An Analysis of Two Complex Surveys to Evaluate Dental Health Status Changes in North Carolina

MAKUC, DIANE M. An Analysis of Two Complex Surveys to Evaluate Dental Health Status Changes in North Carolina. (Under the direction of DENNIS B. GILLINGS.) Dental health status changes in North Carolina are investigated based on data obtained from two cross-sectional prevalence surveys carried out in 1960-63 and 1976. Edentulousness, periodontal disease, and oral hygiene status are used as measures of adult dental health status. Changes over time are evaluated for population subgroups based on age, race, sex, education, urban/rural residence, and geographic zone. The motivation for studying differences in dental disease prevalence among subgroups is to identify those groups with greatest need for improvement in dental health status and aid in establishing program priorities. Since the data arise from two complex surveys, statistical theory indicates that the analysis should take the survey designs into account. The two primary analysis methods employed are weighted least squares as developed by Koch, Freeman, and Freeman, and survey regression as described by Shah, Holt, and Folsom. Taylor series linearization is used for estimation of variances in both cases. Weighted least squares is used when only categorical independent variables are considered, and survey regression is appropriate when the response variable is continuous. All analyses are also carried out using analogous procedures which assume simple random sampling in order to investigate the frequency and magnitude of errors that would be made if the less complicated analysis approach were used. A worsening of dental health status over time was observed among blacks. Edentulousness and periodontal disease have increased, and oral hygiene status has deteriorated among the black population. There has been comparatively little change with respect to periodontal disease and oral hygiene status among whites, and edentu10usness has decreased between 1960-63 and 1976. As a result of these changes, in 1976 black adults have substantially worse oral hygiene status and periodontal disease than white adults. The comparison of analogous statistical procedures indicates that the simple random sampling results more closely approximate the complex survey results for the 1960-63 survey which had a less complicated design than the later survey.