Big Data Research in Neurosurgery: A Critical Look at this Popular New Study Design

Neurosurgery. 2018 May 1;82(5):728-746. doi: 10.1093/neuros/nyx328.

Abstract

The use of "big data" in neurosurgical research has become increasingly popular. However, using this type of data comes with limitations. This study aimed to shed light on this new approach to clinical research. We compiled a list of commonly used databases that were not specifically created to study neurosurgical procedures, conditions, or diseases. Three North American journals were manually searched for articles published since 2000 utilizing these and other non-neurosurgery-specific databases. A number of data points per article were collected, tallied, and analyzed.A total of 324 articles were identified since 2000 with an exponential increase since 2011 (257/324, 79%). The Journal of Neurosurgery Publishing Group published the greatest total number (n = 200). The National Inpatient Sample was the most commonly used database (n = 136). The average study size was 114 841 subjects (range, 30-4 146 777). The most prevalent topics were vascular (n = 77) and neuro-oncology (n = 66). When categorizing study objective (recognizing that many papers reported more than 1 type of study objective), "Outcomes" was the most common (n = 154). The top 10 institutions by primary or senior author accounted for 45%-50% of all publications. Harvard Medical School was the top institution, using this research technique with 59 representations (31 by primary author and 28 by senior).The increasing use of data from non-neurosurgery-specific databases presents a unique challenge to the interpretation and application of the study conclusions. The limitations of these studies must be more strongly considered in designing and interpreting these studies.

MeSH terms

  • Big Data*
  • Databases, Factual*
  • Humans
  • Neurosurgical Procedures*