Introduction. Reference values for pulmonary function parameters, generally applicable to most populations, have shown little validity in the correct interpretation of pulmonary function tests when applied to the adult population in Bogotá. Objective. To identify which of four prediction models generated for use in populations of a similar ethnic background to that found in Colombia is the most accurate for use in adults in Bogotá. Materials and methods. 534 subjects (male and female) between 18 and 65 years of age were recruited from a pool of workers employed by three private Bogotá companies. All subjects had resided in Bogotá for at least five years before the initiation of the study. Smokers and those with altered pulmonary function were excluded. Pulmonary function parameters were measured by flow volume curve. The results were analyzed (specifically, the difference between predicted and observed values) and the limits were calculated using the Bland y Altman method. A maximum average prediction error of 5percent-flag-change was accepted as valid for the observed value of each parameter. Results. The models shown to be valid were as follows: Crapo for forced vital capacity (FVC) in men; Pérez-Padilla for FVC in women and for the ratio forced expiratory volume in the first second (FEV1/FVC) in both sexes, and the Hankinson model for Mexican-Americans, for all parameters in both sexes. Conclusions. The Hankinson model proved to be the most accurate in predicting all spirometry parameters. However, its poor classification average (7percent-flag-change) is a limitation. In the future, new models with a better predictive accuracy will be required.
