Archives

  • 2019-10
  • 2019-11
  • 2020-03
  • 2020-07
  • 2020-08
  • 2021-03
  • br Strengths of this study include the

    2020-08-06


    Strengths of this study include the large sample size, allowing vari-ous adjustment models and rigorous analysis to ensure the durability of the association between statin use and VTE risk. Weaknesses of this study include lack of information on the dose and duration of medica-tions, which may suggest a volume-dependent risk reduction [24]. Data on compliance in medication usage or discontinuation of medica-tion during follow-up was not available. A relatively short follow-up time additionally limits the accuracy of capturing VTE events. Side
    Table 5
    Statin type-specific effects on venous thromboembolism.
    effects related to statin and aspirin use were also not available, and a composite endpoint together with risks and benefits was not assessable.
    An indication bias inherent to this type of retrospective study is a major concern [27]. Our study showed that statin or aspirin users were more likely to have factors associated with decreased risk of VTE (young, minimally-invasive surgery, or non-chemotherapy). While we addressed these confounding factors with multivariable models, exact risk adjustment for VTE remains immeasurable in retrospective study. A potential solution could be to compare to the active comparator with a new-user design [27]. However, such a study may favor more healthy patients as those with the worst disease prognosis probably do not receive statins/aspirin. We have previously addressed survival effects of statin/aspirin [13]: aspirin use was protective for endometrial cancer survival whereas statin use was not.
    Another limitation is the relatively small event number for VTE, which made sub-analyses difficult to conduct. Thus, there may be a pos-sible type II error in the association between aspirin use and VTE risk. With an α-level of 0.05, the power of our study to detect a statistically significant impact of aspirin use on occurrence of VTE was b30%, and N400 aspirin users would be needed to reach a power of 80%. Similarly, fewer than expected VTE events in the statin as SCH 58261 well as aspirin groups in our study may possibly make the interpretation of adjustment models less reliable due to the possibility of over-adjustment. Given the large number of comparisons, some statistically significant findings would be possibly expected by chance alone. However, throughout the layers of adjustment, the magnitude of statistical significance of the protective effects of statins on VTE was consistent (HR range, 0.33–0.42), implying that the association of statin use and decreased VTE risk holds likely true.
    Our study SCH 58261 was predominantly Asian and Hispanic, lying in two continents. Thus, generalizability and reproducibility in different populations may be limited. In a post-hoc analysis, we assessed country
    Adjustment model
    Demographics alone
    Demographics and
    Demographics,
    Demographics,
    Demographics,
    comorbidity r> comorbidity, and tumor
    comorbidity, tumor factors,
    comorbidity, tumor factors,
    factors
    and treatment type
    treatment type, and
    survival events
    Characteristic
    Adjusted-HR P-value Adjusted-HR P-value
    Adjusted-HR P-value
    Adjusted-HR P-value
    Adjusted-HR P-value
    None 1
    Simvastatin
    Atorvastatin
    Other statins
    Cox proportional hazard regression models for P-values. Significant P-values are emboldened. Demographics included age (every quartile), race/ethnicity (Asian versus non-Asian), and obesity (b30.0, 30–34.9, 35.0–39.9, and ≥40.0). Comorbidities included hypertension (yes versus no), diabetes mellitus (yes versus no), and hypercholesterolemia (yes versus no). Tumor factors included histology (type I versus type II), cancer stage (I–II versus III–IV), and CA-125 (b35 versus ≥ 35 IU/L). Treatment factors included hysterectomy (none, minimally-invasive, and laparotomy), chemotherapy use (yes versus no), and aspirin use (yes versus no). Survival events included endometrial cancer recurrence (yes versus no). Country type was not entered in the model because of concern for multicollinearity for race. Abbreviations: HR, hazard ratio; and CI, confidence interval.
    Table 6
    Interaction of patient demographics and statin use for venous thromboembolism risk.
    Characteristic Adjusted-HR (95%CI) P-value (interaction)
    Obesity
    Diabetes mellitus
    Non-diabetic/statin(−) 1
    Histology
    Surgery type
    Chemotherapy
    Survival events
    Cox proportional hazard regression models for venous thromboembolism risk. P-values represent interaction. All covariates shown in Table 1 were examined, and only significant covariates with P b 0.05 are listed. Significant P-values are emboldened. Demographics, co-morbidity, tumor factors, treatment type, and survival events were entered in the final model. Demographics included age (every quartile), race/ethnicity (Asian versus non-Asian), and obesity (b30.0, 30–34.9, 35.0–39.9, and ≥40.0). Comorbidities included hyper-tension (yes versus no), diabetes mellitus (yes versus no), and hypercholesterolemia (yes versus no). Tumor factors included histology (type I versus type II), cancer stage (I–II versus III–IV), and CA-125 (b35 versus ≥35 IU/L). Treatment factors included hysterectomy (none, minimally-invasive, and laparotomy), chemotherapy use (yes versus no), and aspirin use (yes versus no). Survival events included endometrial cancer recurrence (yes versus no). Country type was not entered in the model because of concern for multicollinearity for race. Abbreviations: CA-125, cancer antigen 125; MIS, minimally invasive surgery; HR, hazard ratio; and CI, confidence interval.