The authors respond: Evaluating and analyzing outcomes accurately
American Journal of Emergency Medicine 38 (2020) 2225
Contents lists available at ScienceDirect
American Journal of Emergency Medicine
journal homepage:
The authors respond: Evaluating and analyzing outcomes accurately
Regarding the last concern, there is no conflict in the odds ratio (OR) if the readers could refer to the footnote of the Table 2, where we have
pointed out the GCS on-site and GCS at emergency room are modeled
using the “15-GCS” value in the logistic regression. As a higher GCS score means less Severe coma, if we use the original GCS score in the model, its OR would be <1. From the statistical perspective, when the
Thanks for the interest from the readers and we appreciate this op- portunity to further clarify some of the content of our research.
First, we agree the term “correlation” might be imprecisely used from a rigorous statistical point of view. But it has been widely applied in the literature to express a connection between a risk factor and the outcome where the cause-effect relationship is not yet demonstrated [1,2]. In the univariate analysis, we screened possible factors that might be associated with DNS. Before these factors were tested in the lo- gistic regression for their independence, we think the term “correlative factors” would represent a better expression than “independent predictors”.
Second, the reader raised an important issue in recognizing the DNS cases. In fact, there is a diagnostic standard in China to identify DNS (GBZ 23-2002, Diagnostic Criteria of Occupational Acute Carbon Mon- oxide Poisoning) which has been widely applied by all levels of hospi- tals and doctors [3]. As we mentioned in the Method section, all files were independently reviewed by the research group, and the confirma- tion of DNS was based on a clear history of CO exposure, a distinct “lucid interval“, and newly-onset neuroPsychiatric symptoms which consti- tute the main aspects of the GBZ 23-2002 criteria. Our research group also included a neurologist who was constantly consulted whenever an uncertainty occurred in the diagnosis.
OR is below 1, the corresponding factor would be considered as a “pro- tective” factor, and its AUC would be <0.5 in testing the sensitivity and specificity. To avoid such dilemma, we applied instead the reversed “15- GCS” score, and the results could be then interpreted as “lower GCS on- site is an independent predictor of DNS with an OR of 2.06”, and similar with the GCS at emergency room.
References
- Rose JJ, Wang L, Xu Q, et al. Carbon monoxide poisoning: Pathogenesis, management, and future directions of therapy. Am J Respir Crit Care Med. 2017;195(5):596-606.
- Hampson NB, Piantadosi CA, Thom SR, Weaver LK. Practice recommendations in the
diagnosis, management, and prevention of carbon monoxide poisoning. Am J Respir Crit Care Med. 2012;186(11):1095-101.
Xu P, Wang Y, Cao L, et al. Glasgow coma scale is a better delayed neurological se-
quelae risk factor than Neurological examination abnormalities in carbon monoxide poisoning. Am J Emerg Med. 2020. https://doi.org/10.1016/j.ajem.2020.02.047.
Yongxue Zhang MD, PhD Department of Surgery, Handan Medical Center of Bethune International Peace Hospital, 3 Nanhuan Road, 054000 Handan, China
E-mail address: [email protected]
31 July 2020
https://doi.org/10.1016/j.ajem.2020.07.088
0735-6757/(C) 2020