Data Analysis and Interpretation
A statistical working group considered a number of issues with respect to data analysis and interpretation. These were discussed at two training sessions: December 2011 in Brussels and March 2012 in Copenhagen.
To enable the compiling of a single European database, all national DEMOCOPHES databases had to be set up in the same way. Detailed instructions were provided in a centrally developed codebook and in guidelines for quality control. A process was set up to check the quality of the national databases and to merge them into one European database. The QC program screened the info from all questionnaires and the biomarker data.
The program checked e.g. if each questionnaire contained all variables, if these variables had the correct format or if there were numbers outside the pre-set ranges. The errors reported were checked and if necessary input was asked from the fieldworkers which got back to the paper questionnaires.
For the biomarker data, additional checks were done for the values that were indicated to be below or above the limit of quantification (LOQ). A table with the extreme outliers was displayed. Outliers were confirmed by re-analysis in the analysis labs
A plan for statistical analysis was applied to the biomarker and questionnaire data and guidelines were given on interpretation of the results, including:
- Calculation of response rate which was obtained on the basis of the inventory made during the recruitment on the number of subjects that were invited, that refused to take part, that were not eligible, …
- Description of the general characteristics of the study population (i.e. age, gender, anthropometry, social class, etc).
- Statistical analysis of the biomarkers which includes:
- Basic data of biomarkers: basic statistical values are calculated for each biomarker. This is done separately for mothers and children, both on the total study population and on subgroups of the population.
- Multiple regression models.
- Comparison between data of mothers and children.
- Comparison with used health-based guidance values : the proportion of participants above and below the guidance values was calculated.
Achievements during the data analysis and interpretation
- All countries were able to use the delivered QC programmes. Some unexpected problems occurred, e.g. with relatively high LOQ values and with rounding of the values, which required additional steps.
- All countries performed their statistical analysis and delivered their questionnaire and biomarkers data to the European database.
- At European level, the statistical analysis was performed in a similar way as in the individual countries.