You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
ClinicalTrialsPaper/Latex/Paper/sections/03_CausalIdentification.tex

83 lines
4.2 KiB
TeX

\documentclass[../Main.tex]{subfiles}
\graphicspath{{\subfix{Assets/img/}}}
\begin{document}
Because running experiments on companies running clinical trials is not going
to happen anytime soon, causal identification will depend on creating a
structural causal model.
In \cref{Fig:CausalModel} I diagram the directed acyclic graph that describes
the data generating model.
The proposed data generating model consists of a decision maker, the study
sponsor, who must decide whether to let a trial run to completion or terminate
the trial early.
While receiving updates regarding the status of the trial, they ask questions
such as:
\begin{itemize}
\item Do I need to terminate the trial due to safety incidents?
\item Does it appear that the drug is effective enough to achieve our
goals, justifying continuing the trial?
\item Are we recruiting enough participants to achive the statistical
results we need?
\item Does the current market conditions and expectations about returns on
investment justify the expenditures we are making?
\end{itemize}
When appropriate, the study sponsor terminates the trial.
If there are not enough issues to terminate the trial, it continues until it
is completed.
While conducting a trial, the safety and efficacy of a drug are driven by
fundamental pharmacokinetic properties of the compounds.
These are only imperfectly measured both prior to and during any given trial.
Previously measured safety and efficacy inform the decision to start the trial
in the first place while currently observed safety and efficiency results
help the sponsor judge whether or not to continue the trial.
Of course, these decisions are both affected by the specific condition being
treated due to differences in the severity of the symptoms.
When a trial has been started, it comes time to recruit participancts.
Participants frequently depend on the advice of their physician when deciding
to join a trial or not.
As these physicians have a duty to seek their patients best interest; they, along
with their patients will evaluate if the previously observed safety and efficacy
results justify joining the trial over using current standard treatments.
Thus the current market conditions may affect the rate at which participants
enroll in the trial.
The enrollment of participants in a trial depends on a few other factors.
The condition or disease of interest and how it progresses will determine how long
recruitiment will be held open versus just an observation of treatment arms.
Aditionally, a trial that has already reached a high enough enrollment will often
close recruitment by switching to an "Active, not recruiting" stage to manage costs.
Finally, enrolling participants depends on how difficult it is to find people
who suffer from the condition of interest.
The preceeding issue of population size also affects the number of alternatives available.
When there are less people affected by the disease, the smaller market reduces
possible profitability, all else equal.
Thus the likelihood of companies paying the sunk costs to develop drugs for
these conditions may be lower.
Finally, the number of alternatives on the market may affect the return on
investment directly, causing a trial to terminate early if the return is
not high enough.
\begin{figure}[H] %use [H] to fix the figure here.
\includegraphics[width=\textwidth]{../assets/img/dagitty-model.jpg}
\caption{Causal Model}
\label{Fig:CausalModel}
\end{figure}
%
By using Judea Pearl's do-calculus, I can show that by choosing an adjustment
set of the decision to condut a phase III trial, the condition of interest,
the current status of the trial, and the population size will casually
identify the direct effects of enrollment and market alternatives on the
probability of termination.
This is easily verified through the backdoor criterion, which states that
if every path between the exposure and outcome that starts with an arrow
flowing into the exposure is blocked by one of the values in the adjustment
set, then the effect of the exposure on outcome is causally identified
(\cite{pearl_causality_2000}).
It can be easily visually verified by the DAG on the graph that this is the case.
\end{document}