1 / 14

The Dutch RIF

The Dutch RIF. Experiences and possibilities. Introduction. RIF implementation part of Smarhagt project Development of toolkit for small area health analysis Comparison with more complex methods using R, Winbugs Cluster analysis. Practical issues.

gamada
Download Presentation

The Dutch RIF

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The Dutch RIF Experiences and possibilities

  2. Introduction • RIF implementation part of Smarhagt project • Development of toolkit for small area health analysis • Comparison with more complex methods • using R, Winbugs • Cluster analysis

  3. Practical issues • RIVM software standards versus RIF requirements • First test: Scottish Lip Cancer data set • Second test: existing RIVM data set • hospital admissions (icd, 2001-2004) • postal code areas (surrounding Schiphol Airport) • age, sex, SES, ethnicity • aircraft noise exposure

  4. Risk Analysis in RIF • Focuses on relation between exposure and RR • Exposure to be available as: • Location point source • Dispersion map • Exposure levels per small area (which we use) • Maximum 7 exposure categories • Test for heterogeneity and linear trend • Graphs and spreadsheet of RR by exposure • GIS required to interpret location/dispersion from map

  5. Risk Analysis Exposure Data

  6. Risk Analysis Result

  7. Disease Mapping in RIF • Map based on number of cases per PC4 area • Number of residents per PC4 area differs • Resulting in a difference in precision of the numbers • ‘1 out of 10’ is less precise than ‘100 out of 1000’ • This causes spurious outliers on the map • Solution: smoothing -> see next sheet • In which imprecision determines severity

  8. Smoothing in RIF • Calculates mean number of cases in whole study area • From which follows expected number of cases per PC4 area • Compares this to actual number of cases in PC4 area • Adjusts differences according to numbers of residents: • many residents – smaller adjustment • few residents – bigger adjustment • Smoothing based purely on statistical grounds • Spatial patterns in disease may be lost in smoothing! • ‘Empirical Bayes model’

  9. Smoothing from RIF in Winbugs • Uses information from neighbouring PC4 areas • To calculate local average number of cases for PC4 area • Also gives an expected number of cases for PC4 area • Compares to the actual number of cases in PC4 area • Also adjusts differences according to number of residents • This is spatial smoothing • Spatial patterns in disease are maintained! • ‘Fully Bayes model’ (Besag-York-Mollië)

  10. Disease Mapping Result

  11. Disease Mapping Result

  12. Disease Mapping Result

  13. Health Registration Data RIF Rapid Inquiry Facility Environmental Data Population Data Disease Mapping Risk Analysis Geographical Data Covariate Data (f.i. SES)

  14. Data obstacles • Postal code areas vary, no useful hierarchy • Privacy rules limit resolution for health data • Area size limits • Accuracy of exposure data • Usefulness of point source data • Use of modified ICD9 for Dutch registry • Data set size limit of 2 Gb • Consensus on / availability of exposure maps

More Related