||5R01CA207375-04 Interpret this number
||University Of Washington
||The Effectiveness, Safety, and Costs of Guideline-Concordant Lung Nodule Care
Each year 1.5 million people are diagnosed with an incidentally (not screen) detected lung nodule. The
diagnosis is important because it may represent an early-stage lung cancer, but 90% of incidentally detected
lung nodules are benign. Accordingly, the intensity of a lung nodule evaluation must weigh the benefits of
early-detection and treatment (e.g. cure) against the risks of diagnostic tests (e.g. radiation exposure,
procedure-related adverse events). Practice guideline recommendations are intended to optimize risks and
benefits but adherence rates are only 55%. One reason for this poor adherence is the low level of evidence
supporting guidelines leading to legitimate uncertainty about their effectiveness, safety, and impact on health
care resources. This uncertainty is significant because it is unclear whether interventions should be developed
to increase guideline adherence or if new approaches to varying the intensity of a nodule evaluation are
needed. Until recently, an important barrier to generating a higher level of evidence has been an inability to
identify and longitudinally follow a cohort of individuals with an incidentally detected lung nodule. Investigators
from the Cancer Research Network (CRN) have determined that individuals with an incidentally detected lung
nodule can be efficiently and accurately identified using a combination of administrative data, electronic
radiology reports, natural language processing, and some chart abstraction. Additionally, investigators found
that providers routinely document lung cancer risk factors (e.g. age and smoking status) that allow for a study
of adherence to Fleischner Society guidelines, and this information is readily available in a structured format
within the electronic medical record. Furthermore, other CRN investigators have developed methods that allow
for estimation of radiation exposure from imaging studies and the costs of care delivery within integrated
healthcare systems. This work allows us to propose the first-ever multi-site comparative-effectiveness study of
individuals with an incidentally detected lung nodule diagnosed between 2005 and 2015. The study aims to
compare the: 1) effectiveness (e.g. incidence of early-stage lung cancer), 2) potential harms (e.g. radiation
exposure, procedure-related adverse events), and 3) two-year total costs of care of varying intensities of lung
nodule evaluation (e.g. guideline concordant versus more intense versus less intense evaluations). We
hypothesize that less intense nodule evaluations are associated with a lower incidence of early-stage lung
cancer compared to guideline concordant care, and more intense nodule evaluations are associated with
greater radiation exposure, more procedure-related adverse events, and higher costs. Findings from this study
will determine whether limited resources should be invested in developing system-level interventions designed
to increase guideline adherence or studying alternative approaches to lung nodule evaluation (e.g. risk-
prediction models, biomarkers). This line of investigation is expected to ultimately improve the care and
outcomes of individuals with an incidentally detected lung nodule.
Pathways to breast cancer screening artificial intelligence algorithm validation.
, Houssami N.
, Elmore J.G.
, Buist D.S.M.
Breast (Edinburgh, Scotland), 2020 Aug; 52, p. 146-149.