ABSTRACT
OBJECTIVE: Is the simple mean of the costs per diabetes patient a suitable tool with which to compare care groups? Do the total costs of care per diabetes patient really give the best insight into care group performance? DESIGN: Cross-sectional, multi-level study. METHOD: The 2009 insurance claims of 104,544 diabetes patients managed by care groups in the Netherlands were analysed. The data were obtained from Vektis care information centre. For each care group we determined the mean costs per patient of all the curative care and diabetes-specific hospital care using the simple mean method, then repeated it using the 'generalized linear mixed model'. We also calculated for which proportion the differences found could be attributed to the care groups themselves. RESULTS: The mean costs of the total curative care per patient were 3,092 - 6,546; there were no significant differences between care groups. The mixed model method resulted in less variation (2,884 - 3,511), and there were a few significant differences. We found a similar result for diabetes-specific hospital care and the ranking position of the care groups proved to be dependent on the method used. The care group effect was limited, although it was greater in the diabetes-specific hospital costs than in the total costs of curative care (6.7% vs. 0.4%). CONCLUSION: The method used to benchmark care groups carries considerable weight. Simply stated, determining the mean costs of care (still often done) leads to an overestimation of the differences between care groups. The generalized linear mixed model is more accurate and yields better comparisons. However, the fact remains that 'total costs of care' is a faulty indicator since care groups have little impact on them. A more informative indicator is 'costs of diabetes-specific hospital care' as these costs are more influenced by care groups.
ABSTRACT
OBJECTIVES: A precise knowledge of the risk factors for childhood and adolescent smoking is crucial for the development of appropriate preventive measures. This study investigated current smoking prevalence and the social and regional correlates for smoking among minors (children and adolescents aged 12-17 years) in Germany. METHODS: Bivariate data analysis was performed on the basis of a representative national cross-sectional study performed in 2004, and multivariable logistic regression models were calculated separately for boys and girls. All correlates identified as significant in the bivariate model were used in the multivariable analysis. STUDY DESIGN: The database used in this research was from the study 'Drug Affinity of Young People in the Federal Republic of Germany 2004', with approximately 1298 children and adolescents aged 12-17 years. RESULTS: Twelve percent of male and 9% of female adolescents in Germany reported that they are habitual smokers, and 12% of male and 13% of female adolescents reported that they are occasional smokers. Multivariable data analysis shows that living in a large city is protective for adolescents in terms of local disparities. The educational level of the respondents also correlates significantly with smoking behaviour. The percentage of adolescent smokers is lowest among those with a high level of education. The presence of smokers in the household is associated with a significantly higher prevalence of smoking among adolescents compared with those growing up in a non-smoking household. CONCLUSION: Smoking is a major public health problem among German children and adolescents. Control measures must tackle the structural and social pressures that shape smoking behaviour during childhood.