Frailty, Blood Pressure & Clinical Outcomes: SPRINT Trial Analysis
Summary of Statistical Analysis Methods Used in the Study
Here’s a breakdown of the statistical methods employed in the study, as described in the provided text:
1.Descriptive Statistics:
* Continuous Variables: Means (standard deviations – SDs) or Medians (interquartile ranges – IQRs) were used.
* Categorical Variables: Counts (percentages) were used.
* Group Comparisons: One-way ANOVA (for continuous variables) and Chi-square tests (for categorical variables) were used to compare characteristics across diffrent frailty status groups (robust-to-frail, frail-to-robust, stable-frail, and stable-robust).
* Multiple Testing Correction: The Hochberg method was applied to adjust for multiple comparisons when comparing groups to the stable-robust group.
2. Survival Analysis:
* Kaplan-Meier Curves: Used to visualize the cumulative incidence of outcomes (major CVD events, all-cause mortality, SAEs) based on changes in frailty status.
* Log-Rank Test: Used to test for statistically significant differences between groups in the Kaplan-Meier curves.
* Cox Proportional Hazards Regression: Used to examine the association between changes in frailty status and the incidence of outcomes.
* Hazard Ratios (HRs) & 95% Confidence Intervals (95% CIs): Calculated, using the stable-robust group as the reference.
3. Analysis of Frailty Index change (∆FI):
* Restricted Cubic Splines: Used to illustrate the relationships between baseline Frailty index (FI) and change in FI (∆FI) with the clinical outcomes.
* Categorization of ∆FI: ∆FI was categorized in two ways:
* Tertiles: Divided into three groups based on the distribution of ∆FI values.
* Minimal clinically Important Difference (MCID): Categorized using thresholds of -0.03 and 0.03, based on prior research ([25, 26]).
* Cox Regression with ∆FI Categories: Cox proportional hazards regression was used to assess the association between these ∆FI categories and clinical outcomes.
4. Confounding variables:
* Multivariable Cox regression models included adjustments for potential confounders such as:
* Age (years)
* Sex (men or women)
* Race (non-Hispanic black, Hispanic, non-Hispanic white, or others)
* Education (lower than high school, high school graduate, post-high school training, or college graduate or higher education)
in essence, the study used a combination of descriptive statistics, survival analysis techniques, and regression modeling to investigate the relationship between changes in frailty and clinical outcomes, while controlling for potential confounding factors.