A 76-year-old woman was admitted with septic shock due to perforated diverticulitis. She underwent emergent laparotomy and hemicolectomy, after which she was transferred to the ICU, where she was mechanically ventilated for nine days. She remains unable to take adequate nutrition enterally and has a central venous catheter in place for parenteral nutrition.

It is now hospital day 12. She has had intermittent fevers for most of the past five days. She has received piperacillin-tazobactam since admission. Blood, urine, and sputum cultures have been negative.

In the past hour, she has become hypotensive, and vasopressors are begun.

Read more

The U.S. FDA recently approved cefepime-zidebactam, the latest in a new generation of antibiotics targeting multidrug-resistant Gram-negative bacilli, a growing clinical problem and public health threat.

On its face, it looks like big pharma and regulatory business as usual. But there are changes afoot. The expansion of agents effective against MDR bacteria opens the door for new practice standards and health care operations that would permit much faster optimization of antimicrobials for hard-to-treat infections.

When will this happen? Don’t worry, you’ve got time—this is American health care.

The New ‘Bigger Guns’

The newest antimicrobial, marketed in the U.S. as Zaynich™, adds the β-lactamase inhibitor zidebactam to cefepime. Zidebactam also has bactericidal activity itself. This makes it broadly useful against multidrug-resistant gram-negative pathogens, including Pseudomonas and some carbapenem-resistant Enterobacterales.

Although approved for complicated urinary tract infections, it’s already being used off-label for other infections including:

• ESBL-producing Enterobacterales

• AmpC β-lactamase producers

• Many carbapenem-resistant Enterobacterales (depending on resistance mechanism)

• Difficult Pseudomonas aeruginosa isolates

Zaynich joins a trio of other powerful antibiotic combinations approved since 2023, each with its own niche:

These have further expanded the armory of (still relatively new) antibiotics targeting drug-resistant infections:

• Ceftolozane–tazobactam

• Ceftazidime–avibactam

• Meropenem–vaborbactam

• Imipenem–cilastatin–relebactam

• Cefiderocol

How to Use the New MDR-Active Antimicrobials

In the face of this massive expansion in complexity, the intensivist’s role remains clear. When a patient’s antimicrobial resistance report has a lot of Rs in it, red and/or bold font, or acronyms like MBL, KPC, AmpC, CTX-M, NDM, CRE, or OXA-48, call ID and ask for something extra-hyphenated.

But seriously

The good news is, it’s not necessary to memorize lists of organisms that each drug covers. Instead, laboratories and antimicrobial stewardship programs now think in terms of resistance mechanisms:

New Drugs, Old Practice Patterns

Prescribing these drugs is restricted to infectious diseases specialists. This reduces overuse and preserves the new antimicrobials’ effectiveness, but at the expense of delays in care.

Speaking of delays: hospitalists and intensivists check labs in the morning, while microbiology updates, including speciation and drug resistance, are at many centers updated silently in the electronic record in the afternoon—without notifying the treating clinicians that the patient is on an ineffective antibiotic.

Operationally, that has traditionally resulted in a clinical timeline that often looks something like this:

• Day 0, 9 pm: A patient is admitted to ICU from a skilled nursing facility or long-term acute care hospital, or a long-stay ICU patient develops new sepsis physiology. Pip-tazo is ordered; cultures are collected.

• Day 2, 6 pm: Gram-negative rods are noted to be growing in blood cultures. The patient is not improving. Pip-tazo is changed to meropenem; ID is consulted electronically, with a sign-out to call them in the morning.

• Day 3, 8 am: ID changes antibiotics to a newer-generation agent. At 1 pm, cultures and sensitivity reveal extensively resistant Enterobacterales species.

Initial empiric coverage was appropriate. An ID specialist was consulted. The new agent covers the infection. The care standard was met!

But after 48 hours without optimal antibiotic therapy, the patient’s sepsis has worsened significantly.

It Could Look Like This Instead

At many centers today, blood cultures incubate continuously in an automated instrument (e.g., BACTEC™, BacT/ALERT™, or VersaTREK™). When bacteria multiply enough to produce detectable CO₂ or other metabolic changes (in as little as eight to 24 hours for Gram-negative rods), the instrument flags blood cultures as positive.

Then, mass spectrometry (MALDI-TOF) can measure a protein “fingerprint” (primarily abundant ribosomal proteins) unique to each organism, identifying the species within 30 minutes.

Instead of waiting a day for resistance profile (the classic MICs on a Petri dish), PCR-based panels (BioFire’s BCID2™; Verigene™; ePlex™) can detect major resistance genes directly from the positive blood culture, within two hours after the culture turns positive.

If 12 hours after blood culture collection, these tests strongly suggest Klebsiella pneumonia expressing the KPC resistance gene, it’s highly likely to be carbapenem-resistant.

That’s the time to switch from piperacillin-tazobactam or meropenem to an advanced combination agent—not 24 hours and an ID consult later.

Even if the results come back at three in the morning.

Re-Engineering Resistant Infection Management

Although the algorithms would be complex, EMR-based decision support tools and human workflows could and should address this clinical problem/opportunity.

Identification of resistance genes in a patient with newly positive blood cultures isn’t a bland background update, especially if the patient displays sepsis physiology. It should require the same documented chain of communication and readbacks as any other critical lab.

EMR logic could display popups warning of the health threat, suggesting a menu of potential antimicrobials, with any clinician authorized to prescribe the appropriate agent while ordering an infectious diseases consult.

Cefepime-zidebactam’s broader use cases compared to the other more focused combination agents raise the question if it should be used empirically by intensivists instead of meropenem for very specific situations (e.g., high risk for carbapenem-resistant Enterobacterales), or for patients who deteriorate while on meropenem.

When otherwise serious people argue that antibiotics must be given within one hour of arrival to the ED, it’s hard to watch the sickest and most vulnerable patients in the hospital go days without receiving appropriate antimicrobials.

Only a few years ago, that was an unfortunate but understandable limitation of technology. But we have the tools to do better now.

Will we?

References

Genov P, Mladenov B, Slaitas D, et al. P-1209. Efficacy of β-lactam Enhancer Based Zidebactam-Cefepime Combination (WCK 5222) versus Meropenem in Adults with Complicated Urinary Tract Infection (cUTI) or Acute Pyelonephritis (AP) in a Global, Randomized, Double-blind, Phase 3 Trial. Open Forum Infectious Diseases2026;13(Supplement_1). Accessed July 29, 2026. https://academic.oup.com/ofid/article/13/Supplement_1/ofaf695.1402/8421223

Matrix-Assisted Laser Desorption/­Ionization Time of Flight Mass Spectrometry (MALDI-TOF MS) Analysis for the Identification of Pathogenic Microorganisms: A Review. Microorganisms. 2021. Chen XF, Hou X, Xiao M, et al

Performance of BioFire Blood Culture Identification 2 Panel (BCID2) for the Detection of Bloodstream Pathogens and Their Associated Resistance Markers: A Systematic Review and Meta-Analysis of Diagnostic Test Accuracy Studies. BMC Infectious Diseases. 2022. Peri AM, Ling W, Furuya-Kanamori L, Harris PNA, Paterson DL.

Multidrug-Resistant Gram-Negative Bacterial Infections. Lancet. 2025. Macesic N, Uhlemann AC, Peleg AY.

Usefulness of BioFire FilmArray BCID2 for Blood Culture Processing in Clinical Practice. Journal of Clinical Microbiology. 2021. Berinson B, Both A, Berneking L, et al.

Emerging Technologies for Rapid Phenotypic Antimicrobial Susceptibility Testing of Clinical Isolates of Bacteria. Journal of Clinical Microbiology. 2025. Rattin J, Boswell M, Rhoads D.

Staphylococcus aureus is a particularly virulent gram-positive bacterium that lives on the skin or mucous membranes of about one-third of people, causing no harm. When S. aureus gains access to the bloodstream, however, it can become life-threatening within a short period.

Worse, S. aureus is adapted to adhere to and grow on deep-seated tissues or prosthetics, producing biofilms, vegetations, or abscesses. These can be difficult to detect and require several weeks or months of antibiotics (and often device removal) to eradicate. Inadequate treatment courses can result in even more prolonged infections and poor outcomes.

Because of the difficulty in ruling out so-called metastatic foci of staph infection, infectious diseases specialist guidance has historically been very conservative, recommending prolonged treatment courses for any patient not considered at very low risk for persistent infection. (Patients with no risk factor or evidence of metastatic foci were generally prescribed 14 days of antibiotics.)

But providing four to six weeks of antibiotics, usually in the outpatient setting, creates logistical headaches for health systems and enormous inconvenience for patients. It’s always been an open secret that many of these patients don’t benefit from the long courses—and that a few will even be harmed—but there’s been no way to confidently identify or advise whose course can be shortened.

There’s reason to believe this situation may soon change.

Wait, Give That Back

In July 2026, a leading infectious diseases specialty society soft-launched or prematurely released a new guideline on staphylococcal bacteremia. The documents were withdrawn and the links redirected to dead-end pages. The guideline is now listed as “In Development,” and the manuscripts are unavailable on the relevant websites.

Search engines and web scrapers index newly posted content almost instantly these days, though, with the escaped content then rendered into various semi-accessible formats, so it’s possible to guess at what the contours of the new guidance will be.

We should presume the information was withdrawn for good reasons, so the following is speculative, informational only, and should not be used to guide clinical practice until it’s published in a peer-reviewed journal.

Downgrading the “risk = disease” mindset

By the strict letter of the 2011 guidelines for MRSA bacteremia (which have guided treatment of MSSA as well), any patient at risk for deep S. aureus infection is considered as if they have one. That has included anyone with:

Read more

A 67-year-old man with severe COPD is admitted to the ICU with hypercapnic respiratory failure, presumed to be from an acute exacerbation of COPD. He is intubated and mechanically ventilated with the following settings:

• Volume assist-control, tidal volume 450 mL (6.5 mL/kg predicted body weight)

• Respiratory rate 24/min

• FiO₂ 40%

• PEEP 8 cm H₂O

Peak inspiratory pressure is 42 cm H₂O, and plateau pressure is 22 cm H₂O.

Over the next 24 hours, the patient becomes increasingly tachypneic and appears to be “fighting” the ventilator despite adequate sedation. The ventilator frequently fails to trigger despite obvious inspiratory effort. Blood pressure remains stable, but heart rate has increased from 95 to 110.

An end-expiratory hold maneuver demonstrates a total PEEP of 17 cm H₂O. When you evaluate the patient, the respiratory therapist has already disconnected and reconnected the ventilator circuit after a brief period of manual bag ventilation.

Read more

“… And, pharmacy?”

For the overadrenalized, attentionally voracious intensivist, the enforced low-gear idle of interdisciplinary rounds may sometimes seem worse than Sisyphean in its repetitive monotony, a rock to be rolled up the mountain only to be found at the bottom again not tomorrow, but at the very next bed. The insipidity of the form, its evocations of apathetic classroom recitations, someone now with the rapid-fire monotone of an auctioneer, the next a low talker, each of their tics offering its own endearing evidence of a humanity resisting the homogenizing compulsory mode, with its enforced anti-rhetoric, its narrative flatness a constrictor snake asphyxiating each innocent wriggling datum as it struggles to escape the unfurling, featureless sonic tapestry entombing it, yearning instead to soar in the chaos of context, that jostling muddle where it can insist on its own truth and meaning, become real.

I had so much potential, the intensivist sighs invisibly. What am I doing here?

You really don’t know—you, who always had all the answers?

This one’s easy: saving lives.

Critical care researchers have spent decades searching for heroically transformative therapies in the ICU. N-of-1-omics. Stem cells. Brain-cooling. But in the ICU, as in many areas of medicine and surgery, the interventions with the largest impact have been the least glamorous: avoiding oversedation, liberating patients from ventilators, and removing unnecessary lines and tubes.

A new randomized trial in JAMA is the latest reminder that getting the basics of critical care consistently right matters more than any experimental therapy so far.

Researchers in Brazil launched a “tele-rehab” intervention at 20 public hospitals in random order (a stepped-wedge cluster-randomized trial, which is the right way to do this), ultimately enrolling 1,916 adults requiring invasive mechanical ventilation for acute hypoxemic respiratory failure in 2024 and 2025.

Hospitals sequentially crossed over from usual care to an integrated “telehealth rehabilitation program” spanning the full phase of care from the ICU to the hospital ward, to post-discharge follow-up for two months.

But It Wasn’t “Rehab”; It Was “Pre-hab”

The intervention basically amounted to increased diligence and accountability for consistently maintaining commonly accepted best practices in the ICU:

• Reducing sedation and optimizing analgesia

• Systematically performing spontaneous breathing trials

• Delirium prevention efforts

• Early mobilization and physical therapy

• Removal of unnecessary invasive devices

The extra encouragement/pressure to adhere to these practices was delivered by trained local staff on multidisciplinary rounds that would probably be quite familiar to U.S. clinicians (if they spoke Portuguese).

A remote intensivist reviewed adherence weekly for three weeks in a one-hour videoconferencing session and provided targeted guidance and encouragement.

On the medical ward and after discharge, the focus shifted to true rehabilitation assessment and service delivery, provided via smartphone videoconferencing.

During usual care phases (in a variable lead-in period at each facility), the intervention was not performed.

Menos Pessoas Morreram

Fewer people died during the intervention periods overall: 71.8% all-cause mortality at 90 days vs. 78.3% during usual care periods (with a 95% confidence interval of −14.7% to −0.6%; P = .03).

They also had far shorter durations of mechanical ventilation: 9.9 days vs 15.5 days, adjusted difference, −6.2 days; 95% CI, −8.5 to −3.9; P < .001.

They had more days alive and out of the hospital (a mean of 17 vs. 12).

The primary outcome was health-related quality of life at 90 days after discharge using the EQ-5D-3L utility score, and this was minimally improved in the intervention group. The difference was entirely derived from the survival benefit. That is, patients who survived did not experience an improvement in self-reported quality of life as a result of the intervention.

Bundles: Baffling Black Boxes

Which component of the complex intervention caused the mortality difference—or was it due to something else entirely? There’s no way to know with clarity. The intervention spanned the entire care course and is impossible to unpack.

But it seems darned unlikely that one in every 13 patients’ lives was saved just by extra attention to their nutrition and PT plan after ICU discharge.

And we know that six fewer days on a ventilator will do wonders for anyone’s general health and life expectancy.

We can cautiously conclude that increased adherence with the ICU bundle’s multiple evidence-based practices was at least partly responsible for the improvements in survival and shortened duration of mechanical ventilation.

But in one of the biggest limitations of this paper, the authors didn’t record or report bundle compliance, so we don’t even know if the intervention increased it.

If Not the Bundle, What?

For system-level interventions like this, cluster randomization has major advantages over individually randomizing patients. But stepped-wedge cluster randomization is vulnerable to secular changes in care (e.g., unmeasured general improvements) over the course of the trial. Because all the units “flip” from usual care into the intervention (and never vice versa), albeit at varying times, generalized improvements unrelated to the intervention can spuriously favor the intervention in the results. Standard statistical adjustments are made for this, but they aren’t impervious to (e.g.) entire units becoming more engaged (or fearful, or both) and delivering better care overall during the intervention.

Clinicians were not blinded—obviously, as they were the ones performing rounds and were expected to change their behavior, by design. Surely their psychology and interpersonal interactions also changed, along with the local culture at each ICU, etc. Humans with other humans: the ultimate confounders.

Mortality approached a bone-shivering 75% at 90 days. In trials of patients requiring mechanical ventilation in the U.S. and Europe, mortality has generally been about 30-40%.

When mortality is that high, of course, there are often more opportunities to reduce it.

Conclusions

This trial’s results can’t be directly extrapolated from Brazil to health systems in wealthy nations like the U.S.

Because of the limitations in reporting, we don’t even have proof that ICU care bundles saved lives, or even reduced time on the ventilator.

But something did. Someone did.

Whether it was greater adherence to best practices in critical care, or the collaborative effort itself improved teamwork and care quality over time, doesn’t matter in the end.

We already have evidence that these measures help (albeit not nearly as impressively as here).

They help almost in direct proportion to how boring they are, how tedious and repetitive. And to then have to talk about them over and over again to say they were done, or notice they weren’t, can seem even more tedious and repetitive.

Yet it turns out that’s what good care is. It turns out tedium can be an act of love.

You will likely never see the difference you make, and after everything you’ve gone through to get where you are, that can be deeply disappointing and may seem unfair.

But every nth time you roll that rock up the hill, someone lives who would have died. You’ll never know who, and neither will they.

As PulmCCM reader Luciano Pilla put it:

So let’s keep rolling the rock.

Reference

Rosa, R. G., et al (2026). Integrated Telehealth Rehabilitation and Quality of Life in Mechanically Ventilated Adults. JAMA. https://doi.org/10.1001/jama.2026.10617

A 58-year-old man with atrial fibrillation underwent an abdominal wall hernia repair three weeks ago. Two weeks ago, he presented to the E.D. with abdominal pain, nausea, and vomiting, and was admitted to the medical unit after a CT of the abdomen showed dilated loops of bowel consistent with a small bowel obstruction. His original surgeon was consulted and recommended conservative management with gastric decompression by NG tube to suction.

One week ago, he was admitted to the ICU for intermittent hypotension that responded to volume resuscitation and a brief infusion of vasopressors. He was given seven days of piperacillin-tazobactam. A CT without contrast at that time showed colonic thickening and persistent (improved?) dilation of the small bowel, which along with the stomach was fluid-filled. Blood cultures were negative.

Today, he has worsening hypotension and tachycardia, now requiring two vasopressors, afebrile and without new localizing symptoms. Abdominal exam is unchanged: significant distension with mild diffuse tenderness, no rebound. He denies any change in his symptoms, but has been receiving regular opioids for pain. Meropenem is started, and blood and urine cultures are obtained.

The surgeon’s note describes an unchanged abdominal exam and recommends a CT abdomen with oral contrast if the abdominal exam worsens. Creatinine was 1.5 mg/dL on admission and is now 3.1 mg/dL with low-normal urine output. The nephrologist’s note advises “avoid nephrotoxins, including contrast”.

Read more

Monitoring patients remotely after they’ve survived critical illness is supposed to keep them out of the hospital.

According to a new study out of UPMC in JAMA Network Open, for the elderly after sepsis it’s more like

Since 2019, the U.S. Centers for Medicare and Medicaid (CMS) has expanded a reimbursement infrastructure for remote care (tagged as “hospital at home” and “virtual care,” e.g.) through new billing codes for remote monitoring, video communication, care management, and home-based telehealth.

These diverse programs expanded during the Covid era. Though the pandemic’s proximate pressures driving telehealth decreased, CMS has continued to pin its hopes for efficiencies in care quality and cost on remote patient monitoring.

Sepsis patients, who have up to a 40% one-year mortality rate and high rates of readmission, would seem to be ideal candidates for post-discharge remote patient monitoring.

The enthusiasm may have run ahead of the evidence, however.

The So Ya Had Sepsis Didja? Trial

First, let’s be clear: that was not the real name of the trial. It wasn’t formally named.

That’s just how I (respectfully and fondly) imagine the kind, friendly Pittsburgh-based care teams opening the conversations with the 1,286 patients recently discharged from one of 19 UPMC hospitals after surviving an episode of sepsis. (Only a third had been admitted to ICUs.)

Almost 900 of them were randomized to the intervention, and the 529 who opted in were subsequently pinged with over 10,000 twice-weekly questionnaires delivered to a customized app on their smartphones. They reported their symptoms and could request a call from a nurse. (There was no physiologic telemetry like heart rate monitoring, pulse oximetry, etc.)

Intervention-arm patients were stratified into two intensities of questionnaire (infection symptoms only, or with heart/lung symptoms included) and also two levels of response teams: one had 4 staff nurses; the other included two nurse practitioners who were authorized to diagnose and treat medical conditions they recognized. All that splitting produced four midsized intervention groups.

About 400 patients received usual care, which did not include the questionnaires or ongoing contact with response teams.

Results

Overall, there was an impressive consistency in median days spent at home after discharge across all intervention groups: 90 days. Usual care? Also 90 days.

A relatively narrow range of readmission rates occurred among almost all the groups:

• Usual care: 37.8%

• Shorter questionnaires, staff nurses responding: 39.7%

• NPs responding to either short or long questionnaires: 36-37% readmitted

The outlier, by gestalt, was when staff nurses responded to patients completing the more detailed questionnaires including heart/lung symptoms: 44% of those patients were readmitted.

Sounds Like Yinz Need Come On Back In, Hon

Notably, though, patients older than 65 who were enrolled in remote monitoring were significantly more likely to be readmitted, and had fewer days at home: 39-48% were readmitted among the intervention groups, compared to only 30% for usual care, with an odds ratio around 0.6 for days at home with the intervention (unfavorable and statistically significant).

The staff nurses were again more likely than the NP teams to recommend readmission for the elderly, although all were more likely to readmit than patient- or PCP-led behaviors (i.e., usual care).

Remote monitoring was even more likely to call back patients who had been discharged to skilled nursing facilities: ~42% in the intervention groups vs. 25% who were not enrolled.

Did any of this help?

The intervention did not clearly influence mortality at 90 days, which was 6.5% with usual care and ranged from 5.4% to 8.8% in the intervention arms among all patients.

However, the data suggests monitoring may have had different effects for older vs. younger patients.

Among those <65 years old, mortality was numerically slightly lower in the intervention group than with usual care: no signal of harm.

For those older than 65, there was a concerning numerical increase in mortality among patients assigned to remote monitoring: 6.6% with usual care (~identical to that with usual care in the overall population), but rising to about 10% in the intervention groups. (See table e14 in the supplemental appendix. Again, it was not powered for mortality, and this could easily be due to chance.)

We’ll Just Keep An Eye On Ya N’at

As hospital-based physicians are reminded daily, there is an increasing population in the U.S. of semi-ambulatory patients whose physiology is so abnormal at baseline that many would meet common criteria for hospitalization on an ordinary day.

When dehydration or an ordinary respiratory or GI illness intrudes, causing their pulse to rise from its usual 109 to 125, say, or they turn their oxygen up to 5 liters from 2 to ‘get more air’—and they then push a button in an app to request a call back for assistance—who’s going to tell that elderly and chronically ill person (who was just recently discharged from the hospital with sepsis) to wait it out, based solely on a phone conversation?

Not most staff nurses. NPs might assuage their own anxiety by ordering a chest film, e.g., and avoid advising immediate readmission. Physician behavior was not tested in this study, but “just go to the E.R.” has been recited by enough half-asleep on-call docs (including newsletter authors) that it might as well be a pre-recorded mantra.

Once the patient arrives at the E.D., the psychological bias toward admission for a patient who was funneled there through a hospital-sanctioned remote monitoring program would presumably also be high.

More Care is Not Always Better Care

If remote monitoring increases readmissions that are just in time to treat emerging illnesses and avert death, then bring on the monitoring.

But that’s not what happened here.

Not only was mortality not improved, it was nominally higher for elderly patients in all the intervention groups. That raises the possibility that they were harmed by the additional inpatient care they received (or the difference could be simply due to chance).

In the closest analogue to this trial, ENCOMPASS, remote monitoring paired with a more robust response framework also led to more readmissions; however, mortality was significantly lower among readmitted patients (a good result). It was conducted at seven North Carolina hospitals by the (ethically challenged) Atrium Health.

Similar programs targeted toward COPD and heart failure have for the most part failed to find a benefit in reducing readmissions or quality of life.

The unsuccessful programs have generally shared a lack of infrastructure providing for meaningful action based on changes in patient status. Building that infrastructure — the “hospital at home” model — faces large barriers, which reduce to the fact that homes aren’t, and can’t be, hospitals.

None of this is to say that remote monitoring can’t help; it probably could.

Emerging home-based monitoring technologies (e.g., wearables) will spawn a vast new body of research on how and whether monitoring can prevent not just hospital readmission, but the need for readmission, by enabling more effective post-discharge management of chronic illnesses.

For now, though, products face large barriers of their own, primarily integration with the technologically primitive and highly regulated and innovation-averse U.S. health system, which is dominated by proprietary platforms (e.g., Epic) that have thus far appeared resistant to enabling interoperability with potentially disruptive technologies.

References

MS, S. Y. M. (2026). Remote Monitoring Approaches to Reduce Readmissions After Infection and Sepsis: A Randomized Clinical Trial. JAMA Network Open, 9(6), e2616641–e2616641. https://doi.org/10.1001/jamanetworkopen.2026.16641

Taylor, S. P., Eaton, T., Rios, A., Boyd, D., Tapp, H., McWilliams, A., Chou, S.-H., Halpern, S., Angus, D. C., McCurdy, L., Ganesan, A., Nguyen, H., Connor, C. D, & Kowalkowski, M. (2025). Proactive Telehealth-Based Sepsis Transition and Recovery Support, Hospital Readmission, and Mortality. JAMA Internal Medicine, 185(10), 1238–1238. https://doi.org/10.1001/jamainternmed.2025.3699

Ong, M. K., Romano, P. S., Edgington, S., Aronow, H. U., Auerbach, A. D., Black, J. T., Marco, T. De, Escarce, J. J., Evangelista, L. S., Hanna, B., Ganiats, T. G., Greenberg, B. H., Greenfield, S., Kaplan, S. H., Kimchi, A., Liu, H., Lombardo, D., Mangione, C. M., Sadeghi, B., … Fonarow, G. C. (2016). Effectiveness of Remote Patient Monitoring After Discharge of Hospitalized Patients With Heart Failure: The Better Effectiveness After Transition–Heart Failure (BEAT-HF) Randomized Clinical Trial. JAMA Internal Medicine, 176(3), 310–318. https://doi.org/10.1001/jamainternmed.2015.7712

Asch, D. A., Troxel, A. B., Goldberg, L. R., Tanna, M. S., Mehta, S. J., Norton, L. A., Zhu, J., Iannotte, L. G., Klaiman, T., Lin, Y., Russell, L. B., & Volpp, K. G. (2022). Remote Monitoring and Behavioral Economics in Managing Heart Failure in Patients Discharged From the Hospital. JAMA Internal Medicine, 182(6), 643. https://doi.org/10.1001/jamainternmed.2022.1383

Pinnock, H., Hanley, J., McCloughan, L., Todd, A., Krishan, A., Lewis, S., Stoddart, A., van der Pol, M., MacNee, W., Sheikh, A., Pagliari, C., & McKinstry, B. (2013). Effectiveness of telemonitoring integrated into existing clinical services on hospital admission for exacerbation of chronic obstructive pulmonary disease: researcher blind, multicentre, randomised controlled trial. BMJ, 347(oct17 3), f6070–f6070. https://doi.org/10.1136/bmj.f6070

Sodium bicarbonate is routinely administered to patients during in-hospital cardiac arrest. Acidemia is one of the “Hs” on the list of “Hs and Ts” associated with pulseless electrical activity, the most common in-hospital arrest rhythm.

There has been no strong evidence to support this practice, but (the thinking goes), virtually all patients experiencing cardiac arrest are either acidemic or will soon become so; acidemia can worsen cardiac function and blunt the response to catecholamine vasopressors; treating the acidemia with buffer might help and seems exceedingly unlikely to harm; when the alternative is death, an agent with even a small chance of benefit seems worth using; we should therefore give sodium bicarbonate.

Observational evidence can be cited to support bicarbonate use, with a strong association between bicarbonate administration and survival and ROSC in non-shockable rhythms among 23,000 out-of-hospital cardiac arrests (Niederburger et al Resuscitation 2023).

There’s data pointing in the opposite direction, as well: bicarbonate was associated with a 15% reduced chance of survival among 319,000 in-hospital arrests in the AHA’s registry (Holmberg et al, Resuscitation 2023), and with worsened neurologic recovery among 26,000 out-of-hospital arrests in Canada (Kawana et al, Resuscitation 2017).

Leading resuscitation guidelines don’t recommend bicarbonate routinely for cardiac arrests, reserving it as a treatment for hyperkalemia and certain drug overdoses.

Defying AHA’s “not recommended” red box in its guideline updates (next to its discouragements for routine calcium and magnesium), clinicians continue to provide bicarbonate in about half of the in-hospital cardiac arrests in the U.S.

The Bicarbonate for In-Hospital Cardiac Arrest trial was recently published, providing the strongest evidence available on the matter by far. Is it enough to answer the question, though?

The BIHCA Trial

At 21 hospitals in Denmark, 913 adults experiencing in-hospital cardiac arrest who had received at least one dose of epinephrine were randomized to receive either 50 mmol sodium bicarbonate (50 mL pushes of 8.4% NaHCO3) or placebo, as soon as possible after each dose of epinephrine, for a total of two doses.

About 130 patients had ROSC, termination of resuscitation, recognition of a DNR order, or other reasons not to give bicarbonate, and were excluded.

Among the 779 analyzed:

• About 86% had non-shockable rhythms (PEA or asystole).

• The buffer dose was “effective” in that pH rose, with alkalosis and hypernatremia occurred more often in the sodium bicarbonate arm.

• ROSC (the primary outcomes) was achieved by 39% receiving bicarbonate vs. 37% receiving placebo (risk ratio, 1.05 [95% CI, 0.88-1.24]; P = .62);

• 30-day survival was 12% in the bicarbonate group and 9.1% with placebo (risk ratio, 1.25 [95% CI, 0.84-1.88]);

• Good neurologic outcomes occurred in 8.1% (30 patients) receiving bicarbonate vs 5.4% (22 patients) getting placebo (risk ratio, 1.39 [95% CI, 0.82-2.34]).

The authors’ conclusion:

“There was no significant difference in sustained return of spontaneous circulation between sodium bicarbonate and placebo in adults with in-hospital cardiac arrest. These findings do not support routine administration of sodium bicarbonate for patients with in-hospital cardiac arrest.”

Sounds perfectly reasonable, in a research paper.

And then there’s the world of clinical practice.

So You’re Saying There’s a Chance

Read more

Vitamin C would seem to be an excellent candidate intervention for patients with severe burns. The reactive oxygen species released in massive quantities during burn injuries cause endothelial damage, capillary leakage, edema, shock, and widespread inflammation. Vitamin C scavenges these oxygen-free radicals, potentially mitigating this pathophysiologic cascade.

In a tiny pseudo-randomized trial of 37 patients with severe burns (Tanaka et al, JAMA Surg 2000), those receiving high-dose vitamin C supplementation required much lower volumes of infused fluids to reach resuscitation targets and had shorter durations of mechanical ventilation, without a mortality difference. Patients received the undeniably high dose of 66 mg/kg/hr for 24 hours (about 5 g per hour, or about 120 g per day for a 70 kg patient). The trial was unblinded, and randomization allocation was not concealed.

In the absence of higher-quality evidence, and in a foreshadowing of what was to transpire in medical ICUs fifteen years later, high-dose vitamin C gained popularity in burn units in the wake of the small trial.

A 2017 retrospective review of 40 patients at UW Seattle’s Harborview seemed to support the finding, with authors concluding “vitamin C is associated with a decrease in fluid requirements and an increase in urine output during resuscitation after thermal injury, [and] can be safely used without an increased risk of renal failure.”

A 2022 scoping review analyzed ten published papers: virtually all suggested a benefit of high-dose vitamin C supplementation in burn patients.

Guidelines and reviews mentioned the Tanaka data as plausible, while also highlighting the need for more rigorous study of vitamin C supplementation in severe burns.

Twenty-six years later, that trial was completed and published in JAMA.

The VICTORY Trial

At 24 burn centers in the Americas, Europe, and Asia, 666 patients with severe burns were planned to be randomized to receive either vitamin C (50 mg/kg every 6 hours for 96 hours, i.e., ~200 mg/kg/day or ~14 g/day for a 70 kg patient; i.e., a lower dose for longer than in Tanaka et al) or placebo for four days.

The trial was stopped early at the first prespecified interim analysis after 238 patients had been enrolled, for apparent harm.

Mortality at 28 days was significantly higher in the vitamin C group (15.0% vs 7.6%; adjusted RR, 1.96 [95% CI, 1.32-2.90]; P = .001), as was hospital mortality (23.3% vs 16.1%; adjusted RR, 1.44 [95% CI, 1.03-2.00]; P = .03).

These figures implied at least a 32% relative increased risk for death from receiving vitamin C (and probably significantly higher).

The composite outcome (death or persistent organ dysfunction at 28 days) occurred in 40.8% (49 of 120) patients assigned to vitamin C, compared with 29.7% (35 of 118) assigned to placebo (adjusted relative risk 1.28 with a 95% confidence interval from 0.99 to 1.65). The P value was .06, but the result crossed the trial’s prespecified boundary for futility or harm.

Renal failure occurred numerically more frequently in the vitamin C arm (15% vs. 11%), and numerically almost twice as many vitamin C patients required dialysis (10.8% vs. 5.9%); these did not reach statistical significance.

Discussion

Many more patients with severe burns died while receiving high-dose vitamin C supplementation than placebo. Vitamin C patients experienced more persistent organ dysfunction, with a strong suggestion of an increase in renal failure. Oxalate crystal formation is a known risk from high-dose vitamin C, although this was not a proven mechanism of harm in VICTORY.

The flameout of high-dose vitamin C in VICTORY is a cautionary coda that concludes the inspiring but imaginary story first told in Tanaka et al in 2000.

In an ironic conceptual cross-contamination between specialties, the Tanaka data was cited by authors of various papers touting vitamin C for sepsis as a basis for its putative therapeutic effects. This apparent evidentiary support helped to launch the vitamin C cargo cult that seized the critical care community for a half-decade starting in 2017.

In an interesting parallel to the burn literature, multiple randomized trials in ICUs suggested that vitamin C therapy could save lives in sepsis and critical illness generally—producing a 27% relative risk reduction for mortality, according to a meta-analysis of 16 randomized trials.

The large multicenter randomized LOVIT trial (NEJM 2022, n=872) then reported the counterfactual: a 17% increase in mortality and 30% increase in persistent organ dysfunction (relative) in patients receiving vitamin C monotherapy for sepsis. This followed the ACTS trial, which also showed no benefit for vitamin C in sepsis.

Although it took 26 years after the first kindling of enthusiasm, results from the VICTORY and LOVIT trials seem likely to definitively end any defensible use of high-dose vitamin C for critical illness in either clinical practice or research.

References

High-Dose Intravenous Vitamin C and Mortality and Organ Dysfunction in Severe Burn Injury. JAMA. 2026 Jun 10:e2610728. doi: 10.1001/jama.2026.10728. Epub ahead of print. PMID: 42267884.

Lee ZY, Ortiz-Reyes L, Lew CCH, et al. Intravenous vitamin C monotherapy in critically ill patients: a systematic review and meta-analysis of randomized controlled trials with trial sequential analysis. Annals of Intensive Care. 2023;13(1):14. doi:https://doi.org/10.1186/s13613-023-01116-x

‌Soltany A, Al Aissami M. A scoping review of the role of ascorbic acid in modifying fluid requirements in the resuscitation phase in burn patients. Ann Med Surg (Lond). 2022 Mar 2;75:103460. doi: 10.1016/j.amsu.2022.103460. PMID: 35386786; PMCID: PMC8978049.

Tanaka H. Reduction of Resuscitation Fluid Volumes in Severely Burned Patients Using Ascorbic Acid Administration. Archives of Surgery. 2000;135(3):326. doi:https://doi.org/10.1001/archsurg.135.3.326

‌Kahn, Steven et al . Resuscitation After Severe Burn Injury Using High-Dose Ascorbic Acid: A Retrospective Review. Journal of burn care & research : official publication of the American Burn Association. 2011

WARNING: This post says the word “mucus” about 1,427 times.

Plugging of the small airways (and sometimes the large ones) by mucus is a major problem for many mechanically ventilated patients. Mucus plugging is a notorious source of relatively rapid worsening of ventilation and gas exchange, initiating a cascade of concern and diagnostic and therapeutic maneuvers by the care team.

In a common scenario, at 8 a.m., it looks like the patient is going to experience a respiratory-induced cardiac arrest; at lunchtime, after very little if any meaningful intervention, the patient looks like they will soon extubate themselves and walk home. (Well, maybe not home: most patients experiencing significant mucus plugging are elderly, frail, have severe chronic lung disease, or a combination of these.)

Mucus plugging is a stealthy bogeyman whose presence in ventilated patients is usually inferred. Garden-variety mucus plugging of the smaller airways can’t be easily identified on chest radiographs or CT scanning. Even bronchoscopy has major limitations in identifying or treating the issue, because the bronchoscope only reaches about a third of the distance into the tracheobronchial tree; mucus plugging can often be diffuse and compromise beds of smaller airways unreachable by the endoscope.

Severe, obvious, abundant mucus production requiring frequent suctioning helps predict extubation failure in mechanically ventilated patients. At the much more common lesser magnitudes, however, evidence is lacking as to whether the degree of clinically observable mucus production correlates with patient-centric metrics like duration of ventilation, ICU length of stay, etc.

The traditional, pragmatic response to this uncertainty in vented patients with abundant mucus production has been to instill N-acetyl-cysteine (NAC, still commonly called by its old brand name Mucomyst™ in the U.S.) down the endotracheal tube. NAC liquefies mucus through a direct chemical reaction in which its free sulfhydryl group attacks the disulfide bonds connecting mucin polymers, breaking down the viscous, stringy strands into a thinner, more liquid state. I’m sorry I had to write that.

NAC works great in test tubes but has never been demonstrated to be clearly beneficial to vented patients, either as a routine part of respiratory care or as a targeted therapy for those with high mucus production.

Further, nebulized NAC is not always benign: its chemical properties cause irritation to tracheobronchial mucosa, infrequently and unpredictably triggering bronchospasm.

In the largest randomized trial (NEBULAE, n=922, NEJM 2018), scheduled acetylcysteine (with the bronchodilator salbutamol) did not result in fewer ventilator days than as-needed mucolytics + salbutamol.

Mucolytic agents can also be delivered I.V. (NAC, bypassing its high first-pass metabolism by the liver) or enterally (carbocysteine or erdosteine).

A meta-analysis of very low-quality evidence (i.e., of many small underpowered trials of varying methodology and rigor, testing multiple delivery methods, I.V., nebulized, oral) did not suggest reduced vent days from mucoactive agents in the ICU generally (but did suggest shortened ICU stays).

A large, high-quality randomized trial was then performed.

The MARCH Trial

At 71 ICUs in the U.K., 1,956 mechanically ventilated adults who were considered by their physicians to have difficult-to-clear secretions were randomized into a 2 x 2 factorial design to receive either-or-both enteral carbocysteine 750 mg q8h, 4 mL nebulized hypertonic saline q4h, or placebo. (I.E., they could randomly get either agent and a placebo, both agents, or two placebos).

The first important finding was that there was no interaction effect between the two agents (increasing the confidence in the independent effect of each agent on the results).

Neither carbocysteine nor hypertonic saline reduced ventilator days over placebo. For both agents, the numerical direction favored placebo. There were no improvements in secondary outcomes, either.

Both agents appeared to independently harm patients:

• Carbocisteine increased clinically important upper GI bleeding (1.4% with carbocisteine vs 0.2% without, P=0.01).

• Hypertonic saline increased bronchoconstriction requiring bronchodilator use (2.4% with saline vs. 0.4% without, p=0.001)

Hypertonic saline is known to cause bronchoconstriction in susceptible patients, so the result was not overly surprising.

Carbocysteine has a known association with GI bleeding: in its UK product labeling, peptic ulcer disease is a contraindication, and GI hemorrhage is an explicit warning. The putative mechanism is by breaking down helpful, protective mucus in the GI tract, exposing epithelium to acid-induced ulceration with resulting bleeding.

But Remember Those Great Ads With the Mucus-Monster Guy? That Stuff Must Work

Carbocysteine is not approved by the U.S. FDA, but is (or was) in use in the U.K., Europe, Asia, and South America under brand names like Mucodyne.

Oral erdosteine has weak evidence to support its use in outpatients with moderate-to-severe COPD, based on the RESTORE trial (U.K./Italy). Erdosteine has orphan drug status in the U.S., and is not approved for use as a mucolytic in the ICU.

In the U.S., guaifenesin is sometimes deployed in vented patients in the hope that it will improve secretion clearance. Long used as an over-the-counter expectorant in the ambulatory setting, a widely cited randomized trial (in 2014) showed no effect of guaifenesin on either sputum characteristics or symptom resolution in outpatients with productive coughs. There is virtually zero evidence to support its use in mechanically ventilated patients.

Conclusion

Mucus plugging is a prominent and difficult problem affecting many mechanically ventilated patients.

Unfortunately, no therapy has been shown to be consistently beneficial, and some can have harms.

Common sense preventive and mitigation measures like adequate humidification, appropriately frequent turns and suctioning, (maybe) targeted use of chest physiotherapy and bronchoscopy (for suspected central obstructions) are the primary tools at the clinician’s disposal.

Once preventive measures are in place, airway clearance is yet another of the major physiologic functions that critically ill patients must regain on their own, largely independent of care teams’ best intentions and efforts. Whether and when a severely ill mechanically ventilated patient does so will often determine their outcome.

References

Connolly B, Dickson N, Campbell C, et al. Carbocisteine or Hypertonic Saline for Acute Respiratory Failure. New England Journal of Medicine. Published online June 10, 2026. doi:https://doi.org/10.1056/nejmoa2603406

‌Dal Negro RW, Wedzicha JA, Iversen M, Fontana G, Page C, Cicero AF, Pozzi E, Calverley PMA; RESTORE group; RESTORE study. Effect of erdosteine on the rate and duration of COPD exacerbations: the RESTORE study. Eur Respir J. 2017 Oct 12;50(4):1700711. doi: 10.1183/13993003.00711-2017. PMID: 29025888; PMCID: PMC5678897.

Hoffer-Schaefer A, Rozycki HJ, Yopp MA, Rubin BK. Guaifenesin has no effect on sputum volume or sputum properties in adolescents and adults with acute respiratory tract infections. Respir Care. 2014 May;59(5):631-6. doi: 10.4187/respcare.02640. Epub 2013 Sep 3. PMID: 24003241.

Anand R, McAuley DF, Blackwood B, Yap C, ONeill B, Connolly B, Borthwick M, Shyamsundar M, Warburton J, Meenen DV, Paulus F, Schultz MJ, Dark P, Bradley JM. Mucoactive agents for acute respiratory failure in the critically ill: a systematic review and meta-analysis. Thorax. 2020 Aug;75(8):623-631. doi: 10.1136/thoraxjnl-2019-214355. Epub 2020 Jun 8. PMID: 32513777; PMCID: PMC7402561.

van Meenen DMP, van der Hoeven SM, et al. Effect of On-Demand vs Routine Nebulization of Acetylcysteine With Salbutamol on Ventilator-Free Days in Intensive Care Unit Patients Receiving Invasive Ventilation: A Randomized Clinical Trial. JAMA. 2018 Mar 13;319(10):993-1001. doi: 10.1001/jama.2018.0949. PMID: 29486489; PMCID: PMC5885882.

Post-cardiac arrest research and care delivery are guided by the hope that a therapeutic window exists between an episode of hypoxic brain injury and irreversible neuronal death, i.e., permanent brain damage.

This is entirely rational, because hypoxic-ischemic brain injury is a complex process that proceeds in phases: the initial hypoxic insult (occurring over minutes between arrest and ROSC), followed by secondary damage from the inflammatory processes of reperfusion injury.

In ischemic stroke, for example, an at-risk but viable penumbra of salvageable brain tissue can be visualized surrounding the ischemic core of an infarcted vessel’s territory. Thrombectomy can restore perfusion, saving these at-risk areas from tissue death.

No analogous zone of revivable brain can be visualized from the more diffuse injuries resulting in hypoxic-ischemic encephalopathy. Still, animal models have suggested that interventions (e.g., induced hypothermia) might mitigate damage from the secondary post-reperfusion processes that evolve over days after cardiac arrest.

And at least one intervention that might worsen that damage: excessive oxygen delivery has been suspected of producing toxicity via reactive oxygen species generation that overwhelms antioxidant defenses, amplifying reperfusion injury and neuronal death.

Hyperoxia after cardiac arrest has been associated with worse neurologic injury and increased neuronal death in experimental animals.

In humans, the subset of patients in the PILOT trial with cardiac arrest who received conservative oxygen (target <96% SpO2) had almost double the rate of a good neurologic outcome compared to those targeted to SpO2 96%-100%.

Patients in the ICU-ROX trial with cardiac arrest had greater 6-month survival and a trend toward improved neurologic outcomes.

On the other hand, hypoxemia carries obvious risks of worsening an evolving brain injury. Targeting lower oxygen targets might allow a portion of patients to slip into harmful levels of hypoxemia.

The LOGICAL trial was conducted to discern the potential benefits or harms of lower oxygenation targets in patients suffering cardiac arrest.

At 53 ICUs in Australia, New Zealand, and Ireland, 1,840 patients who were comatose and vented after cardiac arrest were randomized to either conservative (target SpO2 90-95%, allowed to receive room air) or liberal oxygen therapy (target SpO2 90-100%, minimum FiO2 0.3).

This was elegantly nested within the Mega-ROX research program (an ongoing megatrial/platform testing oxygenation strategies in critical illness).

About three quarters had out-of-hospital cardiac arrest, roughly half had a shockable first rhythm, and the median time from ROSC to randomization was about 7 hours.

LOGICAL Results

The two groups achieved adequate separation, with the conservative group spending substantially less time with SpO₂ ≥97% and much more time breathing room air.

More liberal oxygen patients had hyperoxia (PaO2 >100mm Hg) (78% vs. 58%), and more conservative patients had episodes of hypoxemia (PaO2 <60 mm Hg), 43% vs 28%. Time spent with SpO2 < 88% was a median of zero hours in both groups.

But at 180 days, the rate of survival with a favorable functional outcome (Extended Glasgow Outcome Scale score of 5 to 8) was nearly identical at 38.2% in the conservative group vs. 39.7% in the liberal arm.

If anything, the point estimate slightly favored liberal oxygen, though the confidence interval was compatible with modest benefit or modest harm.

Secondary outcomes were also similar: survival at 180 days was 48.0% with conservative oxygen and 49.7% with liberal oxygen. Quality-of-life scores, cognitive outcomes, mechanical ventilation duration, ICU length of stay, hospital length of stay, and discharge home were all similar.

Subgroup analyses did not identify a group clearly benefiting from conservative oxygen. Results were consistent by arrest location, initial rhythm, medical versus nonmedical cause, and earlier versus later randomization.

Conclusions

Brain injury evolves for hours or days after the sudden diffuse ischemic injury caused by hypoperfusion during cardiac arrest.

However, this secondary process of reperfusion injury has thus far been stubbornly resistant to intervention. Induced hypothermia was strongly believed to mitigate reperfusion injury, until multiple randomized trials failed to show a benefit.

Based on the LOGICAL trial, it is exceedingly unlikely that avoidance of hyperoxia can make a meaningful difference, either.

That being said, since most patients were randomized more than six hours after achieving ROSC, the LOGICAL trial is vulnerable to the same criticisms of most of the negative hypothermia trials: the intervention may have been delivered too late to make a meaningful difference, and this isn’t proof that conservative oxygen doesn’t help.

In the larger context of an absent signal for benefit from differential oxygenation strategies, however, most intensivists will likely log the trial results as supporting a logical heuristic of managing oxygenation for vented cardiac arrest patients: not too high, not too low, with the knowledge that almost all of the eventual outcome has already been determined before the patient’s arrival to the ICU.

Reference

LOGICAL Investigators and the Australian and New Zealand Intensive Care Society Clinical Trials Group; Hodgson CL, Mackle D, Mather AM, Bailey M, Beasley R, Beehre N, Bernard S, Brickell K, Crichton B, Deane AM, Eastwood G, Finfer S, Henderson S, Higgins AM, Hunt A, Kasza J, Lawrence C, Linke NJ, Litton E, McDonald CF, Moore J, Nichol AD, Olatunji S, Parke R, Peake S, Secombe P, Seppelt IM, Serpa Neto A, Trapani T, Turner A, Udy A, Young PJ. Conservative Oxygen for Unresponsive Patients after Cardiac Arrest. N Engl J Med. 2026 Jun 10. doi: 10.1056/NEJMoa2513814. Epub ahead of print. PMID: 42267831.

A 67-year-old man with pulmonary fibrosis and chronic obstructive pulmonary disease and heart failure who uses ambulatory oxygen at 2 liters/minute with exertion presents to the ED with worsening dyspnea and cough.

He has been hospitalized for respiratory exacerbations (about twice per year, most recently six months ago), with no history of resistant organisms or Pseudomonas, and has never required mechanical ventilation.

He is febrile, hypotensive, and has an elevated lactate; he requires high-flow nasal cannula oxygen (FiO2 50% / 50L) to maintain oxygen saturation >90%. His chest film shows a new right lower lobe infiltrate. He is provided one liter of crystalloid, norepinephrine, ceftriaxone, azithromycin, and hydrocortisone, and is admitted to the ICU.

On hospital day 3, he has improved significantly. He has been afebrile for almost 24 hours; he requires only 4 liters/minute of oxygen; blood pressure is normal off vasopressors; heart rate is 110 (down from 140); respiratory rate is 24 and unlabored. Blood cultures are negative thus far. Sputum cultures and nasal respiratory PCR were not collected. His mental status is at baseline, according to his wife.

You decide he is ready for transfer to the medical unit.

Read more

In the 1990s, an elite rower had a recorded venous pH of 6.74 after a 2,000-meter race. Rather than collapsing or requiring hospitalization, he caught his breath, chatted with the crowd, and drove home.

Elite sprinters routinely hit pHs of 7.10 with lactate levels exceeding 10 mmol/L. Even weekend-warrior marathoners can easily enter the pH range of 7.20 to 7.30 near the end of a race.

So who’s afraid of a few extra stray hydrogen ions?

Well, the etiology matters: exertional metabolic acidosis is brief, self-limited, and occurring in healthy people with intact cardiopulmonary-renal function. By contrast, critically ill patients have ongoing acid generation from organ dysfunction with diminishing physiologic reserves. Acidemia can depress myocardial function and blunt catecholamine responsiveness, although the thresholds at which this occurs have been debated.

But many critically ill patients with acidemia have something distantly in common with those elite athletes: their acidemia is an adaptive response to stress, not necessarily harmful, and will resolve without consequence after the underlying stressors are relieved.

Others have severe acidemia from an unrelenting metabolic decompensation near death. For them, buffering out some H+ may make their labs look marginally better for a short time, but will have zero impact on the ultimate outcome.

Is there a third group, the Goldilockses of the ICU, who have acidemia that is both maladaptive (i.e., worsening an already bad situation) and remediable?

A series of randomized trials have sought to find this elusive group: BICARICU-1 and BICARICU-2, which we reviewed here —

— and most recently the SODa-BIC trial in NEJM.

The SODa-BIC Trial

In 55 ICUs in seven countries (in Australasia, Brazil, Japan, Canada, and Ireland), 500 critically ill adults with vasopressor-dependent shock and metabolic acidosis (pH < 7.30) were enrolled.

It was a sick group. About 40% had septic shock; ~25% had metabolic acidosis attributed to cardiogenic shock or cardiac arrest. Most had mild acute kidney injury (Cr ~1.5 mg/dL). About 80% were mechanically ventilated.

They were randomized to receive either 5% sodium bicarbonate (600 mmol/L) or placebo (5% dextrose) infusions for ≤5 hours to achieve pH ≥7.30, and followed for 30 days to assess the incidence of a “major adverse kidney event”: either death, need for dialysis or CRRT, or a persistent doubling (or greater) of baseline creatinine. (This composite endpoint, reused across trials, is referred to as MAKE30.)

SODa-BIC Results:

There was separation between groups, and bicarbonate improved acidosis. During the first 24 hours, the sodium bicarbonate group had pH values ~0.05 higher with serum bicarbonate levels than placebo.

But after 30 days, the rate of the MAKE30 composite outcome was no different between arms — 40.2% in the NaHCO3 group and 39.4% in the placebo group.

In-hospital mortality was near-identical (25.4% for NaHCO3 vs 24.0% with placebo).

However, sodium bicarbonate-treated patients had persistent renal dysfunction less often, numerically (14% vs 18% with placebo), required dialysis/CRRT 4% less often overall (16.8% vs 20.9%) and were less often dialysis-dependent at 30 days (5.7% vs 9.1%)—none of which were statistically significant.

There were no significant safety events, but four patients in the sodium bicarb group had significant hypokalemia (<3.0 mmol/L) requiring replacement, vs. none in the placebo arm.

Headlines announced “Sodium Bicarbonate Failed to Improve Kidney Outcomes in ICU” — which is technically true, but there’s more to consider.

Read more

Sepsis resuscitation decisions exist in a tension between two competing instincts: give fluids aggressively to restore perfusion, or limit initial fluid boluses in favor of early vasopressors to avoid potentially harmful fluid overload.

In the 1999 EGDT trial (Rivers NEJM), by repeatedly bolusing patients to a central venous pressure goal, the intervention arm received five liters of crystalloid (and the controls three-and-a-half) in the first six hours, with both groups exceeding +13 liters infused in the first 72 hours. Perhaps unsurprisingly, more than 60% required mechanical ventilation.

With its likely spurious signal of a mortality benefit in the goal-directed arm, this trial encouraged a generation of intensivists to continue the longstanding practice (which predated Rivers) of infusing unnecessarily large volumes of fluids into septic patients in the hope of restoring and maintaining perfusion.

Fifteen years later, the Process, Arise, and Promise trials (c. 2014) refuted the Rivers “goal-directed” package of dobutamine, transfusion, and central venous catheterization, without challenging the dominant fluid-forward paradigm of sepsis care. Patients in those trials received ~2 liters before randomization and another 2-3 liters in the first six hours (~3.5 to 5.5 liters total in 6 hours, generally).

Yet those three trials also showed clinical practice moderating toward less-aggressive fluids, with (e.g.) patients in Process and Arise “only” receiving an additional 4.5 to 5 liters from 6 to 72 hours (compared to ~8 liters in Rivers). There was no difference in mortality between arms in any of the three trials.

This set the stage for two large randomized trials testing a more restrictive (i.e., vasopressors-forward) approach to fluid resuscitation in sepsis after an initial bolus of ~2-3 liters. Neither showed a significant difference between the two fluid strategies:

• CLASSIC (NEJM 2022, n=1,554, Europe): identical 90-day mortality (very high at ~42%), with only 500 mL (restrictive) vs 1,300 mL (liberal) infused on the first day after randomization (plus ~3 liters in each group before enrollment);

• CLOVERS (NEJM 2023, n=1,563, U.S.): 90-day mortality 14% (restrictive) vs 14.9% (liberal), patients received ~2 liters before randomization, then 500 mL (restrictive) vs. 2,300 mL (liberal) in the first 6 hours; only 37% of liberal-arm patients required vasopressors vs 59% (restrictive) on the first day

Not to be outdone, the Australasian critical care consortium ANZICS recently published results from the ARISE FLUIDS trial, which will not extinguish the controversy over this deceptively simple-seeming clinical question.

ARISE FLUIDS Trial

At 51 sites in Australasia, 1,000 adults presenting to the ED receiving a diagnosis of sepsis, high lactate and hypotension after ≤1,000 mL I.V. fluid (but ≤2,000 mL) were randomized to six hours of either a restrictive (vasopressor-forward) or liberal fluid strategy (+1,000 mL then 500 mL boluses, 30 mL/kg in 3 hours advised, plus vasopressors after that or when deemed fluid-unresponsive).

There was meaningful separation: after the ~1,500 mL pre-randomization, restricted patients received only an additional ~500 mL in the first 6 hours vs. ~1,500 mL in the liberal arm (~2 L vs. ~3 L), with a ~1.1 L difference at 24 hours (~2,640 mL vs ~3,740 mL).

Fluid restriction led to higher vasopressor use in the first 24 hours (87% vs 68%)—about an additional one in five patients requiring vasopressors in the restricted arm.

No difference in the primary outcome: both arms had an identical number of 76 days alive and out of the hospital through day 90.

Mortality at 28 days (12.9% vs 10%) and 90 days (16.4% vs 14.4%) was nominally higher in the vasopressor group (non-significant).

There was eightfold more pulmonary edema in the liberal group: 5% vs. 0.6%, statistically significant. Pulmonary edema was identified by clinician opinion, based on hypoxemic respiratory failure, chest radiography, or other supporting findings. Clinicians were unblinded to treatment assignment, but were usually caring for study patients well past the six-hour mark, and thus not involved directly in the trial.

Despite this increase in clinician-identified pulmonary edema, the rate of mechanical ventilation was equal in both groups (15%).

Vasopressor-forward patients required ICU admission more often (77% vs 68%). Understandable, as many hospitals require vasopressors to be managed in ICUs.

Patients in the liberal group also required about one additional day of mechanical ventilation and renal replacement therapy—although these were not statistically significant.

Central lines were largely unnecessary. Most patients received vasopressors through a peripheral IV, without any major safety signal reported in the trial.

What to Take From Arise-Fluids?

Arise-Fluids “settles” the controversy of fluids-versus-vasopressors for septic shock by again demonstrating their general equivalence when all patients’ outcomes are combined.

But partisans can find favor in its data for their preferred tilt in either direction:

1. Restrict Fluids! After an initial fluid bolus of one or two liters for patients without volume depletion, restricting fluids and relying on vasopressors to maintain perfusion is reasonable. 30 cc/kg was always an arbitrary figure; over-resuscitation causes pulmonary edema and possibly adds time on the ventilator or dialysis circuit. Iatrogenic volume overload is an under-recognized cause of disability and incomplete recovery from critical illness, which Arise-Fluids did not measure.

2. Bolus Away! Liberal fluid resuscitation is generally superior from a practical clinical standpoint. A fluid-forward approach reduces the need for vasopressors, which in turn can reduce the need for ICU admission, conserving resources and improving care delivery system-wide. Pulmonary edema is not a hard clinical outcome and did not increase the need for mechanical ventilation either in Arise-Fluids or in Clovers, the most comparable major trial. The trend toward faster weaning from mechanical ventilation in the restrictive group in Arise-Fluids was not significant. Further, the numerically higher mortality in the restrictive arm is a potential warning of the limits of safe fluid restriction.

Fluids vs Vasopressors: A False Dichotomy

U.S. clinicians have appropriately chafed at policymakers’ attempts at national standardization of sepsis care via the governmental “quality” measure called SEP-1; this resistance has been in large part because of its mandate for providing three, four, or five-liter boluses to severely obese patients with sepsis who may also have concomitant cardio-pulmonary-renal failure.

Subsequent iterations of the rule permit clinicians to deliver smaller fluid boluses they consider beneficial, with documentation of their clinical reasoning.

The Arise-Fluids and Clovers trials together support the safety of just such an individualized, patient-centered approach to resuscitation and hemodynamic management. Deferring vasopressors for aggressive fluids for tolerant or volume-depleted patients; limiting fluid boluses and favoring vasopressors in patients with apparent impending respiratory failure or significant cardio-renal disease: the complexity of sepsis and each patient’s physiology necessitate thoughtful tailoring of their plan, rather than blindly following a trial result.

Rather than picking a winner in the fluids-vs-vasopressors debate, the Arise-Fluids trial reaffirms the truth that the best approach to sepsis resuscitation is the one tailored to the patient in front of you.

References

ARISE-FLUIDS trial. Vasopressors or Fluids in Early Septic Shock. New England Journal of Medicine. Published online June 11, 2026. doi:https://doi.org/10.1056/nejmoa2516225

Meyhoff TS, Hjortrup PB, Wetterslev J, et al. Restriction of Intravenous Fluid in ICU Patients with Septic Shock. New England Journal of Medicine. 2022;386(26):2459-2470. doi:https://doi.org/10.1056/nejmoa2202707

‌‌The National Heart, Lung, and Blood Institute Prevention and Early Treatment of Acute Lung Injury Clinical Trials Network. Early Restrictive or Liberal Fluid Management for Sepsis-Induced Hypotension. New England Journal of Medicine. 2023;388(6). doi:https://doi.org/10.1056/nejmoa2212663

‌

Attending to the complex needs of patients after their ICU stays—physical, social, and psychological—seems intuitively certain to improve their long-term recovery.

Clinical evidence has not clearly shown that to be the case, but the overall quality of research in this underfunded area has been limited.

At first glance, the latest data from a rare multicenter randomized trial may seem uninspiring, but a closer look provides some encouragement that focused post-ICU interventions might yield benefits to some patients.

The iRehab Trial

At 52 hospitals in the U.K.’s National Health System, 429 ICU survivors of critical illness with ≥48 hours of mechanical ventilation were recruited within 12 weeks of their hospital discharge.

They were randomized to receive either usual care, or a lot of support for six weeks. This included a remotely administered individualized rehabilitation plan that focused on symptom management, exercise targeted at their weak areas, psychological support, and connections to a peer support group.

After eight weeks, there was no improvement in health-related quality of life (0.69 vs. 0.67 on the EuroQoL 5-dimension 5-level questionnaire).

However, the prespecified subgroup who were intubated for 7 days or less (about half in each arm) did experience a benefit (0.74 vs. 0.66 on the QOL metric, statistically significant).

Leg strength and exercise capacity were improved in the intervention group: they could perform 10.6 sit-to-stand repetitions in 30 seconds, vs 9.9 in the controls. They reported less anxiety and fatigue on validated scales, as well.

A Highly Selected Group: Relatively Young and Strong

The patients enrolled in iRehab were highly selected: out of 3,705 screened, 429 were randomized.

But all 3,705 had already been screened with a much more difficult test: surviving an episode of critical illness with mechanical ventilation (>7 days for half of them).

In most trials of early mobilization, the enrolled patients haven’t yet passed this test, which is a major signal of robustness, reserve, and resilience.

The most telling proof of this fact? After six months, only three patients had died (<1%). Their youth (mean age of 55) also aligns with this. They were already relatively strong: able to sit, stand, sit, etc every 3 seconds, 10 times in 30 seconds.

They were also motivated: a third of the excluded patients (~1,100) declined to participate, further distilling the group to those most dedicated to enhancing their quality of life.

Hope for Post-ICU Syndrome

Today, about 35 clinics in the U.S. are dedicated to helping people recover or manage their symptoms after ICU discharge. Almost all are at academic medical centers or large integrated healthcare systems (e.g., Brigham, Pittsburgh, UCLA, Mount Sinai, U of M, Geisinger, and Intermountain).

Until the iRehab trial, evidence has been lacking to support their resource-intensive, multidisciplinary, high-touch approach (e.g., three-hour intake visits including speech therapy, PT, OT, a pharmacist, nutritionist, and multiple physicians and nurses, plus peer support groups and additional services outside the clinic).

A recent JAMA Insights piece (a mini-review article) leans heavily on caveats like “difficult to ascertain … data are sparse … may promote … may decrease … may prevent … may enhance” … Clinicians have always agreed that patients with specific impairments like dysphagia, weakness, malnutrition, or anxiety should receive targeted therapies and support. But the limited research on PICS hadn’t yet established a benefit from a more holistic approach to the syndrome, nor the clinics that coordinate care.

iRehab provides the first convincingly encouraging signal that these efforts can yield benefits.

But not for everyone. iRehab enrolled younger patients who had passed what for most of them would be the defining stress test of their lives: surviving days or weeks of mechanical ventilation.

Intensive multimodal rehabilitation helped some of them recover more strength than patients who did not receive the intervention. And the subgroup analysis suggests that those with a shorter duration of ventilation may have seen small but meaningful gains in quality of life.

Post-ICU clinics prescribe hope, not just clinical interventions. The iRehab trial adds evidence that at least some ICU survivors can benefit from this focused additional care after they make it back home.

Whether the frail or elderly can also benefit would require another trial more representative of the standard population of today’s ICUs.

References

O’Neill et al. Remote Multicomponent Rehabilitation in Intensive Care Unit Survivors: A Randomized Clinical Trial: JAMA 2026

Post–Intensive Care Syndrome. JAMA. Published online January 15, 2026. doi:https://doi.org/10.1001/jama.2025.23666. Butcher BW.

Outcomes after Critical Illness. The New England Journal of Medicine. 2023. Herridge MS, Azoulay É.

Prevalence of Post-Intensive Care Syndrome Among Intensive Care Unit-Survivors and Its Association With Intensive Care Unit Length of Stay: Systematic Review and Meta-Analysis. PloS One. 2024. Ayenew T, Gete M, Gedfew M, et al.

Post-Intensive Care Syndrome as a Burden for Patients and Their Caregivers: A Narrative Review. Journal of Clinical Medicine. 2024. Schembari G, Santonocito C, Messina S, et al.

Post-Intensive Care Syndrome: A Concept Analysis. International Journal of Nursing Studies. 2021. Yuan C, Timmins F, Thompson DR.

Post-Intensive Care Syndrome and Its New Challenges in Coronavirus Disease 2019 (COVID-19) Pandemic: A Review of Recent Advances and Perspectives. Journal of Clinical Medicine. 2021. Nakanishi N, Liu K, Kawakami D, et al.

Communicating to Patients and Families About Post-Intensive Care Syndrome. Chest. 2025. Rolfsen ML, Wilcox ME, Mart MF, et al.

Delirium-Related Psychiatric and Neurocognitive Impairment and the Association With Post-Intensive Care Syndrome-a Narrative Review. Acta Psychiatrica Scandinavica. 2023. Ramnarain D, Pouwels S, Fernández-Gonzalo S, Navarra-Ventura G, Balanzá-Martínez V.

Prevalence and Incidence of Post-Intensive Care Syndrome Among Intensive Care Unit Survivors: A Systematic Review and Meta-Analysis. Annals of Medicine. 2026. Zare-Kaseb A, Sanaie N, Sarmadi S.

Resource Document on the Neuropsychiatric Symptoms of Subacute and Chronic Long COVID. American Psychiatric Association (2024). 2024. Aisha Gillan MD, Melissa Peace MD, Davin Quinn MD, Jon Levenson MD, Thida Thant MD

Post Intensive Care Syndrome (PICS): An Overview of the Definition, Etiology, Risk Factors, and Possible Counseling and Treatment Strategies. Expert Review of Neurotherapeutics. 2021. Ramnarain D, Aupers E, den Oudsten B, et al.

Post-Intensive Care Syndrome After Critical Illness: Incidence and Predictors in a Nationwide Cohort. Anaesthesia. 2026. Oh, Song IA.

Guideline on Multimodal Rehabilitation for Patients With Post-Intensive Care Syndrome Critical Care. 2023. Renner C, Jeitziner MM, Albert M, et al.

Needham DM, Davidson J, Cohen H, et al. Improving long-term outcomes after discharge from intensive care unit: report from a stakeholders’ conference. Critical Care Medicine. 2012.

A 79-year-old woman nursing home resident with chronic kidney disease and chronic back pain on long-term oxycodone is admitted to the ICU with septic shock from healthcare-associated pneumonia. She is intubated for hypoxemic respiratory failure and provided cefepime, vancomycin, hydrocortisone, and a fentanyl infusion.

On hospital day 3, fentanyl is discontinued for somnolence (RASS -3). Over the next 24 hours, she awakens but remains inattentive and does not follow commands. She has prolonged episodes of staring. An EEG during an episode shows epileptiform discharges >2.5/sec. She has no past seizure history. Lorazepam and levetiracetam are provided, and neurology consultation is requested.

Laboratory studies (including sodium, magnesium, calcium, pH, and glucose) are unremarkable except for creatinine 3.1 mg/dL, increased from 1.4 mg/dL at admission, with BUN 82 mg/dL. Urine output is 0.7 mL/kg/hr. CT head shows no acute abnormality.

Read more

Most critically ill patients either arrive anemic or become so during their ICU stays. Many years ago, they were routinely transfused to near-normal hemoglobin targets until the TRICC trial (NEJM 1999) strongly suggested that withholding red cell transfusion until hemoglobin fell to ≤7 g/dL was safe (but not superior, from a mortality or secondary-outcome standpoint).

In a rare example of replicability, dozens of subsequent randomized trials and meta-analyses generally confirmed the safety of restricting transfusion to Hb ≥ 7 or 8 g/dL for most hospitalized patients, including the critically ill and those with active GI bleeding. Restrictive strategies can reduce the total volume of transfused blood by ~50%, in aggregate, without apparent harm (or benefit).

Guidelines from multiple specialty societies have since advised a restrictive approach to transfusion for hospitalized patients generally.

Patients with ischemic heart disease have been a lingering question mark. Those with significant cardiac morbidity, and especially active cardiac ischemia, were either excluded from these trials or were treated off-protocol with liberal transfusions.

The MINT trial is the only randomized study of any significant size to test transfusion targets in patients with acute myocardial infarction. Its few exclusion criteria produced a very ill, real-world population.

In MINT (n=3,500), liberally transfused patients received more than three times as much blood (~2.5 units per patient vs 0.7 per patient). Blood was spared, but the results were not reassuring, safety-wise:

• Recurrent myocardial infarction or death, the composite primary outcome (~17% in restrictive, 14.5% in liberal)

• Death (~10% in restrictive, ~8% in liberal)

• Recurrent MI (8.5% in restrictive, ~7% in liberal)

• Cardiac death (5.5% in restrictive, ~3% in liberal)

Confidence intervals included 1.0, but just barely, suggesting occasional harm.

Guidelines Tilt Toward Liberal Transfusion for Acute MI

Swayed by the MINT data, major professional societies gently nudged clinicians toward more liberal transfusion of patients with acute coronary syndromes:

• AABB recommended transfusing to Hb ≥10 g/dL for hospitalized patients with acute myocardial infarction, in 2025.

• ESICM weakly recommended a target hemoglobin of 9 to 10 g/dL, citing low-certainty evidence.

• A U.S. critical care society recommended against using a restrictive strategy (Hb 7 to 8 g/dL) based on trials using 10 g/dL as the permissive comparator.

A new randomized trial recently added to the body of data on red cell transfusions for patients at high risk for perioperative MIs.

The TOP Trial

TOP enrolled adults at 16 Veterans Affairs medical centers in the United States from 2018 through 2023 who had recently undergone major vascular or general surgery, were anemic (Hb <10g/dL), and at elevated cardiac risk (by RCRI).

More than 1,400 patients were randomized to either transfusion and maintenance of ≥10 g/dL until discharge, or transfusion only when hemoglobin was <7 g/dL.

More than 90% underwent vascular surgery, half had diabetes, and peripheral arterial disease and coronary artery disease were common (i.e., at high risk for periop MIs and strokes).

Transfusion rates (2 units vs 0 median) and hemoglobin separation (~2 g/dL divergence at day 5) were meaningful.

But at 90 days, the composite of all-cause death, myocardial infarction, coronary revascularization, acute kidney failure, or ischemic stroke had occurred almost equally often in 9% of patients in the liberal group vs. 10% in the restrictive group, a nonsignificant difference.*

*To build their power calculations, the authors over-pessimistically assumed the primary outcome would occur in 30% of the restricted patients, and over-optimistically assumed that liberal transfusion would reduce this occurrence by 25% (to 22.5%). This decision had the salutary effect (for the trialists) of drastically reducing the required enrollment to an attainable level, but reduced the TOP trial’s power to detect smaller effects.

Fewer Non-MI Cardiac Complications in the Liberally-Transfused

The liberally transfused patients had a lower rate of cardiac complications other than MI (arrhythmias, heart failure, and nonfatal cardiac arrest): 5.9% with liberal transfusion, 9.9% with restriction. This was a prespecified endpoint and was statistically significant.

Rates of the individual components of the primary outcome (mortality, MI, AKI, stroke) occurred infrequently and did not differ statistically between groups.

Does TOP “Refute” MINT?

TOP was not an acute MI trial, instead enrolling postoperative patients at high cardiac risk, almost none of whom had an active myocardial infarction at enrollment. MINT enrolled patients already experiencing MIs; TOP sought to prevent them in the perioperative period.

TOP’s topline negative finding suggests that high cardiac risk alone isn’t reason enough to routinely transfuse every postoperative patient to hemoglobin ≥10 g/dL.

But nor does it prove that hemoglobin 7 g/dL is always safe in cardiovascularly high-risk surgical populations. The confidence interval leaves room for benefit or harm, and the secondary cardiac-complication signal aligns with MINT.

TOP was negative, but it was also designed with unreasonably high event rate assumptions and was underpowered to detect small but potentially clinically important differences. (It was also stopped slightly early for funding concerns, but even at full enrollment would have remained underpowered.)

Other Trials Didn’t Show Benefits of Prophylactic Transfusions Either

More support for the thesis that every high-risk patient need not be liberally transfused perioperatively comes from the FOCUS trial (Carson et al NEJM 2011). In more than 2,000 older patients undergoing hip fracture surgery who had cardiovascular disease or cardiovascular risk factors, liberal and restrictive transfusion strategies had similar outcomes (although MIs were, again, about an absolute 1% lower in the liberally transfused).

The TRICS-III trial (Mazer et al NEJM 2017) found no significant difference in its composite outcome in 5,243 cardiac bypass surgery patients transfused to either Hb ≥7.5 or ≥10 g/dL (numerically favoring the restricted-transfusion arm by 1%).

Conclusions

For stable postoperative patients with cardiac risk factors but no active ischemia, TOP, FOCUS and TRICS-III support continuing a restrictive approach rather than reflexively transfusing to 10 g/dL.

For a patient with acute MI, MINT and subsequent professional society guidance point toward a higher threshold, perhaps hemoglobin 9 to 10 g/dL, though the evidence remains low certainty.

For the patient in between (postoperative anemia plus severe coronary disease, tachyarrhythmias, heart failure, dynamic troponin elevation, ischemic symptoms, or tenuous hemodynamics), careful individualized decision-making is required. TOP offered something for both the restrictive (virtually identical mortality in both groups) and the liberal clinician (fewer non-MI complications as a secondary outcome).

There is no ongoing trial comparable to MINT testing transfusion in active MI. The TRICS-IV trial is further testing restrictive vs. liberal transfusion strategies in ~1,400 younger adults undergoing cardiac bypass surgeries. Results are expected in 2026.

References

Kougias P, ; TOP Trial Investigators. Liberal or Restrictive Postoperative Transfusion in Patients at High Cardiac Risk: The TOP Randomized Clinical Trial. JAMA. 2025 Dec 23;334(24):2197-2207. doi: 10.1001/jama.2025.20841. PMID: 41205227; PMCID: PMC12596742.

Carson et al. Restrictive versus Liberal Transfusion in Myocardial Infarction — A Patient-Level Meta-Analysis. NEJM Evid 2024

Carson JL, Brooks MM, Abbott JD, Chaitman B, Kelsey SF, Triulzi DJ, Srinivas V, Menegus MA, Marroquin OC, Rao SV, Noveck H, Passano E, Hardison RM, Smitherman T, Vagaonescu T, Wimmer NJ, Williams DO. Liberal versus restrictive transfusion thresholds for patients with symptomatic coronary artery disease. Am Heart J. 2013 Jun;165(6):964-971.e1. doi: 10.1016/j.ahj.2013.03.001. Epub 2013 Apr 8. PMID: 23708168; PMCID: PMC3664840.

Ducrocq G, Gonzalez-Juanatey JR, Puymirat E, Lemesle G, Cachanado M, Durand-Zaleski I, Arnaiz JA, Martínez-Sellés M, Silvain J, Ariza-Solé A, Ferrari E, Calvo G, Danchin N, Avendaño-Solá C, Frenkiel J, Rousseau A, Vicaut E, Simon T, Steg PG; REALITY Investigators. Effect of a Restrictive vs Liberal Blood Transfusion Strategy on Major Cardiovascular Events Among Patients With Acute Myocardial Infarction and Anemia: The REALITY Randomized Clinical Trial. JAMA. 2021 Feb 9;325(6):552-560. doi: 10.1001/jama.2021.0135. PMID: 33560322; PMCID: PMC7873781.

Carson et al. Restrictive or Liberal Transfusion Strategy in Myocardial Infarction and Anemia. MINT Trial. NEJM 2024

FOCUS trial (Carson et al NEJM 2011)

TRICS-III trial (Mazer et al NEJM 2017)

For patients with severe hemorrhage, transfusion with whole blood has theoretical advantages over component blood products: whole blood is a complete, balanced resuscitation fluid that can also reverse coagulopathy with native plasma proteins.

Observational studies both in military and civilian settings have suggested whole blood may be superior to component blood transfusion for hemorrhage.

In forward-deployed military units, whole blood has the advantage of not always requiring refrigeration—it can be drawn from active-duty personnel and transfused warm within hours.

Such a “just-in-time,” on-demand approach to transfusion creates its own logistical challenges, however, and so in U.S. military forward deployments today, warm blood transfusion from matched active-duty donors tends to be paired with stored low-titer group O whole blood, which requires refrigeration but still retains logistical simplicity compared to a component strategy.

The relative simplicity and theoretical physiologic advantages of a whole-blood approach to transfusion led to its expansion into the civilian trauma community in the 2010s.

Prehospital whole blood transfusion has been tested in two randomized trials to date.

SWIFT Trial

The SWIFT trial (Smith et al NEJM 2026) did not demonstrate benefit from transfusing up to two units of whole blood in hemorrhaging civilian trauma victims in the U.K. en route by air ambulance to trauma centers, as compared to transfusion with standard blood components.

Among 616 patients transfused prehospital, those who were randomized to whole blood had numerically equal or higher mortality at all time points measured, from 6 hours up to 90 days after randomization. They also had near-identical rates of massive transfusion as those receiving component blood. Death and massive transfusion were combined into a composite outcome, which did not differ between groups.

TOWAR Trial

In the TOWAR trial, 1,020 traumatically hemorrhaging casualties were transfused with either up to two units of low-titer group O whole blood or component blood products, in 44 air medical bases that initiated the intervention in a cluster-randomized design.

At 30 days, all-cause mortality was numerically higher in the whole-blood group (25.9% vs 20.5%; non-significant, confidence interval 0.87 to 1.76; P=0.24). There was no detectable difference in adverse events between the two groups.

In an observational substudy, the age of the blood (at 1 to 14 days old, vs. 15 to 21 days old) had no association with mortality (an odds ratio of 0.99).

TOWAR was a pragmatic trial and was consequently “messy” to the expected extent, with numerous crossovers in both directions; many patients were transfused before enrollment; coagulation parameters were not obtained.

Is Two Units of Whole Blood Enough? TROOP Trialists Intend to Find Out

In both the SWIFT and TOWAR trials, a maximum of two units of whole blood were transfused prehospital, raising the question of whether this was an adequate dose to produce a benefit.

Total volumes transfused were likewise low. In both SWIFT and TOWAR, 75% of patients were transfused a total of four units of blood or fewer in the first 24 hours.

Importantly, neither trial tested a whole-blood–predominant massive transfusion strategy. Most patients subsequently received conventional component-based resuscitation after hospital arrival.

Resuscitation from traumatic hemorrhage with whole blood is currently under more definitive study in the TROOP trial.

In TROOP, around 1,100 severely injured adults at 15 major U.S. trauma centers will be randomized to resuscitation with either low-titer type O whole blood or component blood products, with a primary outcome of 6-hour mortality and numerous secondary outcomes.

TROOP began enrolling patients in 2023 and is expected to be complete by mid-2027.

The Verdict

Whole blood transfusion for severe hemorrhage doesn’t clearly improve outcomes compared to component transfusion … when only two units of blood are transfused.

Whole blood provides logistical advantages in the military setting (and possibly civilian as well), but any health benefits have yet to be established in a randomized trial.

The TROOP trial will help to establish whether whole blood provides any real physiologic or survival benefits over component blood transfusion.

References

Sperry JL, Guyette FX, Cotton BA, et al. Prehospital Resuscitation with Type O Whole Blood for Trauma and Hemorrhage. New England Journal of Medicine. Published online May 18, 2026. doi:https://doi.org/10.1056/nejmoa2602167

‌Smith JE, Cardigan R, Sanderson E, et al. Prehospital Whole Blood in Traumatic Hemorrhage — a Randomized Controlled Trial. New England Journal of Medicine. Published online March 17, 2026. doi:https://doi.org/10.1056/nejmoa2516043

‌‌Guyette FX, Sperry JL. Prehospital low titer group O whole blood is feasible and safe: Results of a prospective randomized pilot trial. J Trauma Acute Care Surg. 2022 May 1;92(5):839-847. doi: 10.1097/TA.0000000000003551. Epub 2022 Jan 25. PMID: 35081595; PMCID: PMC9038638.

Hazelton JP, et al. Use of Cold-Stored Whole Blood is Associated With Improved Mortality in Hemostatic Resuscitation of Major Bleeding: A Multicenter Study. Ann Surg. 2022 Oct 1;276(4):579-588. doi: 10.1097/SLA.0000000000005603. Epub 2022 Jul 18. PMID: 35848743.

Brill JB, et al . Impact of Incorporating Whole Blood into Hemorrhagic Shock Resuscitation: Analysis of 1,377 Consecutive Trauma Patients Receiving Emergency-Release Uncrossmatched Blood Products. J Am Coll Surg. 2022 Apr 1;234(4):408-418. doi: 10.1097/XCS.0000000000000086. PMID: 35290259.

Dorken-Gallastegi A, et al . Whole Blood and Blood Component Resuscitation in Trauma: Interaction and Association With Mortality. Ann Surg. 2024 Dec 1;280(6):1014-1020. doi: 10.1097/SLA.0000000000006316. Epub 2024 May 6. PMID: 38708894; PMCID: PMC11538373.

Spinella PC, Perkins JG, Grathwohl KW, Beekley AC, Holcomb JB. Warm fresh whole blood is independently associated with improved survival for patients with combat-related traumatic injuries. J Trauma. 2009 Apr;66(4 Suppl):S69-76. doi: 10.1097/TA.0b013e31819d85fb. PMID: 19359973; PMCID: PMC3126655.

Gurney J, Staudt A, Cap A, Shackelford S, Mann-Salinas E, Le T, Nessen S, Spinella P. Improved survival in critically injured combat casualties treated with fresh whole blood by forward surgical teams in Afghanistan. Transfusion. 2020 Jun;60 Suppl 3:S180-S188. doi: 10.1111/trf.15767. Epub 2020 Jun 3. PMID: 32491216.

Nessen SC, et al . Fresh whole blood use by forward surgical teams in Afghanistan is associated with improved survival compared to component therapy without platelets. Transfusion. 2013 Jan;53 Suppl 1:107S-113S. doi: 10.1111/trf.12044. PMID: 23301962.

Torres CM, Kenzik KM, Saillant NN, Scantling DR, Sanchez SE, Brahmbhatt TS, Dechert TA, Sakran JV. Timing to First Whole Blood Transfusion and Survival Following Severe Hemorrhage in Trauma Patients. JAMA Surg. 2024 Apr 1;159(4):374-381. doi: 10.1001/jamasurg.2023.7178. Erratum in: JAMA Surg. 2024 Apr 1;159(4):470. doi: 10.1001/jamasurg.2024.0324. PMID: 38294820; PMCID: PMC10831629.

Hanna K, Bible L, Chehab M, Asmar S, Douglas M, Ditillo M, Castanon L, Tang A, Joseph B. Nationwide analysis of whole blood hemostatic resuscitation in civilian trauma. J Trauma Acute Care Surg. 2020 Aug;89(2):329-335. doi: 10.1097/TA.0000000000002753. PMID: 32744830.

Ibrahim W, Meza Monge K, Menzel J, et al. Whole-Blood vs Component Therapy in Adult Trauma: An Updated Systematic Review and Meta-Analysis. JAMA surgery. Published online November 2026:e260197. doi:https://doi.org/10.1001/jamasurg.2026.0197

Secondary outcomes include 24-h and 30-day or hospital mortality (whichever is earlier); prespecified complications; adjudicated cause of death; time to death; length of stay (ICU and hospital); and hospital-, ventilator- and ICU-free days; the incidence of major surgical procedures; time to hemostasis in those undergoing procedures with a hemostatic component; number and type of blood products used until hemostasis is achieved (and randomized products are discontinued), as well as after hemostasis has been achieved, to 24 h post-admission; discharge destination and functional status and quality of life at hospital discharge or 30 days, as measured by Glasgow Coma Scale (GCS) and EuroQol (EQ-5D) quality of life measurement.

is Trauma resuscitation with Low-Titer Group O Whole Blood Or Products: study protocol for a randomized clinical trial (the TROOP trial)

“MR. JOHNSON! IF YOU KEEP DOING THAT, WE’RE GOING TO HAVE TO TIE YOU DOWN!”

Restraining the wrists of mechanically ventilated patients with cloth wrist cuffs strapped to bedrails has become a standard practice in many, if not most ICUs in the U.S. This has been defended as necessary for patients (for their safety), nurses (overstressed by attending to fidgety patients), or the hospital (reducing liability claims).

None of these purported benefits has ever been demonstrated to be true. Until 2026, the use of physical restraints in critically ill patients had not been examined in any randomized trial.

In small observational studies, most patients who self-extubated were restrained. (This confounded data point doesn’t prove restraints don’t prevent self-extubation, because caregivers may have been more likely to restrain agitated patients they thought might self-extubate, who then did. It doesn’t tell us if or how often restraints prevented self-extubation in the many others who did not succeed at the task.)

Restraints have also been associated with higher sedative use, longer lengths of stay, and delirium or disorientation—all of which likewise have obvious potential for confounding with the need for restraints. A 2021 meta-analysis of low-quality studies suggested an association between restraint use and PTSD.

A mere six decades after mechanical ventilation and physical restraints entered wide use in developed nations’ ICUs, a randomized trial was performed to test whether the practice provided any benefit.

The R2D2 Trial

Among 405 mechanically ventilated adults in 10 French ICUs, patients were randomized to a restrictive strategy (wrist restraints avoided unless severe agitation occurred, defined as RASS ≥3) or a liberal strategy (in which wrist restraints were applied by default and reassessed daily).

In the “low-use” (unrestrained) group, wrist-straps were used on at least one day in 36% of patients (i.e., 64% of vented patients went completely restraint-free throughout their ICU stay). Most of these were from caregiver noncompliance with the protocol — only 3% of patients were recorded as requiring restraints for severe agitation.

There was no difference in the primary endpoint of “days alive without coma or delirium at 14 days”. This is a muddy endpoint to parse in this unblinded, dynamic trial, due to the complex interactions between restraints, agitation, delirium, and sedation levels.

What everyone wants to know instead is, did the unrestrained patients self-extubate more often? They did not: 18 patients self-extubated in the unrestrained group (9.2%) vs 17 in the restrained (8.5%).

Nor did patients remove devices like I.V.s or urinary catheters more often (two devices in the unrestrained group vs. one in the restrained group). They did not have more agitation, either.

The second most salient question: did the restrained patients receive more sedation (to “snow” them pre-emptively or reactively, to reduce the perceived risk for self-extubation or other self-harm)? They did not. (See eFigures 3-7 in the supplementary appendix for the closely comparable dosing of multiple sedatives.)

The unrestrained patients got up and moved ever so slightly more in the first 14 days, but at two weeks, mobilization was nearly identical between groups.

Mortality and 90-day functional, cognitive, and psychological outcomes were also similar. About 4% (absolute) more patients in the restrained group developed PTSD at 90 days, but this was not statistically significant.

Pressure ulcers were reported in 30 unrestrained patients (15.3%) and 34 restrained patients (17.0%), nonsignificant.

But Let’s Also Show Some Restraint

It would be premature to draw broad conclusions about the non-utility (or disutility) of restraints outside the scope of this trial performed exclusively in French ICUs.

Many in the U.S. will raise a skeptical eyebrow at the report that fewer than 1% (only three patients out of ~400) removed their I.V.s, nasogastric tubes, urinary catheters, or other devices. Anecdotally speaking, critically ill patients in many U.S. ICUs seem to be significantly more unruly than this.

The caregivers participating in the trial were doubtless influenced by the knowledge that their behavior was being observed and the outcomes recorded. This may have resulted in extra vigilance toward reduction of adverse events in the intervention (low-restraint) arm.

The accepted way to account for this and other potential secular (over time) changes in clinician behavior is to randomly start the intervention in each of the ICUs over several months or a year, in a cluster-randomized design. That wasn’t done here.

Should Restraints Be So Routine in the ICU?

Per the CMS regulatory code, restraints may only be “imposed to ensure the immediate physical safety of the patient, a staff member, or others,” and never “as a means of coercion, discipline, convenience, or retaliation by staff. [They] must be discontinued at the earliest possible time.”

One hopes that restraints are virtually always imposed with some motivation to ensure ventilated patients’ safety. But there exists an uncomfortably large overlap between a concern for patient safety and a desire to maximize care teams’ convenience.

The R2D2 trial raises real doubts about whether restraints are required to maintain the safety of most mechanically ventilated patients.

Hundreds of thousands of people are physically restrained during their ICU stays every year around the world, without any good evidence to support the practice. A meta-analysis (albeit of low-quality studies) suggested that restraint use may be associated with the development of PTSD.

Now that the benefits for restraints have been called into question, conducting a properly powered cluster-randomized trial (in thousands of patients) would seem to be the bare minimum we should do to ethically justify this common but often disturbing and conflicted act of care.

References

Restrictive vs Liberal Physical Restraint Strategies in Critically Ill Patients. The Journal of the American Medical Association. 2026. Sonneville R, Couffignal C, Sigaud F, et al.

Clinical Practice Guidelines for the Prevention and Management of Pain, Agitation/Sedation, Delirium, Immobility, and Sleep Disruption in Adult Patients in the ICU. Critical Care Medicine. 2018. Devlin JW, Skrobik Y, Gélinas C, et al.

Physical Restraints and Post-Traumatic Stress Disorder in Survivors of Critical Illness. A Systematic Review and Meta-Analysis. Annals of the American Thoracic Society. 2021. Franks ZM, Alcock JA, Lam T, et al.

Unplanned Endotracheal Extubations in the Intensive Care Unit: Systematic Review, Critical Appraisal, and Evidence-Based Recommendations. Anesthesia and Analgesia. 2012. da Silva PS, Fonseca MC.

Non-Pharmacological Interventions for Minimizing Physical Restraints Use in Intensive Care Units: An Umbrella Review. Frontiers in Medicine. 2022. Cui N, Yan X, Zhang Y, et al.

Revisiting Unplanned Endotracheal Extubation and Disease Severity in Intensive Care Units. PloS One. 2015. Chuang ML, Lee CY, Chen YF, Huang SF, Lin IF.

Critical care nurses’ experiences of physical restraint in intensive care units: A qualitative systematic review and meta‐synthesis. Journal of Clinical Nursing. 2023. Lao Y, Chen X, Zhang Y, et al.SR

CMS regulatory code

A 72-year-old man who resides in a nursing home is brought to the ED because of worsening lethargy and fevers. He has a history of recurrent urinary tract infections that is presumed due to prostatic hypertrophy, and was admitted with sepsis attributed to a UTI last month.

You review his urine culture data for the E. coli isolated on his admission last month:

MRSA PCR screen from a nasal swab was negative at that time.

He is tachycardic, febrile, and has borderline hypotension. He is confused compared to his baseline. WBC is 20K, and creatinine is 2.9. In the ED, he is diagnosed with sepsis and provided a liter of I.V. crystalloid, piperacillin-tazobactam 3.5 g, and vancomycin 1.5 g I.V.

A CT of the abdomen shows no hydronephrosis or calculi.

Extended-spectrum beta-lactamase (ESBL) is an enzyme that can crack open the four-membered β-lactam ring, rendering the antibiotic ineffective.

ESBLs are encoded by genes located on large transferable modules (plasmids) that are easily shared between bacterial populations, most often Escherichia coli, Klebsiella pneumoniae, or Proteus mirabilis.

Infections by ESBL-producing gram-negative bacilli are notoriously lethal in the ICU, but as they tend to afflict the most vulnerable patients with longer ICU stays, more severe illness, and higher expected mortality, and also may often be present as colonizers alongside pathogenic bugs, their independent contribution to mortality is difficult to tease out precisely.

Generally speaking, urinary infections with ESBL-producing organisms are less often fatal than respiratory infections or primary bacteremias. Klebsiella is usually more virulent than E. coli. Septic shock with ESBL organisms has a higher mortality rate than with sensitive bacteria.

Microbiology labs may test for and report ESBL production, but it should be assumed anytime a gram-negative rod on the “usual suspects” list (E. coli, Klebsiella pneumoniae, Klebsiella oxytoca, and Proteus mirabilis) shows resistance to ceftriaxone (and often ceftazidime and aztreonam) but susceptibility to carbapenems.

Ceftriaxone resistance in Enterobacter cloacae, Klebsiella aerogenes, Citrobacter freundii, Serratia, etc., often suggests AmpC, not necessarily ESBL.

Narrower resistance (e.g., to ampicillin and/or cefazolin) does not imply ESBL, as many E. coli isolates are resistant to these agents.

Failure to adequately cover ESBL early is considered a likely major contributor to the excess mortality. As such, early appropriate antibiotic coverage for patients at risk is vital.

Meropenem is recommended for critically ill patients at high risk for ESBL organisms. Ertapenem is also effective, but does not cover Pseudomonas, which is often a concern in severe infections. Meropenem’s dose is reduced for renal impairment.

Although beta-lactamase inhibitors like tazobactam do thwart the ESBL enzyme in many patients, tazobactam has an unpredictable failure rate in ESBL-producing Enterobacteriaceae (aka Enterobacterales) in serious infections, especially bacteremia, sepsis, pneumonia, or ICU-level illness.

Gentamicin is considered an effective and appropriate agent against ESBL, but due to its risks for nephrotoxicity and ototoxicity, is considered second-line to meropenem for severe ICU-level infections.

Multiple newer β-lactam + β-lactamase inhibitor combination agents have entered the U.S. market in recent years. Ceftazidime-avibactam, meropenem-vaborbactam, imipenem-cilastatin-relebactam, ceftolozane-tazobactam, and cefiderocol all exhibit activity against ESBL organisms. These are generally reserved for carbapenem-resistant organisms to preserve their utility.

Oral nitrofurantoin or TMP-SMX may be a good option for uncomplicated ESBL UTIs, but are not recommended for severe infections. Fosfomycin is not recommended for K. pneumoniae or other organisms that may carry fosA genes, after a randomized trial found single-dose fosfomycin had more clinical failures than a 5-day course of nitrofurantoin.

Reflect to earn CME with Learner+

Sample reflection: I reviewed and reflected on infections due to known or suspected extended-spectrum beta-lactamase-producing gram-negative bacilli in the ICU, and the considerations for appropriate antimicrobial therapy.

References

Infectious Diseases Society of America 2023 Guidance on the Treatment of Antimicrobial Resistant Gram-Negative Infections. Clinical Infectious Diseases : An Official Publication of the Infectious Diseases Society of America. 2023. Tamma PD, Aitken SL, Bonomo RA, et al.

Infectious Diseases Society of America 2024 Guidance on the Treatment of Antimicrobial-Resistant Gram-Negative Infections. Clinical Infectious Diseases : An Official Publication of the Infectious Diseases Society of America. 2024. Tamma PD, Heil EL, Justo JA, et al.

Association Between Minimum Inhibitory Concentration, Beta-Lactamase Genes and Mortality for Patients Treated With Piperacillin/­Tazobactam or Meropenem From the MERINO Study. Clinical Infectious Diseases : An Official Publication of the Infectious Diseases Society of America. 2021. Henderson A, Paterson DL, Chatfield MD, et al.

Ceftolozane-Tazobactam Versus Meropenem for Definitive Treatment of Bloodstream Infection Due to Extended-Spectrum Beta-Lactamase (ESBL) and AmpC-producing Enterobacterales (”MERINO-3”): Study Protocol for a Multicentre, Open-Label Randomised Non-Inferiority Trial. Trials. 2021. Stewart AG, Harris PNA, Chatfield MD, Littleford R, Paterson DL.

Multidrug-Resistant Gram-Negative Bacterial Infections. Lancet. 2025. Macesic N, Uhlemann AC, Peleg AY.

Extended-Spectrum Β-Lactamase-Producing Enterobacteriaceae: Update on Molecular Epidemiology and Treatment Options. Drugs. 2019. Peirano G, Pitout JDD.

Carbapenem Use in Extended-Spectrum Cephalosporin-Resistant Enterobacterales Infections in US Hospitals and Influence of IDSA Guidance: A Retrospective Cohort Study. The Lancet. Infectious Diseases. 2024. Walker MK, Diao G, Warner S, et al.

Preserving the Antimicrobial Arsenal: Exploring Alternatives to Carbapenems in ESBL Battles Within the Southeast of Ireland. Journal of Medical Microbiology. 2025. Ali S, Tobin A, Lapthorne S, et al.