Predicting postoperative day 1 hematocrit levels after hysterectomy

Area of Science:

• Gynecology
• Surgical Outcomes
• Hematology

Background:

• The Swenson model is a linear regression tool designed to estimate postoperative day 1 hematocrit levels following benign hysterectomy.
• The model utilizes preoperative hematocrit, patient weight, estimated blood loss, crystalloid volume, platelet count, and hysterectomy route for predictions.
• Internal validation demonstrated high accuracy (±5% for 100% of patients).

Purpose of the Study:

• To externally validate the Swenson model's predictive accuracy in a distinct hysterectomy population.
• To assess the model's performance in a cohort including robotic-assisted hysterectomy cases.
• To evaluate the model's accuracy in predicting low hematocrit levels (≤30%) requiring potential intervention.

Main Methods:

• Retrospective cohort analysis of benign hysterectomies performed between April 2014 and May 2016.
• Comparison of predicted versus measured postoperative day 1 hematocrit levels using the Swenson equation.
• Statistical analysis including chi-square/Fisher's exact tests, receiver-operating characteristic curve analysis, and stratification by hematocrit ranges.

Main Results:

• The Swenson model demonstrated significantly lower accuracy in the external validation cohort compared to internal data (76.8% within ±5% vs. 100%).
• Accuracy deteriorated for predicted hematocrit levels outside the 32-38% range, particularly for levels ≤30% (55.3% within ±5%).
• The model tended to overestimate hematocrit in patients with actual levels ≤30%, potentially leading to false reassurance.

Conclusions:

• The Swenson model's predictive performance for postoperative day 1 hematocrit levels is suboptimal in an external hysterectomy population.
• The model may serve as a screening tool for anemia if projected hematocrit levels are ≥35%.
• Further research is required to develop a more robust and clinically applicable predictive model for postoperative hematocrit.