Question

An investigator conducted a case-control study in 1000 participants (500 cases and 500 controls) to examine the association between childhood head trauma (exposure) and Parkinson’s disease(outcome). Information on head trauma was self-reported by study participants. Cases were able to correctly report exposure to head trauma 96% of the time, while controls correctly reported exposure to head trauma 80% of the time. Cases and controls correctly reported no exposure to head trauma 96% and 98% of the time, respectively.

The truth(which is unbeknownst to the investigator) is given below:

	Cases	Controls
Head Trauma	75	50
No Head Trauma	425	450
	500	500

Specifically, what type of information bias is encountered in this study of childhood head trauma and Parkinson's disease? In this study, is the misclassification that results from this bias differential or non-differential? Explain.

Answer 1

The confidence interval of odds ratio information bias is encountered in this study of childhood head trauma and Parkinson's disease

Given Head No head425 Odds Ration (75 / 50) / (425 / 450) = 1.58824 Calculate Natural log of odds Ration-Ln (1.58824)= 0.46262 This is the point estimate for the confidence interval SE În(Odds ratio) ,|_+_+_+_ = 0.19470 ME= 1.96*0.19470=0.3816 95% confidence interval of the odds ratio is exp (0.46262+0.3816) (1.0844,2.32616) Cases 75 Controls 50 450 23 2010 75 50 425 450

In this study, the mis-classification that results from this bias is differential. Because, Differential (non-random) mis-classification occurs when the proportions of subjects mis-classified differ between the study groups. That is, the probability of exposure being mis-classified is dependent on disease status, or the probability of disease status being mis-classified is dependent on exposure status.