diff --git a/Latex/Paper/Main.tex b/Latex/Paper/Main.tex index e2da6b2..63bb57d 100644 --- a/Latex/Paper/Main.tex +++ b/Latex/Paper/Main.tex @@ -26,6 +26,9 @@ \title{The effects of market conditions on enrollment and completion of clinical trials\\ \small{Preliminary Draft}} \author{William King} +\usepackage{multirow} +\usepackage{multicol} + \begin{document} \maketitle @@ -42,10 +45,18 @@ \subfile{sections/01_introduction} %--------------------------------------------------------------- -\section{Literature Review}\label{SEC:LiteratureReview} +%\section{Literature Review}\label{SEC:LiteratureReview} %--------------------------------------------------------------- \subfile{sections/05_LitReview} +The paper proceeds as follows. +Then section \ref{SEC:data} covers the data sources and the proposed +data generating process as well as the causal identification. +Section \ref{SEC:EconometricModel} describes the econometric model +used. +Section \ref{SEC:Results} discusses the results of the analysis. +\todo{Review this after writing a few mor sections.} + %--------------------------------------------------------------- \section{Causal Story and Data}\label{SEC:Data} %--------------------------------------------------------------- diff --git a/Latex/Paper/sections/01.1_introduction.tex b/Latex/Paper/sections/01.1_introduction.tex new file mode 100644 index 0000000..2241ca2 --- /dev/null +++ b/Latex/Paper/sections/01.1_introduction.tex @@ -0,0 +1,98 @@ +\documentclass[../Main.tex]{subfiles} +\graphicspath{{\subfix{Assets/img/}}} + +\begin{document} +% hook - what makes drugs expensive? Mention high failure rate +% describe current research +% - Examine mechanisms by which clinical trials fail. +% - Mention data +% - Results +How to best address the high cost of pharmaceuticals is a crucial health +and fiscal policy question that has been debated for +decades. +Due to the complicated legal and competitive landscape, unintended consequences +are common +\cite{vandergronde_addressingchallengehighpriced_2017}. +One essential step to introduce a novel pharmaceutical - or even +to begin selling a generic compound - is to establish that the drug as packaged and sold will +have acceptable safety and efficacy profiles. +When evaluating these compounds in a clinical trial, multiple outcomes are possible: +\begin{enumerate} + \item The compound demonstrates sufficient safety and efficacy, and proceeds in the appoval process. + \label{Item:EndSuccess} + \item The compound fails to demonstrate sufficient safety and efficacy, and the approval process halts. + \label{Item:EndFail} + \item The trial is terminated before it can acheive one of the first two + outcomes, for reasons unrelated to safety and efficacy concerns. + \label{Item:Terminate} +\end{enumerate} + + +\begin{table} + \caption{Potential States of Knowledge from a clinical trial}\label{tab:StatesOfKnowledge} + \begin{center} + \begin{tabular}{p{0.15\textwidth} p{0.2\textwidth}||p{0.25\textwidth}|p{0.25\textwidth}|} + \cline{3-4} + \multicolumn{2}{c|}{Drug-Indication Match} & safe and efficacious & not safe or not efficatious \\ + \hline + \hline + \multirow{2}{0.15\textwidth}{Operations} & Success & Known good & Known bad \\ + \cline{2-4} + & Failure & \multicolumn{2}{c|}{Unkown} \\ + \cline{2-4} + \end{tabular} + \end{center} +\end{table} + + +\begin{table} + \caption{Clinical Trial end states}\label{tab:ClinicalTrialEndStates} + \begin{center} + \begin{tabular}{p{0.15\textwidth} p{0.2\textwidth}||p{0.25\textwidth}|p{0.25\textwidth}|} + \cline{3-4} + \multicolumn{2}{c|}{Drug-Indication Match} & safe and efficacious & not safe or not efficatious \\ + \hline + \hline + \multirow{2}{0.15\textwidth}{Operations} & Success & Completion & Completion or Termination \\ + \cline{2-4} + & Failure & \multicolumn{2}{c|}{Termination} \\ + \cline{2-4} + \end{tabular} + \end{center} +\end{table} + +While it is known that pharmaceutical companies withdraw some drugs from +their development pipeline due to commercialization concerns +( +\cite{khmelnitskaya_competition_2021} +and +\cite{van_der_gronde_addressing_2017} +), there are likely unseen +effects that might affect the overall drug pipleline. +One of these is the concern that when there are already approved therapies on +the market, patients might be loath to enroll in clinical trials, +causing the trial to fail for reasons unrelated to the scientific or +commercial viability of the therapy. + + +To adequately guide public policy it is crucial that robust, causally-identified +statistical models are available to describe the interaction between +various players within the space. + +This work endeavors to estimate the change in probability of successful completion +of a clinical trial due to the existence of alternative drugs on the market. +In particular, it seeks to establish whether such an impact is mediated +by enrollment patterns or is caused more directly. + + +The paper proceeds as follows: a brief literature review in \cref{SEC:LiteratureReview}, +a description of the caual model in \cref{SEC:CausalIdentification}, +followed by a description of the data (\cref{SEC:Data}) and the +econometric model (\cref{SEC:EconometricModel}). +Preliminary results are presented in \cref{SEC:Results} and a discussion +of proposed improvements is included in \cref{SEC:Improvements}. + + + + +\end{document} diff --git a/Latex/Paper/sections/01_introduction.tex b/Latex/Paper/sections/01_introduction.tex index 40f2b2a..4b8fda1 100644 --- a/Latex/Paper/sections/01_introduction.tex +++ b/Latex/Paper/sections/01_introduction.tex @@ -2,52 +2,61 @@ \graphicspath{{\subfix{Assets/img/}}} \begin{document} -% hook - what makes drugs expensive? Mention high failure rate -% describe current research -% - Examine mechanisms by which clinical trials fail. -% - Mention data -% - Results -How to best address the high cost of pharmaceuticals is a crucial health -and fiscal policy question that has been debated for -decades. -Due to the complicated legal and competitive landscape, unintended consequences -are common -\cite{van_der_gronde_addressing_2017}. -One critical aspect to successfully introduce a novel pharmaceutical or even -a generic compound is to establish that the drug as packaged and sold will -have acceptable safety and efficacy profiles. -This is done using clinical trials. - -To adequately guide public policy it is crucial that robust, causally-identified -statistical models are available to describe the interaction between -various players within the space. -While it is known that pharmaceutical companies withdraw some drugs from -their development pipeline due to commercialization concerns -( -\cite{khmelnitskaya_competition_2021} -and -\cite{van_der_gronde_addressing_2017} -), there are likely unseen -effects that might affect the overall drug pipleline. -One of these is the concern that when there are already approved therapies on -the market, patients might be loath to enroll in clinical trials, -causing the trial to fail for reasons unrelated to the scientific or -commercial viability of the therapy. - -This work endeavors to estimate the change in probability of successful completion -of a clinical trial due to the existence of alternative drugs on the market. -In particular, it seeks to establish whether such an impact is mediated -by enrollment patterns or is caused more directly. - - -The paper proceeds as follows: a brief literature review in \cref{SEC:LiteratureReview}, -a description of the caual model in \cref{SEC:CausalIdentification}, -followed by a description of the data (\cref{SEC:Data}) and the -econometric model (\cref{SEC:EconometricModel}). -Preliminary results are presented in \cref{SEC:Results} and a discussion -of proposed improvements is included in \cref{SEC:Improvements}. - - +Developing new, effective pharmaceutical compounds is a fundamentally +difficult task. +Starting with challenges identifying promising treatment targets and potential +compounds to ensuring the drug can be properly delivered within the body, the +scientific work that needs to succeede is massive. +The regulatory and market conditions in which they exist add to this difficulty. +For example, regulations are designed to reduce the number of drugs released +to market with significan issues, such as in the case of VIOXX +\cite{krumholz_whathavewe_2007} +or the Perdue Pharma scandal +\cite{officepublicaffairsjusticedepartment_2020}. +These regulations, such as clinical trial standards +\todo{add citation to clinical trials here}, +increase the costs of developing new drugs, adding to the business concerns +already present, including competitors already in the market or close to +entering and the overall demand to address a given condition. + +%begin discussing failures +%I am thinking I'll discuss marketing and operational failures +%I somehow need to step away from the drug development framing and soften it to ... what? drug investigation? +From these general challenges we can begin to classify failures in drug +development into a hierarchy of causes. +\cite{khmelnitskaya_competitionattritiondrug_2021} +described two general causes for a drug to exit the drug-development pipline, +strategic exits and scientific failure. +\cite{hwang_failure_2016} +described failues of Phase III trials in a similar way, +ascribing drug development failures to issues with safety, +efficacy, or other (buisness) concerns. + +% The only one most ameniable to being targeted by policy +% is those ``other concerns''. +Although decisions to continue drug development are driven +by long term profit analyses, +pharmaceutical companies face short term operational challenges. +% As an example, while a drug may have few competitors and +% strong evidence of safety, difficulties recruiting trial participants may +% prevent the clinical trials process from being completed successfully. +For example, even with few competitors and strong safety evidence, recruitment difficulties can still derail a drug's clinical trial process. +\todo{Clean up that hypothetical, it doesn't seem clean} +Thus being able to isolate the effect of operational challenges from +strategic decisions allows us to predict the intended or unintended effects +of a given policy on clinical trials. + +In this work, I focus on separating the effects of enrollment and +competing drugs on clinical trial completion, specifically Phase III trials. +To do this, I create a + dataset extracted from +\url{ClinicalTrials.gov} +that tracks individual clinical trials as they progress towards completion +as well as a novel causal model of individual clinical trial progression. +Unlike previous research which is focused on the drug development pipeline, I +restrict my investigation to modelling individual clinical trials. +The goal of this restriction is to provide a way to predict the impact +of changes that affect enrollment independent of other confounding effects. \end{document} diff --git a/Latex/Paper/sections/05_LitReview.tex b/Latex/Paper/sections/05_LitReview.tex index 5a624ec..788396c 100644 --- a/Latex/Paper/sections/05_LitReview.tex +++ b/Latex/Paper/sections/05_LitReview.tex @@ -3,71 +3,81 @@ \begin{document} -This paper sits within an intersection of health and industrial organization economics -that is frequently studied. -Encouraging a strong supply of novel and generic pharmaceuticals contributes -in important ways to both public health and fiscal policy. -Not only to the pathway to drug approval long, as many as 90\% of compounds -that begin human trials fail to gain approval -(\cite{khmelnitskaya_competition_2021}). -Complicating this is the complex regulatory and competitive environment in -which pharmaceutical companies operate. - -%%%%%%%%% Why are drugs so expensive? - -% van der Grond, Uyle-de Groot, Pieters 2017 -% - What causes high costs of drugs? -% - High level synthesis of discussion regarding causes -% - Academic and non-academic sources - - %%%%%%%%%%%%%%%% What do we know about clinical trials? +\subsection{What do we know about clinical trials and their success rates?} + +Most studies of clinical trials attempt to model only those trials +which are involved in the drug approval process. % Hwang, Carpenter, Lauffenburger, et al (2016) % - Why do investigational new drugs fail during late stage trials? -\citeauthor{hwang_failure_2016} (\citeyear{hwang_failure_2016}) +\cite{hwang_failure_2016} investigated causes for which late stage (Phase III) clinical trials fail across the USA, Europe, Japan, Canada, and Australia. They found that for late stage trials that did not go on to recieve approval, 57\% failed on efficacy grounds, 17\% failed on safety grounds, and 22\% failed on commercial or other grounds. -For context, this current work hopes to be able to distinguish some of the -mechanisms behind those commercial or other failures. % Abrantes-Metz, Adams, Metz (2004) % - What correlates with successfully passing clinical trials and FDA review? % - -In \citeyear{abrantes-metz_pharmaceutical_2004}, -\citeauthor{abrantes-metz_pharmaceutical_2004} +\cite{abrantes-metz_pharmaceutical_2004} described the relationship between various drug characteristics and how the drug progressed through clinical trials. -This non-causal estimate was notable for using a -mixed state proportional hazard model and estimating the impact of +This descriptive estimate used a +mixed state proportional hazard model and estimated the impact of observed characteristics in each of the three phases. They found that as trials last longer, the rate of failure increases for -Phase I \& II trials, while Phase 3 trials generally have a higher rate of +Phase I and II trials, while Phase 3 trials generally have a higher rate of success than failure after 91 months. + +\cite{hay_ClinicalDevelopment_2014} tracks clinical trials based on +the number of indications studied. +They find that 10.4\% of all novel drug development paths for an indication, +studied in a phase I trial, are ultimately approved by the FDA. + +\cite{wong_EstimationClinical_2019} +constructed a model where they estimated each, which they used to estimate the +probability of completing a given phase, conditional on starting a previous phase. +In doing so, they found that 13.8\% of all drug development programs +completed successfully, which is higher than the approximately 10\% rate +others have found\cite{hay_ClinicalDevelopment_2014}. +One cause of this may be that they considered that a single drug might +be used tested for multiple indications. +% Large dataset. +% they found lower estimates than previous work. + % Ekaterina Khmelnitskaya (2021) % - separates scientific from market failure of the clinical drug pipeline -In her doctoral dissertation, Ekaterina Khmelnitskaya studied the transition of +%In her doctoral dissertation, Ekaterina Khmelnitskaya +\cite{khmelnitskaya_CompetitionAttrition_2021} approaches a slightly +different problem. +She created a multistage model to track the transition of drug candidates between clinical trial phases. -Her key contribution was to find ways to disentangle strategic exits from the -development pipeline and exits due to clinical failures. -She found that overall 8.4\% of all pipeline exits are due to strategic +Her key contribution was to find ways to disentangle strategic exits where +firms remove novel from the development pipeline and +exits due to scientific failures +(where safety and efficacy did not prove sufficient). +She estimates that overall 8.4\% of all pipeline exits are due to strategic terminations and that the rate of new drug production would be about 23\% -higher if those strategic terminatations were elimintated -(\cite{khmelnitskaya_competition_2021}). - -% Waring, Arrosmith, Leach, et al (2015) -% - Atrition of drug candidates from four major pharma companies -% - Looked at how phisicochemical properties affected clinical failure due to safety issues -%not in this version - - - +higher if those strategic terminatations were elimintated. %%%%%%%%% What do we know about drug development incentives? +\subsection{What do we know about drug development incentives?} +% Introduce section +% key points +% - multiple types of drugs (generic and brand named) +% - These respond differently +% - Dranov et al 2022 - demand pull seems to bias follow up drug development. +% - increasing demand doesn't necessarily result in new compounds (check this). Risks. +% - acemoglu and linn 2004 - population size matters. +% - Note then that separating effects is difficult at the drug development level. +% - Population ties into the number of drugs available, and operational (recruitment) concerns +% - In general, there are going to be many confounding variables. +% - +% +% % Dranov, Garthwaite, and Hermosilla (2022) % - does the demand-pull theory of R&D explain novel compound development? @@ -88,10 +98,22 @@ Among non-generics, a 1\% increase in potential market size % Gupta % - Inperfect intellectual property rights in the pharmaceutical industry -%\cite{GupaPhd2023} +\cite{gupta_OneProduct_2020} +\todo{Sumarize how intellectual property rights affect things} +% - link to difference between novel and generics from acemoglu and linn % Agarwal and Gaule 2022 % - Retrospective on impact from COVID-19 pandemic % Not in this version +\subsection{What do we know about how Clinical Trials proceed?} +%interview with Adam George +% - clinical trials are often handled by contractors +% - they plan sites, start times, etc from beginning. +% - Running late is normal. + +% Results on enrollment projection +% - nothing really good exists. +% - no cross validation, only tested on a few trials. + \end{document}