Background: When doing systematic reviews with continuous variables we commonly focus on the mean effect size. A positive additional outcome of the work of a systematic review can be achieved by carefully looking at the variability of the data. In our case this was the standard deviation (SD) of blood pressure measurements.

Results: In performing a systematic review of the blood pressure lowering effect of thiazides we learned the following lessons about SD: 1. SD of systolic is greater than SD of diastolic blood pressure, but they are similar when expressed as a coefficient of variation (CV) 8-9%. 2. SD of baseline blood pressure is dependent on the entry criteria eg. If diastolic pressure is the entry criteria the SD of that measurement is significantly decreased (CV 5%) and not a valid measure of true blood pressure variability. 3. SD of the change from baseline associated with treatment is similar to the SD of the measurement itself. 4. SD of the change from baseline is not different in drug treatment and placebo groups. 5. Effect size is relatively insensitive to how the SD of the change from baseline is imputed. 6. Knowledge of the expected variability of blood pressure measurements can be used to detect data, which is of questionable validity.

Conclusions: Systematic reviews often reveal much more than the original objectives of the work. Blood pressure variability as estimated by standard deviation is an important measure, and researchers in the area should be familiar with the average magnitude of that variability.