Feasibility of Combining Common Data Elements Across Studies to Test a Hypothesis. Issue 3 (23rd February 2017)

• Record Type: Journal Article
• Authors: Corwin, Elizabeth J.; Moore, Shirley M.; Plotsky, Andrea; Heitkemper, Margaret M.; Dorsey, Susan G.; Waldrop‐Valverde, Drenna; Bailey, Donald E.; Docherty, Sharron L.; Whitney, Joanne D.; Musil, Carol M.; Dougherty, Cynthia M.; McCloskey, Donna J.; Austin, Joan K.; Grady, Patricia A.
• Is Part Of: Journal of nursing scholarship. Volume 49:Issue 3(2017)
• Page Start: 249
• Page End: 258
• Publication Date: 2017-02-23
• Subjects: Informatics -- case studies -- meta‐analysis/data pooling -- common data elements -- data repository -- nursing science; Nursing -- Periodicals; Nursing -- United States -- Periodicals; 610.73
• DOI: 10.1111/jnu.12287
• Languages: English
• ISSNs: 1527-6546

Abstract

Abstract: Purpose: The purpose of this article is to describe the outcomes of a collaborative initiative to share data across five schools of nursing in order to evaluate the feasibility of collecting common data elements (CDEs) and developing a common data repository to test hypotheses of interest to nursing scientists. This initiative extended work already completed by the National Institute of Nursing Research CDE Working Group that successfully identified CDEs related to symptoms and self‐management, with the goal of supporting more complex, reproducible, and patient‐focused research. Design: Two exemplars describing the group's efforts are presented. The first highlights a pilot study wherein data sets from various studies by the represented schools were collected retrospectively, and merging of the CDEs was attempted. The second exemplar describes the methods and results of an initiative at one school that utilized a prospective design for the collection and merging of CDEs. Methods: Methods for identifying a common symptom to be studied across schools and for collecting the data dictionaries for the related data elements are presented for the first exemplar. The processes for defining and comparing the concepts and acceptable values, and for evaluating the potential to combine and compare the data elements are also described. Presented next are the steps undertaken in the second exemplar to prospectively identify CDEs and establish the data dictionaries. Methods for common measurement and analysis strategies are included. Findings: Findings from the first exemplar indicated that without plans in place a priori to ensure the ability to combine and compare data from disparate sources, doing so retrospectively may not be possible, and as a result hypothesis testing across studies may be prohibited. Findings from the second exemplar, however, indicated that a plan developed prospectively to combine and compare data sets is feasible and conducive to merged hypothesis testing. Conclusions: Although challenges exist in combining CDEs across studies into a common data repository, a prospective, well‐designed protocol for identifying, coding, and comparing CDEs is feasible and supports the development of a common data repository and the testing of important hypotheses to advance nursing science. Clinical Relevance: Incorporating CDEs across studies will increase sample size and improve data validity, reliability, transparency, and reproducibility, all of which will increase the scientific rigor of the study and the likelihood of impacting clinical practice and patient care.