Temporal changes in the prevalence of different diseases and potentially causal risk factors are usually moderate. Epidemiology provides the tools to measure the prevalence of childhood obesity and of potentially causal risk factors. Some of the limitations of these data are inherent to the type of study used. Types of Epidemiological Studies This section describes different study types applied in epidemiology research (table 3). A first approach to population specific issues is surveys. These are designed to assess the prevalence of diseases or their occurrence. The latter is usually called incidence rate. If such prevalence or incidence data are plotted over time and compared to the changes in other observations (exposures) over the same time period, this is conventionally called a correlation or ecological study. Such ecologic studies offer the opportunity to look for associations of exposure and disease on a collective level, e.g. differences in the prevalence rates of obesity and ownership of TV sets between countries or over time. No information can be obtained on an individual level. A simultaneous increase of the prevalence of overweight and the number of TV sets does not necessarily mean that overweight individuals have a TV set. As populations are only considered as a whole, adjustment for confounders (see below) is not possible. The major part of analytical epidemiological studies in obesity research is cross-sectional studies. These studies include all subjects of a population at a certain time point (or small time interval) and they have etiologic objectives [90], e.g. to assess the association of socio-economic status and obesity. Classical cross-sectional studies on adiposity are the study of Sonne-Holm and Sorensen [60] on draftees in Denmark and their socio-economic status or the investigation of Park on obesity in minorities [33]. As they measure exposure and disease at the same time point, it is not always evident that exposure precedes the outcome. The chicken-egg question may not be an issue for genuine traits of the persons studied such as genotypes whereas, for example, eating habits may have been changed in order to treat or prevent obesity. Cohort studies ascertain the exposure first and then determine the outcome – usually over long time periods. Cohort studies, also called longitudinal studies, are costly and time consuming as they have to follow-up individuals over decennial periods, and there is always an important loss of study participants due to mobility or lack of interest. Bias may result as a consequence of losses to follow-up related to either the exposure or outcome of interest. Such cohort studies might help to assess the influence of childhood obesity on the development of coronary heart disease in adulthood. Case-control studies are similar to cross-sectional studies in ascertaining exposure and outcome at the same time point. The difference is that they start with the cases and try to identify adequate controls to reflect the exposure the cases would have had if they had not become cases. Neither cross-sectional nor case control nor longitudinal studies, however, can establish causality, since all of these studies are prone to bias. Well designed and performed interventional studies are the only studies that can prove causality.The already mentioned study on restriction of television viewing and its impact on childhood obesity is a good example of how causality may be established by an interventional study [77]. Such experimental studies may be designed as field trials, community intervention trials or clinical trials.
0 komentar:
Posting Komentar