Moine Pierre, Timsit Jean-François, De Lassence Arnaud, Troché Gilles, Fosse Jean-Philippe, Alberti Corrine, Cohen Yves
Département d'Anesthésie Réanimation, CHU de Bicêtre, 78 rue du Général Leclerc, 94275 Le Kremlin Bicêtre cedex, France.
Intensive Care Med. 2002 Feb;28(2):154-63. doi: 10.1007/s00134-001-1172-7. Epub 2002 Jan 16.
To evaluate the attributable mortality associated with late-onset nosocomial pneumonia (LOP) while taking into account the severity at admission, the evolution of the patients during the first 4 days after admission to the ICU and the appropriateness of initial empiric antibiotic treatment.
Multicenter cohort study with prospective standardization of diagnostic interventions when nosocomial pneumonia develops.
Medical and surgical ICUs of four university-affiliated teaching hospitals.
Seven hundred sixty-four consecutive patients requiring ICU hospitalization for at least 4 days.
The clinical and biological data as well as the therapeutic data and the outcome were prospectively recorded from the day of admission to ICU discharge. Simplified Acute Physiologic Score (SAPS II) and Logistic Organ Dysfunction (LOD) score were collected and computed within the first 4 calendar days of ICU admission. Variables associated with the outcome were selected using a stepwise Cox model. The time to acquisition of the first LOP was then introduced in the final model as a time-dependent covariate. The analysis was stratified by ICU center. Finally, as initial antibiotic therapy could have an impact on the increased risk of death induced by LOP, the Cox model was applied again introducing LOP immediately adequately treated and LOP not immediately adequately treated as two different time-dependent covariates.
Late-onset pneumonia developed in 89 patients (12%). A McCabe score of more than 1, SAPS II score and increases in SAPS between days 1 and 2, days 2 and 3, and days 3 and 4 were significantly associated with an increased risk of death. When the time to acquisition of the first episode of LOP was introduced into the Cox model, the LOP occurrence was associated with increased mortality, even adjusted over the selected prognostic parameters and after stratification by center (hazard ratio (HR)=1.53, 95% CI 1.02-2.3, p=0.04). When LOP immediately adequately treated and LOP not immediately adequately treated were separately introduced into the Cox model, inappropriately treated LOP remained significantly associated with an increased risk of mortality (HR=1.69, 95% CI 1.08-2.65, p=0.022), whereas appropriately treated LOP did not (HR=1.44, 95% CI 0.75-2.76, p=0.27).
These data suggest that, in addition to severity scores, the underlying medical conditions and the evolution of severity within the first 4 days in ICU, late-onset pneumonia independently contribute to ICU patient mortality when empirical antibiotic treatment is not immediately appropriate.
评估晚发性医院获得性肺炎（LOP）相关的归因死亡率，同时考虑入院时的严重程度、入住重症监护病房（ICU）后前4天患者的病情变化以及初始经验性抗生素治疗的合理性。
多中心队列研究，当发生医院获得性肺炎时对诊断干预措施进行前瞻性标准化。
四家大学附属医院的内科和外科ICU。
764例连续入住ICU至少4天的患者。
从入住ICU至出院当天前瞻性记录临床和生物学数据、治疗数据及结局。在ICU入院后的前4个日历日内收集并计算简化急性生理学评分（SAPS II）和逻辑器官功能障碍（LOD）评分。使用逐步Cox模型选择与结局相关的变量。然后将首次发生LOP的时间作为时间依赖性协变量引入最终模型。分析按ICU中心进行分层。最后，由于初始抗生素治疗可能会影响LOP所致死亡风险的增加，再次应用Cox模型，将立即得到充分治疗的LOP和未立即得到充分治疗的LOP作为两个不同的时间依赖性协变量引入。
89例患者（12%）发生晚发性肺炎。McCabe评分大于1、SAPS II评分以及第1天和第2天、第2天和第3天、第3天和第4天SAPS的升高与死亡风险增加显著相关。当将首次发生LOP的时间引入Cox模型时，即使在根据选定的预后参数进行调整并按中心分层后，LOP的发生仍与死亡率增加相关（风险比（HR）=1.53，95%可信区间1.02 - 2.3，p = 0.04）。当将立即得到充分治疗的LOP和未立即得到充分治疗的LOP分别引入Cox模型时，未得到适当治疗的LOP仍与死亡风险增加显著相关（HR = 1.69，95%可信区间1.08 - 2.65，p = 0.022），而得到适当治疗的LOP则不然（HR = 1.44，95%可信区间0.75 - 2.76，p = 0.27）。
这些数据表明，除了严重程度评分外，潜在的医疗状况以及ICU内前4天严重程度的变化，当经验性抗生素治疗不立即适当时，晚发性肺炎会独立导致ICU患者死亡。
