碩士 / 國立陽明大學 / 公共衛生研究所 / 101 / The Global Youth Tobacco Survey (GYTS) has been conducted in hundreds of countries for the past decade. In Taiwan, four and five waves of surveys were performed for the junior and senior high school students respectively. The GYTS is based on a clustered sampling scheme, and different scales of the smoking behavior have been collected. There are two types of methods in modeling clustered data: Random effect models and Marginal models. The estimates from these two types of models were not necessarily consistent. The purpose of this study was to compare different estimates of the effects of the 2009 tobacco control intervention on youngster smoking in Taiwan.
In our study, for binary outcomes, we compared the odds ratio estimated by logistic regression and the risk ratio by modify Poisson models with robust variance estimation. For ordinal and multinomial outcomes, we compared the estimates using marginal models by generalized estimating equations and those using generalized mixed models by numerical integration and pseudo-likelihood. Since the routines of estimating a multinomial logistic regression using generalized estimating equation are not available, a bootstrap method was applied for estimating the confidence intervals.
By different models and different estimate methods, the smoking rate of junior high school students before 2008 increased 5% in average per year. As the policy been conducted in 2009, in a year after, the smoking rate decreased 10% in average per year, and the test of slope difference is significant (P value is between 0.04 and 0.06). So, the effects of the 2009 tobacco control intervention was profound. Also, the results of ordinal outcome estimates have the same trend. The results of tobacco control intervention estimated by nominal outcome has more significant effect than the result of binary outcome. As expected, when the smoking rate is low enough, the odds ratio from logistic regression estimates and relative risk from the modified Poisson estimates is very close.
For the effect of tobacco control intervention in high school students, the smoking rate decrease only in night school students, however, the test of smoking rate difference is significant only on the night school’s girls (P value between 0.0001 and 0.07). On the other hand, the smoking rates in other school types’ students increased after 2009. The results were not under the expectation of the government. These indicated that the effects of the intervention in 2009 is more significant on junior than on senior high school students.
| Identifer | oai:union.ndltd.org:TW/101YM005058027 |
| Date | January 2013 |
| Creators | Yen-Ling Liu, 劉妍伶 |
| Contributors | I-Feng Lin, 林逸芬 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 154 |
Page generated in 0.0016 seconds