A4-4: Visualizing VDW Lab Results Data: Why You Should, and How You Can - Easily!

  • September 2013,
  • 177.1;
  • DOI: https://doi.org/10.3121/cmr.2013.1176.a4-4

Abstract

Background/Aims One of the most difficult tasks facing the VDW lab results file implementer is figuring out which records from local lab data should be included in their VDW file, and which should be left out. These decisions frequently hinge on neat points of clinical science that are often outside the expertise of the programmer. We will describe a set of graphics that implementers can use to shed light on these decisions, useful during implementation, and afterwards as indication of data quality and even as documentation.

Methods SAS’ new Statistical Graphics procedures allow unprecedented control and ease of use in the creation of descriptive graphs and charts. The ODS Graphics Designer utility, paired with the SGDESIGN procedure make it easy to create a single image composed of multiple different graphs, each of which can use its own dataset. These tools allowed us to create something of a “data report card” for each VDW test_type, depicting: (1) number of result records over time; (2) number of result values that are numeric vs. character, stratified by the local lab codes used; 3) distributions of numeric result values, by unit; (4) number of numeric result values, by unit and local lab code; (5) values of character results by local lab code. The report card is produced by a standard VDW program available from the author, which can be run at any VDW lab results-implementing site with access to SAS version 9.2 or greater.

Results The graphics produced by this program allow both implementers and end-users to evaluate at a glance, how cohesive the data from various different local codes are, how much data there is, how it waxes and wanes over time, whether the values are of the expected types, and whether units and character values are within valid values. Being able to depict all of this disparate information in a single, compact display allows users to glean insights that, for example, viewing series of larger graphs separately would not afford.

Conclusions The author has found this program an extremely useful tool in implementing VDW labs by quickly pinpointing misclassified local codes. Other sites surely will also.

Loading
  • Print
  • Download PDF
  • Article Alerts
  • Email Article
  • Citation Tools
  • Share
  • Bookmark this Article

Keywords