Partial profile best for 40-50 attribute HCP study?

As the subject line mentions, we have a client who is asking for a conjoint that is 40-50 attributes. Given the specs below, wanted some guidance on if partial profile or some other form of custom approach may work best to tackle this many attributes. A little background:
• Pharma drug and its efficacy, side effects, etc characteristics that may get launched in the next few years
• A big chunk of the attributes are things that might be included on the label - co-morbidities, side effects, contraindications, etc. etc.
• Some of course will also be efficacy metrics - if a partial profile, will likely want to fix a few of these big hitter attributes
• They are looking to understand how all of these different attributes might impact share of prescribing - so likely a chip allocation
• Sample size is probably maxed out at n=200, this is HCPs and so feasibility is limited
• In consideration of partial profile, we think physicians can probably handle 15 attributes since its aspects they are dealing with all of the time.
o Just as a stake in the ground, 4 fixed and 11 random
• I think many attributes will have 2 levels, but certainly a bunch that will have 3

Does it sound like partial profile is the best option?
I realize we need to test reliabilities in lighthouse, but any ballpark estimate for number of exercises needed for (1) individual level utilities, (2) aggregate level utilities?
Am assuming that chip allocation doesn't introduce any additional obstacles here?

Objective is to build a simulator that allows us to determine the prescribing share for any combination of a single drug - and realizing there is a little loss in fidelity, but to simulate more than one product in the market.

Thanks!

Resolved
4 replies