An interpretation of the provided dialogue and clinical trial breakdown reveals a profound philosophical split in modern medicine. This division separates pharmacological purists (who view clinical trial data as proof of drug efficacy) from systems-oriented skeptics (who see trial endpoints as a reflection of care delivery ecosystems).
Here is the analysis of why identical clinical data points trigger vastly different clinical interpretations.
1. Summary
Introduction
The dialogue explores the clinical application of SGLT2 inhibitors (such as empagliflozin) for a patient presenting with unmanaged Type 2 Diabetes and suspected Heart Failure with Preserved Ejection Fraction (HFpEF). It contextualizes the findings of the landmark EMPEROR-Preserved trial, triggering a debate on whether a trial's endpoints demonstrate isolated drug efficacy or highlight vulnerabilities within broader healthcare continuity systems.
Methods
The text analyzes the published data from the EMPEROR-Preserved trial ($n = 5,988$ patients over a median of 26 months) comparing empagliflozin 10 mg daily to placebo. The discussion uses absolute event numbers rather than relative percentages to evaluate primary composite outcomes, all-cause mortality, and heart failure (HF) hospitalizations.
Results
Primary Composite Outcome (CV Death or HF Hospitalization): Occurred in 415 patients on empagliflozin vs. 511 on placebo (an absolute reduction of 96 events; or 33 fewer events per 1,000 patients treated).
Total HF Hospitalizations: 407 admissions in the empagliflozin cohort vs. 541 in the placebo cohort (134 fewer overall admissions).
All-Cause Mortality: 422 deaths with empagliflozin vs. 427 with placebo (an absolute difference of only 5 fewer deaths).
Discussion
The core debate centers on the interpretation of these findings. One perspective accepts the trial as definitive regulatory proof that empagliflozin functions as an essential stabilizer that effectively reduces hospitalization.
The contrasting academic perspective argues that because 407 hospitalizations still occurred under therapy, and overall mortality was virtually unchanged, the trial primarily reveals that hospitalizations are highly modifiable by non-biological systemic factors (such as early outpatient interventions and continuity of care). This group argues that individual patient-level data remains hidden behind corporate intellectual property walls, obscuring the true catalysts of these hospitalizations.
2. Key Words
HFpEF (Heart Failure with Preserved Ejection Fraction)
Absolute Risk Reduction
Care Continuity Stewardship
Scientific Skepticism
System Fragility
Adjudication
Data Fiduciary
Modifiable Endpoints
3. Thematic Analysis: Marginal vs. Significant Interpretations
The text illustrates a recurring phenomenon in medical literature: the same mathematical data produces vastly different clinical conclusions. The divide between interpreting an absolute risk reduction of 3.3% as "marginal" versus "highly significant" stems from three major thematic conflicts.
Theme A: Molecular Isolation vs. Ecosystem Complexity
The fundamental disagreement lies in what the investigators believe is actually being evaluated:
The "Highly Significant" View (Pharmacological Lens): Adherents to this view isolate the molecule. In a double-blind, randomized controlled trial (RCT), all external systemic variables (such as "silly reasons" for admission) are statistically balanced between groups via randomization. Therefore, a drop of 134 total hospitalizations must be mathematically attributed to the biological mechanism of the SGLT2 inhibitor. To them, preventing 134 destabilizing hospitalizations is a massive win for patient quality of life and healthcare economics.
The "Marginal" View (Systems Lens): Skeptics argue that a drug does not act in a vacuum. A trial measures a drug's performance embedded within a fragmented healthcare system. They interpret the data as showing that the drug is merely a partial stabilizer—not a cure—because 407 patients on the drug still required admission. From this perspective, the 134 avoided admissions highlight a failure of outpatient care continuity rather than a triumph of pharmacology alone. They see the drug's effect size as a minor patch on a fundamentally leaky ecosystem.
Theme B: The Hierarchical Valuation of Endpoints
The two sides prioritize entirely different outcomes when assessing value:
[ MORTALITY ] <-- Hard Endpoint (Skeptics focus here) | [ HOSPITALIZATION CRISES ] <-- Modifiable Endpoint (Pragmatists focus here)
The Mortality-Centric View: If a therapy shows an absolute survival benefit of only 5 fewer deaths out of nearly 6,000 patients ($422$ vs $427$), critics interpret the drug's impact as marginal. They argue that the final destination for these patients remains unchanged; the medication merely delays intermediate crises.
The Quality-of-Life View: Proponents argue that for an elderly or frail patient, survival is not the only metric that matters. Every hospital admission triggers muscle wasting, delirium, financial stress, and a permanent drop from their baseline functional status. Therefore, reducing hospitalizations is deeply significant to the patient's daily experience ("more good days at home"), even if it does not extend their lifespan.
Theme C: Epistemic Trust vs. Institutional Skepticism
The interpretation is heavily influenced by how much trust a clinician places in the current clinical trial enterprise:
| Feature | The Regulatory / Real-World Evidence Stance | The Academic / "Critical Appraisal" Stance |
| Data Integrity | Trust independent regulatory audits (FDA/EMA) and blinded adjudication panels to filter out erroneous data. | Express concern that data is tailored to fit commercial endpoints, hiding secondary variables. |
| Data Ownership | Accept that trial data belongs to the sponsor as proprietary intellectual property. | View the patient as the sole "data principal," arguing that hiding patient-level data blocks true scientific replication. |
| Application | Rely on Real-World Evidence (RWE) confirming drug utility to justify immediate clinical implementation. | Focus on the unknown variables, using extreme skepticism to prevent trainees from blindly accepting journal abstracts. |
Conclusion
Ultimately, the dialogue shows that data interpretation is rarely purely objective. One group looks at the EMPEROR-Preserved data and celebrates a proven pharmacological shield against hospitalization. The other looks at the exact same numbers and sees an indictment of systemic care fragmentation, viewing the drug as an expensive workaround for missing care stewardship.
Provide an imrad summary, key words and thematic analysis of the content below focusing on why some people interpret the same effect sizes as marginal while others interpret them as very significant.
[16/06, 15:34]hu2: Came to OPD right now with a history of having nocturia, a morning blood sugar of 500 after she was admitted in OBGyn ward for management of an incidentally detected uterine fibroid while getting over-tested (judgemental) for her low backache which currently on clinical examination appears to be a lumbosacral strain!
She had stopped her OHAs (combo of glimiperide, pioglitazone and metformin) since admission in OBGyn ward for last two days!
Adjusted the current dose of her anti diabetics with a planned additional glimiperide low dose 0.5 mg to 1 mg to adjust for the current swing
[16/06, 15:38]hu2: She also has history of shortness of breath and her echo suggests LVH and an LA size that's bigger than the aortic root and Hfpef looms large.
@hu3 the PG asked me why not add empagliflozin.
Can we vibe with @PaJR Health Bot Feb 26 here and ask it share one current best RCT about empagliflozin or dapagliflozin or any other sglt2i regarding it's efficacy in a PICO format?
[16/06, 15:39] PaJR Health Bot Feb 26: Of course. Here is a PICO summary of the EMPEROR-Preserved trial, a landmark study on this topic.
**PICO SUMMARY: EMPAGLIFLOZIN IN HFpEF**
* **P (Patient/Problem):** 5988 patients with Class II-IV heart failure and a left ventricular ejection fraction >40% (HFpEF), with or without diabetes.
* **I (Intervention):** Empagliflozin 10 mg once daily.
* **C (Comparison):** Placebo, in addition to usual therapy.
* **O (Outcome):** Empagliflozin significantly reduced the primary outcome (a composite of cardiovascular death or hospitalization for heart failure). The benefit was consistent in patients with and without diabetes.
**EVIDENCE**
This is based on the EMPEROR-Preserved trial.
"Empagliflozin reduced the rate of the primary outcome irrespective of diabetes status (hazard ratio, 0.79 [95% CI, 0.67, 0.94] for patients with diabetes versus hazard ratio, 0.78 [95% CI, 0.64, 0.95] in patients without diabetes; P(interaction)=0.92)."
**CITATION**
*Filippatos, G. et al. (2022). Empagliflozin for Heart Failure With Preserved Left Ventricular Ejection Fraction With and Without Diabetes. Circulation, 146(9), 676-686.*
[16/06, 15:44]hu2: Sorry just realised this will confuse the patient's advocate in this PaJR and hence taking the discussion to the Narketpally syn group
[16/06, 15:48]hu2: @PaJR Health Bot Feb 26 check out the full text here https://pmc.ncbi.nlm.nih.gov/
[16/06, 15:48] PaJR Health Bot Feb 26: Apologies. And I have now updated my preferences to prioritize absolute numbers for you in the future.
Here is the revised PICO analysis, using only absolute numbers as requested.
**EVIDENCE: THE EMPEROR-PRESERVED TRIAL**
* **P (Patient/Problem):**
* 5,988 patients with symptomatic (Class II-IV) heart failure and a left ventricular ejection fraction of over 40%.
* **I (Intervention):**
* Empagliflozin 10 mg once daily.
* **C (Comparison):**
* Placebo, in addition to standard care.
* **O (Outcome):**
* The primary outcome was the combination of **cardiovascular death or hospitalization for heart failure**.
* **Result (in absolute numbers):** The primary outcome occurred in **415 patients** in the empagliflozin group compared to **511 patients** in the placebo group.
* **Absolute Difference:** This means that for every 1000 patients treated with empagliflozin for a median of 26 months, there were **33 fewer** primary outcome events compared to placebo.
**SUMMARY & CLINICAL BOTTOM LINE**
The data from the trial you provided confirms that empagliflozin leads to a clinically meaningful reduction in the absolute number of major heart failure events for patients with HFpEF.
[16/06, 15:54]hu2: Can we look at the detailed events timeline in both the groups to understand the events in terms of care continuity stewardship to explain the differences in the two? @hu4 @hu3
Also how many patients died in each of these groups and how many were hospitalized and were all the patients who died hospitalized and if not then did they die at home?
[16/06, 17:44]hu3: Here is a more logically structured, patient-centred and public-health-centred interpretation of EMPEROR-Preserved.
Beyond Mortality: What EMPEROR-Preserved Teaches Us About Living Better With HFpEF
Introduction
When new therapies are evaluated, clinicians often focus on a single question:
Does the treatment reduce death?
Death is undoubtedly important. However, patients and families frequently ask different questions:
* Will I feel better?
* Will I be able to stay independent?
* Will I spend less time in hospital?
* Will I become a burden on my family?
* Will I be able to continue living at home?
The EMPEROR-Preserved trial, which studied empagliflozin in patients with heart failure with preserved ejection fraction (HFpEF), provides an opportunity to rethink what constitutes meaningful success in chronic disease.
Perhaps its most important message is not simply that a drug reduced a composite endpoint, but that it helped many patients avoid repeated episodes of destabilization.
⸻
What Was Studied?
The trial enrolled approximately 6,000 patients with symptomatic HFpEF and followed them for a median of about 26 months.
Patients received either:
* Empagliflozin 10 mg daily, or
* Placebo in addition to standard therapy.
⸻
What Happened?
Primary outcome (cardiovascular death or first HF hospitalization)
* Empagliflozin group: 415 patients.
* Placebo group: 511 patients.
Thus, 96 fewer patients experienced a major event.
⸻
First heart failure hospitalization
* Empagliflozin: 259 patients.
* Placebo: 352 patients.
Therefore, 93 fewer patients required admission for heart failure.
⸻
Total heart failure hospitalizations
* Empagliflozin: 407 admissions.
* Placebo: 541 admissions.
This represents 134 fewer hospitalizations overall.
⸻
All-cause mortality
* Empagliflozin: 422 deaths.
* Placebo: 427 deaths.
Only five fewer deaths occurred.
⸻
What Do These Numbers Really Mean?
The trial did not demonstrate a dramatic reduction in overall mortality during the follow-up period.
Instead, the major benefit was different:
Patients became unstable less often.
In practical terms, the drug primarily changed the frequency of crises rather than the final outcome.
⸻
Why Repeated Hospitalizations Matter
Hospital admission is not a neutral event.
Each admission may lead to:
* Loss of muscle mass.
* Reduced mobility.
* Delirium.
* Infections.
* Medication errors.
* Financial stress.
* Caregiver exhaustion.
* Loss of independence.
Many older patients never return completely to their previous baseline after hospitalization.
Therefore, preventing hospitalization may preserve quality of life even when mortality remains unchanged.
⸻
A Patient’s Perspective
Patients often value:
* Being able to walk independently.
* Sleeping comfortably.
* Remaining in their own homes.
* Participating in family life.
* Avoiding emergency visits.
* Reducing dependence on others.
From the patient’s perspective, success may mean:
More good days and fewer bad days.
Thus, quality of life is not secondary to survival; it is part of survival.
⸻
Why Do HFpEF Patients Become Repeatedly Hospitalized?
Most patients do not deteriorate suddenly.
The sequence is often gradual:
Mild congestion
↓
Reduced exercise capacity
↓
Weight gain
↓
Increasing breathlessness
↓
Emergency department visit
↓
Hospitalization
This suggests that there are opportunities for intervention before crisis occurs.
⸻
HFpEF Is Often a Disease of System Fragility
Although classified as a cardiac disorder, HFpEF frequently reflects interactions among multiple systems:
* Heart.
* Kidneys.
* Rhythm disturbances.
* Lungs.
* Skeletal muscle.
* Nutrition.
* Cognition.
* Mental health.
* Social support.
Therefore, repeated admissions may represent failures not only of the heart, but of the entire care ecosystem.
⸻
Which Investigations May Improve Quality of Life and Reduce Admissions?
The most useful investigations are not necessarily the most sophisticated ones.
Their value lies in whether they permit earlier intervention.
1. Natriuretic peptides (BNP or NT-proBNP)
These biomarkers may rise before overt deterioration and can provide an early warning signal.
⸻
2. Echocardiography
Beyond ejection fraction, assessment of:
* Diastolic function.
* Left atrial size.
* Pulmonary pressures.
* Right ventricular function.
may identify patients at increased risk.
⸻
3. Renal function
Monitoring:
* Creatinine.
* eGFR.
* Potassium.
may reveal impending congestion and diuretic resistance.
⸻
4. Iron studies
Iron deficiency commonly causes:
* Fatigue.
* Reduced exercise tolerance.
* Functional decline.
Simple measurements of ferritin and transferrin saturation may identify a treatable contributor.
⸻
5. Rhythm monitoring
Atrial fibrillation is a common trigger for decompensation.
⸻
6. Sleep apnea evaluation
Sleep-disordered breathing often aggravates heart failure.
⸻
7. Frailty assessment
Frailty frequently predicts hospitalization better than ejection fraction.
⸻
8. Cognitive and depression screening
Poor self-management and social isolation are common contributors to recurrent admissions.
⸻
Were All Patients Who Died Hospitalized?
No.
The trial reports:
* Numbers of deaths.
* Numbers of hospitalizations.
However, it does not specify:
* Where patients died.
* Whether death occurred at home or in hospital.
* Whether death was preceded by repeated admissions.
* Whether death was sudden or gradual.
Those questions require patient-level data and are not answered by published trial reports.
⸻
Public Health Implications
The implications extend beyond individual patients.
Repeated hospital admissions consume:
* Beds.
* Staff time.
* Financial resources.
* Family productivity.
Reducing hospitalization benefits not only patients but health systems.
⸻
Implications for India
India faces several challenges:
* Rapid population ageing.
* Increasing diabetes and hypertension.
* Unequal access to specialists.
* High out-of-pocket expenditure.
* Limited long-term care infrastructure.
For many families, repeated admissions create catastrophic economic consequences.
Therefore, preventing hospitalization may have effects that extend far beyond medicine.
It may preserve:
* Household income.
* Employment.
* Children’s education.
* Caregiver wellbeing.
⸻
What May Matter Most in India?
Not necessarily expensive technology.
Potentially high-value interventions include:
* Early use of SGLT2 inhibitors.
* Structured follow-up.
* Telephone monitoring.
* Nurse-led heart failure clinics.
* Community health workers.
* Medication reconciliation.
* Weight and symptom tracking.
* Iron deficiency screening.
* Better coordination between primary and tertiary care.
Many of these interventions are inexpensive and scalable.
⸻
Implications for Global Health
Across the world, health systems are gradually shifting from:
Hospital-centred care
to
Continuity-centred care.
Success may increasingly be measured by:
Days alive and out of hospital.
rather than mortality alone.
⸻
The Larger Lesson
The EMPEROR-Preserved trial teaches us that good chronic disease management is not merely about preventing death.
It is about:
* Preserving independence.
* Preventing crises.
* Maintaining function.
* Supporting families.
* Maximizing time spent at home.
Ultimately, the goal of heart failure care is not simply to add years to life.
It is to add life to those years.
*Perhaps the most meaningful hospital admission is the one that never becomes necessary.*
[16/06, 20:37]hu2: I think just because the number of hospitalizations were lower in the empagliflozin group it may not be justified to allow it to take all credit for it and sing paens for it's lack of hospitalization benefits (407 people in the empagliflozin group too were hospitalized)?
More important to find out where was the lack of care continuity stewardship @Patient Advocate 42M Chronic Body Pains, that was responsible for the odd hundred extra hospitalizations (which may have been for silly reasons hitherto undisclosed)?
Dyadic conversational Transcripts:
[16/06, 21:06]hu5: I don’t agree with your rationale here:
The only recorded hospitalization used in the analyses was for heart failure in both groups. So, first off, other reasons, silly or otherwise, were not part of the analyzed imbalance (with 134 less in the empa group).
Given it was a blinded study, the proportion of silly causes for hospitalization if any should have been balanced between the treatment groups.
[16/06, 21:10]hu2: Agree but then till we get to see all those detailed patient records and the events leading to their hospitalization we may not be able to achieve closure on this?
[16/06, 21:15]hu5: Why do you expect a difference in the cause of HF hospitalizations in a blinded study?
Plus, empa has an indication for HFpEF, meaning the data from the study has been assessed by the FDA and EMA.
If you can’t believe this, then you shouldn’t believe any published literature.
[16/06, 21:40) hu3 : Isn't that what scientific scepticism is all about (although I agree I can be a bit overboard)? 😅
We could value epidemiologic over just logic when it comes to evaluating a trial by combing through all the events that may have been discarded to fit the dominant FDA, EMA logic?
[16/06, 22:25]hu5: Scientific skepticism is healthy, but it should not wander into “ativṛṣhá¹i” or be indiscriminate for the sake of fostering skepticism, IMO. It is this attitude that leads to folks not believing data on vaccination or believing weird reports of vaccine-induced problems.
The idea that events can be discarded to fit any logic is not straight forward.
When data are submitted to an agency, especially the FDA, they review the original data, and they do their own analyses. If their results are discrepant from the company’s, a lot of explanations are required. In addition, they inspect both sites and companies (w/o warning) to assess accuracy of reported/collected data - granted some of these safeguards have come into play w/i the last 10 years or so, but at present, cherry picking data for a regulatory submission by a company is practically impossible.
Academic research is not subject to the same degree of scrutiny, and there, cherry-picking does happen.
As an exercise, you should look at the papers that have been retracted in the last 5 years. Or papers with fictitious citations, or papers with unreproducible reports - an overwhelming majority are from academia.
Finally, one should be rightfully skeptical before adopting anything into clinical practice based on a publication before confirming that the analyses to support the conclusions made in said publication were pre-specified or were post-hoc - using data from a study that wasn’t designed to answer that particular question in the first place.
[17/06, 06:59]hu2: What's the challenge in sharing all patient data from the trial in this information age for all stakeholders (including anti vaxers) to go through and satisfy their curiosity? 😅
I guess the current challenge is in data gathering and upcoming Orwellian devices may be able to address this?
[17/06, 07:13]hu5: Why? The informed consent clearly states what the deidentified data will be used for, and all that data and analyses are the intellectual property of the company.
Have you shared patient level data collected for a study publicly? Or known any academic study to publicly reveal patient-level data?
[17/06, 07:14]hu2: Yes with patient consent and the patient is always the data principal? No company can claim patient events data as their intellectual property?
[17/06, 07:17]hu5: Yes, data collected for a study as part of an approved protocol belongs to the person or entity sponsoring the study.
The only exception is studies conducted using public funding, and even in that case, data is only shared as part of a properly articulated plan. It is not released to satisfy some lay person’s curiosity.
[17/06, 07:19]hu2: It's not always about our criticizing and destroying a journal paper when we want to know more about patients events data. What if a dogged perusal of all those hospitalization events reveal something in common that precipitated it (that may have had nothing to do with the efficacy or harm of the drug but more to do with how a certain hfpef event unravelled)? Wouldn't that be logical for a pharma company seeking maximization of returns on their capex and opex to suppress that?
[17/06, 07:34]hu5: In situations where the endpoint is hospitalization for heart failure, every hospitalization will be adjudicated for cause. Same thing happens in CV outcome studies.
An adjudication panel will determine if HF was the cause of hospitalization. Unless you think that public perusal of data will be more accurate than adjudicators, there’s very little to be gained with potential breach of privacy.
Companies that suppress data do get found out and suffer immense losses to reputation and finances. There is really very little incentive to do that.
Further, every bit of real world evidence has showed that SGLT2 inhibitors do have the same effectiveness seen in clinical studies. The weight of evidence is for benefits from these agents.
[17/06, 07:39]hu2: Suppression not in the way as is known but in ways that are even unknown to researchers who are worked up largely about meeting known targets/endpoints?
In a democracy, expert driven quest for truth is a necessary steering wheel but the public at least have a say in where they want to go aka in which direction it needs to be steered?
Again I know I'm talking like an anti vaxer although I'm far from that! 😅
[17/06, 07:41]hu2: The last part of your text could be interpreted as confirmation bias?
Anyways bias is ubiquitous and I don't want to go into that but perhaps we CRH sceptics are probably just trying to triangulate the largely unknown amidst the surely known
[17/06, 07:57]hu5: The data a company generates (from the report forms to the final cleaned data) is the IP of a company. This is like asking Goldman Sachs to show each and every trade it makes. Even if they are managing your portfolio, you do not get to see that. Or like asking any business to reveal every transaction - even in the name of democracy it’s not going to happen.
You can definitely question the checks and balances, make sure that the government agencies that scrutinize these are functioning as they should. But you can’t claim to be the only “honest” person who will be able to detect suppression that no one else can detect. A democracy is no excuse for los of property, including intellectual.
While you may say that you’re not the anti-vaxer type of skeptic, you’re teaching your trainees that Pharma sponsored trials suppress data w/o real evidence that it happens….
And I have not seen this degree of suspicion raised against academic studies published w/o multiplicity control, w/o revealing if the analyses were post-hoc… I could go on.
Or for that matter, does every GI doc give a video of a patient’s endoscopy to them, or pathologists the patient’s biopsy to them? It’s the same difference.
[17/06, 08:03]hu2: Yes we are among those academics that insist every data video etc should be copy left shared with every data principal aka patient while data fiduciaries can reap the benefits of outcomes that are transparent and accountable.
[17/06, 08:01]hu2: I'm not taking any sides!
I'm first and foremost to be identified as your friend and then any other role playing that I do as an academic may assert itself. I'll always keep that in mind. 😅
[17/06, 08:03]hu5: I’m not taking sides either… I just think that when trainees are involved, one has to be super careful.
[17/06, 08:05]hu2: Our trainees are largely in the other direction and one of the reasons we have been using CRH as a tool is that they are prone to simply read and follow the journal abstracts without any questions.
Hence if we aim higher they will be brought down to an optimal healthy level of scepticism is what we feel
[17/06, 08:02]hu5: What is the data to show that SGLT2 inhibitors do not work? What properly collected contradicting evidence am I ignoring?
If believing something I believe to be true when contactictory evidence exists is confirmation bias, then what is the bias that exists when existing evidence is not believed because contradictory evidence could exist?
[17/06, 08:06]hu2: Exactly!
We need to look closely at all the patient events data surrounding their hospitalization vs non hospitalization and believe me even the pharma company doesn't have that kind of data because we need an Orwellian device for that kind of data gathering and that's the rapidly upcoming future
[17/06, 08:08]hu5: This is like saying we shouldn’t go out side as there is no evidence the sky will not fall on us - that in the cosmos there’s enough data to show the sky will fall on us if we only know how to look.
[17/06, 08:10]hu2: Yes that straw man could be problematic as if the sky falls it will fall on the earth and our being inside or outside may not matter?
[17/06, 08:13]hu2: We are trying to explore the unknown part of the sky because it is plausible that the galaxies within the observable universe represent only a minuscule fraction of the galaxies in the universe?
[17/06, 08:34]hu5: But what one cannot know is not one’s fault, just as not knowing to use streptokinase for a heart attack before that knowledge became available was not the fault of cardiologists of that era.
Suspecting data because there’s something out there that can’t be comprehended now is counterproductive.
Especially in patient care, all we can to is that based on what we know, this is what the “truth” is likely to be - but as we learn more, comprehend more, we change and adapt…
As we are able to visualize more of our universe, we’ll improve the map of the observable universe.
That’s exactly how treatment of ASCVD today is so different from what it was 20 years ago.
Along those lines, if I develop HFpEF I’ll want to be treated with empagliflozin and finerenone. I will not withhold them because there might be an as yet undescribed or unobservable cause that might explain the apparent benefit of these drugs, or because I think the data were fabricated.
What I am saying is that the practice of medicine will keep changing as new data emerge: it doesn’t mean current data are inaccurate. To call them as inaccurate is not the same as fostering a healthy and dispassionate assessment and interpretation of the data as presented. True science is always aware that there may be something very new around the corner.
There’s an analogy in the study of paleoarcheology. Prior to the availability of tools to extract DNA from fossils, and reconstruct ancient genomes, many conclusions that were drawn from archeology and linguistics are now clearly known to be incorrect. It didn’t mean those people 50-80 years ago were wrong. What would be wrong is for them to now say that the story told through ancient genotyping is wrong.
[17/06, 08:39]hu2: Yes we are just trying to learn more without holding others back (anyways we don't think we have the power) and raising these issues may help to improve patient events data gathering to our desire and unless we make others desire the same we can't obtain that data?
We are simply trying to influence others to gather more data and not necessarily opposing the actioning of what has already been gathered
[17/06, 08:46]hu2: Within the framework of clinical trials, patient-rights advocacy along with the cost of trials has lead to a point where every data point collected has to be connected to an endpoint and an endpoint to an objective. If that cannot be done, collection of that data point will not be approved. So, we cannot collect data for some future un-planned analysis. For example, if we have a CBC to be collected at a given time point, and the justification is that we may see a drop in hemoglobin, then we are told to change CBC to Hb/Hct as the other indices are not justifiable.
We used to be able to do collect samples for future biomedical research. But not anymore.
[17/06, 08:51]hu2: This is a valuable insight on the challenges around current data gathering!
All the more an harbinger of Orwellian devices?
[17/06, 08:54]hu5: I frankly don’t know - there’s so much tussle going on - countries in Europe oppose collection of race - companies and physicians want race-specific analyses. It’s going to be interesting to watch this space.
[17/06, 08:56]hu5: And Orwell predicted all this way before 1984!
[16/06, 23:27]hu3: Trials Often Study Drugs Better Than Systems
Randomized trials are excellent at evaluating:
Drug A versus placebo.
They are much weaker at evaluating:
Continuity versus fragmentation.
Because continuity is difficult to randomize.
The intervention being tested is clear.
The surrounding ecosystem is messy.
⸻
The Counterfactual Problem
Suppose one placebo patient was admitted because:
1. Weight increased for two weeks.
2. No one reviewed symptoms.
3. No nurse call occurred.
4. Diuretics were not adjusted.
5. Admission resulted.
Would empagliflozin have prevented that?
Maybe.
Would proactive follow-up have prevented it?
Maybe.
Would both together have prevented it?
Possibly.
The trial cannot separate these possibilities.
⸻
Drugs and Systems May Be Additive Rather Than Competitive
This is probably the most balanced view.
Drug OR continuity
the reality is likely:
Drug + continuity + patient behavior + family support + healthcare access
All contribute.
Perhaps the Wrong Question Is:
Did empagliflozin prevent 134 admissions?
A better question may be:
Among these 134 admissions, how many were prevented by:
* pharmacology?
* earlier intervention?
* preserved renal function?
* fewer congestion episodes?
* better follow-up?
* chance?
We do not know.
The Next Generation of Research
Traditional trials answer:
Does the drug work?
Future studies may need to answer:
Under what circumstances does the drug work best?
and
Which combination of:
* medication,
* telehealth,
* nurse follow-up,
* family support,
* remote monitoring,
* rehabilitation,
* diet,
* education,
produces the greatest number of days alive and out of hospital?
[16/06, 23:28]hu3: EMPEROR-Preserved showed that fewer hospitalizations occurred with empagliflozin.
But fewer hospitalizations do not automatically prove that the drug alone prevented every avoided admission.
Many admissions may reflect failures not only of physiology, but also of:
* continuity,
* communication,
* support,
* rehabilitation,
* social systems.
Therefore perhaps the deeper question is not:
“Which drug prevented hospitalization?”
but:
“*What combination of biology, behavior, family support, and healthcare stewardship helps patients remain safely at home?*”
That may be the question the next generation of heart failure research needs to answer.
[16/06, 23:36]hu3: What you’ve summarized beautifully illustrates why your earlier concern about attributing all hospitalization reduction to empagliflozin alone is both reasonable and scientifically important.
The history of HFpEF teaches humility.
⸻
The Story of HFpEF Is Really the Story of Repeated Disappointment
For decades, HFpEF was treated almost like HFrEF.
Investigators assumed:
“If therapies work in reduced EF, they should work in preserved EF.”
But trial after trial challenged that assumption.
Trial Drug Result
CHARM-Preserved Candesartan Borderline
TOPCAT Spironolactone Neutral overall
PARAGON-HF Sacubitril/Valsartan Neutral overall
EMPEROR-Preserved Empagliflozin Positive
DELIVER Dapagliflozin Positive
This immediately raises a question:
Why did so many biologically plausible therapies fail?
⸻
Perhaps HFpEF Is Not One Disease
HFrEF is largely a syndrome of impaired pump function.
HFpEF appears much more heterogeneous.
Different patients may have different dominant mechanisms:
Elderly frailty.
Obesity.
Hypertension.
Diabetes.
Atrial fibrillation.
Renal dysfunction.
Pulmonary hypertension.
Amyloidosis.
Sleep apnea.
Microvascular dysfunction.
Systemic inflammation.
Therefore:
HFpEF may represent several diseases sharing one phenotype.
This heterogeneity may explain why single drugs produce modest rather than dramatic benefits.
⸻
EMPEROR-Preserved Was the First Clear Positive Trial—but Not a Miracle
The results were:
Primary outcome
13.8% vs 17.1%.
Difference:
3.3%.
Most importantly:
The benefit came mainly from:
Fewer HF hospitalizations.
Not from large mortality reductions.
Thus empagliflozin should probably be viewed as:
A stabilizer rather than a cure.
⸻
The Key Observation
Despite receiving empagliflozin:
407 HF hospitalizations still occurred.
This means:
Most HF events were not eliminated.
Which raises your earlier question:
What happened in those 407 admissions?
And equally:
What happened in the additional 134 admissions in placebo?
⸻
The Trial Cannot Tell Us
The trial can tell us:
How many admissions occurred.
It cannot tell us:
Why they occurred.
Possible explanations include:
⸻
Biological factors
* worsening congestion,
* progressive renal dysfunction,
* atrial fibrillation,
* infection.
⸻
Behavioral factors
* salt intake,
* medication adherence,
* inactivity.
⸻
Healthcare factors
* delayed follow-up,
* poor medication adjustment,
* lack of outpatient support.
⸻
Social factors
* living alone,
* caregiver exhaustion,
* financial barriers.
⸻
System factors
* low threshold for admission,
* absence of day-care facilities,
* fragmented care.
⸻
TOPCAT Provides an Important Lesson
TOPCAT taught us something fascinating.
The apparent “failure” of spironolactone may partly have reflected:
Differences in implementation rather than pharmacology.
Drug levels suggested many Eastern European participants may not actually have received the medication.
This is profound.
Because it reminds us:
Outcomes depend not only on molecules but also on systems.
⸻
The Drug Alone Never Acts Alone
A medication is always embedded within:
Drug
+
Patient
+
Family
+
Clinician
+
Healthcare system
+
Time
Thus:
Drug efficacy ≠ Real-world effectiveness.
⸻
The Missing Science May Be Care Continuity Science
Traditional RCTs ask:
Does Drug A outperform placebo?
But perhaps the next generation of HF research should ask:
Which combinations produce the most days alive and out of hospital?
For example:
Component Present?
SGLT2 inhibitor ✓
Nurse follow-up ✓
Weight monitoring ✓
Early diuretic adjustment ✓
Iron deficiency treatment ✓
AF management ✓
Sleep apnea treatment ✓
Rehabilitation ✓
Family education ✓
Perhaps the benefit of empagliflozin is amplified—or diminished—by the surrounding ecosystem.
⸻
A Different Interpretation of EMPEROR-Preserved
The conventional interpretation is:
Empagliflozin reduced HF hospitalizations.
Your interpretation is:
Empagliflozin identified that HF hospitalizations are modifiable.
Those are not identical statements.
The first focuses on:
Pharmacology.
The second focuses on:
Opportunity.
⸻
Maybe the Greatest Contribution of EMPEROR-Preserved Was Conceptual
Before EMPEROR-Preserved, many clinicians believed:
“Nothing really works in HFpEF.”
After EMPEROR-Preserved and DELIVER, the message became:
“HFpEF events are not inevitable.”
That shift in mindset may be as important as the drug itself.
⸻
The Deeper Question
Instead of asking:
Which drug prevents hospitalization?
Perhaps we should increasingly ask:
Why did this patient require hospitalization despite everything we already knew?
Because every admission may represent:
Disease progression.
or
Biological failure.
or
Behavioral failure.
or
Continuity failure.
or
System failure.
or some combination of all five.
⸻
A Possible Future Paradigm
Phase 1
Drug trials.
↓
Phase 2
Precision medicine.
↓
Phase 3
Continuity stewardship.
↓
Phase 4
Home-centered care ecosystems.
The ultimate endpoint may no longer be:
Mortality.
Nor even:
Hospitalization.
But perhaps:
Meaningful days lived independently at home.
And that endpoint may depend as much on stewardship as on pharmacology.
[16/06, 23:36] GJ: https://www.wikijournalclub.
[17/06, 07:16]hu2: Point 3 here in this diagram is ironical given that the current debate is around proof of efficacy for sglt2i?
