Estimating Sepsis-Related Mortality Risk in Preoperative Patients Using Serum Albumin and Delayed Hypersensitivity Skin Testing

 

Patients studied:

(1) 245 patients admitted to Royal Victoria Hospital in Montreal, Canada in the late 1980's.

(2) Most of the deaths were associated with sepsis occurring postoperatively.

(3) The postoperative period was defined as the 30 days after surgery.

(4) 12 deaths were related to sepsis

(5) 2 patients died from other causes but sepsis was contributory

 

Factors found to be different between survivors and those who died:

(1) delayed hypersensitivity skin testing results

(2) serum albumin

(3) age

(4) hemoglobin

(5) total hemolytic complement

 

Factors found independently associated with sepsis-related postoperative death after logistic regression analysis:

(1) serum albumin g/L

(2) diameter in delayed hypersensitivity skin testing

 

Method for determining the DTH score:

(1) Skin testing is performed on the volar surface of the forearm or the lateral aspect of the upper arm.

(2) 5 antigens are tested: (a) Candidin 1:100 dilution, (b) Mumps skin test antigen undiluted, (c) purified protein derivative PPD at 5 TU, (d) Trichophytin, and (e) Streptokinase-Streptodornase (Veridase) at 100 units/mL

(3) control was normal saline

(4) 0.1 mL of the test material was injected intradermally.

(5) The greatest diameter of the area of induration was measured in mm at 24 and 48 hours. The two values were added together, then divided by 2 to give a mean value.

(6) The final value for each antigen was the (mean value for induration) minus (the mean value for the saline control).

(7) Areas of erythema were ignored.

(8) An antigen response was termed positive if the area of induration was >= 5 mm

 

Classification of patient based on response to antigens:

(1) reactive: positive (induration >= 5 mm) for >= 2 antigens

(2) relatively anergic: positive for 1 antigen

(3) anergic: negative (< 5 mm) for all 5 antigens

 

DTH score =

= (diameter for Candidin) + (diameter for Mumps antigen) + (diameter for PPD) + (points for Trichophytin) + (points for Streptokinase-Streptodornase)

 

X =

= 3.45 - (1.75 * (serum albumin in g/L)) - (0.30 * LN(DTH score))

 

where:

• DTH scores of 0 are set to 0.1 to prevent an error when taking the logarithm

 

probability of sepsis-related death =

= 1 / (1 + EXP((-1) * X))

 

Problem with calculations as given in the original reference:

(1) When the above calculation is implemented, the estimated risk of mortality is always 0.

(2) If albumin is represented as g/dL and not g/L, then results approximate that seen in the paper.

(3) If albumin is entered in the equation in g/dL, and is set to 3.6, then the plot of the probability of death is exactly that shown in Figure 2, page 75.

(4) Similarly, if the DTH score is set to 22 and the serum albumin is varied, use of g/dL gives the same curve as Figure 3, page 75.

 

Thus, my implementation uses

 

X =

= 3.45 - (1.75 * (serum albumin in g/dL)) - (0.30 * LN(DTH score))

 

Limitations:

• Use of skin testing prior to surgery is unlikely to be widely used in many centers.

• Advances in therapy to address defective host defenses would reduce risk of death from sepsis following surgery.