PREDICTORS OF LEGIONELLA OCCURRENCE IN DENTAL UNIT WATERLINES OF A HIGHLY COLONIZED DENTAL HOSPITAL

Introduction. Legionella is frequently detected in Dental Unit Waterlines (DUWLs). Although such a high occurrence is not necessarily associated with high risk for Legionnaire’s disease among patients and staff, it is prudent to monitor DUWLs for Legionella periodically. Since this procedure is long and expensive, surrogate markers are frequently


Introduction
Aquatic biofilm is a potential source of nos ocomial and communityacquired infections, due to the occurrence of opportunistic patho gens.More specifically, Pseudomonas aerugi nosa, well known in the hospital environment, is the cause of 10-20% of hospital acquired in fections (HAI), its extreme resistance to antibi otics is responsible for high mortality rate.Le gionella pneumophila causes 4-20% of cases of community-acquired pneumonia and has been ranked as the second-third most frequent cause of pneumonia requiring hospitalization 1,2 .A pe culiar characteristic of aquatic biofilms is that bacteria, organized in sessile phase, produce extracellular polysaccharides that help increase the resistance to antimicrobial agents 3 .In ad dition, bacteria organized in biofilms may sur vive for long periods.Indeed, P. aeruginosa can survive up to five months without nutrients and 10°C and L. pneumopila more than fifteen days without protozoa, essential for replication 4 .
Legionellaceae are distributed worldwide in low numbers in all types of aquatic habitats.Legionellae may enter the dental unit water lines (DUWLs) from the mains drinking water or from the air conditioning system.Legionella multiplication into DUWL requires the pres ence of aquatic biofilm, which, in turn, is pro moted by water stagnation, a condition typical of most dental units, which are frequently not used for more than 12 h daily, during workdays, and 60 h during weekends 5 .Legionellae also require amoebae and other protozoa, few nu trients and temperature ranging between 20°C and 45°C.Incidentally, intracellular growth protects these microorganisms from the bac tericidal action of biocides, chlorination or relatively elevated temperatures.The afore mentioned conditions are frequently found in dental healthcare settings and Legionella lev els in water samples from DUWLs as high as 102-105 colony forming units (CFU)/mL are reported 6 .Once established, legionellae can be come resident members of the aquatic biofilm for years 4,7 .All these characteristics imply that frequency of Legionella detection in DUWLs is necessarily high.For example, an Italian na tional survey reported a detection rate as high as 33.3% 8 , similar values are reported from US and UK (reviewed by Pankhurst and Coulter 6 ).However, these data come from hospital-based studies where the environmental conditions are not the same as those found in the office of the general dental practitioner (GDP) and inferring that Legionella is frequent in all dental health care settings is, therefore, misleading.Indeed, low detection rates are reported in surveys from GDPs' offices in UK (0.4%) 9 and EU (4%) 10 .
Although legionellae are more or less fre quently detected in DUWLs, a different issue is to assess the risk for infection associated with DUWLs contaminated by legionellae 11 .Indeed, there is no conclusive evidence that a patient has ever contracted legionnaire's dis ease inhaling L. pneumophila serogroup 1 com ing from DUWLs.One fatal case is reported of an 82-year-old woman, with no apparent risk factor for Legionnaire's disease, who died after having contracted L. pneumophila-associated pneumonia.During the incubation period (2-10 days before pneumonia development), the pa tient attended to two appointments at a dental practice.The investigation of L. pneumophila sources found no traces in the patient's home, while this microorganism was detected in DU WLs and cold water tap of the dental office at 103-104 CFU/L level 12 .This event raised the level of awareness and fear among dental healthcare workers and public health experts.However, the reported case does not provide conclusive evidence that Legionnaire's disease can be transmitted during dental treatment.Firstly, the reported data do not allow to assess whether L. pneumophila contamination of the dental healthcare setting preceded patient's in fection -and, therefore, was the source for the infection-or was consequent to the patient's in fection.In addition, the air conditioning system was not tested for L. pneumophila occurrence, while tap water resulted contaminated, thus, in the event that environmental contamination preceded patient's infection, it is not possible to assess whether the source of infection was the air conditioning system, the tap water or the DUWL.Finally, the level of L. pneumophila in water necessary for the development of Legion naire's disease is estimated to be at 106-108 CFU/L for showering events of 15 min with hot water 13 .Such an infective dose is almost one thousand times higher than the level previously reported in DUWLs.There is another fatal case of legionellosis developed by a dentist who died from pneumonia.Legionella dumoffii was detected in domestic water and in water from the dental office.It was, therefore, not possible to assess the source of infection 14 .There is no final evidence that prevalence of anti-Legionel la antibodies, a sign of past infection, is higher among dental healthcare workers than in the general population or in non-dental controls 9,15 .These data suggest that, in absence of further evidence, exposure to legionellae is frequent in dental healthcare settings, but the risk for infec tion and Legionnaire's disease is minimal.
These data do not mean that precautions are not necessary to control for Legionella in DUWL, particularly in public health settings where these and other environmentresistant microorganisms are likely to develop and spe cial care patients are frequently treated 16 .In deed, it was previously suggested that DUWL monitoring must follow two directions.One fi nalized at investigating oral fluid retraction and the consequent risk for blood-and air-borne infections, another at investigating the risk for biofilm-associated opportunistic pathogens 17 .However, routine Legionella and Pseudomonas monitoring in dental healthcare settings is not simple and cheap.This problem led to inves tigate for effective predictors or risk markers [18][19][20] .

Aim
The aim of this study was to investigate the factors associated with Legionella occurrence in DUWLs in a dental hospital where these 1238 microorganisms were detected for a 2-year pe riod 4 .

Material and Methods
The water distribution system of a dental hospital in Rome was tested four times.Name ly, wintertime 2002, summertime 2002, win tertime 2003, summertime 2003.At that time, DUWLs were connected to municipal water and hot water came from tanks located at the last floor of the building.The hot water tanks were the ideal environment where Legionella species probably survived for more than ten years, as it is suggested by the fact that these microorganisms were detected during two dif ferent surveys made at 10year interval.The hot tap water was, in turn, the source of contamina tion of dental offices.In 1992 the water system was disinfected through hyper-chlorination and legionellae were not detected for a long period.However, these microorganisms were detected again in 2002.It is important to highlight that no case of legionellosis among patients, dental and administrative staff was reported during this period 4,15 .The water system of the dental hospital was then revised, tanks were removed, most dental chairs were changed and water in DUWLs was periodically disinfected.After that, Legionella and Pseudomonas were no more detected in DUWLs and in tap water.
During the 2002-2003 survey, the level of residual chlorine, pH and total hardness ranged between 0.02-0.06mg/L, 7.45-7.81and 1.0-1.5 mmol/L, respectively.No water disinfection procedure was made during this period.
At every sampling occasion, 25 water sam ples (1.2 L) from DUWLs (air-water syringe, turbine and oral rinsing) from different dental chairs were aseptically collected into sterile bot tles avoiding that the tube touched the margins of the bottle.The residual disinfectant activity of chlorine was neutralized by sodium thiosul phate at a final concentration of 0.01% (w/v).The bottles were stored at 4°C, transported to the laboratory and processed within two hours.
One litre of every sample was used to test for Legionella occurrence and was filtered (poly amide filters, pore size 0.2 μm), re-suspended in 10 mL of the original sample, vortexed for 30 s, treated at 50°C for 30 min, diluted, plated in du plicate on Charcoal-Yeast Extract Agar (CYE; Oxoid, Ltd., Basingstoke, UK) supplemented with Legionella BCYE-α Growth Supplement (Oxoid) and incubated ten days at 37°C with 2.5% CO2.Colonies with typical Legionella morphology were sub-cultured in CYE and BCYE and only those not grown on CYE were serologically identified, by means of agglutina tion tests and counted.
Pseudomonas occurrence was tested using 100 mL of each sample which were filtered (nitrocellulose filters, pore size 0.45 μm), resuspended in 10 mL, vortexed for 30 s, diluted, plated on to Pseudomonas Agar Base with C-F-C supplement (Oxoid) and incubated three days at 30°C.The colonies were counted, subcul tured on Tryptone Soy Agar and biochemically identified by API 20 NE (API System S.A., La Balme Les Grottes, Montalieu Vercieu, France).
Total cultivable flora at 37°C and 22°C was assessed plating aliquots of sampled water on Plate Count Agar (Oxoid).
The explanatory variables used to inves tigate the factors associated with Legionella in the output water of DUWLs were, Pseudo monas, total viable flora at 37°C, total viable flora at 22°C and season.A bivariate analysis was initially made, explanatory variables were dichotomized into Pseudomonas occurrence (no vs. yes) and season (summertime vs. win tertime).As for total viable flora, we chose the threshold value suggested by the Ameri can Dental Association with the Statement on Dental Unit Waterlines (available at, http:// www.ada.org/1856.aspx):"water delivered to patients during nonsurgical dental procedures consistently contained no more than 200 CFU/ mL of aerobic mesophilic heterotrophic bacte ria at any point in time in the unfiltered output of the dental unit".Given the skewed distribution of Legionella values, the outcome variable also was dichotomized into Legionella occurrence (no vs. yes).Unadjusted odds ratios (ORs) were assessed and statistically analysed through χ2 test with Yates correction for continuity.Mul tiple logistic regression analysis was used to assess the ORs adjusted for covariates for Le gionella occurrence.On the basis of previously published data 4 it was likely that summertime was the stronger predictor of Legionella occur rence.Therefore, data were split into winter time and summertime and two separate models for each season were made.The goodness of fit of regression models was statistically analysed through likelihood ratio χ2 test and pseudo-R2.For statistical tests a significance level of 95% was chosen.

Results
Legionella was detected in 26% sampling occasions, specifically, 12/25 (48%) times in summer 2002 and 14/25 (56%) times in sum mer 2003, it was never detected in wintertime.Psuedomonas was detected in 24% sampling occasions, while total viable flora level was higher than the ADA threshold in 33% samples at 37°C and 29% samples at 22°C.Legionella species serologically identified were L. pneu mophila serogroup 6 and L. micdadei, while P. aeruginosa was the only Pseudomonas spe cies biochemically identified.Legionella levels

Discussion
The present study is one of the papers pre sented at the workshop "Advances in Infection Epidemiology and Control in Dental Healthcare Settings", Department of Public Health and In fectious Diseases, Sapienza University, Rome, Italy on February 9th, 2013 2127 .This is the first study which sought to in vestigate the probability to detect legionellae in ranged between 0 and 200 CFU/L, Pseudomo nas between 0 and 1,040 CFU/mL, total viable flora at 37°C between 0 and 1,356 CFU/mL and total viable flora at 22°C between 0 and 988 CFU/mL (data not in Table ).
Table 1 displays the bivariate associations between the investigated factors and Legionel la occurrence.Total viable floras at 37°C and 22°C were not significantly associated with le gionellae, while absence of Pseudomonas and summertime were strongly associated with this condition, with unadjusted ORs of 25 and 108, respectively.The logistic regression analysis provided a highly predictive model (pseudo-R2=0.45),but all explanatory variables were not significantly associated with detection of le gionellae (Table 2).The estimated coefficients were used to assess the probability to detect Legionella in the most favourable conditions, that is, in summertime, with Pseudomonas un detected and levels of total viable flora at 37°C and 22°C greater than the ADA limit of 200 CFU/mL.In these conditions, the probability was 29.3% (data not in Table ).
the output water of DUWLs in a building which was highly contaminated by these microorgan isms.Indeed, the hot water tanks contained >200 CFU/L legionellae and hot tap water was contaminated almost everywhere in summer time.Very interesting, legionellae were never detected in wintertime, even in the hot water tanks 4 .One would expect that these favourable conditions for Legionella development would result in ubiquitous detection in DUWLs which were not routinely disinfected.Nevertheless, only 48% (in 2002) and 56% (in 2003) of the dental units resulted contami nated in summertime, when legionellae devel oped and flourished throughout the hospital.In addition, the detected levels ranged between 4 and 200 CFU/L with a mean level of 31.2CFU/L (data not in Table ), a value lower than the estimated infectious dose for showering events, which could be assimilated to dental treatments for Legionellacontaining aerosol production and inhalation 13 .Indeed, during the period of the study no case of infection was reported among staff and patients, while in 1992 antiLegionella antibodies, suggestive of past infection, were not detected among the dental healthcare workers, but were detected in 4% of the administrative staff 15 .Although legionellae were never detected in DUWLs contaminated by Pseudomonas and were never detected in wintertime, these variables were not significantly predictive of Legionella oc currence in DUWLs.The addition of the two other microbiological indices, that is, total vi able floras at 37°C and 22°C >200 CFU/mL did not increase the predictive power of the model and it was only possible to infer that in output water of DUWLs without Pseudomonas, with high levels of aerobic heterotrophic bacteria and in summertime, that is, the most favourable Legionella development promoting conditions, the probability to detect these microorganisms was of 29%.
In previous studies, Zanetti and colleagues found that in DUWLs with L. pneumophila, cul tivable floras at 37°C and 22°C showed lower levels than in DUWLs where legionellae were not detected.Nevertheless, this association was not statistically significant and, therefore, was not predictive of L. pneumophila detection 18 .Göksay and colleagues did not detect legionel lae in DUWLs, but reported high Pseudomonas detection rate (24%), the same values reported in the present survey, and of total viable flora at 27°C higher than the ADA threshold (97%), suggesting that, as in the present study, Psu domonadaceae could prevent Legionelleaceae growth in waterlines 19 .Similar results, that is, high total viable flora and high Pseudomo nas detection level associated with Legionella undetected in water reservoirs of dental units was reported by Szymańska and Sitkowska 28 and by Veronesi and colleagues, who reported that 24% dental units were contaminated by Legionella spp., 54% by P. aeruginosa, while only 0.6% were contaminated by both micro organisms 29 .An inverse association between L. pneumophila and P. aeruginosa also was re ported by Aprea and colleagues.These authors found legionellae and P. aeruginosa in 76% and 33% water samples from DUWLs of a dental hospital highly contaminated by Legionella, re spectively 20 .

Conclusion
The reported studies corroborate the results of the present survey that despite the more or less significant associations between microbio logical parameters and legionellae in the out put water of DUWLs, none of these parameters seems to be an effective predictor of Legionella detection and, most importantly, none of them could be used as risk marker for legionellosis, Legionella infection, or even Legionella trans mission.This is an important result which raises doubts regarding the use of the level of aerobic heterotrophic bacteria as risk marker of water quality of DUWLs.Indeed, the number of sur veys reporting high levels of these bacteria in DUWLs is uncountable.Yet, neither cultivable heterotrophic bacteria are effective predictors of legionellosis or P. aeruginosa infection, as there are no reported cases in the world litera ture, nor they per se pose any risk for infection, once again because there are no reported cases in the world literature.
These data suggest that the risk for opportu nistic pathogen infection or infectious disease transmission mediated by water from DUWLs is evaluable only using specific tests, such as legionellae detection at high levels.

Table 1 .
Unadjusted ORs for Legionella occurrence (95% confidence intervals between parentheses) in DUWLs from a highly contaminated dental hospital.

Table 2 .
Logistic regression analysis of the association between the investigated variables and probability of Legionella occurrence in DUWLs from a highly contaminated dental hospital.