Last week we went through nine vaccine claims and let the evidence answer each one. Not everyone liked what the evidence said. That is how evidence works.

This week we step back from the specific claims and look at the system that generates the evidence in the first place. Because if you are going to evaluate any health claim about a treatment, a vaccine, a drug, or a device, you need to understand how treatments and treatment devices actually get tested and approved.

That system has a name, or names actually: the drug development pipeline, or the clinical research pipeline if that’s what you would prefer. It is one of the most complex, expensive, and counterintuitive systems humans have ever built. It is also the reason the pill you pick up at the pharmacy exists at all. Or the brain implant that helps modulate Parkinson’s tremors. Or the inhaler used for allergy medications, plus the allergy medication in that inhaler. And it is the reason the treatment for your condition might not exist yet.

Welcome back to Root to Rx. If you are reading this, you are getting “Rooted”, a little deeper with every issue. This week we are going all the way through the journey: from the first molecule that catches a scientist's attention to the moment an approved drug reaches the pharmacy shelf. It is a 10-to-15-year story, and most of it happens completely out of sight until you see a drug commercial talking about you could experience x, y, z. We are going to change that.

The average approved drug takes 10 to 15 years to reach a pharmacy shelf.

The average cost, across all the compounds that succeed and the many more that fail along the way: approximately $2.6 billion per approved drug. (DiMasi et al., Journal of Health Economics, 2016.)

Of every 10,000 compounds that enter early discovery research, roughly 250 will make it to preclinical testing. Of those 250, about 5 will enter human clinical trials. Of those 5, only 1 will receive FDA approval.

That is not a failure rate. That is the system working. It is designed to stop treatments that do not work, or are not safe enough, before they reach the people who need them. The 9 out of 10 that fail along the way are not wasted. They are the reason the 1 that succeeds is trustworthy.

For a visual overview of the full FDA drug approval pipeline, see the official FDA Drug Approval Process Infographic (Horizontal).

Every drug begins with a question: what, exactly, is causing this disease at the molecular level?

Before anyone develops a treatment, someone must understand the mechanism. Cancer cells grow uncontrollably because a specific protein is misfiring. A pathogen invades because it can attach to a specific receptor. A neurological disease progresses because a particular biological pathway has broken down. The molecule, protein, or pathway at the center of the disease is called the target.

Finding it requires years of basic research, cell biology, genetics, and biochemistry. Most of it is publicly funded. Universities, the NIH, and research institutes around the world do this work long before any pharmaceutical company is involved. The publicly funded research that identified the molecular basis of hundreds of diseases, including the BCR-ABL protein at the center of a particular leukemia, is what makes targeted drug development possible at all.

Once a target is identified, it has to be validated, meaning researchers have to confirm that interfering with this target actually changes the disease, not just in a petri dish, but in living biological systems. Target validation can take as long as target identification. This phase alone averages three to six years.

Now that researchers know what they are trying to affect, they need something that can affect it. This is where the scale of pharmaceutical research becomes difficult to comprehend.

High-throughput screening systems, essentially robotic laboratories, can test tens of thousands of chemical compounds against a target in a matter of weeks. They are looking for hits: compounds that show some measurable interaction with the target. Out of 10,000 compounds screened, a handful, maybe 10 to 100, will show enough activity to investigate further.

From those hits, medicinal chemists select the most promising candidates and begin reshaping them. They modify atoms, add functional groups, test variations, and iterate. The goal is to increase potency, meaning it works at a lower dose. Improve selectivity, meaning it affects the target without disrupting other biological processes. And reduce toxicity, meaning it does what it is supposed to do without harming the cells it is not supposed to touch.

The compound that emerges from this process is called the lead compound. It is the molecule that will go into a living system for the first time. It has never touched a human being. That is about to change.

Before any compound enters a human body, it has to prove itself in cell cultures and animal models. This is the preclinical phase.

Preclinical studies answer a specific set of questions. Is it toxic at doses that would have a therapeutic effect? How does the body absorb, distribute, metabolize, and excrete it? Does it do what it is supposed to do in a living biological system? What happens at higher doses? What is the no-observed-adverse-effect level?

These are not bureaucratic requirements. They are the questions that protect the first human being who will ever receive this compound from harm. The Thalidomide disaster of the 1950s and 1960s, in which a drug prescribed to pregnant women caused severe birth defects in thousands of children, is the historical reason preclinical reproductive toxicology studies are now mandatory before human trials. Every requirement in the preclinical process exists because something went wrong before it existed.

If the preclinical data is satisfactory, the sponsor, the company or institution funding the research, compiles everything they know into a formal document called the Investigational New Drug application, or IND. It contains the preclinical safety data, the proposed clinical study design, the manufacturing information, and the plan for protecting participants.

This document goes to the FDA. The FDA has 30 days to review it. If they do not object, the trial can proceed to humans.

The clinical trial phases are not arbitrary bureaucratic steps. Each one is designed to answer a specific question, in a specific population, with a specific amount of certainty, before asking the next question in a larger group of people.

The first time a new compound enters a human body is called a First in Human trial. Phase I trials typically enroll 20 to 100 participants, often healthy adult volunteers, and the primary question is not whether the drug works. It is whether the drug is safe at any dose.

Researchers start with a very small dose, far below what they expect to be therapeutic, and escalate slowly and carefully, watching for any sign of harm at each level. They are measuring how the body processes the compound, what side effects appear at which doses, and what the maximum tolerated dose appears to be. This is called dose escalation, and it can take one to two years.

For certain serious diseases, including some cancers, Phase I participants are often patients who have no other treatment options. They are not volunteers in the casual sense. They are people who have run out of approved paths and are choosing to be the first. That choice is a specific kind of courage, and the entire regulatory framework governing clinical trials exists in part to honor it.

If Phase I establishes that the drug is safe at a therapeutic dose, Phase II tests whether it actually has an effect in patients with the target disease. Studies typically enroll 100 to 500 participants across multiple clinical sites.

Phase II is where the data starts telling a story. Researchers are looking for response rates, biomarker changes, dose-response relationships. They are refining the patient population, identifying which subgroups respond best, and sharpening the outcome measures that will be used in the definitive trial.

This is also where many compounds fail. A treatment that looks promising in animal models does not always translate to humans. A drug that reduces tumor size in a mouse may have no effect on the same tumor in a person. Phase II failure is not a mistake. It is the system catching a dead end before investing hundreds of millions of dollars in a Phase III trial for a drug that does not work.

Phase III is where a drug earns its approval, or fails to. These are large, randomized, controlled trials, often involving 1,000 to 5,000 or more participants across dozens of clinical sites in multiple countries, comparing the investigational drug against the current standard of care or a placebo.

The goal is not to show that the drug helps some patients sometimes. It is to generate statistically robust evidence, at the population level, that the drug's benefits outweigh its risks for a defined patient population. Pre-specified primary and secondary endpoints. Blinded assessments. Independent data monitoring committees watching for safety signals throughout.

Running a global Phase III trial is one of the most complex human endeavors that does not make headlines. I spent over a decade managing them. A single trial can involve 40 clinical sites across 15 countries, thousands of patients, and data collected at hundreds of thousands of individual timepoints. Every data point has to be verified against source documents by a trained clinical monitor before it can enter the analysis. The reason that verification exists, and the standards that govern it, is the subject of the next section.

Most people assume clinical trials are governed by trust and reputation. They are not. They are governed by a document.

Good Clinical Practice, or GCP, is the international standard that defines how every clinical trial must be conducted: how participants must be protected, how data must be collected and verified, how investigators must be trained, and what happens when something goes wrong. It is not optional. It is not aspirational. Sponsors, investigators, and clinical sites that do not comply with GCP can have their trial data rejected by the FDA and other regulatory agencies, meaning years of work and billions of dollars can be invalidated by a single documentation failure.

The current version is ICH-GCP E6(R3). It was finalized by the International Council for Harmonisation in November 2023 and took effect in March 2025. It is the most significant update to GCP standards since 2016, and it reflects how dramatically clinical research has changed since then.

Three changes are worth understanding, even if you are not in the industry.

First: risk-based quality management. Previous GCP required exhaustive monitoring of every data point in every trial, regardless of how critical that data point was to patient safety or trial integrity. E6(R3) formalizes a more targeted approach. Sponsors identify which data is critical, meaning it directly affects patient safety or the validity of the trial result, and concentrate monitoring resources there. The principle is not that other data does not matter. It is that not everything poses the same risk if it is wrong.

Second: decentralized clinical trials. Participants can now complete certain study visits remotely, using telemedicine, digital monitoring tools, and home health visits, rather than traveling to a clinic for every appointment. This matters for the enrollment crisis described below. If a patient with a rare disease has to fly to a major academic medical center every two weeks for a two-year trial, many will not enroll. If they can complete some visits from home, the population that can participate expands.

Third: electronic systems and data integrity. E6(R3) provides specific, updated guidance on how electronic health records, wearables, and remote data collection tools can be validated and used in trials. As more trials collect data outside traditional clinical settings, the standards for what counts as reliable data have to keep pace.

The history of GCP is the history of what happens when research goes wrong.

The Nuremberg Code of 1947 was written in direct response to experiments conducted on prisoners in Nazi concentration camps. It established, for the first time, that voluntary consent from research participants is absolutely essential. The Declaration of Helsinki, adopted by the World Medical Association in 1964, built on the Nuremberg Code and established physician ethics in human research. The Belmont Report of 1979, written in the aftermath of the Tuskegee Syphilis Study, codified the principles of respect for persons, beneficence, and justice that still form the ethical foundation of clinical research today.

ICH-GCP E6 was first harmonized across the US, European Union, and Japan in 1996, establishing a single international standard that allows data from trials conducted in one country to be accepted by regulators in another. The 2016 revision, E6(R2), introduced risk-based monitoring. The 2023 revision, E6(R3), addressed decentralized trials and digital data.

Each revision represents something the system learned. The pace of updates has accelerated because the pace of change in clinical research has accelerated. The rulebook is not static because the science is not static.

When a sponsor believes the clinical evidence is sufficient to support approval, they submit a New Drug Application, or NDA, to the FDA. For biologic drugs, including vaccines and monoclonal antibodies, the equivalent is a Biologics License Application, or BLA. These documents are enormous. A typical NDA contains data from every clinical trial conducted, manufacturing quality information, proposed labeling, and a complete safety database from every patient who received the drug.

The FDA assigns a team of physicians, pharmacologists, statisticians, and chemists to review the application. For drugs designated as Priority Review, addressing serious conditions with unmet medical need, the review target is six months. For standard reviews, the target is ten months.

For many applications, the FDA convenes an advisory committee: a panel of independent scientists and clinicians who review the data and provide a non-binding recommendation on whether to approve. These meetings are public. Anyone can watch the advisory committee for a drug that could affect them, hear the evidence debated, and understand the basis for the regulatory decision.

The way drugs get approved is the system answering four questions from the Skeptic's Toolkit on your behalf:

Q4. Who funded this? Phase 3 trial funders are disclosed in the published paper.

Q8. How many people were studied? Replicated? Phase 3 trials enroll hundreds to thousands. FDA approval typically requires two pivotal trials.

Q9. What do independent experts say? The FDA Advisory Committee. Independent scientists with no employment ties to the sponsor reviewing the data on the public record.

Q10. If we are wrong about this, what would that mean? Post-market surveillance, recall authority, and the ability to pull a drug from the shelf.

When the system works, it is auditing a claim the way a careful reader would audit a claim. When it fails, it fails by skipping one of these questions.

The Skeptic's Toolkit and other worked examples are in Issue 003b.

The drug pipeline is not a conveyor belt. It breaks. Often. And it breaks at a specific, uncomfortable place: the clinical trial.

55 percent of clinical trials are prematurely terminated. Not because the science failed. Not because the drug was proven unsafe. Because not enough people enrolled. The trial could not fill its seats. The data was never completed. The question the trial was trying to answer went unanswered. Permanently.

This is not an industry statistic. It is a treatment gap. When a Phase III trial closes for poor enrollment, the drug it was testing goes back to the shelf. In most cases, it never comes back. The next generation of patients with that condition does not have an approved option that might have existed. The evidence died with the trial.

More than 1 in 4 rare disease trials were terminated between 2016 and 2020 due to low patient rates. For pediatric cancers, diseases affecting children, the number is even higher, because the patient population is small and geographically scattered. 48,027 patients enrolled in trials that closed unable to answer their primary research question. Their participation was real. Their data was collected. The answer was never reached.

There is another reason trials sit unfilled, and it is quieter than distrust. Many people simply do not know to look. A patient receiving a serious diagnosis is not always told that a trial exists for their condition. The treating physician may not know either, because keeping current with thousands of active studies across every therapeutic area is its own full-time job, and most clinicians do not have the time for it. Caregivers searching for options for a parent or a child rarely encounter ClinicalTrials.gov on their own until they may be lucky enough to connect with a Patient Advocacy Group that does know about a possible trial. Trials with available seats sit half-empty while the very people they were designed for sit in waiting rooms unaware the trial exists. The pipeline goes silent not only because people refuse to enroll, but because no one has told them enrollment is possible.

The most common reason people do not enroll in clinical trials is distrust. Distrust built from real betrayals: the opioid crisis, Tuskegee, Vioxx, supplement labeling that bore no relationship to the contents. That distrust is not irrational. It is the residue of genuine institutional failures. But when distrust of the research process prevents enrollment, it does not punish the institutions that earned the distrust. It punishes the patients who never received the treatment the trial was trying to prove.

This is what health misinformation costs. Not just confusion at a dinner table. Not just arguments in a group chat. Treatments that were tried, tested, and abandoned because not enough people showed up to complete the test.

Root to Rx exists to close that gap. Not by telling people what to believe about any specific treatment. By giving people the tools to evaluate the evidence themselves. A public that can read a study is a public that can trust a process, not just an institution. That trust is what fills clinical trials. That is what gets treatments over the finish line.

Out of every 10,000 compounds tested in early drug discovery, about 1 makes it to FDA approval. The 9,999 that fail along the way are not waste. They are the reason the one that succeeds is trustworthy.

The average approved drug takes 10 to 15 years and costs around $2.6 billion to bring to market. That figure is from a peer-reviewed study (DiMasi et al., 2016) that included every failure in the count, not just the successes.

55 percent of clinical trials are terminated before they can answer the question they were designed to answer. Not because the drug failed. Because not enough people enrolled.

ClinicalTrials.gov lists every active trial in the United States, searchable by condition. Most patients have never heard of it. Many physicians have not either. That is one of the reasons good trials fail to fill.

We are building something called the Wall of Sams. Real stories from real people who used to be skeptical of clinical research, pharmaceutical development, or vaccine science, and who have since gotten Rooted. Not converts. Not cheerleaders. People who learned to evaluate evidence rather than just accept or reject conclusions.

If you have a story, we want it. Reply to this email with three sentences: what you used to believe, what changed it, and where you are now. The best ones will appear in a future issue with your permission.

55 percent of clinical trials fail because not enough people enroll. Root to Rx exists to close that gap by building a public that understands the process. One person you forward this to is one more Rooted reader. One more branch.

Branch someone in. One referral earns you the Skeptic's Toolkit, 10 questions that will change how you evaluate any health claim.

This issue continues with two web-only sections that close out the dual-track structure. The Root Room is where Debby has her first encounter with the Skeptic's Toolkit. The Informed walks through the full regulatory and statistical mechanics behind everything you just read. Both live on the website at the link below.

Click to read: The Root Room and The Informed editions online only.

Everything in this issue is sourced. Nothing is cited from memory alone.

1. DiMasi JA, Grabowski HG, Hansen RW. Innovation in the pharmaceutical industry: new estimates of R&D costs. Journal of Health Economics. 2016;47:20-33. doi.org/10.1016/j.jhealeco.2016.01.012

2. Hwang TJ, Carpenter D, Lauffenburger JC, Wang B, Franklin JM, Kesselheim AS. Failure of investigational drugs in late-stage clinical development and publication of trial results. JAMA Internal Medicine. 2016;176(12):1826-1833.

3. Lamberti MJ, Vaitkevicius H, Ashford C, et al. Assessing clinical trial termination due to poor enrollment. Therapeutic Innovation and Regulatory Science. 2020;54(2):498-504.

4. ICH. E6(R3) Guideline for Good Clinical Practice. International Council for Harmonisation, November 2023. ich.org/page/efficacy-guidelines

5. US FDA. Step 3: Clinical Research. fda.gov/patients/drug-development-process/step-3-clinical-research

6. US FDA. 21 CFR Part 312: Investigational New Drug Application. ecfr.gov

7. The Belmont Report: Ethical Principles and Guidelines for the Protection of Human Subjects of Research. HHS, 1979. hhs.gov/ohrp/regulations-and-policy/belmont-report

8. National Cancer Institute. Cancer Statistics. seer.cancer.gov

9. ClinicalTrials.gov. National Library of Medicine. clinicaltrials.gov

10. World Medical Association. Declaration of Helsinki. Ethical Principles for Medical Research Involving Human Subjects. 2013 revision. wma.net