Archived Comments for:
Incidence of cardiac arrests and unexpected deaths in surgical patients before and after implementation of a rapid response system
Inappropriate conclusions due to omission of value of statistics
Armand Girbes, University Hospital VU medical center
18 July 2012
Dear Sir,
It is with interest we read the paper by Simmer et al. where they retrospectively describe the incidence of cardiac arrest and unexpected deaths in surgical patients before and after implementation of a rapid response system (RRS). We congratulate the authors for their effort to describe their experience with the RRS in their hospital. Unfortunately this paper is apparently subjected to serious bias and incorrect interpretation of data and statistics, as if statistics do not matter. What the authors found in fact was the absence of any demonstrable effect of the introduction of a RRS. We are disappointed that statements as ¿Introduction of a RRS resulted in a 50% reduction of mortality¿ have passed the editors and reviewers. A reduction in mortality was not found at all, as can be derived from the data in the paper and the given Odds Ratios. The bias of the authors could not be better demonstrated than by their statement ¿the positive results were abated by the delayed activation of the medical emergency team activation¿ and ¿by the low baseline incidence¿. Apparently the believe in the positive effect of an RRS convinces them that only a type II error can explain the results without mentioning it as such.
Furthermore, important flaws in and related to the retrospective design are present in the study. E.g. so-called early warning scores were, as the authors point out not in use before the introduction of the RRS and therefore not systematically recorded. This makes an adequate comparison between groups rather difficult. Furthermore, the definition of ¿unexpected death¿ can be debated. The authors have defined this as: ¿death in the surgical ward or death in the ICU after an unplanned ICU admission¿. Whether death is expected yes or no depends on many factors, among others medical expertise, how well you look at a patient. Death in a surgical ward can be anticipated and therefore expected, but according to the authors it is by definition unexpected. It should be realized that RRS is in fact about detecting patients who are potentially in the wrong department, the ward instead of the ICU. But not all patients who fulfil the EWS criteria will benefit from an ICU admission. Finally, what especially worries us about this paper, that in our view ignores the value of statistics, is the interpretation of lay people and non-scientist who read the conclusion of this paper. It is of note that the Dutch Inspectorate has made the introduction of a RRS imperative in Dutch hospitals, although scientific proof is lacking and it has not been advocated by the scientific Dutch societies. We therefore plead for a careful representation of data and we would advice that the conclusion of the paper should be rephrased in: ¿Introduction of an RRS did not result in a reduction in cardiac arrest rates and/or unexpected deaths. A type II error cannot be excluded¿.
Prof. dr. Armand R.J. Girbes, Dr Albertus Beishuizen, Prof. dr. Jan G. Zijlstra
Inappropriate conclusions due to omission of value of statistics
18 July 2012
Dear Sir,
It is with interest we read the paper by Simmer et al. where they retrospectively describe the incidence of cardiac arrest and unexpected deaths in surgical patients before and after implementation of a rapid response system (RRS). We congratulate the authors for their effort to describe their experience with the RRS in their hospital. Unfortunately this paper is apparently subjected to serious bias and incorrect interpretation of data and statistics, as if statistics do not matter. What the authors found in fact was the absence of any demonstrable effect of the introduction of a RRS. We are disappointed that statements as ¿Introduction of a RRS resulted in a 50% reduction of mortality¿ have passed the editors and reviewers. A reduction in mortality was not found at all, as can be derived from the data in the paper and the given Odds Ratios. The bias of the authors could not be better demonstrated than by their statement ¿the positive results were abated by the delayed activation of the medical emergency team activation¿ and ¿by the low baseline incidence¿. Apparently the believe in the positive effect of an RRS convinces them that only a type II error can explain the results without mentioning it as such.
Furthermore, important flaws in and related to the retrospective design are present in the study. E.g. so-called early warning scores were, as the authors point out not in use before the introduction of the RRS and therefore not systematically recorded. This makes an adequate comparison between groups rather difficult. Furthermore, the definition of ¿unexpected death¿ can be debated. The authors have defined this as: ¿death in the surgical ward or death in the ICU after an unplanned ICU admission¿. Whether death is expected yes or no depends on many factors, among others medical expertise, how well you look at a patient. Death in a surgical ward can be anticipated and therefore expected, but according to the authors it is by definition unexpected. It should be realized that RRS is in fact about detecting patients who are potentially in the wrong department, the ward instead of the ICU. But not all patients who fulfil the EWS criteria will benefit from an ICU admission. Finally, what especially worries us about this paper, that in our view ignores the value of statistics, is the interpretation of lay people and non-scientist who read the conclusion of this paper. It is of note that the Dutch Inspectorate has made the introduction of a RRS imperative in Dutch hospitals, although scientific proof is lacking and it has not been advocated by the scientific Dutch societies. We therefore plead for a careful representation of data and we would advice that the conclusion of the paper should be rephrased in: ¿Introduction of an RRS did not result in a reduction in cardiac arrest rates and/or unexpected deaths. A type II error cannot be excluded¿.
Prof. dr. Armand R.J. Girbes, Dr Albertus Beishuizen, Prof. dr. Jan G. Zijlstra
Competing interests
We declare that we have no competing interests