Martínez García M A, Soler Cataluña J J, Román Sánchez P, Cabero Salt L, Giménez Ibáñez I, Gastaldo Palop T
Unidad de Neumología. Hospital General de Requena. Valencia. Spain.
Arch Bronconeumol. 2003 Oct;39(10):449-54. doi: 10.1016/s0300-2896(03)75427-1.
To analyze the predictive value of clinical data for identifying patients suspected of sleep apnea-hypopnea syndrome with an apnea-hypopnea index (AHI)> or = 30.
Patient characteristics, cardiorespiratory medical history, and clinical signs and symptoms were recorded for all patients. Exclusion criteria were daytime respiratory insufficiency or heart failure. All patients underwent polysomnographic testing (AutoSet Portable Plus II, ResMed Corp, Sydney, Australia) for automatic AHI calculation and manual determination of central and obstructive apneas. A logistic regression model was constructed to calculate the likelihood of an individual's presenting an AHI> or = 30 as well as the predictive value of each variable and of the final model.
Three hundred twenty-nine patients with a mean +/- SD age of 58 +/- 13.45 years were studied; 76.4% were men. Data for 207 patients were used to construct the logistic regression model: logit (P) = 2.5 blood pressure + 1.5 Epworth test + body mass index + 0.6 repeated observed episodes of apnea 2.1. Logit(P) was loge (1-P)/P and variables were dichotomized with cut points of 11 for the Epworth test and of 30 kg/m2 for body mass index. The diagnostic sensitivity of the model was 80.2% (75%-86%), specificity was 93.4% (89%-95%), positive predictive value was 89.6% (84%-93%) and negative predictive value was 86.9% (81%-90%), such that 89.6% of the patients were correctly classified. The variable with the greatest predictive value was high blood pressure. The model was validated prospectively in the remaining 102 patients.
Prior to diagnostic tests for SAHS, clinical data can be useful for identifying patients suspected to have a AHI> or = 30.
分析临床数据对于识别呼吸暂停低通气指数（AHI）≥30的疑似睡眠呼吸暂停低通气综合征患者的预测价值。
记录所有患者的特征、心肺病史以及临床体征和症状。排除标准为白天呼吸功能不全或心力衰竭。所有患者均接受多导睡眠图测试（AutoSet Portable Plus II，瑞思迈公司，悉尼，澳大利亚），以自动计算AHI并手动确定中枢性和阻塞性呼吸暂停。构建逻辑回归模型以计算个体出现AHI≥30的可能性以及每个变量和最终模型的预测价值。
对329例平均年龄为58±13.45岁的患者进行了研究；76.4%为男性。使用207例患者的数据构建逻辑回归模型：logit（P）=2.5×血压+1.5×爱泼沃斯嗜睡量表测试+体重指数+0.6×重复观察到的呼吸暂停发作次数-2.1。Logit（P）为loge（1-P）/P，变量进行二分法划分，爱泼沃斯嗜睡量表测试的分界点为11，体重指数的分界点为30kg/m²。该模型的诊断敏感性为80.2%（75%-86%），特异性为93.4%（89%-95%），阳性预测值为89.6%（84%-93%），阴性预测值为86.9%（81%-90%），89.6%的患者被正确分类。预测价值最大的变量是高血压。该模型在其余102例患者中进行了前瞻性验证。
在进行睡眠呼吸暂停低通气综合征的诊断测试之前，临床数据有助于识别疑似AHI≥30的患者。
