**2. Results**

## *2.1. Overall Results*

A total of 148 water samples were cultured on MWY and BCYE α agar produced by two di fferent manufacturers (i.e., Xebios and Oxoid). Bacteria were isolated from 70/148 samples (47.3%) using either type of medium, showing comparable levels of *Legionella* spp. detection. Specifically, we detected 64 (43.2%) and 62 (41.9%) positive cultures using Xebios and Oxoid mediums, respectively, with excellent agreemen<sup>t</sup> between the two brands (90.5%; κ = 0.807).

According to the Wilcoxon test analysis performed on 148 results obtained with each medium, the detection sensitivity increased when the samples were plated on Xebios medium ( *P* = 0.0054, median di fference between log-counts: Δ = 0.192, CI95%: 0.055, 0.394). Consistently, the 56 concordant positive samples also displayed the highest counts on Xebios agar plates ( *P* = 0.0006, Δ = 0.159, CI95%: 0.068, 0.295).

#### *2.2. Results Relative to Medium Type*

To further investigate the di fferences in the recovery of *Legionella* spp., we assessed the sensitivity and selectivity of Oxoid vs. Xebios alpha-ketoglutarate agar (BCYE α) and bromocresol purple (MWY) in detecting *Legionella* spp. from 148 environmental water samples (Table 1).


**Table 1.** Relative sensitivity and selectivity of BCYEα and MWY media from Xebios or Oxoid for *Legionella* spp. isolation.

**a** Sensitivity was calculated by comparing the number of positive plates for a given medium with the cumulative yield of *Legionella* spp. from all four media (n = 70). **b** Selectivity for each media was defined as the number of plates suppressing the growth of organisms that were not *Legionella* spp. over the total number of plates (n = 148).

The detection sensitivity was significantly higher in samples plated on MWYXebios agar compared to that of samples grown on MWYOxoid agar (McNemar test: *P* = 0.0042). No difference was found between BCYEαOxoid and BCYEαXebios media (McNemar test: *P* = 0.03588); no difference in suppressing non-*Legionella* bacteria (i.e., selectivity) was found between the two brands of MWY agar.

A Wilcoxon signed-rank test comparison (Table 2) revealed a significant difference in the number of *Legionella* spp. colonies, not only between BCYEαXebios and MWYXebios media but also between MWYOxoid and MWYXebios media. In either case, the recovery of *Legionella* spp. was significantly higher when the samples were plated on MWYXebios agar.

**Table 2.** Wilcoxon signed-rank test analysis of *Legionella* spp. counts (CFU/L) obtained with four different types of mediums.


Comparison between corresponding media from the two companies (a,b) and between different types of medium from each company (c,d). Δ is the median of the differences of log-counts, where the left column is the reference.

All data were disaggregated according to medium and manufacturer. Table 3 classifies the samples according to the presence or absence of *Legionella* spp. in MWY and BCYE media from Oxoid (Table 3a) or Xebios (Table 3b). There is a greater agreemen<sup>t</sup> between the two Xebios media (47/49) in comparison with that between the two Oxoid media (34/44).


**Table 3.** Assessment of *Legionella* spp. recovery according to culture medium and manufacturer.

Pearson's chi-squared test comparison between Oxoid and Xebios results, after a Deming–Stephan adjustment, revealed a statistically significant difference between the two brands (χ<sup>2</sup> = 9.0824,

*P*= 0.0026; Table 3c).

The number of *Legionella* spp. colonies (CFU/L) growing or not on BCYEα and MWY is shown in Tables 4 and 5.


**Table 4.** Number of *Legionella* spp. colonies (CFU/L) appearing on different plates: BCYEα positive results vs. MWY negative results.

Background (B.) flora was measured by semiquantitative counting: four categories were determined according to the visual density of colonies spread onto the plate, where zero is no background flora and 3+ is massive contamination (see supplementary materials). Lp 1 = *L. pneumophila* serogroups 1; Lp 3 = *L. pneumophila* serogroups 3; Lp 6 = *L. pneumophila* serogroups 6; L.spp. = *Legionella* spp. non-*pneumophila.*

**Table 5.** Number of *Legionella* spp. colonies (CFU/L) appearing on different plates: BCYEα negative results vs. MWY positive results.


Background (B.) flora was measured by semiquantitative counting: four categories were determined according to the visual density of colonies spread onto the plate, where zero is no background flora and 3+ is massive contamination (see supplementary materials). Lp 1 = *L. pneumophila* serogroups 1; Lp 3 = *L. pneumophila* serogroups 3; Lp 6 = *L. pneumophila* serogroups 6; Lp 2–14 = *L. pneumophila* serogroups 2–14; Lp 7–14= *L. pneumophila* serogroups 7–14; L.spp = *Legionella* spp. non-*pneumophila*.

The observation that some water samples were *Legionella* spp. positive on BCYEα medium but negative on MWY agar (7/10 cultures, samples 4–10, Table 4) indicates that the use of selective medium can affect the recovery of non-*Legionella pneumophila* species. Cell concentrations from these water samples ranged from 2.0 × 10<sup>2</sup> to 3.4 × 10<sup>3</sup> CFU/L. These results only refer to samples cultured on Oxoid media (Table 4).

Table 5 shows that high levels of background flora can challenge the results: on BCYEα agar, the results were affected by the presence of concomitant background flora that prevented the growth or masked the observation of *Legionella* colonies in BCYEα agar from both manufacturers. No qualitative data are available for these aquatic bacteria. Contaminating non-*Legionella* bacteria were rarely recovered on MWY agar from either manufacturer.

The influence of the media on the detection time was also assessed. Typically, antibiotics added to the selective media suppress the accompanying flora but at the same time slow down the growth of the target organism. The average colony size of *Legionella* spp. cultured on Xebios media was greater than that of *Legionella* spp. plated on Oxoid medium, especially for bacteria plated onto MWY agar. With regard to colony count, the number of *Legionella* spp. was higher on MWYXebios agar compared to MWYOxoid. As for the Oxoid media, even though the supplemented antibiotics suppressed the accompanying flora, it slowed down excessively the growth of the target organisms. No particular differences in colony size or count were observed between the two brands of BCYEα agar, as already shown in Table 2 (see also the supplementary materials).
