📈 Analyse your 🥘 food-borne outbreak data



This data frame is expected to have consistent coding of exposures.

For example, all exposures should be consistently coded as 'yes' or 'no', 'True' or 'False' or other labels with two levels. 


            

          

        

        

          

            

              
            

            

              

              

              It is worthwhile checking your data for duplicates. 
        Choose a unique identifier, or combination of identifiers to check for possible duplicate entries.
              

              

              If you find duplicate entries, it is worthwhile looking in the original file.
              

              

              

              

                

                

                

              

            

          

        

        

          

            
Plot of spreadsheet with variables displayed along horizontal axis.

              Black cells indicating missing observations. 


          

          

            
Check for suspicious patterns that may indicate a 

              need to recheck data in spreadsheet 

              for data entry errors.


          

          

            
The plot sorts variables (columns) with those with the most missing 

             on the left and vice versa

          

          

            

              

            

          

          

            

              

            

          

          

            
This plot shows the types of data in your input file. 


          

          

            
'Factor' means categorical variable. 


          

          

            
'Numeric' means continuous. 


          

          

            
Grey cells indicate missing observations.


          

          

            
The plot sorts the observations (rows) with those with the most missing on the top 

             and vice versa.

          

          

            

              

            

          

        

        

          

            

              
            

            

              I prefer the use of odds ratios, over relative risks in analysing food borne outbreak data.
              Odds ratios are not bounded by the prevalence of the disease in the unexposed cases.
              Relative risks, conversely, are bounded by the prevalence of disease in the unexposed.
              

              

              For multi-level exposure variables (3-5 levels), each non-reference level is compared to the selected reference (baseline) level.
              The PAR is estimated once per variable using all non-reference levels combined.
              

            

          

        

        

          

            

              
            

            

              Euler diagrams summarizing exposure-disease relationships.