Table 1

Population characteristics and the matching equation
Variable
 Mean Values
 Coefficient
 P-value
 Mean Value Matched Controls (SA<1%)
 Difference Between Cases And Matched Control P value
 Mean Value Unmatched Controls (SA<1%)







All
 Cases (SA>5%)




Number of Practices
 702
 39

140
523
Proportion of South Asian patients
 0.01
 0.10
 -
 -
 0
 <0.001
Carstairs
 -0.17
 2.3
 0.45
 <0.001
 2.6
 0.22
 -0.7
Census not good health
 98.9
 122
 0.11
 <0.001
 134
 0.12
 91.4
Census limiting long term illness
 98.8
 107
 -0.2
 <0.001
 121
 0.09
 95.2
Temporary residents/practice list size
 0.03
 0.01
 -3.23
 0.22
 0
 0.31
 0.05
Proportion of list eligible for road mileage payments
 0.11
 0
 -2.88
 <0.001
 0.01
 0.15
 0.19

Notes: Analysis weighted by list size. Dependent variable in the matching equation is the log-odds of the population proportion in South Asian ethnic groups. Adjusted R-squared for the model = 0.59* Figures for matched controls obtained by weighted average using kernel matching.

McLean International Journal for Equity in Health 2008 7:18   doi:10.1186/1475-9276-7-18

Open Data