This retrospective study analyzed laboratory records of patients admitted to six public hospitals in Hong Kong during the SARS epidemic from March to June 2003. The first inclusion criterion was serologic evidence of SARS-CoV infection. Altogether, 433 patients who exhibited either seroconversion or a fourfold rise in anti–SARS-CoV immunoglobulin (Ig)G antibody titer were identified. Detection of anti-SARS-CoV IgG antibody was based on an in-house immunofluorescence assay that used virus-infected cells. Of the 433 patients with positive serologic test results, 18 were excluded because no samples had been collected for virus detection. As a result, 415 patients were included in this study. Twelve were pediatric patients 3–16 years of age (mean 11.3, standard deviation [SD] 4.1), divided equally between girls and boys. Three hundred thirty-five were adult patients 17–64 years of age (mean 37.1, [SD 11.2), with 60.9% females. The remaining 68 were elderly patients 65–97 years of age (mean 76.7, SD 8.2), with 37 (54.4%) women. Altogether, 48/335 (14.3%) of the adult group and 2/68 (2.9%), respectively, of the elderly group required ventilation and received intensive care but recovered; 4/335 (1.2%) of adults and 21/68 (30.9%) of elderly patients died of the infection. All children recovered without requiring mechanical ventilation or intensive care.

Respiratory, stool, and rectal swab specimens were collected in viral transport medium, and urine samples were transported in sterile containers. For some patients, throat and nasal swab samples were pooled into a single specimen container and processed as a single specimen. These samples were referred as “pooled throat and nasal swabs” for the purpose of analysis in this study. Specimens collected were refrigerated (approximately 10°C) until delivery, which were done on the same day in most circumstances. Specimens were kept in iceboxes during delivery to the designated centralized laboratory. SARS-CoV investigations were performed on fresh specimens without prior freezing and thawing.

SARS-CoV detection by RT-PCR was conducted in two laboratories based on the same primer set COR-1 (sense) 5′ CAC CGT TTC TAC AGG TTA GCT AAC GA 3′, and COR-2 (antisense) 5′ AAA TGT TTA CGC AGG TAA GCG TAA AA 3′ (11). The specimens were centrifuged at 10,000 x g for 1 min, and 140 μL of the supernatant were used for RNA extraction using the QIAamp viral RNA Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s protocol. Reverse transcription of RNA was conducted in a 20-μL reaction mix containing 4.2 μL of extracted RNA preparation, 2.5 μmol/L of random hexamer and 50 U of reverse transcriptase (Applied Biosystems, Foster City, CA). After incubation at room temperature for 10 min and at 42°C for 30 min, the reaction was stopped by heating at 95°C for 5 min and chilled on ice. The subsequent PCR was conducted in a 50-μL reaction mix containing 5 μL of cDNA template, 1 μmol/L of each primer COR-1 and COR-2, 1.5 U Taq polymerase (Amersham Biosciences, Uppsala, Sweden), 0.2 mmol/L of each deoxynucleoside triphosphate and 2.0 mmol/L magnesium chloride. The cycling conditions were 94°C for 3 min; 45 cycles of 94°C for 30 sec, 60°C for 30 sec, and 72°C for 1 min, and a final extension at 72°C for 7 min. The PCR amplicons were visualized by ethidium bromide staining after agarose gel electrophoresis.

The same RNA extraction method was used in laboratory B. The RT-PCR was carried out in a single-tube system (Superscript One-Step RT-PCR with Platinum Taq, Invitrogen, Carlsbad, CA), in a 25-μL reaction mix containing 0.6 μmol/L of each COR-1 and COR-2 primer, 0.2 mmol/L of each deoxynucleoside triphosphate and 1.2 mmol/L magnesium sulphate. The reverse transcription was conducted at 54°C for 30 min. After the mixture was held at 94°C for 3 min, it underwent 45 cycles of amplification at 94°C for 45 sec, 60^oC for 45 sec, 72°C for 45 sec, and final extension at 72°C for 7 min. The PCR amplicon was also detected by agarose gel electrophoresis as in laboratory A.

All reagent preparation, sample extraction, amplification, and amplicon detection procedures were conducted in separate areas and under strengthened precautions to avoid cross-contamination. The lower detection limit of the RT-PCR assays was determined by testing preparations with known copies of SARS-CoV as determined by real-time RT-PCR. Both laboratories showed a lower detection limit of 50 viral copies per reaction. In all test runs, positive controls containing approximately 100 copies of viral RNA in viral transport medium were included, and double distilled water was used as a negative control. Positive samples were confirmed by repeating the RNA extraction and RT-PCR from the original samples.

Virus isolation for SARS-CoV was performed in laboratory B. Specimens were injected into African green monkey (Vero E6) cell monolayers. For stool or rectal swab samples, the suspension was passed through a 0.45-μm filter before injection. The cell culture tubes were examined daily for diffuse, refractile, rounding cytopathic effects characteristic of SARS-CoV. When cytopathic effects were observed, the cells were stained by the indirect immunofluorescence technique with a convalescent-phase serum sample collected from a SARS patient. The identity of the isolate was further confirmed by RT-PCR.

Statistical tests were performed by using the Statistical Package for the Social Sciences software (SPSS 10.1.0, Inc., Chicago, IL). The chi-square test was used to analyze categorical variables. All statistical tests were two-tailed and p values ≤0.05 were regarded as significant.

Altogether, 624 respiratory specimens, 671 stool or rectal swab specimens, and 314 urine specimens were collected from the 415 study patients for RT-PCR; 738 respiratory, 810 stool or rectal swab, and 531 urine specimens were submitted for virus isolation; and 558 respiratory, 318 stool or rectal swab, and 296 urine specimens were tested by both RT-PCR and virus isolation. The mean number of specimens collected from each patient was 5.3 (range 1–32, SD 5.1). The mean time of collection of the first specimen was 13.5 days (range 1–88, SD 16.5) after the onset of symptoms (Figure 1). Patients whose first specimens were collected at a later stage of illness had become ill early in the outbreak when no diagnostic test was available.

To analyze the profile of viral shedding, specimens were grouped into categories: respiratory, stool or rectal swab, and urine. Respiratory specimens included tracheal aspirate, nasopharyngeal aspirate, throat swab, throat washing, nasal swab, and pooled throat and nasal swabs. The viral shedding profile is shown in Figure 2; the number of specimens tested is shown in Table 1. Stool/rectal swab specimens provided the highest positive rate by RT-PCR, followed by respiratory and urine specimens. The RT-PCR positive rate for respiratory specimens increased slightly from week 1 to week 2 after the onset of illness and then dropped to lower levels at week 3 and week 4. The positive rate for stool/rectal swab peaked at week 2 and week 3 and then dropped sharply. The positive rate for urine specimens increased gradually and peaked at week 4. Viral shedding beyond week 6 was rare, with only three stool samples (collected on day 54, day 67, and day 75, respectively) and one respiratory sample collected on day 50 positive by RT-PCR. As for virus isolation, the latest positive specimen was collected on day 31 for a respiratory specimen, day 23 for a urine specimen, and day 6 for a stool sample.

The RT-PCR results of specimens collected within the first 3 weeks after the onset of illness were analyzed to further clarify the viral shedding pattern. The positive rate for respiratory specimens began to increase on day 5 and remained high during the second week. The positive rate for stool/rectal swab specimens peaked at days 9 and 10 and remained high during the second and third week, whereas the detection rate for urine specimens started to increase at the end of the second week (Figure 3).

The RT-PCR and virus isolation results of different specimen types collected within the first 4 weeks after the onset of symptoms are shown in Table 2. When RT-PCR was used for virus detection, tracheal aspirate and stool provided a high diagnostic yield, with an average positive rate of 66.7% and 56.5%, respectively, for the first 2 weeks. Pooled throat and nasal swabs, rectal swab, nasal swab, throat swab and nasopharyngeal aspirate provided a moderate yield with average positive rates ranging from 29.7% to 40.0% for the first 2 weeks, whereas throat washing and urine specimens provided a lower yield with an average positive rate of 17.3% and 4.5%, respectively. The yield from virus isolation was much lower than from RT-PCR, and no specimen was positive by culture but negative by RT-PCR.

To compare the sensitivity of RT-PCR and virus isolation for detecting SARS-CoV, a subgroup analysis was performed on 1,172 specimens that had been submitted for both RT-PCR and virus isolation. The isolation/RT-PCR index, defined as the number of isolation-positive specimens per RT-PCR-positive specimens, was highest for respiratory samples, particularly for pooled throat and nasal swabs, tracheal aspirate, and nasopharyngeal aspirate. The isolation/RT-PCR index for stool or rectal swab samples was approximately 5- to 10-fold lower when compared with that for respiratory specimens (Table 3).

Altogether, 132 (31.8%) of the 415 study patients had SARS-CoV detected by RT-PCR or virus isolation. To analyze factors associated with positive virologic testing results, a subgroup analysis was performed on 342 patients whose first specimens were collected within 4 weeks of illness onset. Within this subgroup, 128 (37.4%) patients had one or more positive results by RT-PCR or virus isolation. The mean number of positive specimens among these patients was 1.8 (range 1–10, SD 1.7). The characteristics of patients with and without positive specimens are shown in Table 4. A higher positive detection rate for SARS-CoV was observed for patients with more severe disease (p < 0.001 by chi-square test).