The site of the Media Sphera Publishers contains materials intended solely for healthcare professionals.
By closing this message, you confirm that you are a certified medical professional or a student of a medical educational institution.

A.O. Sivkov

Medical and Sanitary Unit «Neftyanik»

I.N. Leyderman

Almazov National Medical Research Center

O.G. Sivkov

Tyumen Cardiology Research Center, Branch of the Tomsk National Research Medical Center of the Russian Academy of Sciences

Malnutrition markers as negative outcome predictors in critically ill patients undergoing prolonged mechanical ventilation

Authors:

A.O. Sivkov, I.N. Leyderman, O.G. Sivkov

More about the authors

Views: 1499

Downloaded: 114


To cite this article:

Sivkov AO, Leyderman IN, Sivkov OG. Malnutrition markers as negative outcome predictors in critically ill patients undergoing prolonged mechanical ventilation. Russian Journal of Anesthesiology and Reanimatology. 2022;(6):52‑57. (In Russ., In Engl.)
https://doi.org/10.17116/anaesthesiology202206152

Recommended articles:
Features of nutritional support in stress hype­rglycemia and diabetes mellitus in the inte­nsive care unit. Russian Journal of Anesthesiology and Reanimatology. 2023;(3):59-66
Predictors of the need for correction of therapy during neonatal transport. Russian Journal of Anesthesiology and Reanimatology. 2023;(4):12-18
Seda­tion of patients in inte­nsive care units. Guidelines. Russian Journal of Anesthesiology and Reanimatology. 2023;(5):6-24
Three main components of initial ente­ral support for neurological patients in critical illness. Russian Journal of Anesthesiology and Reanimatology. 2023;(6):84-90
Role of nutritive support in the reha­bilitation of cancer patients. P.A. Herzen Journal of Onco­logy. 2023;(5):79-88

Introduction

Critically ill patients, especially those undergoing prolonged mechanical ventilation, are at high risk of malnutrition as they are deprived of proper nutrition. According to some authors, about 30% of patients in developed countries admit to intensive care units with signs of malnutrition, and other ones have a high risk of this complication at further stages of treatment [1]. Incidence of malnutrition in ICU patients is 38-78% [2]. According to multiple-center studies, early nutritional support within 48 hours after admission to ICU reduces hospital-stay, duration of mechanical ventilation and mortality [3, 4]. Modern clinical guidelines suggest nutritional support of ventilated patients within 24-48 hours to provide them with necessary macro- and micronutrients in a timely manner [5–7]. Early detection of disorders in protein-energy metabolism and nutritional status is an important factor in the choice of nutritional support strategy [8, 9]. To achieve this goal, various markers of malnutrition are used (prealbumin, albumin, total protein, transferrin, lymphocyte count, Nutritional Risk Index (NRI) [10], Maastricht Index (MI) [11], The Prognostic Nutritional Index (PNI) [12], Nutritional Risk Screening 2002 (NRS-2002) [13], Modified Nutrition Risk in the Critically Ill (mNUTRIC) scales [14]). Available methods for screening and diagnosis of malnutrition in ICU patients have different validity regarding the risk of adverse outcomes in patients undergoing prolonged ventilation. Searching for the most informative predictors (or their combinations) of adverse outcomes in critically ill patients is important problem.

The purpose of the study was to assess predictive value of malnutrition markers for negative clinical outcomes in critically ill patients undergoing prolonged mechanical ventilation.

Material and methods

A prospective single-center study was carried out between 2012 and 2017 at the JSC Neftyanik Medical Unit (Tyumen). Inclusion criteria: age 18-80 years, mechanical ventilation for more than 48 hours (all criteria are required). Exclusion criteria: acute cerebral insufficiency, terminal state, chronic renal failure, pregnancy. On the first day after admission, we analyzed clinical severity and nutritional status using clinical and laboratory data (serum prealbumin, albumin, lymphocyte count), specialized indices (NRI, MI PNI) and integral scales (APACHE II, SOFA; NRS-2002, mNUTRIC). The study enrolled 62 patients (31 women) including 39 (62.9%) surgical and 23 (37.1%) therapeutic ones. Diseases are summarized in Table 1.

Table 1. Clinical forms in surgical and therapeutic patients

Surgical profile

n

Therapeutic profile

n

Secondary peritonitis

29

Community-acquired pneumonia

8

Urinary tract infections (urosepsis)

1

Acute decompensation of chronic heart failure

2

Purulent-inflammatory soft tissue lesions

4

Decompensated liver cirrhosis

6

Gastrointestinal bleeding

3

Epilepsy (epileptic status)

1

Arterial thrombosis

1

Acute leukemia

1

Acute mediastinitis

1

Acute kidney injury

1

Pulmonary embolism

1

Total:

39

23

At the next stage, we distinguished 2 groups (deceased patients (n=39) and survivors (n=23)) (Table 2). Statistical analysis was performed using the SPSS software. Distribution normality was tested using the Shapiro-Wilk test. Data are presented as mean with 95% confidence interval or median with quartiles [Q25; Q75]. Univariate and multivariate binary logistic regression was used to identify predictors of poor outcomes. Appropriate predictors were included in the model by sequential input. Quality of predictive model was assessed using ROC analysis. The null hypothesis was rejected at p-value<0.05.

Table 2. Malnutrition markers in critically ill patients with unfavorable and favorable outcomes undergoing prolonged mechanical ventilation

Variable

Dead (n=39)

Survivors (n=23)

p

Age, years

62.5 (58.25—68.46)

49.4 (55.78—61.7)

0.011*

Male/female, n

21/18

10/13

APACHE-II scorea

15.5 (15.22—18.46)

9.17 (8.67—10.72)

<0.001*

SOFA scoreb

4 [3; 7]

2 [1.5; 4.5]

0.005**

NRS-2002 scorec

5 [4; 5.5]

3 [2.5; 51]

0.001**

mNUTRIC scored

3 [2; 5]

1 [0; 2]

<0.001**

Serum prealbumin, mg/dL

17 [13; 39.3]

32.6 [24.7; 44.3]

0.007**

Serum albumin, g/L

28 (25.82—30.75)

32.6 (31.87—34.74)

0.023*

NRIe

97.6 (93.84—103.01)

107.5 (105.3—112.68)

0.68*

MIf

6.3 [3.5; 7.7]

0.91 [–1.1; 5.33]

0.007**

PNIg

33.2 (31.33—36.6)

39.0 (38.14—41.47)

0.033*

Lymphocyte count, thousand/µl

0.85 [0.64; 1.17]

1.12 [0.86; 1.57]

0.15**

Time of ventilation, days

4 [2.5; 5]

3 [2; 5.5]

0.28**

Note. a — Acute Physiology And Chronic Health Evaluation; b — Sequential Organ Failure Assessment; c — Nutritional Risk Screening 2002; d — Modified Nutrition Risk in the Critically Ill; e — Nutritional Risk Index; f — Maastricht Index; g — The Prognostic Nutritional Index; * — Student's T-test; ** — Mann-Whitney test.

Results

An analysis of the performed studies showed that such indicators as age, scores on the APACHE II, SOFA, NRS-2002, mNUTRIC, MI scales were statistically significantly higher in the group of deceased patients (see Table 2). Duration of hospital stay, as well as serum albumin and prealbumin concentrations, were statistically higher in the survivor group. The PNI index turned out to be statistically significantly lower in patients of the group with an unfavorable clinical outcome.

At the next stage, we assessed prognostic significance of predictors of adverse outcomes in patients on prolonged ventilation using univariate logistic regression (Table 3).

Table 3. Univariate logistic regression analysis of predictors of adverse outcomes in patients undergoing prolonged mechanical ventilation

Variable

p

Constant

B

R2 Nigelkirk coefficient of determination

Exp (B)

95% CI

Se, %

Sp, %

Accurate response rate, %

Lower limit

Upper limit

Albumin

0.065

2.374

–0.061

0.084

0.941

0.882

1.004

21.7

97.4

69.4

Lymphocyte count

0.251

1.01

–0.042

0.029

0.657

0.321

1.347

8.7

92.3

61.3

Prealbumin

0.038

1.554

–0.034

0.097

0.96

0.937

0.998

17.4

79.5

56.5

NRIb

0.066

3.15

–0.26

0.077

0.97

0.94

1.002

21.7

87.2

62.9

PNIc

0.042

2.66

–0.059

0.1

0.942

0.89

0.998

26.1

94.9

69.4

MId

0.013

0.063

0.149

0.142

1.161

1.032

1.305

34.8

79.5

62.9

mNUTRICe

<0.001

–1.686

1.037

0.458

2.82

1.614

4.927

61

84.6

75.8

NRS-2002f

0.002

–2.365

0.706

0.241

2.027

1.29

3.185

61

87.2

77.4

APACHE IIg

0.001

–2.435

0.246

0.356

1.279

1.11

1.473

56.5

87.2

75.8

SOFAh

0.015

–0.681

0.296

0.15

1.344

1.060

1.705

52.2

84.6

72.6

Note. a — Nutritional Risk Index; b — The Prognostic Nutritional Index; c — Maastricht Index; d — Modified Nutrition Risk in the Critically Ill; e — Nutritional Risk Screening 2002; f — Acute Physiology And Chronic Health Evaluation; g — Sequential Organ Failure Assessment. Se — sensitivity; Sp — specificity.

The mNUTRIC, NRS-2002, APACHE II and SOFA scales demonstrated the best ability to predict unfavorable outcomes with correct answer rate over 70%. To identify predictive abilities of independent predictors, we applied ROC analysis (Table 4).

Table 4. ROC analysis of integrated scales, scores and laboratory markers of malnutrition as independent predictors of poor clinical outcomes

Variable

p

AUC

95% CI

Cutt-off value

Se, %

Sp, %

Lower limit

Upper limit

PNIa

0.047

65.2

0.511

0.793

34.9

61.5

60.9

MIb

0.007

70.5

0.575

0.834

4.21

69.2

69.6

Prealbumin

0.007

70.6

0.578

0.834

25.8

66.7

69.6

Scores

mNUTRICc

<0.001

84.8

0.75

0.946

3.5

46.2

95.7

NRS-2002d

0.002

74

0.602

0.878

3.5

87.2

69.6

APACHE IIe

<0.001

81.7

0.703

0.931

11.5

79.5

78.3

SOFAf

0.005

71.5

0.575

0.854

3.5

64.1

65.2

Note. a — The Prognostic Nutritional Index; b — Maastricht Index; c — Modified Nutrition Risk in the Critically Ill; d — Nutritional Risk Screening 2002; e — Acute Physiology and Chronic Health Evaluation; f — Sequential Organ Failure Assessment.

Multivariate analysis was carried out to improve the accuracy of predicting unfavorable outcomes by several independent predictors. Of all combinations, serum prealbumin together with NRS-2002 and mNUTRIC scores ensured the best rate of accurate answers (79%) (Table 5).

Table 5. The best model for predicting adverse outcomes in patients undergoing prolonged mechanical ventilation (multivariate analysis)

Variable

p

95% CI

Se, %

Sp, %

Correct answer rate, %

Lower limit

Upper limit

mNUTRIC scorea

<0,001

1,5

5,15

69,6

84,6

79

NRS-2002 scoreb

1,01

3,33

Note. a — Modified Nutrition Risk in the Critically Ill; b — Nutritional Risk Screening 2002; B — standard error of the coefficient; Se — sensitivity; Sp — specificity.

According to regression coefficients, prealbumin decrease by 1 mg / dl increases the risk of mortality by 3.3%. NRS-2002 increment by 1 score increases the risk of unfavorable outcome by 1.83 times, mNUTRIC score — by 2.77 times. The threshold value of logistic function was determined by ROC analysis. As a result, the area under curve was 0.88±0.43 (95% CI 0.797–0.967, p<0.001). The cutt-off value was 63.3. Equal or higher value corresponds to high risk of mortality. Sensitivity of the model for threshold value was 82.1%, specificity 82.6%. Thus, combined use of NRS-002 and mNUTRIC scales improves prediction of mortality compared to any score alone.

Discussion

Serum prealbumin (transthyretin) measured on the first day of ICU-stay is the most informative biochemical marker predicting unfavorable outcomes in critically ill patients undergoing prolonged ventilation. These findings do not contradict other studies showing low prealbumin as effective criterion for predicting the course [15] and outcomes of disease [16, 17]. A recent meta-analysis of 19 studies and 4616 patients with COVID-19 showed that low serum transthyretin is associated with severe disease and mortality [18]. The Prognostic Nutritional Index based on serum albumin and lymphocyte count is a simple and objective indicator of postoperative outcomes in cancer patients undergoing surgery [19]. Cheng Y.L. revealed PNI as an independent predictor of overall and cardiac mortality in patients with heart failure [20]. In addition, PNI had good predictive capacity regarding in-hospital mortality in infective endocarditis [21] and COVID-19 [22]. In our study, PNI also predicted mortality (OR 0.942; 95% CI 0.89–0.998). The Maastricht Index was developed in 1985. This index includes serum concentration of albumin and prealbumin, lymphocyte count and ideal body weight. The Maastricht Index verifies malnutrition with sensitivity of 93% and specificity of 94% [11]. Patients with the Maastricht index > 0 (malnutrition) had high risk of postoperative complications. In 2006, Kuzu M.A. et al. [23] assessed malnutrition in patients undergoing major surgery. MI- and PNI-based incidence of malnutrition was comparable (67.4 and 63.5%, respectively) with risk of mortality 2.81 (95% CI 0.79–9.95) and 2.3 (95% CI 1.43–3.71), respectively. In ROC analysis, area under curve was 0.671 for MI and 0.66 for PNI [23]. We found no studies that used MI and NRI as predictors of outcomes in ICU patients on long-term mechanical ventilation. In our study, these indicators demonstrated independent relationship with prediction of mortality in critically ill patients on long-term mechanical ventilation and confirmed moderate and good quality of the model. To date, the NRS-2002 and mNUTRI scales are the most informative in assessing malnutrition and predicting the outcome of disease. G. Liu et al. [24] assessed the risk of malnutrition and mortality in patients with COVID-19 using the NRI and NRS-2002 scales. Patients with high risk of malnutrition were characterized by longer hospital-stay and higher mortality [24]. In a retrospective observational study, the mNUTRIC score (AUC 0.875 (95% CI 0.81-0.95), sensitivity 93.2%, specificity 63%) was better in predicting mortality than the NRS-2002 score (AUC 0.73 (95% CI 0.65-0.82), sensitivity 91%, specificity 50%) in patients with severe pneumonia [25]. In 2017, Kalaiselvan M.S. et al. [26] analyzed the risk of malnutrition in a prospective observational study of 678 patients on prolonged mechanical ventilation. Patients with mNUTRIC score ≥5 spent much time in ICU (9.0±4.2 vs. 7.8±5.8 days, p<0.01) and had higher risk of mortality (41.4% vs. 26.1%, p<0.01). The mNUTRIC score ≥5 predicted mortality with area under curve of 0.582 (95% CI 0.535–0.628). In a recent multiple-center retrospective study, validity of the mNUTRIC scale was assessed in 737 patients on long-term mechanical ventilation. Patients were divided into groups with APACHE-II score ≥25, mNUTRIC score ≥5, mechanical ventilation ≥72 h, unsuccessful weaning from ventilator and dead, as well as patients with APACHE-II score <25, mNUTRIC score <5, mechanical ventilation <72 h, successful weaning from ventilator, survivors. The mNUTRIC score > 4.5 predicted mortality with area under curve of 0.67 (95% CI 0.57–0.7), sensitivity of 68% and specificity of 54.8%. Score > 4.5 was also predictive regarding unsuccessful weaning from ventilator with area under curve of 0.64 (95% CI 0.56–0.71), sensitivity of 68% and specificity of 55.3%. Score ≥5 predicted mechanical ventilation for more than 72 h with area under curve of 0.56 (95% CI 0.51–0.6), sensitivity of 62.7% and specificity of 31% [27]. In our study, mNUTRIC and NRS-2002 scales gave the highest correct answer rate, and their combination significantly improved prediction of unfavorable outcomes.

Conclusion

1. Malnutrition markers can be used as predictive criteria of poor outcomes in critically ill patients on long-term mechanical ventilation.

2. Serum prealbumin, PNI and MI scores, as well as NRS-2002, mNUTRIC, SOFA and APACHE II scores are independent predictors of unfavorable outcomes.

3. The mNUTRIC and NRS-2002 scales have the best predictive value. Their combination improves prediction of mortality in critically ill patients undergoing long-term mechanical ventilation.

Author contribution:

Concept and design of the study — Sivkov A.O., Leiderman I.N.

Collection and analysis of data — Sivkov A.O.

Statistical analysis — Sivkov A.O.

Writing the text — Sivkov A.O., Leiderman I.N., Sivkov O.G.

Editing — Leiderman I.N., Sivkov O.G.

The authors declare no conflicts of interest.

Email Confirmation

An email was sent to test@gmail.com with a confirmation link. Follow the link from the letter to complete the registration on the site.

Email Confirmation



We use cооkies to improve the performance of the site. By staying on our site, you agree to the terms of use of cооkies. To view our Privacy and Cookie Policy, please. click here.