To the Editor:
The Problem
The 1952 epidemic of poliomyelitis in Copenhagen resulted in critical interest in acid–base disturbances1 and the creation of intensive care.2 Technical advances3 in measurement were accompanied by improvements in the diagrammatic representation.4 The vigorous “Great Transatlantic Debate” faded with the introduction of “standard base excess” (SBE) to represent metabolic acidosis. Nevertheless, the potential for confusion remains. Partial pressure of carbon dioxide (PCO2) is measured in different units by geographical location (mm Hg or kPa), and metabolic acidosis may be represented by SBE, BEB (base excess of blood), BD (base deficit), BEECF (base excess of extracellular fluid), or BDECF (base excess of extracellular fluid), and less reliably by HCO3− (bicarbonate).
The Acid–Base Chart presented in 1971 by Siggaard-Andersen5 plotted pH against log(PCO2). It generated a value for standard base excess and offered curved outlined areas where clinical conditions (and normal) might be found. In 1976, a new diagram was created6 plotting partial pressure of carbon dioxide against standard base excess. This produced linear pH lines as well as linear zones for six typical clinical conditions (and normal); the partially compensated zones were placed midway between complete compensation and none.
In 1998, Schlichtig et al.7 refined the location of the partially compensated zones using Schlichtig’s comprehensive collection of data from 21 published reports (see fig. 1).
There is one published attempt to validate this diagram.8 The authors found that it presented the “higher diagnostic agreement” but also reported that their own “two expert physicians disagreed with each other in the diagnosis of 1/3 of the cases.” This last statement, in conjunction with abundant anecdotes of error and confusion in critical care, emergency care, and anesthesia, prompted the present article.
The Proposal
Artificial intelligence already generates various clinical text reports. For example, text may be added to an electrocardiogram report and a tentative diagnosis may be created from a mammogram. Accuracy has improved, but such reports cause concern9–11 and caution persists.12 In striking contrast, there appear to be no text reports associated with acid–base values despite the level of certainty that can be achieved. The proposal is that values for pH and partial pressure of carbon dioxide should generate accurate text descriptions. These are examples of reports generated by this system:
Ideally, a diagram showing the location of the values should accompany the text. For further examples of such text reports, use a laptop or desktop computer to visit the site http://www.acid-base.com/diagramlarge.php (accessed December 11, 2018).
The proposed text presents four separate items:
The direction and magnitude of the overall change in blood acidity
The primary, or major, component (respiratory or metabolic), magnitude, and value
The secondary, or minor, component (metabolic or respiratory), magnitude, and value
Description of a characteristic zone where one exists
The first three items are not controversial. They are accurate text descriptions that reduce the possibility of confusion. Although the fourth line is soundly derived from published data, the nature of acid–base disturbances means that outliers will exist. For this reason line four is preceded with the word “Consider….”
Conclusion
Environments in which acid–base values are interpreted are often far from ideal: in many hospitals in different parts of the world, fatigue, staff shortage, and inexperience are common. Blood gas reports should include a version of this text report: it may not have the vivid colors shown on the website, text may continue on a single line, and the units used will depend on location (e.g., rarely, if ever, will the acidity be shown both as pH and [H+]).
There is no longer any reason for providing acid–base results as just numbers with, in some cases, a string of confusing options for metabolic acidosis. Minimizing confusion in high-stress environments must be a priority. This proposal provides a solution.
Acknowledgments
The author is grateful to Peter Grogono, B.A., M.A., M.Comp.Sc., Ph.D. (Department of Computer Science and Software Engineering, Concordia University, Montreal, Quebec, Canada) for the introduction to animating diagrams using JavaScript.
Competing Interests
The author is not supported by, nor maintains any financial interest in, any commercial activity that may be associated with the topic of this article.