Skip to main content

Table 2 Poisson multilevel model results exploring socioeconomic interactions with age and gender, and geographic variations in AMI incidence rates

From: Increasing socioeconomic inequalities in first acute myocardial infarction in Scotland, 1990–92 and 2000–02

 

1990–92

2000–02

Model

Estimate

p-value

ICC†

RR‡

Estimate

p-value

ICC†

RR‡

Full

        

Fixed

        

Gender

 

<0.001

   

0.001

  

DEPCAT

 

<0.001

   

<0.001

  

Age

 

<0.001

   

<0.001

  

Gender*DEPCAT

 

<0.001

   

0.027

  

Gender*Age

 

<0.001

   

<0.001

  

DEPCAT*Age

 

<0.001

   

<0.001

  

Gender*DEPCAT*Age

 

<0.001

   

0.005

  

Random

        

Council Area

0.003

0.003

0.29

1.24

0.002

0.001

0.38

1.19

Postcode Sector

0.007

<0.001

0.71

1.39

0.001

<0.001

0.62

1.13

Age/Sex group

1

   

1

   

Men

        

Fixed

        

DEPCAT

 

<0.001

   

<0.001

  

Age

 

<0.001

   

<0.001

  

DEPCAT*Age

 

<0.001

   

<0.001

  

Random

        

Council Area

0.008

0.001

0.51

1.42

0.002

0.002

0.55

1.19

Postcode Sector

0.008

<0.001

0.49

1.42

0.002

<0.001

0.45

1.19

Age/Sex group

1

   

1

   

Women

        

Fixed

        

DEPCAT

 

<0.0001

   

<0.001

  

Age

 

<0.0001

   

<0.001

  

DEPCAT*Age

 

<0.0001

   

<0.001

  

Random

        

Council Area

0.004

0.007

0.31

1.28

0.003

0.005

0.28

1.23

Postcode Sector

0.009

<0.001

0.69

1.45

0.014

<0.001

0.72

1.59

Age/Sex group

1

   

1

   
  1. † The Intraclass Correlation Coefficient (ICC) is the percentage of unexplained residual variance attributable to each higher level in the model.
  2. ‡ The Risk Ratio (RR) given at the higher level of local council area (LCA) compares the risk of incidence in a notional LCA lying on the 97.5th centile with a LCA on the 2.5th centile. The RR at the postcode sector (PS) level compares the risk of incidence in a notional PS (within a given LCA) lying on the 97.5th centile with a PS (within the same LCA) on the 2.5th centile.
  3. Note that coefficient estimates are not given for the fixed effects as these consist of numerous dummy variables.
Back to article page

BMC Public Health

ISSN: 1471-2458

Contact us