Evaluation of factors associated with mortality in adult intensive care unit patients: A single-center retrospective study

Abstract

Predicting mortality in intensive care unit (ICU) patients is crucial for optimizing treatment strategies. This study aimed to evaluate the prognostic value of comorbidities, CRP/albumin ratio and lactate levels in predicting mortality, providing comprehensive insights into these parameters. This retrospective study was conducted between 2016 and 2021 in the ICU at a university hospital. Demographic data, comorbidities, and ICU length of stay were recorded, while APACHE II scores and laboratory parameters were measured within the first 24 hours following ICU admission. The study compared survivors and non-survivors using a 1:1 matching methodology. The association of these variables with mortality was statistically analyzed. The study included 540 patients, with a mean age of 64.1 ± 16.6 years and 62.4% were male. The mortality rate was 50%, and non-survivors were slightly older than survivors (65.3 ± 12.5 vs 63.1 ± 13.5 years, P = .12). Hypertension (43.3% vs 24.1%, P < .001), coronary artery disease (16.7% vs 6.7%, P < .001), and chronic renal failure (13% vs 3%, P < .001) were significantly higher in non-survivors. Increased mortality was significantly associated with higher lactate levels (5.78 ± 2.98 vs 3.11 ± 2.11 mmol/L, P < .001), higher CRP/albumin ratios (36.45 ± 26.42 vs 18.15 ± 14.57, P < .001), and lower albumin levels (3.21 ± 0.56 vs 3.61 ± 0.55 g/dL, P < .001). Additionally, univariate logistic regression analysis showed that each unit increase in WBC (OR: 1.03, 95% CI: 1.01–1.05, P < .001), NEU (OR: 1.08, 95% CI: 1.05–1.11, P < .001), PLT (OR: 1.02, 95% CI: 1.01–1.03, P < .001), and lactate (OR: 1.51, 95% CI: 1.38–1.62, P < .001) was associated with an increased risk of mortality. This study highlights that CRP/albumin ratio, lactate levels and comorbidities are significantly associated with increased mortality in ICU patients. Moreover, unit increases in WBC, NEU, and PLT were also linked to a higher mortality risk. These findings can improve clinical management through the early identification of high-risk patients in ICUs.

1. Introduction

The prognosis of intensive care unit (ICU) patients is significantly influenced by the severity of inflammation caused by infections, trauma and the associated immunoinflammatory response.^[1] Accurately predicting ICU patient outcomes plays a critical role in optimizing treatment strategies.

Recent advancements in medicine have facilitated the development of novel biomarkers that measure inflammatory responses and predict patient outcomes. Studies have evaluated the importance of inflammatory markers in predicting complication risks in clinical conditions such as sepsis.^[2] Key biomarkers such as C-reactive protein (CRP), which is produced in response to interleukin-6 (IL-6), and decreased albumin levels have been found to be essential for diagnosis.^[2] Additionally, the neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, lymphocyte-to-monocyte ratio, and red blood cell distribution width (RDW) are easily accessible indicators that show significant prognostic value in various clinical conditions, including stroke.^[3] These findings emphasize the critical role of inflammatory markers in clinical practice, especially for critically ill patients in ICUs.

C-reactive protein (CRP) is an acute-phase reactant that increases in response to infection or tissue injury, regardless of its underlying cause. Its levels are particularly elevated in severe diseases.^[4] Meanwhile, albumin, a marker associated with inflammation and nutritional status, is linked to outcomes in various inflammatory conditions. The CRP-to-albumin ratio (CAR) has been identified as a novel and potentially more effective indicator of inflammatory status than each marker alone. Its significance has been investigated in various clinical contexts, including diabetes mellitus (DM), coronary artery disease, cancer, and vasculitis. Although there are studies on CRP/albumin and lactate levels in determining the prognosis of specific patient groups, further research is needed to clarify their prognostic significance in critically ill patients admitted to ICUs.^[5]

Elevated lactate levels in ICUs are typically a marker of poor prognosis, often indicating respiratory or circulatory failure. In conditions such as shock, the inadequate delivery of oxygen to tissues and organs leads cells to shift to anaerobic glycolysis, resulting in elevated serum lactate levels and the development of hyperlactatemia.^[6,7] Studies have identified lactate as an independent predictor of mortality, particularly in hospitalized patients with conditions such as sepsis, cardiac arrest, hypoxia, and heart failure. These studies consistently emphasize the association between elevated serum lactate levels and increased mortality risk across various conditions.^[7]

This study aims to evaluate the prognostic significance of the CRP/albumin ratio, comorbidities, and other laboratory parameters in relation to mortality to provide a more comprehensive understanding of prognostic markers and valuable data for the management of critically ill patients.

2. Material and methods

2.1. Study design

This study was conducted in the ICU of the Department of Anesthesiology and Reanimation at Hatay Mustafa Kemal University Faculty of Medicine between 2016 and 2021. The data used in this study was retrospectively screened through the Hospital Information Management System. This study was approved by the Non-Interventional Research Ethics Committee of Hatay Mustafa Kemal University Faculty of Medicine (Approval No: 21, dated June 17, 2021) and conducted in accordance with the ethical principles of the Declaration of Helsinki. Owing to the retrospective and anonymous nature of the data, the requirement for informed consent was waived.

2.2. Data collection

After obtaining approval from the local ethics committee, all patients admitted to the anesthesia and reanimation ICU were retrospectively screened using the hospital’s electronic database. For patients with multiple ICU admissions, only the first admission was included in the analysis. Patients under 18 years old and those with missing data (e.g., incomplete demographic information, missing laboratory parameters, or APACHE II scores) were excluded from the study.

Demographic data (age and gender), comorbidities (DM, hypertension, coronary artery disease, chronic renal failure, cerebrovascular events, and chronic obstructive pulmonary disease (COPD)), and ICU length of stay were recorded, while Acute Physiology and Chronic Health Evaluation II (APACHE II) scores and laboratory parameters (platelet (PLT), white blood cell (WBC), neutrophil (NEU), lactate (LAC), C-reactive protein (CRP), CRP/albumin ratio, and albumin levels) were measured within the first 24 hours following ICU admission.

After data collection, patients were categorized as survivors and non-survivors based on their ICU discharge status. To minimize selection bias and ensure comparability between groups, a 1:1 matching between survivors and non-survivors was performed using Propensity Score Matching based on age and gender. All data were analyzed.

2.3. Statistical analysis

Patients who became non-survivors were compared with survivors, and the odds ratio for the exposure rate of CRP/ALB and other factors associated with non-survivor status was considered clinically significant at a minimum of 1.5. The study was powered to detect a type I error at a maximum of 5% and a minimum power of 80%, requiring a total of 540 patients.

A 1:1 matching between survivors and non-survivors was performed using Propensity Score Matching based on age and gender to reduce the risk of selection bias and control for potential covariate. Continuous variables are presented as mean and standard deviation, while median, minimum, and maximum values are used to describe characteristics. Categorical variables are expressed as numbers and percentages. The normality of continuous variables according to survivor/non-survivor status was assessed using the Kolmogorov-Smirnov test. For continuous variables that follow a normal distribution, the Student t test was used to compare the mean differences according to survivor/non-survivor status, while the Mann–Whitney U test was used for non-normally distributed data. The Chi-square test was employed to assess the relationship between survivor/non-survivor status and other variables. Factors related to survival were evaluated with univariate binary logistic regression analysis with odds ratios and 95% confidence intervals. A P-value of <0.05 was considered statistically significant. The data were analyzed using the IBM SPSS 25 statistical software package.

3. Results

Among the 540 patients, 62.4% (n = 337) were male and 37.6% (n = 203) were female, with ages ranging from 18 to 111 years. The mean age was 64.1 ± 16.6 years, and the median age was 65 years. The length of ICU stay varied from 1 to 99 days, with a mean of 15.4 ± 18.68 days and a median of 5 days. The APACHE II score ranged from a minimum of 3 to a maximum of 88, with a mean APACHE II score of 11.2 ± 13.6 and a median score of 7.

The patients (17.6%) had DM, 33.7% had hypertension (HT), 11.7% had coronary artery disease, 8% had chronic renal failure, 2.4% had a cerebrovascular event (CVE), and 7% had COPD (Table 1). The mean age of non-survivors was 65.3 ± 12.5 years, while the mean age of survivors was 63.1 ± 13.5 years. No significant difference was observed between the 2 groups regarding age (P > .05). The gender distribution was similar between the 2 groups; 63% of non-survivors were male and 37% were female, compared to 61.9% male and 38.1% female in the survivor group (P = .79).

Table 1.

Distribution of socio-demographic characteristics and clinical information (n = 540).

Characteristics	Characteristics	Mean ± SD*	Median (Min–Max)
Age		64.1 ± 16.6	65 (18–111)
		n	(%)
Gender	Male	337	62.4
	Female	203	37.6
Survivor status	Non-Survivor	270	50
	Survivor	270	50
DM	Absent	445	82.4
	Present	95	17.6
HT	Absent	358	66.3
	Present	182	33.7
Coronary artery disease	Absent	477	88.3
	Present	63	11.7
Chronic renal failure	Absent	497	92
	Present	43	8
Cerebrovascular event	Absent	527	97.6
	Present	13	2.4
COPD	Absent	502	93
	Present	38	7
Length of stay (days)	–	15.4 ± 18.6	5 (1–99)
APACHE II	–	11.2 ± 13.6	7 (3–88)

COPD = chronic obstructive pulmonary disease, DM = diabetes mellitus, HT = hypertension, SD = standard deviation.

Hypertension (HT) was found in 43.3% of the non-survivors, compared to 24.1% of the survivors, indicating a significant association between HT and mortality (P = <.001). Similarly, coronary artery disease was observed in 16.7% of the non-survivors and 6.7% of the survivors, demonstrating a significant relationship between these conditions and mortality (P = <.001). Furthermore, chronic renal failure was present in 13% of the non-survivors compared to 3% of the survivors, further highlighting its significant association with mortality (P = <.001). However, no significant associations were identified between mortality and the presence of DM, cerebrovascular events (CVE), or COPD (P > .05).

In our study, the mean APACHE II score was 13.6 ± 12.4 in non-survivors and 8.7 ± 6.64 in survivors, with a significant difference observed between the 2 groups (P = <.001). The mean WBC count (WBC) was 14.71 ± 10.98 in non-survivors, compared to 11.86 ± 9.29 in survivors, with a significant difference in WBC values (P = <0.001). The neutrophil percentage (NEU) was 91.18 ± 7.05 in non-survivors and 85.72 ± 11.09 in survivors, with a statistically significant difference between the 2 groups (P = <.001).

Platelet count (PLT) was 375.48 ± 221.44 in non-survivors, compared to 250.53 ± 218.78 in survivors, showing a significant difference in PLT levels (P = <.001). Additionally, lactate (Lac) levels were 5.78 ± 2.98 in non-survivors and 3.11 ± 2.11 in survivors, with a statistically significant difference (P = <.001).

C-reactive protein (CRP) levels were 110.82 ± 71.97 in non-survivors, while in survivors, the levels were lower at 61.64 ± 46.33, with a significant difference in CRP values (P = <.001). Albumin (ALB) levels were 3.21 ± 0.56 in non-survivors and 3.61 ± 0.55 in survivors, indicating a significant difference (P = <.001). The CRP/ALB ratio was 36.45 ± 26.42 in non-survivors and 18.15 ± 14.57 in survivors, with a statistically significant difference between the 2 groups (P = <.001) (Table 2).

Table 2.

Assessment of differences and relationships according to survivor and non-survivor status.

Characteristics	Characteristics	Survivors mean ± SD (n = 270)	Non-survivors mean ± SD (n = 270)	P-value
Age*		63.1 ± 13.5	65.3 ± 12.5	.12
		n	(%)
Gender**	Male	167 (61.9)	170 (63)	.79
	Female	103 (38.1)	100 (37)
DM**	Absent	226 (83.7)	219 (81.1)	.43
	Present	44 (16.3)	51 (18.9)
HT**	Absent	205 (75.9)	153 (56.7)	<.001
	Present	65 (24.1)	117 (43.3)
Coronary artery disease**	Absent	252 (93.3)	225 (83.3)	<.001
	Present	18 (6.7)	45 (16.7)
Chronic renal failure**	Absent	262 (97)	235 (87)	<.001
	Present	8 (3)	35 (13)
CVE**	Absent	265 (98.1)	626 (97)	.41
	Present	5 (1.9)	8 (3)
COPD**	Absent	251 (93)	251 (93)	.99
	Present	19 (7)	19 (7)
APACHE II***	–	8.7 ± 6.64	13.6 ± 12.4	<.001
WBC***	–	11.86 ± 9.29	14.71 ± 10.98	<.001
NEU***	–	85.72 ± 11.09	91.18 ± 7.05	<.001
PLT***	–	250.53 ± 218.78	375.48 ± 221.44	<.001
LAC***	–	3.11 ± 2.11	5.78 ± 2.98	<.001
CRP***	–	61.64 ± 46.33	110.82 ± 71.97	<.001
ALB*	–	3.61 ± 0.55	3.21 ± 0.56	<.001
CRP/ALB***	–	18.15 ± 14.57	36.45 ± 26.42	<.001

ALB = albumin, APACHE = acute physiology and chronic health evaluation, COPD = chronic obstructive pulmonary disease, CRP = C-reactive protein, CVE = cerebrovascular event, DM = diabetes mellitus, HT = hypertension, LAC = lactate, NEU = neutrophil, PLT = platelet, WBC = white blood cell.

^*Student’s t-test.

^**Chi-square test.

^***Mann–Whitney U.

Among non-survivors, hypertension, coronary artery disease and chronic renal failure was 2.41 times (95% CI: 1.67–3.49; P = <.001), 2.81 times (95% CI: 1.57–4.98; P = <.001), and 4.88 times (95% CI: 2.22–10.73; P = <.001) increased, respectively, when compared to survivors.

The analyses also revealed that unit increases in WBC, NEU, PLT, Lac, CRP, and the CRP/ALB ratio significantly affected the risk of mortality. Specifically, each 1-unit increase in WBC raised the mortality risk by 1.03 times (95% CI: 1.01–1.05; P = <.001), each 1-unit increase in NEU raised it by 1.08 times (95% CI: 1.05–1.11; P = <.001), each 1-unit increase in PLT raised it by 1.02 times (95% CI: 1.01–1.03; P = <.001), and each 1-unit increase in Lac raised it by 1.51 times (95% CI: 1.38–1.62; P = <.001). Conversely, each 1-unit increase in ALB significantly reduced the risk of mortality by 0.30 times (95% CI: 0.21–0.41; P = <.001). Each 1-unit increase in the CRP/ALB ratio significantly increased the risk of mortality by 1.06 times (95% CI: 1.04–1.07; P = <.001) (Table 3).

Table 3.

Assessment of factors associated with non-survivor status.

Characteristics	Odds ratio (95% CI: lower–upper)	P-value
Hypertension	2.41 (1.67–3.49)	<.001
Coronary artery disease	2.81 (1.57–4.98)	<.001
Chronic renal failure	4.88 (2.22–10.73)	<.001
WBC	1.03 (1.01–1.05)	<.001
NEU	1.08 (1.05–1.11)	<.001
PLT	1.02 (1.01–1.03)	<.001
Lac	1.51 (1.38–1.62)	<.001
CRP	1.02 (1.01–1.25)	<.001
ALB	0.30 (0.21–0.41)	<.001
CRP/ALB	1.06 (1.04–1.07)	<.001

ALB = albumin, CI = confidence interval, CRP = C-reactive protein, LAC = lactate, NEU = neutrophil, PLT = platelet, WBC = white blood cell.

4. Discussion

The prognosis of critically ill patients in ICUs is influenced by multiple factors including comorbidities and biochemical markers.^[8,9] In our analysis, comorbid conditions such as hypertension, coronary artery disease and chronic renal failure were associated with significantly increased odds of mortality, with respective risk increases of 2.41, 2.81, and 4.88 times. These findings are consistent with earlier studies showing that chronic conditions, including hypertension and coronary artery disease, aggravate ICU mortality by increasing cardiovascular instability and inflammation.^[10] Additionally, chronic renal failure has frequently been recognized in the literature as a significant predictor of adverse outcomes in ICU settings, particularly in cases with elevated serum creatinine levels. Although creatinine was not included in our statistical analysis due to data limitations, previous reports have demonstrated that even minimal increases in creatinine are associated with higher mortality and prolonged ICU stays.^[11] The association between chronic renal failure and mortality was clearly observed in our study (odds ratio: 4.88), whereas Barrantes et al reported a high odds ratio (7.9) for mortality in patients with acute renal failure, further supporting the prognostic importance of renal dysfunction.^[12]

In addition to comorbidities, several laboratory markers have been identified as significant predictors of prognosis in ICU patients.^[13] Our findings are consistent with the existing literature, demonstrating that increases in specific biochemical markers are significantly associated with higher mortality risk in critically ill patients. Unit increases in WBC, NEU, PLT, lactate (Lac), CRP and the CRP/ALB ratio were all found to correspond to an elevated mortality risk. WBC, NEU, and PLT increases resulted in 1.03, 1.08, and 1.02 times higher mortality risk, respectively. Similarly, increases in Lac, CRP and the CRP/ALB ratio were associated with 1.51, 1.02, and 1.06 times higher mortality risk, respectively. Conversely, an increase in albumin (ALB) was associated with a 0.30 times reduction in mortality risk, underscoring its protective role. These biomarkers are important for early intervention and treatment in ICU settings.

The extensive literature on serum lactate levels supports our study’s findings by emphasizing its significant prognostic value in critically ill patients. In our study, elevated lactate levels were associated with a 1.51 times higher risk of mortality, consistent with the findings of Ma et al who identified serum lactate as a reliable predictor of poor outcomes.^[6] Haas et al demonstrated that hyperlactatemia often results from multiple causes and is linked to both reversible and irreversible conditions, such as sepsis and cardiac failure, reflecting the complex relationship between comorbidities and ICU prognosis.^[7] Elevated lactate levels indicate an imbalance between oxygen delivery and consumption, leading to poor clinical outcomes.^[6,14] For example, a German population study demonstrated that high lactate levels at emergency department visits were associated with increased mortality, independent of infection status, while Aduen et al showed that a lactate concentration > 4 mmol/L reliably predicted mortality in non-hypotensive patients.^[6,15] These findings suggest that routine lactate monitoring may improve patient management and outcomes in critically ill patients.

The C-reactive protein/albumin ratio (CAR) is a widely accessible and cost-effective inflammatory marker that plays a key role in both diagnosis and prognosis. It is particularly valuable for guiding clinical decisions in critically ill patients at high risk of mortality.^[16] Several studies have shown that CAR, as a marker of both inflammation and nutritional status, demonstrates strong predictive accuracy for mortality, especially in sepsis and septic shock.^[5,17] In this study, CAR was analyzed for its prognostic value in predicting mortality among critically ill patients. Our results are consistent with the findings of Xu et al, who reported that CAR offers superior prognostic accuracy compared to CRP alone, particularly in cases of sepsis and septic shock.^[18] Similarly, Park et al found that the sensitivity and specificity of CAR for predicting ICU mortality were moderate, suggesting that its predictive performance may improve when combined with other biomarkers.^[19] Qu et al also highlighted that elevated CRP levels at ICU admission independently predict mortality, supporting the value of inflammatory markers in patient risk assessment.^[20] These findings suggest that CAR, particularly when combined with other biomarkers, may serve as a more reliable tool for supporting clinical decision-making.

Studies have found that CRP, albumin, and lactate levels are associated with the degree of inflammation and prognosis, even in critically ill pediatric patients.^[20–22] Although tests for these markers are available, they have not been fully integrated into current scoring systems, indicating potential opportunities for improving prognostic evaluations.^[22] Our findings reveal the multifactorial nature of prognosis in ICU patients, emphasizing the role of comorbidities and laboratory markers in improving outcomes.

This study has several limitations. First, it was conducted at a single center with a relatively small cohort, which may limit the generalizability of the findings. Due to its retrospective design, the study relied on available medical records, and some important clinical variables were missing because they were not routinely recorded. Variations in patient characteristics and underlying conditions may also have affected the consistency of the results. In addition, the exclusive focus on selected biomarkers may have caused other potentially important prognostic indicators to be missed. Lastly, the absence of body mass index data may have influenced the evaluation of laboratory findings, particularly when examining the relationship between metabolic status and inflammation.

5. Conclusion

In conclusion, this study highlights the multifactorial nature of ICU mortality, demonstrating the combined impact of comorbidities and laboratory markers. Our findings confirm previous research and underscore the need for comprehensive assessments that integrate clinical, laboratory, and hemodynamic parameters to improve the prediction of patient outcomes and mortality. Future multicenter studies with larger cohorts and detailed data collection including body mass index and additional biomarkers are essential to validate these findings and clarify their significance in ICUs.

Acknowledgments

We are grateful to Prof Dr Cemil Kurekci (Hatay Mustafa Kemal University) for his detailed reading and comments on our manuscript.

Author contributions

Conceptualization: Senem Urfali, Onur Koyuncu.

Data curation: Senem Urfali, Bircan Kara, Elif Ertas.

Formal analysis: Senem Urfali, Bircan Kara, Elif Ertas.

Investigation: Senem Urfali, Mehmet Esat Duymus, Selim Turhanoglu.

Methodology: Senem Urfali, Selim Turhanoglu.

Resources: Senem Urfali, Mehmet Esat Duymus.

Writing – original draft: Senem Urfali, Onur Koyuncu

Writing – review & editing: Senem Urfali, Onur Koyuncu, Selim Turhanoglu.