Should we perform bronchoscopy to augment pulmonary toilet?

The scene:

Rounds are proceeding. The team arrives at the patient’s room upon a flotilla of computers.

“Look at that pO2,” laments someone in a white coat. “How’s the chest xray?”

“Bad,” replies another. They all gather around to look. Indeed, a portion of the patient’s left lung is whited out by infiltrates.

“Well,” remarks the intern, “should we… bronch him?”

There is a pause. Then the attending shrugs. “Meh.”

An incisive summary. Let us unpack it.

Bronchoscopy to suck out pus

The concept is sound. Funky mucoid pulmonary exudates, whether infectious (i.e. pneumonia) or non-infectious (copious secretions for whatever reason—which have a habit of becoming infected if left to dwell) need to exit the lungs. This is both because they impair oxygen exchange by creating regions of shunt, and because they can induce or prolong pneumonia if “source control” is not obtained by de-consolidating the chest film.

Toward this end, we grant that passing a suction catheter down the endotracheal tube, or even through the nose in the non-intubated patient (much to their consternation), is a laudable intervention. Chest physiotherapy, where patients are thumped and pounded and rotated to loosen secretions, is probably worth the candle as well. (Perhaps more on the interesting topic of chest PT another time.) Once extubated, we encourage and insist that patients perform grueling regimens of deep breathing, incentive spirometry, tooting on flutter valves, and most of all the simple expediency of coughing up their crud. In short, pulmonary toilet (aka pulmonary hygiene) is a simple, non-glamorous, but essential cornerstone of improving pulmonary function and preventing complications.

But if we endorse blindly passing a catheter into the airway to suction mucus, why not kick it up a notch by doing it under guidance and visualization? That is, by pushing down a flexible bronchoscope, finding the junk, and sucking it out directly?

It makes sense, and has made sense ever since the flexible bronchoscope became standard ICU paraphernalia. (With rigid bronchoscopy alone, the smaller airways were usually inaccessible, and the whole enterprise was generally more challenging and traumatic.) Since the 1970s (12), intensivists and eager house staff have been sucking out pus to reexpand lungs. But does it work?

Why might it not work? The most obvious argument is that mucus and proteinaceous crud is not present in the airway because it fell off the back of a turnip truck, or because somebody won it in a sweepstakes. It is generated by an underlying cause, whether bacterial pneumonia, direct lung trauma, an extrapulmonary SIRS state, or simply poor pulmonary clearance. Whatever it may be, none of these mechanisms are eliminated by bronchoscopy. Hence, though pus you may suck, likely it will return. It’s like treating bleeding by wiping away the blood.

That being said, sometimes symptomatic care is appropriate to keep a patient on smooth seas while the underlying pathology is addressed. (After all, mechanical ventilation doesn’t usually “treat the underlying problem” either.)

We therefore find ourselves in a state of equipoise. Authoritative sources (Irwin and Rippe 2011, Parrillo and Dellinger 2013, Dr. UpToDate) give the facts and “teach the controversy,” declining to give a recommendation on this practice. The Society of Critical Care Medicine, American College of Chest Physicians, and Eastern Association for the Surgery of Trauma have no comment on the matter.

What do we know?

The evidence

The data is—like a vegetarian living in the deep South—rather slim.

The only RCT was by Marini et al. in 1979 (yes, nearly 40 years ago; an alarming amount of the relevant literature is from the ’70s). They enrolled 31 patients from their single ICU with lobar atelectasis, randomizing them to receive either chest PT maneuvers alone, or chest PT plus bronchoscopy. In both groups, about the same number of patients—around 38%—reexpanded their atelectatic lung after therapy. Likewise, at 24 and 48 hour follow-up there was no difference in atelectasis. They even crossed over the non-bronch group to receive a delayed bronchoscopy if they had not improved at 24 hours, with rather minimal effect (only 11% had a therapeutic result).

Notably, the Marini study also found that the presence of air bronchograms in the area of atelectasis—suggesting that larger airways were open and the obstruction was more distal—was predictive of a longer time to resolution. Presumably this is because it’s harder to get rid of diffuse distal crud than one big proximal mucus plug . One wonders whether this distinction might help guide bronchoscopy, but Marini actually examined this question (albeit with an underpowered subgroup analysis), finding that the absence of air bronchograms did predict a therapeutic result from bronchoscopy… but also predicted successful resolution with chest PT alone, with no added benefit from the bronch even in that subgroup.

Clearly, most atelectasis is not due to a proximal mucus plug with clear airways beyond, just waiting to be ventilated; often the problem is distal in the pulmonary tree, involving fluid, diffuse exudates, chest wall and diaphragm expansion, and other factors. And it seems that no matter how we manage them, diffuse distal plugging simply takes longer to resolve than more proximal obstructions.

Weinstein et al. (1977) further explored that question, looking specifically at 43 bronchoscopies that performed lavage of the distal airways when there was no proximal mucous obstruction noted. (They do report—and seemed to rather enjoy—retrieving junky sediment, sometimes containing “small mucous plugs or casts of the smaller airways.”) Despite some initial post-bronch worsening in oxygenation, more than half of their patients showed improvements in both pO2 and compliance after 7–12 hours. Of course, many patients have improvements over the course of a day, so without a control group it’s hard to make much of this; plus, pO2 and static compliance are not exactly patient-oriented outcomes.

Numerous non-controlled case series from the 1970 and 1980s report similar results. Snow et al. prospectively considered 67 bronch cases from a SICU, performed for various indications (mostly lobar collapse and suspected aspiration). In their cohort, 58% had radiographic improvement overall; in general they found more benefit for patients with fully lobar atelectasis versus smaller infiltrates or collapse. (That being said, there was actually no significant difference in post-bronch A-a gradient between the two groups.) Wanner et al (1973) described 37 similar cases, with 79% experiencing clinical improvement and 85% experiencing radiologic improvement. Lindholm et al (1974) described 53 cases, with radiographic improvements in 67%. Barrett (1978) reported a large number of cases, and in “all cases [where secretions were cleared from the airway],” suctioning resulted in radiographic improvement and improved oxygenation.

Further reading: Kreider et al (2003) and Jelic et al (2008), the two most current reviews on this topic.


This somewhat less-than-satisfying body of literature has led to the current state of affairs, where bronchoscopy for pulmonary toilet is performed largely in accordance with the provider’s personal training and preferences, perhaps influenced in many cases by ignoble considerations like time available in the shift, how much the house staff wants to practice bronchoscopy, and who might be able to bill.

My takeaway from the data available is this:

  1. Bronchoscopy willy-nilly for everyone with pulmonary infiltrates may improve their oxygenation, but is probably not worthwhile.
  2. Bronchoscopy in patients you suspect (i.e. by radiographic criteria) have a proximal mucus plug has a reasonable chance of opening the atelectatic portion of the lung. If their atelectasis has created intractable hypoxemia, this should probably be done. However, the effect may not last, and in any case may not be any better than other pulmonary hygiene maneuvers (chest PT, etc); there is no clear evidence that bronchoscopy will ultimately benefit anyone.
  3. In general, atelectasis is a very broad and heterogeneous disease, and the patients who experience it are very different. Broadly asking whether bronchoscopy for pulmonary toilet is good or bad is probably a silly question. Limiting it to “bronchoscopy for large-scale atelectasis, such as an entire lung or lobe” is a better question, but still may be too broad. We should instead be asking “what [patient-oriented outcome] do we hope to achieve by bronchoscopy today in this particular patient, and what are the risks?” For example, if your most realistic hope is to reduce the patient’s FiO2 from .6 to .4 within the next several hours, that may be achievable. However, it may or may not be worthwhile, even if most people would consider it a therapeutic result.
  4. Thus, in individual cases, performing a fairly safe procedure to effect a reasonable (if perhaps transient) improvement in oxygenation and reduction in bioburden might make sense. In other cases—maybe in most—it does not.