A clinical trial in Delhi compared the mean fasting blood glucose (FBG) levels of 35 newly diagnosed type 2 diabetic patients before and after a 12-week structured diet and exercise program. The baseline mean FBG was 156 mg/dL (SD = 18 mg/dL) and post-intervention mean FBG was 132 mg/dL (SD = 16 mg/dL). The researcher wants to determine if the reduction in FBG is statistically significant. Which statistical test is most appropriate?

## Test Selection for Paired Continuous Data **Key Point:** When the **same subjects** are measured at **two time points** (before and after), a **paired t-test** is the appropriate test of significance. ### Study Design Analysis - **Outcome variable:** Fasting blood glucose (FBG) — **continuous/quantitative** - **Measurement design:** Same 35 patients measured twice (baseline and post-intervention) — **paired/dependent samples** - **Sample size:** n = 35 (adequate for t-test; n > 30) - **Data distribution:** Assumed approximately normal (reasonable for FBG in a clinical population) ### Why Paired t-Test? Paired t-test is used when: 1. Data is **continuous** (FBG in mg/dL) 2. Same subjects measured at **two time points** (before-after design) 3. Observations are **dependent** (not independent) 4. Sample size is adequate (n ≥ 10, preferably ≥ 30) **High-Yield:** The paired t-test accounts for **within-subject variation** and is more powerful than unpaired t-test because it controls for individual baseline differences. ### Paired t-Test Formula $$t = \frac{\bar{d}}{SE_d} = \frac{\bar{d}}{s_d / \sqrt{n}}$$ where $\bar{d}$ = mean difference, $s_d$ = SD of differences, n = number of pairs ### Data Summary | Parameter | Baseline | Post-intervention | Difference | |-----------|----------|-------------------|------------| | Mean FBG (mg/dL) | 156 | 132 | 24 | | SD | 18 | 16 | — | | n | 35 | 35 | 35 | **Mnemonic:** **PAIR-ed t-test** = **PAIR**ed subjects, **PAIR**ed measurements, **PAIR**ed design **Clinical Pearl:** In clinical trials evaluating intervention efficacy (diet, exercise, medication), before-after designs in the same cohort require paired t-test, not unpaired t-test.