Abstract
OBJECTIVES:
To investigate the variability in diagnostic and therapeutic approaches to posthemorrhagic ventricular dilatation (PHVD) among Canadian neonatal centers, and secondary exploration of differences in approaches between Canadian and European practices.
METHODS:
We conducted a survey among Canadian tertiary neonatal centers on their local practices for managing very preterm infants with PHVD. The survey covered questions on the diagnostic criteria, timing and type of interventions and resources utilization (transfer to neurosurgical sites and neurodevelopmental follow-up). In a secondary exploration, Canadian responses were compared with responses to the same survey from European centers.
RESULTS:
23/30 Canadian centers (77%) completed the survey. There was no consensus among Canadian centers on the criteria used for diagnosing PHVD or to initiate intervention. The therapeutic interventions also vary, both for temporizing procedures or permanent shunting. Compared to European practices, the Canadian approach relied less on the sole use of ultrasound criteria for diagnosing PHVD (43 vs 94%, p < 0.0001) or timing intervention (26 vs 63%, p = 0.007). Majority of European centers intervened early in the development of PHVD based on ultrasound parameters, whereas Canadian centers intervened based on clinical hydrocephalus, with fewer centers performing serial lumbar punctures prior to neurosurgical procedures (40 vs 81%, p = 0.003).
CONCLUSION:
Considerable variability exists in diagnosis and management of PHVD in preterm infants among Canadian tertiary centers and between Canadian and European practices. Given the potential implications of the inter-center practice variability on the short- and long-term outcomes of preterm infants with PHVD, efforts towards evidence-based Canada-wide practice standardization are underway.
Keywords
List of abbreviations
Ānterior horn width Canadian Neonatal Network Cerebrospinal fluid Frontal horn ratio, Frontal-temporal horn ratio Intraventricular hemorrhage Neurodevelopmental impairment Neonatal Intensive Care Unit Posthemorrhagic ventricular dilatation Thalamo-occipital distance Ventricular index Ventriculoperitoneal
Introduction
Neurologic injury and subsequent neurodevelopmental impairment (NDI) remain well-known complications of prematurity in spite of improved survival [1–3]. Severe intraventricular hemorrhage (IVH) (also called grade 3 or 4 IVH) affects around 10–15%of preterm infants, and approximately 30–50%of those infants develop subsequent posthemorrhagic ventricular dilatation (PHVD) requiring intervention [1–3]. PHVD in preterm infants is an independent predictor for NDI including: cerebral palsy, epilepsy, cognitive and behavioral delay [4–6]. Preterm infants with PHVD may require multiple interventions, prolonged hospitalization and long-term care resulting in high health services utilization [7].
Center variability in the rates of PHVD in preterm infants has been previously reported [8]. This is compounded by diverse criteria for diagnosis and treatment of PHVD [9–17], with little guidance from large prospective clinical trials on the best approach. PHVD diagnosis is commonly based on different ultrasound dimensions of lateral ventricules (ventricular index (VI), anterior horn width (AHW), thalamo-occipital distance (TOD), frontal-temporal horn ratio (FTHR) [18–21]. Compared to ultrasound, PHVD diagnosis based on clinical hydrocephalus (enlarged head circumference > 2 cm/week and splayed sutures) and/or signs of increased intracranial pressure (bulging fontanelle, apnea/bradycardia, irritability, vomiting or seizures) results in considerable delay in diagnosis and intervention [2, 12]. Interventions for PHVD include drainage of intraventricular blood [13], cerebrospinal fluid (CSF) tapping (serial lumbar punctures) [14], temporizing neurosurgical CSF diversion (ventricular access device) or permanent ventriculoperitoneal (VP) shunt [15].
There is lack of data on the management practices for preterm infants with PHVD across Canadian tertiary Neonatal Intensive Care Units (NICUs). Data from the largest neurosurgical referral center in Canada suggested that the local approach for PHVD was associated with worse short and long-term outcomes compared to the European approach [22]. Therefore, we conducted this survey study to gain a better understanding of the variability in institution-specific diagnostic and therapeutic practices applied to very preterm infants (<33 weeks’ gestation) with PHVD across Canadian tertiary NICUs. As a secondary exploration, we compared the diagnostic and therapeutic practices between Canadian and European neonatal centers, using the responses to the same survey previously sent out in Europe [11].
Methods
Design
A cross-sectional survey comprising questions on Canadian institutional diagnostic and therapeutic practices for preterm infants with PHVD was administered between October 2017 - May 2018. PHVD was defined as ventricular dilatation due to accumulation of CSF in the ventricular system following unilateral or bilateral IVH. In view of our secondary exploration, we used the same survey previously used to explore practice variability across 17 European NICUs (with permission from de Vries LS) [11]. Study approval from the Research Ethics Board at the IWK Health Centre, Halifax, Canada, was obtained. Requirement for informed consent from each NICU was waived, as the survey did not request patient or participant information, but focused on exploring the local practices in their NICUs.
Participants
One Neonatologist (Canadian Neonatal Network (CNN) Site Investigator/delegate) from each tertiary NICU in Canada (n = 30) was approached to participate in the survey. The CNN Site Investigators were chosen because of their involvement in a working group implementing evidence-based practices for prevention of preterm brain injury as part of a national quality improvement collaborative [23].
Data collection
The survey was divided into four main sections (Appendix): Local criteria (clinical and/or radiological) used for diagnosing PHVD; criteria applied to decide on timing/threshold of intervention; types of locally implemented interventions (temporizing and/or permanent); and resources utilization including: referral for neurosurgery, transport to other facilities for surgical intervention and long-term follow-up
Statistical analysis
Descriptive statistics were used to present the Canadian survey results with parameters presented as proportion, mean (standard deviation) or median (inter-quartile ranges), as appropriate. Differences between Canadian and European responses were compared using Chi square or Fisher exact test. Significance was set as p < 0.05.
Results
Respondents’ demographics
23/30 sites completed the survey representing 77%of all tertiary NICUs across Canada. The Canadian and European (86%; 32/37 tertiary NICUs) response rates were comparable.
The Canadian Survey
PHVD diagnosis
Candian centers commonly (74%) used combined clincal and radiologic criteria for diagnosing PHVD. When survey participants were asked about which ultrasound parameter(s) of lateral ventricules used for diagnosing PHVD, there was lack of consensus [Table 1]. Canadian participants commonly used multiple parameters for diagnosing PHVD (29%used > 2 paramters), but infrequently used third or combined third and fourth ventricles dilatation for diagnosis (13%and 26%, respectively).
Cranial ultrasound parameters for diagnosing PHVD
Cranial ultrasound parameters for diagnosing PHVD
Data are presented as number (percentage). *2 Canadian centers and 2 European centers used visual assessment of ballooning of lateral ventricles, as per radiology report. # Canadian survey: combination of > 2 parameters = 5, TOD+AHW=1. European survey;VI+AHW+TOD=2. Abbreviations: AHW (anterior horn width), FHR (frontal horn ratio), PHVD (posthemorrhagic ventricular dilatation), TOD (thalamo-occipital distance), VI (ventricular index).
Table 2 shows the variability among Canadian NICUs for criteria applied to time intervention for PHVD. None of Canadian NICUs practiced early approach based solely on ultrasound (defined as VI > 97 percentile and AHW > 6 mm) and only 26%(6/23) provided a cutoff value from standardized ultrasound plots (mostly Levene or Davies charts) [Table 2]. The Canadian approach relied commonly on clinical hydrocephalus either alone (17%), or in combination with ultrasound (53%) or other criteria (4%) for initiating intervention.
Criteria for defining the timing for intervention for PHVD
Criteria for defining the timing for intervention for PHVD
Data are presented as number (percentage). # Increased intracranial pressure + ultrasound measures = 7, large head size + ultrasound measures = 1, combination of these factors = 4. Abbreviations: PHVD (posthemorrhagic ventricular dilatation).
Table 3 reports the temporizing CSF diversion procedures reported by 20 Canadian NICUs, with on-site access to pediatric neurosurgery. Ventricular reservoir was the most common temporizing device used (80%of responses). Following surgical reservoir insertion, 22 Canadian NICUs reported tapping by a variety of neonatal/neurosurgical clinicians under aseptic precautions in the vast majority (95%). The volume of CSF per tap varied widely among Canadian NICUs, ranging from 5-20 ml/kg, with frequency of tapping being as per clinical discretion.
Therapeutic interventions practiced for PHVD (excluding VP shunt)
Therapeutic interventions practiced for PHVD (excluding VP shunt)
Data are presented as number (percentage). More than one practice can be selected by each survey participant. *3 Canadian centers did not provide data as they transferred infants with PHVD to other hospitals for management. # Canadian survey: endoscopic third ventriculostomy + choroid plexus coagulation = 3. # European survey: endoscopic third ventriculostomy = 1, DRIFT = 2. Abbreviations: PHVD (posthemorrhagic ventricular dilatation), VP (ventriculoperitoneal).
The criteria for VP shunt insertion varied among Canadian NICUs. Weight, either alone or combined with clinical or other criteria, was the most commonly reported indicator. None of Canadian centers used solely concentration of CSF protein or eryhtrocytes to decide for VP shunt insertion [Table 4].
Criteria used for determining eligibility for VP shunt insertion
Criteria used for determining eligibility for VP shunt insertion
Data are presented as number (percentage). *3 Canadian and 2 European Centers with missing data. (3 Canadian Centers did not provide data as they transferred infants with PHVD to other hospitals for management). # Combined clinical indicators + other criteria (not specified). Abbreviations: CSF (cerebrospinal fluid), VP (ventriculoperitoneal).
74%(17/23) of Canadian NICUs referred preterm infants with PHVD to neurodevelopmental follow-up programs, with the majority reporting follow-up until 24-36 months post-term age. Standardized developmental assessments included: Bayley Scale for infant and Toddler Development 3rd Edition (15/17), Alberta Infant Motor Scale (5/17), CAT/CLAMS (4/17), and neurobehavioral tests (4/17).
Secondary comparaison of candian and european surveys
Both surveys reported similar rates of severe IVH (9%versus 7%), as per Papile classification [24], with wide variability in ultrasound parameter(s) used to diagnose PHVD [Table 1]. Fewer Canadian centers reported the sole use of standardized ultrasound parameters with their cutoffs to diagnose PHVD (26%vs 94%, p < 0.0001) or to initiate intervention (26%vs 63%, p = 0.007) than European centers [Table 2]. Both surveys did not provide intervention for isolated dilatation of the occipital horns of the lateral ventricles.
On site Neurosurgical availability was different between Canadian and European surveys (70%vs 88%). Resultant, the rate of transfer to other hospitals for surgical intervention was higher among Canadian centers (30%vs 12%, p = 0.09). Ventricular reservoirs were the commonest temporizing neurosurgical procedure reported in both surveys (80%and 66%), but Canadian centers reported less freequent lumbar punctures prior to surgical CSF diversion (40%vs 81%, p = 0.003) and higher rates of subgaleal shunts (30 vs 3%, p = 0.006). [Table 3] Temporizing devices were almost always (22/23; 96%) inserted by neurosurgeons in Canada, while inserted by either surgeons (12/28; 43%) or neurosurgeons (16/28; 57%) across European centers.
Canadian practice relied less frequently on concentration of CSF protein or erythrocytes, either alone (0%vs 17%, p = 0.05) or combined with infant’s weight (20%vs 57%, p = 0.01) as criteria for VP shunt compared to European practice [Table 4]. European participants provided specific details in relation to indications for VP shunt placement including: reservoir complications, failure of tapping or requiring tapping for > 4 weeks.
Discussion
Our study showed marked variability among Canadian NICUs regarding the diagnosis, timing and type of interventions when managing preterm infants with PHVD. Canadian practice differed substantially from European practice in relation to the sole use of ultrasound standardized parameters for diagnosing PHVD or to time intervention. European centers intervene early in the trajectory of PHVD development, based on ultrasound cutoffs, when the infant may still be asymptomatic.Whereas, Canadian centers tend to intervene late awaiting clinical signs of hydrocephalus. This finding is compounded by lack of consensus among healthcare providers regarding the benefits of early treatment versus the risk of more interventions and complications.
Canadian and European surveys reported similar rates of severe IVH (9%and 7%, respectively) comparable to published reports [1–3]. However, Canadian and European diagnostic practices varied in relation to the sole use of ultrasound to diagnose PHVD, with a lack of consensus on both surveys regarding which parameter(s) to use. Serial measures of standardized indices with their reference plots/charts, predominantly used by European centers, strengthen quantified assessment of PHVD severity and progression [18–21, 25]. Regardless of the ultrasound parameter(s) used, standardizing the diagnostic approach and defining the cutoff/threshold for intervention would contribute to reduction of practice variability.
The criteria for timing PHVD intervention differed between Canadian and European surveys; relying solely on ultrasound parameters of early ventricular dilatation in Europe, while based mainly on clinical hydrocephalus in Canada. Previous studies reported the lack of correlation between head size and ultrasound evidence of PHVD in preterm infants, with hydrocephalus occuring at a late stage due to compliance of preterm brain and skull to ventricular dilatation [12, 26]. Awaiting head enlargement or increased intracranial pressure would delay intervention with the potential for secondary periventricular whilte matter injury from breakdown of intraventricular blood products, prolonged compression/ischemia, inflammation and release of oxygen free radicels [3]. The resultant brain injury is characterized by ependymal disruption, axonal injury, microglial infiltration, impaired mylenation, gliosis and fibrosis antecedent to NDI [3]. Previous studies showed that CSF drainage and ventricular decompression resulted in improved cerebral oxygenation, electroencephalogram background patterns and white matter diffusion in infants with PHVD [27–29]. Recently, ELVIS trial showed significantly lower death/NDI at 2 years favoring early intervention [30, 31]. A nested sub-study also showed that late intervention for PHVD is associated with delayed myelination, thinning of corpus callosum and high global brain injury scores [32].
Our survey provided insight of the underlying practices between the clinically-focused Canadian approach and the ultrasound-based European approach. Notably, both the European survey and ELVIS trial defined thresholds for PHVD intervention as early (VI > 97th centile and AHW > 6 mm or TOD > 24 mm) or late (VI > 97th centile+4 mm and AHW > 10 mm) based on ultrasound, while the infant may be asymptomatic [11, 30]. In this survey, none of the Canadian centes practiced early ultrasound approach, only 26%practiced late ultrasound approach, while the majority (70%) relied on clinical or combined clinical/radiological hydrocephalus. The Canadian approach in this survey is” later” than both the late intervention arm of the ELVIS trial and the late approach of the European survey. This finding is in agreement with recent prospective study from the largest neurosurgical referral center in Canada comparing the “early Dutch” versus the “late Canadian” approach (median age at first intervention for PHVD of 13 vs 47 days and median corrected age 29 vs 33 weeks, respectively) [22]. In this study, Leijser et al showed that early intervention was associated with significantly lower VP shunt rates, lower shunt complications and better neurodevelopment [22]. Our survey findings are concerning with evidence from observational studies and clinical trial showing higher death/NDI with late approach [22, 30–34].
The type of treatment modalities for PHVD differed between Canadian and European practices in agreement with previous studies showing choice of surgical modality (primary VP-shunt vs primary temporizing diversion) as per Neurosurgeon’s preference [9, 15–17]. Although temporizing neurosurgical CSF diversion was commonly used as the primary mode of treatment in both surveys, serial lumbar punctures prior to surgical CSF diversion was more commonly practiced in Europe. Recent data from ELVIS trial suggested that serial lumbar punctures may reduce the need for surgical intervention [30].
Both surveys showed similarity in relation to ventricular reservoirs being the most common CSF-drainage device used with different healthcare professionals performing reservoir tapping. A recent study showed that repeated reservoirs tapping did not increase the risk of infection among preterm infants with PHVD [35]. Subgaleal shunts were implanted more frequently in Canada, agreeing with multiple studies showing no difference between reservoirs and subgaleal shunts in the rate of conversion to VP shunt or other short-term outcomes [9, 15].
The criteria for VP shunt insertion was different in both surveys. Neurosurgeon’s threshold and preference, the availability of neurosurgeons on site and different surgical referral systems between Canadian and European practices may contribute to the reported variability. Notably, both surveys reported low absolute numbers of infants operated upon at each center (<5 per year), substantiating published reports of temporal decrease in VP shunt rates among preterm infants with PHVD [10, 36]. However, the proportion of infants with PHVD receiving VP shunt remains high in Canada (29%) with variable reported rates in literature ranging from 20%to 60%[9, 37].
In preterm infants, progressive PHVD is an independent predictor of NDI and those who received surgical interventions being at the highest risk [5, 37]. However, there is no consensus between healthcare providers, including neonatologists and neurosurgeons, on the type and time of intervenions and the benefits of early approach versus the risks for more procedures and complications. Recently, guidelines with stepwise approach for managing PHVD based on risk categorization were published [38]. The Canadian Neonatal Follow-up Network reported center variability in the rates of NDI among preterm infants < 29 weeks’ gestation [39]. The relation between practice variations highlighted in this survey and the long-term outcomes of preterm infants with PHVD needs further exploration. A study to assess the neurodevelopmental outcomes of Canadian preterm infants with PHVD is underway.
Strengths and limitations
This survey study has several strengths including good response rate and representation from the majority of tertiary NICUs caring for very preterm infants in Canada. However, this study is not without limitations; surveys are known for recall bias. One Neonatologist from each NICU in Canada was approached to reduce sampling bias, but limiting our ability to assess within center variability when managing PHVD. The survey did not include details on the local ultrasound protocols and the time/corrected age at surgical referral or intervention. We explored the difference between the current Canadian practices to the 2012-reported European practices, which may not reflect the change in practice over time. However, we do not anticipate major changes in the European approach for PHVD inbetween the two surveys; as the same approach was practiced widely in Europe during conduction of ELVIS trial (2010-2016). Finally, we did not report the responses from Canadian Neurosurgeons due to poor response rates, limited to a small subgroup of Neurosurgeons not reflecting the Canadian geographic surgical coverage.
Conclusion
There is considerable variability in the diagnostic criteria and the type and timing of therapeutic interventions when managing PHVD in preterm infants among Canadian NICUs and between Canadian and European centers. The impact of those practice variations on the long-term outcomes of preterm infants with PHVD needs further exploration.
Future research and clinical implications
This descriptive study provides information to aid future research exploring the impact of center variability on the neurodevelopmental outcomes of preterm infants with PHVD. Future research should focus on the availability and adherence to local ultrasound protocols, guidelines standardizing the diagnostic and therapeutic approach, the details of neurosurgical referral, the time and type of interventions and indications for VP shunts. Future research should also explore the impact of the newer technology (amplitude integrated electroencephalography, near-infrared spectroscopy or 3-dimensional ultrasound) in determining the optimal timing for intervention for PHVD.
Our survey raised the awareness of the Canadian practice variability and its potential impact on outcomes of preterm infants with PHVD. Therefore, efforts towards evidence-based standardized Canada wide practices for PHVD are ongoing. Guidelines standardizing ultrasound screening protocols for preterm infants, including PHVD definition and grading was recently published (JA and LL are co-authors) [40]. Furthermore, a national quality improvement collaborative to standardize the diagnostic and therapeutic practices for PHVD is underway.
Footnotes
Acknowledgments
The authors would like to acknowledge the site investigators of the CNN for their participation in this study.
Disclosure
This project was conducted with no specific financial support. No honorarium, grant, or other form of payment was given to anyone to produce this manuscript. The authors indicate that they have no financial relationships and no potential, perceived, or real conflict of interest relevant to this manuscript.