Haemorrhage pattern predicts complications in angiogram-negative subarachnoid haemorrhage patients

This systematic review and meta-analysis examined the clinical course of 5921 patients with angiogram-negative subarachnoid haemorrhage. The analysis focused on how the haemorrhage pattern, categorized as perimesencephalic, non-perimesencephalic, diffuse, or convexity, influences patient outcomes. The review aimed to determine if these patterns are associated with distinct complication profiles that could guide clinical surveillance. The study population consisted of patients who underwent angiography to rule out aneurysms but remained negative for a definitive vascular source. The setting encompassed various clinical environments where these patients were managed over a follow-up period of 3 to 6 months.

The primary intervention of interest was the haemorrhage pattern itself, which served as a stratification variable rather than a therapeutic agent. The comparator was the presence of other haemorrhage patterns within the same cohort. The study did not test a specific drug or device but rather utilized the radiological classification of the bleed to predict risk. This approach allows clinicians to tailor monitoring intensity based on the initial angiogram-negative diagnosis.

Primary outcome assessment focused on favourable functional outcome. The meta-analysis reported that 95% to 97% of patients achieved a favourable functional outcome. The specific breakdown by haemorrhage pattern was not detailed for this primary metric in the provided data, but the overall high rate suggests a generally benign course for this population. Secondary outcomes included rebleeding, clinical vasospasm, acute hydrocephalus requiring external ventricular drainage, and delayed vascular lesion detection.

Rebleeding rates varied significantly by haemorrhage pattern. The rate was 1% for perimesencephalic haemorrhage, 2% for non-perimesencephalic haemorrhage, and 3% for diffuse haemorrhage. Clinical vasospasm incidence was 3% in perimesencephalic cases, 11% in non-perimesencephalic cases, and 13% in diffuse cases. Acute hydrocephalus requiring external ventricular drainage occurred in 7% of perimesencephalic patients, 29% of non-perimesencephalic patients, and 44% of diffuse haemorrhage patients. These differences highlight the prognostic value of the haemorrhage pattern classification.

Delayed vascular lesion detection was another critical finding. A total of 82 vascular lesions were identified during the follow-up period. No lesions were detected within 7 days of the ictus. All identified lesions were found at or beyond 7 days after the initial event. This temporal pattern suggests that early negative imaging does not rule out delayed vascular pathology, particularly in non-perimesencephalic and diffuse patterns.

Safety and tolerability were assessed through the incidence of adverse events. The main adverse events observed were rebleeding, clinical vasospasm, and acute hydrocephalus. The rates of these events were directly linked to the haemorrhage pattern. No specific serious adverse events or discontinuation rates were reported beyond the complication profiles. The study notes that repeat imaging was selectively performed and timing varied across the included studies, which introduces potential heterogeneity.

A key methodological limitation is the selective nature of repeat imaging and the variation in timing across the included studies. This variability may affect the comparability of rebleeding and vasospasm rates. The study does not provide p-values or confidence intervals for the primary outcome, and absolute numbers for some metrics were not fully detailed. Prospective studies are required to define optimal imaging intervals and to confirm these associations with greater certainty.

The clinical implication is that haemorrhage pattern may assist in individualising surveillance strategies. Clinicians can use the pattern to anticipate risks such as hydrocephalus or vasospasm. For instance, diffuse haemorrhage carries a higher risk of hydrocephalus requiring drainage compared to perimesencephalic haemorrhage. However, the evidence is observational, and causality cannot be definitively established. Questions remain regarding the optimal timing for repeat imaging in each pattern to balance detection of delayed lesions against resource utilization.

In conclusion, this meta-analysis provides evidence that haemorrhage pattern is associated with distinct complication profiles in angiogram-negative subarachnoid haemorrhage. The high rate of favourable functional outcomes is encouraging, but the risk of rebleeding and hydrocephalus varies by pattern. The identification of 82 delayed vascular lesions underscores the importance of extended surveillance. Future research must address the limitations of selective imaging and establish standardized protocols for follow-up.