Comparison of SES measures for targeting dental services

Objective: The aim of the study was to examine targeting of dental services on the basis of the household, school and/or area SES. Methods: The cross-sectional study used a multi-stage, stratified (school and area) random sample. Data were collected during 2007/8 as part of the NSW child survey. Data about household SES were collected by parent-completed questionnaire. School SES data were obtained from the ‘MySchools' website. Postcodes were attributed SEIFA scores. Clinical data were collected by trained examiners. The social gradient in caries across household, school and area SES measures was examined. Logistical regression models for significant caries (SiC10 - top decile of caries experience – dmfs for 5-8 and DMFS for 6-12 year olds) were run against SES measures. The targeting accuracy of the models for SiC cases was examined using relative risk (RR) and population attributable risk (PAR%) for the SES measures alone and in combination. Results: Response rate was 66%, yielding 5,243 5-12 year olds. Children in the SiC10 had a mean dmfs score of 20.39 (18.97,21.81) and mean DMFS score 5.01 (4.70,5.32). In logistic regression modelling for SiC10, school, area and household SES were all significant. RR was highest for disadvantaged school (deciduous 1.93, permanent 1.78) and lowest for disadvantaged area (deciduous 1.63, permanent 1.43). Deciduous PAR%s were 15.8% (school), 13.7% (area) and 17.3% (low SES household) while permanent PAR%s were 13.6% (school), 10.0% (area) and 14.0% (household). Combining school and area resulted in a higher deciduous PAR% (24.0%) but lower PAR% for permanent (13.2%). Conclusion: Disadvantaged school had the greatest strength of association with SiC cases while low household SES was associated with the highest fraction of SiC cases. However the accuracy in targeting dental services for a directed vulnerable populations strategy on the basis of the household, school and/or area socio-economic characteristics was low.