Validation of prediction models for risk stratification of incidentally detected pulmonary subsolid nodules: a retrospective cohort study in a Korean tertiary medical centre
Compose Response

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Author Information
First or given name, e.g. 'Peter'.
Your last, or family, name, e.g. 'MacMoody'.
Your email address, e.g.
Your role and/or occupation, e.g. 'Orthopedic Surgeon'.
Your organization or institution (if applicable), e.g. 'Royal Free Hospital'.
Statement of Competing Interests


  • Responses are moderated before posting and publication is at the absolute discretion of BMJ, however they are not peer-reviewed
  • Once published, you will not have the right to remove or edit your response. Removal or editing of responses is at BMJ's absolute discretion
  • If patients could recognise themselves, or anyone else could recognise a patient from your description, please obtain the patient's written consent to publication and send them to the editorial office before submitting your response [Patient consent forms]
  • By submitting this response you are agreeing to our full [Response terms and requirements]

Vertical Tabs

Other responses

  • Published on:
    Author response to Dr. Wada and Dr. Kataoka
    • Hyungjin Kim, Radiologist Seoul National University Hospital
    • Other Contributors:
      • Chang Min Park, Radiologist

    We thank Dr. Wada and Dr. Kataoka for showing interest in our validation study. Dr. Wada and his colleague raised two important questions: 1) how was the pathologic diagnosis established in our study; and 2) the effect of selection bias (e.g. non-surgically diagnosed patients).
    First, in our study, all pathologic diagnoses were determined by the attending pathologists according to the 2011 lung adenocarcinoma classification described by the International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society.1 In our institution, 10% buffered formalin was infused to inflate and fix all surgical specimens containing subsolid nodules (SSNs) via the transpleural and transbronchial approach to precisely measure the invasive adenocarcinoma component.2 Furthermore, pathologic examination was performed independently as a part of routine clinical practice. Thus, pathologic diagnosis was not affected by the predictor variables we collected and vice versa.
    Second, we admit that exclusion of the non-surgically diagnosed patients might have caused selection bias in our study. The current study population comprised solely surgically resected lung nodules. However, this was inevitable given the indolent nature of SSNs and the unique characteristics of SSNs’ definitive diagnosis. SSNs grow slowly and even the SSNs with invasive adenocarcinoma components may remain unchanged over years of CT surveillance. Thus, definitive diagnosis of S...

    Show More
    Conflict of Interest:
    None declared.
  • Published on:
    Comment on, “Validation of prediction models for risk stratification of incidentally detected pulmonary subsolid nodules: a retrospective cohort study in a Korean tertiary medical center”
    • Nobuko Wada, MD Hyogo Prefectural Amagasaki General Medical Center
    • Other Contributors:
      • Yuki Kataoka, MD/MPH

    To the Editor:
    We read the recent articles from Kim et al. with great interest and appreciate the authors’ efforts to evaluate the suitability of the two models regarding the prediction of incidentally-detected pulmonary subsolid nodules (SSNs), as well as their reports that there were substantial differences. However, we would like to highlight two concerns that we have regarding their study.

    First, there is a lack of description regarding whether an adequate pathological diagnosis was performed. We would like to know who performed the diagnosis and how it was made. In predictive model research, it is preferable that outcomes are evaluated with masked predictors, as there might be bias in estimating associations between predictors and outcomes. [1]

    Secondly, there might have been a sampling bias before surgery selection. Among patients with SSNs, surgery might be preferentially performed, especially for patients who show a high possibility of lung cancer. Further, additional upper lobes and peripheral nodules, which were difficult to diagnose by bronchoscopy examination, might be selected and resected. Thus, cases of atypical adenomatous hyperplasia (AAH)/ adenocarcinoma in situ (AIS) might comprise a smaller portion of the study cohort, and minimally invasive adenocarcinoma (MIA)/ invasive pulmonary adenocarcinoma (IPA) might be diagnosed more frequently. Clinically, we often struggle to decide whether the nodule is malignant in a case where surgery c...

    Show More
    Conflict of Interest:
    None declared.