By Justin Yamashita, MSc. Benchtop, site, CRO: three levels of basic and clinical research, explained without spin.

The receipts for everything Monday and Wednesday claimed.

The deeper dive: trial methodology, mechanisms, and the science behind the story.

The single most useful fact in this issue is a distinction almost nobody draws. One of the two oversight bodies discussed this week is required by federal regulation. The other one, for most trials, is not.

The Two Bodies, By Regulation

Institutional Review Board (IRB). Required under 21 CFR Part 56 for FDA regulated clinical investigations, and under 45 CFR Part 46 for federally funded human subjects research. Under 21 CFR 56.109(a) an IRB has authority to approve, require modifications in order to secure approval, or disapprove research covered by the regulations. Under 21 CFR 56.109(f) it conducts continuing review at intervals appropriate to the degree of risk and not less than once per year.

Composition is specified, not left to the institution. Under 21 CFR 56.107, an IRB must have at least 5 members, including at least one member whose primary concerns are scientific, at least one whose primary concerns are nonscientific, and at least one member who’s not otherwise affiliated with the institution and isn’t part of the immediate family of a person who is affiliated with it.

Data Monitoring Committee (DMC), also called a Data Safety Monitoring Board (DSMB) or an Independent Data Monitoring Committee (IDMC). A group of individuals with pertinent expertise, typically statisticians and clinicians, with no financial or professional stake in the trial's outcome, that reviews accumulating unblinded outcome data at intervals set out in a written charter and makes recommendations to the sponsor about continuing, modifying, or terminating the trial.

Note the verb, “recommends”. FDA's guidance describes the committee's function as advising the sponsor. The sponsor holds the decision. In practice, a sponsor that overrides its own independent committee has to explain that later to regulators, to investigators, to journal editors and to an IRB, which is a real constraint, but it’s a constraint of accountability rather than a switch on the wall.

Who Actually Requires a Monitoring Committee

This is where most explanations of clinical trial oversight get sloppy.

There is one clear federal mandate. Under 21 CFR 50.24, research conducted under an exception from informed consent for emergency research must have an independent data monitoring committee. That is a narrow category, and it’s the only place US regulation flatly requires one.

Outside it, the FDA's position is a strong recommendation. Its guidance advises sponsors to consider a DMC and recommends one particularly where trial participants are at risk of serious morbidity or mortality. NIH policy sits in a similar place: its 1998 data and safety monitoring policy provides that Institutes and Centers may require a monitoring board, and treats one as expected for multi site trials involving interventions that carry potential risk to participants. Neither of those is the same as a blanket statutory requirement for every phase 3 trial, and it should not be described as one.

There’s a date problem here too. FDA's final DMC guidance, "Establishment and Operation of Clinical Trial Data Monitoring Committees," was issued in March 2006. The FDA issued a Revision 1 draft, "Use of Data Monitoring Committees in Clinical Trials," in February 2024. As of this issue's publication that revision is still a draft and not in effect, so the operative final guidance in this area is 20 years old.

What an IRB Is Not For

When the FDA published the final IRB rule on 27 January 1981, the preamble said something that has largely fallen out of the public conversation about trial oversight. The FDA deleted from the rule any reference to an IRB responsibility to assure the validity and reliability of data, stating that IRBs have no such obligation. It stated that IRBs are generally not in a position to follow human subjects. And it stated that IRBs generally will not have the scientific competence to judge whether one treatment has been shown safer or more effective than another, that determination resting with investigators, sponsors and ultimately the FDA.

That’s the regulator defining the boundary of its own safeguard, when it created it. An IRB clears a study to begin and reviews it periodically. It’s not a real time data monitor, and it was never designed to be. Anyone who cites IRB approval as evidence that a running trial's data look fine is citing the wrong body.

The Office Behind the Required Board

The IRB requirement has 2 legal homes and 2 different enforcers. For FDA regulated clinical investigations it is 21 CFR Part 56, enforced by FDA. For research conducted or supported by HHS it is 45 CFR Part 46, the Common Rule, enforced by the Office for Human Research Protections, OHRP. A sponsor funded trial at a site with no federal funding is FDA territory. Many academic sites sit under both.

OHRP does 4 things. It investigates allegations that an institution violated the Common Rule. It writes the interpretive guidance that tells institutions what the regulation means in practice, which is the function IRBs actually run on day to day. It educates IRB members and investigators. And it administers the Federalwide Assurance, the written promise an institution makes to follow the regulations, without which that institution cannot conduct HHS supported human subjects research. Its remit covers more than 13,000 institutions.

That office is now very small. STAT reported on 5 June 2026 that OHRP has 10 employees, has gone close to a year without a director in any of its 3 divisions, policy, compliance and education, and has lost more than half its staff since February 2025. It has 3 investigators. Its budget has been flat at roughly $6 million since 2013, down from about $7.5 million in 2003. Between 1 October 2025 and 31 March 2026 it received 103 complaints and 352 incident reports, and had 3 active cases. Free educational programming has stopped. The comparison that frames all of it: when OHRP was established in 2000, the expert recommendation was a minimum of 42 staff.

Those figures come from one outlet's reporting, not from an HHS budget justification or an org chart, and I cannot verify them against a primary federal record. They are specialty journalism, of which I am not. I am using them to establish that an institutional event happened, which is what that tier of source is good for, and I am telling you their provenance so you can weigh them yourself.

The advisory committee is gone too. The Secretary's Advisory Committee on Human Research Protections, SACHRP, an 11 member volunteer panel of bioethics and regulatory experts that met 3 times a year to advise OHRP, was disbanded in spring 2025. STAT reported it on 3 April 2025 from an internal email it obtained. HHS made no public announcement. The committee's charter had been renewed in October 2024, so it was ended mid charter. SACHRP is the body that would normally pressure test regulatory language like exemption categories in public, on the record, before a proposal went out.

What Is Queued Behind It

On 14 August 2026, HHS published a Unified Agenda entry under RIN 0937-AA16, titled "Human Research Protections: Exemptions and Clarifying Provisions Related to Institutional Review Board Oversight." The abstract proposes revising 45 CFR part 46 subpart A to clarify terminology, expand exemptions for certain low risk research activities, and enable flexibilities for regulatory review of de minimis protocol changes. HHS designated it Deregulatory under Executive Order 14192, the January 2025 order requiring agencies to identify regulations for removal. The RIN status field reads "first time published in the Unified Agenda," meaning this is a new item rather than a long running technical correction.

Four things about that entry are easy to get wrong, and they are the reason this issue is covering it at all. It’s not a proposed rule. The Unified Agenda is a planning document an agency publishes about twice a year listing what it intends to write, and entries routinely slip, change or disappear. There’s no regulatory text, no docket and no comment period, so there’s nothing to submit an opinion on today. The target for a proposed rule was July 2026 and nothing has published as of this issue. And a search of OIRA's review records returns no entry for this RIN, which matters because review at OMB normally precedes publication by 30 to 90 days. On that evidence a proposal does not look imminent. A negative search result proves less than it feels like it proves, so check the Federal Register directly before repeating that.

Expanding exemptions is also not, by itself, a weakening. The revisions that created the current 8 category exemption structure and added limited IRB review were published as a final rule in January 2017 and were themselves a burden reduction exercise. IRB capacity is finite, and time spent on a minimal risk survey is time not spent on a first in human dose escalation. That complaint is decades old and it crosses party lines. None of this touches FDA regulated interventional research either, which lives under 21 CFR 50 and 56 and is not amended by a change to 45 CFR 46.

Here’s the part that’s not speculation. An exemption does not take a study out of the rules, it moves who decides. The judgment about whether the rules apply shifts from an independent board to the institution itself, and the whole arrangement rests on an office of 10 people checking that more than 13,000 institutions judged it right. An exemption regime is safe when 3 conditions hold: the categories are drawn tightly, somebody independent checks that a study actually fits the category it claims, and guidance exists telling institutions how to apply them. The second and third are OHRP functions. The rule is a plan and the staffing is a fact, and those are different kinds of claim that this issue is not going to blur. But they land on the same office. The same regulatory text is a different object depending on who is administering it.

One more thing, for anyone who wants to follow this rather than react to it. When the proposal publishes, don’t read the exemptions list first. Read §46.102, the definitions. What counts as research, what counts as a human subject, what counts as identifiable private information. A change to a definition removes more from oversight than an exemption expansion does, and it reads as housekeeping.

The Legal Timeline

1932 to 1972. The US Public Health Service conducts the Tuskegee Study of untreated syphilis in Black men in Macon County, Alabama, without informed consent.

1943. Penicillin becomes the treatment of choice for syphilis and becomes widely available, per CDC's own timeline of the study. The study does not offer it to participants.

25 July 1972. Associated Press reporter Jean Heller publishes the story. A federally appointed advisory panel reviews the study, and it is terminated in November 1972.

12 July 1974. The National Research Act, Public Law 93-348, is signed. It requires IRB review of federally funded human subjects research and establishes the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research.

18 April 1979. The Belmont Report is published at 44 Federal Register 23192. Respect for persons, beneficence, justice.

27 January 1981. FDA publishes the final IRB rule at 46 Federal Register 8958, including the language above on what IRBs are not responsible for.

31 May 2002. The Women's Health Initiative estrogen plus progestin arm is stopped early on the recommendation of its monitoring board. Detailed below.

21 January 2019. The revised Common Rule takes general effect. Its single IRB requirement for cooperative research takes effect 20 January 2020.

February 2025 onward. OHRP loses more than half its staff, per STAT reporting published in June 2026, and goes close to a year with no director in any of its 3 divisions.

3 April 2025. STAT reports that the Secretary's Advisory Committee on Human Research Protections has been disbanded, based on an internal email it obtained. HHS issues no public notice. The committee's charter had been renewed in October 2024.

9 September 2025. FDA announces availability of ICH E6(R3) Good Clinical Practice as final guidance. E6(R3) restructures GCP around principles and annexes and pushes trial oversight toward risk based quality management rather than uniform monitoring of everything. It is guidance, not law, and its effect runs through sponsor and regulator adoption.

14 August 2026. HHS publishes RIN 0937-AA16 in the Unified Agenda, a planned rewrite of Common Rule subpart A, designated Deregulatory under Executive Order 14192. As of this issue no proposed rule has published.

The 2002 Stop, In Full

The Women's Health Initiative estrogen plus progestin trial randomized 16,608 postmenopausal women aged 50 to 79 at 40 US clinical centers, 8,506 to conjugated equine estrogens plus medroxyprogesterone acetate and 8,102 to placebo. Planned duration was 8.5 years.

At a regularly scheduled meeting on 31 May 2002, the Data and Safety Monitoring Board found that invasive breast cancer had crossed a pre specified monitoring boundary and that the global index of harms and benefits was unfavorable, and recommended stopping. The arm ended after a mean of 5.2 years of follow up. Participants began receiving letters on 8 July 2002 instructing them to stop study medication, and the principal results were released publicly on 9 July 2002 ahead of print publication.

Hazard ratios, with nominal 95 percent confidence intervals, from Rossouw et al.

Invasive breast cancer 1.26, CI 1.00 to 1.59.

Coronary heart disease 1.29, CI 1.02 to 1.63.

Stroke 1.41, CI 1.07 to 1.85.

Pulmonary embolism 2.13, CI 1.39 to 3.25.

Hip fracture 0.66, CI 0.45 to 0.98.

Colorectal cancer 0.63, CI 0.43 to 0.92.

Two things a careful reader should notice. The breast cancer interval's lower bound sits at 1.00, which is exactly the kind of detail that gets sanded off in retellings, and the decision did not rest on that number alone. It rested on a monitoring boundary agreed before enrollment plus a global index weighing harms against benefits. And 2 of these results point the other way. Hip fracture and colorectal cancer both fell. This trial is routinely described as proving hormone therapy is dangerous, and what it actually produced was a mixed ledger that an independent group judged, on balance, unfavorable for this population at this dose over this duration.

The structural point stands regardless of how anyone reads the ledger. The investigators seeing those women could not have made that call, because they could not see the comparison. That’s not an accident of that trial. It’s the design of every blinded trial.

How Often Does This Actually Happen

Rarely, and it has been getting less rare.

A systematic review published in JAMA in 2005 identified 143 randomized controlled trials stopped early for benefit and found that the proportion of trials in high impact journals stopped early for benefit rose from 0.5 percent in 1990 to 1994 to 1.2 percent in 2000 to 2004. Truncated trials had recruited a median of 63 percent of their planned sample and stopped after a median of 13 months of follow up.

Two limits on that number, stated plainly. It measures stopping for benefit, not for harm, and I could not find an aggregate estimate for stopping for harm of comparable rigor, so this issue does not offer one. And the review is now 21 years old, which is exactly the kind of stale anchor this newsletter spent last issue warning about. It is the best sourced figure I have, and its date is part of the claim.

A recent example, with its date attached. In the NIH sponsored STOMP trial of tecovirimat for clade II mpox, the study's monitoring board recommended halting randomized enrollment after an interim analysis found no difference in lesion resolution time and a very low probability that continuing would show benefit. NIH announced those interim results in December 2024. The drug was safe. It simply did not work for that indication, and the board saw it before the field did.

Citation Tier Breakdown

Tier 1, peer reviewed. Rossouw et al., JAMA 2002, the load bearing citation for the WHI stop. Montori et al., JAMA 2005, for early stopping frequency.

Tier 2, regulatory and legal. 21 CFR 56.107, 56.109 and 50.24. 45 CFR Part 46. Public Law 93-348. The Belmont Report at 44 FR 23192. FDA's 1981 IRB final rule at 46 FR 8958. FDA's 2006 DMC guidance and its February 2024 draft revision. ICH E6(R3) as adopted by FDA in September 2025. The Unified Agenda entry for RIN 0937-AA16, read directly off reginfo.gov. Executive Order 14192. The January 2017 Common Rule final rule.

Tier 3, agency and institutional record. CDC's Tuskegee timeline. NIH's 1998 data and safety monitoring policy. NIH's December 2024 STOMP announcement. ClinicalTrials.gov protocol data element definitions. The October 2024 SACHRP charter renewal notice.

Attribution note. The 25 July 1972 publication date for the Heller story comes from AP's own retrospective account, not from CDC's timeline, which gives the year only. Separately, every figure in this issue about OHRP's staffing, budget and caseload, and the fact of SACHRP's disbanding, comes from STAT's reporting and not from a primary federal record. That’s specialty journalism, and it’s labeled inline every time it appears below. It’s good enough to establish that an institutional event happened. It’s not peer reviewed and it’s not carrying any clinical claim here.

Skeptic's Toolkit, This Issue

Question 1: what is the actual source of this claim? For everything above, the source is either the Code of Federal Regulations, a Federal Register notice, an agency page, or a peer reviewed paper, and every one is in the list below. If a claim in this issue is not in that list, treat it as my opinion.

Question 9: does this apply to my specific situation? The IRB and DMC framework is a description of a regulatory structure. It is not a statement about any particular trial. Whether a specific study has a monitoring committee is a fact about that study, findable in its registry record and answerable by its coordinator.

Next issue: the manufacturing floor, and why the same molecule made in 2 different plants is not automatically the same product.

Refer a friend

If you read this far, you're the reader this thing is for. The best thanks is sending it to one person who'd read it too. Series 1 cards are the reward.

NEW TO ROOT TO RX? START HERE

Issue 003b: how to evaluate any health claim using the Skeptic's Toolkit. Start here.

Issue 001: why medical distrust is rational, and what to do with it anyway.

Issue 004: how a drug gets from a lab to your pharmacy. The pipeline in plain language.

All back issues are at RootToRx.com.

References

1. US Food and Drug Administration. Institutional Review Boards, 21 CFR 56.107. ecfr.gov

2. US Food and Drug Administration. IRB review, 21 CFR 56.109. ecfr.gov

3. US Food and Drug Administration. Exception from informed consent requirements for emergency research, 21 CFR 50.24. ecfr.gov

4. US Department of Health and Human Services. Protection of Human Subjects, 45 CFR Part 46. ecfr.gov

5. US Food and Drug Administration. "Protection of Human Subjects; Standards for Institutional Review Boards for Clinical Investigations." Final rule, 46 Federal Register 8958, 27 January 1981. fda.gov

6. National Research Act of 1974. Public Law 93-348, signed 12 July 1974. govinfo.gov

7. The Belmont Report. 44 Federal Register 23192, 18 April 1979. hhs.gov/ohrp

8. US Department of Health and Human Services, Office for Human Research Protections. "Revised Common Rule," compliance dates 21 January 2019 and 20 January 2020. hhs.gov/ohrp

9. Centers for Disease Control and Prevention. "The Tuskegee Timeline." cdc.gov

10. Rossouw JE, Anderson GL, Prentice RL, et al. "Risks and Benefits of Estrogen Plus Progestin in Healthy Postmenopausal Women: Principal Results From the Women's Health Initiative Randomized Controlled Trial." JAMA. 2002;288(3):321-333. PMID 12117397

11. Montori VM, Devereaux PJ, Adhikari NKJ, et al. "Randomized Trials Stopped Early for Benefit: A Systematic Review." JAMA. 2005;294(17):2203-2209. PMID 16264162

12. US Food and Drug Administration. "Establishment and Operation of Clinical Trial Data Monitoring Committees." Guidance for clinical trial sponsors, final, March 2006. fda.gov

13. US Food and Drug Administration. "Use of Data Monitoring Committees in Clinical Trials." Draft guidance for industry, Revision 1, February 2024. Draft, not in effect. fda.gov

14. US Food and Drug Administration. "E6(R3) Good Clinical Practice." ICH guidance for industry, availability announced 9 September 2025. fda.gov

15. National Institutes of Health. "NIH Policy for Data and Safety Monitoring." NOT-98-084, 1998. grants.nih.gov

16. National Institutes of Health. STOMP trial of tecovirimat for mpox, interim results announcement, December 2024. nih.gov

17. ClinicalTrials.gov. Protocol registration data element definitions, oversight information. clinicaltrials.gov

18. US Department of Health and Human Services, Office for Human Research Protections. "Human Research Protections: Exemptions and Clarifying Provisions Related to Institutional Review Board Oversight." Unified Agenda entry, RIN 0937-AA16, Proposed Rule Stage, published 14 August 2026. Designated Deregulatory under Executive Order 14192. RIN status: first time published in the Unified Agenda. reginfo.gov

19. Executive Order 14192, "Unleashing Prosperity Through Deregulation," January 2025. federalregister.gov

20. Office of the Federal Register. "Introduction to the Unified Agenda of Federal Regulatory and Deregulatory Actions," 14 August 2026. federalregister.gov

21. US Department of Health and Human Services. "Federal Policy for the Protection of Human Subjects." Final rule, Federal Register, 19 January 2017. The revision that created the current 8 category exemption structure. federalregister.gov

22. STAT. "Tiny HHS office tasked with protecting research participants' safety is running on fumes." 5 June 2026. Specialty journalism, not peer reviewed. Source for all OHRP staffing, budget and caseload figures in this issue. statnews.com

23. STAT. "Federal advisory panel on ethical, legal issues in human health research disbanded." 3 April 2025. Specialty journalism, reported from an internal email obtained by the outlet. HHS issued no public notice. statnews.com

24. US Department of Health and Human Services. "Charter Renewal of the Secretary's Advisory Committee on Human Research Protection." Federal Register, 3 October 2024. federalregister.gov

New here? Root to Rx runs three pieces a week. Monday you watch a real health claim try to work on someone. Wednesday you get the tools to run it yourself. Thursday you get the full evidence so you can check my work. Written by a clinical researcher, free, at RootToRx.com.

For educational purposes only. Nothing in this newsletter is medical advice. Talk to your doctor before making any health decision.

Instagram: instagram.com/roottorx | LinkedIn: linkedin.com/company/roottorx | Facebook Lab: facebook.com/groups/theroottorxlab | Subscribe: RootToRx.com

You are receiving this because you subscribed at roottorx.com.

Disclosure: Root to Rx is published independently by Open Label Media LLC. Views expressed are personal views of Justin Yamashita and do not represent his employer or any affiliated organization. No employer resources or proprietary information are used. Every claim is sourced from publicly available materials.

Root to Rx | 228 Park Ave S, #29976, New York, New York 10003, United States