From the authors:
We would like to thank S.N.C. Barra and co-workers for their thoughtful response to our article [1]. We agree with the authors that prognostic assessment of patients with acute pulmonary embolism is of pivotal importance and, thus, is an area of active investigation for our group. We would like to comment on just a few of the points raised by the authors. First, the authors comment that the Low-Risk Pulmonary Embolism Decision score is the only model derived completely from haemodynamically stable patients, which they define as those without any evidence of myocardial necrosis or echocardiographic right ventricular dysfunction (RVD). We agree that risk stratification tools should be derived from haemodynamically stable pulmonary embolism patients. However, we disagree that markers of myocardial injury and RVD should be performed prior to clinical scores. In fact, there is increasing evidence that the Pulmonary Embolism Severity Index (PESI) and Simplified PESI (sPESI) identify low-risk patients who might benefit from outpatient therapy of their disease without the need for echocardiography or troponin tests [2,3].
S.N.C. Barra and co-workers seem to take away from our study that the PESI score 48 h after admission (PESI48) and sPESI48are more accurate than the PESI and sPESI. This is not really true. As the authors note later in their correspondence, we did not compare PESI with PESI48but retested PESI in those who were initially PESI class III. We were not interested in finding a score better than PESI but rather a way to increase our ability to identify patients who will have a good outcome regardless of the treatment setting. Thus, our intention in the current study was to further risk-stratify patients who are at intermediate risk on presentation. In essence, calculation of PESI48can identify a portion of intermediate risk patients who have responded favourably to their initial treatment and can be safely discharged early from the hospital. From a clinical application standpoint, we would suggest that PESI and/or sPESI can be used to identify a very low-risk population that can be treated entirely as outpatients. Calculation of the PESI48/sPESI48can then be used to identify patients appropriate for early discharge after an initial hospitalisation.
The authors asked us to consider further assessment of PESI48by calculating discrimination, calibration and accuracy scores. We have gladly done so. For PESI48, the area under the curve is 0.75 (95% CI 0.66–0.84), Hosmer–Lemeshow goodness-of-fit is 5.38 (p=0.72) and Brier score is 0.073. Finally, we thank the authors for identifying a typographical error in our manuscript. They note that we reported a negative integrative discrimination index (IDI). In fact, the IDI was positive. In other words, the new score improves classification by 2%. This error has been corrected in this issue of the欧洲呼吸杂志, both in print and online.
Thank you again for the correspondence and the opportunity to further clarify our findings.
Footnotes
Statement of Interest
A statement of interest for R. Yusen is available atwww.www.qdcxjkg.com/site/misc/statements.xhtml
- ©ERS 2013