EP124 - Glycated hemoglobin as a predictor of postoperative outcomes after elective colorectal surgery

Purpose/Background: Diabetes mellitus (DM) has been associated with poor outcomes following elective colorectal surgery. The use of preoperative glycated hemoglobin (HbA1c) may provide a more precise estimate of DM control. The objectives of this study were to 1) determine whether DM control, using preoperative HbA1c values, could risk-stratify postoperative outcomes, and 2) to evaluate whether the use of DM control, rather than DM status alone, could improve the ACS-NSQIP risk calculator's model performance in predicting postoperative outcomes, following elective colorectal surgery.

Methods/Interventions: An observational multicenter cohort study using data from the ACS-NSQIP dataset between 2021-2023 was performed. Adult patients who underwent an elective colectomy or proctectomy for neoplastic or diverticular disease, and who had HbA1c measured preoperatively, were included. The primary exposure was DM control, defined as no DM (HbA1c < 6.5%), well-controlled DM (6.5% ≤ HbA1c ≤ 7.5%), or poorly-controlled DM (HbA1c > 7.5%). Several 30-day outcomes were assessed, including ACS-NSQIP risk calculator-defined composite outcomes of any complication, serious complications, cardiac complications, and surgical site infection. Multiple logistic regression was performed to adjust for relevant patient and disease covariates. As a subgroup analysis, outcomes were assessed among patients with a negative DM status only, comparing "true non-DM" (HbA1c < 6.5%) to "undiagnosed DM" (HbA1c >6.5%). Finally, the ACS-NSQIP risk prediction model (which currently uses DM status) was re-calibrated using DM control and HbA1c (continuous), and model discrimination and calibration were evaluated using the c-statistic and Brier score for 10 selected postoperative outcomes.

Results/Outcomes: In total, 32,578 patients were included: 19,261 (59.1%) had no DM, 8,999 (27.6%) had well-controlled DM, and 4,318 (13.3%) had poorly-controlled DM. Table 1 presents baseline patient and surgical characteristics between the three cohorts. On multivariate analysis, compared to patients with no DM, those with well-controlled DM had a similar risk of any complication (aOR 1.01; 95% CI 0.93-1.10), serious complications (aOR 1.01; 95% CI 0.92-1.10), cardiac complications (aOR 1.19; 95% CI 0.91-1.57), and surgical site infection (aOR 0.94; 95% CI 0.85 - 1.05). In contrast, compared to patients with no DM, poorly-controlled diabetes was associated with higher risk of any complication (aOR 1.17; 95% CI 1.06-1.29), serious complications (aOR 1.13; 95% CI 1.01-1.25), and surgical site infection (aOR 1.16; 95% CI 1.02-1.31), but not cardiac complications (aOR 1.32; 95% CI 0.94 - 1.84). On subgroup analysis, those with "undiagnosed DM" had a higher incidence of any complication (17% vs. 13%, p< 0.001), serious complications (14% vs. 11%, p< 0.001), and cardiac complications (1.1% vs. 0.7%, p=0.053) compared to "true non-DM". When comparing the ACS-NSQIP risk calculator prediction model with DM status, DM control, or HbA1c (continuous), model performance was fairly similar across the three (Table 2).

Conclusion/Discussion: In patients who underwent elective colorectal surgery, DM control based on preoperative HbA1c levels may improve postoperative risk stratification compared to DM status alone. Despite this, the use of DM control did not improve the ACS-NSQIP risk calculator's model performance. The increased risk for negative postoperative outcomes among those with "undiagnosed DM" may support universal preoperative testing of HbA1c, even among those thought to have no DM.

Table 1
Baseline Demographic and Surgical Characteristics
Table 2
Performance of Diabetes Predictors for Postoperative Complications