Occurrence of chemical substances in water supply systems of Brazil: a nonparametric approach for statistical analysis of Sisagua data

Abstract

The objective of this work was to develop a methodology for statistical analysis of monitoring data of chemical compounds in drinking water supply systems in Brazil, using data from Sisagua (Drinking Water Quality Surveillance Information System). Initially, the inconsistencies in the database were identified and adjusted. Then, the descriptive statistics were estimated using the Kaplan-Meier (KM) method, evaluating its applicability in different censored data sets. The descriptive parameters were compared with the substitution method. The substitution method showed susceptibility to biased estimates, especially for groups of compounds containing high percentage of censored data and with high limits of quantification and detection, leading to higher descriptive parameters compared to KM method. This work reinforces the need to use appropriate methods for analyzing environmental data and evidences that the analysis of this type of data may be complex. The methods proposed here can help environmental scientists to deal with this issue, providing a systematic procedure to check and solve consistency problems, as well as presenting a nonparametric approach for computing descriptive statistics for environmental monitoring data.

Diurom e tebuconazol no Brasil:

“…Para isso, foram obtidas as concentrações máximas detectadas de cada agrotóxico nas diferentes matrizes dos municípios brasileiros. Para o caso de dados detectados que se mantiveram abaixo dos LQ dos métodos analíticos utilizados, adotou-se a metade do maior LQ reportado na literatura como valor aproximado para a concentração máxima do agrotóxico, tendo em vista que a técnica de substituição de valores de dados detectados por metade do valor de LQ é frequentemente utilizada em estudos estatísticos para análise de dados ambientais (GOMES et al, 2022;MIKKONEN et al, 2018, SABINO et al, 2014.…”