Highlights
-
This study describes our institution’s experience of implementing an early intervention management protocol for infants with post-hemorrhagic ventricular dilatation
-
The protocol was found to be safe and areas for improvement were identified
Introduction
Intraventricular hemorrhage (IVH) occurs in 20%–30% of preterm infants with a birth weight of <1500 g, Reference Koschnitzky, Keep and Limbrick1 and up to 40% of those with IVH develop post-hemorrhagic hydrocephalus (PHH). Reference Chen, Feng and Tan2 The occurrence of PHH has been linked to significant psychomotor delay and epilepsy. Reference Maunu, Lehtonen and Lapinleimu3
Since permanent CSF diversion with a ventriculoperitoneal shunt (VPS) is not usually feasible as an initial intervention in these patients due to their poor immunity, thin skin, and high risk for developing necrotizing enterocolitis, a number of temporizing measures have been recommended in the literature. Reference Deger, Goethe, LoPresti and Lam4 Of these, the ventricular access device (VAD) and the ventriculo-subgaleal (VSG) shunt have arguably been employed most routinely in a North American setting, although there remains no consensus regarding the exact timing of such interventions.
There is some evidence suggesting that early insertion of a VAD based on neurosonographic criteria, as opposed to initial intervention after the development of symptoms and signs of elevated intracranial pressure (usually at a much larger ventricular size), may result in better neurocognitive outcomes and a decreased likelihood of requiring subsequent VPS insertion in preterm infants with post-hemorrhagic ventricular dilatation (PHVD). Reference de Vries, Liem and van Dijk5–Reference Leijser, Miller and Van Wezel-Meijler7 Based on this data and ongoing discussions with our neonatology colleagues, a decision was made to standardize our approach to managing infants with PHVD and to adapt and test an early intervention management protocol for infants using a quality improvement framework. In this publication, we evaluate our experience with implementation of this new protocol, examine process and compliance measures, identify areas of improvement based on our initial results, and assess resource use and safety implications.
Methods
Family of measures for this quality improvement initiative
We sought to evaluate our adherence to a new institutional protocol to manage PHVD, with early placement of a VAD based on specific neurosonographic ventricular measurements. The process measures were the time taken to insert a VAD after neurosonographic criteria were met, and compliance with postoperative head ultrasounds and VAD taps. The balancing measure was the complication rate related to earlier VAD insertion.
Setting
At the beginning of 2020, after 12 months of multidisciplinary protocol development, an early intervention management pathway for the management of preterm infants with IVH was instituted at our center. It is the quaternary referral site for pediatric neurosurgical care for the province of British Columbia, Canada. It is equipped with a large level IV neonatal intensive care unit (NICU) with 56 acute care and 14 step-down beds, and is a member of the Hydrocephalus Clinical Research Network (HCRN).
Management of patients in historical cohort
Prior to the implementation of this new management pathway, infants with PHVD at our institution, like most centers in North America, were managed with temporizing procedures if they manifested symptoms and/or signs of elevated intracranial pressure Reference Wellons, Shannon and Holubkov8 ; radiologic assessment of ventricular size played an ancillary role in evaluation of a patient for temporizing therapy. Similarly, tapping of the VAD was usually done to mitigate clinical symptomatology rather than to achieve a target ventricular size. Management of these infants was not standardized across neonatologists or neurosurgeons, and there was evidence of intra-center variation in how these infants were treated, both anecdotally at our site as well as in the literature. Reference El-Dib, Limbrik and Inder9
Protocol development
Our neonatologists expressed an interest in standardizing the care of these infants and exploring the possibility of adapting an early intervention management philosophy for PHVD for use in our local environment. A working group comprised of a lead neurosurgeon, neonatologist and neuroradiologist iteratively refined the details of a management protocol, based on the Early versus Late Ventricular Intervention Study (ELVIS), Reference de Vries, Groenendaal and Liem10 that was eventually agreed upon by all stakeholders. Key elements of discussion regarding how to operationalize an ELVIS-based protocol in our local environment included establishing a ventricular size threshold for surgical temporization, the frequency of ultrasound surveillance, standardization of radiologic reporting of ultrasound evaluations, developing policies and procedures to facilitate the transport of infants from peripheral NICUs for possible surgical intervention and how to train and maintain competence in VAD tapping procedures amongst our NICU colleagues. As a part of protocol implementation, the neurosurgical team, who previously performed VAD taps at our institution, published a tapping procedure document on the hospital intranet and demonstrated proper tapping technique to a critical mass of neonatologists who then propagated the technique to other faculty, staff and trainees in their department. Members of the neurosurgical team provided ongoing training and assistance with this intervention when requested.
Because this new management approach would entail an increased intensity of intervention in fragile premature infants compared to our historical practice, we decided to develop and evaluate the protocol using a quality improvement methodology focused primarily on safety. The fully developed protocol was posted on the hospital intranet, and educational sessions for neonatologists at our and peripheral sites were held pre-implementation and on a regular basis thereafter.
Selection of patients
Preterm infants (born < 32 weeks of gestation) with grade III/IV IVH Reference Papile, Burstein, Burstein and Koffler11 or those with any grade of IVH associated with ventricular dilatation on ultrasonography were included in the new management pathway.
Screening ultrasound recommendations
Screening for IVH in premature infants was in accordance with the Canadian Neonatal Network’s clinical practice guideline regarding ultrasound screening for brain injury in preterm infants. Reference Mohammad, Scott and Leijser12 Once an ultrasound scan detected IVH in an infant, the Levene index was measured to assess for ventriculomegaly. Reference Levene13 If the ventricular index measured in mm was less than the 97th percentile (p97) for post-menstrual age, a scan was repeated in a week. If the repeat scan did not show any change in ventricular size, subsequent imaging was done two weeks later. If at that point, there was no progression, the next ultrasound scan was done 4–6 weeks later to look for periventricular leukomalacia. However, if the ventricles were enlarging, scans were repeated weekly. Anterior horn width Reference Davies, Swaminathan, Chuang and Betheras14 was also documented in all patients but was not a criterion considered for VAD insertion.
Apart from radiological assessment, routine clinical assessments were also performed. This included daily measurement of head circumference and documentation of symptoms or signs of raised intracranial pressure by the neonatologists. These comprised apnea/bradycardia out of proportion to degree of immaturity and without any other known cause, bulging of the anterior fontanel, splaying of the calvarial sutures, or additional specific neurological findings such as altered level of consciousness, downward deviation of the eyes and seizures. The presence of two or more of these signs was considered clinically significant. Reference Flanders, Kimmel and Lang15
Intervention recommendations
If the ultrasound showed that the Levene index in either of the ventricles had crossed the p97 threshold, indicating a diagnosis of PHVD, the neurosurgery team was consulted so that they could follow the subsequent scans. Repeat ultrasound scans thereafter were done twice a week or earlier if the infant developed any clinical sign of raised intracranial pressure. If the p97 + 4 mm threshold was crossed, VAD insertion was arranged with a goal of placement within three days of crossing the threshold. In the meantime, daily serial lumbar punctures (up to a maximum of 3) were recommended, with an aim of draining 10 ml/kg of CSF each time; the decision to perform a lumbar puncture was left to the discretion of the neonatology team, based on factors such as the presence of symptoms of elevated intracranial pressure, the general medical status of the infant and/or whether VAD placement was imminent. The three-day VAD-placement target was felt to be reasonable and realistic and was put in place to avoid a prolonged rise in ventricular size prior to surgical temporization.
If the patient was in a peripheral NICU, the process of transfer to our institution was initiated when they crossed the p97 threshold (i.e. at an ultrasonographic diagnosis of PHVD), optimistically allowing for their arrival before they crossed the p97 + 4 mm surgical intervention threshold, but also recognizing that some infants who were transferred on this basis would stabilize their ventricular size below the intervention threshold and therefore not require surgery.
Surgical procedure
The VAD was placed with the infant on the operating room table under general anesthesia. All procedures occurred during the day, with experienced anesthesia and nursing teams readily available. A curvilinear incision was made, centered on the corner of the anterior fontanel. A unitized reservoir with a 3–4 cm ventricular catheter was inserted into the lateral ventricle under ultrasound guidance, and 10 ml/kg of CSF was aspirated to confirm patency. Intravenous antibiotic administration before the skin incision, hand scrub and double gloving by all participants in the operation, preparation of the surgical field with non-alcohol iodine-based prep and the use of iodine-impregnated incision tape were ensured in all patients. These were adapted from the HCRN CSF shunt infection prevention protocol. Reference Chu, Jensen and Holubkov16 Postoperatively, the child was transferred back to the NICU for monitoring.
Ventricular access device tapping
Portable ultrasound scans were performed in the NICU, daily in the morning on the weekdays and once over the weekend. The VAD was tapped daily at the bedside by one of the neonatologists after the final ultrasound report with ventricular size indices (Levene index, anterior horn width) was reviewed and the measurements recorded, with 10 ml/kg of CSF being aspirated at a rate of 1 ml/kg/min. A 25-gauge butterfly needle connected to a 10 ml syringe was used, ensuring sterile precautions. Some infants required modification of the daily 10ml/kg tapping standard to accommodate two 10ml/kg taps per day, depending on the unique clinical circumstance. Clear stopping rules for VAD taps were put in place; taps were to be held, and neurosurgical opinion sought, if new intraventricular, intraparenchymal or extraaxial blood was seen on ultrasound, there was clinical concern of infection at the VAD site, there was technical difficulty in accessing the reservoir or aspirating the full volume of CSF, or the fontanel was sunken at the beginning of assessment.
Daily taps were continued until the Levene index fell below p97 (our prescribed target). Occasionally, we observed instances where the Levine index appeared to stabilize below p97 + 4 mm but above p97 despite repeated taps and attempts to achieve the p97 target. In these instances, one or more tap “holidays” of variable duration were then observed, whereby ultrasound assessments were carried out per protocol but tapping was only reinstituted if the Levene index again increased above p97 + 4 mm. As infants approached term-equivalent age, consideration was given to whether the infant required a permanent CSF diversionary procedure on the basis of evaluating the overall clinical status of the child (head circumference trajectory, presence or absence of clinical symptoms or signs of hydrocephalus, radiologic assessment of ventricular size) rather than on ventricular size criteria alone. If the infant did not require a permanent procedure whilst in the NICU, they continued outpatient neurosurgical surveillance for hydrocephalus with serial clinical and imaging assessments. The decision to proceed with a permanent CSF diversionary procedure was not strictly protocolized, but was consistent with our management of other infants being assessed for hydrocephalus treatment.
Data collection and statistical analysis
Data were collected retrospectively from the electronic medical record and a patient database for infants treated after introduction of the new protocol in 2020. For the historical cohort of patients treated from 2018 to 2019, relevant data were retrieved from their electronic medical records. Variables collected from both patient groups included gestational age and weight at birth, gender, IVH grade, head circumference percentile at the time of surgery and complications related to the intervention (VAD obstruction, infection, CSF leak from the wound and fresh intraventricular/parenchymal/extraaxial hemorrhage documented on imaging). The time intervals between the ultrasound finding of Levene index crossing p97 (i.e. diagnosis of PHVD) and when neurosurgery was consulted, and the interval between the Levene index crossing the p97 + 4 mm surgical threshold and VAD insertion, were determined. Where appropriate, these data were compared between both groups. For the patients managed with the new protocol, compliance rates to prescribed head ultrasound frequency and VAD tapping rules were calculated, and the reasons for protocol deviation/violation were documented.
Data were analyzed using R (version 4.3.3, R Foundation for Statistical Computing, Vienna, Austria). Frequency and percentage were used for categorical variables. Median with inter-quartile range (IQR) was used for continuous variables. Categorical variables were compared using either Fisher’s exact test or Chi-square test, and continuous variables were compared using the Mann-Whitney U (Wilcoxon rank-sum) test. A P value of<0.05 was considered significant.
Results
Patient characteristics
There were 16 infants who underwent VAD placement since the institution of the protocol. They comprised nine (56.2%) male and seven (43.8%) female infants. Their median gestational age at birth was 29.3 weeks (IQR 27.7–30.6, range 24–36.6), and the median birth weight of this cohort was 1435 g (IQR 1057.5–1971, range 510–2160). Table 1 shows a detailed comparison of the patient characteristics between the 16 infants who had VAD insertion as per the new protocol and the nine patients who underwent VAD placement in the two years prior to protocol implementation. There were no statistically significant baseline differences between the two groups, although infants treated under the new protocol had a significantly smaller ventricle size at VAD placement than the historical control group (median Levene index 20 mm vs. 25 mm, p < 0.001; median anterior horn width 18 mm vs. 23 mm, p < 0.001). Despite this result, infants treated under the new early intervention protocol continued to undergo VAD placement with severely dilated ventricles (Levene index significantly above p97 + 4 mm for gestational age and anterior horn width > 10 mm in most infants).
Table 1. Comparison of patients before and after implementation of an early intervention management protocol for neonatal PHVD
* Presence of at least two of the following – bulging fontanel, splayed sutures, rapidly increasing occipitofrontal circumference, or bradycardia not attributable to causes other than raised intracranial pressure.
IVH = intraventricular hemorrhage; ICP = intracranial pressure; VAD = ventricular access device.
Evaluation of process measures – time from diagnosis of PHVD to neurosurgical consultation and time to VAD insertion after surgical intervention threshold surpassed
The median time between the Levene index on ultrasound crossing p97 and the neurosurgery team being consulted was 1.5 days (IQR 0–6.5) as compared to an 8-day interval (IQR 5.5–14) for those infants managed under the historical, clinical symptom-driven treatment paradigm (p = 0.01). The median time interval between the Levene index crossing p97 + 4 mm and VAD insertion was 6 days (IQR 3.5–9.5) amongst infants treated under the early intervention management protocol, similar to what was observed in the historical cohort as well. The proportion of patients treated under the new protocol who had their VAD insertion within the target of 3 days after the Levene index crossed p97 + 4 mm was 4/16 (25%), and four (25%) out of the 16 patients in the new protocol group had lumbar punctures while awaiting VAD placement.
Post-VAD compliance with protocol
Patients received a median of 92% (IQR 85%–100%) of the ultrasounds required by protocol. Complete data regarding the number of VAD taps and the volume aspirated were available in only 11/16 patients. Out of a total of 341 indicated VAD taps in these 11 patients, 335 were performed as per protocol and 6 (2%) were missed. Of these 335 taps, 288 were of adequate volume, while the remaining 47 (14%) were inadequate. None of the taps were over the standard prescribed volume of 10 ml/kg. On two occasions in the same patient, an unindicated tap was done despite the Levene index dropping below p97 on ultrasound. Overall, for the patients with complete tapping data, the taps were 89% (IQR 77%–96%) compliant with protocol.
Complications
We did not observe an increase in complications for those infants treated under the new protocol despite an increased intensity of treatment-related interventions. The overall complication rate was 25% in the protocol group and 33% in the historical cohort; the breakdown of complications is shown in Table 2. Rates of VAD obstruction, infection or CSF leak from the wound were comparable. One patient had an asymptomatic recurrence of IVH on ultrasound for which VAD taps were temporarily withheld. Two patients in the historical group died due to causes unrelated to the PHVD, while there were no deaths in the protocol group.
Table 2. Comparison of VAD-related complications before and after implementation of an early intervention management protocol for neonatal PHVD
Discussion
Types of intervention for PHVD
VADs serve as a temporizing measure for CSF diversion in infants with PHVD who are deemed too immature to undergo permanent CSF diversion. Their insertion has been shown to be safe in premature infants and can even be done at the bedside if necessary. Reference Flanders, Kimmel and Lang15 Studies have shown that VADs may reduce morbidity, mortality and shunt dependence in patients with PHVD. Reference Hudgins, Boydston and Gilreath17 Large multi-center studies have shown no significant difference in shunt requirement or infection rates between VAD or VSG shunts, Reference Wellons, Shannon and Holubkov8,Reference Wellons, Shannon and Kulkarni18 the two most commonly utilized surgical temporizing measures employed in a North American setting. Others, most notably those in European practice, have explored the utility of neuro-endoscopic lavage and intraventricular fibrinolytic therapy in preterm infants with PHH and found that these treatment modalities may also be effective. Reference Schulz, Bührer, Pohl-Schickinger, Haberl and Thomale19–Reference Whitelaw, Jary and Kmita21 Unfortunately, there is still a lack of Level 1 evidence demonstrating superiority of any of these modalities, so as of now, the choice of temporizing measure for CSF diversion in these patients remains institution-dependent, with significant practice variation being observed. Reference Mazzola, Choudhri and Auguste22
Timing of intervention for PHVD
A 2002 study retrospectively analyzed data from five NICUs in the Netherlands and found that early intervention via either serial lumbar punctures or VADs for preterm IVH with ventricular dilatation based on neurosonographic criteria (PHVD) led to significantly better neurocognitive outcomes and a reduced VPS requirement. Reference de Vries, Liem and van Dijk5 Another multi-center observational study in 2018 found that in preterm infants with IVH and ventricular dilatation, those with early intervention guided largely by ventricular size criteria, even when eventually requiring a shunt, had better neurodevelopmental outcomes than those who had late intervention based largely on clinical symptomatology. Reference Leijser, Miller and Van Wezel-Meijler7 A subsequent randomized controlled trial on 126 preterm babies with PHVD by the ELVIS study group found similar rates of VP shunt placement, regardless of whether temporizing treatment was instituted at a p > 97 or p > 97 + 4 mm threshold. Reference de Vries, Groenendaal and Liem10 However, on post-hoc analysis in 113 of these patients, earlier intervention was associated with lower odds of death or severe neurodevelopmental disability. Reference Cizmeci, Groenendaal and Liem6
Rationale for the choice of surgical intervention threshold and for standardizing an early intervention pathway for PHVD management
While established guidelines exist for some aspects of the management of these patients, Reference Whitelaw, Jary and Kmita21 there is no consensus regarding the critical question of the optimal timing for instituting a temporizing intervention – whether it should be introduced at the onset of PHVD (when infants are generally asymptomatic) or reserved for when the infant exhibits symptoms/signs of elevated intracranial pressure (usually with a significantly larger ventricular size at intervention).
After reflecting on our historical practice, we decided to adopt the p97 + 4 mm threshold in our early intervention protocol for several reasons. Firstly, this choice of surgical intervention threshold would represent an incremental change from our historical care of these infants; most of the neurosurgeons felt that a move to VAD placement once the p97 threshold was crossed would be an “aggressive” shift in practice, especially since the focus of protocol implementation was primarily safety rather than efficacy. In addition, concern was raised about the potential of intervention at the p97 threshold leading to possibly unnecessary intervention (transport from another facility or surgery) in a subset of patients. It was ultimately decided to adopt the p97 + 4 mm threshold at the outset, assess shunting and neurocognitive outcomes in a future iteration of this work and revise the surgical intervention threshold based on review of our quality improvement data.
Based on evolving literature extolling the virtues of standardized, protocolized care in improving patient outcomes, there was universal agreement that we should embark on standardizing the care of these infants regardless of the surgical intervention threshold that was selected. Reference Chu, Jensen and Holubkov16,Reference Kestle, Riva-Cambrin and Wellons23–Reference Kestle, Holubkov and Douglas Cochrane25 This shared philosophy amongst neonatology and neurosurgery providers certainly contributed positively to the successful and sustained implementation of this care protocol at our site, and is felt to be an important prerequisite for successful protocol implementation at other sites.
Evaluation of our early intervention protocol
Our new early intervention management pathway for preterm IVH with ventricular dilatation has been implemented for over four years now and has shown itself to be sustainable and safe. It has led to a reduction in ventricle size at the time of VAD placement, as would be expected given the fact that we shifted our criteria for intervention to earlier in the continuum from the onset of ventricular dilatation to the development of clinical symptoms of elevated intracranial pressure. However, there is still room for improvement, given the fact that the median Levene index at the time of VAD placement was still well above the p97 + 4 mm threshold for most infants, the median anterior horn width was still indicative of severe ventricular dilatation at VAD placement (noting that anterior horn width was not a criterion for intervention in our protocol) and the fact that only 25% of infants had a VAD inserted within what was felt to be a reasonable 3-day target. Although some factors, such as the time to transport critically ill premature infants across the vast geographic expanse of our province, may be difficult to influence, other factors, such as improved access to the operating room, may be more amenable to data-driven change, and we will certainly be advocating for such. For the safety of the neonate, we opted to perform the procedure during standard weekday business hours in the operating room (as opposed to the NICU) because of the better availability of personnel should an issue arise during transport to/from the operating room, or in the operating room itself. Recognizing that this practice may have contributed to VAD placement delays in a proportion of neonates, we have recently liberalized our philosophy to accommodate off-hours VAD placement.
With respect to compliance measures, we were pleased to see that there was 92% compliance with ultrasound scans mandated by the protocol. We did, however, note that our fairly rigorous ultrasound surveillance program led to a change in only one patient’s management by detecting the presence of new IVH and causing a temporary cessation of VAD taps until the hemorrhage resolved. Based on this observation, we are considering decreasing surveillance ultrasound frequency to 3 times a week. This less-intensive use of ultrasound resources may facilitate protocol implementation at other sites.
Median compliance with tapping of the VADs by NICU providers was almost 89%, which was quite satisfactory. The majority of the non-compliant taps were due to inadequate volume being aspirated. These were not considered protocol deviations, since collapse of the ventricles often results in inability to aspirate more than a certain amount of CSF. Less frequent notable occurrences, namely missed indicated VAD aspirations and unindicated VAD taps, both of which were considered protocol deviations, occurred less than 2% of the time. Ongoing provider education campaigns should minimize the incidence of protocol deviations as we as we continue refining protocol implementation at our site.
We were disappointed with the relatively small number of lumbar punctures that were conducted under the new protocol. In the next iteration of the protocol, we plan to be more prescriptive regarding lumbar punctures rather than recommending that they be performed “at the discretion of the treating neonatologist,” acknowledging that there may be a population of infants who achieve ventricular size control through lumbar punctures alone, thereby eliminating the requirement for VAD placement.
Study limitations
Due to the relatively low incidence of PHVD at our institution, the number of patients in our series is small, leading to underpowered statistical analysis, although this was not the primary aim of this work. The data that we analyzed were retrospectively collected and therefore prone to the inherent deficiencies thereof. For example, during the first year of implementation of the new management pathway, information on individual VAD taps and the volume of CSF aspirated were not uniformly collected; this led to missing tap data for the first five out of the 16 patients treated under the new protocol. Nevertheless, this study documents an initial North American experience with the implementation of an early intervention pathway based solely on sonographic ventricular size criteria for the management of PHVD and provides an important foundational template for others seeking to replicate this experience at their site.
Conclusion
Implementation of an early intervention management pathway for PHVD has led to a meaningful reduction in ventricle size at the time of VAD placement and the time interval between detection of PHVD and insertion of a VAD compared to institutional historical controls. The management protocol was found to be safe and sustainable in our quaternary care setting. After reviewing process and compliance measures, we have identified several areas for improvement and are currently working on a new iteration of this management pathway.
Our experience demonstrates that a new standardized management protocol for the treatment of infants with PHVD can be successfully implemented if all relevant stakeholders provide input at the outset and all stakeholders demonstrate an ongoing commitment to improving care through standardization. We anticipate that the data and discussion presented here may be of use to others trying to achieve the same aim at their institution.
Author contributions
Conceived and designed the study: MST
Data collection: ME, APA, AW, IW
Statistical analysis: APA, ME, MST
Preparation of the manuscript: APA
Critical revision of the manuscript for important intellectual content: MST, APA, AS, FH.
Funding statement
No funding was received for this study.
Competing interests
The authors report no conflicts of interest.
Open access
