The Westfriese Flora is a one week flower show held every year in Bovenkarspel, in the north west of the Netherlands, with agricultural and consumer demonstrations on the same grounds. This year’s show, held from 19 to 28 February, attracted 80 000 visitors from all over the country. Subsequently 181 people who had visited the show in the second half of the week developed legionella pneumonia and 21 died (1,2). This article describes the immediate investigations and action in the first two days that led to a national alert and to the formulation of research questions for further investigation.

At 2330 on Wednesday 10 March 1999 the medical director of the Westfriese hospital (Westfriese Gasthuis, WFG) informed the public health service Westfriesland (Gemeenschappelijke Gezondheidsdienst, (GGD)) that the clinical condition of 12 people admitted to hospital with community acquired pneumonia in the preceding three days had deteriorated quickly and unexpectedly. As the facility’s respirators were all in use, patients had had to be transferred to other hospitals in the vicinity. Through the coordinator of communicable disease control of the GGD the national coordinating centre (Landelijke Coordinatiestructuur voor de Infectieziektebestrijding (bureau LCI)) was informed the same evening. On 11 March a meeting was held at the WFG-hospital with staff members of the WFG-hospital, the Division of Infectious Diseases, Tropical Medicine and AIDS of the Academic Medical Centre Amsterdam, the GGD, and the bureau LCI.

Hospital staff reported that 12 cases of severe community acquired pneumonia had been admitted since 7 March (see timetable). The cases came from various villages in a geographical area known as Westfriesland, the catchment area of the 500 bed general hospital. They were aged from 41 to 82 years; six were men and six women. Eight required mechanical ventilation (four were transferred for that reason to other hospitals). Patients had radiographic signs of bilobar infiltration. Microbiological testing of sputum and bronchoalveolar lavage had yielded no growth of specific microorganisms by Thursday (in particular no L. pneumophila). Specimens had been sent to another laboratory for polymerase chain reaction (PCR) test for L. pneumophila. Results were expected the next day. Serum from five patients sent in the morning to the virology department of the Academic Medical Centre in Amsterdam for testing for influenza, respiratory syncytial virus, adenovirus, and Chlamydia pneumoniae were reported negative the same afternoon. During the meeting it was decided that urine from cases should be tested for L. pneumophila serogroup (sg) antigen. Six of the eight patients tested were positive, one result was inconclusive, and one was negative. Urine from two of the patients transferred to another hospital was positive for L. pneumophila type 1 antigen (E. Yzerman, personal communication). During the afternoon a 13th patient with pneumonia was admitted and one of the original 12 patients died. The same evening a second patient died.

Date 1999

Time

Number of admissions
to hospital

Communication

07 March

1

08 March

4

09 March

4

10 March

0900

1

1100

8/10 MV*

2330

11 March

1000

1

1500

1

1700

1900

Informing all general practitioners in the area

2000

12 March

1200

1630

1800

13 March

1100

1430

There was no initial indication of common exposures that could be related to L. pneumophila. No recent increase in notified cases of legionellosis in the Netherlands had been observed. A case control study was set up immediately using an existing legionellosis questionnaire consisting of 40 items. Public health nurses of the GGD interviewed relatives of patients the same evening. The next morning they identified and interviewed controls. Controls were matched by street or neigbourhood of address, age (+ or - 5 years), and sex. By 2300 on Thursday 11 March it was clear that there were no secondary cases in the families of index cases nor previous respiratory illnesses in their families, and that cases had visited the Westfriese Flora held from 19 to 28 February. As the flower show is a major event it was possible that many local inhabitants in this age group had attended.

Medical professionals in the area were informed in the early evening before information about the possible place of exposure emerged. The general practitioners on call in the area (Westfriesland) were notified by telephone about the existence of a legionella epidemic. In a wider area (the complete province of north Holland) all hospitals were informed and asked about admissions of atypical pneumonia (active surveillance). Several hospitals reported that one or two patients with severe community acquired pneumonia had been admitted.

On Friday morning 13 March it became clear that all cases had visited the flower show. Public health services in the areas where cases had been admitted administered the same questionnaire to cases and controls (selected as above). By noon the case control study carried out by the GGD Westfriesland showed that all ten cases interviewed and three of 21 controls had visited the flower show (odds ratio (OR)=infinite, p< 0.001) (3).

On Friday afternoon preparations were started for a national campaign to inform medical professionals and the public. GGDs were advised by fax (followed up by telephone) to inform the general practitioners the same afternoon or evening that people with respiratory symptoms might ask for visits and should be diagnosed and treated promptly. The public information campaign advised people with respiratory symptoms who had visited the flower show to seek medical attention.

The coordination of communicable disease control is undertaken nationally by the bureau LCI, which consists of a professional outbreak management team (OMT) and a board of administrators (Bestuurlijk Afstemmings Overleg, (BAO)) (4). A telephone conference with BAO was held at 1630 and all BAO participants agreed to inform the public. A press conference started at 1700 and the outbreak was reported on the six o’clock evening news.

The OMT on Saturday 13 March met to plan further intervention (control of all used demonstration equipment) and investigations to find the source of the legionella (5). Decisions were agreed later by the BAO. Intervention was brought under the control of the health inspectorate (Inspectie Gezondheidszorg) and consumer inspectorate (Inspectie Gezondheidsbescherming). Investigation was to be carried out by the public health laboratory, (Rijksinstituut voor Volksgezondheid en Milieu, (RIVM)). The international community was informed through the channels of the health inspectorate and the diplomatic representatives to those countries that had sent special delegations to the flower show. An electronic bulletin (ProMED) was used later to inform a wider range of professionals (6) as people from various countries might have visited the flower show.

It will remain a matter of debate whether it would have been better, from the public health point of view, if the medical specialists of the WFG-hospital had informed the GGD earlier. The immediate investigation in the first 24 hours after informing the GGD brought a probable if not definitive working diagnosis and relevant medical professionals in a wide area around the signalling hospital were informed immediately. Within 48 hours the place where exposure to legionella had occurred had been found and a national information campaign for medical professionals and the public had begun.

The international professional community was informed using an existing rapid alert system on the internet. The European Working Group on Legionella Infections was informed after 48 hours by the participating microbiologist.

This epidemic revealed some shortcomings in the infrastructure of the public health reaction force. To inform all 52 GGDs by fax takes the bureau LCI about three hours. We had not previously realised how long three hours could be. With a second fax machine and a separate telephone line the faxing speed has since doubled. Most GGDs can inform all their local general practitioners by fax but the system does not work at weekends. Telephoning general practitioners on call on Friday night took considerable time. As a result of both shortcomings some general practitioners were informed sooner by news media than by the GGD.

By the time of the first investigation and action (12 March) further exposure was no longer a risk as the flower show had closed on 28 February. As the incubation period of legionellosis is two to ten days any patients were expected to be in medical care already. In retrospect this assumption was incorrect. As will be reported by the RIVM, cases continued to be admitted with severe community acquired pneumonia related to the Flora until 16 March.

Another important retrospective finding is that the case register revealed that 71 patients with community acquired pneumonia related to this outbreak had been admitted to hospital all over the country by 12 March. Since April 1999 a new infectious disease law in the Netherlands has required medical institutions to notify unusual numbers of infectious diseases to the GGD. Even with this new law, however, no signals would have come to national attention. Only in two hospitals were three or more cases admitted. Most of the cases of legionellosis were not microbiologically confirmed and therefore did not fulfil the criteria for obligatory notification. Even if patients had been diagnosed and notified, the information would not have been percolated to the national level as clinicians notify to the regional GGD on paper and send by mail. The Dutch surveillance system in March 1999 was inadequate to detect a large continuing epidemic early. Arrangements have been made between RIVM and GGDs to adapt and improve the early warning capacity of the surveillance system in the Netherlands.

1. Conyn-van Spaendonck M. Onderzoek epidemie legionellose, eindrapportage 23 augustus 1999. [Legionellosis epidemic investigation, final account 23 August 1999.] Infectiekziekten Bulletin 1999; 10: 157-8.

2. Conyn-van Spaendonck M. Tussenrapportage RIVM-onderzoek epidemie legionellose. [Interim report RIVM investigations legionellosis epidemic.] Infectiekziekten Bulletin 1999; 10: 135-8.

3. Boer JW den, Slijkerman FAN. Ijzerman EPF. De epidemie van Legionella pneumonie onder bezoekers van de Westfriese Flora in Bovenkarspel. Stand van zaken na vier weken. [The epidemic of legionella pneumonia among visitors of the Westfriese Flora in Bovenkarspel. State of affairs after four weeks.] Infectiekziekten Bulletin 1999; 10: 72-5.

4. Steenbergen JE van, Kraayeveld AG, Spanjaard L. Vaccination campaign for meningococcal disease in a rural area in the Netherlands - January 1998. Eurosurveillance 1999; 4: 18-21.

5. Hoepelman IM. Epidemie van legionellose in Nederland. [Epidemic of legionellosis in the Netherlands.] Ned Tijdschr Geneeskd 1999; 143: 1992-6

6. Steenbergen JE van. Legionellosis, flower show - Netherlands (Bovenkarspel): request for info. ProMED Digest 1999; 99: 990315. (http://www.healthnet.org/programs/promed.html)