Cooling tower Legionella pneumophila surveillance results: Vancouver, Canada, 2021

Legionnaires' disease is a serious type of pneumonia: Nearly all patients diagnosed with the disease, regardless of age, are hospitalised, and about 7% die (Barskey et al. 2020). There is no vaccine. For various reasons, it is likely underreported – even outbreaks may be missed – and up to 70,000 cases annually have been estimated in the United States (NASEM 2020; Spiegelman et al. 2020). It is the leading cause of reportable waterborne disease outbreaks in the United States associated with drinking water and its incidence has been rising over the last two decades (Barskey et al. 2022). The disease results from the exposure of a susceptible individual to a virulent strain of Legionella bacteria at an infective concentration in, most often, contaminated aerosols (NASEM 2020).

Legionella pneumophila, and more particularly L. pneumophila serogroup 1 (sg1), is associated with the majority of Legionnaires' disease cases in North America and Europe, but other serogroups and non-pneumophila species of Legionella can cause the disease (Yu et al. 2002; NASEM 2020). The dominance of L. pneumophila (and of sg1) may be an artefact of clinical diagnosis by the most commonly used urinary antigen test (UAT) kits, and recent advances in UAT kits may provide a diagnosis to a broader range of Legionella bacteria (Hamilton et al. 2018; Barskey et al. 2020; Kim et al. 2022).

Aerosols are produced by building mechanical systems called ‘cooling towers’ – an imprecise but convenient shorthand used commonly in the literature for the various configurations of evaporative heat rejection or so called ‘wet cooling’ systems – which have been documented as sources of Legionnaires' disease cases from as far as 12 km away (Walser et al. 2014). Cooling towers were associated with 60% of Legionnaires' disease outbreak-associated deaths in a review over a decade (Hamilton et al. 2018). These building-scale mechanical systems are ubiquitous in (but not exclusive to) the urban environment. Bhopal et al. (1991) estimated that those living within 0.5 km of a cooling tower have a relative risk of Legionella infection over three times that of those living >1 km away, and Ricketts et al. (2012) estimated that 19.6% of non-outbreak, community-acquired cases may be associated with residential proximity to a cooling tower.

Legionella grows in building water systems that are not well maintained (Garrison et al. 2016). The general understanding that controlling Legionella at the building scale is an effective prevention strategy has catalysed the development of regulations, standards and guidance for mechanical systems such as cooling towers. A key recommendation of the NASEM (2020) consensus study report is the mandatory registration of cooling towers and regular monitoring of these systems for Legionella.

The fatal 2018 Legionnaires' disease outbreak in the City of Surrey, Canada (neighbouring the City of Vancouver) was associated with a cooling tower, but the public health response was hampered by the non-existence of a cooling tower registry (McCabe 2020). Similarly, when in 2020, the City of New Westminster, Canada (also neighbouring the City of Vancouver) recorded cases of Legionnaires' disease, no cooling tower registry was available to assist the same local health authority's investigation (personal communication). A number of European and North American jurisdictions have instituted mandatory cooling tower registries after an outbreak (Ricketts et al. 2008). Indeed, lamenting the absence of a cooling tower registry is a consistent theme of post-mortem outbreak reports (Walser et al. 2014; Paschke et al. 2019; Wüthrich et al. 2019); for jurisdictions with authority, a template for a cooling tower registry is available (USDN 2020).

In the City of Vancouver, Canada, operating permits are compulsory for cooling towers, which are defined by the regulations (see ‘Methods’). Operating permits are renewable annually and serve as a registry of cooling towers (http://www.vancouver.ca/operating-permit). Additionally, as a public health protection measure, the City implemented on 1 January 2021, an L. pneumophila standard of <10 CFU/mL in cooling towers (Vancouver Plumbing By-law 2019). This follows the Province of Quebec, Canada, which after a lethal cooling tower outbreak in 2012, adopted a province-wide registry and an L. pneumophila standard of <10 CFU/mL for cooling towers, verified by monthly culture-based testing (Racine et al. 2019). Similarly, Public Services and Procurement Canada (PSPC), an arm of the federal government, established an L. pneumophila standard of <10 CFU/mL for its nationwide portfolio of cooling towers, seven of which are in Vancouver, with water samples analysed monthly by a culture-based method to verify compliance (MD 15161-2013; NASEM 2020).

The <10 CFU/mL L. pneumophila standard is consistent with the NASEM (2020) report which recommended that, to prevent an outbreak, swift remediation should be initiated with a Legionella spp. concentration of 50 CFU/mL or more. The report advised that a lower ‘action level’ may be desirable for higher-risk populations, such as hospital patients (NASEM 2020). Indeed, in a review of 19 legionellosis outbreaks in a dozen countries, Walser et al. (2014) noted an L. pneumophila concentration of as low as 2 CFU/mL in an air scrubber associated with cases 10 km away.

In Vancouver, the <10 CFU/mL L. pneumophila standard is concordant with the detection limits of local commercial laboratories (see ‘Methods’). Cooling tower results of 10 CFU/mL or more are termed ‘exceedances.’ With culture-based methods, the warning of Lucas et al. (2011) merits repetition: Environmental laboratories are likely to underestimate the actual level of contamination. Nevertheless, culture-based testing (ISO 11731 method), despite its limitations, was selected over molecular-based methods in Vancouver for compliance verification. Llewellyn et al. (2017) found that 84% of the cooling towers they sampled across the United States were PCR-positive for Legionella spp., and, unable to correlate DNA presence with the ability to culture the bacteria, concluded that DNA presence alone may not be an accurate gauge of the effectiveness of a cooling tower maintenance program.

It is acknowledged that Legionella spp. rather than solely L. pneumophila could be better suited for a cooling tower standard: Non-pneumophila species of Legionella can also be pathogenic (Yu et al. 2002; NASEM 2020), and Llewellyn et al. (2017) cautioned that ‘the risk of concurrent or future growth of [L. pneumophila sg1] should be considered high’ if a non-pneumophila species of Legionella is isolated from a cooling tower. The City of Vancouver chose to harmonise with the <10 CFU/mL L. pneumophila standard established by the Province of Quebec and PSPC – an important co-benefit of which has been to preclude a regulatory patchwork in Canada for building owners and managers – with a phased implementation. In 2021, cooling towers in Vancouver were required to provide one L. pneumophila culture-based test result during the calendar year. Starting 1 January 2022, the cooling tower reporting frequency increased to once per calendar month of operation (Vancouver Plumbing By-law 2019). This article presents the 1 January–31 December 2021 dataset.

Cooling towers in the City of Vancouver, as in other jurisdictions, typically use municipally supplied potable water (though alternative sources may also be exploited, such as on-site groundwater, storm water, adjoining surface water or even reclaimed wastewater, though the latter source is prohibited in the City of Vancouver). Metro Vancouver supplies the City of Vancouver with treated surface water from protected watersheds and chlorinates for distribution. The City of Vancouver's municipal water distribution system comprises eight water pressure zones, and water can only flow between zones by passing through pressure-regulating valves. Water samples are taken weekly or every 2 weeks from the City of Vancouver's network of sampling stations for various field measurements and laboratory analyses.

In the present study, for those cooling towers which used municipal drinking water and reported L. pneumophila ≥10 CFU/mL, conventional distribution system water quality parameters – Escherichia coli and total coliform concentrations, free chlorine residual level, temperature, pH, turbidity and conductivity – were examined for the 3 months preceding each cooling tower exceedance, for possible correlations with the cooling tower L. pneumophila concentration.

Overall, the city-wide L. pneumophila surveillance data shared in this article are intended to facilitate insights into cooling tower operations and maintenance practices in the context of water supply and the potential public health implications. It is hoped that the City of Vancouver's transparency with these data is of assistance to other jurisdictions and researchers.

The City's regulatory definition of a ‘cooling tower’ is ‘a direct (open circuit) cooling tower, indirect (closed circuit) cooling tower, evaporative condenser, adiabatic cooler which recirculates non-evaporated water or fluid cooler that is part of a recirculated water system incorporated into a building's cooling, industrial process, refrigeration or energy production system, and may comprise one or more cooling tower cells’ (Vancouver Plumbing By-law 2019). The origin of this definition is The Rules of the City of New York (Title 24, Chapter 8).

Operating permit holders were responsible for sampling, arranging analysis with a commercial laboratory, and reporting their cooling tower L. pneumophila results online to the City of Vancouver. The regulations require that the commercial laboratory be accredited to ISO/IEC 17025, use the ISO 11731 culture-based method to quantify L. pneumophila (all serogroups), participate in a bi-annual external proficiency testing program such as CDC ELITE and agree to inform immediately the City of Vancouver and Medical Health Officer of any L. pneumophila result >1,000 CFU/mL (Vancouver Plumbing By-law 2019). When L. pneumophila was detected, most of the laboratories enumerated its serogroups and ‘other Legionella species,’ though this was not required by the regulations. The data are reproduced in Table 1 as they were presented in the laboratory reports. The commercial laboratories were coded and their self-reported method detection limits (MDL) for Legionella are as follows: L1 (10 CFU/mL); L2 (1 or 10 CFU/mL); L3 (‘typical MDL = 0.04 CFU/mL’); L4 (10 CFU/mL); L5 (1 or 5 CFU/mL) and L6 (5 CFU/mL). L. pneumophila data in this article were current with the City's database as of 25 April 2022.

Permit holders entered their L. pneumophila sample date and result and uploaded their laboratory report via the City's permitting portal (Computronix POSSE Land Management System v7.4.0). More than half of the online entries for 2021 were audited by City staff. Errors were corrected by City staff; the most common was the entry of the laboratory report date instead of the sample date (a difference of 2–5 weeks). Much rarer was the entry of an L. pneumophila result without a corresponding laboratory report, and this was consistently for a ‘not detected’ result. In such cases, the entry was deleted by City staff if the permit holder could not supply the laboratory report. All L. pneumophila reports of 10 CFU/mL or greater were assigned to a Plumbing Inspector who, typically in collaboration with the local health authority, conducted one or more site inspections and followed up with the permit holder on the prescribed corrective actions (Vancouver Plumbing By-law 2019), sometimes into the 2022 calendar year.

Water quality data are presented for the 3 months preceding the sample date of a cooling tower L. pneumophila exceedance, from the municipal water sampling station closest to the corresponding building and in the same municipal pressure zone. This time-frame was informed by the predictive modelling of Racine et al. (2019) who had tested cooling tower basin water quality data against subsequent cooling tower L. pneumophila concentrations. In the present study, the distance between a building and a water quality sampling station was estimated (±50 m) by tracing the shortest horizontal path along water mains between the sampling station and the service line, up to the point of building entry (http://maps.vancouver.ca). For buildings with multiple service lines, this distance was estimated for each service line, and the largest of these estimates is presented.

Water quality data from sampling stations were extracted on 14 April 2022 from the City of Vancouver database (Infor Hansen v11.2). Water samples were field tested by City staff for free chlorine, turbidity, conductivity and pH using a Hach Pocket Colorimeter II, based on the USEPA DPD Method; a Hach 2100Q Portable Turbidimeter, meeting EPA Method 180.1; a Thermo Scientific Orion Star A222 Conductivity Portable Meter; and a VWR sympHony H10P meter, respectively. The latter also provided water temperature data. Samples from the municipal water distribution system were also collected aseptically by City staff in bottles containing sodium thiosulphate and delivered to the Metro Vancouver Microbiology Laboratory (Burnaby, BC, Canada) for analysis by membrane filtration and Chromocult^® Coliform Agar (Austin 2020), with a detection limit of 1 CFU/mL for each of Escherichia coli and total coliforms. The City of Vancouver does not monitor for L. pneumophila in the distribution system. Available in the Supplementary Data are sampling station field measurements of total chlorine and heterotrophic plate count results for samples processed by the Metro Vancouver Microbiology Laboratory.

For statistical analyses, the dataset comprised each reported cooling tower exceedance in 2021 (Table 1). Spearman's rank-order correlation coefficient was used to evaluate the relationship between variables (two-tailed), with a significance level of 0.005 (IBM SPSS v29.0.0.0). This significance level is more rigorous than the conventional alpha of 0.05 and was selected to reduce false-positive results, acknowledging the increased potential of false-negative results (Miller & Ulrich 2019). For the Legionella data used in these analyses, ‘not detected’ reports were assigned a value of 0 CFU/mL, and the ‘greater than’ symbol was ignored (for example, >1,500 CFU/mL was treated as 1,500 CFU/mL). Notes 4 and 5 to Table 1 provide additional information on data preparation.

In 2021, of the 607 operating permits issued for cooling towers in the City of Vancouver, 557 provided at least one L. pneumophila laboratory report during the calendar year and 15 were not required to supply a report (because the cooling tower was not operational or not subject to the City's regulations), for an overall 94% compliance (557/592) with the sampling requirement. (Of the 35 operating permits which failed to meet this regulatory obligation, 24 provided an L. pneumophila report in early 2022, of which two cooling towers reported exceedances (data not shown).)

L. pneumophila results of ≥10 CFU/mL were reported in 2021 for 30 cooling towers (5.4%), of which six measured >1,000 CFU/mL. While more research is needed to correlate infection risk with measured Legionella concentrations and other factors in cooling towers, higher Legionella concentrations are associated with higher disease risk (NASEM 2020). All of the cooling towers in exceedance were on rooftops. For the City of Vancouver in 2021, there were no reported Legionnaires' disease outbreaks, and data on non-outbreak cases are unavailable.

With an exceedance, the City's regulations require another L. pneumophila sample after corrective actions (Vancouver Plumbing By-law 2019). The 30 cooling towers in exceedance are indexed in Table 1 to facilitate referencing in this article; for five of these cooling towers (C01e, C02b, C02c, C09a, C14), a follow-up sample taken in 2021 after cleaning and disinfection was also ≥10 CFU/mL, giving a total of 35 reports of ≥10 CFU/mL in 2021 (Figure 1).

Follow-up samples were taken in 2021 for 21 of the 30 cooling towers which reported an L. pneumophila result of 10 CFU/mL or greater; the other 9 failed to resample in 2021, in some cases because the exceedance had been in late 2021 and the corrective actions were in progress at year's end (Table 1). Of the 21 which submitted a 2021 follow-up report, 16 met the City of Vancouver standard of <10 CFU/mL of L. pneumophila. Five were again in exceedance (noted above), but each with a lower L. pneumophila concentration than the initial report, prompting another round of corrective actions, including the retesting requirement. Two of these five were resampled in 2021 and met the City of Vancouver standard (C09a, C14); the other three (at the two hospitals) were expected to be retested in the new year.

L. pneumophila serogroups were enumerated in the laboratory reports for 28 of the 30 cooling towers with exceedances (Table 1). Of these 28, 17 cooling towers (61%) contained L. pneumophila serogroup 1 (sg1). This was intermingled with another serogroup (described collectively herein as ‘non-serogroup 1’ L. pneumophila) in only three, unambiguous cases (C02a, C03, C09b; C01e had an unidentified serogroup, which may or may not have been sg1) and with non-pneumophila species of Legionella in four other cooling towers (C01a, C01b, C01c and C12). Non-pneumophila species of Legionella were not detected when a cooling tower in exceedance contained ‘non-serogroup 1’ L. pneumophila.

There was a negative correlation between the concentrations of L. pneumophila sg1 and ‘non-serogroup 1’ L. pneumophila, and this was statistically significant (r_s = −0.708, p < 0.001, n = 29). There was not a statistically significant correlation between the concentrations of non-pneumophila species of Legionella and L. pneumophila (r_s = 0.065, p = 0.716, n = 34). When the latter was considered based on serogroup, the analyses were suggestive of associations, but with α = 0.005, there was not a statistically significant correlation between the concentrations of non-pneumophila species of Legionella and either L. pneumophila sg1 (r_s = 0.400, p = 0.029, n = 30) or ‘non-serogroup 1’ L. pneumophila (r_s = −0.400, p = 0.032, n = 29). A caveat to these conclusions is their limitation to cooling towers containing at least 10 CFU/mL of L. pneumophila. Importantly, since the City of Vancouver cooling tower standard is based on L. pneumophila, a cooling tower containing non-pneumophila species of Legionella in the absence of L. pneumophila would not have been reported as an exceedance and would therefore not have been included in the correlation analyses.

The vast majority (94.6%) of reporting cooling towers met the <10 CFU/mL L. pneumophila standard, and a clear pattern emerged of highly localised, building-scale Legionella issues. The 30 cooling towers reporting exceedances were concentrated within 16 facilities – including two hospital campuses which together housed 12 of the cooling towers reporting exceedances. The other 14 facilities were diverse in both built form and occupancy classification and included a hotel, an industrial facility, a public utility, a warehouse and a variety of multi-storey commercial, residential and mixed-use buildings (Table 1). Taken together, the 16 facilities were dispersed geographically, encompassed by a circle with a 5 km radius and located within eight of the City's 22 local planning areas.

All of the cooling towers with exceedances used municipal water, and the associated 16 facilities were located in four of the City's eight water pressure zones (Table 1). Pressure zone 407 includes the entire downtown peninsula and hosts the majority of cooling towers in the City. In pressure zone 570, two cooling towers (C05, C08) reported exceedances, in buildings 2.3 km apart – and each recovered different L. pneumophila serogroups. Within pressure zone 442, three cooling towers reported exceedances, with two (C09a, C09b) co-located at an industrial facility 3.3 km away from a warehouse hosting the third cooling tower (C10). In pressure zone 499, exceedances were isolated to just one facility, Hospital A, accounting for nearly one-quarter of all exceedances in 2021. City-wide, there were only four cooling towers with exceedances that also reported non-pneumophila species of Legionella: Three of these were co-located at Hospital A, and the fourth cooling tower (C12) was in a commercial building 2.8 km away – and in a different municipal water pressure zone.

All of the municipal water quality results were compliant with applicable drinking water standards. During the 3 months preceding each of the reported L. pneumophila cooling tower exceedances, in the municipal water supplied to each of the corresponding 16 facilities: (1) neither Escherichia coli nor total coliforms were detected in any sample; (2) the free chlorine residual was ≥0.46 mg/L (and ≥0.54 mg/L for the two hospitals); (3) the water temperature was <20 °C; (4) the pH was 7.6–8.5; (5) the conductivity was <50 μS/cm; and (6) the turbidity was ≤0.71 NTU (and ≤0.34 NTU in the water used by the two hospitals), with three exceptions. On June 9, a turbidity of 2.79 NTU was measured at the municipal water sampling station nearest to C07, dropping to 0.18 NTU the next day; on June 24, a turbidity of 1.21 NTU was measured at the sampling station closest to C06, declining to 0.81 NTU when the next sample was taken a week later, and then to 0.31 NTU a week after that (Table 1). It is unknown what caused these temporary turbidity increases, but the free chlorine residual in the municipal water was ≥0.67 mg/L and ≥0.60 mg/L, respectively, during the 3 months preceding each cooling tower exceedance (Table 1).

No statistically significant correlation was found between the cooling tower concentration of L. pneumophila and any of the following municipal water quality parameters over the 3 months preceding each cooling tower exceedance: free chlorine residual (median or minimum), turbidity (median or maximum), temperature (median or maximum), pH (median) or conductivity (median or maximum) (Table 2). Considering L. pneumophila based on serogroup 1 and testing these parameters, there were two statistically significant associations: a negative correlation between turbidity (maximum) and L. pneumophila sg1 (r_s = −0.602, p < 0.001, n = 30) and a positive correlation between turbidity (maximum) and ‘non-serogroup 1’ L. pneumophila (r_s = 0.669, p < 0.001, n = 29). A statistically significant correlation was not found between the cooling tower concentration of non-pneumophila species of Legionella and any of the aforementioned municipal water quality parameters (Table 2).

There were no reported Legionnaires' disease outbreaks in the City of Vancouver in 2021, but just as alluded to by Spiegelman et al. (2020) in the context of the City of Toronto, Canada, not all cases of pneumonia in the City of Vancouver are subject to urinary antigen and Legionella culture-based testing – especially in 2021, in the midst of widespread respiratory illness caused by the coronavirus SARS-CoV-2.

Nonetheless, that two hospitals had systemic cooling tower L. pneumophila issues – identifying up to 880 CFU/mL of serogroup 1 (sg1) – was concerning (Table 1). Seven registered cooling towers at Hospital A and five at Hospital B were in exceedance, accounting for 40% of exceedances in the City in 2021. All of these cooling towers underwent cleaning and disinfection. Follow-up samples were provided for seven of these cooling towers in 2021, and in three instances, there was again an L. pneumophila exceedance, of 10–70 CFU/mL (C01e, C02b, C02c, Table 1). Through 2022, the cooling towers at Hospitals A and B will be followed closely by inspectors.

The most lethal Legionnaires' disease outbreak in Canadian history was in the City of Toronto in the fall of 2005, with a cooling tower at a long-term care home implicated in 23 deaths (Walker et al. 2005). Reviewing 19 cooling tower outbreaks in 12 countries (of which the Toronto 2005 outbreak was one), Walser et al. (2014) concluded that an avoidable public health risk is Legionella-contaminated cooling towers in proximity to medical care facilities: The authors described an outbreak fatality rate averaging 6.3%, but in the two medical facility outbreaks reviewed, the rate was about 4.5 times higher than this average. Indeed, summarising 27 Legionnaires' disease outbreak field investigations, Garrison et al. (2016) noted that all of the outbreaks were caused by L. pneumophila sg1 and that 85% of deaths were healthcare associated.

Of the Legionnaires' disease cases in 2016–2017 in the United States, Barskey et al. (2020) found that 20–21% resulted from healthcare exposure and 15–16% were associated with travel accommodations. To the latter, there are 74 hotels in the City of Vancouver (as of April 2022, and not all have cooling towers). Only one hotel, representing approximately 1% of total hotel rooms, reported exceedances in 2021, for both of its cooling towers, and follow-up samples after cleaning and disinfection did not detect L. pneumophila (C15a, C15b, Table 1).

Overall, 5.4% of cooling towers reporting L. pneumophila ≥10 CFU/mL was on the lower end when compared – imperfectly, due to differing contexts, such as climate – to other surveillance datasets derived from a culture-based test. Representing the lower end of the expected range is PSPC, with its well-established, robust Legionella prevention program. For the nationwide PSPC portfolio, 10 of its approximately 200 cooling towers ‘required unscheduled cleaning and disinfection’ in 2018, presumably because testing revealed ≥10 CFU/mL of L. pneumophila (NASEM 2020). Another domestic dataset is from a water treatment company servicing cooling towers in the Province of Quebec. L. pneumophila results of ≥10 CFU/mL were reported (on a quarterly basis) in 11–14% of 323 cooling towers in 2015, and in 8–12% in 2016; presumably, had this dataset also incorporated cooling towers without professional water management, the percentages would have been higher (Racine et al. 2019). Internationally, Mouchtouri et al. (2010) reported that 23% of 96 cooling towers tested in Greece ‘required remedial action,’ either because Legionella spp. was ≥10 CFU/mL or the ‘aerobic count’ was >100,000 CFU/mL. The results of the present study are for L. pneumophila, which is of course a subset of Legionella spp., so a comparison is inexact. Similarly, Hong Kong reported that 24% of water samples from cooling towers in 2021 had Legionella spp. ≥10 CFU/mL and 2.4% had >1,000 CFU/mL (n = 875, EMSD 2022). These samples were collected by inspectors and it is unlikely that the cooling towers were selected on a random basis; conversely, Vancouver's samples were collected city-wide by permit holders and self-reported, so a comparison is once again approximate. Lastly, and cited for completeness, cooling tower L. pneumophila surveys have been conducted using other enumeration methods, but such results cannot be assumed to be comparable to those derived from a culture-based method. For example, with a direct fluorescent antibody method, Gilpin & Gilpin (2014) surveyed 472 cooling towers across the United States over a 4-year period, with some cooling towers sampled repeatedly. Of the 1,336 samples acquired, 54% had L. pneumophila counts ≥10/mL and 3.1% had counts >1,000/mL (Gilpin & Gilpin 2014).

With the aim of characterising cooling tower microbial communities, Llewellyn et al. (2017) surveyed 196 cooling towers located in multiple U.S. climate regions using PCR as a screening tool and then a culture-based method. Llewellyn et al. (2017) reported that when L. pneumophila sg1 was recovered, 72% of samples also contained another L. pneumophila serogroup or non-pneumophila species of Legionella. In comparison, this was true for less than half of the samples in the present study (with a sample set limited to cooling towers from one climate region, with L. pneumophila ≥10 CFU/mL and for which serogroup-level analysis was available). Additionally, the present study identified a negative correlation between the concentrations of L. pneumophila sg1 and ‘non-serogroup 1’ L. pneumophila in cooling towers, hinting at a potentially competitive relationship. This proposition has a basis in laboratory research: Levin et al. (2019) discovered that L. pneumophila can antagonise the growth of Legionella bacteria – including other L. pneumophila – through secretion of an inhibitor called homogentisic acid (HGA). Levin et al. (2019) found that ‘high-density, established’ L. pneumophila communities were resistant to HGA and postulated that HGA was deployed ‘to protect against low-density, invading Legionella competitors’ in the contest for an ecological niche. Outside of the laboratory, Wéry et al. (2008) demonstrated from a study of a facility's cooling towers that L. pneumophila proliferation was associated with a decrease in the concentration (and diversity) of non-pneumophila species of Legionella (a correlation that was not statistically significant in the present study across cooling towers reporting exceedances). The speculation of Wéry et al. (2008), predating the HGA discovery, was that this might have been due to an antagonistic relationship, divergent temperature growth ranges or differences in amoebal hosts. Such ‘active’ or ‘passive’ competition (Bauer et al. 2018) could also be the basis for the results of the present study. This is explored further, below.

Overall, that 94.6% of cooling towers in the City of Vancouver did not contain L. pneumophila ≥10 CFU/mL in 2021 may suggest a general proficiency of building owners and managers in operating and maintaining their mechanical equipment. However, to assert this necessitates more than a single water sample from each cooling tower and demands acknowledgement that L. pneumophila concentrations can fluctuate – sometimes rapidly – with weather conditions and operational decisions (Bentham & Broadbent 1993; Wéry et al. 2008; Gilpin & Gilpin 2014; Llewellyn et al. 2017; Racine et al. 2019). It is noteworthy that through the policy development process culminating in the City of Vancouver's Legionella prevention regulations, the local tourism industry and the local association of building owners and managers were supportive, hosting educational seminars for their members and writing to City Council in favour of the measures (which were passed unanimously by City Council).

The low number of exceedances in 2021 may also be attributable partly to the high quality of municipal water that supplied nearly all of the City's registered cooling towers. Illustrative of an extreme is the 2014–2015 Legionnaires' disease outbreak in the City of Flint, Michigan contemporaneous with changes to the municipal water supply. Nevertheless, Smith et al. (2019) noted that while the Flint water distribution system could have fostered Legionella growth with its low chlorine and high iron content, residential proximity to only certain cooling towers – but not to others – was associated with some of the Legionnaires' disease cases, and for this and other reasons concluded that a key barrier to outbreaks is building-scale design and maintenance of water systems.

This being said, maintaining a minimum disinfectant residual in municipal drinking water systems is recommended so that the best water quality is delivered to buildings (NASEM 2020). The City of Vancouver and the local health authority have agreed that a minimum free chlorine residual of 0.2 mg/L in the municipal potable water is to be delivered to the property line (but this is not formally established as a system permit condition). This is a defensive measure intended to eliminate any faecal pathogen contamination downstream of the treatment plant and to forestall opportunistic pathogen growth. In the City of Vancouver in 2021, there was a free chlorine residual of ≥0.46 mg/L in the municipal water serving each of the cooling towers reporting an exceedance; moreover, for the six cooling towers reporting L. pneumophila >1,000 CFU/mL, the free chlorine residual in the municipal water was even higher, at ≥0.60 mg/L. These levels surpass the recommendation of LeChevallier (2019) for a minimum free chlorine residual of 0.1 mg/L in municipal distribution systems. LeChevallier (2019) had conducted surveillance exercises with several American water utilities and detected L. pneumophila in distribution systems only when water temperature exceeded 18 °C; furthermore, in positive samples, L. pneumophila was consistently <0.1 MPN/mL except when the free chlorine residual was <0.1 mg/L.

NASEM (2020) also suggests regular monitoring of L. pneumophila in municipal distribution systems, ‘triggered once water temperatures consistently exceed 20 °C.’ This condition was not met: For the 3 months preceding each cooling tower exceedance, the municipal water temperature never exceeded 20 °C. In fact, the highest measured temperature was 19.3 °C, and this municipal water, at pH 8.4, still had a robust 0.56 mg/L free chlorine residual. For reference, illustrating the susceptibility of L. pneumophila to free chlorine inactivation, Buse et al. (2019) reported a Chick–Watson Ct value of 0.3 min·mg/L for 99.99% (4-log₁₀) reduction of L. pneumophila by free chlorine at pH 8 and 21 °C.

No statistically significant correlations were found between cooling tower Legionella populations and conventional municipal water quality parameters measured in the 3 months preceding cooling tower exceedances, with the exception of turbidity. For cooling towers reporting exceedances, as the maximum measured turbidity increased in municipal water (range: 0.19–2.79 NTU), L. pneumophila serogroups exhibited divergent tendencies: L. pneumophila sg1 levels tended to decrease while ‘non-serogroup 1’ L. pneumophila concentrations tended to increase, and overall L. pneumophila concentration appeared to be uncorrelated. If these associations are not due to chance or an artefact, they of course do not imply causation; understanding what the relationship might be would be of public health interest, as L. pneumophila sg1 has been shown to be overwhelmingly dominant in culture-confirmed legionellosis (Yu et al. 2002). There is furthermore a general implication for studies on L. pneumophila: aggregating its serogroups may mask real differences (in the present study, laboratory reports did not facilitate the resolution of individual serogroups within the aggregate ‘non-serogroup 1’ L. pneumophila). L. pneumophila serogroups are defined by the architecture of lipopolysaccharide (LPS) associated with the outer leaflet of the bacteria's outer membrane (Shevchuk et al. 2011) and these serogroups may interact differentially with the environment. For example, Buse et al. (2018) observed a wide range in surface charge amongst L. pneumophila serogroups, with LPS diversity likely contributing to this; the researchers noted that surface charge impacts mobility, surface adherence and disinfectant resistance, and the variability observed may affect survival and ecological distribution. In another study, Buse et al. (2022) reported differences in L. pneumophila sg1, sg4 and sg6 inactivation by UV-C light-emitting diodes and cited the importance of understanding outer membrane properties. From the present study, one could conjecture that within the context of a potentially competitive relationship in cooling towers between L. pneumophila sg1 and other L. pneumophila serogroups, the long-term competitive outcome might be affected by a selective pressure (Bauer et al. 2018) exerted by unidentified suspended solids on the outer membrane, disadvantaging L. pneumophila sg1. Further investigation is warranted.

The cooling tower exceedances observed in the City of Vancouver in 2021 are most likely attributable to the building-specific design, maintenance and/or operational factors. Logan-Jackson & Rose (2022) concluded that ‘water management in the building is the most critical factor that contributes to the growth of Legionella species.’ Various studies have shown associations between L. pneumophila colonisation and growth in cooling towers and building-specific factors such as a cooling tower's exposure to sunshine, age and permanently-wet surface area-to-volume ratio; operator training; the existence of a risk management plan; biocide levels; and basin water temperature, pH and microbial community composition (Yamamoto et al. 1992; Bentham & Broadbent 1993; Mouchtouri et al. 2010; Llewellyn et al. 2017; Racine et al. 2019). Moreover, airborne construction debris, freeway pollution and wildfire particulates can challenge cooling tower water treatment regimens (Scanlon et al. 2020; personal communication with the Los Angeles Department of Water and Power). Intriguingly, Wüthrich et al. (2019) postulated from an outbreak investigation that contaminated aerosols from one cooling tower could potentially colonise another with Legionella bacteria.

An example is illustrative of the potential consequences of building-specific factors on L. pneumophila concentration. In the present study, a newly installed cooling tower (C13) reported >1,500 CFU/mL of L. pneumophila, and the municipal water had a free chlorine residual of ≥0.60 mg/L in the 3 months preceding the exceedance. The service provider explained that, owing to partial occupancy of the new building, valves to certain condenser loops had been closed, and the cause of L. pneumophila growth might have been the uncirculated water held in these loops (after corrective actions, no L. pneumophila was detected). No reasonable level of free chlorine residual in the municipal water delivered to this building could have been expected to thwart such a plumbing configuration.

Accordingly, Holsinger et al. (2022) encourage building-scale water management programs, concluding from their review of 184 legionellosis outbreaks that water utility ‘[d]isinfectant management alone should not be considered to control Legionella.’ Indeed not all cooling towers use municipally supplied potable water. Additionally, in various jurisdictions, there are pressures such as drinking water scarcity, rising utility costs and sewer capacity limitations compelling building owners to improve water efficiency (for cooling towers, by increasing the ‘cycles of concentration’) and to consider alternative water sources – placing even more importance on building-scale water management programs to control Legionella. Some new developments in the City of Vancouver, for example, are intending to supply their cooling towers with captured rainwater as their contribution to assuaging combined sewer overflows. The NASEM (2020) consensus study report recommends that authorities with jurisdiction, such as building departments, codify requirements for water management plans in all public buildings, and recommends regulations to register and monitor cooling towers.

No inferences are drawn from the ostensible proportion of cooling towers in exceedance in each calendar month (Figure 1), because permit holders were free to choose the month in 2021 for their one required sample (such that a cooling tower reporting an exceedance in a given month may well have also had L. pneumophila ≥10 CFU/mL earlier in the year). The L. pneumophila data presented in this paper are subject to several other limitations. The registration of cooling towers, collection of water samples, organisation of testing by a competent laboratory, and reporting of results were the responsibility of building owners. While pragmatic from the City of Vancouver's perspective, this arrangement introduced an inherent variation in proficiency and engendered a conflict-of-interest. There are possibilities of unregistered cooling towers, unrepresentative samples, unreported exceedances and incorrect sample dates or L. pneumophila results in online reports, and there is no reason to assume equivalent Legionella recovery or enumeration accuracy amongst the six commercial laboratories which conducted culture-based testing, which has its own methodological limitations (Lucas et al. 2011). It is acknowledged that a more robust dataset might have been derived by third parties such as City inspectors collecting cooling tower water samples, with appropriate quality assurance and quality control measures like travel blanks, field duplicates, and reference samples, but available resources precluded such arrangements. For statistical analyses, a stringent significance level (0.005) was chosen to reduce chance findings, but at the cost of potentially failing to identify real correlations.

The City of Vancouver regulations pertain to L. pneumophila concentrations, and permit holders were not required to supply information about other Legionella species that may have been detected. When L. pneumophila was detected, serogroup-level analysis was not required (but nonetheless five of the six laboratories provided such analysis). Ideally, information about sequence types could have facilitated a more refined analysis of Legionella distribution amongst the City's cooling towers, and even a comparison to past clinical isolates by the health authorities (Wüthrich et al. 2019).

Municipal water quality data were extracted from the closest sampling station in the same water pressure zone as a building that had reported a cooling tower exceedance, and the estimated distance between each building and its corresponding sampling station ranged from 50 to 2,300 m (±50 m, Table 1). The further a sampling station is from a building, the less indicative its data are for the municipal water supplied to that building.

As of 1 January 2022, as part of the phased implementation of Legionella prevention regulations, all cooling towers in the City of Vancouver require monthly L. pneumophila reporting, with the same sample collection, testing and reporting arrangements as for the data presented here. The 2022 dataset would facilitate a seasonal analysis of L. pneumophila positivity and concentrations, and its larger sample size would support more robust correlation analyses to validate the conclusions reached in the present study. Because cooling towers colonised by Legionella spp. have been found to be ‘at higher risk of future positivity’ (Racine et al. 2019), there is an opportunity to also validate this finding by tracking the subsequent results of cooling towers through 2022. Lastly, Mouchtouri et al. (2010) found that, when Legionella spp. testing was not conducted regularly (for the authors, this meant at least quarterly), there was a positive association with Legionella spp. colonisation of cooling towers. The City of Vancouver 2022 dataset would help to test the hypothesis of others that regular Legionella testing of cooling towers drives down Legionella positivity, perhaps by the increased sense of awareness and accountability (Mouchtouri et al. 2010; Gilpin & Gilpin 2014; Racine et al. 2019; NASEM 2020).

As the present study illustrates, even when high-quality municipal water is supplied to a facility, this may not counteract issues with building-scale design, operations or maintenance that facilitate Legionella growth. In totality, for the City of Vancouver in 2021, cooling tower and municipal water quality data suggest highly-localised, building-scale L. pneumophila issues. There was almost a 95% compliance of cooling towers with the L. pneumophila standard of <10 CFU/mL; the 30 cooling towers which did report exceedances were clustered within 16 facilities that were scattered geographically, and Legionella results considered in the context of the City's municipal water pressure zones and water quality data do not point to a municipal water distribution system issue.

An unexpected finding was that 12 of the registered cooling towers at two hospitals measured an L. pneumophila concentration of between 10 and 880 CFU/mL. Without a cooling tower registry and a regulatory requirement to report L. pneumophila concentrations, it is unknown for how long these issues, and those of the other 14 facilities with cooling towers in exceedance, would have persisted. To this point, while there was no reported Legionnaires' disease outbreak in Vancouver in 2021, Wüthrich et al. (2019) established that a strain recovered from a cooling tower during a Swiss outbreak investigation had been causing infections over several years. In 2021, all of the cooling towers in exceedance were required by the City of Vancouver's regulations to undergo corrective actions including cleaning, disinfection and retesting.

This is the first article, to the best of the authors' knowledge, summarising environmental surveillance data for L. pneumophila (analysed by a culture-based method) in cooling towers city-wide, coupled to information from the municipal water quality distribution system. It underscores the crucial role of building owners and managers in the design, operations and maintenance of mechanical systems such as cooling towers which have the potential for growing and spreading lethal waterborne pathogens. It also demonstrates the public health protection benefits from implementing proactively recommendations to catalogue, monitor and establish minimum operational expectations for cooling towers.

The City of Vancouver operating permit program team comprises Leora Mistry, Gurwinder Gadey, Anthony Leung, Mike Shaul, Tyler Carmichael and Aaron Hart, with technical support from Arne Faremo of the Vancouver Coastal Health Authority (VCHA), and the authors thank the team for their work and commitment. Neil Visser is thanked for the municipal water quality data. The authors are grateful to Kimberley Beck, Megan Bullock, Darius Chiu, Rebecca Doel and Ruth Suurkask for their assistance, and to Patrick Ryan, Kevin Lau and Saul Schwebs for their support.

The authors also wish to recognise local building owners and managers, and their consultants, contractors and laboratories for participating in the operating permit program. Singled out for special acknowledgement are Destination Vancouver and the Vancouver Tourism Facility Managers Association (Gwendal Castellan), the Building Owners and Managers Association of British Columbia (Damian Stathonikos) and the QuadReal Property Group (Matthew Strand and Alan Walls).

For assistance in the development of the operating permit program, the authors thank, from the VCHA and the B.C. Centre for Disease Control, Randy Ash, Shelley Beaudet, Linda Dix-Cooper, Eleni Galanis, Linda Hoang, Jessica Ip, David Jantzen, Emily Peterson, Natalie Prystajecky, Michael Schwandt, Christine Tchao, Esther Tong, Frankie Tsang and Michael Wu. Beyond Vancouver, the authors thank, from the U.S. Centers for Disease Control and Prevention, Laura Cooley, Jasen Kunz and Claressa Lucas for valuable discussions. Within Public Services and Procurement Canada, Jeff Moffat is recognised for his technical assistance and strong collaborative spirit. And last but certainly not least, the authors thank Christopher Boyd of MARC Strategic Solutions for his many years of benevolent support, guidance and friendship. Thank you to Christopher Boyd, Arne Faremo, Neal Wells and two anonymous peer-reviewers for providing comments to drafts of this manuscript.

Arne Faremo was seconded to Development, Buildings and Licensing in the City of Vancouver through the generous financial assistance of the Vancouver Coastal Health Authority. This work benefited from the support of the Urban Sustainability Directors Network (USDN) through a 2017 Peer Exchange Award (PE33) and a 2019–2020 Innovation Fund Grant (GR51), the latter of which was also supported by NSF International.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.