Search

Objective: To assess the reliability of 29 prediction models for vital capacity when these are applied as reference for young females, natives and residents in Mexico City (altitude = 2,240 m)., Design: Transversal, prospective and comparative study., Setting: Metropolitan Autonomous University, Mexico., Subjects: Sixty females clinically healthy, 17 to 29 years old, non-smokers., Procedure: Vital capacity was predicted from 29 models and measured with a bell spirometer (Collins 9 L) according to the American Thoracic Society guidelines. Statistical comparison was done between measured and predicted values by linear regression, analysis of variance and analysis of residuals., Results: Mean of measured vital capacity was 3,625 mL while mean values of predicted vital capacities varied from 2,644 mL to 3,962 mL. Actual and reference values showed a small but significative correlation (r = 0.40 to 0.49). By Dunnett's test, means from six of the 29 models showed non-significant differences (p > 0.05) with the mean of the measured vital capacity; however, all 29 models presented significative bias in the regression line between actual values and the residuals of predicted values (correlation coefficient, slope and intercept statistically different from zero, p < 0.05)., Conclusion: Due to the large bias found, models tested in the present study have limitations to predict accurate reference values for young mexican females.

The reliability of 33 prediction models for vital capacity was assessed in university students living in Mexico City (altitude = 2,240 m). Forced vital capacity (FVC) was predicted and measured in 83 healthy male students, 18 to 30 years old and non-smokers. Spirometric tests were performed with a 9 liter bell spirometer, according to ATS guidelines. Mean measured FVC was 5,094 +/- 555 mL (BTPS) while mean predicted FVC by the models ranged from 3,700 mL to 5,350 mL (BTPS) with a small but significative correlation between actual and reference values (r = 0.398 to 0.474, p < 0.05). The residual standard deviation showed values from 503 to 577 mL which were similar to those seen in the original population. Although 11 of the 33 models were able to predict mean observed FVC by Dunnett's test (p < 0.05), all of them showed significant bias on the regression between measured values and residuals of predicted values, i.e. intercepts (2,435 to 4,127 mL), slopes (-0.61 to -0.87) and correlations (-0.59 to -0.95) were statistically different from zero (p < 0.01). Our data remark the need to obtain appropriate reference FVC values in young Mexican males due to the questionable predictability of the models tested.