Active Bacterial Core Surveillance for Legionellosis — United States, 2011–2013
Vol. 64, No. 42
October 30, 2015
|PDF of this issue|
Active Bacterial Core Surveillance for Legionellosis — United States, 2011–2013
WeeklyOctober 30, 2015 / 64(42);1190-3
1,2; , MPH1; , DO1; , MPH1; , MPH3; , PhD4; , MD5; , MD6; , MPH7; , MPH8; , MD9; , MPH10; , MPH11; , MPH12; , MPH13; , MD1, MD
During 2000–2011, passive surveillance for legionellosis in the United States demonstrated a 249% increase in crude incidence, although little was known about the clinical course and method of diagnosis. In 2011, a system of active, population-based surveillance for legionellosis was instituted through CDC's Active Bacterial Core surveillance (ABCs) program. Overall disease rates were similar in both the passive and active systems, but more complete demographic information and additional clinical and laboratory data were only available from ABCs. ABCs data during 2011–2013 showed that approximately 44% of patients with legionellosis required intensive care, and 9% died. Disease incidence was higher among blacks than whites and was 10 times higher in New York than California. Laboratory data indicated a reliance on urinary antigen testing, which only detects Legionella pneumophila serogroup 1 (Lp1). ABCs data highlight the severity of the disease, the need to better understand racial and regional differences, and the need for better diagnostic testing to detect infections.
Legionellosis is acquired by inhalation of Legionella bacteria in aerosolized water. The two main clinical syndromes associated with legionellosis are Legionnaires' disease, a severe form of pneumonia, and Pontiac fever, a milder, self-limited illness without pneumonia. National Notifiable Diseases Surveillance System (NNDSS) data reported during 2000–2011 demonstrated a 249% increase in crude incidence of legionellosis in the United States, from 0.39 to 1.36 cases per 100,000 persons (1,2). NNDSS is a passive public health reporting system that relies on laboratories and physicians to report cases and does not capture testing method, clinical course, or information about underlying medical conditions. Most of what is known about the clinical course and outcomes of legionellosis comes from published reports of case series and outbreaks, which might not be representative of the overall epidemiology of legionellosis. In 2011, active surveillance for legionellosis was initiated through CDC's ABCs program to describe the incidence and epidemiologic and clinical characteristics of legionellosis in a large, geographically diverse population. Data from the first 3 years of ABCs surveillance were analyzed and compared with those collected through the NNDSS passive legionellosis surveillance system.
ABCs, part of the Emerging Infections Program network of CDC, is an active, laboratory- and population-based surveillance system at 10 sites in the United States. A catchment area is found in every region of the United States (statewide in Connecticut, Maryland, Minnesota, New Mexico, and Oregon, and in selected counties in California, Colorado, Georgia, New York, and Tennessee), covering a population of approximately 36 million persons (http://www.cdc.gov/abcs/methodology/index.html). NNDSS covers the entire U.S. population; however, NNDSS relies on laboratories and physicians to report legionellosis cases to local or state public health authorities, who in turn, transmit the data to CDC. Unlike ABCs legionellosis surveillance, NNDSS does not include testing method, clinical course, or information about patients' underlying medical conditions.
ABCs personnel actively contacted laboratories that serve persons who live in the surveillance catchment areas to identify legionellosis cases that were confirmed by a laboratory test during January 1, 2011–December 31, 2013. For surveillance purposes, ABCs defined a confirmed case of legionellosis as the isolation ofLegionella from respiratory culture, detection of Legionella antigen in urine, or seroconversion (a more than fourfold rise in antibody titer between acute and convalescent sera) to Lp1. The NNDSS case definition differs slightly: cases must have an illness that is clinically compatible with legionellosis in addition to the laboratory criteria mentioned for the active system.
ABCs personnel reviewed medical records for all cases using a standardized form to collect information on demographics, underlying medical conditions, diagnostic tests performed, clinical courses, and outcomes. Race was recorded from the medical record and categorized as white, black, or other (American Indian/Alaska Native, Asian, Native Hawaiian/Other Pacific Islander). Missing race data (approximately 8%) were imputed using sequential regression imputation. Incidence was calculated using 2013 U.S. postcensal population estimates. NNDSS data were obtained from the Summary of Notifiable Diseases (2).
ABCs identified 1,426 legionellosis cases during 2011–2013, for an incidence of 1.3 cases per 100,000 population over the 3 years. For 2011, 2012, and 2013, the rates were 1.3, 1.1, and 1.4 cases per 100,000, respectively. In 2012, the most recent year that NNDSS rates were available, legionellosis incidence was the same as that found in ABCs (1.1 per 100,000 population), and the number of cases reported to NNDSS and projected to the U.S. population from ABCs were similar (3,688 cases versus 3,362 cases, respectively). ABCs incidence rates in whites (1.0 per 100,000) and blacks (1.5 per 100,000) were similar to NNDSS rates in whites (1.0 per 100,000) and blacks (1.4 per 100,000); 17% of cases reported to NNDSS were missing race categorization. Rates increased with age. In ABCs, rates per 100,000 population (by age category) were 0.4 (<50 years), 2.5 (50–64 years), 3.6 (65–79 years), and 4.7 (≥80 years). Legionellosis incidence by ABCs sites varied during 2011–2013, from 0.4 per 100,000 population in California to 4.0 per 100,000 population in New York. The three highest incidence sites during this period, New York, Maryland, and Connecticut, are all located in the Northeast or Mid-Atlantic United States (Table 1). This is consistent with NNDSS data that show a higher incidence of legionellosis in these regions compared with other regions (1).
Among cases identified during 2011–2013, 79% occurred in persons aged >50 years, 65% were in males, and 72% of patients were white (Table 2). Seven percent of patients were residents of health care facilities (e.g., acute care hospitals, long-term care facilities, or long-term acute care facilities) during at least part of the period they were likely exposed to Legionella. Current smoking was the most common underlying condition (38%), followed by diabetes (30%), chronic obstructive pulmonary disease (16%), immune compromise (14%), and former smoking (14%). Almost all patients with legionellosis (1,354 [95%]) had a diagnosis of pneumonia; 98% were hospitalized, 44% were admitted to an intensive care unit (ICU), and 27% required mechanical ventilation. The median duration of hospitalization was 7 days. Overall, 134 (9%) patients with legionellosis died (Table 2).
Among all patients, 1,300 (91%) received a diagnosis of legionellosis on the basis of urine antigen testing, which only detects Lp1 species (Table 3). Cultures were performed on respiratory specimens from 330 (23%) patients. Among these, specimens from 140 patients (42%, representing 10% of all cases) tested positive for Legionella, 112 (80%) of which were Lp1. Specimens from 13 (9%) of these 140 patients were identified as non-Lp1, and the remainder (15) had Legionellaspecies that were not further identified.
The first 3 years of active population-based surveillance for legionellosis demonstrated similar rates of disease compared with the rates detected through passive surveillance, including regional and racial/ethnic differences (1,2). However, the data from ABCs provided additional information on clinical history, disease severity, and diagnostic testing. The reported incidence in both systems is likely an underestimate because of reliance on urine antigen testing, which only detects Lp1. In addition, not all patients at risk for legionellosis are likely tested by any diagnostic method.
The racial/ethnic differences in legionellosis incidence might reflect disparities in the prevalence of underlying medical conditions, socioeconomic determinants, and environmental exposures (3). Geographic differences in incidence might be influenced by regional differences in environmental exposures, testing practices, or the prevalence of underlying medical conditions. Future analyses will take into account underlying conditions, area-level socioeconomic status, and location of residence when calculating rates, to determine whether racial/ethnic and geographic disparities persist.
Approximately 40% of patients with legionellosis required ICU admission, and 9% died. Previous estimates of disease severity did not include rates of ICU admissions, and reported death rates that ranged from <1% in community settings to >60% in nosocomial outbreaks (4,5). Conditions known to be risk factors for legionellosis, including smoking, alcohol abuse, diabetes, and immune compromise were relatively common among legionellosis patients (6). Future ABCs analyses will determine actual rates by health conditions using population-based denominators.
The findings in this report are subject to at least three limitations. First, ABCs is population-based, but covers only a portion of the U.S. population, so results might not be generalizable to the entire population. However, the similar rates identified in NNDSS suggest that ABCs is representative. Second, urine antigen testing, which only detects Lp1 infections and is approximately 70%–90% sensitive (7), was the most common method for detecting legionellosis cases. Therefore, some cases of legionellosis likely were missed. Finally, in addition to missing cases because of the test sensitivity, other cases likely were missed because patients with legionellosis were not tested for Legionella bacteria by any diagnostic method. Therefore, the rates reported likely represent an underestimate of the actual disease burden of legionellosis.
These findings highlight the importance of developing more sensitive laboratory tests for legionellosis because proper diagnosis is needed for treatment and public health action. In 1998, the proportion of patients who received a diagnosis of legionellosis on the basis of urine antigen testing was 69% (8); during 2011–2013, this proportion had increased to >90%. With fewer patients being tested by culture, the likelihood that more non-Lp1 cases are being missed exists. Development of molecular-based tests that can detect more species and serogroups from respiratory specimens will likely improve detection (9). Because up to half of patients with legionellosis might not produce sputum, more sensitive urine diagnostics are needed (10). Until such diagnostic tests are developed, validated and implemented, obtaining respiratory specimens for culture from persons suspected to have legionellosis infection is important for diagnosis and initiation of appropriate treatment. Clinical cultures also are important to establish linkage between individual patients and environmental sources in outbreak settings. The underlying reasons for geographic and racial differences in legionellosis incidence need further exploration, which can be done through additional analyses in ABCs.
1Division of Bacterial Diseases, National Center for Immunization and Respiratory Diseases, CDC; 2Epidemic Intelligence Service, CDC; 3Maryland Department of Health and Mental Hygiene; 4New York State Department of Health; 5Tennessee Department of Health; 6Vanderbilt University School of Medicine, Nashville, Tennessee; 7California Emerging Infections Program; 8Connecticut Department of Public Health; 9Oregon Public Health Division; 10Georgia Emerging Infections Program, Atlanta, Georgia; 11New Mexico Emerging Infections Program; 12Minnesota Department of Health; 13Colorado Department of Public Health and Environment.
Corresponding author: Gayle Langley, firstname.lastname@example.org, 404-639-8092.