BACKGROUND Protein-energy malnutrition (PEM), resulting from depleted energy and nutrient stores, compromises the body's defense systems and may exacerbate sepsis and its impact. However, population-based studies examining the association of PEM on the prevalence and health-care burden of sepsis are lacking. OBJECTIVE To investigate the relationship between PEM and sepsis, influence of PEM on clinical outcomes of sepsis, and impact of PEM on trends in sepsis mortality. DESIGN The primary study is a retrospective cohort analysis of the 2012-2014 National Inpatient Sample (NIS) patient discharge records. Secondary analyses are cross-sectional study on the 2014 NIS and trend analysis on 2007-2014 NIS. PARTICIPANTS/SETTING The primary study included adult inpatient hospitalizations for sepsis in the United States. MAIN OUTCOME MEASURES Mortality, complicated sepsis, and 10 other metrics of clinical outcomes and health care utilization. STATISTICAL ANALYSIS First, patients with sepsis (2014 NIS) were stratified into two groups: uncomplicated (without shock) and complicated (with shock). The adjusted odds ratio of having sepsis (total, uncomplicated, and complicated) was estimated with PEM as predictor using logistic regressions (binomial and multinomial). Second, among patients with sepsis (2012-2014 NIS), PEM cases were matched to cases without PEM (no-PEM) using a greedy-algorithm based propensity-matching methodology (1:1), and the outcomes were measured with conditional regression models. Finally, the trend in mortality from sepsis was calculated, stratified by PEM status, as an effect modifier, using Poisson models (2007-2014 NIS). All models accounted for the complex sampling methodology (SAS 9.4). RESULTS In 2014, PEM was associated with higher odds for sepsis (3.97 [3.89 to 4.05], P<0.0001) and complicated vs uncomplicated sepsis (1.74 [1.67 to 1.81], P<0.0001). From 2012-2014, about 18% (167,133 of 908,552) of hospitalizations for sepsis had coexisting PEM. After propensity matching, PEM was associated with higher mortality (adjusted odds ratio: 1.35 [1.32 to 1.37], P<0.0001), cost ($160,724 [159,517 to 161,940] vs $86,650 [85,931 to 87,375], P<0.0001), length of stay (14.8 [14.9 to 14.8] vs 8.5 [8.5 to 8.6] days, P<0.0001), adverse events, and resource utilization. Although mortality in sepsis has been trending down from 2007-2014 (-1.19% per year, P trend<0.0001), the decrease was less pronounced among those with PEM vs no-PEM (-0.86% per year vs -1.29% per year, P<0.0001). CONCLUSIONS PEM is a risk factor for sepsis and associated with poorer outcomes among patients with sepsis. A concerted effort involving all health care workers in the prevention, identification, and treatment of PEM in community-dwelling people before hospitalization might mitigate against these devastating outcomes.

中文翻译：

---

蛋白质能量营养不良与脓毒症更糟糕的结果相关——一项全国性分析

背景蛋白质-能量营养不良(PEM)，由耗尽的能量和营养储存引起，损害身体的防御系统并且可能加剧败血症及其影响。然而，缺乏基于人群的研究来检验 PEM 与败血症的患病率和医疗负担之间的关联。目的 调查 PEM 与脓毒症之间的关系、PEM 对脓毒症临床结果的影响以及 PEM 对脓毒症死亡率趋势的影响。设计 主要研究是对 2012-2014 年全国住院样本 (NIS) 患者出院记录的回顾性队列分析。次要分析是2014年NIS的横断面研究和2007-2014年NIS的趋势分析。参与者/地点 主要研究包括在美国因败血症住院的成人住院患者。主要结局指标死亡率，复杂的败血症，以及其他 10 个临床结果和医疗保健利用指标。统计分析 首先，脓毒症患者（2014 NIS）分为两组：无并发症（无休克）和复杂（有休克）。使用逻辑回归（二项式和多项式）以 PEM 作为预测因子估计了脓毒症（总体、非复杂性和复杂性）的调整优势比。其次，在脓毒症患者（2012-2014 NIS）中，使用基于贪婪算法的倾向匹配方法（1：1）将 PEM 病例与没有 PEM 的病例（无 PEM）进行匹配，并使用条件回归测量结果楷模。最后，使用泊松模型 (2007-2014 NIS) 计算脓毒症死亡率趋势，并按 PEM 状态分层，作为影响修饰符。所有模型都考虑了复杂的抽样方法 (SAS 9.4)。结果 2014 年，PEM 与脓毒症（3.97 [3.89 至 4.05]，P<0.0001）和复杂性与非复杂性脓毒症（1.74 [1.67 至 1.81]，P<0.0001）的较高几率相关。从 2012 年到 2014 年，大约 18%（908,552 例中的 167,133 例）因败血症而住院的患者同时存在 PEM。倾向匹配后，PEM 与更高的死亡率相关（调整后的优势比：1.35 [1.32 至 1.37]，P<0.0001）、成本（160,724 美元 [159,517 至 161,940] 比 86,650 美元 [85,931 至 1.5001]，长度<80001）停留（14.8 [14.9 至 14.8] 天 vs 8.5 [8.5 至 8.6] 天，P<0.0001）、不良事件和资源利用率。尽管脓毒症的死亡率从 2007 年到 2014 年一直呈下降趋势（每年 -1.19%，P 趋势 <0.0001），但在 PEM 与非 PEM 患者中，这种下降不太明显（每年 -0.86% 与 -1。每年 29%，P<0.0001)。结论 PEM 是脓毒症的危险因素，并且与脓毒症患者的较差预后相关。所有卫生保健工作者在住院前共同努力预防、识别和治疗社区居民的 PEM，可能会减轻这些破坏性后果。