Predicting the occurrence of major adverse cardiac events within 30 days of a vascular surgery: an empirical comparison of the minimum p value method and ROC curve approach using individual patient data meta-analysis

Abstract

We aimed to compare the minimum p value method and the area under the receiver operating characteristics (ROC) curve approach to categorize continuous biomarkers for the prediction of postoperative 30-day major adverse cardiac events in noncardiac vascular surgery patients. Individual-patient data from six cohorts reporting B-type natriuretic peptide (BNP) or N-terminal pro-B-type natriuretic peptide (NTproBNP) were obtained. These biomarkers were dichotomized using the minimum p value method and compared with previously reported ROC curve-derived thresholds using logistic regression analysis. A final prediction model was developed, internally validated, and assessed for its sensitivity to clustering effects. Finally, a preoperative risk score system was proposed. Thresholds identified by the minimum p value method and ROC curve approach were 115.57 pg/ml (p < 0.001) and 116 pg/ml for BNP, and 241.7 pg/ml (p = 0.001) and 277.5 pg/ml for NTproBNP, respectively. The minimum p value thresholds were slightly stronger predictors based on our logistic regression analysis. The final model included a composite predictor of the minimum p value method's BNP and NTproBNP thresholds [odds ratio (OR) = 8.5, p < 0.001], surgery type (OR = 2.5, p = 0.002), and diabetes (OR = 2.1, p = 0.015). Preoperative risks using the scoring system ranged from 2 to 49 %. The minimum p value method and ROC curve approach identify similar optimal thresholds. We propose to replace the revised cardiac risk index with our risk score system for individual-specific preoperative risk stratification after noncardiac nonvascular surgery.

Keywords: Biostatistics; Cardiovascular epidemiology; Minimum p value; Pre-operative risk; ROC curve approach; Vascular surgery.

Fig. 1

A minimum p value analysis demonstrating potential BNP threshold values and corresponding p values. This graph presents the corresponding p value of each Chi square test performed on a series of potential B-type natriuretic peptide (BNP) thresholds to predict 30-day MACE after vascular surgery. The threshold with the smallest p value is set as the optimal threshold for dichotomizing NTproBNP

Fig. 2

A minimum p value analysis demonstrating potential BNP threshold values and corresponding p values. This graph presents the corresponding p value of each Chi square test performed on a series of potential NTproBNP thresholds to predict 30-day MACE after vascular surgery. The threshold with the smallest p value is set as the optimal threshold for dichotomizing NTproBNP

Fig. 3

Forest plot of final model, internal validation model, and sensitivity models. Four models were created: (1) a final prediction model, (2) internal validation (IV) model, generalized estimating equations (GEE), and (3) mixed effects logistic regression (MELR) model with predictors MPM threshold, surgery type, and history of diabetes mellitus. This plot presents the odds ratio (OR), 95 % confidence interval (CI) and p value for each predictor of each model. MPM threshold is a composite predictor of BNP and NTproBNP thresholds determined by the minimum p value method [MPM threshold = 0 (reference) if BNP < 115.57 pg/ml or NTproBNP < 241.7 pg/ml; MPM threshold = 1 if BNP > 115.57 pg/ml or NTproBNP > 241.7 pg/ml). Surgery type is the type of noncardiac vascular surgery [infrainguinal (reference) vs. aortoiliac]. Diabetes is an indicator of whether or not an individual was diagnosed with diabetes [no (reference) vs. yes]