We believe that when updating the SOFA score, the fundamental principles outlined above should be retained. The score should be kept as simple as possible by including a limited number of objective variables—acknowledging the presence of iatrogenic confounders, such as sedation for the Glasgow Coma Scale (GCS) score—which are easily obtained and routinely measured in every institution, and retaining the same 0–4 scale for each organ system.
Many (bio)markers of organ function have been studied since the initial SOFA score was developed but have not been extensively validated and are clearly not available everywhere. Some of the variables proposed in the original SOFA score may therefore still represent the most widely available and reliable indicator of function for that organ system albeit with acknowledged limitations. For example, bilirubin concentrations may still be the best choice for the hepatic system even though raised bilirubin can be due to hemolysis rather than liver dysfunction and hyperbilirubinemia takes time to develop. Similarly, although the platelet count can be normal despite an abnormal prothrombin or partial thromboplastin time and clearly does not provide a full picture of coagulopathy, it may still represent the best option for assessing function of the coagulation system.
Assessment of central nervous system function is particularly challenging given the lack of available objective measures. The GCS score, although subjective, remains an obvious choice given its relative simplicity and extensive validation. Nevertheless, the use of sedative agents makes its interpretation difficult in some patients, in particular those receiving mechanical ventilation. In these circumstances, an assumed GCS, i.e., the score that the patient would have in the absence of sedation, could be used, as is currently recommended , recognizing that this may not be compatible with the fully automated data collection systems that are increasingly employed .
As noted during the development of the original SOFA score , the variables selected for each organ should ideally be independent of therapy, as management practices vary across units and patients depending on availability and hospital and/or physician preference. However, for the cardiovascular and respiratory systems this may not be possible. If the current use of therapeutic agents for the cardiovascular system is retained in a SOFA 2.0, several changes in use of vasopressor and inotropic agents to primarily correct hypotension or cardiac output merit consideration for inclusion. For example, vasopressin and its derivatives are now used in many centers [7, 8]. Although less widely used, metaraminol, phenylephrine and angiotensin II are other vasopressors that could be considered for inclusion [3, 8, 9]. While use of dopamine has declined considerably worldwide, it may still be used sufficiently to warrant retention [10, 11]. Inclusion of other inotropic agents, such as levosimendan and phosphodiesterase (PDE)-3 inhibitors , may also be considered in addition to dobutamine. At which degree of severity these variables should be included and which doses/cutoffs should be used would need prospective validation. Use of venoarterial extracorporeal membrane oxygenation (VA-ECMO), cardiac assist devices, or other support systems may also be considered in the assessment of the cardiovascular system , although such support may impact on the evaluation of the function of other organ systems. For example, a patient with severe cardiogenic shock receiving VA-ECMO will often have a very high PaO2 (i.e., > 400). When considering these issues, it will be important to not over-complicate the score while, at the same time, being generic for the different approaches in current use.
Another variable that could be considered to quantify the severity of cardiovascular dysfunction is blood lactate concentration. This can be easily monitored, values are related to morbidity and mortality in almost every critically ill patient, and a decrease during initial resuscitation generally indicates a good response to treatment . However, changes in lactate concentration are relatively slow and values may remain elevated after apparently adequate resuscitation. Moreover, concentrations may be raised by factors other than tissue hypoxia, for example, liver function and drugs , so their inclusion in a SOFA 2.0 would need prospective evaluation of utility.
A key change in clinical practice has been the gradual shift toward less invasive monitoring. Hence, for the respiratory system, the use of a PaO2 value obtained from blood gas analysis could potentially be replaced by SpO2 measured by pulse oximetry . However, this value is an approximation, as SpO2 is subject to more bias than is PaO2 , especially in the absence of positive end-expiratory pressure (PEEP). If the SpO2/FiO2 ratio were to be included as an alternative to evaluate and score oxygenation, as recently recommended , a relatively complex mathematical conversion is necessary using nonlinear equations [19, 20]. Conversion tables are, however, available to simplify this process (see Additional file 1: Table S1).
The need for “respiratory support” is currently a criterion for a respiratory sub-SOFA score of 3 or 4, which could now include use of high-flow oxygen therapy (HFOT) , non-invasive mechanical ventilation, and even venovenous extracorporeal membrane oxygenation (VV-ECMO) [22, 23] as these are more widely used. Similarly, renal replacement therapies are now widely available and could be considered as an indicator of renal dysfunction, unless used for non-renal indications (e.g., removal of toxic products). Use of other organ support techniques, such as liver replacement therapies, may need to be considered in the future, but these remain experimental at present.
Addition of other organ systems, such as gastrointestinal, metabolic or immune, could be considered, but it is unclear which variables could be used at the bedside to objectively evaluate function. Indeed, the gut was considered in the initial SOFA score, but excluded for these reasons . Moreover, the simplicity of SOFA is one of its key features; adding more organ systems would increase complexity and thus reduce its global accessibility.
We are fully aware that some of the variables in any scoring system may not be measured every day, especially in low resource countries. The variable that is most frequently missing from the current SOFA score is the bilirubin concentration [24, 25], usually because the clinician assumes the level is normal so does not measure it. This is in agreement with the general rule from the original score developers that missing values are considered as normal for calculation of the SOFA score. Other options for dealing with missing data are available and need to be considered when creating SOFA 2.0, particularly in an era with increased use of automated data abstraction.