Donini Lorenzo M, Savina Claudia, Ricciardi Laura Maria, Coletti Cecilia, Paolini Maddalena, Scavone Luciano, De Felice Maria Rosaria, Laviano Alessandro, Rossi Fanelli Filippo, Cannella Carlo
Department of Medical Physiopathology (Food Science Section), Sapienza University of Rome, Italy.
Nutrition. 2009 Jan;25(1):11-9. doi: 10.1016/j.nut.2008.07.001. Epub 2008 Oct 9.
Artificial nutrition (AN) is now considered medical therapy and has progressively become one of the mainstays of the different therapeutic options available for home or hospitalized patients, including surgical, medical, and critically ill patients. The clinical relevance of any therapy is based on its efficacy and effectiveness and thus on the improvement of its cost efficiency, i.e., the ability to provide benefits to the patients with minimal wasting of human and financial resources. The aim of the present study was to identify those indices, clinical, functional, or nutritional, that may reliably predict, before the start of AN, those patients who are likely not to benefit from nutritional support.
Three hundred twelve clinical charts of patients receiving AN between January 1999 and September 2006 were retrospectively examined. Data registered before starting AN were collected and analyzed: general data (age, sex), clinical conditions (comorbidity, quality of life, frailty), anthropometric and biochemical indices, type of AN treatment (total enteral nutrition, total parenteral nutrition, mixed AN), and outcome of treatment.
The percentage of negative outcomes (death or interruption of AN due to worsening clinical conditions within 10 d after starting AN) was meaningfully higher in subjects >80 y of age and with reduced social functions, higher comorbidity and/or frailty, reduced level of albumin, prealbumin, lymphocyte count, and cholinesterase and a higher level of C-reactive protein. The multivariate analysis showed that prealbumin and comorbidity were the best predictors of AN outcome. The logistic regression model with these variables showed a predictive value equal to 84.2%.
Proper prognostic instruments are necessary to perform optimal evaluations. The present study showed that a patient's general status (i.e., comorbidity, social quality of life, frailty) and nutritional and inflammatory statuses (i.e., lymphocyte count, albumin, prealbumin, C-reactive protein) have good predictive value on the effectiveness of AN.
人工营养（AN）现被视为一种医学治疗方法，并已逐渐成为可供居家或住院患者选择的不同治疗手段的主要支柱之一，这些患者包括外科手术患者、内科患者和危重症患者。任何治疗方法的临床相关性都基于其疗效和有效性，进而基于其成本效益的提高，即能够以最少的人力和财力资源浪费为患者带来益处。本研究的目的是确定那些临床、功能或营养指标，这些指标能够在开始人工营养之前可靠地预测哪些患者可能无法从营养支持中获益。
回顾性检查了1999年1月至2006年9月期间接受人工营养治疗的312例患者的临床病历。收集并分析了开始人工营养治疗前记录的数据：一般资料（年龄、性别）、临床状况（合并症、生活质量、虚弱程度）、人体测量和生化指标、人工营养治疗类型（全肠内营养、全肠外营养、混合性人工营养）以及治疗结果。
年龄大于80岁、社会功能减退、合并症较多和/或身体虚弱、白蛋白、前白蛋白、淋巴细胞计数和胆碱酯酶水平降低以及C反应蛋白水平升高的患者中，负面结果（死亡或在开始人工营养后10天内由于临床状况恶化而中断人工营养）的百分比明显更高。多因素分析表明，前白蛋白和合并症是人工营养治疗结果的最佳预测指标。包含这些变量的逻辑回归模型显示预测价值为84.2%。
需要合适的预后评估工具来进行最佳评估。本研究表明，患者的一般状况（即合并症、社会生活质量、虚弱程度）以及营养和炎症状态（即淋巴细胞计数、白蛋白、前白蛋白、C反应蛋白）对人工营养治疗的有效性具有良好的预测价值。
