Characterization of reported legionellosis outbreaks associated with buildings served by public drinking water systems: United States, 2001–2017

Legionella bacteria are responsible for legionellosis with mild to severe illness (NASEM 2020). Legionella occur naturally in the environment and can be introduced into drinking water from water entering the distribution system or through distribution system breaches. Legionella can grow in biofilms where certain conditions in distribution and plumbing systems support the proliferation of the bacteria, including low disinfectant residual, favorable water temperature, high water age, sediment accumulation, and free-living amoeba (Wang et al. 2012; USEPA 2016a; LeChevallier 2020). Disinfectant residuals can be depleted as a result of long residence times (e.g., in storage tanks and extensive plumbing systems), reactions with piping materials, demand from contaminants entering the distribution system, nutrients, and microbial activity (Li et al. 2019). The optimal growth temperature for Legionella ranges between 25 and 45°C (77–113°F), which is commonly found in hot water lines of building water systems (BWSs). BWSs include water lines that supply water throughout a building for various uses. Water at the optimal growth temperature could also be present in some areas of drinking water distribution systems during warm months (USEPA 2016a; NASEM 2020; CDC 2021).

In certain water features and fixtures, such as sink faucets or showerheads, water containing Legionella can become aerosolized and inhaled or aspirated into a person's lungs, which may cause legionellosis (CDC 2021). Legionellosis is classified as a respiratory disease that varies in severity from a self-limiting febrile illness known as Pontiac fever, to a more serious and sometimes fatal form of pneumonia called Legionnaires’ disease (NASEM 2020). People who are more susceptible to Legionnaires’ disease include current or former smokers, males, people who are 50 years of age or older, and people who are immunocompromised (CDC 2021). Legionnaires’ disease is the leading cause of reportable waterborne disease outbreaks associated with drinking water in the United States (Benedict et al. 2017). CDC estimated in 2014 that there were 11,000 cases of Legionnaires’ disease due to waterborne exposure (Collier et al. 2021), with an estimated one in 10 cases leading to death (CDC 2021).

Legionnaires’ disease incidence has been rising since 2000, with the number of reported cases increasing nine-fold between 2000 and 2018 (CDC 2021). Legionnaires’ disease cases are believed to be underdiagnosed since the disease presentation is indistinguishable from that of other bacterial infections and treatment can be non-specific, with broad-spectrum antibiotics. There are likely large numbers of unreported cases of Pontiac fever since the illness is relatively mild and self-limiting. Therefore, national surveillance systems likely underestimate the magnitude of the public health burden of legionellosis.

In 2017, 95% of legionellosis cases reported to CDC's Supplemental Legionnaires’ Disease Surveillance System (SLDSS) were Legionnaires’ disease (CDC 2020). While legionellosis is a nationally notifiable disease in the United States, more data are needed on how distribution systems and BWSs may influence the occurrence of drinking water-associated cases. Legionellosis outbreaks are voluntarily reported by states and territories to CDC's Waterborne Disease and Outbreak Surveillance System (WBDOSS, a national level surveillance system). Public health professionals have used the web-based National Outbreak Reporting System (NORS) to report waterborne disease outbreaks to the CDC since 2009, prior to which paper forms were used. The NORS outbreak reports include the following: type of water exposure, epidemiologic data, setting of exposure, suspected or confirmed outbreak etiology (genus/chemical/toxin) and species (if applicable), and water sampling testing related to the outbreak investigation. Drinking water outbreaks may collect additional information such as type of water system, Public Water System Identification Number (PWSID), water source, water treatment, and water quality (report of violations). The CDC defines a legionellosis outbreak as two or more people with legionellosis and linked epidemiologically to a common time and source of exposure (CDC 2021). According to CDC (2021), nearly 10,000 total cases were reported in 2018, and Collier et al. (2021) estimated that there are 2.3 cases of legionellosis in the United States for each reported case due to underdiagnosis. No estimates of underdiagnosis or underreporting are available for legionellosis outbreak surveillance, but annual numbers of outbreak-associated cases reported in the WBDOSS comprise about 4% of cases reported in case surveillance systems. This suggests that outbreaks of legionellosis are substantially underreported.

The Environmental Protection Agency (EPA) has different disinfectant requirements for public water systems (PWSs), depending upon type of source water. EPA defines a PWS as a system that has at least 15 service connections or serves an average of at least 25 people for at least 60 days per year (USEPA 2020a). PWSs are categorized into community water systems (CWSs) and non-community water systems (non-transient and transient) (NCWSs). CWSs supply water to the same population year-round. For surface waters (and ground water under the influence of surface water), the EPA regulates Legionella in PWSs under the surface water treatment rule (SWTR) through a treatment technique with a maximum contaminant level goal (MCLG) of zero (USEPA 1989). While there are no monitoring requirements for Legionella, the SWTR requires a detectable residual in distribution systems, which is effective in reducing waterborne enteric organisms. PWS treatment and its residual impact the water quality found in distribution systems and BWSs (Kool et al. 1999). Under the SWTR, drinking water systems are required, in most instances, to filter, disinfect, and maintain a measurable disinfectant in the distribution system (USEPA 1989). In comparison, the ground water rule (GWR) does not require remedial action (which can include disinfection) for ground water systems unless a well is deemed vulnerable to fecal contamination or a corrective action is needed (USEPA 2006).

According to the EPA's Safe Drinking Water Information System (SDWIS), there were 142,985 small PWSs serving ≤10,000 people (130,999 systems with a ground water source and 11,906 with a surface water source) and 4,369 large PWSs serving >10,000 (1,591 systems with a ground water source and 2,778 with a surface water source) in 2017 (USEPA 2017). Of systems that were CWSs, 38,305 used ground water and 11,658 used surface water (USEPA 2017). EPA's third Six-Year Review, a process that the Safe Drinking Water Act requires to provide the review of each national primary drinking water regulation for adequate public health protection, evaluated disinfection practices of PWSs based on the third Unregulated Contaminant Monitoring Rule (UCMR 3) that included a census of drinking water systems serving a population >10,000 people and a survey of 800 systems serving ≤10,000 people (USEPA 2020b). Of the 9,838 ground water entry points to distribution systems included in the UCMR 3, chlorine disinfection was used 8.8 times more often than chloramine (n = 7,881 for chlorine exclusively and n = 896 for chloramines or both chlorine and chloramines) (USEPA 2016b). For the 3,179 surface water entry points to distribution systems in the UCMR 3, chlorine was used 1.9 times more than chloramine (n = 1,648 for chlorine exclusively and n = 879 for chloramines or both chlorine and chloramines) (USEPA 2016b). A 2017 disinfectant practices survey by the American Water Works Association (AWWA) showed that among the 375 responding PWSs, chlorine was used in 70% of systems (n = 259) and chloramine in 21% of systems (n = 77) (AWWA 2021).

Maintaining a disinfectant residual in the distribution system can reduce Legionella exposures (NASEM 2020). Residual disinfectant has been shown to decrease the occurrence and concentration of Legionella in distribution systems (Flannery et al. 2006; LeChevallier 2019) and prevent outbreaks of Legionnaires’ disease in hospitals (Kool et al. 1999). There is also growing evidence that chloramine residual drinking water disinfection better controls Legionella risk in BWSs compared with chlorine; the reasons for this remain unclear (Kool et al. 1999; Donohue et al. 2019; NASEM 2020). Kool et al. (1999) found in a case–control study that hospitals supplied with drinking water containing chlorine as a residual disinfectant were more likely to have a reported outbreak of Legionnaires’ disease than those that used water with chloramine as a residual disinfectant. In another study of 573 distribution system samples, Legionella pneumophila was detected in 14 samples: 13 from 317 samples from chlorinated systems (4.1%) and one from 256 samples from chloraminated systems (0.4%) (LeChevallier 2019). Additionally, the prospective environmental study of San Francisco's switch from chlorine to chloramine in the drinking water system detected Legionella in fewer sites of the distribution system after the switch to chloramine (Flannery et al. 2006). Chloramine has also been found to penetrate biofilms that may harbor Legionella better than chlorine (NASM 2020). Though these findings show a difference in chlorine and chloramine efficacy, additional studies would improve understanding of optimal disinfection practices and their impact on Legionella.

Legionellosis outbreaks have been reported to the WBDOSS since the 1970s but did not expand to include more severe disease manifestations (i.e., Legionnaires’ disease) until 2001 (CDC 2019). For this study, we examined legionellosis outbreak data reported to the WBDOSS for 2001–2017, restricted to BWSs receiving water from PWSs, by several characteristics, including the type of PWS (CWS and NCWS). Legionellosis outbreak exposures have been identified in healthcare facilities, apartment complexes, hotels, and many other settings. To date, no comprehensive national study on drinking water-associated legionellosis outbreaks has been conducted to determine the drinking water system characteristics associated with PWSs and BWSs.

The overall study goal was to evaluate PWS characteristics that could impact the incidence of legionellosis outbreaks. Legionellosis outbreaks were characterized by source water type, drinking water disinfectant type, settings of exposure, geographical distribution by U.S. Census Divisions, and PWS size (population served). This national, systematic assessment of reported legionellosis outbreaks from drinking water exposures provides new insights on PWS characteristics that may influence building water quality.

The authors compiled the following information from CDC's 2001 to 2017 WBDOSS legionellosis outbreak data: count of outbreak-associated cases, year, state/territory, county (when available), city (when available), type of exposure (drinking water), setting of exposure (healthcare facilities, hotels/inns, apartment/condominiums, gym, prison, or other), type of PWS serving the BWS (community or non-community), PWS source water (ground water, surface water, mixed sources, or unknown), and the PWSID (when available) for additional dataset linking activities. This information was collected by local and state/territorial public health professionals during the outbreak investigation. This analysis included reported drinking water-associated legionellosis outbreaks and did not include outbreaks associated with recreational waters or water from systems not intended for drinking such as cooling towers.

Starting with outbreaks reported to the WBDOSS, the authors used SDWIS databases (USEPA 2017) to match the PWS providing drinking water to the BWSs implicated in the outbreak first by using the PWSID if available in the WBDOSS. If the PWSID was not available the authors used state, county, and city information to identify the PWSID. The PWS was determined by verifying the type of water system (CWS or NCWS), type of source water (surface water, ground water, or mixed sources), and other information collected by the WBDOSS. Using the SDWIS information for each PWS, water system characteristics were collected including the system size (size of the population served) and the PWSID (when not available from the WBDOSS). Although some information on disinfectant residual type was available in the WBDOSS dataset, it was unclear if this disinfectant type was due to PWS disinfection or additional treatment by the BWS.

The authors used multiple EPA datasets to determine PWS disinfectant type, including the Disinfection Byproduct Information Collection Rule (DBP ICR) (1998 data only), the UCMR 2 (2008–2010), the UCMR 3 (2013–2015), and the UCMR 4 (2018–2020) datasets (USEPA 2019, 2020b). If disinfectant type was unavailable from EPA ICR or UCMR databases, three other resources were used including PWS Consumer Confidence Reports (CCRs) (from the outbreak year if available or from a recent CCR (typically 2018–2019)), PWS websites, and EPA's SDWIS. If multiple disinfectant types were reported for a system (either treatment system or residual) within UMCR databases, the disinfectant type was further explored using other available resources. Recognizing data were not always available for each outbreak year, the authors assigned the type of disinfectant used by the PWS serving the BWS associated with each outbreak based on that known to have been used closest to before, during, or after the outbreak year. For example, if the same disinfectant type was used before and after the outbreak year, then that type of disinfectant was assumed for the outbreak year. As another example, the authors assumed that once a PWS began using chloramine, unless indicated after the year the outbreak occurred, the PWS remained a chloraminated system. If the system could not be identified, then the disinfectant type was categorized as unknown (see Supplementary Material for more detail on the disinfectant category assignment). Although information on additional BWS water treatment is not available in the SDWIS or systematically collected in the WBDOSS, we examined outbreak reports for any mention of additional treatment in a BWS.

One hundred and eighty-four reported legionellosis outbreaks and 1,030 cases of illness associated with water from public drinking water systems were reported to the WBDOSS for 2001–2017. The authors were able to link outbreak etiology to 146 outbreaks. Of those outbreaks, 142 were confirmed and four were suspected for Legionella. The species were determined as pneumophila (141 (137 confirmed; four suspected)), pneumophilia and anisa (1), and unknown species (4). The PWSIDs associated with the reported outbreaks were reported in 62 (34%) of the 184 reported outbreaks. The authors were able to determine PWSIDs for 87 (47%) of the remaining outbreaks and unable to determine 35 (19%) of them. Figure 1 shows the number of reported outbreaks (bar graph) and associated cases (line graph) of legionellosis for the years 2001–2017 associated with drinking water exposures. Figure 1 also details the source water type used by PWSs serving the BWSs implicated in the outbreaks.

Figure 2 shows reported outbreaks of legionellosis by U.S. Census divisions and territories (US Census Bureau 2010). The majority of outbreaks occurred in South Atlantic (n = 65, 35%), Middle Atlantic (n = 58, 32%), and East North Central (n = 29, 16%), regions which, combined, contain a population of 147 million based on the 2010 Census (total U.S. population, 308.7 million). One outbreak occurred in a U.S. territory, with the remaining occurring in the 50 states.

Figure 3 displays the types of exposure settings for reported legionellosis outbreaks (n = 184) and associated cases (n = 1,030). The pie chart on the left in Figure 3 displays exposure settings associated with outbreaks and the pie chart on the right displays exposure settings associated with outbreak-associated cases. The majority of exposure settings were healthcare-related facilities, including hospitals, nursing homes, long-term care facilities, or assisted-living/senior housing complexes. The second highest number of outbreak settings were in a combined category of hotels, lodges, and inns. Thirteen outbreaks were associated with apartment buildings and the remaining exposure settings were classified as ‘other,’ consisting of clubs (n = 3), casinos (n = 2), prisons (n = 2), a camp (n = 1), a restaurant (n = 1), a gym (n = 1), a factory (n = 1), a private residence (n = 1), a municipality (n = 1), and other, unspecified settings (n = 5). ‘Municipality,’ in this instance, was used to indicate a general community setting.

CWSs comprised the majority (94%) of PWSs supplying water to the settings where the outbreaks occurred, followed by NCWS (4%), and unknown (2%).

Of the 184 reported legionellosis outbreaks, 54% (n = 100) were associated with BWSs served by surface water systems, 34% (n = 63) by ground water systems, 2% (n = 4) by mixed source (surface and ground water) systems, and 9% (n = 17) by an unspecified type of water source. Most outbreaks (81%) (n = 149) were linked to a specific PWS based on descriptive data or a PWSID in the outbreak report (34% contained PWSIDs (n = 62)). There were a total of 88 unique PWSs: 66 PWSs served water to BWSs that were implicated in one outbreak each and 22 PWSs served water to multiple BWSs implicated in outbreaks. The 22 PWSs associated with multiple legionellosis outbreaks accounted for 83% of the total outbreaks evaluated; in most instances, multiple outbreaks associated with the same PWS occurred in different years in multiple settings of exposure. Five PWSs were associated with six or more outbreaks, including one PWS associated with nine outbreaks.

Focusing on the 149 outbreaks with PWS information available, BWSs obtaining water from large (serving 10,001–100,000 people) and very large (greater than 100,000 people) (n = 44 and n = 79, respectively) sized PWSs were implicated in 83% of reported outbreaks (n = 123). Very small (n = 7), small (n = 10) and medium (n = 9) sized PWSs provided water to BWSs implicated in 17% of outbreaks. A comparison of outbreaks across PWS characteristics, including source water, disinfectant type, and system size is shown in Table 1.

Among the 149 reported outbreaks with PWS information available, 85% (n = 127) were associated with BWSs receiving chlorinated water, 10% (n = 15) were receiving chloraminated water, and 5% (n = 7) were receiving water either from PWSs whose disinfectant type could not be determined or systems that did not provide disinfection. For a subset of outbreaks (around 15%), data were available related to disinfectant residual concentrations in the BWS, but collection times relative to the outbreak, sampling locations within buildings, and analytical methods were not identified.

Of the 22 PWSs that were individually associated with multiple BWS outbreaks, 91% (n = 20) used chlorine, the remaining 9% (n = 2) included one that used chloramine and one that used chlorine during two outbreaks and chloramine during one outbreak. Seventy-seven percent of the 22 outbreaks associated with multiple BWSs were supplied by surface water systems (n = 17) and 23% were supplied by ground water systems (n = 5). Sixty-four percent (n = 14) of the 22 PWSs were reported as serving more than 100,000 people, 32% (n = 7) served more than 10,001 but less than 100,000 people, and 5% (n = 1) served 500 or fewer people. A few BWSs were identified as supplementing treatment onsite with chlorine dioxide (n = 4) and copper silver ionization (n = 4).

This study examined characteristics of 184 legionellosis outbreaks associated with drinking water that were reported to CDC's WBDOSS from 2001 to 2017. Reported outbreaks and outbreak-associated cases of legionellosis have been on the rise for the last two decades in the United States as shown in Figure 1. This increase could be related to several factors including increased number of persons at risk (such as an aging population), aging plumbing or distribution system infrastructure, or changes in the climate leading to warmer water in drinking water distribution systems. Public health surveillance of legionellosis outbreaks, which became more comprehensive in 2001, may have been bolstered by increased use of diagnostic testing due to greater awareness among clinicians and availability of diagnostic tests, as well as increased public health capacity to detect, investigate, and report these outbreaks. Currently, it is not known which factors are significant contributors to the increase in reporting for this subset of drinking water-associated outbreaks.

The number of reported outbreaks fluctuated through 2001–2017, but overall, the trend showed an increase over time. This upward trend is concerning in light of the recent annual estimate of Legionnaires’ disease cases (NASEM 2020), which included outbreaks and sporadic cases for which the primary exposure source was not identified. Furthermore, the data in Figure 1 do not include undiagnosed cases or cases not associated with outbreaks. Likely more cases with drinking water exposures are associated with outbreaks that were not identified, as well as community-acquired cases that were not recognized or diagnosed. Source attribution is higher for outbreak-associated cases than for cases not known to be outbreak-associated, and exposure source is unlikely to be determined for the latter, whether drinking water or other.

Most outbreaks were reported from South Atlantic (n = 65), Middle Atlantic (n = 58), and East North Central (n = 29) regions. Factors that could potentially contribute to this geographical clustering include differences in legionellosis surveillance across states and localities, higher population density and associated healthcare facilities or other large buildings with complex plumbing systems (CDC 2021), environmental factors (such as climate), age distribution of the population (Neil & Berkelman 2008), quality of source waters, corrosion control practices, aging infrastructure, and other characteristics such as pipe material or condition (USEPA 2016a). This information was not available from the outbreak reports. Uncertainty associated with these factors and others suggest a need for further examination and data collection during outbreaks where feasible.

From the WBDOSS, 81% (n = 149) of the PWSs providing water to the BWSs implicated in the outbreaks were determined using and comparing water system information in SDWIS databases or by the provided PWSIDs. The remaining 19% (n = 35) of PWSs were not identified due to insufficient information. The large systems (population >10,000 people) had more associated outbreaks (n = 123, 83%) than the small systems (population ≤10,000 people) (n = 26, 17%). Nationally, there are many more small than large PWSs, but the large systems were over-represented at 83% of drinking water-associated outbreaks in this dataset. To compare these outbreaks to the overall number of systems based on size, the 2017 SDWIS database included 142,985 small systems (97%) and 4,369 large systems (3%) (USEPA 2017). While the majority of PWSs are small systems, the majority of the population is served by large systems. Although there are more small PWSs than large PWSs, the large PWSs may serve many more BWSs. For example, if there are 10 small PWSs that serve 10 BWSs each, and one large PWS that serves 1,000 BWSs, more outbreaks would still be expected among BWSs served by the large PWS. The lower outbreak count for small systems may also be due to the types of BWSs that are present in such communities as well as the capacity of local public health agencies to identify and report legionellosis outbreaks. Large PWSs are more likely to serve hospitals, hotels, and apartment complexes since they typically have larger numbers of users. Additionally, the larger population associated with large systems could overrepresent those systems, such that larger populations lead to more cases which lead to more detections and more identified outbreaks.

CDC's WBDOSS outbreak data contain limited information on other water variables that could influence the presence of Legionella such as water temperature, water age, pipe age, the presence of biofilms and free-living amoeba, dead legs in the plumbing (sections of potable water piping systems through which water cannot flow that may lead to stagnation), corrosion control, cross connections, or disinfectant residual levels (although some residual level information was available, the timeframe in relation to the outbreak was unclear). Additional studies on the impact of these variables would further inform the impact of disinfectant residuals on Legionella. Furthermore, PWS demographic-related information was not available for this analysis, including the range of customer age and immune status. Increasing linkages to connect multiple data systems or streamlining measures to look at health and environmental data across systems would support further understanding of legionellosis outbreaks.

Of the 149 reported outbreaks with PWS information available, 85% were associated with BWSs receiving chlorinated water. Some of the identified PWSs did not have disinfectant residual type information available. Noteworthy in this analysis is that only 4% of the PWSs using surface water used chloramine as a distribution system residual disinfectant whereas in general (based on the UCMR 3 data) 35% (879/2527) of all surface water system entry points to the distribution system used chloramine as a residual disinfectant (USEPA 2016b). We found that 22% (11/56) of ground water PWSs used chloramines as a residual disinfectant whereas according to the UCMR 3, 9% (896/9838) of all ground water system entry points to the distribution system used chloramine. It is important to note that typical chloramine concentrations used in the distribution system are greater than those used for chlorine (USEPA 2019) and that chloramine can be more stable and persistent compared to chlorine in the distribution system (USEPA 2016a). This may confound the comparison of disinfectant type alone as an explanation for the observed differences in the outbreaks and suggests additional data collection and reporting, such as disinfectant types and levels, during outbreak investigations.

The settings of outbreaks suggest that larger buildings (with intermittent usage, increased water age, and complex premise plumbing) may be particularly at risk for legionellosis outbreaks regardless of the system size or disinfection type. Some outbreaks were associated with BWSs that had additional disinfectant treatment and with BWSs that had measurable disinfectant residual according to reported data, which could point to other factors related to Legionella growth and spread. The detection of disinfectant residual in the building water might not reflect conditions at the time of the exposure due to lag time between exposure and illness onset or outbreak detection; BWS water quality may have changed and may not reflect the conditions that led to the outbreak. Therefore, residual levels were not evaluated as part of this study.

Twenty-two PWSs had repeated outbreaks mostly occurring in different BWS categories and across multiple years. It is not possible to determine if the BWSs are the same when they are in the same category per the available data. One PWS (Mountain Census Division) using chlorine served BWSs implicated in nine outbreaks from 2001 to 2017 that included two types of exposure settings. Within the exposure setting category, it is unknown if the multiple outbreaks occurred at the same BWS over time. This one PWS was very large and may have served many BWSs. Of the 22 PWSs with multiple outbreaks, 20 PWSs used chlorine, one PWS used chloramine, and one PWS used chlorine during two outbreaks and chloramine during one outbreak. This is in comparison to the UCMR 3 entry point information previously cited that identified chlorine being used 1.9 times more than chloramine (USEPA 2016b). This distinct difference in incidence for legionellosis outbreaks between chlorinating and chloraminating PWSs shows a potential vulnerability in systems using chlorine that is generally consistent with Kool et al. (1999), Flannery et al. (2006), and LeChevallier (2019) findings on Legionella occurrence. However, as described previously and further below, we also observed outliers in this national trend showing a lower outbreak incidence among chloraminating systems than chlorinating systems.

Eleven PWSs associated with outbreaks in the South Atlantic Census Division used ground water and chloramine for residual disinfection. Previous research has also shown a greater occurrence of Legionella when chlorine is used, but our findings indicate that outbreaks can occur when chloramine is used. Perhaps most revealing is the occurrence of outbreak incidence in BWSs associated with large PWSs (>10,000 people). This could be due to large PWSs that are more likely to serve BWSs with complex plumbing or BWSs with higher susceptible populations, but further research is needed. Factors such as populations with an increased risk of legionellosis, large water systems (BWSs and PWSs), challenging water quality conditions (i.e., naturally occurring ammonia and/or high organic carbon in groundwater), and warmer water temperatures could also be responsible for this deviation from the national trend and warrants further attention. Disinfectant management alone should not be considered to control Legionella, as this is just one aspect of recommendations for water management programs and Legionella control.

This study is subject to several limitations. System misclassification was unlikely as 81% of outbreaks could be linked with reasonable confidence to a second data source that validated the classifications for critical system characteristics. Limitations in reporting practices may have led to misidentifying the drinking water system and their other characteristics. Additional limitations include the lack of information available on variables that may impact disinfectant residuals and the presence of Legionella, limited information on BWS conditions, and potential discrepancies in outbreak investigations.

The goal of this study was to evaluate characteristics related to PWSs that could impact the rise in drinking water legionellosis outbreaks. Our analyses suggest that more targeted opportunities to evaluate risks from Legionella, such as increased monitoring for Legionella or disinfectant residual, might exist within PWSs and BWSs served by surface water and ground water systems with certain characteristics including a high number of BWSs with complex plumbing. Measures to prevent legionellosis outbreaks might also benefit from targeted approaches. For example, large drinking water systems serving people who are at increased risk of legionellosis may need increased attention. Additionally, increased monitoring and reporting could increase our understanding of public health risks from Legionella and developing linkages for national databases would further our understanding of exposures and public health impacts. While this analysis focused on PWSs, there are tools specifically designed to mitigate risk factors in BWSs, including the CDC toolkit for controlling Legionella (CDC 2021). Building water management programs are also encouraged to minimize growth and transmission of Legionella (CDC 2021). Further examination of the relationships between the factors examined and other site-specific attributes is warranted.

The views expressed in this paper are those of the individual authors and do not necessarily reflect the views and policies of the U.S. Environmental Protection Agency or the Centers for Disease Control and Prevention.

Data cannot be made publicly available; readers should contact the corresponding author for details.