Machine learning predicts risk of cerebrospinal fluid shunt failure in children: a study from the hydrocephalus clinical research network

Object The Hydrocephalus Clinical Research Network (HCRN), which comprises 7 pediatric neurosurgical centers in North America, provides a unique multicenter assessment of the current outcomes of CSF shunting in nonselected patients. The authors present the initial results for this cohort and compare them with results from prospective multicenter trials performed in the 1990s. Methods Analysis was restricted to patients with newly diagnosed hydrocephalus undergoing shunting for the first time. Detailed perioperative data from 2008 through 2012 for all HCRN centers were prospectively collected and centrally stored by trained research coordinators. Historical control data were obtained from the Shunt Design Trial (1993–1995) and the Endoscopic Shunt Insertion Trial (1996–1999). The primary outcome was time to first shunt failure, which was determined by using Cox regression survival analysis. Results Mean age of the 1184 patients in the HCRN cohort was older than mean age of the 720 patients in the historical cohort (2.51 years vs 1.60 years, p < 0.0001). The distribution of etiologies differed (p < 0.0001, chi-square test); more tumors and fewer myelomeningoceles caused the hydrocephalus in the HCRN cohort patients. The hazard ratio for first shunt failure significantly favored the HCRN cohort, even after the model was adjusted for the prognostic effects of age and etiology (adjusted HR 0.82, 95% CI 0.69–0.96). Conclusions Current outcomes of shunting in general pediatric neurosurgery practice have improved over those from the 1990s, although the reasons remain unclear.

Download Full-text

Predictors of fast and ultrafast shunt failure in pediatric hydrocephalus: a Hydrocephalus Clinical Research Network study

Journal of Neurosurgery Pediatrics ◽

10.3171/2020.7.peds20111 ◽

2020 ◽

pp. 1-10

Author(s):

Jason S. Hauptman ◽

John Kestle ◽

Jay Riva-Cambrin ◽

Abhaya V. Kulkarni ◽

Samuel R. Browd ◽

...

Keyword(s):

Clinical Research ◽

A Priori ◽

Multivariable Analysis ◽

Mitigation Strategies ◽

Research Network ◽

Fast Time ◽

Shunt Failure ◽

Slit Ventricle Syndrome ◽

Clinical Research Network ◽

Shunt Placement

OBJECTIVEThe primary objective of this study was to use the prospective Hydrocephalus Clinical Research Network (HCRN) registry to determine clinical predictors of fast time to shunt failure (≤ 30 days from last revision) and ultrafast time to failure (≤ 7 days from last revision).METHODSRevisions (including those due to infection) to permanent shunt placements that occurred between April 2008 and November 2017 for patients whose entire shunt experience was recorded in the registry were analyzed. All registry data provided at the time of initial shunt placement and subsequent revision were reviewed. Key variables analyzed included etiology of hydrocephalus, age at time of initial shunt placement, presence of slit ventricles on imaging at revision, whether the ventricles were enlarged at the time of revision, and presence of prior fast failure events. Univariable and multivariable analyses were performed to find key predictors of fast and ultrafast failure events.RESULTSA cohort of 1030 patients with initial shunt insertions experienced a total of 1995 revisions. Of the 1978 revision events with complete records, 1216 (61.5%) shunts remained functional for more than 1 year, and 762 (38.5%) failed within 1 year of the procedure date. Of those that failed within 1 year, 423 (55.5%) failed slowly (31–365 days) and 339 (44.5%) failed fast (≤ 30 days). Of the fast failures, 131 (38.6%) were ultrafast (≤ 7 days). In the multivariable analysis specified a priori, etiology of hydrocephalus (p = 0.005) and previous failure history (p = 0.011) were independently associated with fast failure. Age at time of procedure (p = 0.042) and etiology of hydrocephalus (p = 0.004) were independently associated with ultrafast failure. These relationships in both a priori models were supported by the data-driven multivariable models as well.CONCLUSIONSNeither the presence of slit ventricle syndrome nor ventricular enlargement at the time of shunt failure appears to be a significant predictor of repeated, rapid shunt revisions. Age at the time of procedure, etiology of hydrocephalus, and the history of previous failure events seem to be important predictors of fast and ultrafast shunt failure. Further work is required to understand the mechanisms of these risk factors as well as mitigation strategies.

Download Full-text

This month at the NIH: NIDDK forms hepatotoxicity clinical research network

Gastroenterology ◽

10.1016/s0016-5085(03)01302-7 ◽

2003 ◽

Vol 125 (4) ◽

pp. 1008-1008

Author(s):

Stephen P. James

Keyword(s):

Clinical Research ◽

Research Network ◽

Clinical Research Network

Download Full-text

Rare Diseases Clinical Research Network

AWHONN Lifelines ◽

10.1111/j.1552-6356.2004.tb00171.x ◽

2004 ◽

Vol 8 (1) ◽

pp. 63-64

Keyword(s):

Clinical Research ◽

Rare Diseases ◽

Research Network ◽

Clinical Research Network

Download Full-text

Use of the Asthma Control Questionnaire by the Asthma Clinical Research Network

European Respiratory Journal ◽

10.1183/09031936.00054711 ◽

2011 ◽

Vol 38 (5) ◽

pp. 1242-1242

Author(s):

M. J. Walter ◽

M. Castro ◽

E. Israel ◽

C. A. Sorkness ◽

Keyword(s):

Clinical Research ◽

Asthma Control ◽

Research Network ◽

Asthma Control Questionnaire ◽

Clinical Research Network

Download Full-text

Treatment strategies for hydrocephalus related to Dandy-Walker syndrome: evaluating procedure selection and success within the Hydrocephalus Clinical Research Network

Journal of Neurosurgery Pediatrics ◽

10.3171/2020.11.peds20806 ◽

2021 ◽

pp. 1-9

Author(s):

Aaron M. Yengo-Kahn ◽

John C. Wellons ◽

Todd C. Hankinson ◽

Jason S. Hauptman ◽

Eric M. Jackson ◽

...

Keyword(s):

Clinical Research ◽

Treatment Success ◽

Treatment Strategies ◽

Research Network ◽

Csf Shunt ◽

Third Ventriculostomy ◽

Separate Analysis ◽

Complication Rates ◽

Primary Analysis ◽

Clinical Research Network

OBJECTIVE Treating Dandy-Walker syndrome–related hydrocephalus (DWSH) involves either a CSF shunt-based or endoscopic third ventriculostomy (ETV)–based procedure. However, comparative investigations are lacking. This study aimed to compare shunt-based and ETV-based treatment strategies utilizing archival data from the Hydrocephalus Clinical Research Network (HCRN) registry. METHODS A retrospective review of prospectively collected and maintained data on children with DWSH, available from the HCRN registry (14 sites, 2008–2018), was performed. The primary outcome was revision-free survival of the initial surgical intervention. The primary exposure was either shunt-based (i.e., cystoperitoneal shunt [CPS], ventriculoperitoneal shunt [VPS], and/or dual-compartment) or ETV-based (i.e., ETV alone or with choroid plexus cauterization [CPC]) initial surgical treatment. Primary analysis included multivariable Cox proportional hazards models. RESULTS Of 8400 HCRN patients, 151 (1.8%) had DWSH. Among these, the 102 patients who underwent shunt placement (79 VPSs, 16 CPSs, 3 other, and 4 multiple proximal catheter) were younger (6.6 vs 18.8 months, p < 0.001) and more frequently had 1 or more comorbidities (37.3% vs 14.3%, p = 0.005) than the 49 ETV-treated children (28 ETV-CPC). Fifty percent of the shunt-based and 51% of the ETV-based treatments failed. Notably, 100% (4/4) of the dual-compartment shunts failed. Adjusting for age, baseline ventricular size, and comorbidities, ETV-based treatment was not significantly associated with earlier failure compared with shunt-based treatment (HR for failure 1.32, 95% CI 0.77–2.26; p = 0.321). Complication rates were low: 4.9% and 6.1% (p = 0.715) for shunt- and ETV-based procedures, respectively. There was no difference in survival between ETV-CPC– and ETV-based treatment when adjusting for age (HR for failure 0.86, 95% CI 0.29–2.55, p = 0.783). CONCLUSIONS In this North American, multicenter, prospective database review, shunt-based and ETV-based primary treatment strategies of DWSH appear similarly durable. Pediatric neurosurgeons can reasonably consider ETV-based initial treatment given the similar durability and the low complication rate. However, given the observational nature of this study, the treating surgeon might need to consider subgroups that were too small for a separate analysis. Very young children with comorbidities were more commonly treated with shunts, and older children with fewer comorbidities were offered ETV-based treatment. Future studies may determine preoperative characteristics associated with ETV treatment success in this population.

Download Full-text

The Diabetic Retinopathy Clinical Research Network (DRCR.net) and Its Contributions to the Treatment of Diabetic Retinopathy

Ophthalmic Research ◽

10.1159/000502779 ◽

2019 ◽

Vol 62 (4) ◽

pp. 225-230 ◽

Cited By ~ 6

Author(s):

Jennifer K. Sun ◽

Lee M. Jampol

Keyword(s):

Diabetic Retinopathy ◽

Clinical Research ◽

Research Network ◽

Clinical Research Network

Download Full-text

Implementing a hash-based privacy-preserving record linkage tool in the OneFlorida clinical research network

JAMIA Open ◽

10.1093/jamiaopen/ooz050 ◽

2019 ◽

Vol 2 (4) ◽

pp. 562-569 ◽

Cited By ~ 3

Author(s):

Jiang Bian ◽

Alexander Loiacono ◽

Andrei Sura ◽

Tonatiuh Mendoza Viramontes ◽

Gloria Lipori ◽

...

Keyword(s):

Clinical Research ◽

Open Source ◽

Real World ◽

Record Linkage ◽

Gold Standard ◽

Privacy Preserving ◽

Research Network ◽

Clinical Research Network ◽

Real World Setting ◽

Hash Codes

Abstract Objective To implement an open-source tool that performs deterministic privacy-preserving record linkage (RL) in a real-world setting within a large research network. Materials and Methods We learned 2 efficient deterministic linkage rules using publicly available voter registration data. We then validated the 2 rules’ performance with 2 manually curated gold-standard datasets linking electronic health records and claims data from 2 sources. We developed an open-source Python-based tool—OneFL Deduper—that (1) creates seeded hash codes of combinations of patients’ quasi-identifiers using a cryptographic one-way hash function to achieve privacy protection and (2) links and deduplicates patient records using a central broker through matching of hash codes with a high precision and reasonable recall. Results We deployed the OneFl Deduper (https://github.com/ufbmi/onefl-deduper) in the OneFlorida, a state-based clinical research network as part of the national Patient-Centered Clinical Research Network (PCORnet). Using the gold-standard datasets, we achieved a precision of 97.25∼99.7% and a recall of 75.5%. With the tool, we deduplicated ∼3.5 million (out of ∼15 million) records down to 1.7 million unique patients across 6 health care partners and the Florida Medicaid program. We demonstrated the benefits of RL through examining different disease profiles of the linked cohorts. Conclusions Many factors including privacy risk considerations, policies and regulations, data availability and quality, and computing resources, can impact how a RL solution is constructed in a real-world setting. Nevertheless, RL is a significant task in improving the data quality in a network so that we can draw reliable scientific discoveries from these massive data resources.

Download Full-text