Publication Details |
Reference Category | Journals |
Title (Primary) | Automatic re-coding of variable values for pooled meta-analysis of epidemiological data |
Author | Röder, S. |
Journal | Journal of Epidemiology and Community Health |
Year | 2004 |
Department | EXPOEPID |
Volume | 58 |
Issue | Supplement 1 |
Page From | 155 |
Language | englisch |
Abstract | Introduction: Epidemiological studies often use multiple datasets from different periods (such as longitudinal studies). For pooled meta-analysis of these datasets, it is important to concatenate them. If variables have different specifications over periods, then this concatenation is difficult, therefore a method for automatic re-coding of variable values from different periods or studies to match a common specification is needed. Method: The re-coding algorithm analyses the specification of each concatenated variable and harmonises them to a common specification. Furthermore, constraints such as missing value codes or values outside specification have to be considered. Afterwards, the harmonised variables will be used for further meta-analyses. Results: Using this algorithm, we are able to perform a meta-analysis on four studies with 149 variables and over 2500 cases over 12 periods. Furthermore, the error prone process of manually re-coding is now supported by software, and is thus much more reliable. Conclusions: The shown algorithm allows easier re-coding of variable specifications in complex data structures for pooled meta-analysis. The risk of errors in manually re-coding prior to statistical analysis can be significantly reduced. |
Persistent UFZ Identifier | https://www.ufz.de/index.php?en=20939&ufzPublicationIdentifier=4482 |
Röder, S. (2004): Automatic re-coding of variable values for pooled meta-analysis of epidemiological data J. Epidemiol. Community Health 58 (Supplement 1), 155 |