Robotic bronchoscopy against CT-guided lung biopsy: the yield is the same, the pleura is not

Design

Systematic review and meta-analysis restricted to direct comparisons, ROBINS-I and GRADE, with a cohort-genealogy step to exclude patient overlap

Population

Adults with peripheral pulmonary lesions across five retrospective studies from one US health system; at most three cohorts mutually independent

Primary outcome

Strict 2024 ATS/CHEST diagnostic yield

Effect

Yield RR 0.99 (95% CI 0.93 to 1.06, I² 24%); pneumothorax needing drain or admission RR 0.25 (0.14 to 0.46, I² 0%)

This is the first systematic review restricted to studies that compared robotic-assisted bronchoscopy and CT-guided transthoracic biopsy head to head in the same institution. Five retrospective studies qualified, all from a single US health system, and four of them shared patients — so the authors added a cohort-genealogy step and pooled only the largest mutually independent set, which was three cohorts. Diagnostic yield under the strict 2024 American Thoracic Society and CHEST definition was the same, risk ratio 0.99 (95% CI 0.93 to 1.06, I² 24%). Pneumothorax requiring a chest drain or admission was far less common with the robotic approach, RR 0.25 (0.14 to 0.46, I² 0%). Procedure time was about 50 minutes longer for robotic bronchoscopy where staging endobronchial ultrasound in the same session was counted in its time, and about 8 minutes longer where it was not.

The methodological care here is unusual and worth noting, because it works against the headline. Most reviews would have pooled all five cohorts and reported a larger, spuriously precise result; excluding overlapping patients is why the numbers are believable. The authors are equally direct about what remains: certainty is low for pleural complications and very low for everything else, the evidence is retrospective and from one system, and no cohort reported yield by bronchus sign or lung zone — which is precisely the stratification that would tell you which lesion belongs with which technique.

The practical reading is about who a lesion gets sent to. Where both services exist, a peripheral lesion in a patient with emphysema, poor reserve, or a previous pneumothorax is the one where a fourfold difference in drain-requiring pneumothorax should weigh. Where robotic bronchoscopy does not exist — which is most of India — nothing here argues against CT-guided biopsy, whose yield was no worse. The absolute yield figures are the other thing to carry: applying the strict definition dropped yield from 88% to between 74% and 84%, for both techniques.

• Weigh the pleural complication difference most in patients with emphysema, poor reserve or previous pneumothorax.
• Quote strict-definition yield, 74% to 84%, not the 88% that looser criteria produce.
• Do not read this as equivalence — it was not formally tested, and certainty was low to very low.
• Ask whether the robotic time quoted locally includes same-session staging endobronchial ultrasound.
• Where robotic bronchoscopy is unavailable, CT-guided biopsy yield was no worse in these comparisons.

The statistics, in plain English

A risk ratio of 0.99 with an interval of 0.93 to 1.06 is a tight null: the data exclude any difference in yield larger than about 7% in either direction. That is not the same as proving equivalence, which requires a prespecified margin, and the authors say so. The pneumothorax result, 0.25 with an interval of 0.14 to 0.46, is large and consistent — but note what the Hartung-Knapp correction does to it, widening the interval to 0.07 to 0.96, because three studies is very few to estimate between-study variance from. Two other outcomes were reported by only two cohorts each and did not survive that correction at all. The pattern to take away is that the effect is probably real and its size is not yet knowable.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.