Leung Gabriel M, Rainer Timothy H, Lau Fei-Lung, Wong Irene O L, Tong Anna, Wong Tai-Wai, Kong James H B, Hedley Anthony J, Lam Tai-Hing
University of Hong Kong, Prince of Wales Hospital, Shatin, China.
Ann Intern Med. 2004 Sep 7;141(5):333-42. doi: 10.7326/0003-4819-141-5-200409070-00106. Epub 2004 Aug 23.
Accurate, objective models of triage for patients with suspected severe acute respiratory syndrome (SARS) could assess risks and improve decisions about isolation and inpatient treatment.
To develop and validate a clinical prediction rule for identifying patients with SARS in an emergency department setting.
Retrospective analysis using a 2-step coefficient-based multivariable logistic regression scoring method with internal validation by bootstrapping.
2 hospitals in Hong Kong.
1274 consecutive patients from 1 hospital and 1375 consecutive patients from another hospital.
Points were assigned on the basis of history, physical examination, and simple investigations obtained at presentation. The outcome measure was a final diagnosis of SARS, as confirmed by World Health Organization laboratory criteria.
Predictors for SARS on the basis of history (step 1) included previous contact with a patient with SARS and the presence of fever, myalgia, and malaise. Age 65 years and older and younger than 18 years and the presence of sputum, abdominal pain, sore throat, and rhinorrhea were inversely related to having SARS. In step 2, haziness or pneumonic consolidation on chest radiographs and low lymphocyte and platelet counts, in addition to a positive contact history and fever were associated with a higher probability of SARS. A high neutrophil count, the extremes of age, and sputum production were associated with a lower probability of SARS. In the derivation sample, the observed incidence of SARS was 4.4% for those assigned to the low-risk group (in steps 1 or 2); in the high-risk group, incidence of SARS was 21.0% for quartile 1, 39.5% for quartile 2, 61.2% for quartile 3, and 79.7% for quartile 4. This prediction rule achieved an optimism-corrected sensitivity of 0.90, a specificity of 0.62, and an area under the receiver-operating characteristic curve of 0.85.
The prediction rule may not apply to isolated cases occurring during an interepidemic period. Generalizability of the findings should be confirmed in other SARS-affected countries and should be prospectively validated if SARS returns.
Our findings suggest that a simple model that uses clinical data at the time of presentation to an emergency department during an acute outbreak predicted the incidence of SARS and provided good diagnostic utility.
针对疑似严重急性呼吸综合征（SARS）患者的准确、客观的分诊模型可以评估风险，并改善关于隔离和住院治疗的决策。
开发并验证一种用于在急诊科环境中识别SARS患者的临床预测规则。
采用基于系数的两步多变量逻辑回归评分方法进行回顾性分析，并通过自抽样法进行内部验证。
香港的两家医院。
来自一家医院的1274例连续患者和来自另一家医院的1375例连续患者。
根据患者就诊时的病史、体格检查和简单检查进行评分。结局指标是根据世界卫生组织实验室标准确诊的SARS最终诊断结果。
基于病史（第一步）的SARS预测因素包括既往接触过SARS患者以及发热、肌痛和不适的存在。65岁及以上和18岁以下的年龄以及痰液、腹痛、咽痛和流涕的存在与患SARS呈负相关。在第二步中，除了有阳性接触史和发热外，胸部X线片上的模糊或肺炎实变以及低淋巴细胞和血小板计数与SARS的较高概率相关。高中性粒细胞计数、极端年龄和咳痰与SARS的较低概率相关。在推导样本中，对于被分配到低风险组（在第一步或第二步）的患者，观察到的SARS发病率为4.4%；在高风险组中，第一四分位数的SARS发病率为21.0%，第二四分位数为39.5%，第三四分位数为61.2%，第四四分位数为79.7%。该预测规则的乐观校正敏感性为0.90，特异性为0.62，受试者操作特征曲线下面积为0.85。
该预测规则可能不适用于流行间期出现的个别病例。研究结果的可推广性应在其他受SARS影响的国家得到证实，如果SARS再次出现，应进行前瞻性验证。
我们的研究结果表明，一个在急性疫情期间使用患者就诊时临床数据的简单模型可以预测SARS的发病率，并具有良好的诊断效用。
