Diagnostic Yield: Apples, Oranges, and Paradigm
Incommensurability
WABIP Newsletter
S E P T E M B E R 2 0 2 2 V O L U M E 1 0 , I S S U E 3
EXECUTIVE BOARD
Hideo Saka, MD
Japan, Chair
Stefano Gasparini,
MD
Italy, Vice-Chair
Silvia Quadrelli, MD
Argenna, Immediate
Past-Chair
David Fielding MD
Australia, Treasurer
Naofumi Shinagawa,
MD
Japan,
Secretary General
Philippe Astoul, MD
France, President
WCBIP 2022
Menaldi Rasmin, MD
Indonesia, President
WCBIP 2024
STAFF
Michael Mendoza
General Manager
Judy McConnell
Administrator
Kazuhiro Yasufuku
Newsleer Editor-in-
chief
P A G E 2
In the past two decades, unprecedented advances in
technology have led to the commercializaon of
new bronchoscopy plaorms which have been wide-
ly adopted as standard of care procedures by inter-
venonal pulmonologists around the globe. In the
US and in Europe, these novel medical devices were
cleared by regulatory stakeholders in the absence of
comparave eecveness studies, based on the
assumpon that similaries in form and funcon to
already available devices obviated the need for such
data. Thus, most studies published to date reporng
on the diagnosc test performance of these novel
devices have been single-arm observaonal studies
using the surrogate endpoint of diagnosc yield
(DY). (1)However, DY captures more than the tech-
nical ability of these devices to navigate to the tar-
geted lesion, and is aected by many other relevant
variables, which include targeted nodule characteris-
cs, operator skills and cancer prevalence, among
others. Perhaps more importantly, DY has been de-
ned in many dierent ways in studies published to
date which further complicate comparisons across
single-arm studies.
Diagnosc yield (DY) is strictly dened as the likeli-
hood that a biopsy will provide a histopathological
diagnosis sucient to guide management with con-
dence. Typically, this would include malignant diag-
noses, usually, although not always, uncontroversial,
and benign diagnoses deemed specic enough to
convince the pathologist that the lesion in queson
was indeed adequately sampled. These would in-
clude granulomatous inammaon, robust neutro-
philic inltraon, or hamartomas among others.
Diagnosc yield is oen confused with diagnosc
accuracy, which represents the degree to which histopatho-
logical diagnosis reects the ground truth.(2) Inammaon,
while non-specic and thus excluded from the prior deni-
on, could in fact represent lesional ssue, but oen re-
main inconclusive at the me of diagnosis, leading to subse-
quent intervenons. Thus, diagnosc accuracy requires
sucient clinical follow-up to adjudicate non-specic be-
nign diagnoses. Studies on diagnosc test performance of
bronchoscopy have oen conated these two noons and
used them interchangeably, leading to incommensurable
esmates of diagnosc test performance even when ac-
counng for dierences in paent populaon. We have
been comparing apples to oranges.
The impact of these dierent denions on DY was recently
evaluated in a Monte-Carlo simulaon by Vachani and col-
leagues. In “The Impact of Alternave Approaches to Diag-
nosc Yield Calculaon in Studies of Bronchoscopy”, pub-
lished in CHEST in 2022, a hypothecal cohort of 1000 pa-
ents undergoing diagnosc bronchoscopy was assessed
using 3 commonly used but disnct methods for DY calcula-
on: (1) a strict approach, which corresponds to actual DY
as dened above, (2) an intermediate approach, in which
non-specic benign diagnoses were considered diagnosc if
clinical follow-up did not reveal malignancy, and (3) a liberal
approach, according to which even non-diagnosc samples
(such as normal lung or pleura) were considered diagnosc
if follow-up did not reveal malignancy. Expectedly, es-
mates of “DY” varied considerably, ranging from 67% to
88%, a 21% absolute dierence simply due to an arbitrarily
chosen method of reporng the data.(3)
The implicaon of these observaons is obvious: studies
using dierent denions of DY are dicult, if not impossi-
ble to compare. Denions of DY should be explicitly and
Crisna D. Salmon, MD
Department of Allergy, Pulmonary and
Crical Care Medicine,
Vanderbilt University Medical Center
Fabien Maldonado, MD, MSc
Department of Allergy, Pulmonary and
Crical Care Medicine,
Vanderbilt University Medical Center