Analysis of large databases in vascular surgery

Abstract

Large databases can be a rich source of clinical and administrative information on broad populations. These datasets are characterized by demographic and clinical data for over 1000 patients from multiple institutions. Since they are often collected and funded for other purposes, their use for secondary analysis increases their utility at relatively low costs. Advantages of large databases as a source include the very large numbers of available patients and their related medical information. Disadvantages include lack of detailed clinical information and absence of causal descriptions. Researchers working with large databases should also be mindful of data structure design and inherent limitations to large databases, such as treatment bias and systemic sampling errors. Withstanding these limitations, several important studies have been published in vascular care using large databases. They represent timely, "real-world" analyses of questions that may be too difficult or costly to address using prospective randomized methods. Large databases will be an increasingly important analytical resource as we focus on improving national health care efficacy in the setting of limited resources.

Effect of Socioeconomic Status on Inpatient Mortality and Use of Postacute Care After Subarachnoid Hemorrhage

“…19 The limitations of using large administrative data sets for health services research have been highlighted elsewhere. 25,26 Of particular relevance is our inability to adjust for case severity. Although low SES groups have been shown to present with more severe stroke in comparison with higher SES groups, 27 there is inconsistent evidence in support of a causal effect of stroke severity on incomemortality associations after ischemic stroke.…”

Pediatric urolithiasis associated with acute gastroenteritis: an inpatient database study in Japan

“…This discrepancy between studies can be explained by the differences in the study design. Patients who were reported in the previous literature represented instructive cases requiring intensive management at specialty hospitals, while our administrative database research patients are likely to represent the patient population in daily clinical practice [16,19].…”

So You Have a Research Idea

“…Each database was assessed for the availability of outcomes and/or complications. We also recorded whether the database collected longitudinal surgeon or patient information that may allow for tracking of surgeon procedure 11 Limited data set of deidentified 5 percent random sample ($5000) or full data set of identifiable data ($25,000). 12 Commission on Cancer accreditation is free.…”

“…Further, the scope of large databases often circumvents the selection bias inherent in case series from single institutions 10 and increases external validity. 11 Although these databases capture a great deal of information, there is enormous heterogeneity between databases. For example, differences may include the number and type of variables, processes of data extraction, and represented patient populations.…”