Comment: an CEA of using procalcitonin in intensive care
Source:    Publish Time: 2012-08-11 07:16   1629 Views   Size:  16px  14px  12px
Comment by Xuanqian Xie. This comment has not been peer-reviewed. Wilker et al. 20111 estimated the cost savings for s

Author: Xuanqian Xie. This comment has not been peer-reviewed.

Wilker et al. 20111 estimated the cost savings for septic patients in intensive care by using procalcitonin (PCT). Authors clearly stated their setting (DRG-based in Germany), assumptions made, patient selection, parameter inputs, and so on. Everything was transparent. The results were also very understandable. More importantly, I am glad to know that authors will implement algorithm in this paper to 5-10 hospitals, and will measure the effects of PCT in reality.

However, I still have some concerns of this study.

  • The overall cost savings were mainly derived from the reductions of length of stay (LoS) in ICU. But, I do not think the method of estimating LoS reduction is reliable.  Authors simply calculated the difference of the weighted means by number of patients for PCT and control arm. Those simple analyses cannot provide the P value and confidence interval. Therefore, I do not know whether PCT arm with less 1.8 days of ICU stay was statistically shorter than control arm. If there were no statistical differences, it would be misleading to estimate the cost-saving due to reductions of LoS in ICU. Authors may consider use the appropriate method in the meta-analysis of the difference in means.
  • Before conducting the meta-analysis, authors may check whether the 3 studies reporting the LoS in ICU can be combined. Those studies reported the LoS in ICU as one of the secondary outcomes. As there are quite a few secondary outcomes in one study, it is not uncommon to get some random significant results. It is necessary to justify the rational for pooling them, and test the credibility of significant pooled results, if any.
  • Authors used both the primary diagnosis (1 diagnosis) and secondary diagnosis (up to 50 diagnoses) of the administrative datasets to detect the sepsis patients, and extracted the healthcare costs of those patients. It would be consistent to use the primary diagnosis only to identify sepsis patients, rather than mix the patients with primary diagnosis and the secondary diagnosis together. According to my knowledge, most studies use the secondary diagnosis to identify the potential confounders, not the eligible patients with certain disease for inclusion. The severity levels of the disease can be quiet different between patients with primary and secondary disease, and usually the secondary diagnosis is less accurate.  
  • This study was lack of sensitivity analysis. It is important to explore the uncertainties in any cost analysis. Some journals, for example Value in Health, do not accept any economic article without sensitivity analysis.

Due to a number of obvious limitations, I think the findings of this article may be not robust. Fortunately, authors applied their simulated model to reality, so they can calibrate some parameters with real data later.   


Disclaimer: The views expressed in this comment reflect those of the author and do not reflect those of the institute which the author is affiliated with. 



Wilke MH, Grube RF, Bodmann KF. The use of a standardized PCT-algorithm reduces costs in intensive care in septic patients - a DRG-based simulation model. Eur J Med Res 2011; 16(12):543-548.