Hi Dr. Newman,
Are you able to clarify question 8, part d, subpart i: the question states:
d. To supplement their propensity analysis, the authors also did an instrumental variable analysis, using neonatal unit preference for early screening (in quartiles) as the instrument for actual screening. An alternative approach would be to use screening itself as an instrument for PDA treatment in the propensity-matched groups.
i. If we used this latter approach, what would we need to assume about the relation between PDA treatment, screening, and in-hospital mortality? [2]
Are we to assume for this part that the exposure is PDA treatment and the outcome is mortality? In the abstract both PDA treatment and mortality were described as outcomes.
Thank you in advance for your help!
Kim