A team of neurologists and psychologists researching intellectual function is reviewing the results of a study on the association between childhood lead exposure and adult intellectual function.
Based on the abstract, which of the following is the most concerning factor regarding the design of the study?
A)Contamination bias
B)External validity
C)Lack of statistical power
D)Measurement error
E)Selection bias

Question

A team of neurologists and psychologists researching intellectual function is reviewing the results of a study on the association between childhood lead exposure and adult intellectual function. 
Based on the abstract, which of the following is the most concerning factor regarding the design of the study?
A)Contamination bias
B)External validity
C)Lack of statistical power
D)Measurement error
E)Selection bias

Quizplus · Accepted Answer

Explanation: Statistical power, the ability to detect an association if that association exists, is dependent on sample size, with a larger sample size increasing power. A high nonresponse rate reducesthe sample size; it may also influence the distribution of potential confounders so that they become imbalanced between groups. As a result, an inadequate sample size may reducethe statistical power to control for all confounders. This is evidenced by a reduction in the precision of estimates of association (ie, wider confidence intervals or larger p-values) and may result in apparent associations that do not reach statistical significance. In this study, the sample size was considerably reduced as a result of the low response rate, with only 43 (29%) of the 148 participants (or 17.2% of the original cohort of 249) completing the study. As a result, after adjusting for maternal intelligence function (IQ), the initially established apparent association between childhood blood-lead concentrations and adult IQ was still evident (−1.89 before adjusting for maternal IQ versus −1.81 after adjusting for maternal IQ), but the association was no longer statistically significant (95% confidence interval [95%CI] [−2.40, −0.47] before adjusting for maternal IQ versus 95%CI [−2.59, 0.06] after adjusting for maternal IQ). The statistical significance of the results was lost due to lack of sufficient power to adjust for maternal IQ. (Choice A) Contamination occurs when participants in one trial group receive the treatment or intervention meant for participants from another group. This type of bias affects experimental designs, not observational studies. (Choice B) External validity (or generalizability) is defined as the applicability of a study's results beyond the group that was initially assessed. Given that participants were similar to members of the original cohort who did not participate or undergo neuropsychological assessment in the study, external validity is less likely to be a concern in this study. (Choice D) Measurement bias may occur from poor exposure/outcome measurements or data collection processes. In this study, lead levels (exposure) were measured through standardized blood sample testing at different points in time, and IQ (outcome) was measured by a child neurologist trained to use the Wechsler Abbreviated Scale of Intelligence and blinded to subjects' developmental history and lead concentrations. These control measures significantly reduce the likelihood of measurement bias in the study. (Choice E) Selection bias can occur with inappropriate selection methods or through selective attrition or nonresponse of participants. Due to the low response rate in the study, selection bias is a possible limitation of the study. However, the fact that the participants were similar to the members of the original cohort considerably reduces the effect of this potential bias. Educational objective: An inadequate sample size may reduce the statistical power to control for all confounders in a study, which may result in apparent associations that do not reach statistical significance.

A Team of Neurologists and Psychologists Researching Intellectual Function Is