients with subjective improvement of symptoms did not have more pronounced improvement in Evans’ index (-0.01 ± 0.03 vs. −0.02 ± 0.04, p-value = 0.28) or mFHI (-0.00 ± 0.04 vs. −0.01 ± 0.05, p-value = 0.59) following VP-shunt-treatment.

Figure 2. Shunt response in relation to shunt indication. The figure indicates the rate of subjective clinical improvement (positive response) after VP-shunt treatment. Six out of six patients with symptoms reminding of low-pressure hydrocephalus (HC), 7/8 patients in the group with impeded/delayed neurological recovery, 4/5 patients in the high-pressure HC (reduction in consciousness) group, 1/2 patients in the subdural hygroma group, 1/3 patients in the external brain herniation group, and 0/5 patients in the vegetative group experienced a clinical improvement after ventriculo-peritoneal (VP)-shunt surgery.

Post-traumatic hydrocephalus treated with ventriculo-peritoneal-shunt – relation to clinical outcome

Patients receiving shunts had worse clinical outcome (GOS-E) with a lower rate of favourable clinical outcome at 6 months post-injury (5 (17%) vs. 456 (61%), p-value = 0.001) compared to the other patients (Table 1). Multivariate analysis with favourable outcome as dependant variable showed that having a VP-shunt due to PTH was an independent risk factor for a lower rate of favourable outcome (odds ratio (95% confidence interval) = 0.13 (0.04–0.36), p-value < 0.001), after adjustment for age, GCS M, and pupillary status at admission (Table 2). There was no association between PTH/VP-shunt and mortality after 6 months.

Table 2. Prediction of mortality and favourable clinical outcome at 6 months post-TBI – multiple logistic regression analyses.

Variables	Mortality		Favourable clinical outcome
	OR (95%CI)	p Value	OR (95%CI)	p Value
Age (years)	1.06 (1.05–1.08)	<0.001	0.96 (0.95–0.97)	<0.001
GCS M (1–6)	0.57 (0.48–0.67)	<0.001	1.84 (1.55–2.19)	<0.001
Pupillary abnormality (Yes)	2.79 (1.59–4.91)	<0.001	0.35 (0.21–0.59)	<0.001
VP-shunt (Yes)	0.24 (0.05–1.11)	0.07	0.13 (0.04–0.36)	<0.001

Regression 1 – Mortality, n = 771. Nagelkerke R² = 0.35. Regression 2 – Favourable clinical outcome, n = 771. Nagelkerke R² = 0.35. Sixty-five patients with missing outcome data (none in the shunt-group).

Discussion

In this study including 836 TBI patients treated at our NIC unit, 46% exhibited post-traumatic ventriculomegaly at discharge, but only 3.5% developed PTH judged to require VP-shunt treatment. Poor neurological status at NIC admission, greater amount of tSAH, EVD monitoring, meningitis, DC surgery, and ventriculomegaly at NIC discharge were factors related to PTH/VP-shunt operation. Most PTH patients had some subjective improvement following VP-shunt surgery, but few had any improvement in their everyday life according to mRS. Patients with symptoms reminding of low-pressure hydrocephalus had the best shunt response and patients in a poor neurological condition with suspected vegetative state exhibited a minimal shunt response. Considering the limited objective shunt response in a large proportion of the patients and the general risk of shunt complications, improved diagnostic evaluation for better patient selection for surgery is desirable.

Post-traumatic ventriculomegaly and post-traumatic hydrocephalus

We found post-traumatic ventriculomegaly at discharge from our NIC in 46% according to mFHI definition. This is in line with most earlier studies.^4,25 However, post-traumatic ventriculomegaly is not equal to PTH and not solely an indication of VP-shunting.

Post-traumatic ventriculomegaly may be caused by a disturbance in the CSF circulation (true PTH), secondary ventricular dilation due to brain atrophy (hydrocephalus ex vacuo), and a combination of the two mechanisms. The overall radiological picture may give some information regarding the presence of PTH, but the clinical picture is essential for the diagnosis. PTH may have several different clinical presentations, such as those with low-pressure hydrocephalus who develop gait disturbance, urinary incontinence, and cognitive decline or those with high-pressure hydrocephalus presenting with a reduction in consciousness and papilledema.² It may be difficult to differentiate between the neurological sequelae following severe TBI from the symptoms caused by PTH and this difficulty may also delay the detection of PTH.

Some previous studies have reported that 5–23% of their TBI patients required shunt treatment due to PTH,^4–6,8,26 but the study populations and indication for treatment varied between the studies. Those with a lower estimated incidence (5–9%) of shunted PTH patients evaluated the entire TBI populations in the NIC or the rehabilitation unit, and the indication for treatment was based on a combination of neurological symptoms, radiological images, and sometimes CSF dynamics.^4,6,26 The higher incidence (23%) was based on a group of patients with EVD, in which those who required persistent CSF drainage received shunt surgery.⁸ However, in a combined evaluation of the radiological degree of ventriculomegaly and CSF pressure dynamics in severe TBI cases, Marmarou et al. estimated that 20% of the survivors developed either low- or high-pressure PTH.¹⁰ Our conclusion, based on the variable reports on the incidence of PTH and the indication for shunt surgery, is that there is a need to develop better guidelines for this condition.

Factors associated with post-traumatic hydrocephalus and ventriculo-peritoneal-shunt treatment

The underlying mechanisms for PTH involve altered CSF reabsorption e.g., due to fibrosis of the arachnoid granulations, but also obstruction of the CSF circulation in the ventricles and/or the subarachnoid space. Several factors such as patient characteristics, TBI subtype, and NIC treatments contribute to the development of these disturbances. Particularly, previous studies have highlighted older age, lower GCS M, presence of tSAH, meningitis with positive CSF culture, and DC as risk factors.^3–6,8,10 Our results were to a great degree in line with previous findings.

Meningeal fibrosis increases with age and older patients may be more susceptible to symptomatic CSF outflow obstructions following acute brain injury.³ Higher age has also been associated with increased risk of PTH and VP-shunt treatment in some,^3–5,8 but not in all studies.^6,25 We did not find any association between age and the risk of needing a VP-shunt due to PTH in the current study. However, it is possible that the threshold for shunt surgery was lower for younger patients with poor neurological status and because of this possible selection bias for treatment, we cannot exclude an association between higher age and PTH.

Lower GCS M has been associated with higher risk of VP-shunt treatment previously^5,6 and the same association was found in the current study. This might reflect that lower GCS M indicates a more severe structural brain injury with a corresponding development of disturbances in the CSF circulation. However, lower GCS M might also reflect a more severe brain injury with more pronounced Wallerian degeneration with secondary ventriculomegaly.

It has been hypothesized that tSAH and IVH induce scarring in the ventricular system and arachnoid granulations that precipitate for hydrocephalus, similar to the risk of post-hemorrhagic hydrocephalus following aneurysmal SAH, which is also higher in case of greater amount of SAH.²⁷ Some studies support similar associations between tSAH and the risk of PTH,^3,10 but not in other studies.^6,25 In this study, we found a weak association between higher CT Fisher grade and having a VP-shunt due to PTH. Poca et al. suggested that SAH is a relatively common post-traumatic finding and it may be difficult to grade the amount of SAH after TBI, which could explain the mixed results in the literature.²⁵

Meningitis may lead to scarring in the subarachnoid space with a corresponding reduction in CSF outflow.²⁸ In line with this hypothesis and with previous studies,⁸ we found that meningitis with positive CSF culture was associated with having a VP-shunt due to PTH. EVD monitoring was also associated with PTH, but this is more likely explained by EVD-associated meningitis.

In line with our findings, several previous studies have demonstrated an association between DC and PTH,^4,6,29 possibly as opening of the cranial vault might reduce CSF outflow. Furthermore, we found that patients with higher Evans’ index/mFHI at discharge were more likely to develop PTH that required VP-shunt treatment. Clinicians should hence raise suspicion for the risk of PTH even in case of mild ventriculomegaly (Evans’ index at discharge – median value 0.30 IQR 0.27–0.35 for those that required a shunt).

Response to ventriculo-peritoneal-shunt treatment

A VP-shunt is the treatment of choice in PTH. However, there is generally a high incidence of shunt-related complications³⁰ and appropriate patient selection for treatment is therefore crucial. Overall we found hat 66% had some subjective improvement of symptoms following VP-shunt treatment but no significant objective improvement in mRS. Previous studies have reported clinical improvements (as evaluated with GOS or Functional Independence Measure (FIM)), in 52–78% of the patients after shunt surgery in PTH.^12,13,26 The differences in outcome is probably related to differences in outcome measures and patient selection for shunt surgery.

We distinguished among six clinical PTH states that led to indication for VP-shunt-treatment, i.e., impeded neurological recovery after initial improvements, low-pressure hydrocephalus, high-pressure hydrocephalus, external brain herniation after DC, symptomatic SDGs, and poor clinical/vegetative state. Although most of our PTH patients had poor clinical status (high mRS) before treatment, those with impeded neurological recovery, low-pressure hydrocephalus, and high-pressure hydrocephalus had the greatest improvement following shunt-treatment (Figure 2). Thus, the clinical PTH picture may be of help in the selection of patients for VP-shunt surgery.

Six DC patients received a VP-shunt prior to or at the same time as the cranioplasty. However, it would have been interesting to determine if these PTH manifestations would have persisted following cranioplasty. The risk of complications may be higher if shunt surgery and cranioplasty are performed at the same time³¹ and, if possible, it would be advantageous to perform cranioplasty first and later determine if the patient still suffers from PTH. Furthermore, we found that none of the five vegetative patients responded to shunt surgery. These results are in line with previous outcome studies^13,26 and indicate that brain injury may have been too severe or that the PTH diagnosis was incorrect.

Since complications are common following shunt surgery (28% required shunt revision in our study), it is important to improve patient selection. Particularly, in poor-grade patients or in case of atypical symptoms with suspected PTH,¹² additional diagnostic tools should be considered. Studies on CSF pressure dynamics may help in determining if the patient has high-pressure hydrocephalus, low-pressure hydrocephalus (low ICP, but high CSF outflow resistance), or atrophy (low ICP and low CSF outflow resistance). Although the CSF pressure dynamics in the PTH group is not fully elucidated, some studies have found promising results to identify candidates that could benefit from shunt surgery based on e.g., the rate of CSF outflow resistance.^10,32,33 MRI diagnostics of CSF flow in the cerebral aqueduct could also be useful for the PTH diagnosis and in the prediction of shunt-response.³⁴

Limitations

In the absence of gold standard diagnostics for PTH, we defined PTH based on if the patient was treated with a VP-shunt. This selection bias probably led to some “false negatives”, i.e., PTH patients who were not shunted because they were not clinically identified or for some reason not considered to benefit from a shunt. There could also be some “false positives”, i.e., patients treated with a shunt despite a low probability of PTH in the hope for a better neurological recovery.

The evaluation of ventricular size is complex. Evans’ index and mFHI are crude, since the ventricles may sometimes be relatively wide in the axial plane, but still relatively thin, generating a high Evans’index/mFHI, despite a relatively small ventricular volume.¹⁹

The radiological evaluation was done by only one author as it was very time-consuming due to the vast number of patients and images. This might limit the reliability of the results.

Differences in admission criteria for the NIC unit and management protocol among centres may affect the incidence of PTH, which limits the external validity of our findings to some degree.

Conclusions

Although post-traumatic ventriculomegaly is relatively common, few (3.5%) patients were considered to require a ventriculo-peritoneal-shunt due to post-traumatic hydrocephalus. Those with symptoms such as impeded neurological recovery, low-, or high-pressure hydrocephalus symptoms had the greatest clinical improvement following shunt treatment. However, patients in a poor neurological condition with concurrent ventricular dilatation did not respond to shunt treatment. Considering the limited clinical improvement and the general risk of shunt complications, these patients, in particular, could benefit from additional diagnostic assessments with cerebrospinal fluid pressure dynamics to improve patient selection for shunt treatment.

Ethical approval

All procedures performed in the studies involving humans were in accordance with the ethical standards of the national research committee and with the 1964 Helsinki declaration and its later amendments. The study was approved by Uppsala University Regional Ethical Board. Informed consent was obtained by the next of kin.

This work was conducted at the Uppsala University Hospital, Department of Neurosurgery.

Acknowledgement

The authors express our gratitude to the NIC staff for attentive patient care. The study was funded by the Department of Neuroscience, Uppsala University.

Disclosure statement

The authors declare no conflict of interest.

Additional information

Funding

The study was funded by the Department of Neuroscience, Uppsala University.