Bias Study

Learning Objectives

>

Introduction to Study

>

Student Role

>

Dr Stellman's Study

>

Dr Shapiro's Study

>

Discussion Questions

Step 5: Questions for Dr. Shapiro's Study

Table A. Distribution of 149 Cases of Endometrial Cancer and 402 Controls, According to Various Characteristics

1.	Based on the table above you notice that cases and controls are similar in terms of several factors including race, religion, marital status, and years of education. However, cases and controls differ in their parity, 72% of cases have less than 3 children whereas only 58% of controls have less than 3 children. It's likely that women with more children see their doctor more often than others because to receive prenatal care. On the other hand, women who see their doctor more often would be more likely to be treated for precancerous lesions and less likely to progress to full-blown cancer of the endometrium. If this were the case in this study, what type of bias could have been introduced?
		a.	interviewer bias
		b.	selection bias
		c.	volunteer bias

2.	In this study nothing is mentioned regarding the way cases and controls were interviewed. We do not know whether cases and controls were interviewed during the same time or cases were interviewed first and then controls were interviewed later. Which is the best way to interview cases and controls?
		a.	simultaneously
		b.	interview cases first and some time later interview controls
		c.	interview controls first and some time later interview cases

3.	The authors of the study estimated effect of estrogen use on endometrial cancer by categorizing exposure according to the time elapsed since latest use and duration of use. Refer to Table 3 in the synopsis. What potential bias would have been introduced if the authors also asked women to recall doses of estrogens?
		a.	measurement error
		b.	recall bias
		c.	no bias

4.	Errors in the recall of exposures is a common problem in case-control studies. In this study estrogen use for five years and more was evaluated according to the time elapsed since latest use. Refer to Table 3 from the synopsis for detailed results. How could limitations in the recall of estrogen use have affected this study?
		a.	Recall would had been the same regardless of time elapsed since latest use and no recall limitations took place.
		b.	Recall would had been better amongst those women for whom the time elapsed since latest use was shorter than for those women for whom the time elapsed since latest use was longer and thus recall limitations occurred.
		c.	Recall would had been worse amongst those women for whom the time elapsed since latest use was shorter than for those women for whom the time elapsed since latest use was longer and thus recall limitations occurred was.

5.	Authors indicated that cases and controls were matched on decade of age and geographic area. What do you think were the effects of controlling or adjusting for these variables on the effect estimates (ORs)?
		a.	It was possible to estimate effects of age and geographic area as predictors of endometrial cancer.
		b.	It was possible to estimate partial effect of age and geographic area on the endometrial cancer occurrence.
		c.	It was not possible to estimate effect of age and geographic area on the endometrial cancer occurrence.

Learning Objectives

>

Introduction to Study

>

Student Role

>

Dr Stellman's Study

>

Dr Shapiro's Study

>

Discussion Questions