Poor predictability of lactose malabsorption from clinical symptoms for Chilean populations.

Clinical symptoms and lactose malabsorption may occur in persons intolerant to lactose. To predict malabsorption from symptoms and age, the authors used simple regression, multiple regression, and discrimant analysis on data from 436 Chileans. All individual r2 values from simple regression were less than 0.05. For the best stepwise multiple regression program, R2 was only 0.09, indicating that 91% of the variability in malabsorption was "unexplained" by the five best predictor variables. Because of so much unexplained variability, the authors could detect (using multiple regression) only 56 of 90 (62%) malabsorbers and only 14 of 38 (37%) absorbers. In conclusion, for their Chilean populations the authors poorly predicted malabsorption from symptoms.