Consider the DAG below. Identify all the non-causal open backdoor paths between exposure (E) and disease (D).

Consider the DAG above. Is it possible to obtain an unbiased estimate of the causal effect of E on D? If so, how? If not, why not?

Consider the DAG below. Identify all the non-causal open backdoor paths between Hunger (E) and plasma HIV RNA viral load (PVL) nonsuppression (outcome). This is from a study of HIV positive individuals receiving antiretroviral drugs (ART).

For the next questions, use the DAG (above).

i. You collect data on all the variables. In your data, do you expect CD4 count to be associated with Hunger? (Circle the best answer).

a. Yes

b. No

c. Not enough information

ii. In your data, do you expect CD4 count to be associated with outcome (PVL nonsuppression)? (Circle the best answer).

a. Yes

b. No

c. Not enough information

iii. Should you adjust for CD4 count to get the non-biased measure of association between Hunger and the outcome, PVL nonsuppression? (Circle the best answer).

a. Yes

b. No

c. Not enough information

iv. Assuming you have closed all backdoor paths, the RR relating Hunger and PVL nonsuppression is RR_ED = 3.0 while the RR conditioned on ART nonadherence (C) is RR_ED|C = 2.0. Based on these results, do you think there is an effect of Hunger on PVL nonsuppression which is not through ART nonadherence? Explain your answer.
 
 

4 E K D