Epidemiology 200B: Winter 2010 Homework 4: “Practice Quiz”

Name: TA: Joni / Doug

The following questions refer to the article: A matched, case–control study of the association between

Schistosoma japonicum and liver and colon cancers, in rural China; Annals of Tropical Medicine & Parasitology, Vol. 99, No. 1, 47–52 (2005).

You have 30 minutes to complete.

Why weren’t additive interactions between Hepatitis B and schistosomiasis explored in this analysis?

Briefly explain 1 pro and 1 con of excluding those with Hepatitis B from an analysis of the association between schistosomiasis and liver cancer in rural China.

On page 51, in the last sentence before the DISCUSSION section the authors write, “Thus, the present data indicate that hospitalized subjects with a history of schistosomal infection are more than three times more likely to have colon or liver cancer than those with no such history.” Rewrite this statement to express more accurately what this data indicates. (HINT: Consider the study design.)

Is the ‘rare disease assumption’ necessary in this context?

Draw a DAG which demonstrates the Berkson’s bias described in the discussion section.

What were the matching factors in this study?

Comment on the phrase “…but otherwise randomly selected,” located in the top left of page 49 in the paragraph describing the matching factors.

At the top right of page 49 the authors discuss their method for addressing a type of information bias in the exposure assessment (schistosomiasis diagnosis) among colon cancer cases and controls. What kind of misclassification do they think would result from this information bias? Why? In what direction would you expect this misclassification to bias the effect estimate?

On page 50 the authors state, “When data analyses for liver cancer were limited to those case-control pairs that had no record of a positive hepatitis test, only 48 matched pairs remained available.” How many matched groups are used to generate the Odds Ratio? (HINT: Look at Table 2). Explain the discrepancy.

Using the Liver Cancer data from Table 2:

(1) Calculate the Unmatched - Crude Odds Ratio for the association between Schistosomiasis infection and Liver Cancer

(2) Calculate a Matched-Crude OR for the association between Schistosomiasis infection and Liver Cancer using the 48 matched pairs alluded to in the article. (Assume the distribution of exposure among case and controls is the same as shown in Table 2)

(3) Use the number of cases and controls “Investigated” to generate a crude odds ratio. But assume that exposure prevalence for liver cancer cases was assessed from the hospital only, and exposure prevalence for controls was assessed from the control-station only.

Schistosomiasis is transmitted via freshwater that has been contaminated with human urine and/or feces. Based on this information, present an alternative hypothesis on the association between Schistosomiasis and colon/liver cancer. Include a DAG.

1