Calculating Mean and SD of Change from Baseline Using Correlation

Mayta
Jun 9
1 min read

🎯 Purpose

In clinical trials and observational studies, it's often necessary to compare change scores (e.g., improvement in blood pressure or lab values from baseline to follow-up). However, most studies report only mean and SD at baseline and follow-up, not the SD of the change.

This method estimates the mean change and its standard deviation (SD) using known formulas that incorporate an assumed correlation (r) between baseline and follow-up measurements.

🧩 Required Inputs

For each group (e.g., treatment or control), you need:

🧮 Step-by-Step Calculations

🔢 Example Calculation

Inputs:

Baseline mean = 140, SD = 10
Follow-up mean = 125, SD = 12
Correlation rrr = 0.6

So:

Mean change = −15
SD of change = 10

🧠 Interpretation Notes

Negative mean change typically indicates improvement (e.g., lower BP).
The higher the correlation, the lower the SD of the change.
Sensitivity analyses using multiple r-values are advisable when r is unknown.

🧭 Clinical Application

Essential for calculating effect size (Cohen’s d) for change-from-baseline designs.
Allows reconstruction of missing change data for meta-analyses.
Accepted in Cochrane reviews and RevMan5 when direct SD of change is unavailable.