Investigating an Airborne Tularemia Outbreak, Germany

Infectious aerosols can contribute to the transmission of tularemia during processing of dead hares.

I n the last 50 years, few laboratory-confi rmed outbreaks of airborne tularemia have been described. They include outbreaks in workers in sugar cane factories in Ukraine, the Czech Republic, and Austria (1-3); farmers in Sweden and Finland (4,5); and residents of the island of Martha's Vineyard, Massachusetts, USA (6). Small clusters and outbreaks with probable common source exposure may have been associated with disturbance of contaminated animal carcasses (7)(8)(9) and dogs with contaminated fur shaking themselves inside houses (10,11). In Germany, tularemia is rare, with only 184 cases reported during 1955-2004 (12). Starting in late 2004, tularemia reemerged, causing repeated outbreaks in nonhuman primates at different animal facilities in central Germany (13), followed by rising numbers of human cases in 2005, 2007, and 2008. Here we report a point-source outbreak of tularemia among participants of a hare hunt in Hesse, Germany, in November 2005.

The Outbreak
On December 1, 2005, Darmstadt health authorities were notifi ed of a laboratory-diagnosed case of tularemia. The patient had participated in a hare (Lepus europaeus) hunt on October 29, 2005, and cut 1 fi nger while disemboweling and skinning hares. On November 2, the patient became ill with fever >40°C, axillary lymphadenopathy, arthralgia, and headache. Initially treated as an outpatient, he was hospitalized November 21 for progressive lymphadenitis and recurrent fever; Francisella tularensis infection was diagnosed by lymph node biopsy and specifi c antibodies. After the Darmstadt-Dieburg Public Health Authority received notifi cation of this index case, that agency initiated an outbreak investigation.
On October 29, 2005, 29 hunters and 10 beaters, who drove hares out of areas of cover, participated in the hunt. Sixty-three hares were shot. Some hares were disemboweled where they were shot; most were later disemboweled and rinsed with a water hose at a hunting lodge. Disemboweled hares were transported to a game chamber and skinned the next day.

Patients
All participants of the hunt were offered serologic testing. From December 3, 2005, through March 3, 2006, serum was obtained from 29 participants, and DNA was extracted from an affected lymph node of the index casepatient.
Two different case defi nitions were used. Symptomatic participants of the hunt who fell ill during October 30-November 12, 2005, were defi ned as confi rmed case-patients if they had a single high titer of F. tularensis-specifi c antibodies. We defi ned a probable case-patient as either an asymptomatic hunt participant with a single high titer of F. tularensis-specifi c antibodies or a hunt participant who had signs and symptoms suggestive of F. tularensis infection from October 30 through November 12, 2005, but no laboratory confi rmation.

Retrospective Cohort Study
Starting December 13, 2005, we interviewed hare hunt participants using a standardized questionnaire to determine demographic and clinical details and risk factors for F. tularensis infection. For statistical analysis, we combined probable and confi rmed cases; all participants who did not fulfi ll a case defi nition were included as controls. All analyses were performed with Intercooled STATA 10.0 for Windows statistical software (StataCorp, College Station, TX, USA). Fisher exact test was used to analyze the relationship between categorical variables and the 2-sample Wilcoxon rank-sum test used to analyze the relationship between numeric data and the categorical outcome.

Environmental Investigation
Starting in early December 2005, we visited the outbreak area 3 times. We obtained data on elevation, regional mean annual air temperature, precipitation, and sunshine hours  from the Federal Meteorological Service (Offenbach am Main, Germany). Water samples were obtained from a small creek near the hunting lodge and from the water hose used to rinse disemboweled hares. Additionally, 28 samples were taken at the game chamber (Table 1; Figure 1). All samples were stored at 4°C.
Deep frozen parts from 12-14 hares shot during the initial hunt on October 29 were recovered from different households. Additional animals were shot in the same hunting area on December 6, 2005, and January 7 and 14, 2006. In January 2006, all frozen samples were transported on dry ice to a microbiologic laboratory.

Direct Detection of F. tularensis
All animal samples were stored at −20° C until preparation for PCR, antigen detection, or culture. Specimens of spleens, livers, bone marrow, and muscle tissue of the animals were homogenized as described recently (13) and tested for F. tularensis-specifi c lipopolysaccharide (LPS) using a capture ELISA (14) or an immunochromatographic column assay (ABICAP, Senova, Jena, Germany). Purifi ed DNA was prepared from tissue homogenates, blood, and water samples and from fl uids obtained during thawing of the hare samples by using the QIAamp Tissue kit (QIA-GEN, Hilden, Germany).
PCR amplifi cation and product detection were performed in a LightCycler instrument (Roche, Mannheim, Germany) by using a commercially available real-time PCR kit (TibMolBiol, Berlin, Germany) for the detection of a specifi c nucleotide sequence within the 16S rRNA gene of F. tularensis (15). Additionally, real-time PCR protocols targeting the tul4 (16), iglC, ISFtu2, or fopA gene were performed (17). Each run included positive and negative controls. For subspecies identifi cation, a conventional PCR protocol employing primers fl anking the RD1 region of F. tularensis was used (18). To prove the presence of F. tu- Water samples; swab samples; and spleen, liver, and bone marrow homogenates were cultured on cysteine heart agar supplemented with 9% sheep blood, Columbia blood agar, McConkey agar, and modifi ed Thayer-Martin medium containing antimicrobial drugs (Merck, Darmstadt, Germany). Culture plates were incubated at 37°C for 10 days and investigated daily for bacterial growth (13). Serum from 29 participants was examined for F. tularensisspecifi c anti-LPS antibodies by a qualitative screening ELISA and confi rmed by immunoblot (18).

Patients: Clinical Characteristics and Laboratory Results
Characteristics of 9 hare hunt participants met the definition of a confi rmed case; 2 participants had characteristics that met the defi nition of a probable case. The median age of case-patients was 55 years (range 11-73 years); all were male. Illness onsets ranged from November 2 through November 7 ( Figure 2). One probable case-patient (no. 3 in Figure 2) fell ill with high fever, myalgia, and clinically diagnosed bilateral pneumonia; he had chronic heart failure and died during the second week of illness despite treatment with moxifl oxacin. Neither specifi c antibodies nor F. tularensis-specifi c DNA could be detected in a serum specimen taken 8 days after illness onset. The second probable case-patient was asymptomatic but had high levels of anti-LPS-specifi c antibodies (immunoglobulin [Ig] M 32,000; IgA 32,000; IgG 8,000), suggesting a recent subclinical infection. Antibody titers of the 9 confi rmed case-patients ranged from 64,000 to >256,000 (negative <500). All 9 showed a specifi c IgG, IgA, and IgM immune response, all were medically attended, and 1 was hospitalized. They reported fever >38.5°C (range 38.5°C-40.6°C) (8 persons), chills (6), headache (5), weight loss (5), myalgia (5), enlarged lymph nodes (3), and coughing (1). None reported sore throat or pneumonia. Two case-patients had an ulceroglandular form of tularemia: the index patient (case-patient 1 in Figure 2) had cut his fi nger while skinning hares; the other (case-patient 6) had scratched his fi nger before the hunt.
PCR of an affected lymph node specimen and sequencing of the amplifi cate indicated Francisella infection. Realtime PCR (targets 16S rRNA gene, tul4 gene) confi rmed the presence of F. tularensis-specifi c DNA within the sample. Partial amplifi cation of the RD 1 region identifi ed a 923-bp fragment considered to be specifi c for subspecies holarctica (18). Several attempts to amplify F. tularensis DNA fragments from serum of case-patient 3 were unsuccessful.

Retrospective Cohort Study
The analysis included data for 10 of the 11 case-patients and all 28 controls (Table 2). Presence within 5 meters of where disemboweled hares were rinsed was the risk factor most strongly associated with infection. Case-patient 3, who died, was not included in the cohort study; however, he was reported to have disemboweled hares within 5 meters of the area where disemboweled hares were rinsed. Hares were disemboweled and rinsed at the hunting lodge during the lunch break and in the afternoon after the hunt. Ten case-patients were at the lodge at the end of the hunt; 7 were at the lodge during the lunch break. In case-patient 6, who was not at the hunting lodge in the afternoon, ulceroglandular tularemia developed. The asymptomatic hunter (no. 11) had disemboweled ≈12 hares at a distance 8-10 meters from the place where hares were rinsed. One person present at the hunting lodge, whose laboratory tests were negative for F. tularensis, reported that although he 240 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 16 had been within 5 meters of the area where disemboweled hares were rinsed, he preferred to keep some distance from the group that were handling the hares.

Environmental Investigations
The outbreak area has several ecologic characteristics that were shown to correlate with high numbers of tularemia foci in the Czech Republic (Table 3). According to the hunters, all hares shot during the hunt on October 29 appeared healthy and showed no macroscopic signs of systemic infection (e.g., enlarged liver or spleen). No die-off of hares or rodents was observed in the region.
Samples taken in the game chamber and of the water were negative for F. tularensis, whereas samples taken from 11 of 14 parts of hares from the initial hunt were positive ( Table 1). Six of these samples were additionally positive for F. tularensis-specifi c LPS.

Discussion
We investigated an outbreak of tularemia after a hare hunt in Hesse, Germany, for which epidemiologic, clinical, and microbiologic data indicate an airborne origin. The re-sults of the cohort study support this hypothesis on the basis of the association between case status and presence within 5 meters of the location where disemboweled hares were rinsed. During the afternoon session of disemboweling and rinsing hares, 10 of the 11 case-patients were at the hunting lodge; aerosolization of infectious particles may have been limited to this session. Three case-patients, among them the patient who did not participate in the afternoon session, had a glandular or ulceroglandular form of tularemia. They may have acquired infections through skin lesions. The absence of cutaneous lesions or lymphadenopathy in the remaining 8 patients makes a cutaneous route of infection less likely than a respiratory route. The low incidence of respiratory symptoms among our patients is in agreement with fi ndings from previous airborne outbreaks that involved patients infected with the less virulent subspecies F. tularensis holarctica, in which only a minority of case-patients had symptoms suggestive of pneumonia (8,9).
Two hunters met the probable case-patient defi nition. The asymptomatic hunter (no. 11) disemboweled hares at a distance from the group. Severity of clinical tularemia has been correlated with infectious dose (20), and this hunter Table 2. Attack rates among exposed and nonexposed hare hunters, according to potential risk factors for Francisella tularensis infection, Germany, 2005* Exposed Not exposed might have been exposed to a smaller pathogen load or exposed on another recent occasion. Case-patient 3 died during the second week of illness. Antibodies against F. tularensis in most patients appear 6-10 days after onset of symptoms (21,22). Serum available for testing from casepatient 3 was from his eighth day of illness; hence, it was possibly taken before a measurable antibody response developed. We further cannot exclude the possibility that the 10 asymptomatic participants who did not undergo laboratory testing had to be considered as probable case-patients if they provided a serum sample. Detection of F. tularensis in hare specimens, including bone marrow specimens, and lack of F. tularensis detection in samples of the water system used to rinse hares suggest infection of the hares. One or more infected hares, still bloody and wet, may have cross-contaminated additional hares during further processing, e.g., during transport to and storage at the game chamber. All samples taken in the game chamber showed negative results. Casepatient 3 had cleaned the game chamber thoroughly with a pressure washer, possibly exposing himself to a high pathogen load.
Small clusters and outbreaks of airborne tularemia have been associated with hares or rabbits (7)(8)(9)(10)(11). However, most cases of tularemia associated with hares are of the ulceroglandular or glandular form (1,22). In a protracted outbreak in Spain, 97% of patients reported previous contact with hares; 71% of these had a glandular or ulceroglandular form of disease (23). Of 577 case-patients treated at a clinic in Czechoslovakia, 194 had direct contact with hares, and an (ulcero) glandular form of disease developed (1). Different frequencies of the diverse clinical forms of tularemia suggest differences in the main route of pathogen acquisition.
In the retrospective cohort study, presence within 5 meters of the place where disemboweled hares were rinsed was the risk factor most strongly associated with infection. Washing of contaminated produce was found to be an effective mechanism of generating infectious aerosols in tularemia outbreaks in sugar beet factories (1-3), and rinsing >1 hares infected with F. tularensis was the most probable way by which an infectious aerosol was generated. However, we cannot exclude the idea that an infectious aerosol was formed through further hare manipulating activities, e.g., transport.
Previous outbreaks in Germany date back to the 1950s, with the last case reported in the outbreak area in 1957 (24). Environmental characteristics of natural foci of tularemia persisting over >30 years have been described (19,25). The outbreak region in Germany shares several features favoring the occurrence or persistence of F. tularensis in the environment. Recently, the presence of F. tularensis in trapped rodents (2.1%) and in water samples from this hunting area was directly confi rmed, and >10% of rodents in several German regions not previously considered as endemic foci were infected (19). In addition, F. tularensis was repeatedly detected in 22 hares from 5 federal states (Bavaria, Hesse, Baden-Wuerttemberg, Thuringia, and Lower Saxony) after improved diagnostic tools (real-time PCR) had been applied complementary to standard 48-h bacterial cultivation (W.D. Splettstoesser et al., unpub. data). Together with results obtained from serologic studies in the German population (26), the outbreak reported here suggests that tularemia has either reemerged in Germany or is seriously underreported.