Surgical approaches to cerebral decompression: A meta-analysis of hinge craniotomy and decompressive craniectomy

INTRODUCTION

Intracranial hypertension due to various causes, such as traumatic brain injury (TBI), ischemic stroke, or hemorrhage, represents a major determinant of poor neurological outcomes and death. Decompressive craniectomy (DC) has long been the cornerstone surgical intervention for refractory intracranial pressure (ICP), supported by pivotal trials such as randomised evaluation of surgery with craniectomy for uncontrollable elevation of ICP (RESCUE-ICP), which showed that DC reduces ICP and mortality but often at the cost of severe disability.^[1] However, the complication profile of DC and the need for cranioplasty have spurred interest in alternative techniques.

Hinge craniotomy (HC) offers a surgical strategy that allows cerebral expansion while preserving the bone flap. This approach may preserve more normal cranial dynamics, reduce complications, and eliminate the need for a second surgery for cranioplasty.^[2,3] As international practice evolves, surveys suggest increasing adoption of HC across neurosurgical centers, especially in settings where resource constraints demand low-cost, efficient solutions.^[4]

Despite its growing use, high-quality comparative evidence for HC versus DC remains limited. Our meta-analysis synthesizes current literature to evaluate outcomes across multiple domains – ICP control, mortality, favorable functional outcomes, Glasgow outcome scale (GOS), and complications – when comparing HC to DC. In doing so, we intend to provide a robust foundation for clinical decision-making and highlight areas that require further investigation.

A recent meta-analysis^[5] compared HC and DC, but it also included two studies,^[6,7] which included the osteoplastic craniotomy procedures that did not involve decompressive motive or hinge/floating fixation. Therefore, they were excluded from the current review as this technique cannot be considered equivalent to HC or floating craniotomy, which are defined by their capacity to allow cerebral expansion while preserving the bone flap. In addition, five additional studies^[8-12] have been included in this updated analysis.

This review will assess clinical outcomes, including favorable recovery measured by the GOS, where the favorable outcomes are GOS >3, mortality (in-hospital and at last follow-up), complications, and post-operative ICP -1, 2, and 3 days postoperatively, in patients undergoing surgical decompression for elevated ICP. The objective is to assess whether HC yields outcomes comparable to those of DC.

MATERIALS AND METHODS

The meta-analysis is registered on PROSPERO (CRD420251114487) and was carried out according to the procedures outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses.^[13]

RESULTS

Forty-eight articles were identified by a comprehensive search, of which four duplicate articles were excluded from the study selection process. A total of 44 articles were used for title/abstract screening, and 19 articles remained for further evaluation. In the full-text screening, seven articles were excluded due to the lack of relevant outcomes (n = 1) and single-arm studies (n = 6). Finally, 12 articles^[8-12,14-20] fully met the eligibility criteria and were used for data synthesis. Supplementary Figure 1 represents the PRISMA flowchart of the study selection process. The studies included 663 participants (HC: 311 and DC: 352). The study demographics are given in Table 1. The quality assessment for the studies is shown in Table 2 and Supplementary Figure 2.

Mortality

In-hospital mortality

In-hospital mortality was reported in nine studies.^[8-11,14-18] One study^[10] with only two patients in the DC arm (both of whom died) was excluded from the mortality meta-analysis due to extreme arm-size imbalance. When the study was included, the pooled RR was 0.95 (95% confidence interval [CI]: 0.51–1.77; I² = 55%). After excluding it, eight studies contributed to the pooled analysis with a pooled RR of 1.20 (95% CI: 0.75–1.93; I² = 14%), again indicating no statistically significant difference between HC and DC [Figure 1]. The statistical heterogeneity was low (I² = 14%) when that study was excluded, suggesting that the between-study differences were minimal and the effect estimates were fairly consistent. A leave-one-out sensitivity analysis of the eight studies showed similar results, with no single study exerting undue influence on the pooled effect.

Close

Figure 1:

Forest plot of in-hospital mortality comparing hinge craniotomy and decompressive craniectomy. Pooled analysis indicated no significant difference (Risk ratio: 1.20, 95% Confidence interval (CI): 0.75–1.9; I² = 14%), M-H: Mantel-Haenszel, HKSJ: Hartung–Knapp– Sidik–Jonkman, I² : I-squared statistic

Export to PPT

Follow-up mortality

Mortality at the last reported follow-up (excluding instances where in-hospital mortality was the final endpoint) was available in three studies^[8,16,19] excluding one study^[10] with arm-size imbalances. The pooled RR was 0.80 (95% CI: 0.19–3.31; I² = 48%), indicating no statistically significant difference between HC and DC [Figure 2]. The wide CI and moderate heterogeneity reflect imprecision and variability across studies. On including the study,^[10] the pooled RR was 1.07 (95%

Close

Figure 2:

Forest plot of mortality at last reported follow-up comparing hinge craniotomy and decompressive craniectomy. Pooled analysis showed no significant difference (Risk ratio: 0.80, 95% Confidence interval (CI): 0.19–3.31; I² = 48%), with moderate heterogeneity across studies, I²: I-squared statistic, HC: Hinge craniotomy, DC: Decompressive craniectomy

Export to PPT

GOS

A favorable outcome on the GOS (defined by score > 3) was reported at discharge and at 6 months.^{[8,10,12,16-18]} At discharge, the pooled RR was 1.51 (95% CI: 0.34–6.71; I² = 0%), [Figure 3] indicating a non-significant trend favoring HC with wide uncertainty. Similarly, at 6 months, the pooled RR was 1.23 (95% CI: 0.85–1.76; I² = 0%), again showing a non-significant trend toward better functional outcome in the HC group [Figure 4]. In both analyses, statistical heterogeneity was absent. The leave-one-out sensitivity analysis for GOS at 6 months showed that no single study entirely changed the direction of the effect and the leave-oneout analysis for GOS at discharge showed that while the point estimates consistently favored HC, the results remained non-significant. The CIs widened considerably when individual studies were excluded, reflecting the very small number of studies and sparse data, and indicating that the pooled result is highly imprecise and should be interpreted with caution.

Close

Figure 3:

Forest plot of favorable Glasgow outcome scale (GOS) at discharge (GOS >3) comparing hinge craniotomy (HC) and decompressive craniectomy. Pooled analysis showed no significant difference (Risk ratio: 1.51, 95% confidence interval [CI]: 0.34–6.71; I² = 0%). Leave-oneout analysis demonstrated that point estimates consistently favored HC but remained non-significant, with CIs widening dramatically when individual studies were excluded highlighting the imprecision of the evidence base.

Export to PPT

Close

Figure 4:

Forest plot of favorable Glasgow outcome scale (GOS) at 6 months (GOS >3) comparing hinge craniotomy and decompressive craniectomy. Pooled analysis showed no significant difference (Risk ratio: 1.23, 95% Confidence interval: 0.85–1.76; I² = 0%). Leave-one-out sensitivity analysis showed that no single study fully changed the direction of the effect.

Export to PPT

Complications

Hydrocephalus

Hydrocephalus incidence was reported in six studies.^{[8,11,15,16,19,20]} The pooled RR was 0.48 (95% CI: 0.36– 0.65; I² = 0%), indicating that there was a significantly lower risk of hydrocephalus following HC compared to DC [Figure 5]. Statistical heterogeneity was absent across studies. On performing leave-one-out analysis, it was seen that the pooled RRs were similar in magnitude and direction, indicating that the results are robust.

Close

Figure 5:

Forest plot of hydrocephalus incidence comparing hinge craniotomy (HC) and decompressive craniectomy. HC was associated with a significantly lower risk (Risk ratio: 0.48, 95% Confidence interval (CI): 0.36–0.65). M-H: Mantel-Haenszel, HKSJ: Hartung–Knapp–Sidik– Jonkman

Export to PPT

Seizure

Seizure incidence was reported in three studies.^[8,11,20] The pooled RR was 0.64 (95% CI: 0.40–1.04), suggesting fewer seizures with HC compared to DC, although not statistically significant [Figure 6]. Leave-one-out sensitivity analysis was performed and although the overall direction and magnitude of the effect were similar, the precision of the pooled effect was sensitive to study removal due to the dominance of one study.^[8]

Close

Figure 6:

Forest plot of post-operative seizure incidence comparing hinge craniotomy and decompressive craniectomy. Pooled analysis showed no statistically significant difference between the groups (Risk ratio 0.64, 95% confidence interval (CI): 0.40–1.04). M-H: Mantel-Haenszel, HKSJ: Hartung–Knapp–Sidik–Jonkman

Export to PPT

Post-operative ICP

Out of the three studies^[14-16] reporting post-operative ICP for both HC and DC groups [Table 3], the first study^[14] had a mixed cohort and comparable mean ICPs between HC and DC over all 3 days. ICP was 2 mmHg lower in HC on day 1 (10.1 ± 5.9 vs. 12.1 ± 6.7 mmHg), 1 mmHg higher on day 2 (13.4 ± 8.2 vs. 12.4 ± 5.1 mmHg), and 1.8 mmHg lower on day 3 (11.3 ± 3.5 vs. 13.1 ± 5.9 mmHg). These differences were minor compared to the wide within-group variability, which indicated no consistent advantage for either procedure.

Table 3: Post-operative ICP (mmHg) on post-operative days 1–3 in patients undergoing HC and DC.

Study (year)	Indication/population	Group	n (ICP monitored)	POD1 mean±SD (mmHg)	POD2 mean±SD (mmHg)	POD3 mean±SD (mmHg)
Kenning et al. 2009[14]	Mixed (TBI, ICH, infarct)	HC	20	10.1±5.9	13.4±8.2	11.3±3.5
		DC	30	12.1±6.7	12.4±5.1	13.1±5.9
Kenning et al. 2012[15]	Malignant infarction	HC	9	9.7±5.3	13.3±3.4	11.1±2.4
		DC	19	10.3±6.3	12.6±3.1	12.0±3.3
Vankipuram et al. 2020[16]	TBI (severe)	DC	20	8.9 ± 6.7	6.9 ± 6.3	6.5 ± 5.0

POD: Post-operative day, FoQOsD: Four-quadrant osteoplastic decompressive craniotomy, TBI: Traumatic brain injury, ICH: Intracerebral hemorrhage, HC: Hinge craniotomy, DC: Decompressive craniectomy, SD: Standard deviation, ICP: Intracranial pressure

The second study^[15] focused on a cohort with malignant infarction. HC values were again close to those in DC: 0.6 mmHg lower on day 1 (9.7 ± 5.3 vs. 10.3 ± 6.3 mmHg), 0.7 mmHg higher on day 2 (13.3 ± 3.4 vs. 12.6 ± 3.1 mmHg), and 0.9 mmHg lower on day 3 (11.1 ± 2.4 vs. 12.0 ± 3.3 mmHg). The differences were slight and inconsistent in direction, as in the previous study. Although mean HC values in both studies were similar, the within-group variability was lower in the second study,^[15] most likely reflecting the more homogeneous infarct population compared with the mixed cohort in 2009.

The third study^[16] evaluated decompression using the four-quadrant osteoplastic HC technique in severe TBI patients, and it reported lower levels of ICP in the DC group on each of the 3 post-operative days (POD). Differences were ~2 mmHg on day 1 (11.1 ± 6.8 vs. 8.9 ± 6.7 mmHg) and ~2.5–3 mmHg on days 2–3 (HC 9.7 ± 5.3 vs. DC 6.9 ± 6.3 mmHg; HC 9.4 ± 5.2 vs. DC 6.5 ± 5.0 mmHg). Although variability remained large, the direction of effect favored DC in this study.

These findings indicate that post-operative ICP was generally reduced by both HC and DC into a comparable range. However, differences were slight, inconsistent, and frequently overshadowed by within-group variability, except in one study,^[16] where DC values were consistently lower. Firm conclusions are limited by the small number of studies and the varied indications (malignant infarction, trauma, and mixed cohorts).

The three studies ^[14-16] reporting post-operative ICP values were included in the exploratory pooled analysis. The primary pooled MD (random effects) for POD1, POD2, and POD3 were −0.32 mmHg (95% CI: −5.68–5.05; I² = 15%), 1.32 mmHg (95% CI: 1.37–4.02; I² = 0%), and –0.09 mmHg (95% CI: −6.05–5.87; I² = 67%), respectively, indicating no statistically significant differences between HC and DC [Supplementary Figures 3-5]. There was substantial statistical heterogeneity on POD3. In the sensitivity analysis excluding one of the studies^[14] (to remove potential cohort overlap), the pooled MDs changed to 0.88 mmHg (95% CI: –16.66–18.42) for POD1, 1.41 mmHg (95% CI: –11.06–13.89) for POD2, and 0.82 mmHg (95% CI: –23.24–24.89) for POD3. These estimates were imprecise, with wide CIs, reflecting the very limited number of studies.

Overall, post-operative ICP values following HC and DC were broadly comparable across studies. The minor observed differences were inconsistent in direction and overshadowed by within-group variability, and the limited evidence base precludes firm conclusions. Due to the limited number of studies and potential overlap between reports from the same center, these pooled ICP results should be considered exploratory.

DISCUSSION

Our pooled analysis of in-hospital mortality showed low statistical heterogeneity and revealed no significant difference between HC and DC (RR: 1.20, 95% CI: 0.75– 1.93). Sensitivity analysis indicated stability of results. One study ^[10] was excluded due to its extremely small control arm size (DC arm n = 2, both deceased), which would have resulted in statistical instability.

Pooled mortality at follow-up also indicated no significant difference (RR: 0.80, 95% CI: 0.19–3.31), though moderate statistical heterogeneity (I² = 48%) was present, suggesting that there was variability in follow-up durations and populations. Current evidence available does not support the superiority of either technique in survival outcomes.

We synthesized the available evidence on favorable functional outcome, defined by the GOS >3, at both hospital discharge and at 6 months. At discharge, the pooled RR from three small studies ^[10,17,18] suggested a non-significant trend favoring HC (RR: 1.51, 95% CI: 0.34–6.71; I² = 0%). However, the evidence base was extremely limited, with sparse events and one study contributing only two patients in the DC arm. Sensitivity analysis showed that while point estimates consistently in favor of HC, the removal of individual studies resulted in considerably broader CIs. This pattern emphasizes how fragile and imprecise the discharge GOS analysis is, and the results should therefore be treated as exploratory rather than conclusive.

Four studies ^[8,10,12,16] were available at 6 months, providing a somewhat stronger evidence base. The pooled estimate again favored HC (RR: 1.23, 95% CI: 0.85–1.76; I² = 0%), but remained statistically non-significant. Importantly, sensitivity analyses indicated greater stability. Exclusion of individual studies shifted the pooled effect’s magnitude but did not reverse its direction. Only low heterogeneity was seen when studies were excluded, and point estimates consistently favored HC. This suggests that while the precision is still limited, the 6-month data provide a more consistent signal than at discharge.

The evidence that is available indicates that HC may be associated with better functional outcome compared to DC, but the data remain insufficient to establish a clear benefit. The trend favoring HC is consistent across time points, yet the analyses are restricted by small sample sizes, sparse events, and reliance on a few single-center studies. Before we can make firm conclusions, larger, well-powered studies are required.

Hydrocephalus was significantly less frequent following HC compared to DC (RR: 0.48, 95% CI: 0.36–0.65), with no heterogeneity (I² = 0%). This supports the physiological rationale that hinged, or floating bone flaps, may preserve more normal CSF dynamics and prevent reabsorption failure seen after DC.^[21] This is clinically meaningful as hydrocephalus often requires additional interventions.

There was no statistically significant difference in postoperative seizure incidence between HC and DC (RR: 0.64, 95% CI: 0.40–1.04). However, the direction of the effect suggests a possible benefit and warrants further investigation in more and larger comparative studies.

The synthesis of post-operative ICP outcomes highlights important differences in study populations that influence interpretability. Two studies^[14,15] reported broadly similar mean ICP values for HC. However, the 2012 study,^[15] limited to malignant infarction, showed narrower within-group variability, suggesting more precise estimates and better generalizability within that pathology, whereas the 2009 cohort^[14] was mixed (trauma, infarction, and hemorrhage), leading to greater heterogeneity. However, a third study^[16] that examined severe TBI found that DC had lower ICP values than HC, which may be due to the four-quadrant osteoplastic HC approach used as well as the underlying injury profile. Overall, our results show that both HC and DC can lower ICP into similar ranges, but surgical technique and pathology type have a major impact on results.

The results of this meta-analysis reveal that HC and DC offer broadly equivalent outcomes in terms of mortality, post-operative ICP, and short-to-medium term functional recovery. However, HC may have advantages in reducing certain complications, especially hydrocephalus.

CONCLUSION

This meta-analysis reveals that both HC and DC are comparable in terms of mortality. Functional recovery was also similar, but the GOS findings were fragile and imprecise at discharge, while 6-month outcomes were more consistent. Seizure risk was not significantly different between HC and DC, but pooled estimates suggested possibly fewer seizures with HC which requires confirmation in more and larger studies. HC was associated with a lower incidence of hydrocephalus. Post-operative ICP could not be assessed reliably due to insufficient data, highlighting the need for further research.