I have used a square root transformation to successfully correct the positive skew of a predictor variable in a hierarchical multiple linear regression model.

The unstandardized beta coefficient for said predictor variable = -7.84.
My understanding is, for every 1 unit increase in the predictor variable, the dependent variable will decrease by 7.84 units (if the effects of all other predictors are held constant).

However, because the predictor variable has been square root transformed I don't know how to interpret this. I wish to report how a change in the predictor variable (in its original form/units) affects the dependent variable. Any guidance on how to interpret this model would be greatly appreciated.