[Evaluative Matrix of the Longitudinal Bond in primary healthcare: statistical validation in a health territory in the city of Rio de Janeiro, Brazil].

In keeping with the process of institutionalization of evaluation of primary healthcare in Brazil, we developed the Evaluative Matrix of the Longitudinal Bond (MAVIL in Portuguese), a simple and concise evaluative tool. MAVIL is structured according to three dimensions in the concept of longitudinal bond: Dimension 1 - recognition of the basic health unit (UBS in Portuguese) as a regular source of care, Dimension 2 - healthcare worker/patient interpersonal relationship, and Dimension 3 - continuity of information. The instrument consists of a structured set of 12 criteria, 20 indicators and respective questions, and verification items submitted to various validation stages. The current article presents the statistical validation of the MAVIL application in two UBS in a health territory in order to support future applications and also verify the instrument's capacity to capture differences in the longitudinal bond profile between UBS. The method included the following definitions: sample design, choice of the metric, descriptive data analysis, and comparative test of means, through application of the test of normality and nonparametric statistic. As results, the sample by strata, together with option of totaling the points from the MAVIL items as the metric, allowed describing the longitudinal bond profile for three cross-sections: territory and UBS A and B. To verify the capacity of MAVIL to capture differences in the longitudinal bond profile, we tested the hypothesis of the data's normality with the Kolmogorov-Smirnov test. Having rejected the hypothesis of normality, we opted for the Wilcoxon nonparametric test to compare the UBS. Dimension 1 and Dimension 2 showed similar profiles, but there was a significant difference for Dimension 3, attesting to the sensitivity of MAVIL.