Read Berger Chandra Garthwaite #1

REGULATORY APPROVAL AND EXPANDED MARKET SIZE

Outline

Questions: Quality certification.

Does FDA approval increase market size due to the market learning about/trusting the drug? (yes)
Is this due to approval or to the increased testing/safety information. (approval)
Is the effect of approval due to insurer's restrictions? (appears to not be so).

Estimation strategy

Issues

Selection bias in that drugs submitted for approval are expected to be profitable, i.e. expectation of an effect.

How those are addressed

Data

Dataset 1 including all indications for new molecular and biological entities approved by FDA.
The dataset they provide covers from 1995 onward reliably.
The authors limited themselves to 1995-2019.
This allowed them to identify follow up indications that were approved.
These indications were then assigned ICD-10CM diagnosis codes to identify if
indications were for previously approved diagnoses or new diagnoses.

Dataset 2 is de-identified claims and medical history data which allowed them to measure the use of drugs.
Allows a look at

Drug
beneficiary
lab results
some demographics

for medicare advantage enrollees.
The prescribing diagnosis is inferred as it doesn't come on the prescription.

dataset 3 (first mentioned page 11{13}) is clinical trials date.

Results

Differencial utilization between same and other diagnosis does occur after approval and is significant, both statistically and economically.
Indication approval is more impactful than trial.

Thoughts

Dataset 1 might be a good way identify competing products.
I'm not sure why they used a linear model for estimating impacts.

# Outline ## Questions: Quality certification. - Does FDA approval increase market size due to the market learning about/trusting the drug? (yes) - Is this due to approval or to the increased testing/safety information. (approval) - Is the effect of approval due to insurer's restrictions? (appears to not be so). ## Estimation strategy ### Issues - Selection bias in that drugs submitted for approval are expected to be profitable, i.e. expectation of an effect. ### How those are addressed ## Data Dataset 1 including all indications for new molecular and biological entities approved by FDA. The dataset they provide covers from 1995 onward reliably. The authors limited themselves to 1995-2019. This allowed them to identify follow up indications that were approved. These indications were then assigned ICD-10CM diagnosis codes to identify if indications were for previously approved diagnoses or new diagnoses. Dataset 2 is de-identified claims and medical history data which allowed them to measure the use of drugs. Allows a look at - Drug - beneficiary - lab results - some demographics for medicare advantage enrollees. The prescribing diagnosis is inferred as it doesn't come on the prescription. dataset 3 (first mentioned page 11{13}) is clinical trials date. ## Results - Differencial utilization between same and other diagnosis does occur after approval and is significant, both statistically and economically. - Indication approval is more impactful than trial. # Thoughts Dataset 1 might be a good way identify competing products. I'm not sure why they used a linear model for estimating impacts.

Key points for wiki

Data appears to be most important.

the use of ICD-10 codes is a good way to group indications. This would be useful to capture the effect of prior entrants.
Clinicaltrials.gov can be correlated with approvals
drugs@fda has info on when approvals occured. I think this might appear in the orangebook.

# Key points for [wiki](wiki/BergerChandraGarthwaite2021) Data appears to be most important. - the use of ICD-10 codes is a good way to group indications. This would be useful to capture the effect of prior entrants. - Clinicaltrials.gov can be correlated with approvals - drugs@fda has info on when approvals occured. I think this might appear in the orangebook.

Labels Milestones