Predicting postoperative day 1 hematocrit levels after uncomplicated hysterectomy

Abstract

Objective To develop a model for predicting postoperative hematocrit levels after uncomplicated hysterectomy. Methods In a retrospective study, data were analyzed from the Michigan Surgery Quality Collaborative for non-emergent hysterectomies performed for benign indications among women aged at least 18 years between January 1, 2012, and April 4, 2014. Linear mixed models were used for univariate and multivariate analyses. Results The model was developed with data from 4747 hysterectomies and validated on 1184 cases. In the mixed multivariate analysis, higher postoperative day 1 (POD1) hematocrit levels were associated with higher weight (B=0.03222, P<0.001), higher preoperative hematocrit (B=0.6587, P<0.001), and non-vaginal hysterectomy (B=0.2815, P=0.0055). Lower POD1 hematocrit was associated with higher preoperative platelet count (B=–0.00457, P<0.001), greater estimated blood loss (B=– 0.00652, P<0.001), and larger intraoperative crystalloid volume (B=–0.3303, P<0.001). The final model predicted POD1 hematocrit within 4% points of the actual value for 91.7% of cases in the validation set. Conclusion Use of the model after uncomplicated hysterectomy might help to support the practice of selectively conducting postoperative hematocrit tests after hysterectomy in a clinically thoughtful and cost-effective manner.

Predicting postoperative day 1 hematocrit levels after hysterectomy for malignant indication: Validating a previously published model

“…Statistical analysis was performed with SPSS version 21 (IBM Corp., Armonk, NY, USA). Demographic variables were compared with those from Swenson et al 3 to determine whether the populations were similar. Predicted POD1 hematocrit was compared with actual hematocrit by using descriptive statistics and χ 2 or Fisher exact test for the same accuracy thresholds (±1% to ±5%) used by Swenson et al 3 A P value of less than 0.05 was considered statistically significant.…”

“…Svenson et al 3 found that, in their internal validation set, the model predicted the actual POD1 hematocrit value to within ±4% for more than 90% of cases. A subsequent study performed by Rahman et al 1 validated the Swenson model at a different facility.…”

“…Routinely obtaining postoperative complete blood counts (CBC) has been shown to have no influence on patient outcomes 2 . Based on the 2019 Medicaid fee schedule, a CBC test costs US $7.18, suggesting that up to $4.3 million may be spent annually on obtaining routine postoperative day 1 (POD1) hematocrit levels at the time of hysterectomy with unproven clinical benefit 3 …”

“…Many studies have attempted to identify patient and operative factors that may predict POD1 hematocrit values 2,4 . Swenson et al 3 developed the following regression model using factors that should be available at the end of surgery to predict the POD1 hematocrit level:…”

“…A subsequent study performed by Rahman et al 1 validated the Swenson model at a different facility. Whereas the Swenson study 3 was conducted in a facility that performed only abdominal, laparoscopic or vaginal hysterectomies, the Rahman study 1 included a large number of robotic hysterectomies. In the latter validation study, the model predicted POD1 hematocrit to within ±5% for approximately 76% of cases, which was significantly less accurate than observed in the initial dataset.…”