Correspondence| Volume 34, ISSUE 4, P494, October 2007

“On the subject of multilevel modelling in vascular surgery”

Open ArchivePublished:August 07, 2007DOI:
      We would like to thank Tang et al. for their comments. We advocate the idea that patient outcome is not only a function of the patient case-mix but instead predictive models should also adjust for the individual hospital-related factors (structure and process of care). This methodology is essential when there is “clustering” of patients (i.e. patients' results from the same hospital are more likely to be correlated) such as in a national setting. Multilevel models have already been used extensively in comparison studies focusing: on the Bristol enquiry (Aylin P. Lancet 2001); in colorectal cancer survival (Kee F. BMJ 1999, Tekkis PP. BMJ 2003); in gastro-oesophageal surgery (McCulloch P. BMJ 2003).
      Multi-level methodology in our study, was able to rank the 24 ICUs from the development study (Hadjianastassiou VG. Br J Surg 2005) and the 2 ICUs used in the current study (the data from the smaller unit were collected prospectively and could not simply be discarded), according to the level of adjustment that needed to take place for each hospital, and it certainly did not hide differences. On the contrary, contemporary models in vascular surgery only take into account the patient case-mix and ignore this “hospital effect”. We agree with Tang et al that models should only be used for the purpose they were designed for which is why we provided the evidence to warn investigators not to use POSSUM/VBHOM models in the post-operative AAA setting.
      A predictive model attempts to summate into a single value (“predicted risk”) the “case-mix” or the patient-related variables which have a clinical influence on outcome. Physiological data temporally closer (post-operative data) to this outcome more accurately reflect the state of the patient than data before a therapeutic intervention. Otherwise, prediction modelling would be more akin to “guessing” future events.

      Linked Article


      Commenting Guidelines

      To submit a comment for a journal article, please use the space above and note the following:

      • We will review submitted comments as soon as possible, striving for within two business days.
      • This forum is intended for constructive dialogue. Comments that are commercial or promotional in nature, pertain to specific medical cases, are not relevant to the article for which they have been submitted, or are otherwise inappropriate will not be posted.
      • We require that commenters identify themselves with names and affiliations.
      • Comments must be in compliance with our Terms & Conditions.
      • Comments are not peer-reviewed.