- Email: [email protected]
Prediction of prognosis in chronic obstructive pulmonary disease patients with respiratory failure: A comparison of three nutritional assessment methods
European Journal of Internal Medicine xxx (xxxx) xxx–xxx
Contents lists available at ScienceDirect
European Journal of Internal Medicine journal homepage: www.elsevier.com/locate/ejim
Original Article
Prediction of prognosis in chronic obstructive pulmonary disease patients with respiratory failure: A comparison of three nutritional assessment methods Ruiqi Chena,1, Lu Xingb,1, Chao Youa, Xuemei Ouc,
⁎
a
West China School of medicine, West China Hospital of Sichuan University, Chengdu 610041, Sichuan Province, China West China Second University Hospital, Sichuan University; Key Laboratory of Birth Defects and Related Diseases of Women and Children (Sichuan University), Ministry of Education, Chengdu 610041, Sichuan Province, China c Department of Respiratory Medicine, West China Hospital of Sichuan University, Chengdu 610041, Sichuan Province, China b
A R T I C LE I N FO
A B S T R A C T
Keywords: Chronic obstructive pulmonary disease Respiratory failure Nutritional status Prognosis
Objectives: Due to their increased energy expenditure, chronic obstructive pulmonary disease (COPD) patients with respiratory failure are susceptible to malnutrition. This study aimed to compare the predictive values of the following three widely used nutritional assessment methods for the clinical prognosis of COPD patients with respiratory failure: body mass index (BMI), Nutritional Risk Screening 2002 (NRS 2002), and serum albumin (ALB) level. Methods: COPD patients with respiratory failure treated in our center from June 2013 to June 2016 were retrospectively included. Patient BMI, NRS 2002 and ALB values were measured to assess their nutritional status. A multivariable analysis was conducted, and receiver operating characteristic (ROC) curves were generated to explore the predictive factors for clinical prognoses. Results: A total of 438 qualified patients were enrolled in our study. Multivariable analysis revealed that the BMI and ALB values independently predicted in-hospital mortality, the BMI and NRS 2002 predicted 1-year mortality, and all three methods (BMI, NRS 2002, and ALB) predicted 30-day readmission after discharge (P < 0.05). Regarding the results of the AUROC analysis, the optimal cutoff values that maximized the ability to predict the prognosis were an ALB level of 30.5 g/L for in-hospital mortality, an NRS 2002 score of 3 points for 1year mortality, and an ALB level of 30.1 g/L for readmission within 30 days following discharge. Conclusions: For COPD patients with respiratory failure, ALB level was superior for predicting in-hospital mortality and 30-day readmission after discharge, and NRS 2002 was superior for long-term prognosis of 1-year mortality.
1. Introduction Chronic obstructive pulmonary disease (COPD) patients with respiratory failure are susceptible to malnutrition due to their increased energy expenditure [1]. Malnutrition has been associated with a poor prognosis, including high readmission rates and case mortality [2–4]. Therefore, it is essential to adopt an accurate and reasonable method to identify COPD patients with respiratory failure who are nutritionally at risk in a timely manner to provide nutritional support and prevent a poor prognosis.
Currently, several nutritional assessment methods have been developed. These methods generally assess nutritional status based on objective nutritional indicators, including food intake, weight reduction and biochemical indicators. Among these nutritional assessment methods, the body mass index (BMI) is the easiest and most popular method, and it has been extensively studied and conveniently used in COPD patients [5]. Several meta-analyses and original studies have reported that a poor BMI is associated with increased mortality in COPD patients, indicating that it is a predictor of clinical prognosis [6, 7]. However, as a systemic inflammatory disease, COPD includes different
Abbreviation list: ALB, Serum albumin; AUROC, Area under the ROC curve; BMI, Body mass index; CI, Confidence intervals; COPD, Chronic obstructive pulmonary disease; HIS, Hospital information system; IRB, Institutional review board; LR, Likelihood ratio; NRS 2002, Nutritional Risk Screening 2002; OR, Odds ratio; ROC, Receiver operating characteristic; SD, Standard deviation; WCH, West China Hospital ⁎ Corresponding author at: No.37 Guo Xue Xiang, Chengdu, Sichuan Province 610041, China. E-mail address: [email protected] (X. Ou). 1 These authors contributed equally to this work. https://doi.org/10.1016/j.ejim.2018.06.006 Received 2 February 2018; Received in revised form 2 June 2018; Accepted 6 June 2018 0953-6205/ © 2018 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
Please cite this article as: Chen, R., European Journal of Internal Medicine (2018), https://doi.org/10.1016/j.ejim.2018.06.006
European Journal of Internal Medicine xxx (xxxx) xxx–xxx
R. Chen et al.
1. BMI
subgroups (e.g., chronic bronchitis and emphysema) and clinical stages (stable and acute exacerbation). Therefore, the simple assessment of weight reduction by BMI may not accurately reflect the nutritional status of different types and stages of COPD patients. Compared to other nutritional assessment methods, the predictive power of the BMI is not consistent in COPD patients [8, 9]. The Nutritional Risk Screening 2002 (NRS 2002), developed by the European Society for Clinical Nutrition and Metabolism (ESPEN), is the only evidence-based tool to date and is widely used to assess nutritional status with good reliability and validity [10–12]. The NRS 2002 is simple, noninvasive [13, 14], and includes the assessment of age information and severity of COPD, which are considered good tools for reflecting the nutritional status of different types and clinical stages of COPD patients. However, due to a lack of related studies, whether the NRS 2002 can be used to predict the mortality and readmission rate of COPD patients with respiratory failure remains unknown. In addition to the BMI and NRS 2002, nutritional status can also be intuitively reflected by some biomarkers, among which, serum albumin (ALB) is considered accurate and has been widely used in clinical settings [15, 16]. Studies have shown that the ALB level is associated with mortality in COPD patients and is a better predictor of the prognosis than many other biochemical factors [17–19]. Furthermore, compared to other methods, the ALB level can rapidly evaluate a patient's malnourished status, therefore allowing timely nutritional treatment for COPD patients with respiratory failure and especially for those who experience acute exacerbation. However, ALB measurement is an invasive approach, with a higher cost and more complicated operation process than the BMI and NRS 2002. To our knowledge, studies focusing on the predictive value of different nutritional assessment methods for the prognosis of COPD patients with respiratory failure are still lacking. The current study aims to compare three widely used nutritional assessment methods (BMI, NRS 2002, and ALB) based on a cohort of COPD patients with respiratory failure. In addition to assessing in-hospital mortality, we also focused on 1-year mortality and 30-day readmission rates. Such data could help to determine which nutritional assessment method most accurately predicts the prognosis of COPD patients with respiratory failure.
BMI is a universal method for assessing the nutritional status of a subject [7, 21]. It is measured by the following equation: BMI = weight (kg) ÷ [height (m)]2 [22]. For Chinese people, a BMI < 18.5 kg/m2 is considered underweight, 18.5 kg/ m2 ≤ BMI ≤ 23.9 kg/m2 indicates normal weight, and a BMI ≥ 24.0 kg/m2 is considered overweight [23]. 2. NRS 2002 The NRS 2002 is a nutritional screening tool developed by the ESPEN [24]. This tool includes three parts: (1) nutritional status, evaluated by the variation of food intake in the past week, body weight, and BMI changes in the past 1–3 months; (2) severity of the disease; and (3) age ≥70 years. If the BMI < 18.5 kg/m2, the score of the NRS 2002 is 3 [23]. The total possible score of the NRS 2002 is 7, and a patient with a total score ≥ 3 is nutritionally at risk [25]. 3. ALB ALB is the most abundant protein in plasma and is a common marker used to determine a subject's nutritional status [15, 16, 19]. Based on the criteria, an ALB level ≥35 g/L is defined as good nutrition, 30 g/L ≤ ALB ≤ 34.9 g/L is defined as mild malnutrition, 25 g/ L ≤ ALB ≤ 29.9 g/L is defined as moderate malnutrition, and ALB < 25 g/L is defined as severe malnutrition [26]. In this study, ALB was measured by an automatic biochemical analyzer (Roche Modular-P800, Diamond Diagnostics Inc., Holliston, MA, USA) and its corresponding reagents. 2.3. Data collection For every patient, the demographic data on the first day of admission, including the respiratory rate, PaO2, PaCO2, NRS 2002 score, height, weight, and ALB level were obtained and entered into the study database by research assistants. For the patients without results on the first day, the earliest data after admission were collected. All the data were obtained from the hospital information system (HIS) of WCH, except for the NRS 2002 scores, which were assessed and calculated by an experienced nutrition nurse. In addition, research assistants collected outcomes of every patient at discharge and for each follow-up period, including in-hospital death, 12-month mortality after discharge, and readmission within 30 days after discharge. Based on the outcomes, participants were classified as survivors and non-survivors, or the readmitted group and non-readmitted group. Finally, the integrity and completeness of all data were regularly checked by another experienced researcher.
2. Material and methods 2.1. Materials and patients We performed a retrospective, observational study to compare the abilities of the BMI, NRS 2002 and ALB level to predict the prognosis of patients diagnosed with COPD and respiratory failure admitted to the Department of Respiratory Medicine, West China Hospital (WCH), from June 2013 to June 2016. The study was approved by the Institutional Review Board (IRB) of WCH. COPD was diagnosed according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) [20]. We recruited patients who were diagnosed with COPD, had a partial pressure of O2 (PaO2) < 60 mmHg and/or a partial pressure of CO2 (PaCO2) > 50 mmHg, were ≥18 years of age and had a respiratory rate > 23 bpm. Subjects with the following conditions were excluded: pregnancy or lactation; ventilatory dysfunction due to neuromuscular disorders; acute and chronic thromboembolic disease; severe illness in other systems such as malignancies and cardiovascular diseases; conditions causing complications that influence the nutritional status of participants such as diabetes; unconscious patients or those who could not assist in the nutritional assessment; patients with incomplete medical profile; and patients who declined to participate.
2.4. Statistical analyses SPSS statistical software (version 22.0; SPSS Inc., Chicago, Illinois, USA) and MedCalc statistical software (version 15.2; MedCalc Software, Mariakerke, Ostend, Belgium) were used for all statistical analyses. Quantitative data are expressed as the mean ± standard deviation (SD). Categorical data are expressed as frequencies and percentages. A binary logistic regression analysis was performed for the multivariate analysis. The receiver operating characteristic (ROC) curve was adopted, and the area under the ROC curve (AUROC) was used to evaluate the discriminative ability of the three nutritional assessment methods. The z statistic was used to calculate differences in the AUROC derived from the same cases. The sensitivity, specificity, positive likelihood ratio (+LR) and negative likelihood ratio (−LR) were also calculated to test the abilities and accuracy of the three methods9. The 95% confidence intervals (CI) of estimates were calculated for each variable. A value of P < 0.05 indicated statistical significance. When
2.2. Nutritional assessment methods Patient BMI, NRS 2002 scores and ALB levels were measured to assess nutritional status, and all the participants completed their assessments within 1 day after admission. 2
European Journal of Internal Medicine xxx (xxxx) xxx–xxx
R. Chen et al.
Table 1 Characteristics of the study population (n = 438). Variable Age, years (Mean ± SD) Sex, n (%) Marriage
Height, m (Mean ± SD) Weight, kg (Mean ± SD) BMI, kg/m2 (Mean ± SD) BMI category, n (%)
NRS 2002 (Mean ± SD) NRS 2002 category, n (%) ALB, g/L (Mean ± SD) ALB category, n (%)
In-hospital death, n (%) 12-month death, n (%) Readmission, n (%)
Fig. 1. Flowchart for the study.
Value
Male Female Single Married Divorced Widowhood
Underweight Normal weight Overweight With nutritional risk (≥3) Without nutritional risk (< 3) Good nutrition Mild malnutrition Moderate malnutrition Severe malnutrition Survivors Non-survivors Survivors Non-survivors Yes No
70.28 ( ± 10.16) 333 (76.0%) 105 (24.0%) 30 (6.8%) 374 (85.4%) 22 (5.0%) 12 (2.7%) 1.62 ( ± 0.07) 56.30 ( ± 9.62) 21.11 ( ± 3.59) 123 (28.1%) 219 (50.0%) 96 (21.9%) 2.85 ( ± 1.34) 239 (54.6%) 199 (45.4%) 35.11 ( ± 5.95) 225 (51.4%) 131 (29.9%) 48 (11.0%) 34 (7.8%) 406 (92.7%) 32 (7.3%) 381 (87.0%) 57 (13.0%) 31 (7.1%) 417 (95.2%)
BMI = body mass index, NRS = Nutritional Risk Screening, ALB = serum albumin, SD = standard deviation.
the three variables were compared in pairs, the Šidák-adjusted level of approximately 0.017 was used to indicate statistical significance when P < 0.017.
discharge. The AUROC for the NRS 2002 for predicting 1-year mortality was significantly larger than that for the BMI (0.835 (95% CI 0.797–0.869) vs 0.674 (95% CI 0.628–0.718), P = 0.0004) (Fig. 2B). Readmission: The total readmission rate was 4.8% (21 patients). The AUROC values for the BMI, NRS 2002 score and ALB level for predicting readmission were 0.798 (95% CI 0.757–0.834), 0.883 (95% CI 0.850–0.912) and 0.924 (95% CI 0.895–0.947), respectively (Fig. 2C). The Šidák-adjusted significance level (P = 0.017) was used when the three variables were compared in pairs. Statistically, no significant differences were found between the AUROC for the BMI and the NRS 2002 score (P = 0.0923) or the AUROC for the NRS 2002 score and ALB level (P = 0.3274), but the AUROC for the ALB level was significantly larger than that for the BMI (P = 0.0025).
3. Results 3.1. Patient characteristics A total of 480 COPD patients with respiratory failure were admitted to our hospital from December 2012 to June 2015. Of those patients, 438 were enrolled in our study. A flowchart of the reasons for exclusion is presented in Fig. 1. The characteristics of the included cohort are summarized in Table 1. 3.2. Multivariate analysis As shown in Table 2, the BMI and ALB values independently predicted in-hospital mortality in the COPD patients with respiratory failure (odds ratio (OR) 11.498, 95% CI 4.469–29.581, P = 0.000 and OR 2.503, 95% CI 1.356–4.620, P = 0.003, respectively). Patient BMI and NRS 2002 scores predicted 1-year mortality after discharge (OR 6.494, 95% CI 3.273–12.888, P = 0.000 and OR 0.222, 95% CI 0.120–0.410, P = 0.000, respectively). All three nutritional assessment methods (BMI, NRS 2002, ALB) independently predicted readmission within 30 days after discharge (OR 3.835, 95% CI 1.383–10.632, P = 0.010; OR 0.328, 95% CI 0.108–0.994, P = 0.042; OR 2.854, 95% CI 1.509–5.397, P = 0.001, respectively). BMI = body mass index, NRS = Nutritional Risk Screening, ALB = serum albumin, CI = confidence interval, OR = odds ratio.
3.4. Optimal cutoff values The cutoff values that maximized the sum of the sensitivity and specificity for the mortality and readmission prediction were calculated for the BMI, NRS 2002 and ALB. According to the AUROC results, the optimal cutoff value that maximized the ability to predict in-hospital mortality was 30.5 g/L for ALB; for 1-year mortality prediction, the optimal cutoff value was 3 points for the NRS 2002; for re-admittance within 30 days following discharge, the optimal cutoff value was 30.1 g/L for ALB (Table 3). 4. Discussion COPD patients with respiratory failure often also have different degrees of malnutrition, which may lead to treatment failure and could be associated with a poor prognosis and high mortality. Therefore, it is essential to adopt an accurate, appropriate and reasonable method of identifying COPD patients with respiratory failure who are nutritionally at risk in a timely manner and to provide them with the appropriate nutritional support to prevent a poor clinical prognosis. Though many nutritional assessment methods have been developed, few methods have proven to accurately predict the prognosis of COPD patients with
3.3. AUROC curves In-hospital mortality: Fifty-seven (13.0%) patients died during hospitalization. The AUROC for BMI and ALB for predicting in-hospital mortality were 0.704 (95% CI 0.659–0.746) and 0.908 (95% CI 0.876–0.933), respectively (Fig. 2A), and the AUROC for ALB was significantly larger than that for BMI (P < 0.0001). One-year mortality: Eighty (18.3%) patients died within 1 year after 3
European Journal of Internal Medicine xxx (xxxx) xxx–xxx
R. Chen et al.
Table 2 Multivariate analysis of in-hospital mortality, 12-month mortality and 30-day readmission. Variables
Age Sex Marriage BMI NRS 2002 score ALB
In-hospital mortality
12-month mortality
30-day readmission
OR (95% CI)
P
OR (95% CI)
P
OR (95% CI)
P
1.031 (0.956–1.111) 0.892 (0.107–7.467) 2.777 (0.728–10.592) 11.498 (4.469–29.581) 0.537 (0.211–1.364) 2.503 (1.356–4.620)
0.431 0.916 0.135 0.000 0.191 0.003
1.017 0.597 4.364 6.494 0.222 1.140
0.434 0.364 0.090 0.000 0.000 0.180
0.930 2.149 3.569 3.835 0.328 2.854
0.131 0.498 0.167 0.010 0.042 0.001
respiratory failure. Based on the results of our study, ALB and BMI were both independent predictors of in-hospital mortality in COPD patients with respiratory failure, but ALB was superior to BMI (AUROC 0.908 vs 0.704, P < 0.0001), indicating that the ALB level might be more accurate for the prediction of short-term mortality. This result was consistent with several previous studies [27, 28]. In the short term, COPD with respiratory failure presents with relatively faster protein consumption, which can be more quickly and accurately detected by ALB levels rather than by BMI. In our study, the cutoff value that maximized the ability to predict in-hospital mortality was 30.5 g/L for ALB. Therefore, for a COPD patient with respiratory failure whose ALB level is lower than 30.5 g/L on admission, the risk of in-hospital death is higher, and appropriate nutritional support should be provided to reduce this risk. However, the ALB measurement is an invasive method that causes pain and fear in patients and requires skilled nurses for venipuncture. Additionally, compared to other nutritional assessment methods, ALB detection is more expensive. Therefore, if ALB is selected to predict the prognosis, the quality of the blood test should be verified. In addition, a detailed explanation should be provided to patients to obtain their cooperation, and an effective venipuncture technique should be used to ease their pain. Regarding 1-year mortality, both the BMI and NRS 2002 score were independent predictors of COPD patients with respiratory failure, but the NRS 2002 was superior to the BMI for the 1-year mortality prediction (AUROC 0.835 vs 0.674, P = 0.0004), indicating that the NRS 2002 may have a better ability to predict long-term mortality. Compared to the nutritional status assessment by the single factor of BMI, the assessment of the nutritional status of COPD patients by the NRS 2002, includes information on age and the severity of the disease (e.g., presence or absence of severe pneumonia, whether the patient presented with acute exacerbation, use of a ventilator). Thus, the NRS 2002 can comprehensively reflect the nutritional status of COPD
(0.976–1.059) (0.195–1.821) (1.446–13.172) (3.273–12.888) (0.120–0.410) (0.941–1.382)
(0.846–1.022) (0.235–19.643) (0.588–21.667) (1.383–10.632) (0.108–0.994) (1.509–5.397)
Table 3 Optimal cutoff values for the three nutritional assessment methods (BMI, NRS 2002 and ALB). Outcomes
Nutritional assessment methods BMI
In-hospital mortality
Criterion Sensitivity Specificity
12-month mortality
+LR −LR Criterion Sensitivity Specificity
Readmission
+LR −LR Criterion Sensitivity Specificity +LR −LR
≤19.53 kg/m 77.19 (64.2–87.3) 66.67 (61.7–71.4) 2.32 (1.9–2.8) 0.34 (0.2–0.6) ≤19.53 kg/m2 72.50 (61.4–81.9) 68.44 (63.3–73.2) 2.30 (1.9–2.8) 0.40 (0.3–0.6) ≤19.53 kg/m2 95.24 (76.2–99.9) 63.79 (59.0–68.4) 2.63 (2.2–3.1) 2
0.075 (0.01–0.5)
NRS 2002
ALB
>3 80.70 (68.1–90.0) 75.33 (70.7–79.6) 3.27 (2.6–4.1) 0.26 (0.2–0.4) >3 76.25 (65.4–85.1) 77.93 (73.3–82.1) 3.46 (2.7–4.3) 0.30 (0.2–0.5) >4 71.43 (47.8–88.7) 91.85 (88.8–94.3) 8.76 (5.8–13.3) 0.31 (0.2–0.6)
≤30.5 g/L 84.21 (72.1–92.5) 87.93 (84.2–91.0) 6.97 (5.2–9.4) 0.18 (0.1–0.3) ≤30.5 g/L 65.00 (53.5–75.3) 88.27 (84.5–91.4) 5.54 (4.0–7.7) 0.40 (0.3–0.5) ≤30.1 g/L 95.24 (76.2–99.9) 83.69 (79.8–87.1) 5.84 (4.6–7.4) 0.057 (0.008–0.4)
BMI = body mass index, NRS = Nutritional Risk Screening, ALB = serum albumin, +LR = positive likelihood ratio, −LR = negative likelihood ratio.
patients with respiratory failure and accurately predict their long-term prognosis. The AUROC for the NRS 2002 was 0.835, indicating a good predictive ability. This result was confirmed by our previous study [29], which reported that the NRS 2002 could be effectively used as a
Fig. 2. A: ROC curves for the BMI and ALB for predicting in-hospital mortality; B: ROC curves for the BMI and the NRS 2002 for predicting 1-year mortality after discharge; C: ROC curves for the BMI and the NRS 2002 and ALB for predicting readmission rates within 30 days after discharge. AUROC = area under the ROC curve, BMI = body mass index, ALB = serum albumin, COPD = chronic obstructive pulmonary disease, ROC = receiver operating characteristic. 4
European Journal of Internal Medicine xxx (xxxx) xxx–xxx
R. Chen et al.
Declarations of interest
predictive tool for 12-month mortality in COPD patients with respiratory failure. The ROC curves revealed that the cutoff value that maximized predictive ability for long-term mortality was a score of 3 for the NRS 2002, indicating that a COPD patient with respiratory failure is more likely to die within 1 year following discharge when his or her NRS 2002 score is higher than 3 on admission. The NRS 2002 is the ESPEN-recommended nutritional risk screening tool and is more convenient than most of the other available methods due to its simplicity, feasibility and noninvasiveness. However, because the NRS 2002 includes subjective content, the result of the assessment is somewhat dependent on the assessors. Therefore, we suggest that an experienced assessor is highly necessary, and combining the NRS 2002 with other objective nutritional assessment methods may produce more accurate results when assessing a patient's nutritional status in clinical practice. The multivariate analysis showed that in COPD patients with respiratory failure, the BMI, NRS 2002 score and ALB level were all independent predictors of readmission within 30 days after discharge. According to the ROC curves, we found that ALB had the greatest ability to predict readmission within 30 days following discharge (AUROC 0.924, P < 0.0001). This finding is consistent with a study by Adogwa et al. [30], which reported that the ALB level is an independent predictor of the risk of 30-day hospital readmission. Furthermore, ROC curves showed that a cutoff value of 30.1 g/L for ALB maximized the ability to predict readmission within 30 days following discharge, indicating that COPD patients with respiratory failure have a higher risk of readmission within 30 days when their ALB level is lower than 30.1 g/L on admission. Nutritional support is also suggested for these patients. To our knowledge, this study is the first to compare the ability of three nutritional assessment methods (BMI, NRS 2002, ALB) to predict the clinical prognosis of COPD patients with respiratory failure. In addition to in-hospital mortality, the study also included 1-year mortality and readmission within 30 days after discharge as prognostic outcomes, which made this study more integrated and meaningful. However, some limitations of this study should be mentioned. First, this was a retrospective, observational, single-center study; however, we included a large number of patients. This may influence the application of our results in centers with a different case mix, and the identification of certain factors affecting the prognosis may have been eliminated by the homogeneity of the therapeutic environment and study population. Second, the patients were only measured via three nutritional assessment methods on admission, and the results may not represent the dynamic conditions of patients. Third, all the subjects recruited were conscious when included to allow successful completion of the assessments. Therefore, the application of the results of this study may be limited. Finally, because there is no gold standard nutritional assessment method to predict prognosis in COPD patients with respiratory failure, we did not compare the ROC curves of the three methods with a gold standard method. Further prospective multicenter studies are warranted to confirm and improve our findings.
None. Funding This work was supported by the National Science Foundation of China (Grants No.31671189) and Foundation of Sichuan Provincial Science and Technology Committee agency (Grants No.18ZDYF2039). Acknowledgements None. References [1] Cui J, Wan Q, Wu X, Zeng Y, Jiang L, Ao D, et al. Nutritional risk screening 2002 as a predictor of outcome during general Ward-based noninvasive ventilation in chronic obstructive pulmonary disease with respiratory failure. Med Sci Monit 2015;21:2786–93. [2] Hallin R, Gudmundsson G, Ulrik CS, Nieminen MM, Gislason T, Lindberg E, et al. Nutritional status and long-term mortality in hospitalised patients with chronic obstructive pulmonary disease (COPD). Respir Med 2007;101:1954–60. [3] Shuangjun He. The impact of nutrition status on clinical prognosis in elderly patients with acute exacerbation of chronic obstructive pulmonary disease. J Intern Med Concepts Pract 2014;9:407–10. [4] Zapatero A, Barba R, Ruiz J, Losa J, Plaza S, Canora J, et al. Malnutrition and obesity: influence in mortality and readmissions in chronic obstructive pulmonary disease patients. J Hum Nutr Diet 2013;26:16–22. [5] Cao C, Wang R, Wang J, Bunjhoo H, Xu Y, Xiong W. Body mass index and mortality in chronic obstructive pulmonary disease: a meta-analysis. PLoS One 2012;7:e43892. [6] Guo Y, Zhang T, Wang Z, Yu F, Xu Q, Guo W, et al. Body mass index and mortality in chronic obstructive pulmonary disease: a dose–response meta-analysis. Medicine 2016;95:e4225. [7] Yamauchi Y, Hasegawa W, Yasunaga H, Sunohara M, Jo T, Takami K, et al. Paradoxical association between body mass index and in-hospital mortality in elderly patients with chronic obstructive pulmonary disease in Japan. Int J Chron Obstruct Pulmon Dis 2014;9:1337–46. [8] Marquis K, Debigaré R, Lacasse Y, LeBlanc P, Jobin J, Carrier G, et al. Midthigh muscle cross-sectional area is a better predictor of mortality than body mass index in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2002;166:809–13. [9] Vestbo J, Prescott E, Almdal T, Dahl M, Nordestgaard BG, Andersen T, et al. Body mass, fat-free body mass, and prognosis in patients with chronic obstructive pulmonary disease from a random population sample: findings from the Copenhagen city heart study. Am J Respir Crit Care Med 2006;173:79–83. [10] Velasco C, García E, Rodríguez V, Frias L, Garriga R, Álvarez J, et al. Comparison of four nutritional screening tools to detect nutritional risk in hospitalized patients: a multicentre study. Eur J Clin Nutr 2011;65:269–74. [11] Raslan M, Gonzalez MC, Dias MCG, Nascimento M, Castro M, Marques P, et al. Comparison of nutritional risk screening tools for predicting clinical outcomes in hospitalized patients. Nutrition 2010;26:721–6. [12] Almeida AI, Correia M, Camilo M, Ravasco P. Nutritional risk screening in surgery: valid, feasible, easy!. Clin Nutr 2012;31:206–11. [13] Liu P, Zhang Z-F, Cai J-J, Wang B-S, Yan X. NRS2002 assesses nutritional status of leukemia patients undergoing hematopoietic stem cell transplantation. Chin J Cancer Res 2012;24:299–303. [14] Liu P, Yan X, Wang B, Xu X. Three methods assess nutritional status of leukemia patients before hematopoietic stem cell transplantation. Chin Med J (Engl) 2012;125:440–3. [15] Gama-Axelsson T, Heimbürger O, Stenvinkel P, Bárány P, Lindholm B, Qureshi AR. Serum albumin as predictor of nutritional status in patients with ESRD. Clin J Am Soc Nephrol 2012;7:1446–53. [16] Helliksson F, Wernerman J, Wiklund L, Rosell J, Karlsson M. The combined use of three widely available biochemical markers as predictor of organ failure in critically ill patients. Scand J Clin Lab Invest 2016;76:479–85. [17] Herrmann FR, Safran C, Levkoff SE, Minaker KL. Serum albumin level on admission as a predictor of death, length of stay, and readmission. Arch Intern Med 1992;152:125–30. [18] Gunen H, Hacievliyagil S, Kosar F, Mutlu L, Gulbas G, Pehlivan E, et al. Factors affecting survival of hospitalised patients with COPD. Eur Respir J 2005;26:234–41. [19] Fanali G, di Masi A, Trezza V, Marino M, Fasano M, Ascenzi P. Human serum albumin: from bench to bedside. Mol Aspects Med 2012;33:209–90. [20] Rabe KF, Hurd S, Anzueto A, Barnes PJ, Buist SA, Calverley P, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary. Am J Respir Crit Care Med 2007;176:532–55. [21] Jésus P, Achamrah N, Grigioni S, Charles J, Rimbert A, Folope V, et al. Validity of predictive equations for resting energy expenditure according to the body mass index in a population of 1726 patients followed in a nutrition unit. Clin Nutr 2015;34:529–35.
5. Conclusion For COPD patients with respiratory failure, the best nutritional assessment method varies for the prediction of different prognostic outcomes. The ALB level is superior for predicting in-hospital mortality and 30-day readmission after discharge, while the NRS 2002 score better predicts the long-term prognosis, including 1-year mortality.
Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings described in this paper. 5
European Journal of Internal Medicine xxx (xxxx) xxx–xxx
R. Chen et al.
1998;31:997–1006. [28] Zeng H, Luo H, Tang Y, Zhou Z, Kong X. The purpose of the study is to explore the relationship between serum IL-18, TFN-αand nutritional status, prognosis in patients with COPD in severe stage. Am J Respir Crit Care Med 2013;187:A1521. [29] Chen R, Xing L, You C. Nutritional risk screening 2002 should be used in hospitalized patients with chronic obstructive pulmonary disease with respiratory failure to determine prognosis: a validation on a large Chinese cohort. Eur J Intern Med 2016;36:e16-e7. [30] Adogwa O, Elsamadicy AA, Mehta AI, Cheng J, Bagley CA, Karikari IO. Pre-operative nutritional status is an independent predictor of 30-day hospital readmission after elective spine surgery. Spine 2016;41:1400–4.
[22] Stratton RJ, Green CJ, Elia M. Disease-related malnutrition: an evidence-based approach to treatment. J Hum Nutr Diet 2003;16:375. [23] Chen C. Introduction of body mass index classification suggestion in Chinese adult. Zhonghua Yu Fang Yi Xue Za Zhi 2001;35:349–50. [24] Kondrup J, Rasmussen HH, Hamberg O, Stanga Z, Group AahEW. Nutritional risk screening (NRS 2002): a new method based on an analysis of controlled clinical trials. Clin Nutr 2003;22:321–36. [25] Kondrup J, Allison SP, Elia M, Vellas B, Plauth M. ESPEN guidelines for nutrition screening 2002. Clin Nutr 2003;22:415–21. [26] Zhang A. Medical nutrition. Bejing: People's Medical Publishing House; 1998. [27] Leavey SF, Strawderman RL, Jones CA, Port FK, Held PJ. Am J Kidney Dis
6
Contents lists available at ScienceDirect
European Journal of Internal Medicine journal homepage: www.elsevier.com/locate/ejim
Original Article
Prediction of prognosis in chronic obstructive pulmonary disease patients with respiratory failure: A comparison of three nutritional assessment methods Ruiqi Chena,1, Lu Xingb,1, Chao Youa, Xuemei Ouc,
⁎
a
West China School of medicine, West China Hospital of Sichuan University, Chengdu 610041, Sichuan Province, China West China Second University Hospital, Sichuan University; Key Laboratory of Birth Defects and Related Diseases of Women and Children (Sichuan University), Ministry of Education, Chengdu 610041, Sichuan Province, China c Department of Respiratory Medicine, West China Hospital of Sichuan University, Chengdu 610041, Sichuan Province, China b
A R T I C LE I N FO
A B S T R A C T
Keywords: Chronic obstructive pulmonary disease Respiratory failure Nutritional status Prognosis
Objectives: Due to their increased energy expenditure, chronic obstructive pulmonary disease (COPD) patients with respiratory failure are susceptible to malnutrition. This study aimed to compare the predictive values of the following three widely used nutritional assessment methods for the clinical prognosis of COPD patients with respiratory failure: body mass index (BMI), Nutritional Risk Screening 2002 (NRS 2002), and serum albumin (ALB) level. Methods: COPD patients with respiratory failure treated in our center from June 2013 to June 2016 were retrospectively included. Patient BMI, NRS 2002 and ALB values were measured to assess their nutritional status. A multivariable analysis was conducted, and receiver operating characteristic (ROC) curves were generated to explore the predictive factors for clinical prognoses. Results: A total of 438 qualified patients were enrolled in our study. Multivariable analysis revealed that the BMI and ALB values independently predicted in-hospital mortality, the BMI and NRS 2002 predicted 1-year mortality, and all three methods (BMI, NRS 2002, and ALB) predicted 30-day readmission after discharge (P < 0.05). Regarding the results of the AUROC analysis, the optimal cutoff values that maximized the ability to predict the prognosis were an ALB level of 30.5 g/L for in-hospital mortality, an NRS 2002 score of 3 points for 1year mortality, and an ALB level of 30.1 g/L for readmission within 30 days following discharge. Conclusions: For COPD patients with respiratory failure, ALB level was superior for predicting in-hospital mortality and 30-day readmission after discharge, and NRS 2002 was superior for long-term prognosis of 1-year mortality.
1. Introduction Chronic obstructive pulmonary disease (COPD) patients with respiratory failure are susceptible to malnutrition due to their increased energy expenditure [1]. Malnutrition has been associated with a poor prognosis, including high readmission rates and case mortality [2–4]. Therefore, it is essential to adopt an accurate and reasonable method to identify COPD patients with respiratory failure who are nutritionally at risk in a timely manner to provide nutritional support and prevent a poor prognosis.
Currently, several nutritional assessment methods have been developed. These methods generally assess nutritional status based on objective nutritional indicators, including food intake, weight reduction and biochemical indicators. Among these nutritional assessment methods, the body mass index (BMI) is the easiest and most popular method, and it has been extensively studied and conveniently used in COPD patients [5]. Several meta-analyses and original studies have reported that a poor BMI is associated with increased mortality in COPD patients, indicating that it is a predictor of clinical prognosis [6, 7]. However, as a systemic inflammatory disease, COPD includes different
Abbreviation list: ALB, Serum albumin; AUROC, Area under the ROC curve; BMI, Body mass index; CI, Confidence intervals; COPD, Chronic obstructive pulmonary disease; HIS, Hospital information system; IRB, Institutional review board; LR, Likelihood ratio; NRS 2002, Nutritional Risk Screening 2002; OR, Odds ratio; ROC, Receiver operating characteristic; SD, Standard deviation; WCH, West China Hospital ⁎ Corresponding author at: No.37 Guo Xue Xiang, Chengdu, Sichuan Province 610041, China. E-mail address: [email protected] (X. Ou). 1 These authors contributed equally to this work. https://doi.org/10.1016/j.ejim.2018.06.006 Received 2 February 2018; Received in revised form 2 June 2018; Accepted 6 June 2018 0953-6205/ © 2018 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
Please cite this article as: Chen, R., European Journal of Internal Medicine (2018), https://doi.org/10.1016/j.ejim.2018.06.006
European Journal of Internal Medicine xxx (xxxx) xxx–xxx
R. Chen et al.
1. BMI
subgroups (e.g., chronic bronchitis and emphysema) and clinical stages (stable and acute exacerbation). Therefore, the simple assessment of weight reduction by BMI may not accurately reflect the nutritional status of different types and stages of COPD patients. Compared to other nutritional assessment methods, the predictive power of the BMI is not consistent in COPD patients [8, 9]. The Nutritional Risk Screening 2002 (NRS 2002), developed by the European Society for Clinical Nutrition and Metabolism (ESPEN), is the only evidence-based tool to date and is widely used to assess nutritional status with good reliability and validity [10–12]. The NRS 2002 is simple, noninvasive [13, 14], and includes the assessment of age information and severity of COPD, which are considered good tools for reflecting the nutritional status of different types and clinical stages of COPD patients. However, due to a lack of related studies, whether the NRS 2002 can be used to predict the mortality and readmission rate of COPD patients with respiratory failure remains unknown. In addition to the BMI and NRS 2002, nutritional status can also be intuitively reflected by some biomarkers, among which, serum albumin (ALB) is considered accurate and has been widely used in clinical settings [15, 16]. Studies have shown that the ALB level is associated with mortality in COPD patients and is a better predictor of the prognosis than many other biochemical factors [17–19]. Furthermore, compared to other methods, the ALB level can rapidly evaluate a patient's malnourished status, therefore allowing timely nutritional treatment for COPD patients with respiratory failure and especially for those who experience acute exacerbation. However, ALB measurement is an invasive approach, with a higher cost and more complicated operation process than the BMI and NRS 2002. To our knowledge, studies focusing on the predictive value of different nutritional assessment methods for the prognosis of COPD patients with respiratory failure are still lacking. The current study aims to compare three widely used nutritional assessment methods (BMI, NRS 2002, and ALB) based on a cohort of COPD patients with respiratory failure. In addition to assessing in-hospital mortality, we also focused on 1-year mortality and 30-day readmission rates. Such data could help to determine which nutritional assessment method most accurately predicts the prognosis of COPD patients with respiratory failure.
BMI is a universal method for assessing the nutritional status of a subject [7, 21]. It is measured by the following equation: BMI = weight (kg) ÷ [height (m)]2 [22]. For Chinese people, a BMI < 18.5 kg/m2 is considered underweight, 18.5 kg/ m2 ≤ BMI ≤ 23.9 kg/m2 indicates normal weight, and a BMI ≥ 24.0 kg/m2 is considered overweight [23]. 2. NRS 2002 The NRS 2002 is a nutritional screening tool developed by the ESPEN [24]. This tool includes three parts: (1) nutritional status, evaluated by the variation of food intake in the past week, body weight, and BMI changes in the past 1–3 months; (2) severity of the disease; and (3) age ≥70 years. If the BMI < 18.5 kg/m2, the score of the NRS 2002 is 3 [23]. The total possible score of the NRS 2002 is 7, and a patient with a total score ≥ 3 is nutritionally at risk [25]. 3. ALB ALB is the most abundant protein in plasma and is a common marker used to determine a subject's nutritional status [15, 16, 19]. Based on the criteria, an ALB level ≥35 g/L is defined as good nutrition, 30 g/L ≤ ALB ≤ 34.9 g/L is defined as mild malnutrition, 25 g/ L ≤ ALB ≤ 29.9 g/L is defined as moderate malnutrition, and ALB < 25 g/L is defined as severe malnutrition [26]. In this study, ALB was measured by an automatic biochemical analyzer (Roche Modular-P800, Diamond Diagnostics Inc., Holliston, MA, USA) and its corresponding reagents. 2.3. Data collection For every patient, the demographic data on the first day of admission, including the respiratory rate, PaO2, PaCO2, NRS 2002 score, height, weight, and ALB level were obtained and entered into the study database by research assistants. For the patients without results on the first day, the earliest data after admission were collected. All the data were obtained from the hospital information system (HIS) of WCH, except for the NRS 2002 scores, which were assessed and calculated by an experienced nutrition nurse. In addition, research assistants collected outcomes of every patient at discharge and for each follow-up period, including in-hospital death, 12-month mortality after discharge, and readmission within 30 days after discharge. Based on the outcomes, participants were classified as survivors and non-survivors, or the readmitted group and non-readmitted group. Finally, the integrity and completeness of all data were regularly checked by another experienced researcher.
2. Material and methods 2.1. Materials and patients We performed a retrospective, observational study to compare the abilities of the BMI, NRS 2002 and ALB level to predict the prognosis of patients diagnosed with COPD and respiratory failure admitted to the Department of Respiratory Medicine, West China Hospital (WCH), from June 2013 to June 2016. The study was approved by the Institutional Review Board (IRB) of WCH. COPD was diagnosed according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) [20]. We recruited patients who were diagnosed with COPD, had a partial pressure of O2 (PaO2) < 60 mmHg and/or a partial pressure of CO2 (PaCO2) > 50 mmHg, were ≥18 years of age and had a respiratory rate > 23 bpm. Subjects with the following conditions were excluded: pregnancy or lactation; ventilatory dysfunction due to neuromuscular disorders; acute and chronic thromboembolic disease; severe illness in other systems such as malignancies and cardiovascular diseases; conditions causing complications that influence the nutritional status of participants such as diabetes; unconscious patients or those who could not assist in the nutritional assessment; patients with incomplete medical profile; and patients who declined to participate.
2.4. Statistical analyses SPSS statistical software (version 22.0; SPSS Inc., Chicago, Illinois, USA) and MedCalc statistical software (version 15.2; MedCalc Software, Mariakerke, Ostend, Belgium) were used for all statistical analyses. Quantitative data are expressed as the mean ± standard deviation (SD). Categorical data are expressed as frequencies and percentages. A binary logistic regression analysis was performed for the multivariate analysis. The receiver operating characteristic (ROC) curve was adopted, and the area under the ROC curve (AUROC) was used to evaluate the discriminative ability of the three nutritional assessment methods. The z statistic was used to calculate differences in the AUROC derived from the same cases. The sensitivity, specificity, positive likelihood ratio (+LR) and negative likelihood ratio (−LR) were also calculated to test the abilities and accuracy of the three methods9. The 95% confidence intervals (CI) of estimates were calculated for each variable. A value of P < 0.05 indicated statistical significance. When
2.2. Nutritional assessment methods Patient BMI, NRS 2002 scores and ALB levels were measured to assess nutritional status, and all the participants completed their assessments within 1 day after admission. 2
European Journal of Internal Medicine xxx (xxxx) xxx–xxx
R. Chen et al.
Table 1 Characteristics of the study population (n = 438). Variable Age, years (Mean ± SD) Sex, n (%) Marriage
Height, m (Mean ± SD) Weight, kg (Mean ± SD) BMI, kg/m2 (Mean ± SD) BMI category, n (%)
NRS 2002 (Mean ± SD) NRS 2002 category, n (%) ALB, g/L (Mean ± SD) ALB category, n (%)
In-hospital death, n (%) 12-month death, n (%) Readmission, n (%)
Fig. 1. Flowchart for the study.
Value
Male Female Single Married Divorced Widowhood
Underweight Normal weight Overweight With nutritional risk (≥3) Without nutritional risk (< 3) Good nutrition Mild malnutrition Moderate malnutrition Severe malnutrition Survivors Non-survivors Survivors Non-survivors Yes No
70.28 ( ± 10.16) 333 (76.0%) 105 (24.0%) 30 (6.8%) 374 (85.4%) 22 (5.0%) 12 (2.7%) 1.62 ( ± 0.07) 56.30 ( ± 9.62) 21.11 ( ± 3.59) 123 (28.1%) 219 (50.0%) 96 (21.9%) 2.85 ( ± 1.34) 239 (54.6%) 199 (45.4%) 35.11 ( ± 5.95) 225 (51.4%) 131 (29.9%) 48 (11.0%) 34 (7.8%) 406 (92.7%) 32 (7.3%) 381 (87.0%) 57 (13.0%) 31 (7.1%) 417 (95.2%)
BMI = body mass index, NRS = Nutritional Risk Screening, ALB = serum albumin, SD = standard deviation.
the three variables were compared in pairs, the Šidák-adjusted level of approximately 0.017 was used to indicate statistical significance when P < 0.017.
discharge. The AUROC for the NRS 2002 for predicting 1-year mortality was significantly larger than that for the BMI (0.835 (95% CI 0.797–0.869) vs 0.674 (95% CI 0.628–0.718), P = 0.0004) (Fig. 2B). Readmission: The total readmission rate was 4.8% (21 patients). The AUROC values for the BMI, NRS 2002 score and ALB level for predicting readmission were 0.798 (95% CI 0.757–0.834), 0.883 (95% CI 0.850–0.912) and 0.924 (95% CI 0.895–0.947), respectively (Fig. 2C). The Šidák-adjusted significance level (P = 0.017) was used when the three variables were compared in pairs. Statistically, no significant differences were found between the AUROC for the BMI and the NRS 2002 score (P = 0.0923) or the AUROC for the NRS 2002 score and ALB level (P = 0.3274), but the AUROC for the ALB level was significantly larger than that for the BMI (P = 0.0025).
3. Results 3.1. Patient characteristics A total of 480 COPD patients with respiratory failure were admitted to our hospital from December 2012 to June 2015. Of those patients, 438 were enrolled in our study. A flowchart of the reasons for exclusion is presented in Fig. 1. The characteristics of the included cohort are summarized in Table 1. 3.2. Multivariate analysis As shown in Table 2, the BMI and ALB values independently predicted in-hospital mortality in the COPD patients with respiratory failure (odds ratio (OR) 11.498, 95% CI 4.469–29.581, P = 0.000 and OR 2.503, 95% CI 1.356–4.620, P = 0.003, respectively). Patient BMI and NRS 2002 scores predicted 1-year mortality after discharge (OR 6.494, 95% CI 3.273–12.888, P = 0.000 and OR 0.222, 95% CI 0.120–0.410, P = 0.000, respectively). All three nutritional assessment methods (BMI, NRS 2002, ALB) independently predicted readmission within 30 days after discharge (OR 3.835, 95% CI 1.383–10.632, P = 0.010; OR 0.328, 95% CI 0.108–0.994, P = 0.042; OR 2.854, 95% CI 1.509–5.397, P = 0.001, respectively). BMI = body mass index, NRS = Nutritional Risk Screening, ALB = serum albumin, CI = confidence interval, OR = odds ratio.
3.4. Optimal cutoff values The cutoff values that maximized the sum of the sensitivity and specificity for the mortality and readmission prediction were calculated for the BMI, NRS 2002 and ALB. According to the AUROC results, the optimal cutoff value that maximized the ability to predict in-hospital mortality was 30.5 g/L for ALB; for 1-year mortality prediction, the optimal cutoff value was 3 points for the NRS 2002; for re-admittance within 30 days following discharge, the optimal cutoff value was 30.1 g/L for ALB (Table 3). 4. Discussion COPD patients with respiratory failure often also have different degrees of malnutrition, which may lead to treatment failure and could be associated with a poor prognosis and high mortality. Therefore, it is essential to adopt an accurate, appropriate and reasonable method of identifying COPD patients with respiratory failure who are nutritionally at risk in a timely manner and to provide them with the appropriate nutritional support to prevent a poor clinical prognosis. Though many nutritional assessment methods have been developed, few methods have proven to accurately predict the prognosis of COPD patients with
3.3. AUROC curves In-hospital mortality: Fifty-seven (13.0%) patients died during hospitalization. The AUROC for BMI and ALB for predicting in-hospital mortality were 0.704 (95% CI 0.659–0.746) and 0.908 (95% CI 0.876–0.933), respectively (Fig. 2A), and the AUROC for ALB was significantly larger than that for BMI (P < 0.0001). One-year mortality: Eighty (18.3%) patients died within 1 year after 3
European Journal of Internal Medicine xxx (xxxx) xxx–xxx
R. Chen et al.
Table 2 Multivariate analysis of in-hospital mortality, 12-month mortality and 30-day readmission. Variables
Age Sex Marriage BMI NRS 2002 score ALB
In-hospital mortality
12-month mortality
30-day readmission
OR (95% CI)
P
OR (95% CI)
P
OR (95% CI)
P
1.031 (0.956–1.111) 0.892 (0.107–7.467) 2.777 (0.728–10.592) 11.498 (4.469–29.581) 0.537 (0.211–1.364) 2.503 (1.356–4.620)
0.431 0.916 0.135 0.000 0.191 0.003
1.017 0.597 4.364 6.494 0.222 1.140
0.434 0.364 0.090 0.000 0.000 0.180
0.930 2.149 3.569 3.835 0.328 2.854
0.131 0.498 0.167 0.010 0.042 0.001
respiratory failure. Based on the results of our study, ALB and BMI were both independent predictors of in-hospital mortality in COPD patients with respiratory failure, but ALB was superior to BMI (AUROC 0.908 vs 0.704, P < 0.0001), indicating that the ALB level might be more accurate for the prediction of short-term mortality. This result was consistent with several previous studies [27, 28]. In the short term, COPD with respiratory failure presents with relatively faster protein consumption, which can be more quickly and accurately detected by ALB levels rather than by BMI. In our study, the cutoff value that maximized the ability to predict in-hospital mortality was 30.5 g/L for ALB. Therefore, for a COPD patient with respiratory failure whose ALB level is lower than 30.5 g/L on admission, the risk of in-hospital death is higher, and appropriate nutritional support should be provided to reduce this risk. However, the ALB measurement is an invasive method that causes pain and fear in patients and requires skilled nurses for venipuncture. Additionally, compared to other nutritional assessment methods, ALB detection is more expensive. Therefore, if ALB is selected to predict the prognosis, the quality of the blood test should be verified. In addition, a detailed explanation should be provided to patients to obtain their cooperation, and an effective venipuncture technique should be used to ease their pain. Regarding 1-year mortality, both the BMI and NRS 2002 score were independent predictors of COPD patients with respiratory failure, but the NRS 2002 was superior to the BMI for the 1-year mortality prediction (AUROC 0.835 vs 0.674, P = 0.0004), indicating that the NRS 2002 may have a better ability to predict long-term mortality. Compared to the nutritional status assessment by the single factor of BMI, the assessment of the nutritional status of COPD patients by the NRS 2002, includes information on age and the severity of the disease (e.g., presence or absence of severe pneumonia, whether the patient presented with acute exacerbation, use of a ventilator). Thus, the NRS 2002 can comprehensively reflect the nutritional status of COPD
(0.976–1.059) (0.195–1.821) (1.446–13.172) (3.273–12.888) (0.120–0.410) (0.941–1.382)
(0.846–1.022) (0.235–19.643) (0.588–21.667) (1.383–10.632) (0.108–0.994) (1.509–5.397)
Table 3 Optimal cutoff values for the three nutritional assessment methods (BMI, NRS 2002 and ALB). Outcomes
Nutritional assessment methods BMI
In-hospital mortality
Criterion Sensitivity Specificity
12-month mortality
+LR −LR Criterion Sensitivity Specificity
Readmission
+LR −LR Criterion Sensitivity Specificity +LR −LR
≤19.53 kg/m 77.19 (64.2–87.3) 66.67 (61.7–71.4) 2.32 (1.9–2.8) 0.34 (0.2–0.6) ≤19.53 kg/m2 72.50 (61.4–81.9) 68.44 (63.3–73.2) 2.30 (1.9–2.8) 0.40 (0.3–0.6) ≤19.53 kg/m2 95.24 (76.2–99.9) 63.79 (59.0–68.4) 2.63 (2.2–3.1) 2
0.075 (0.01–0.5)
NRS 2002
ALB
>3 80.70 (68.1–90.0) 75.33 (70.7–79.6) 3.27 (2.6–4.1) 0.26 (0.2–0.4) >3 76.25 (65.4–85.1) 77.93 (73.3–82.1) 3.46 (2.7–4.3) 0.30 (0.2–0.5) >4 71.43 (47.8–88.7) 91.85 (88.8–94.3) 8.76 (5.8–13.3) 0.31 (0.2–0.6)
≤30.5 g/L 84.21 (72.1–92.5) 87.93 (84.2–91.0) 6.97 (5.2–9.4) 0.18 (0.1–0.3) ≤30.5 g/L 65.00 (53.5–75.3) 88.27 (84.5–91.4) 5.54 (4.0–7.7) 0.40 (0.3–0.5) ≤30.1 g/L 95.24 (76.2–99.9) 83.69 (79.8–87.1) 5.84 (4.6–7.4) 0.057 (0.008–0.4)
BMI = body mass index, NRS = Nutritional Risk Screening, ALB = serum albumin, +LR = positive likelihood ratio, −LR = negative likelihood ratio.
patients with respiratory failure and accurately predict their long-term prognosis. The AUROC for the NRS 2002 was 0.835, indicating a good predictive ability. This result was confirmed by our previous study [29], which reported that the NRS 2002 could be effectively used as a
Fig. 2. A: ROC curves for the BMI and ALB for predicting in-hospital mortality; B: ROC curves for the BMI and the NRS 2002 for predicting 1-year mortality after discharge; C: ROC curves for the BMI and the NRS 2002 and ALB for predicting readmission rates within 30 days after discharge. AUROC = area under the ROC curve, BMI = body mass index, ALB = serum albumin, COPD = chronic obstructive pulmonary disease, ROC = receiver operating characteristic. 4
European Journal of Internal Medicine xxx (xxxx) xxx–xxx
R. Chen et al.
Declarations of interest
predictive tool for 12-month mortality in COPD patients with respiratory failure. The ROC curves revealed that the cutoff value that maximized predictive ability for long-term mortality was a score of 3 for the NRS 2002, indicating that a COPD patient with respiratory failure is more likely to die within 1 year following discharge when his or her NRS 2002 score is higher than 3 on admission. The NRS 2002 is the ESPEN-recommended nutritional risk screening tool and is more convenient than most of the other available methods due to its simplicity, feasibility and noninvasiveness. However, because the NRS 2002 includes subjective content, the result of the assessment is somewhat dependent on the assessors. Therefore, we suggest that an experienced assessor is highly necessary, and combining the NRS 2002 with other objective nutritional assessment methods may produce more accurate results when assessing a patient's nutritional status in clinical practice. The multivariate analysis showed that in COPD patients with respiratory failure, the BMI, NRS 2002 score and ALB level were all independent predictors of readmission within 30 days after discharge. According to the ROC curves, we found that ALB had the greatest ability to predict readmission within 30 days following discharge (AUROC 0.924, P < 0.0001). This finding is consistent with a study by Adogwa et al. [30], which reported that the ALB level is an independent predictor of the risk of 30-day hospital readmission. Furthermore, ROC curves showed that a cutoff value of 30.1 g/L for ALB maximized the ability to predict readmission within 30 days following discharge, indicating that COPD patients with respiratory failure have a higher risk of readmission within 30 days when their ALB level is lower than 30.1 g/L on admission. Nutritional support is also suggested for these patients. To our knowledge, this study is the first to compare the ability of three nutritional assessment methods (BMI, NRS 2002, ALB) to predict the clinical prognosis of COPD patients with respiratory failure. In addition to in-hospital mortality, the study also included 1-year mortality and readmission within 30 days after discharge as prognostic outcomes, which made this study more integrated and meaningful. However, some limitations of this study should be mentioned. First, this was a retrospective, observational, single-center study; however, we included a large number of patients. This may influence the application of our results in centers with a different case mix, and the identification of certain factors affecting the prognosis may have been eliminated by the homogeneity of the therapeutic environment and study population. Second, the patients were only measured via three nutritional assessment methods on admission, and the results may not represent the dynamic conditions of patients. Third, all the subjects recruited were conscious when included to allow successful completion of the assessments. Therefore, the application of the results of this study may be limited. Finally, because there is no gold standard nutritional assessment method to predict prognosis in COPD patients with respiratory failure, we did not compare the ROC curves of the three methods with a gold standard method. Further prospective multicenter studies are warranted to confirm and improve our findings.
None. Funding This work was supported by the National Science Foundation of China (Grants No.31671189) and Foundation of Sichuan Provincial Science and Technology Committee agency (Grants No.18ZDYF2039). Acknowledgements None. References [1] Cui J, Wan Q, Wu X, Zeng Y, Jiang L, Ao D, et al. Nutritional risk screening 2002 as a predictor of outcome during general Ward-based noninvasive ventilation in chronic obstructive pulmonary disease with respiratory failure. Med Sci Monit 2015;21:2786–93. [2] Hallin R, Gudmundsson G, Ulrik CS, Nieminen MM, Gislason T, Lindberg E, et al. Nutritional status and long-term mortality in hospitalised patients with chronic obstructive pulmonary disease (COPD). Respir Med 2007;101:1954–60. [3] Shuangjun He. The impact of nutrition status on clinical prognosis in elderly patients with acute exacerbation of chronic obstructive pulmonary disease. J Intern Med Concepts Pract 2014;9:407–10. [4] Zapatero A, Barba R, Ruiz J, Losa J, Plaza S, Canora J, et al. Malnutrition and obesity: influence in mortality and readmissions in chronic obstructive pulmonary disease patients. J Hum Nutr Diet 2013;26:16–22. [5] Cao C, Wang R, Wang J, Bunjhoo H, Xu Y, Xiong W. Body mass index and mortality in chronic obstructive pulmonary disease: a meta-analysis. PLoS One 2012;7:e43892. [6] Guo Y, Zhang T, Wang Z, Yu F, Xu Q, Guo W, et al. Body mass index and mortality in chronic obstructive pulmonary disease: a dose–response meta-analysis. Medicine 2016;95:e4225. [7] Yamauchi Y, Hasegawa W, Yasunaga H, Sunohara M, Jo T, Takami K, et al. Paradoxical association between body mass index and in-hospital mortality in elderly patients with chronic obstructive pulmonary disease in Japan. Int J Chron Obstruct Pulmon Dis 2014;9:1337–46. [8] Marquis K, Debigaré R, Lacasse Y, LeBlanc P, Jobin J, Carrier G, et al. Midthigh muscle cross-sectional area is a better predictor of mortality than body mass index in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2002;166:809–13. [9] Vestbo J, Prescott E, Almdal T, Dahl M, Nordestgaard BG, Andersen T, et al. Body mass, fat-free body mass, and prognosis in patients with chronic obstructive pulmonary disease from a random population sample: findings from the Copenhagen city heart study. Am J Respir Crit Care Med 2006;173:79–83. [10] Velasco C, García E, Rodríguez V, Frias L, Garriga R, Álvarez J, et al. Comparison of four nutritional screening tools to detect nutritional risk in hospitalized patients: a multicentre study. Eur J Clin Nutr 2011;65:269–74. [11] Raslan M, Gonzalez MC, Dias MCG, Nascimento M, Castro M, Marques P, et al. Comparison of nutritional risk screening tools for predicting clinical outcomes in hospitalized patients. Nutrition 2010;26:721–6. [12] Almeida AI, Correia M, Camilo M, Ravasco P. Nutritional risk screening in surgery: valid, feasible, easy!. Clin Nutr 2012;31:206–11. [13] Liu P, Zhang Z-F, Cai J-J, Wang B-S, Yan X. NRS2002 assesses nutritional status of leukemia patients undergoing hematopoietic stem cell transplantation. Chin J Cancer Res 2012;24:299–303. [14] Liu P, Yan X, Wang B, Xu X. Three methods assess nutritional status of leukemia patients before hematopoietic stem cell transplantation. Chin Med J (Engl) 2012;125:440–3. [15] Gama-Axelsson T, Heimbürger O, Stenvinkel P, Bárány P, Lindholm B, Qureshi AR. Serum albumin as predictor of nutritional status in patients with ESRD. Clin J Am Soc Nephrol 2012;7:1446–53. [16] Helliksson F, Wernerman J, Wiklund L, Rosell J, Karlsson M. The combined use of three widely available biochemical markers as predictor of organ failure in critically ill patients. Scand J Clin Lab Invest 2016;76:479–85. [17] Herrmann FR, Safran C, Levkoff SE, Minaker KL. Serum albumin level on admission as a predictor of death, length of stay, and readmission. Arch Intern Med 1992;152:125–30. [18] Gunen H, Hacievliyagil S, Kosar F, Mutlu L, Gulbas G, Pehlivan E, et al. Factors affecting survival of hospitalised patients with COPD. Eur Respir J 2005;26:234–41. [19] Fanali G, di Masi A, Trezza V, Marino M, Fasano M, Ascenzi P. Human serum albumin: from bench to bedside. Mol Aspects Med 2012;33:209–90. [20] Rabe KF, Hurd S, Anzueto A, Barnes PJ, Buist SA, Calverley P, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary. Am J Respir Crit Care Med 2007;176:532–55. [21] Jésus P, Achamrah N, Grigioni S, Charles J, Rimbert A, Folope V, et al. Validity of predictive equations for resting energy expenditure according to the body mass index in a population of 1726 patients followed in a nutrition unit. Clin Nutr 2015;34:529–35.
5. Conclusion For COPD patients with respiratory failure, the best nutritional assessment method varies for the prediction of different prognostic outcomes. The ALB level is superior for predicting in-hospital mortality and 30-day readmission after discharge, while the NRS 2002 score better predicts the long-term prognosis, including 1-year mortality.
Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings described in this paper. 5
European Journal of Internal Medicine xxx (xxxx) xxx–xxx
R. Chen et al.
1998;31:997–1006. [28] Zeng H, Luo H, Tang Y, Zhou Z, Kong X. The purpose of the study is to explore the relationship between serum IL-18, TFN-αand nutritional status, prognosis in patients with COPD in severe stage. Am J Respir Crit Care Med 2013;187:A1521. [29] Chen R, Xing L, You C. Nutritional risk screening 2002 should be used in hospitalized patients with chronic obstructive pulmonary disease with respiratory failure to determine prognosis: a validation on a large Chinese cohort. Eur J Intern Med 2016;36:e16-e7. [30] Adogwa O, Elsamadicy AA, Mehta AI, Cheng J, Bagley CA, Karikari IO. Pre-operative nutritional status is an independent predictor of 30-day hospital readmission after elective spine surgery. Spine 2016;41:1400–4.
[22] Stratton RJ, Green CJ, Elia M. Disease-related malnutrition: an evidence-based approach to treatment. J Hum Nutr Diet 2003;16:375. [23] Chen C. Introduction of body mass index classification suggestion in Chinese adult. Zhonghua Yu Fang Yi Xue Za Zhi 2001;35:349–50. [24] Kondrup J, Rasmussen HH, Hamberg O, Stanga Z, Group AahEW. Nutritional risk screening (NRS 2002): a new method based on an analysis of controlled clinical trials. Clin Nutr 2003;22:321–36. [25] Kondrup J, Allison SP, Elia M, Vellas B, Plauth M. ESPEN guidelines for nutrition screening 2002. Clin Nutr 2003;22:415–21. [26] Zhang A. Medical nutrition. Bejing: People's Medical Publishing House; 1998. [27] Leavey SF, Strawderman RL, Jones CA, Port FK, Held PJ. Am J Kidney Dis
6