top of page
hand-businesswoman-touching-hand-artificial-intelligence-meaning-technology-connection-go-

Who Murdered Sepsis?


Image from Unsplash.com


Sepsis is dead.

Who is Sepsis? - Sepsis is a serious medical condition that is caused by the body’s response to an infection.

Who murdered Sepsis?

Let’s find out who is responsible for Sepsis’s death.

First, let us look at all the reasons which could have led to this.


Case Review:

The first clue is:

SIRS – Lets investigate SIRS.

Who is SIRS? – SIRS is Systemic Inflammatory Response Syndrome, related to systemic inflammation, organ dysfunction, and organ failure. SIRS is characterized by the following symptoms:

1. Temperatures >38.5˚ C or <35˚ C (Hyperthermia or Hypothermia)

2. Heart Rate > 90 beats/min (Tachycardia)

3. Respiratory rate > 20 breaths/min or PaCo2 < 35mmHg

4. WBC Count > 11000 cells/mm3 or <4000 cells/mm3

When two or more of these requirements are met, with or without the sign of infection, then it is SIRS.

But can SIRS alone be responsible for Sepsis condition? Is SIRS a good indicator of possible sepsis? Let’s dig deeper.

Detective Mr. Tableau:

Let’s get the help of Detective Mr. Tableau.

We have in our possession, a Sepsis dataset of patients admitted to the hospital, along with their vitals recorded every hour and their laboratory values. The patients can be categorized under 3 conditions:

Patients without sepsis, on admission (Sepsis Label 0).

Patients without sepsis on admission, but later developed Sepsis in ICU (Sepsis Label 0 to 1). Let us call them Onset Sepsis patients.

Patients with sepsis, on admission (Sepsis Label 1).

We’ll ask Mr. Tableau to segregate the patients based on these conditions.

Figure 1: Segregation of patients


From the visualization, we can see that 92.73% are non-Sepsis patients. Around 6.21% are Onset Sepsis and only 1.06% are confirmed, Sepsis patients. We are interested in the non-Sepsis patients who exhibit the symptoms of SIRS.

We can create a calculated field for isolating non-Sepsis patients exhibiting two or more SIRS symptoms. Out of 37,404 patients, 22440 patients (59.99%) had SIRS symptoms.


Figure 2: Count of non-Sepsis patients with SIRS symptoms


We created a calculated field for isolating patients exhibiting two or more SIRS symptoms. Out of 37,404 patients, 22440 patients (59.99%) had SIRS symptoms.


Let us look at the reasoning behind these results. What other causes could result in these symptoms? Why do so many patients exhibit SIRS? Does it mean all these patients are going to develop Sepsis?

A patient with burns or any trauma could easily exhibit any two of these symptoms.

· A burn patient could exhibit an increase in heart rate and temperature along with leukocytosis.

· A patient with a physical injury could also show these symptoms.

· Pancreatitis patients can exhibit high fever with increased heart rate.

By just looking at these symptoms we cannot conclude that SIRS alone leads to Sepsis. Not all patients who start with SIRS get septic. The SIRS criteria seem to be too generalized and on a broader scale and might negate our intention to identify only patients in whom SIRS might develop into SEPSIS.

So, is SIRS the main suspect? No. I think it’s safe to assume SIRS alone could not have led to the Sepsis condition.


Let’s investigate further.

What else could have happened?

· Complications in burn patients could lead to further infections or even renal failure, resulting in sepsis.

· The presence of open wounds in patients with physical injury increases the risk of infection.

On deeper inspection, we came across Sequential Organ Failure Assessment (SOFA) score - the SOFA score is based on six different scores, one for each of the systems (Respiratory, Cardiovascular, Hepatic, Renal, Coagulation, and Neurological systems). This score is usually used to determine the extent of a person’s organ function or rate of failure, during their stay in the ICU.


Figure 3: Table showing biomarker ranges based on Sequential Organ Failure Assessment Score


The above table lists the primary biomarker for each system in our body and the range for each biomarker, which could indicate any sign of infection.

Now, let us go to Mr. Tableau for more insight.

A look at the laboratory values of the patients who developed sepsis later in the hospital (Onset sepsis) might shed some light. Let us check for abnormalities in the readings.


Figure 4: Visualization showing the abnormal biomarker values for Onset sepsis patients


The above visualization shows the bio-marker values recorded every hour for the given patient. Here, we are comparing the biomarkers MAP (Mean Arterial Pressure), the ratio SaO2/FiO2, Platelets count, Creatinine, and Bilirubin of the patient with their respective ranges. The reference line on each pane indicates the minimum/maximum value that must be maintained, so as not to slip into sepsis. The biomarker values in the colored pane are abnormal values, which indicates that there is something wrong in that corresponding system and requires further attention. The abnormal values could indicate an underlying condition or an infection in that respective system.


From the visualization for Onset sepsis patients, we can observe that there are many patients with biomarker values in the colored region, which means that these patients are slipping into sepsis. Now, let us have a look at the same Gantt view plotted against non-sepsis patients, showing SIRS symptoms.


Figure 5: Visualization showing the abnormal biomarker values for non-sepsis patients exhibiting SIRS symptoms


We can see from this visualization, very few patients show abnormal values in the colored pane, which means that out of all these patients exhibiting SIRS symptoms, these are the selected few patients who are entering the sepsis stage. Please click here for an interactive comparison dashboard.

But, unlike the vitals which are recorded every hour, the laboratory values are available only when they are ordered by the provider. So, whenever two or more of the SIRS criteria are satisfied, a further look at the above-listed biomarker values might help us narrow down the initial diagnosis of sepsis. It could even point us in the direction of the system, which is affected or has an underlying infection.

So, it is safe to assume initial SIRS coupled with any suspected or confirmed sign of infection could be a major indicator of Sepsis.

Crime Report: From our findings, we can safely conclude that SIRS played a major role in Sepsis. Sepsis patients unquestionably exhibit SIRS symptoms, though the reverse is not true for all cases. Using SOFA, we can identify and segregate patients entering the sepsis stage, from all the patients exhibiting SIRS symptoms.

SIRS beyond doubt was involved in the death of Sepsis, but the actual events which eventually led to Sepsis’s condition require further investigation.


References:

https://www.kaggle.com/code/namratakapoor1/sepsis-classification/data


https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5472513/


The Sequential Organ Failure Assessment score for predicting outcome in patients with severe sepsis and evidence of hypoperfusion at the time of emergency department presentation - PMC (nih.gov)


https://fpnotebook.com/id/Exam/SqntlOrgnFlrAssmntScr.htm





79 views0 comments

Recent Posts

See All
bottom of page