Environmental Cultures and Hospital-Acquired Legionnaires' Disease: A 5-Year Prospective Study in 20 Hospitals in Catalonia, Spain

Abstract Objective: To determine whether environmental cultures for Legionella increase the index of suspicion for legionnaires' disease (LD). Design: Five-year prospective study. Setting: Twenty hospitals in Catalonia, Spain. Methods: From 1994 to 1996, the potable water systems of 20 hospitals in Catalonia were tested for Legionella, Cases of hospital-acquired LD and availability of an “in-house” Legionella test in the previous 4 years were assessed. After the hospitals were informed of the results of their water cultures, a prospective 5-year-study was conducted focusing on the detection of new cases of nosocomial legionellosis and the availability and use of Legionella testing. Results: Before environmental cultures were started, only one hospital had conducted active surveillance of hospital-acquired pneumonia and used Legionella tests including Legionella urinary antigen in all pneumonia cases. Only one other hospital had used the latter test at all. In six hospitals, Legionella tests had been completely unavailable. Cases of nosocomial LD had been diagnosed in the previous 4 years in only two hospitals. During prospective surveillance, 12 hospitals (60%) used Legionella urinary antigen testing in house and 11 (55%) found cases of nosocomial legionellosis, representing 64.7% (11 of 17) of those with positive water cultures. Hospitals with negative water cultures did not find nosocomial LD. Conclusions: The environmental study increased the index of suspicion for nosocomial LD. The number of cases of nosocomial LD increased significantly during the prospective follow-up period, and most hospitals began using the Legionella urinary antigen test in their laboratories.

Since Legionella pneumophila was first identified as the causative agent of legionnaires' disease, 12 this microorganism has been associated with both community-acquired and nosocomial pneumonia. More than 300 reports of nosocomial Legionella pneumonia have appeared in peer-reviewed journals and public health bulletins. It is now recognized that many cases of nosocomial Legionella pneumonia go undiagnosed. Thus, many authorities actually agree that Legionella could be an important cause of nosocomial pneumonia. 3 " 8 However, there is no general consensus regarding the prevention of legionellosis in hospitals. The position of the Centers for Disease Control and Prevention (CDC), 9 which recommends environmental investigation only after detecting cases of nosocomial Legionella pneumonia, is controversial. The CDC argues that negative environmental cultures may give a false sense of security and that legionnaires' disease could thereby be underestimated by physicians. On the other side, the Allegheny County Health Care Department and the Maryland Scientific Working Group have recommended routine environmen-tal cultures for Legionella in acute care hospitals 1011 ; if the microorganism is detected in the water, the Legionella test should be available in the laboratory and applied to all cases of hospital-acquired pneumonia. These groups argue that knowing that the water is colonized by Legionella, clinicians are more aware of legionnaires' disease and, consequently, more Legionella tests are requested and the possibility of achieving the diagnosis of nosocomial legionellosis increases. The CDC has recommended periodic cultures for Legionella in water samples from high-risk wards housing patients undergoing stem cell transplant, even in the absence of previous cases of legionnaires' disease. 12 In a previous study, L. pneumophila was isolated from potable hot water systems in 17 of 20 hospitals in Catalonia. 13 After all of the hospitals had been informed about the environmental data collected, a 5-year follow-up study was performed to detect cases of hospital-acquired legionnaires' disease and incorporate Legionella tests, especially Legionella urinary antigen, in the laboratories of the hospitals. Cases of hospital-acquired legionnaires' disease before and after the copper-silver ionization system was implemented in September 1999 in hospital 6.

M E T H O D S Sites
Twenty hospitals in Catalonia, an autonomous region of 32,000 km 2 located in northeastern Spain, were studied. Five of these hospitals were located in the city of Barcelona. The sizes of the hospitals ranged from 200 to 2,000 beds. Nineteen were acute care hospitals and one was a psychiatric hospital.

Environmental
Study From November 1994 to April 1996, 196 water samples were taken from the 20 hospitals. The water samples were concentrated, decontaminated by acid treatment, and inoculated on selective modified Wadowsky-Yee buffered charcoal yeast extract plates (Oxoid Ltd., Basingstoke, Hampshire, United Kingdom) in duplicate. Isolates of Legionella were identified by demonstrating growth on buffered charcoal yeast extract but not on sheep blood agar plates (bioMerieux, Paris, France) and by Gram staining. L. pneumophila was determined by the Monofluo L. pneumophila IFA Test Kit (Genetic Systems Corp., Redmond, WA) and then classified into two groups (serotype 1 or serotypes [2][3][4][5][6][7][8][9][10][11][12][13][14] according to the reaction with the immunoagglutination serotyping by the MicroScreen Legionella Latex Kit (Microkit Iberica, Madrid, Spain). A complete description of the points sampled and the methodology applied may be found elsewhere. 13

First
Questionnaire At the time of the environmental sampling, a questionnaire was completed by the staff involved in the control of nosocomial infection. Information about surveillance programs for nosocomial pneumonia, tests available for Legionella in the hospital laboratory, and cases of nosocomial legionnaires' disease diagnosed in the 4 years prior to sampling was collected.

Information
to Hospitals At the end of 1996, each hospital received a written report of its water culture results. In addition, a general information session was held in one of the hospitals and the heads of most of the hospitals studied attended. By request, this session was also held in some of the other hospitals. During these sessions, the results of the study were presented with emphasis on the problem of nosocomial legionnaires' disease. These results were also reported at the community, state, and international levels.

Second
Questionnaire From December 1997 to December 2001, the information requested by the first questionnaire was collected annually.
The timing of the study and the data requested on the first and second questionnaires are detailed in the figure and Table 1, respectively.

Definitions
The diagnosis of pneumonia caused by L. pneumophila was based on the following: isolation of L. pneumophila from respiratory samples, a fourfold increase in antibody titers to greater than 1/128 in the paired serum samples, or positive results for urinary antigen. Nosocomial legionnaires' disease was as defined by the CDC. 9  The environmental study was conducted a few months after disinfection procedures.

Environmental
The Legionella urinary antigen test became available at the end of period 1.

First Questionnaire
According to the first questionnaire, only one hospital (hospital 6) had been performing active surveillance of hospital-acquired pneumonia in the entire institution (ie, intensive care unit and general wards). This same hospital used tests for Legionella, including Legionella urinary antigen, in all cases of hospital-acquired pneumonia. Only one other hospital (hospital 3) had the Legionella urinary antigen test. No Legionella tests had been available in the laboratories of six hospitals. Cases of nosocomial legionnaires' disease had been diagnosed in the 4 years prior to the questionnaire in only two centers (hospitals 3 and 6).

Prospective Follow-Up Study
According to the second questionnaire, the situation of active surveillance of hospital-acquired pneumonia had not changed. However, in 12 (60%) of the 20 hospitals the Legionella urinary antigen test was available in house and in 11 (55%) of the 20 hospitals cases of nosocomial legionellosis had been diagnosed, representing 64.7% (11 of 17) of those that had shown positive results on the environmental study and 63.6% (7 of 11) that had more than 30% of the peripheral points colonized by Legionella. The hospitals with negative results on the environmental study did not diagnose cases of nosocomial legionnaires' disease. One of the hospitals that reported cases on the first questionnaire (hospital 3) did not diagnose cases during the prospective follow-up period. Table 2 provides the results of the environmental study, the percentage of peripheral points positive for Legionella, and the data regarding the availability of the Legionella urinary antigen test and the number of cases of nosocomial legionnaires' disease diagnosed during both periods for each hospital.

DISCUSSION
L. pneumophila can infect anyone, but it has a clear predilection for immunosuppressed patients. Patients who have neoplasms, are receiving immunosuppressive therapy or organ transplants, or are elderly with chronic lung disease are the most susceptible hosts. 1415 Several studies have demonstrated that Legionella is a frequent inhabitant of hot water systems in hospitals, 16 " 19 and thus legionnaires' disease should be a common cause of pneumonia in this setting. However, the prevalence of nosocomial legionellosis remains unknown. 20 When environmental cultures are performed routinely, hospitals report more cases of nosocomial legionnaires' disease. A Canadian study found that the presence of Legionella in the water supply led to a significant number of hospitals discovering cases of nosocomial legionnaires' disease. However, the study was performed for only 9 months. 21 If the study had been performed for a longer period, it might have uncovered more cases in the uninvolved hospitals. In three studies conducted in Pittsburgh, the discovery of a contaminated water supply led to the subsequent discovery of nosocomial legionnaires' disease. 2224 However, the number of hospitals was small: the first study had two hospitals, 22 the second study had three hospitals, 23 and the third study included four hospitals. 24 Our study of 20 hospitals is the largest to correlate the notification of contamination of hospital water with the subsequent discovery of nosocomial legionnaires' disease. Furthermore, our study lasted longer (5 years) than any such study conducted to date.
The results of our environmental study increased the index of suspicion for nosocomial legionnaires' disease among clinicians. This is supported by the fact that the number of cases of nosocomial legionnaires' disease increased significantly during the prospective follow-up period and many hospitals later used the Legionella urinary antigen test in their laboratories. Thus, although only two hospitals (10%) had reported cases of nosocomial legionnaires' disease previously, 55% did so in the prospective follow-up period (rising to 65% for hospitals colonized with Legionella). We also confirmed that hospitals not colonized by Legionella found no cases of legionnaires' disease. This is consistent with five other prospective studies in which no cases of nosocomial legionellosis were observed in any hospital that did not have a contaminated water supply. 20 Many of the hospitals included in our study have since incorporated the use of Legionella urinary antigen testing and this has undoubtedly contributed to the increase in the number of cases of nosocomial legionellosis. Approximately 66% of the hospitals using the Legionella urinary antigen test had diagnoses of nosocomial legionnaires' disease. This method has been shown to be highly useful for diagnosing legionnaires' disease, with a sensitivity ranging from 60% to 100% and a specificity of 100%. 25 ' 26 However, it is mainly useful for the diagnosis of L. pneumophila serogroup 1. Hospital 14 of our study is a clear example of this. The environmental study demonstrated the presence of L. pneumophila serogroups 2-14, yet all of the cases diagnosed were attributed to L. pneumophila serogroup 6 and were diagnosed through sputum cultures. When environmental cultures are positive for species other than L. pneumophila or for L. pneumophila serogroups other than serogroup 1, specialized culture techniques using selective media should be reinforced to identify legionnaires' disease. 2227 Despite the communication with the other hospitals and the intention of performing active surveillance for nosocomial legionnaires' disease, only one hospital (hospital 6) was successful in applying the Legionella diagnostic test to almost all of the cases of hospital-acquired pneumonia. This hospital also discovered the greatest number of cases of legionnaires' disease. These limitations are important because they reinforce the possibility of underdiagnosis of nosocomial legionellosis in our study. The fact that no further environmental survey has been performed since the 1994-1996 period is another limitation of this study. It is unknown whether hospitals with negative environmental cultures became colonized by Legionella or whether those colonized by Legionella had negative cultures later. Complete elimination of Legionella from a water system is difficult to achieve, even using complementary disinfection methods. 28 On the other hand, hospitals that were colonized by Legionella continued reporting cases of nosocomial legionnaires' disease in the prospective follow-up study, indicating that they remained colonized. In September 1999, a copper-silver ionization system was installed in hospital 6 after many years of failure using the hyperchlorination and superheat-and-flush methods. Thereafter, the number of cases of nosocomial legionnaires' disease dropped significantly in this hospital (Figure). Currently, the copper-silver ionization system is in use in five of the hospitals included in this study.
There are increasing reports of unrecognized cases of nosocomial legionellosis in hospitals colonized by Legionella over long periods. 67 In these hospitals, patients may be incorrectly treated and the mortality rate may thus be high.
Discovery of a single case of nosocomial legionnaires' disease is an important sentinel of the possibility of additional undiscovered cases. 7 Thus, the search for these microorganisms in hospital water systems seems irrefutable. If they are found, measures of primary prevention should be applied, ranging from the introduction of adequate diagnostic tests and the incorporation of antibiotics active against Legionella in the therapeutic protocols to the implementation of complementary disinfection measures for the potable water system of the hospital.