br Normality of continuous variables was
Normality of continuous variables was examined with the Kolmogorov-Smirnov test, and continuous variables were expressed with mean (±standard deviation) or median (interquartile range) values as appropriate. Differences in continuous variables were assessed with the Student's t-test or the Mann-Whitney U test, as appropriate. Differences in ordinal and categorical variables were examined by the Fisher's exact test or chi-squared tests, as appropriate.
The cumulative incidence curves of VTE were constructed with the Kaplan-Meier method, and the difference between curves was assessed with log-rank testing. Cox proportional hazard regression models were used for multivariable analysis, and magnitude of statistical significance was expressed with hazard ratios (HR) and 95% confidence interval (CI). Proportional hazard assumption was tested and showed no interaction with time.
To examine the independent association between statin, aspirin use and VTE risk, multiple adjustments were performed in various layers. This stepwise-adjustment in various models was used to assess and vi-sualize the interaction of statin use, aspirin use, and VTE risk in each layer. In a stepwise fashion, adjustment was performed for clinically rel-evant factors in the management of endometrial cancer, including pa-tient demographics, medical comorbidity, tumor characteristics, treatment types, and survival sequentially. That is, the first adjustment model included patient demographics only. In the second model, med-ical comorbidity was added to patient demographics. In the third model, tumor characteristics were further added. In the fourth model, treat-ment type was added. In the last model, survival event was added.
Logic behind this stepwise approach was as follows: First, we as-sumed that (i) medical comorbidities are commonly affected by patient demographics and that (ii) statin/aspirin use largely depends on these two factors. Next, tumorigenesis in endometrial cancer is generally divided into two pathways, type I (obesity and metabolic syndrome) versus type II tumors (elderly), which also depend on patient demo-graphics and comorbidities. Treatment type then depends on tumor and patient factors. Lastly, survival is largely affected by tumor factors. As stated earlier, each layer of variables independently impacts VTE risk. As long as 3X FLAG of HR for VTE risk was observed, we moved for-ward to construct the next adjustment model. In this study, we ob-served that statin-related VTE risk reduction remained constant throughout the five adjustment models. Thus, interpretation of analysis was based on the last adjustment model.
In a sensitivity analysis, the particular impact of statin subtypes on VTE risk was examined. This is based on the rationale that the type-specific risk reduction in VTE has not been examined previ-ously. In addition, the joint effect of combined statin and aspirin use on VTE risk was examined, as use of both medications is common in women with endometrial cancer [15,16]. The interaction between statin and/or aspirin use and clinico-pathological variables was assessed.
All statistical analyses were based upon two-tailed hypotheses, and a P b 0.05 was considered statistically significant. Statistical Package for the Social Sciences (IBM SPSS, version 24.0, Armonk, NY, USA) was used for the analyses. The STROBE guidelines were consulted to outline the results of this retrospective observational study .
Patient demographics based on statin, aspirin use.
There were 219 (8.7%, 95%CI 7.6–9.8) aspirin users recorded (aspirin daily dose: 81–100 mg n = 217 and 325 mg n = 2). Aspirin users were more likely to be young, Caucasian or Hispanic, obese, hypertensive, di-abetic, and dyslipidemic compared to non-users (all, P b 0.05). Aspirin users were less likely to have undergone laparotomy for hysterectomy and to have received chemotherapy (both, P b 0.05). Aspirin users had a significantly lower rates of endometrial cancer recurrence compared to non-users (P = 0.004).
On univariable analysis, statin users had a significantly lower inci-dence of VTE compared to non-users (5-year cumulative rates: 2.5% ver-sus 6.7%, P = 0.005). Similarly, on univariable analysis, aspirin use was significantly associated with decreased VTE risk (5-year cumulative in-cidence: 2.0% versus 6.5%, P = 0.017).
Multivariable analysis was performed to examine the independent association between statin and/or aspirin use and VTE risk in endome-trial cancer (Table 3). After controlling for patient characteristics, med-ical comorbidity, tumor factors, treatment types, and survival events, the association between statin use and decreased VTE risk remained