Role of absolute lymphocyte count in the screening of patients with suspected SARS.

Absolute lymphopenia has been reported in patients with SARS but there are no studies assessing the predictive value of lymphocyte counts on initial admission to hospital. In 54 patients admitted to hospital suspected of SARS, absolute lymphocyte counts were significantly lower in the confirmed SARS group than in the non‐SARS group. With a cut off of ≤ 1.0, the area under the receiver operator characteristic (ROC) curves was 0.937 (95% CI = 0.835–0.984). The sensitivity and specificity for disease were 84.6% (95% CI = 65.1–95.5) and 89.3% (95% CI = 71.7–97.6). The positive predictive value was 88.0% and the negative predictive value was 86.2%. Absolute lymphocyte counts taken on the first day of admission to hospital may be useful in predicting SARS.

Hong Kong has been one of the worst areas in the world to be hit by SARS. Emergency and primary care physicians are faced with the dilemma of dealing with a disease that is novel and easy to miss. The case definition of SARS has been described but is evolving as new information arises and now includes fever, chest X‐ray changes and laboratory tests for coronavirus. This definition differs from that of the Hospital Authority and Department of Health of Hong Kong, and so in the early phase of the crisis there is no single definition or single test that is indicative of the disease.

We investigated patients admitted consecutively with suspected SARS to a hospital in Tuen Mun during a two‐week period. The hospital is situated in the west of Hong Kong and has a catchment population of 1 million people. The annual census of the ED in 2002 was of 270 000 patients.

Suspected SARS was defined as a fever (temperature > 38°C) for more than 2 days, and chest X‐ray changes showing evidence of consolidation with or without respiratory symptoms (cough and shortness of breath), and at least two out of the following: chills, increased cough, malaise or history of exposure.

All patients with suspected SARS were admitted. Data were collected from both the accident and emergency record on admission and from the hospital computer system upon discharge. Demographic data included gender and age. Physiological data included blood pressure, pulse rate, temperature, respiratory rate and oxygen saturation. Clinical data on presentation included cough, diarrhoea, chills, malaise and myalgia. White cell count and absolute lymphocyte counts were collected on the day of admission, whilst PCR results performed on samples taken up to 21 days after onset of symptoms for coronavirus were also recorded. Patients were then divided into two groups: confirmed SARS and non‐SARS. The subsequent in‐patient diagnostic criteria had incorporated the use of PCR assay for coronavirus in saliva, urine and stool, together with serological changes.

Fifty‐four patients were included in this study of which 26 patients (49%) had confirmed SARS. No patients died from this cohort.

The demographic data were comparable between the two groups. The mean ages were 46.4 (SD 20.1) and 50 (SD 28.3) years old for the non‐SARS and SARS groups, and the proportion of female gender was 63% in the SARS and 55% in the non‐SARS group.

All patients had respiratory symptoms and chest X‐ray changes indicative of pneumonia. Ten patients (42%) in the SARS group and 12 patients (43%) in the non‐SARS group had a fever (> 38°C).

Mean absolute lymphocyte count was 0.75 × 109/L (SD 0.33) for the SARS group, and 1.71 × 109/L (SD 0.62) for the non‐SARS group. A cut‐off absolute lymphocyte count ≤ 1.0 gives a sensitivity of 84.6% (95% CI = 65.1–95.5) and a specificity of 89.3% (95% CI = 71.1–97.6; Fig. 1). The positive predictive value was 88.0% and the negative predictive value was 86.2%. The area under the receiver operator curve was 0.937 (95% CI = 0.835–0.984; Fig. 2).

Figure Figure 1. 

Dot plot of absolute lymphocyte counts in patients who did (1) and did not (0) develop SARS.

Figure 2

Receiver operator characteristic curve of absolute lymphocyte counts to detect patients with SARS.

The application of absolute lymphocyte count in the diagnosis of SARS in the early phase of presentation appears promising and may be considered in future guidelines.