The role of blood product removal in intraventricular hemorrhage of prematurity: a meta-analysis of the clinical evidence.

introductionPremature neonates have a high risk of intraventricular hemorrhage (IVH) at birth, the blood products of which activate inflammatory cascades that can cause hydrocephalus and long-term neurological morbidities and sequelae. However, there is no consensus for one treatment strategy. While the mainstay of treatment involves CSF diversion to reduce intracranial pressure, a number of interventions focus on blood product removal at various stages including extraventricular drains (EVD), intra-ventricular thrombolytics, drainage-irrigation-fibrinolytic therapy (DRIFT), and neuroendoscopic lavage (NEL).

methodsWe performed a systematic review and meta-analysis to compare the risks and benefits commonly associated with active blood product removal treatment strategies. We searched MEDLINE, Embase, Scopus, Cochrane Library, and CINAHL databases through Dec 2020 for articles reporting on outcomes of EVDs, thrombolytics, DRIFT, and NEL. Outcomes of interest were rate of conversion to ventriculoperitoneal shunt (VPS), infection, mortality, secondary hemorrhage, and cognitive disability.

resultsOf the 10,398 articles identified in the search, 23 full-text articles representing 22 cohorts and 530 patients were included for meta-analysis. These articles included retrospective, prospective, and randomized controlled studies on the use of EVDs (n = 7), thrombolytics (n = 8), DRIFT therapy (n = 3), and NEL (n = 5). Pooled rates of reported outcomes for EVD, thrombolytics, DRIFT, and NEL for ventriculoperitoneal shunt (VPS) placement were 51.1%, 43.3%, 34.3%, and 54.8%; for infection, 15.4%, 12.5%, 4.7%, and 11.0%; for mortality, 20.0%, 11.6%, 6.0%, and 4.9%; for secondary hemorrhage, 5.8%, 7.8%, 20.0%, and 6.9%; for cognitive impairment, 52.6%, 50.0%, 53.7%, and 50.9%. Meta-regression using type of treatment as a categorical covariate showed no effect of treatment modality on rate of VPS conversion or cognitive disability.

conclusionThere was a significant effect of treatment modality on secondary hemorrhage and mortality; however, mortality was no longer significant after adjusting for year of publication. Re-hemorrhage rate was significantly higher for DRIFT (p < 0.001) but did not differ among the other modalities. NEL also had lower mortality relative to EVD (p < 0.001) and thrombolytics (p = 0.013), which was no longer significant after adjusting for year of publication. Thus, NEL appears to be safer than DRIFT in terms of risk of hemorrhage, and not different than other blood-product removal strategies in terms of mortality. Outcomes-in terms of shunting and cognitive impairment-did not differ. Later year of publication was predictive of lower rates of mortality, but not the other outcome variables. Further prospective and randomized studies will be necessary to directly compare NEL with other temporizing procedures.