Question 4 from the second paper of 2014 and the very similar Question 27 from the first paper of 2005 interrogate the candidates' understanding of the indications for a lung biopsy, and the possible complications of such a course of action. It is the investigation of last resort when you cannot arrive at a diagnosis of a diffuse interstitial infiltrate, or other stubborn CXR blemish.
The LITFL lung biopsy page is a definitive resource for the time-poor exam candidate.
Generally speaking, one resorts to an open lung biopsy when one cannot arrive at the diagnosis, and when the differentials include possibilities with radically different (and potentially mutually exclusive) courses of treatment, for instance steroids versus antibiotics.
The biopsy must be performed in several regions of the lung, and must yield specimens which offer a representative sample, without sampling any areas of irreversible fibrosis or uninformative necrosis. It cannot be performed in patients who cannot be ventilated on one lung for prolonged periods. Risks and contraindications of of thoracotomy apply.
In their answer to Question 4 from the second paper of 2014, the college mention some numbers from "multiple case series in the literature". It would be lovely if they gave a reference. Those multiple case series are quoted as follows:
Even in 1976 (Hill et al) the mortality rate from this procedure was zero, and the morbidity rate was around 4%. In that ancient case series, 33% of the patients enjoyed some sort of positive change in their management because of the biopsy result.
Flabouris and Myburgh (1999) reported something similar. Their complication rate was higher (21%) but these were "proper" mechanically ventilated ICU patients. "Open lung biopsy-guided alteration of therapy directly benefited 39%, and withdrawal was possible in 8.4% of the patients". However, the change to management did not discriminate survivors from non-survivors (i.e. it didn't matter that you changed to an appropriate therapy, the patient died anyway).
A retrospective series by Lim et al (2007) also reported that a specific diagnosis was achieved in 86% of the biopsied patients, and that in 64% changes to management occurred. Those who were biopsied earlier (within 1 week of intubation) did better in terms of mortality (63% survival vs 11%), which contrasts with the earlier studies.