HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Naleway, A. L. Articles by Mullooly, J. P. Search for Related Content PubMed Articles by Naleway, A. L. Articles by Mullooly, J. P.

Abstract C-B3-01: Data Quality Assessment of the Vaccine Safety Datalink Dynamic Data Files

Allison L. Naleway, PhD, Kaiser Permanente Center for Health Research; Lois Drew, Kaiser Permanente Center for Health Research; Brad Crane, MS, Kaiser Permanente Center for Health Research; James Baggs, PhD, Centers for Disease Control and Prevention; Eric Weintraub, Centers for Disease Control and Prevention; John P. Mullooly, PhD, Kaiser Permanente Center for Health Research

Abstract

To enhance our ability to rapidly detect adverse events following immunization, the Vaccine Safety Datalink (VSD) developed dynamic data files (DDF). The DDF are created through weekly extracts of selected electronic medical records data at participating VSD HMOs. However, the timeliness and completeness of these files depend largely on the data storage capacity and infrastructure at each of the HMOs. VSD Data Managers produce a series of weekly, monthly, and annual reports to identify patterns that might signal a data quality problem in the DDF. All sites collect multiple weeks of event data during the weekly data extracts. To assess data quality, we can examine summaries of any particular week of event data repeated over multiple extracts. One data quality report summarizes the number of observations in a dataset for a given event week across multiple extracts, and calculates the percent change in observations across extracts. We used this report to estimate the number of weeks (or ‘lag time’) until the diagnosis and vaccination data at each HMO become relatively stable in the DDF. For each event week, we estimated the lag time until stability for outpatient diagnoses, inpatient diagnoses, and vaccinations in the DDF. Stability was defined as four consecutive extracts with less than 1% change (positive or negative) from the previous week. We created a set of criteria to define anomalous extract weeks and then imputed the percent change for these anomalous weeks in our estimation of lag time. Anomalous weeks included extracts in which datasets were not updated (i.e., the percent change across event weeks was 0%) and extracts in which the percent change across all event weeks was equal to 5%. The average lag time for outpatient diagnoses ranged from 3 to 19 weeks, and the average lag time for inpatient diagnoses ranged from 3 to 25 weeks across HMOs. The average lag time for vaccinations ranged from 2 to 26 weeks. Our findings suggest that the timeliness of diagnosis and vaccination data varies across the HMOs. In spite of this limitation, the DDF have proven to be a useful tool for rapid vaccine adverse event detection. Continuous data quality assessment of the DDF enables problems to be identified rapidly and either corrected or properly documented.