We performed a retrospective case series study. All reported SARS patients who stayed in the medical wards or intensive care unit of Princess Margaret Hospital and Wong Tai Sin Hospital on April 16, 2003, were screened. Patients were excluded if subsequent follow-up serologic tests showed no rise in antibody titer against SARS-associated coronavirus. All eligible SARS patients, except three, were recruited into the study. One healthcare worker refused to be studied, and two patients who were suspected of contracting the infection during their hospital stay were also excluded. This cohort was followed up until May 20, 2003. Data were collected through the hospital authority’s computerized clinical management system, case record review, and a questionnaire survey assisted by the nursing staff of each SARS ward. Age, sex, occupation, residential address, smoking habit, time between onset of fever and start of antiviral therapy, coexisting conditions, and laboratory data were the variables under study. Outcome variables were the following: dependency on high amounts of oxygen (requiring at least 3 L/min of oxygen through a nasal cannula) and admission to an intensive care unit or death.

The study population consisted of 127 male and 196 female patients, ranging in age from 18 to 83 (41±14). Forty-seven (15%) patients were healthcare workers. One hundred thirty-three (41%) were Amoy Gardens residents. Two hundred seventy-three (85%) patients were in good health. The coexisting conditions are listed in Table 2. Psychiatric illness, hepatitis B carrier status, and thalassemia trait status were not classified as coexisting conditions. Fifteen (14%) males and 7 (4%) females were current smokers. The overall prevalence of smoking among SARS patients was 7.6% (9.1% if healthcare workers are excluded). None of the affected healthcare workers smoked. The symptoms exhibited fulfilled the diagnostic criteria of the Hospital Authority’s SARS registry.

All patients had lung involvement, documented either by chest x-ray or high-resolution computed tomographic scan of the thorax. Lymphopenia, found in 221 (68%) patients, was a prominent feature in those who sought treatment. Other initial laboratory findings included thrombocytopenia (41%), elevated creatine kinase level (14%), and elevated lactate dehydrogenase level (42%). Initial bacterial cultures were negative. Virus screening was negative for adenovirus, respiratory syncytial virus, influenza A and B, and parainfluenza virus. Two hundred and seven (64%) patients had reverse transcriptase–polymerase chain reaction (RT-PCR) assays performed for SARS-associated coronavirus, and 128 (62%) of the results were positive. Two hundred and forty-two (75%) patients had completed serologic testing. The diagnosis of recent SARS-associated coronavirus infection was confirmed by either RT-PCR assays or serologic test in 286 (89%) patients. The sensitivity of RT-PCR assays was 58% (95% confidence interval [CI], 50% to 66%).

Our patients sought treatment 3.9±2.7 days after onset of fever. The interval between onset of fever and admission was positively correlated with admission neutrophil count (Pearson r=0.1, p=0.07), admission platelet count (Pearson r=0.1, p=0.06), and initial lactate dehydrogenase level (Pearson r=0.36, p<0.001). An antibiotic was started immediately after admission in all cases. Either levofloxacin, 500 mg once a day, or amoxicillin/clavulinate acid, 375 mg three times a day plus clarithromycin, 500 mg twice a day, was used to protect against community-acquired pneumonia. All patients were also treated with oral or intravenous ribavirin, according to protocol. Most (94%) were given either intravenous hydrocortisone or oral prednisolone, according to protocol. Five patients received intravenous methylprednisolone as a form of steroid therapy. The dose was administered at 3 mg/kg once a day and would be tapered down to 1 mg/kg if the patient showed a clinical response. Pulsed doses of methylprednisolone (500 mg per dose) were given as initial treatment in 12 patients, who then received maintenance steroid therapy. Two patients were treated with ribavirin only. Ribavirin plus steroid therapy was administered 1.2±1.7 days after admission. The interval between admission and initiation of antiviral therapy was negatively correlated with the interval between onset of fever and admission (Pearson r = –0.17, p=0.003).

Clinical Outcome

In 115 (36%) patients, the disease was limited with resolution of fever and pneumonitis. Two hundred and eight (64%) patients had either clinical or radiologic evidence of progression of pneumonitis, and they received 2.9±2 gm pulsed dose methylprednisolone therapy. Maintenance steroid was resumed after pulsed dose therapy. Patients who were given pulsed doses of steroids were treated with potent broad-spectrum intravenous antibiotics (piperacillin and tazobactam) to protect against hospital-acquired infection. Hyperglycemia, hypokalemia, flare up of hepatitis B infection, hospital-acquired infection, and steroid psychosis were the acute side effects encountered. Hepatitis B carriers were treated with lamivudine, 100 mg once a day; no liver failure occurred in members of this cohort. Disease progression was apparently arrested by pulsed dose steroid therapy in 98 (30%) patients. In the remaining 110 (34%) patients, the illness ran a severe and protracted course, and the patient needed high doses of oxygen. Sixty-seven (21%) had been admitted to intensive care unit, and 42 (13%) required ventilator support. Twenty-six patients died (12 males and 14 females). The crude mortality rate of our cohort after 47±8 days of follow-up was 7.9% (95% CI, 5% to 10.8%) and was an underestimation because of sampling bias. Those who died before April 16, 2003, were excluded from our sample, while long-term survivors were retained for study. Among them, 10 had concurrent medical illness. No healthcare worker in this cohort died. Diabetes was found in three patients who died, and hypertension in four who died. Eleven of those who died lived in Amoy Gardens. A young pregnant woman died after delivery, despite aggressive treatment.

Figure 1

Figure 1. Relationship between age and fatal severe acute respiratory syndrome illness, Hong Kong, 2003.

Figure 2

Figure 2. Relationship between neutrophil count and fatal severe acute respiratory syndrome illness, Hong Kong, 2003.

Figure 3

Figure 3. Relationship between lactate dehydrogenase level (LDH) and fatal severe acute respiratory syndrome illness, Hong Kong, 2003

Age, sex, healthcare worker status, Amoy Gardens resident status, presence of coexisting conditions, interval between onset of fever and therapy (ribavirin plus steroid), neutrophil and platelet count on admission, and initial creatine kinase and lactate dehydrogenase levels were the correlates of clinical outcome under study. Variables with a p value of <0.1 by univariate analysis were entered into the multivariate regression model. By multivariate logistic regression, advanced age, high neutrophil count on admission, and high initial lactate dehydrogenase level were independent correlates of high oxygen dependency as well as intensive care unit admission or death (Table 3). By Cox’s backward stepwise regression, young age, low neutrophil count on admission, and healthcare worker status (p=0.05) were favorable independent correlates of survival time (Table 3). A dose-response relationship also existed between the independent correlates and clinical outcome (Figure 1,2,3). We used the term “correlates” instead of “predictors” of outcome because of the method we used, a case series.

The second serology titer obtained after the end of second week was negatively correlated with age (Pearson r=–0.13, p=0.05) and admission lymphocyte count (Pearson r=–0.17, p=0.01). Conversely, the neutrophil count on admission was positively correlated with the second serology titer (Pearson r=0.2, p=0.003). The pulsed dose of steroid was not shown to affect the second serology titer (Pearson r=0.1, p=0.18). Patients who depended on high oxygen therapy had a higher second antibody titer against SARS-associated coronavirus (p = 0.05).

The virus attacked persons of both sexes and all ages. Many were previously in good health and the wage earners in their families. Not infrequently, several members of a family were admitted to the hospital. The need for isolation discouraged close social contact. Unfortunately, some of the patients were also stigmatized. The psychosocial effect of SARS is by no means a lesser problem.

RT-PCR assay for SARS-associated coronavirus is a new test, and its sensitivity and specificity have yet to be established. In our cohort, the sensitivity was 58%, and results depended on sampling technique and stage of disease (10). Contamination of specimen could lead to a false-positive result. A false-negative result could arise from performing the test in the very early or late stage of the disease. Diarrhea was common among Amoy Gardens SARS patients. The virus could be found in stool by RT-PCR assays. A negative test does not rule out the diagnosis, however. The serologic test remains as the standard criterion of definitive diagnosis. Pulsed doses of steroid did not seem to affect the humoral response of SARS patients. In retrospect, the intensity of antibody response was related to clinical outcome and associated pretreatment prognostic factors. The viral load could be a determinant of these prognostic association factors.

Our hematologic and biochemical data, as well as associated prognostic factors, agreed with the work of Lee et al. (4). Both advanced age and high neutrophil count on admission were associated with poor outcome. We found that initial lactate dehydrogenase level was also an associated prognostic factor. The early phase of SARS is characterized by lymphopenia and thrombocytopenia. As the disease progresses, both neutrophil and platelet counts rise, accompanied by an elevation in lactate dehydrogenase level. The neutrophilic response is important in the pathogenesis of hypersensitivity pneumonitis, and thus the initial neutrophil count could also indicate disease progression. Lactate dehydrogenase level reflects tissue necrosis related to immune hyperactivity in SARS and thus relates to poor outcome. Patients with high neutrophil counts and lactate dehydrogenase levels on admission could have been late in seeking treatment or have experienced heavy exposure to the virus.

One third of the SARS patients in our study had a limited disease course. In the remaining two thirds, pneumonitis progressed rapidly after the early use of ribavirin and steroid combination therapy. Apparently, approximately one third responded to pulsed doses of steroids, while the other third depended on treatment with high amounts of oxygen. Intensive care was required for 21% of patients. Advanced age, high neutrophil count on admission, and elevated initial lactate dehydrogenase level were independent correlates of an adverse clinical outcome. Strong evidence to support early and routine use of ribavirin and steroid combination therapy in all SARS patients does not exist.

We need to investigate new antiviral agents and test the efficacy of steroids in randomized controlled trials. SARS is an entirely new emerging disease and its clinical course varies widely. By stratifying our patients according to risk, we could individualize our treatment protocol. In addition, we need a more sensitive and rapid diagnostic test for SARS-associated coronavirus infection, both for treatment and for forming cohorts of patients infected with this deadly disease.