Isolation and identification of nontuberculous mycobacteria from hospitalized patients and drinking water samples--examination of their correlation by chemometrics.

Nontuberculous mycobacteria (NTM) have been found to be widely dispersed in the environment and are being considered potentially pathogenic for humans and animals, while reports of their human to human transmission are absent. Water and aerosols are potential transmission modes of NTM to humans. Hospitalized patients with NTM infections were studied together with drinking water samples from their respective residence areas during 2003-2013. Cluster analysis and factor analysis were used to analyze the data matrix. A total of 367 hospitalized patients living in 30 localities in the Prefecture of Larissa were tested positive for NTM. The most frequently isolated NTM species of the 383 NTM isolates from the clinical specimens were Mycobacterium fortuitum (n = 118, 30.8 %), M. gordonae (n = 87, 22.7 %), M. peregrinum (n = 46, 12.0 %), M. chelonae (n = 11, 2.9 %), M. avium (n = 8, 2.1 %), and M. intracellulare (n = 7, 1.8 %), while 88 (23.0 %) of these isolates were not identified. It is noted that in 8 patients, M. tuberculosis was isolated simultaneously with one NTM, in 15 patients, together with two types of NTM, while in 1 patient, it was found at the same time as three different NTM. In addition, 3360 drinking water samples were collected from 30 localities and analyzed during 2010 to 2013; they were found 11.2 % NTM positive. Cluster analysis and factor analysis results confirm that NTM strains are correlated to each other in both isolated samples from patients and drinking water, while the strength of their correlation varied from weak to moderate (e.g., factor loadings ranged from 0.69 to 0.74 when all data are considered). These results provide indications that drinking water could be linked with NTM cases in humans.

Application of multivariate statistical methods for groundwater physicochemical and biological quality assessment in the context of public health.

Three representative areas (lowland, semi-mountainous, and coastal) have been selected for the collection of drinking water samples, and a total number of 28 physical, chemical, and biological parameters per water sample have been determined and analyzed. The mean values of the physical and chemical parameters were found to be within the limits mentioned in the 98/83/EEC directive. The analysis of biological parameters shows that many of the water samples are inadequate for human consumption because of the presence of bacteria. Cluster analysis (CA) first was used to classify sample sites with similar properties and results in three groups of sites; discriminant analysis (DA) was used to construct the best discriminant functions to confirm the clusters determined by CA and evaluate the spatial variations in water quality. The standard mode discriminant functions, using 17 parameters, yielded classification matrix correctly assigning 96.97% of the cases. In the stepwise mode, the DA produced a classification matrix with 96.36% correct assignments using only ten parameters (EC, Cl-, NO3-, HCO3-, CO3(-2), Ca+2, Na+, Zn, Mn, and Pb). CA and factor analysis (FA) are used to characterize water quality and assist in water quality monitoring planning. CA proved that two major groups of similarity (six subclusters) between 17 physicochemical parameters are formed, and FA extracts six factors that account for 66.478% of the total water quality variation, when all samples' physicochemical data set is considered. It is noteworthy that the classification scheme obtained by CA is completely confirmed by principal component analysis.