Translating evidence into practice: sampling and outcomes

Obesity Discussion

A study on obesity can focus on a specific demographic, community, or subpopulation, but if the study is to be representative of a large and diverse population the sampling constraints and requirements become more stringent. Sampling should not be left entirely up to chance, but should be controlled so as to be representative of the demographics represented in the larger population. Many demographic markers such as age, gender, and ethnicity have been linked to variances in obesity rates and outcomes, meaning that an ample sample of all represented demographics would be essential in a broad population-based study (Paeratakul et al., 2002). In such a study, the only exclusionary criteria would be medical inappropriateness or ethical concerns, with limits placed on enrollment of certain demographics to ensure an adequate analysis of all demographics without overburdening researchers. Recruitment could occur in a variety of ways, including through general practice/healthcare organization outreach, advertisement, and other means of communication and outreach.

Getting an accurate sample does not simply rely on the exclusion and inclusion requirements of the study, however, as this only deals with participants once they have been approached or have registered for initial participation. Recruitment on a widespread and truly pervasive basis would be essential to achieving an accurate population sample, as only with proper spread and depth of participation would a representative sample be available (Paeratakul et al., 2002). Sampling in theory must be supported by effective practical techniques of achieving necessary levels of participation, then, otherwise the theory is for naught.

In response to the posting regarding sampling at a clinic, the general plan seems good. The recruitment description, or rather the lack thereof, is the only area of concern. Wouldn't some level of recruitment process, such as at least ensuring that patients are aware their data will be aggregated and analyzed, be slightly better? This would not only be more ethical, but would also provide an opportunity to get additional information from participants. A depression-screening tool would of course be an effective tool for use in the obesity clinic and would be worthwhile to employ even without any external research project being undertaken, but to have no recruitment procedures at all seems to invite the potential for a host of problems in sampling and in other aspects of research. Otherwise, however, a study-based directly on the population you are gathering for is clearly going to be an effective sample.