[An evaluation of the statistical significance and calculation of the confidence interval for the standardized mortality ratio].

The standardized mortality ratio (SMR) is the ratio of the number of deaths observed (D) to the number expected (E), on the basis of the mortality rates of some reference population. Several procedures have been proposed in order to test its significance and to estimate its confidence intervals. In this study, the SMR of two causes of death in 27 health areas of Castilla-La Mancha have been calculated. The significance has been evaluated by exact Poisson test and by four methods approximating the Poisson distribution by the normal: 1) a Z statistic based on the assumption that a Poisson variate with expectation E has a standard deviation equal to square root of E; 2) the Z statistic with a continuity correction; 3) a Z statistic based on the square root transformation of a Poisson variable and 4) an approximation of the exact test by Byar. Also, the confidence intervals have been estimated by exact method and by three approximate procedures: 1) by Byar; 2) by Z statistic uncorrected and 3) by the square root transformation of the Poisson distribution. With the exact methods and Byar procedure the results were very similar; therefore, using the last to testing significance and estimate the confidence intervals of SMR, is suggested.