Question

Study A) Using pediatricians’ records, randomly select 200 children with DM and 800 children without DM; interview their mothers concerning infant feeding history. Compare the infant feeding history among children with and without DM.

Study B) Review pediatricians’ records for a random sample of 1,000 youth age 12-19 years old and record DM status and infant feeding history for each youth.

Study C) Send letters and make follow-up telephone calls to the mothers of 1,000 12-year old boys and girls who have tested negative for diabetes within the past six months. Request the parents’ consent to review their child’s medical records annually until the child reaches age 20. Obtain information on infant feeding status from the medical record and a parental questionnaire.

Study D) Send letters and make follow-up telephone calls to 1,000 mothers of newborn babies asking the mothers to participate in a study in which the investigators will randomly assign mothers to breastfeed or formula feed and will review the annual pediatric visit records for the child through age 20 years.

Study E) Examine the correlation between per capita formula sales and DM prevalence across the 20 towns in Maramba.

Ethical issues aside, which one of these studies could provide the strongest evidence that formula feeding increases risk of DM?

Study A)

Study B)

Study C)

Study D)

Study E)

Which one of these studies would be the most problematic for inferring the existence of an association between infant feeding practices and subsequent DM at the individual level?

Study A)

Study B)

Study C)

Study D)

Study E)

Which one of these studies would provide data that can be used to estimate the prevalence of DM at the time of the study?

Study A)

Study B)

Study C)

Study D)

Study E)

Answer 1

1. The method that is the best and will provide the strongest evidence is the one that is done in a controlled manner. Hence, this method should have appropriate controls and randomization of subjects to each group. This is to ensure that there is no inherent bias in the experiment. This is observed only in Study D.

2. Study E would be the most problematic for inferring association at an individual level. This is because it is obtaining a population level correlation and not obtaining data at an individual level. Therefore, no information about individuals can be drawn from this study. All other studies, however, are obtaining data from individual subjects.

3. Study D is assigning randomization and then studying the individuals after a certain period of time. Studies A, B and C are being performed in the current time but they will not provide information on the prevalence of DM. This is because they are looking at the case histories of individuals. Study E can be used to study the prevalence as it examines the correlation between per capita sales and DM prevalence at the time of the study. Hence, the answer to this would be Study E.