Co-twin control designs for testing behavioral economic theories of child nutrition: methodological note

To illustrate the use and potential efficiency of the co-twin control design for testing behavioral economic theories of child nutrition.

Co-twin control design, in which participating twins ate an ad libitum lunch on two laboratory visits. At visit 1, child food choices were not reinforced. On visit 2, twins were randomized to conditions such that one twin was reinforced for each fruit and vegetable serving consumed during lunch ('contingent') while his co-twin was reinforced irrespective of food intake ('non-contingent').

Six male twins, 5 years old, from three monozygotic twin pairs.

Ad libitum intake of total energy (kcals), fat (kcals), and fruits and vegetables (servings) from the protocol test meals on the two visits.

Compared to twins receiving non-contingent reinforcement, twins receiving contingent reinforcement increased fruit and vegetable intake by 2.0 servings, reduced fat intake 106.3 kcals, and reduced total energy intake by 112.7 kcals. The relative efficiency of the co-twin control design compared to a conventional between-groups design of unrelated children was most powerful for detecting 'substitution effects' (i.e., reduced total energy and fat intake) more so than for detecting increased fruit and vegetable intake.

Genetically informative studies, including the co-twin control design, can provide conceptually elegant and efficient strategies for testing environmental theories of child nutrition and obesity.