Good morning! Welcome to another edition of Pressure Support, a weekly newsletter simplifying pulmonary and critical care medicine—and the systems that shape it.

I was swamped in the MICU this week, but had some new ideas regarding how to streamline the pre-rounding/handoff process in the ICU.

If you have colleagues who’d benefit from this newsletter, share it with them using the link below:

I’m alpha testing this newsletter—so please let me know your feedback.

As a reminder, here’s how the newsletter will be structured:

September 24th, 2026

I’ve written about the sign-out process before (back in October 2023… three years ago!). At the time, I focused on how a sign-out that should take 15 minutes can easily turn into 30 minutes because we communicate too much information without a consistent structure.

Now that I’m in the ICU, I’ve started thinking about the problem a little differently.

Every morning, I review the charts of every patient in the unit. As the fellow, I want to know what happened overnight, how the clinical picture is changing, and what decisions we may need to make that day. I also want to be prepared to contribute during rounds, catch anything that may have been missed, and teach when the opportunity comes up.

All of that preparation takes time… a lot. I go through the overnight notes, recent labs, imaging, procedures, medication changes, and whatever else happened while the day team was gone. Then the actual formal sign-out starts when I arrive on the floor.

The night team walks through the same patients. I hear the updated one-liner, what happened overnight, the active problems, and what I should watch for during the day. I listen, ask questions, and jot down anything important.

An hour later, I review much of the same information again during rounds.

Each part of this process serves a purpose:

The inefficiency comes from how often we reconstruct and repeat the same patient story. The chart contains the information, but it’s scattered across progress notes, laboratory results, imaging reports, procedure notes, medication records, and flowsheets. Someone has to pull those pieces together into a concise narrative before the rest of the team can use them. Right now, that work is largely manual.

I often wake up earlier or get to the hospital sooner so I can review the unit before sign-out. A 12-hour ICU shift may easily become 13 hours when a patient decompensates or a procedure comes up near shift change. Adding another 30 to 60 minutes of pre-charting makes an already long day even longer.

There is also a cognitive cost. When I’m rapidly reviewing more than a dozen charts, I have to decide which details matter, which changes are new, and which problems require attention that day. Then I compare my understanding with what I hear from the night team and what the residents present on rounds.

That redundancy can help us catch errors (shout out Swiss Cheese model). But it can also consume a large amount of physician time because every clinician is independently rebuilding a version of the same clinical summary.

The problem: The current sign-out and pre-charting process requires clinicians to repeatedly reconstruct the same patient story, creating unnecessary work during an already high-risk transition.

Generative AI can already take a progress note or H&P and turn it into a useful one-liner and hospital-course summary. I use it occasionally when I’m covering a new consult service and need to understand a patient quickly.

The more useful version would connect directly to the EHR, review the full chart, and create a short audio briefing for every patient on the service before sign-out. Think of it as an audio version of pre-charting.

For each patient, the AI would pull together:

The output should sound like a concise patient presentation rather than someone reading the chart aloud. Each briefing could follow the same structure clinicians already use during sign-out and rounds, with the most important information placed first.

The one-liner would also update each day. A patient who arrived with septic shock, required vasopressors, and was intubated should not have the same one-liner after they are extubated, off pressors, and preparing to leave the ICU. The audio briefing should reflect the current clinical picture while preserving enough of the hospital course to explain how the patient got there.

For me, this would fit naturally into the start of the day. I usually have a 20-minute walk to the hospital. During that time, I could listen to briefings on 12 to 14 patients at 1.5x speed. By the time I arrive, I would already know the basic story, what changed overnight, and which patients deserve a closer chart review.

I would still look through the chart myself. The difference is that I could spend that time verifying the important details and thinking through the plan instead of starting from a blank screen and reconstructing every hospital course manually.

Then, during the actual in-person sign-out, the night team would not need to spend as much time rebuilding the basic patient narrative. We could focus on the details that require human context: what worried the team overnight, what remains uncertain, which decisions are pending, and what the incoming team should be prepared to do if the patient worsens.

The same briefing could help prepare residents, fellows, attendings, nurses, and other clinicians without asking each person to independently assemble another version of the same story.

The goal is to make the live handoff more focused and useful. Sign-out carries information that may never appear cleanly in the chart. The outgoing clinician knows which patient looked worse at the bedside, which family conversation changed the plan, and which pending issue is most likely to become a problem in the middle of the night. That judgment still needs to move directly from one clinician to another.

But AI can handle more of the repetitive synthesis before the conversation starts. It can assemble the patient story, update the relevant data, and deliver it in a format clinicians can absorb while commuting or preparing for a shift.

That would leave the actual sign-out for the work that requires a human: uncertainty, prioritization, accountability, and clinical judgment.

The information already exists, and generative AI can already summarize it. The next step is turning that capability into a secure, EHR-integrated audio workflow that helps clinicians arrive prepared without adding another screen, another note, or another hour to the day.

Reply back if you want to build this.

This AJRCCM editorial argues that bronchiectasis exacerbations have a “kindling” effect—each exacerbation (especially severe ones) increases the risk of future exacerbations in a dose-dependent way, as shown in Sibila et al.’s work across diverse populations. This is a good reminder to treat every flare like a chance to break the cycle. Optimize airway clearance, nail down triggers/infection management, and prevent the next hit rather than just putting out today’s fire.

This AJRCCM editorial argues that reduced COPD hospitalization rates—highlighted by the 50–60% drop seen during the COVID-19 era—are a meaningful, patient-centered outcome that should be prioritized when judging interventions and systems of care. I like the focus on hospitalization as a hard endpoint, but it also pushes us to invest in the everyday workflow/process fixes (prevention, access, rapid outpatient rescue) that actually make those reductions sustainable.

This NEJM piece warns that new “vape-to-earn” smart vaping devices can literally pay people in crypto based on use/engagement, which is a pretty alarming mash‑up of addictive nicotine delivery plus behavioral/financial incentives. The authors argue this could turbocharge uptake (especially among young people) and that regulators should scrutinize and modernize oversight before these products scale. This is exactly the kind of perverse incentive we need to get ahead of in pulmonary/critical care—once it spreads, we’ll be stuck dealing with downstream lung injury and nicotine dependence while the system plays catch‑up.

This JAMA Network Open cross-sectional study looks at whether physicians in clinician shortage areas who feel more integrated into their communities—and feel their community needs are being met—are more likely to anticipate staying (retention), with separate analyses for rural vs urban settings. This is a helpful reminder that “retention” isn’t just a compensation problem. We probably need to invest in onboarding, community connection, and practical life supports for clinicians if we want to keep people in the places that need them most.

This JAMA pulmonary medicine piece argues that while GLP-1 weight-loss meds can improve OSA metrics (mostly by helping people lose weight), they’re not the most effective treatment—CPAP remains the gold standard for meaningful, reliable control of obstructive events. I like the pragmatic framing here: GLP-1s feel best as an adjunct (and a great obesity/OSA comorbidity tool) rather than a replacement, and it’s a good reminder not to let shiny new pharmacotherapy distract us from getting patients onto therapies that actually work night-to-night.

Randomized, double-blind, placebo-controlled trial in patients with idiopathic pulmonary fibrosis and chronic cough found gabapentin (titrated up to 900 mg/day over 12 weeks) significantly improved cough symptoms and cough-related quality of life versus placebo, with secondary measures like LCQ/VAS also favoring treatment. Side effects were more common than placebo—mostly drowsiness, fatigue, and dizziness—but were generally tolerable and no serious safety signals were reported. If the benefit holds up in broader practice, gabapentin could be a really practical, low-cost option for a symptom that drives patients crazy, as long as we’re thoughtful about sedation and fall risk in the patients we’re treating.

This ATS Scholar describes an immersive virtual-reality simulation curriculum for managing tracheostomy emergencies, tested as a randomized educational intervention and found to be feasible and well‑received by learners. This could be a really practical, scalable way to get reps on rare-but-high-stakes trach scenarios without tying up sim center resources, and I’d love to see whether it translates into better real-world performance (and patient outcomes) over time.

This CHEST cohort analysis of KOCOSS suggests that even a single moderate COPD exacerbation in year 1 is a real prognostic signal—patients had higher risk of future moderate-to-severe and severe exacerbations and higher 12-year all-cause and respiratory mortality, supporting GOLD 2026’s move to treat 1 moderate exacerbation as “high risk” (group E). My take is this is a nice, practical nudge to take “just one” exacerbation seriously and tighten follow-up/optimization early, even if it feels like a small event in the moment.

Let me know your feedback!

Keeping it simple,

Jared