Factors incorporated into future survival estimation among Europeans


1. Introduction


The supplemental .zip archive for this publication contains the following files:

 - Demographic Research dataset.csv (comma-delimited .csv file. It includes SHARE data)
 - Life tables.csv (comma-delimited .csv file.It includes HMD data)

A more detailed discription of the data and the estimation process is presented below.



2. Data


The dataset is from the wave 6 of the Survey of Health, Ageing and Retirement in Europe (SHARE) whereas life tables are downloaded from Human Mortality Database.

Note that individuals younger that 50 years old should be excluded from the analysis.


The variables included in the dataset are:

- Mergeid
- Country of residence (1:Austria;...; 18:Slovenia)
- Subjective survival probability (as a percentage)
- Objective survival probability (as a percentage)
- Distance of SSPs from OSPs at 10% tolerance level (1:SSP=OSP;2: SSP>OSP, 3: SSP<OSP)
- Distance of SSPs from OSPs at 15% tolerance level (1:SSP=OSP;2: SSP>OSP, 3: SSP<OSP)
- Distance of SSPs from OSPs at 20% tolerance level (1:SSP=OSP;2: SSP>OSP, 3: SSP<OSP)
- Prediction interval (in years)
- Age (in years)
- Gender (1:Male;2:Female)
- Education (1:Primary;...;4;Tertiary)
- Equivalised income quartiles (1:1st quartile;...;4:4th quartile)
- Self-rated health (1:Excellent;...;5:Poor)
- BMI groups (1:Underweight;2:Normal;3:Overweight;4:Obese)
- Number of chronic diseases
- Number of ADLs
- Ever smoked daily
- Physical inactivity indicator (1: Never do vigorous exericise;0:other)
- writing skills score
- Orientation in time score
- Numeracy score
- Memory score
- EURO-D depression index
- Marital status (1:widowed;2:divorced;3:never married;4:separated;5:partnership;6:married)
- Number of children
- Number of times recieved help from others
- Life satisfaction score
- Quality of life CASP index
- Grip strength
- Peak flow



3. Algorithm for the calculation of Objective Survival Probabilities


The following algorithm was developed in Visual Basic for Applications (VBA) in order to calculate the objective survival probabilities for each individual.

	
*---------------------------------------------Algorithm syntax----------------------------------------------------------------------------------*


Public Function ObjectiveSurvivalProbability( _
    		ByVal Age As Variant, ByVal PredictionInterval As Variant, ByVal country As String, _
    		ByVal gender As Integer) As Variant


	Dim thisGender, Identifier As String
	Dim PredictionAge, thisAge As Integer
	Dim SurvivalProbability, CummSurvivalProbability As Double


	CummSurvivalProbability = 1 'Initialise the variable

	
	If PredictionInterval <> "" Then ' mortality rates depend on gender.

    		
		If gender = 1 Then


        		thisGender = "Males"
    
    		Else

        		thisGender = "Females"
    
    		End If
 
    	
	
		PredictionAge = Application.WorksheetFunction.Min(Age + PredictionInterval, 109)


    	
		For thisAge = Age To PredictionAge Step 1 ' go througg all ages in the prediction interval to calculate the OSP 


        		Identifier = "" & country & thisGender & thisAge
    
    
        		SurvivalProbability = _
            			1 - Application.WorksheetFunction.VLookup(Identifier, Life.Range("LifeTables"), 2, False)

        		CummSurvivalProbability = CummSurvivalProbability * SurvivalProbability

    		Next


    		ObjectiveSurvivalProbability = CummSurvivalProbability  ' Output the calculated OSP.
    
	Else

    		ObjectiveSurvivalProbability = ""   ' In case of missing inputs.
    
    
	End If

 
End Function

*--------------------------------------------------------------------------------------------------*


4. Sofrware & Estimation process


All statistical analyses run in SPSS 21. No additional code is required for this step.

The dependent variable for the multinomial models is "Distance of SSPs from OSPs at ...% tolerance level" (see section 2 above). The explanatory variables are included in the models
as provided in the 'Demographic Research dataset.csv' file.