Grand Rounds Archive

Grand Rounds September 27, 2024: Azithromycin for Childhood Mortality: Randomizing Entire Countries (Thomas Lietman, MD)

Posted on by Sophia Ramirez

Speaker

Thomas Lietman, MD
Director
Francis I. Proctor Foundation
University of California, San Francisco

Keywords

Child Mortality; Antibiotics; Large Simple Trial

Key Points

  • Trachoma is a disease that affects the human eye, caused by infection with a strain of chlamydia. The scar tissue that is left behind warps the shape of eyelid, irritating the cornea and eventually causing full blindness. Antibiotics have been shown to reduce trachoma infection, with azithromycin being the preferred treatment.
  • Dr. Lietman described a series of mass azithromycin trials, including the “Azithromycine pour la Vie des Enfants au Niger: Implementation et Recherche” (AVENIR) trial in Niger and the “Macrolides Oraux pour Réduire les Décès avec un Oeil sur la Résistance” (MORDOR) trial in Niger, Tanzania, and Malawi.
  • In one of these trials, the mass distribution of azithromycin for trachoma control was shown to reduce the overall child mortality rate. The research team sought to test these findings in a younger age group (preschool-aged children) and with a larger sample size.
  • The MORDOR trial found that biannual azithromycin distribution reduced overall mortality in 1 – 59 month old children, though more so for the younger children. The WHO amended their guidelines to recommend administration of azithromycin in high mortality Sub-Saharan African settings to children 1 – 11 months old.
  • The release of the WHO guidelines was followed by the AVENIR trial, which – in addition to re-testing whether mass azithromycin distribution reduced childhood mortality in 1 – 59 month-olds – compared the reduction in mortality when treating only 1 – 11 month-olds as opposed to 1 – 59 month-olds. They also monitored antibiotic resistance.
  • The research team found that the 1 – 11 month-olds had significantly better outcomes in the latter treatment arm, in which children 1 – 59 months old received Azithromycin. The herd effect led to an indirect benefit for the younger children when the older children were treated.
  • The researchers successfully adapted toward the 1 – 59 month-old treatment group after 12 months, using a tempering algorithm to avoid dramatic swings in allocation probabilities.
  • They found that trachoma distributions and the childhood mortality distributions do select for macrolide resistance in a number of different bacteria, with resistance decreasing when the programs are stopped. Since they are only treating 1/6 of a population, there is less antibiotic pressure than with trachoma programs that treat an entire community.
  • AVENIR 2 will expand AVENIR to the entire country of Niger.

Discussion Themes

The ethical considerations of the trial were enormous and included consent and post-trial availability. It was very important for the research team A) to have strong, local collaborators and B) not to impose Western academic ways of thinking when partnering with these communities.

The researchers were mindful of the risk that, with adaptive trials, you can lose a treatment arm if you adapt too much early on. Retrospectively, Dr. Lietman noted, they were overly cautious when it came to tempering. Ultimately, everyone was treated by 24 months.

The team is hoping to eventually create a platform to monitor and coordinate treatment efforts between trials.

Grand Rounds September 20, 2024: Similarities and Differences Between Pragmatic Trials and Hybrid Effectiveness-Implementation Trials (John Fortney, PhD)

Posted on by Sophia Ramirez

Speaker

John Fortney, PhD
Professor
Department of Psychiatry and Behavioral Sciences
University of Washington
Senior Research Career Scientist
HSR Center of Innovation for Veteran-Centered and Value-Driven Care
VA Puget Sound Health Care System

Keywords

Pragmatic; Implementation; Hybrid-Effectiveness; Trial Types

Key Points

  • Pragmatic trials are primarily designed to determine the effectiveness of a clinical intervention under the usual conditions in which it will be applied. A hybrid-effectiveness implementation trial assesses both clinical effectiveness and implementation.
  • Dr. Fortney walked through the translational research pipeline, which flows from explanatory trials to pragmatic trials to implementation research. Ideally, the end product is improved processes and outcomes.
  • While the differences between explanatory, pragmatic, and implementation trials is well understood, the differences and similarities between pragmatic trials and hybrid type trials are not well understood.
  • Pragmatic Trials (PTs) and Hybrid Type 1 Trials (HT1) aim to test the effectiveness of clinical interventions, with patient outcomes as their primary outcomes. Hybrid Type 3 Trials (HT3) look at whether a given implementation strategy successfully promotes intervention uptake. Hybrid Type 2 Trials (HT2) compare two clinical interventions and one or two implementation strategies.
  • Whereas HT1s utilize artificial implementation strategies, PTs, HT2s, and HT3s utilize practical implementation strategies that they expect to be replicable outside of a research context.
  • Dr. Fortney provided resources, including a question tree and case examples, to aid researchers in identifying or choosing a trial type. He also provided several guidelines, including:
    • HT1 trial types should be used to determine whether a clinical intervention can be effective when delivered in routine care.
    • Pragmatic and HT2 trial types should be used to determine whether an evidence-based clinical intervention(s) is effective when delivered with practical implementation strategies in routine care.
    • HT2 and HT3 trial types should be used to determine whether practical and novel implementation strategies successfully promote the uptake of evidence-based clinical interventions.

Discussion Themes

The distinction between artificial implementation strategies and evidence-based, or practical, implementation strategies is not always black and white, and will often fall somewhere on the Artificial → Practical continuum.

This research team laid out some of the subtler differences in research objectives between study types, and the elements studies need to successfully address those objectives. Reviewers and researchers alike need to be careful in thinking these distinctions through and understanding them.

Grand Rounds September 6, 2024: Conventional, Complementary, and Integrative Pain Therapies in a Military Population with Chronic Musculoskeletal Pain: Results of a Pragmatic Clinical Trial Using SMART Design (Ardith Z. Doorenbos, PhD, RN; Diane M. Flynn, MD, MPH)

Posted on by Sophia Ramirez

Speakers

Ardith Z. Doorenbos, PhD, RN
Professor
Department of Biobehavioral Nursing Science
College of Nursing
University of Illinois, Chicago

Diane M. Flynn, MD, MPH
Primary Care Pain Management Advisor
Interdisciplinary Pain Management Center
Madigan Army Medical Center
Tacoma, WA

Keywords

Chronic Pain; Military; Rehabilitative Care; Pain Therapies; SMART Design

Key Points

  • Physical and occupational therapies are standard rehabilitative care (SRC) for chronic pain, and a growing body of evidence supports complementary and integrative health (CIH) therapies (such as acupuncture, chiropractic, yoga and massage).
  • Few studies have explored the optimal duration, sequence and combination of SRC and CIH to manage chronic pain.
  • The study team investigated the effectiveness of starting treatment for pain with SRC versus CIH therapies. Their primary outcome was the Pain Impact Scores of active-duty service members with chronic pain.
  • Their study design, a Sequential Multiple Assignment Randomized Trial (SMART) Design, randomized participants to CIH or SRC for three weeks each. At that point, participants who were improving continued on in their assigned treatment arm, while those who were not improving were randomized to either the other treatment arm or a combination arm.
  • Compared to baseline pain levels, both groups improved significantly up to six months out. After three weeks, the group that received SRT showed less improvement the group that received CIH therapies. However, by the end of the six-week mark, as well as at three- and six-month follow-ups, there wasn’t a significant difference in average improvement.
  • The study team’s ability to make assessments with regard to the duration of treatment is limited; participants were given the option to continue therapy for up to six weeks, so some were engaged in active therapy during the follow-up period.
  • These findings lend support to expanding access to CIH approaches. Clinicians can feel confident recommending patients start with CIH therapies if that is the patients’ preference.

Discussion Themes

Patients in the CIH arm and SRT arm received therapy for a total of 6.5 and 3 hours per week, respectively. At first blush, this indicates that a treatment hours effect could have contributed to the benefits seen in the first three weeks of the CIH arm; in practice, due to lower-than-intended participation, the actual treatment hours were only slightly higher in the CIH arm than the SRT arm.

Whether it’s caused by pain, combat experience, or adverse childhood experiences, many members of the study population experience a great deal of allostatic stress and kinesiophobia: “When I move, I hurt.” One of the study team’s theories as to why participants initially responded more positively to the CIH arm is that there was less movement involved.

Future directions may include examining the biological or neurological mechanisms underlying the interventions’ impact.

Dr. Flynn shared a resource that she noted would be helpful in defining SMART Designs and outlining some of the advantages and disadvantages:  https://jamanetwork.com/journals/jama/article-abstract/2800681

Grand Rounds August 23, 2024: Improving Quality of Life in COPD and Heart Failure: Unpacking a Successful Multicomponent Virtual Team Intervention (David B. Bekelman, MD, MPH; Lyndsay DeGroot PhD, RN, CNE)

Posted on by Sophia Ramirez

Speakers

David B. Bekelman, MD, MPH
Professor of Medicine and Psychiatry
Rocky Mountain Regional VAMC and University of Colorado Division of General Internal Medicine

Lyndsay DeGroot PhD, RN, CNE
Postdoctoral Fellow
University of Colorado
Anschutz Medical
Campus School of Medicine

Keywords

COPD; Heart Failure; Collaborative Care; Palliative Care

Key Points

  • Heart failure (HF) and Chronic Obstructive Pulmonary Disease (COPD) are common illnesses that share a number of symptoms, including persistent breathlessness and fatigue; depression; and anxiety.
  • Though HF and COPD patients could benefit from palliative care prior to the end of life, there aren’t enough palliative care specialists to provide care to the patient population.
  • ADAPT is a virtual intervention built on a collaborative care team approach. A nurse and social worker provide direct patient care and meet weekly with a collaborative care team – consisting of a primary care physician, palliative care specialist, and as-needed cardiologist and pulmonologist – to review individual cases.
  • The research team’s primary aim was to determine the effect of ADAPT on participant quality of life, measured using the FACT-G scale.
  • They concluded that a nurse and social worker palliative telecare team demonstrated early, persistent, clinically meaningful improvements in quality of life for high-risk outpatients with COPD, HF, and Interstitial Lung Disease.
  • The ADAPT virtual care model leveraged a team of nurses, social workers, and physicians across two large VA health systems to increase the reach of palliative care to common, serious non-cancer illnesses.
  • The intervention was relatively inexpensive, but implementing a multifaceted, personnel-intensive program posed its own challenges.
  • Future directions include investigating the intervention effect on utilization outcomes and end-of-life outcomes; adapting the model for patients with advanced liver or renal disease; testing other implementation and dissemination strategies; and investigating how to improve adoption.

Discussion Themes

The mortality rate in the population they enrolled was lower than expected. This may be because it is more difficult to enroll very sick patients from an outpatient population.

This study was designed to test the effect of the intervention on quality of life, an outcome that has a significant impact on patients. However, securing funding for work like this can be a challenge, as the program is not designed to reduce costs (unlike interventions that aim to reduce hospitalizations).

There is a need for creative solutions when it comes to funding this model; the research team is looking to collaboration with community partners and examples set by other, similar projects.

Grand Rounds August 16, 2024: Methodological Insights and Lessons Learned from Conducting a Pragmatic Randomized Trial on Surgical Face Masks (Runar Solberg, PhD; Atle Fretheim, PhD)

Posted on by Sophia Ramirez

Speakers

Runar Solberg, PhD
Scientist
Centre for Epidemic Interventions Research (CEIR)
Norwegian Institute of Public Health

Atle Fretheim, PhD
Research Director
Norwegian Institute of Public Health

Keywords

Face Masks; Masking; Respiratory Infection; Public Health; COVID-19

Key Points

  • Observational evidence supports a reduction in respiratory infection with face mask use. However, randomized trials face challenges–especially in achieving sufficient statistical power–that contribute to uncertainty in their findings.
  • The study team conducted a pragmatic randomized trial aimed at assessing the personal protective efficacy of wearing a surgical face mask in public settings.
  • Adherence varied between the control and intervention groups, with higher rates of adherence within the control group (no mask).
  • Participants were recruited from multiple locations across Norway. Advertising and the researchers’ appearance on Norwegian television, radio, and other media helped raise awareness about the study.
  • Reviewers acknowledged that the primary outcome–self-reported respiratory symptoms–was clinically meaningful and well-defined. However, the reliance on self-reported data led to concerns about bias and accuracy, as there was no in-person verification of masking.
  • To mitigate these concerns, future studies may opt to use registry data or incorporate mandatory PCR testing of participants.

Discussion Themes

An open and important question about masking effectiveness relates to the durability of the intervention: Does it flatten the curve early or delay the peak?

Collecting data on components of the primary outcome–i.e. if more objective symptoms like fever were more or less impacted by the intervention than more subjective outcomes like malaise–may help address concerns about the reliability of self-reported outcomes.

Efficacy depends on the degree of risk. Proving efficacy in a randomized controlled trial or pragmatic trial is tricky unless there is something to prevent. This depends on many factors that may cause heterogeneity of treatment effects (e.g. degree of exposure, physical proximity, virulence, etc.).

Grand Rounds July 19, 2024: Lessons From PRECIDENTD: A Pragmatic Comparative Effectiveness Trial (Deborah J. Wexler, MD, MSc; Brendan M Everett, MD, MPH)

Posted on by Sophia Ramirez

Speakers

Deborah J. Wexler, MD, MSc
Associate Professor of Medicine and Diabetes Unit Chief
Massachusetts General Hospital and Harvard Medical School

Brendan M. Everett, MD, MPH
Associate Professor of Medicine
Divisions of Cardiology and Preventive Medicine
Brigham and Women’s Hospital and Harvard Medical School

Keywords

Cardiovascular Health; Kidney Disease; Patient Engagement

Key Points

  • Comparative effectiveness of SGLT2 inhibitors (SGLT2i) and GLP-1 receptor agonists (GLP-1 RA) for cardiac and kidney outcomes is a major question in cardiac-kidney-metabolic health.
  • More broadly, comparative effectiveness of on-patent medications is crucial, yet not mandated, with little incentive (and some disincentive) for pharmaceutical companies to participate.
  • The PRECIDENTD trial will seek to test one approach to evaluating the comparative effectiveness of expensive new therapies. Their primary aim is to conduct a head-to-head evaluation of SGLT2i versus GLP-1 RA for the prevention of major adverse cardiovascular/kidney events and death.
  • Partnerships with patients, providers and stakeholders will be crucial to the success of PRECIDENTD. Patient engagement strategies include Community Engagement Studios, a Center for Effective Health Communication, and an interactive text messaging system to assess adherence.
  • The research team sees the trial, which involves two expensive, brand-name drugs and requires extensive collaboration, as a true test of a pragmatic trial of branded pharmaceutical medications.
  • Obtaining informed consent isn’t going to be pragmatic; it is time-consuming, challenging, and individualized.
  • Constrained funding in the feasibility phase may hamper the growth of the trial. In a bureaucratic research environment, the administrative burden can take time and energy from a study.

Discussion Themes

Patients have been more interested in participating given that this trial compares one effective, well-understood active drug to another, so there’s no risk of being assigned to a placebo group.

The pragmatic nature of the trial means that the study team is recruiting participants who can go either way when it comes to the two treatment options. The highest-risk people won’t be enrolled.

Grand Rounds August 9, 2024: Does Starting Buprenorphine Prevent Suicidal Behavior: What Trial Should We Emulate? (Gregory Simon, MD, MPH; Susan Shortreed, PhD)

Posted on by Sophia Ramirez

Speakers

Gregory Simon, MD, MPH
Investigator
Kaiser Permanente Washington Health Research Institute

Susan Shortreed, PhD
Investigator
Kaiser Permanente Washington Health Research Institute

Keywords

Buprenorphine; Self-Harm; Overdose; Target Trial Emulation

Key Points

  • There is some overlap between the two public health crises of our time: suicidal behavior and the Opioid Use Disorder (OUD)/opioid overdose crisis.
  • The research team looked at whether starting (vs not starting) buprenorphine reduces the subsequent risk of a self-harm event or suicide attempt among people diagnosed with OUD.
  • They conducted an observational study with target trial emulation, emulating a randomized trial where people with an OUD diagnosis might be randomly assigned to start or not start buprenorphine during an outpatient visit. They tracked the rates of self-harm, adverse outcomes more broadly, and overdose over 90 days.
  • Their analysis found that starting buprenorphine didn’t reduce the risk of self-harm behavior and (surprisingly) had no protective or beneficial effect when it came to opioid poisoning or overdose.
  • They found that when they stopped tracking outcomes for those who changed course – i.e. had initiated buprenorphine but stopped or had initially declined but initiated later – they observed protective effects consistent with previous findings. The risk of self-harm for people who stopped or had their treatment interrupted increased four-fold, and the risk of opioid overdose or poisoning increased five- or six-fold.
  • These findings reinforce concerns about high risk of adverse outcomes soon after stopping or interrupting buprenorphine treatment.
  • These findings also raise questions about the kind of target trials researchers should be emulating and what pragmatic trials might be considered for follow-up. Given these results, the research team is interested in research questions that center on the stopping or interruption of buprenorphine treatment.

 

Discussion Themes

Given that buprenorphine is often prescribed in shorter supply (7 – 14 days worth), the researchers were able to track the timing of events more accurately than they’d be able to with other chronic conditions. The profit motive of healthcare systems (i.e. billing insurance for treating an overdose) also improved their ascertainment of those events.

One limitation of the study is that they don’t have any way of ascertaining self-harm events or overdoses that never came to medical attention.

At the individual clinician level and at the policy level, the decision to treat OUD with buprenorphine is informed by the knowledge that interruptions in treatment are fairly common and can have adverse effects.

Another study design option is to compare people before and after initiating buprenorphine treatment, as opposed to comparing people who initiate and people who don’t. The tricky thing about that design is that an individual’s state is not consistent; there may have be an initiating event that precipitates the decision to start treatment.

 

 

Grand Rounds August 2, 2024: Precision Health to Population Health: Opportunities and Challenges for Gene Editing Therapies (Adrian Hernandez, MD, MHS)

Posted on by Sophia Ramirez

Speaker

Adrian Hernandez, MD, MHS
Vice Dean and Executive Director
Duke Clinical Research Institute
Duke University School of Medicine

Keywords

CRISPR; Gene Editing Therapies; Population Health

Key Points

  • The epidemiologic forecast for 2050 is gloomy, with projected increases in the prevalence of diabetes, high cholesterol, hypertension, obesity, and resulting cardiovascular conditions. Innovative technologies and therapies – particularly gene editing – could potentially prevent these conditions and have a positive long-term impact.
  • Gene editing technology could be used to develop targeted treatments for ultra-rare, rare, and common diseases. Several clinical trials are already underway, testing CRISPR-based therapies for cancers, chronic bacterial infections, lupus, and others.
  • A little over a decade after the first discovery using CRISPR technology was published in Science, the FDA approved the first CRISPR-based medicine for the treatment of any disease. The recent, rapid, and revolutionary advances in gene editing therapies simultaneously hold promise and indicate the importance of ensuring that this technology is developed safely and effectively.
  • When thinking through trial design, researchers need to consider durability. This means developing assessment methods for off-target effects and on-target safety validation, as well as planning for long-term monitoring.
  • There are also ethical and regulatory points to consider, especially around obtaining informed consent, navigating the unknowns and potential permanence of gene editing interventions, engaging underserved communities, preventing misinformation, and defining long-term follow-up.
  • The Precision Health Alliance was pulled together by Dr. Hernandez and his colleagues in an effort to bring together a variety of stakeholders – clinicians, researchers, healthcare systems, and life-science leaders – to think through what’s needed to develop research-to-care pathways with gene editing.
  • Many questions still need to be addressed: How do we handle more and more personalized health data appropriately? If we know we can cure someone, should we? How do we ensure we don’t create greater health disparities? How do we overcome mistrust in revolutionary science? How do we address common questions, commonly?

 

Discussion Themes

– What’s your view on the current temperature of public perception in this space? And as you look to the future, how important do you think it’ll be to improve public understanding and acceptance of these types of therapies as a safe and effective strategy in clinical practice? There’s been some interesting early survey data in terms of, if you had an option for a permanent therapy that you took once, would you like to do it? And the answer was largely yes. Now if you were to say, well, we’re going to edit your genes, then the answer kind of changes. Some of the confusion comes from germline editing. So I think there’s a lot of education that has to be done in this area.

– As a person with a rare disease, I’ve been spending some time thinking about what I would do if I had this option, what the considerations are, and the ableism that all of these possibilities bring up. Contemplating and discussing with impacted persons is so critical. How would your initiative address that? It’s critically important to have lived experience involved, especially when you’re thinking about something that’s potentially durable over a lifetime. You want to make sure that you’re deriving meaningful benefit for patients that’s substantially beneficial rather than incremental. That’s also important for engaging the population.

 

 

Tags

#pctGR, @Collaboratory1

Grand Rounds July 26, 2024: Interventions for Optimization of Guideline-Directed Medical Therapy for Heart Failure (Gregg C. Fonarow, MD, FACC, FAHA, FHFSA)

Posted on by Sophia Ramirez

Speaker

Gregg C. Fonarow, MD, FACC, FAHA, FHFSA
Eliot Corday Professor of Cardiovascular Medicine and Science
Director, Ahmanson-UCLA Cardiomyopathy Center
Codirector, UCLA Preventative Cardiology Program
UCLA Division of Cardiology

Keywords

Heart failure; HFrEF; guideline-directed medical therapy; implementation; systematic review

Key Points

  • Despite the availability of effective therapies for heart failure (HF), a large number of eligible patients are not receiving one or more evidence-based, guideline-recommended therapies. This implementation gap has stymied improvements in morbidity and mortality.
  • The benefits of evidence-based heart failure medications are cumulative. If all therapies are used, there is a relative risk reduction of 74.0% and an absolute risk reduction of 25.9%. Simultaneous or rapid initiation of therapies reduces HF hospitalizations, rehospitalizations, and mortality.
  • Reasons for underutilization of Guideline-Directed Medical Therapy (GDMT) for heart failure include a lack of systems to reliably implement therapies; gaps in knowledge and awareness of guidelines; therapeutic inertia; and insufficient urgency.
  • Dr. Fonarow’s team conducted a systematic review of interventions for the optimization of GDMT, including 28 randomized clinical trials with an aggregate sample of over 19,000 patients. They found that the initiatives that used interdisciplinary teams, largely comprised of nurses and pharmacists, most consistently led to improvements in GDMT. Clinician education, electronic health record initiatives, and patient interventions, on the other hand, resulted in no or modest improvements in GDMT.
  • Implementation of GDMT needs to improve in all clinical settings. Programs that successfully implement GDMT often have access to current and accurate data on treatment outcomes; administrative and clinician support; use care maps and pathways; and use data to provide feedback. There is also a need for further implementation science innovation and testing.

 

Discussion Themes

– Do you still see lack of education as a major problem that needs to be approached differently, or should we really be focusing on something else? I do think it need to be further studied. I think our educational efforts around the clinician need to center on the really practical aspects, on what the common pitfalls are in trying to apply and uptitrate these medications. Interdisciplinary teams and cross-discipline education are critically important. [The study results] don’t mean that education shouldn’t be part of further interventions, but I think we need to look at what’s truly going to see a meaningful increase in medication titration and persistence of GDMT.

– The polypill could address some of the clinical inertia issues and patient resistance. Patients hate it when you keep adding on four different medications and you’re increasing the doses. Is it being studied in this context? There is work going on, editorials have been written about it, the concept has been laid out. There are small pilot studies that have been done. There are challenges; is titration necessary? You’ll need a series of polypills to uptitrate, or you’ll get the therapy started and then you’ll bring them together as a polypill. One thing I do want to highlight, though, is that starting medications simultaneously defeats that patient question of, “Why are you adding this extra medication? Are the prior medications not working?” It’s a different mindset upfront: “Here are these four medications you’re going to be on.”

 

 

Tags

#pctGR, @Collaboratory1

Grand Rounds July 12, 2024: Causal Estimands: Should We Ask Different Causal Questions in Randomized Trials and in the Observational Studies That Emulate Them? (Miguel Hernán, MD)

Posted on by Sophia Ramirez

Speaker

Miguel Hernán, MD
Professor of Biostatistics and Epidemiology
CAUSALab
Harvard T.H. Chan School of Public Health

Keywords

Statistical Analysis; Causal Estimands; Research Questions; Randomized Clinical Trials; Observational Studies; Intention-to-Treat Effect; Per-protocol Effect

Key Points

  • Researchers often look at the intention-to-treat (ITT) effect – i.e., the effect of being assigned to an intervention – for randomized trials, and at the per-protocol effect – i.e., the effect of receiving an intervention – for observational studies.
  • Why are we giving estimates for 2 different effects? If a causal question is important enough to be asked in an observational study, Dr. Hernán posed, then we should also ask it in a randomized trial, and vice versa.
  • He walked through some of the field’s justifications for asking different causal questions. If adherence patterns vary, the ITT effect may differ across trials that study the same treatment. Arguments in favor of using the ITT effect often include that it preserves the null; it’s conservative; and it measures effectiveness in the real world.
  • But null preservation is not guaranteed; most pragmatic trials are not blinded, for instance, so assignment may make patients and doctors alter their behavior. Through hypotheticals and case examples, Dr. Hernán made the case that ITT effects are not necessarily a measure of effectiveness, nor are they of primary interest for doctors and patients.
  • The per-protocol effect represents another causal question that randomized trials could answer. The issue with per-protocol analysis is that valid estimate of the effect requires adjustment for confounding. Historically, trialists have been suspicious of these analyses as potentially biased.
  • In the past few decades, however, statistical methods have been developed that allow researchers to adjust for confounders at baseline and after baseline. Provided a few criteria are met, there is hope for the estimation of per-protocol effects as a way to contrast generic treatment strategies.
  • For a long time, regulators were reluctant to consider causal effects other than the ITT effect. Under increasing pressure from industry, a group of methodologists from pharma and regulatory agencies worked together to generate the “estimand framework.” This manifested as an addendum to the ICH E9 guidelines.
  • The addendum was a great step forward for more rigorous causal influence, Dr. Hernán noted. But it has room for improvement, specifically in terms of choice (of standard terminology and of estimands) and a lack of emphasis on both treatment strategies and trial design.
  • Researchers do not compare treatments but treatment strategies. Therefore, the causal contrast needs to specify the treatment strategies of interest; typically, these will be the treatment strategies specified in the protocol. Observational studies for causal inference already try to estimate well-defined per-protocol effects when emulating target trials. It’s time to turn to randomized trials.

Discussion Themes

The intention-to-treat effect is not always the primary interest. When there is a safety outcome or in a non-inferiority trial, the per-protocol effect is necessarily the estimand of interest.

When we say the ITT effect is biased towards the null, we mean it’s biased for the per-protocol effect. I.e., “I’m interested in the per-protocol effect because it’s going to be harder to find differences between treatments using ITT analysis.”

One issue with the estimand framework is that it doesn’t give the per-protocol effect much consideration; it is not named and is part of a set of estimands, all of which are given equal importance. Dr. Hernán argued that the per-protocol effect should be made more explicit, as it’s the main question that we can ask of a trial outside of the ITT effect.

The person who designs the trial is responsible for pre-specifying the procedure when a participant switches treatments.