A randomized controlled study of chronically sleep-deprived (5 = abnormal. There were statistically significant differences in all 3 parameters between visit 1 and visit 2. Which of the following is the best explanation for the difference between findings at the screening and randomization visits?

Question

A randomized controlled study of chronically sleep-deprived (<6.5 hr per night) , morbidly obese subjects examined whether sleep extension is associated with weight loss and metabolic and endocrinal improvements. The intervention consisted of education regarding sleep hygiene and a gradual change to subjects' bedtimes to earlier in the evening. Participants underwent general screening at visit 1 with collection of baseline parameters; then, 2 months later, at visit 2, subjects were randomized into either an intervention group or a comparison group. Researchers evaluated several parameters, including sleep duration, sleepiness, and sleep quality. The following results were reported: Note: Data is reported as mean ± standard deviation.
^*p < 0.05.
¹ ESS (Epworth Sleepiness Scale [daytime sleepiness]) >10 = sleepy.
² PSQI (Pittsburgh Sleep Quality Index [subjective sleep quality]) >5 = abnormal.
There were statistically significant differences in all 3 parameters between visit 1 and visit 2. Which of the following is the best explanation for the difference between findings at the screening and randomization visits?

Quizplus · Accepted Answer

Explanation: The Hawthorne effect is related to subjects' participation in research, the subsequent awareness of being observed, and the possible impact on behavior. This type of bias is commonly seen in studies that concern behavioral outcomes. In this case, subjects reported substantial improvements in sleep in the 2-month period between the screening and randomization phases, when they were candidates for participation but not yet subject to the actual intervention being implemented. It is likely that their awareness of their potential for participation (prior to any intervention) had a significant effect on the outcome under study, a characteristic of the Hawthorne effect. Alterations in behavior due to participation (or potential for participation) awareness may significantly impact how subjects report on the outcomes under study. In most instances, an overestimation of the effect of the intervention will be reported, which is a threat to the internal validity of the study. To minimize the Hawthorne effect, subjects can be kept unaware they are being studied; however, at times this is not reasonably possible. (Choice A) Allocation bias can result from the way subjects are assigned to the treatment and control groups. It may occur when subjects are nonrandomly assigned to the treatment groups of a clinical trial (eg, physicians may preferentially enroll sicker patients into the treatment group). Allocation bias is different from selection bias, which occurs when the studied sample does not represent the target population. (Choice B) A confounder is an extraneous variable that is independently associated with both the dependent variable (ie, outcome) and the independent variable (ie, exposure). Confounding bias occurs when the study design or analysis does not control for the potential effect these confounders may have on the association under study. However, in this case, researchers observed substantial improvements in sleep prior to randomization and to any intervention. (Choice D) The placebo effect is a perceived outcome (eg, beneficial or harmful) resulting from a participant's personal expectations rather than from the treatment itself. In this case, the design did not include a placebo treatment group. Moreover, the observed improvements in sleep were not related to having any particular intervention. (Choice E) Selection bias can occur with inappropriate (eg, random or nonrandom) selection methods or through differential attrition rates between study groups. This would not be a likely source of bias in this case, as subjects were not yet participating in the study but rather were candidates for participation. Educational objective: The Hawthorne effect, which describes the impact on behavior from awareness of being observed, may result in an overestimation of an outcome not only among participants, but also among candidates for participation. Such overestimation of the effect of the intervention poses a threat to the internal validity of any study.

A Randomized Controlled Study of Chronically Sleep-Deprived (<6