The Prince of Wales Hospital (PWH) has been at the forefront of the outbreak of severe acute respiratory syndrome (SARS) in Hong Kong. 1 We relate our experience at this hospital. A working definition of SARS is important, 2 although clinical conditions rarely remain within artificial boundaries. Some patients might not have all features, others may present unusually. Fever is a cardinal symptom but not always so, and is sometimes absent in elderly patients. Some patients have presented with diarrhoea or, in at least two cases, with severe acute abdominal pain requiring exploratory laparotomy. All these patients developed typical SARS. Patients presenting with other respiratory infections must now all be regarded as potential SARS cases until proven otherwise. Contact with a known case is an important discriminator but, if emphasised too strongly in the diagnostic process, may lead to false positives or negatives. The difficulty of making a firm diagnosis until chest radiographic changes appear has important implications for health-care personnel and for surveillance. Three major reasons for spread of infection to health-care workers have been: failure to apply isolation precautions to cases not yet identified as SARS, breaches of procedure, and inadequate precautions. Every patient must now be assumed to have SARS, which has major long-term implications for the health-care system. Another reason for spread among health-care workers is infected workers continuing to work despite symptoms, such as mild fever. Such individuals must now cease working. However, staying at home can also have disastrous consequences for exposed family members. Potential cases therefore require early isolation from both workplace and household. Extreme measures are required to protect health-care workers, who account for about 20% of cases. Early diagnosis by virus isolation or serological testing is essential to halt the spread of SARS. Progress has been made with the isolation of the coronavirus.3, 4, 5 A metapneumovirus was also identified in Canada 4 and in many of the cases at PWH. Coronavirus appears to be the main pathogen, but dual infections may be possible. Such situations are uncommon in human disease, apart from HIV-related infections, but in veterinary medicine combined infections with coronavirus and other agents have been described.6, 7 The first cases probably occurred in Guangdong Province in southern China in November, 2002. 8 The term SARS appears to have been first used for a patient in Hanoi who became ill on Feb 26, 2003, and was evacuated back to Hong Kong where he died on March 12. The physician who raised the alarm in Hanoi, Carlo Urbani, subsequently contracted SARS and died. The first case in Hanoi had stayed at a hotel in Kowloon, Hong Kong, at the same time as a 64-year-old doctor who had been treating pneumonia cases in southern China. This doctor was admitted to hospital on Feb 22, and died from respiratory failure soon afterwards. 9 He was the first known case of SARS in Hong Kong and appears to have been the source of infection for most if not all cases in Hong Kong as well as the cohorts in Canada, Vietnam, Singapore, USA, and Ireland, and subsequently Thailand and Germany. 10 The index patient at PWH was admitted on March 4, 2003, and had also visited this hotel. He had pneumonia which progressed initially despite antibiotics, but after 7 days he improved without additional treatment. 1 On March 10, 18 health-care workers at PWH were ill and 50 potential cases among staff were identified later that day. Further staff, patients, and visitors became ill over the next few days and there was subsequent spread to their contacts. By March 25, 156 patients had been admitted to PWH with SARS, all traceable to this index case. 1 One important factor in the extensive dissemination of infection appears to have been the use of nebulised bronchodilator, which increased the droplet load surrounding the patient. Overcrowding in the hospital ward and an outdated ventilation system may also have contributed. The second major epicentre in Hong Kong, accounting for over 300 cases, has been an apartment block called Amoy Gardens. The source has been attributed to a patient with renal failure receiving haemodialysis at PWH who stayed with his brother at Amoy Gardens. 11 He had diarrhoea, and infection may have spread to other residents by a leaking sewage drain allowing an aerosol of virus-containing material to escape into the narrow lightwell between the buildings and spread in rising air-currents. Sewage also backflowed into bathroom floor drains in some apartments. Spread to people in nearby buildings also occurred, probably by person-to-person contact and contamination of public installations. Although the rapid spread of the disease in some situations may have been explained, many uncertainties remain. Why the disease spread in the Kowloon hotel has not been clarified, and there are many other important issues. “Super-spreaders” may be prone to carry a high viral load because of defects in their immune system, as could be the case in the patient with end-stage renal failure implicated in the Amoy Gardens outbreak and another with renal failure at the centre of an outbreak in Singapore. Subclinical infections may also occur and will not be recognisable until reliable diagnostic tests are available. Procedures causing high risk to medical personnel include nasopharyngeal aspiration, bronchoscopy, endotracheal intubation, airway suction, cardiopulmonary resuscitation, and non-invasive ventilation procedures. Cleaning the patient and the bedding after faecal incontinence also appears to be a high-risk procedure. Treatments have been empirical. Initial patients were given broad-spectrum antibiotics but, after failing to respond for 2 days, were given ribavirin and corticosteroids. Patients who continued to deteriorate with progression of chest radiographic changes or oxygen desaturation, or both, were given pulsed methylprednisolone. 1 Steroids were used on the rationale that progression of the pulmonary disease may be mediated by the host inflammatory response, similar to that seen in acute respiratory distress syndrome, and produced by a cytokine or chemokine “storm”. The clinical impression is that pulsed steroids sometimes produce a dramatic response. However, apparent benefits of steroid treatment have proven to be incorrect before, as in infection with respiratory syncytial virus. 12 Lack of knowledge of SARS' natural history adds to the difficulty of determining the effectiveness of therapy. Some patients have a protracted clinical course with potential for relapses continuing into the second or third week, or beyond. Long hospital stays, even in less ill patients, are required, and the high proportion of patients requiring lengthy intensive care, with or without ventilation (23% in the 138 cases from PWH 1 ), and the susceptibility of health-care workers bodes ill for the ability of health-care systems to cope. Even when the acute illness has run its course, unknowns remain. Continued viral shedding and the possible development of long-term sequelae, such as pulmonary fibrosis or late post-viral complications, means that patients will require careful surveillance.