Association between glycemic variability in patients with traumatic lung injury and risk of in-hospital death: a retrospective analysis based on the Medical Information Mart for Intensive Care IV (MIMIC-IV) database
Highlight box
Key findings
• The results suggested that higher glycaemic variability (GV) was related to a heightened risk of in-hospital mortality (IHM) in traumatic lung injury (TLI) patients. Nonetheless, further large-scale prospective studies are needed to confirm the causal relationship.
What is known and what is new?
• Studies have demonstrated that notable fluctuations of blood glucose may have an even greater impact on the individual’s health condition. For instance, a recent study suggests that GV is related to a higher risk of IHM in intensive care unit (ICU) patients.
• Thus, high GV is considered an indicator of poor prognosis. However, the relationship between GV and prognosis in ICU patients with TLI remains unclear. This retrospective study seeks to investigate the association between GV and the prognosis of TLI patients.
What is the implication, and what should change now?
• Our findings may assist clinicians in closely monitoring TLI patients and providing timely interventions to reduce death rates and improve clinical outcomes. As a retrospective cohort study, despite efforts to minimize reverse causality in the study design, the findings should be validated through prospective longitudinal cohort studies.
Introduction
Trauma stands as a major cause of mortality and disability in individuals under 45 worldwide, with its death toll surpassing that of cancer in young people (1). Chest trauma is a major injury among severely injured trauma patients, affecting more than 50% of polytrauma patients with certain types of chest injury (2,3). Traumatic lung injury (TLI) refers to a pathological condition caused by direct or indirect external forces acting on the chest or body, leading to damage to the structure and function of lung tissue. TLI often results from motor vehicle accidents, falls from heights, crush injuries, or penetrating trauma. TLI can affect multiple organ systems, and even become life-threatening, with post-trauma death rates for patients with severe chest injuries reaching up to 30% (4). TLI patients always require prolonged intensive care unit (ICU) stays, multiple surgeries, or rehabilitation, resulting in high medical costs (5). Additionally, some patients may lose their ability to work, which not only reduces their quality of life but also affects their mental well-being. As a result, TLI patients have notably lower psychological scores than healthy individuals (6). Trauma patients are more likely to experience stress-induced hyperglycemia (HG). The trauma-related stress response can activate the sympathetic nervous system, increase catecholamine release, and cause insulin resistance, ultimately leading to HG (7). Elevated levels of blood glucose (BG) increase reactive oxygen species, damage alveolar cells, promote the release of inflammatory cytokines (such as tumor necrosis factor-alpha and interleukin-6), trigger oxidative stress (OS), and induce neutrophil infiltration, ultimately disrupting the alveolar-capillary barrier (8). This process leads to increased permeability and pulmonary edema (PE), further worsening lung injury severity. Moreover, HG is related to an elevated risk of in-hospital mortality (IHM) and long-term death rate (9). Studies have demonstrated that notable fluctuations of BG may have an even greater impact on the individual’s health condition. For instance, a recent study suggests that glycemic variability (GV) is related to a higher risk of IHM in ICU patients (10).
GV is a key pattern of BG fluctuations over a specific period in seriously ill patients (11). Studies have indicated that increased GV is notably linked to the death rate and multi-organ dysfunction in severely ill patients (12,13). Thus, high GV is considered an indicator of poor prognosis. However, the relationship between GV and prognosis in ICU patients with TLI remains unclear. This retrospective study seeks to investigate the association between GV and the prognosis of TLI patients. Our findings may assist clinicians in closely monitoring TLI patients and providing timely interventions to reduce death rates and improve clinical outcomes. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-743/rc) (14).
Methods
Data source
The clinical data of individuals with TLI were extracted from the publicly available and freely accessible Medical Information Mart for Intensive Care IV (MIMIC-IV) version 3.0 database (15). One of the authors (L.D.) completed the necessary training to gain access to the database (record ID: 68779432). As the data were anonymized, ethical approval was waived, and no additional approval was required. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.
Inclusion and exclusion criteria
Inclusion criteria: patients diagnosed with TLI upon first admission based on ICD-9 or ICD-10 codes. Exclusion criteria: (I) patients not admitted to the ICU; (II) patients with missing BG data; (III) patients with fewer than 3 BG measurements; (IV) patients with an admission duration of more than 30 days.
Clinical data collection
Relevant medical information was extracted from the MIMIC-IV database using Structured Query Language. The following data were collected: (I) demographics: age, gender, race, and body mass index (BMI); (II) vital signs: diastolic blood pressure (DBP) and systolic blood pressure (SBP); (III) comorbidities defined based on ICD-9 or ICD-10 codes: acute respiratory distress syndrome (ARDS), acute kidney injury (AKI), chronic kidney disease (CKD), septic shock (SS), hypertension (HP), hypovolemic shock (HS), and diabetes; (IV) laboratory indicators: hemoglobin (HGB), white blood cells (WBCs), prothrombin time (PT), activated partial thromboplastin time (PTT), hematocrit (HCT), platelet (PLT) count, creatinine, creatine kinase (CK-x), international normalized ratio (INR), lactate, total bilirubin (TB), alanine aminotransferase (ALT), albumin, aspartate aminotransferase (AST), BG, and calcium. For repeated measurements, only the first set of data recorded within the first 72 hours of admission was analyzed.
To avoid bias caused by sample exclusion, multiple imputation using the “mice” package in R with random forest methods was applied for variables with less than 25% missing data. For variables with more than 25% missing data, missing values were treated as a separate category and analyzed as an independent subgroup. Among these variables, BMI, SBP, DBP, HCT, HGB, albumin, ALT, AST, CK-x, lactate, and TB had missing values exceeding 25%.
Exposure and outcome
The included patients were followed from the time of their ICU admission until discharge or death. The exposure variable was GV, which was calculated based on random BG levels measured from plasma during hospitalization and recorded in real time. Due to heterogeneity in the specific parameters of BG records across patients, there was no standardized timing for glucose monitoring. GV was calculated as the ratio of the standard deviation (SD) to the mean of all recorded BG values. The outcome was defined as the risk of IHM for TLI patients.
Statistical analysis
Continuous variables that were normally distributed were represented by SD, and a Student’s t-test was applied to assess the difference between groups. Non-normally distributed continuous variables were presented as the interquartile range (IQR), and the difference between groups was compared via the Mann-Whitney U or Kruskal-Wallis tests. Categorical variables were assessed via Pearson’s Chi-squared test or Fisher’s exact test.
The association between GV and the risk of IHM in TLI patients was examined via the Cox proportional hazards (CPH) models. The models were controlled for potential confounders: model 1 was uncontrolled; model 2 controlled for race, age, and gender; model 3 controlled for the variables in model 2 and further controlled for BMI, AKI, ARDS, CKD, HS, HP, SS, and diabetes.
Before constructing model 3, the variance inflation factor (VIF) was calculated for each variable to avoid multicollinearity, and variables with a VIF >5 were removed. Subsequently, restricted cubic spline (RCS) models were employed to investigate the nonlinear relationship between GV and the outcome. Finally, subgroup analyses were carried out to investigate the exposure-outcome relationship in different subgroups. Kaplan-Meier (KM) survival analysis was applied to assess the differences in survival rate (SR) between groups. All data analyses were conducted via the R version 4.4.1, and the two-sided P value was less than 0.05.
Results
Study population
The inclusion and exclusion criteria resulted in the inclusion of 1,232 patients. Figure 1 illustrates the patient selection process.
The study cohort consisted of 821 males and 411 females, with a median age of 58 years. The majority of the patients were White (n=756). The eligible individuals were grouped based on whether the death occurred during hospitalization. As illustrated in Table 1, patients who died in the hospital had a shorter length of stay [4 (2, 10) vs. 7 (5, 12) days], were more likely to be males (47% vs. 32%), and non-White individuals (57% vs. 37%). Furthermore, these individuals had a higher incidence of AKI (40% vs. 11%), ARDS (5.5% vs. 0.4%), and SS (16% vs. 0.4%).
Table 1
| Characteristic | Overall (n=1,232) | Alive (n=1,141) | Dead (n=91) | P value |
|---|---|---|---|---|
| Hospital stays (days) | 7 (5, 12) | 7 (5, 12) | 4 (2, 10) | <0.001 |
| Gender | 0.003 | |||
| Female | 411 [33] | 368 [32] | 43 [47] | |
| Male | 821 [67] | 773 [68] | 48 [53] | |
| Age (years) | 58 (35, 75) | 57 (34, 74) | 66 (47, 81) | 0.003 |
| Race | <0.001 | |||
| Non-White | 476 [39] | 424 [37] | 52 [57] | |
| White | 756 [61] | 717 [63] | 39 [43] | |
| BMI (kg/m2) | <0.001 | |||
| <18.5 | 19 [1.5] | 16 [1.4] | 3 [3.3] | |
| ≥30.0 | 117 [9.5] | 116 [10] | 1 [1.1] | |
| 18.5–24.9 | 191 [16] | 186 [16] | 5 [5.5] | |
| 25.0–29.9 | 148 [12] | 143 [13] | 5 [5.5] | |
| Missing | 757 [61] | 680 [60] | 77 [85] | |
| SBP (mmHg) | <0.001 | |||
| <90 | 4 [0.3] | 4 [0.4] | 0 [0] | |
| >120 | 335 [27] | 330 [29] | 5 [5.5] | |
| 90–120 | 243 [20] | 236 [21] | 7 [7.7] | |
| Missing | 650 [53] | 571 [50] | 79 [87] | |
| DBP (mmHg) | <0.001 | |||
| <60 | 50 [4.1] | 49 [4.3] | 1 [1.1] | |
| >90 | 38 [3.1] | 38 [3.3] | 0 [0] | |
| 60–90 | 494 [40] | 483 [42] | 11 [12] | |
| Missing | 650 [53] | 571 [50] | 79 [87] | |
| AKI | 164 [13] | 128 [11] | 36 [40] | <0.001 |
| ARDS | 9 [0.7] | 4 [0.4] | 5 [5.5] | <0.001 |
| CKD | 78 [6.3] | 68 [6.0] | 10 [11] | 0.058 |
| Hypovolemic | 11 [0.9] | 11 [1.0] | 0 [0] | >0.90 |
| HP | 382 [31] | 351 [31] | 31 [34] | 0.50 |
| SS | 20 [1.6] | 5 [0.4] | 15 [16] | <0.001 |
| Diabetes | 158 [13] | 144 [13] | 14 [15] | 0.40 |
| Albumin (g/dL) | <0.001 | |||
| <3.5 | 179 [15] | 140 [12] | 39 [43] | |
| 3.5–5.2 | 92 [7.5] | 88 [7.7] | 4 [4.4] | |
| Missing | 961 [78] | 913 [80] | 48 [53] | |
| ALT (U/L) | <0.001 | |||
| ≤40 | 194 [16] | 175 [15] | 19 [21] | |
| >40 | 229 [19] | 187 [16] | 42 [46] | |
| Missing | 809 [66] | 779 [68] | 30 [33] | |
| AST (U/L) | <0.001 | |||
| ≤40 | 113 [9.2] | 104 [9.1] | 9 [9.9] | |
| >40 | 312 [25] | 260 [23] | 52 [57] | |
| Missing | 807 [66] | 777 [68] | 30 [33] | |
| CK-x (IU/L) | <0.001 | |||
| <40 | 2 [0.2] | 2 [0.2] | 0 [0] | |
| >200 | 286 [23] | 248 [22] | 38 [42] | |
| 40–200 | 35 [2.8] | 28 [2.5] | 7 [7.7] | |
| Missing | 909 [74] | 863 [76] | 46 [51] | |
| HCT (%) | <0.001 | |||
| <37 | 81 [6.6] | 65 [5.7] | 16 [18] | |
| <40 | 184 [15] | 167 [15] | 17 [19] | |
| >50 | 1 [<0.1] | 0 [0] | 1 [1.1] | |
| 37–48 | 6 [0.5] | 5 [0.4] | 1 [1.1] | |
| 40–50 | 31 [2.5] | 25 [2.2] | 6 [6.6] | |
| Missing | 929 [75] | 879 [77] | 50 [55] | |
| HGB (g/dL) | ||||
| <11 | 22 [1.8] | 18 [1.6] | 4 [4.4] | |
| <12 | 154 [13] | 139 [12] | 15 [16] | |
| >15 | 45 [3.7] | 36 [3.2] | 9 [9.9] | |
| >16 | 2 [0.2] | 1 [<0.1] | 1 [1.1] | |
| 11–15 | 20 [1.6] | 16 [1.4] | 4 [4.4] | |
| 12–16 | 60 [4.9] | 52 [4.6] | 8 [8.8] | |
| Missing | 929 [75] | 879 [77] | 50 [55] | |
| INR | 1.20 (1.10, 1.30) | 1.20 (1.10, 1.30) | 1.30 (1.10, 1.60) | <0.001 |
| Lactate (mmol/L) | <0.001 | |||
| >2 | 353 [29] | 293 [26] | 60 [66] | |
| 0.5–2 | 288 [23] | 268 [23] | 20 [22] | |
| Missing | 591 [48] | 580 [51] | 11 [12] | |
| PT (s) | 13.10 (11.90, 14.60) | 13.00 (11.90, 14.50) | 14.00 (13.10, 17.60) | <0.001 |
| PTT (%) | 28 (25, 31) | 28 (25, 31) | 31 (27, 42) | <0.001 |
| WBC (109/L) | 11.4 (8.6, 15.3) | 11.3 (8.6, 15.2) | 13.3 (8.7, 18.1) | 0.11 |
| Calcium (mmol/L) | 8.40 (7.90, 8.80) | 8.40 (7.90, 8.80) | 8.00 (7.20, 8.70) | <0.001 |
| Creatinine (mg/dL) | 0.90 (0.70, 1.10) | 0.90 (0.70, 1.10) | 1.20 (0.80, 1.50) | <0.001 |
| Platelets (109/L) | 192 (147, 236) | 194 (150, 237) | 158 (116, 213) | <0.001 |
| Hemoglobin (g/dL) | 11.60 (10.20, 12.90) | 11.60 (10.30, 12.90) | 10.80 (9.00, 12.60) | 0.005 |
| Hematocrit (%) | 34.9 (30.9, 38.5) | 35.0 (31.2, 38.4) | 32.9 (27.2, 39.1) | 0.01 |
| TB (mg/L) | <0.001 | |||
| ≤1.5 | 361 [29] | 313 [27] | 48 [53] | |
| >1.5 | 43 [3.5] | 33 [2.9] | 10 [11] | |
| Missing | 828 [67] | 795 [70] | 33 [36] | |
| GV | 0.16 (0.12, 0.23) | 0.16 (0.12, 0.22) | 0.23 (0.15, 0.33) | <0.001 |
Data are presented as n [%] or median (Q1, Q3). AKI, acute kidney injury; ALT, alanine aminotransferase; ARDS, acute respiratory distress syndrome; AST, aspartate aminotransferase; BMI, body mass index; CKD, chronic kidney disease; CK, creatine kinase; DBP, diastolic blood pressure; GV, glycaemic variability; HCT, hematocrit; HGB, hemoglobin; HP, hypertension; INR, international normalized ratio; PT, prothrombin time; PTT, partial thromboplastin time; SBP, systolic blood pressure; SS, septic shock; TB, total bilirubin; WBC, white blood cell.
Table 2 illustrates the baseline characteristics of eligible patients grouped by the GV quartiles. The results revealed that individuals in the highest GV quartile (Q4) had a higher risk of IHM, elevated incidence of AKI, SS, and diabetes, longer hospital stays, lower albumin levels, and increased levels of lactate, ALT, CK-x, AST, and were more likely to be males and White individuals.
Table 2
| Characteristic | Overall (n=1,232) | Q1 (n=324) | Q2 (n=303) | Q3 (n=301) | Q4 (n=304) | P value |
|---|---|---|---|---|---|---|
| Dead | 91 [7.4] | 9 [2.8] | 17 [5.6] | 17 [5.6] | 48 [16] | <0.001 |
| Hospital stays (days) | 7 (5, 12) | 6 (4, 8) | 8 (5, 14) | 8 (5, 13) | 8 (5, 15) | <0.001 |
| Gender | <0.001 | |||||
| Female | 411 [33] | 95 [29] | 82 [27] | 109 [36] | 125 [41] | |
| Male | 821 [67] | 229 [71] | 221 [73] | 192 [64] | 179 [59] | |
| Age (years) | 58 (35, 75) | 59 (35, 77) | 52 (29, 70) | 59 (35, 76) | 61 (46, 77) | <0.001 |
| Race | 0.40 | |||||
| Non-White | 476 [39] | 128 [40] | 125 [41] | 105 [35] | 118 [39] | |
| White | 756 [61] | 196 [60] | 178 [59] | 196 [65] | 186 [61] | |
| BMI (kg/m2) | ||||||
| <18.5 | 19 [1.5] | 4 [1.2] | 4 [1.3] | 8 [2.7] | 3 [1.0] | |
| ≥30.0 | 117 [9.5] | 34 [10] | 27 [8.9] | 26 [8.6] | 30 [9.9] | |
| 18.5–24.9 | 191 [16] | 49 [15] | 52 [17] | 46 [15] | 44 [14] | |
| 25.0–29.9 | 148 [12] | 46 [14] | 32 [11] | 39 [13] | 31 [10] | |
| Missing | 757 [61] | 191 [59] | 188 [62] | 182 [60] | 196 [64] | |
| SBP (mmHg) | ||||||
| <90 | 4 [0.3] | 1 [0.3] | 1 [0.3] | 0 [0] | 2 [0.7] | |
| >120 | 335 [27] | 99 [31] | 80 [26] | 80 [27] | 76 [25] | |
| 90–120 | 243 [20] | 51 [16] | 71 [23] | 63 [21] | 58 [19] | |
| Missing | 650 [53] | 173 [53] | 151 [50] | 158 [52] | 168 [55] | |
| DBP (mmHg) | >0.99 | |||||
| <60 | 50 [4.1] | 14 [4.3] | 12 [4.0] | 11 [3.7] | 13 [4.3] | |
| >90 | 38 [3.1] | 12 [3.7] | 8 [2.6] | 9 [3.0] | 9 [3.0] | |
| 60–90 | 494 [40] | 125 [39] | 132 [44] | 123 [41] | 114 [38] | |
| Missing | 650 [53] | 173 [53] | 151 [50] | 158 [52] | 168 [55] | |
| AKI | 164 [13] | 21 [6.5] | 36 [12] | 45 [15] | 62 [20] | <0.001 |
| ARDS | 9 [0.7] | 0 [0] | 4 [1.3] | 2 [0.7] | 3 [1.0] | 0.20 |
| CKD | 78 [6.3] | 18 [5.6] | 14 [4.6] | 17 [5.6] | 29 [9.5] | 0.06 |
| HS | 11 [0.9] | 2 [0.6] | 3 [1.0] | 5 [1.7] | 1 [0.3] | 0.30 |
| HP | 382 [31] | 100 [31] | 79 [26] | 100 [33] | 103 [34] | 0.20 |
| SS | 20 [1.6] | 0 [0] | 1 [0.3] | 6 [2.0] | 13 [4.3] | <0.001 |
| Diabetes | 158 [13] | 23 [7.1] | 17 [5.6] | 27[9.0] | 91 [30] | <0.001 |
| Albumin (g/dL) | <0.001 | |||||
| <3.5 | 179 [15] | 26 [8.0] | 42 [14] | 45 [15] | 66 [22] | |
| 3.5–5.2 | 92 [7.5] | 16 [4.9] | 28 [9.2] | 26 [8.6] | 22 [7.2] | |
| Missing | 961 [78] | 282 [87] | 233 [77] | 230 [76] | 216 [71] | |
| ALT (U/L) | <0.001 | |||||
| ≤40 | 194 [16] | 35 [11] | 43 [14] | 57 [19] | 59 [19] | |
| >40 | 229 [19] | 32 [9.9] | 58 [19] | 59 [20] | 80 [26] | |
| Missing | 809 [66] | 257 [79] | 202 [67] | 185 [61] | 165 [54] | |
| AST (U/L) | <0.001 | |||||
| ≤40 | 113 [9.2] | 22 [6.8] | 26 [8.6] | 36 [12] | 29 [9.5] | |
| >40 | 312 [25] | 48 [15] | 74 [24] | 80 [27] | 110 [36] | |
| Missing | 807 [66] | 254 [78] | 203 [67] | 185 [61] | 165 [54] | |
| CK-x (IU/L) | ||||||
| <40 | 2 [0.2] | 0 [0] | 0 [0] | 0 [0] | 2 [0.7] | |
| >200 | 286 [23] | 51 [16] | 71 [23] | 74 [25] | 90 [30] | |
| 40–200 | 35 [2.8] | 3 [0.9] | 9 [3.0] | 15 [5.0] | 8 [2.6] | |
| Missing | 909 [74] | 270 [83] | 223 [74] | 212 [70] | 204 [67] | |
| HCT (%) | ||||||
| <37 | 81 [6.6] | 10 [3.1] | 18 [5.9] | 24 [8.0] | 29 [9.5] | |
| <40 | 184 [15] | 23 [7.1] | 63 [21] | 47 [16] | 51 [17] | |
| >50 | 1 [<0.1] | 0 [0] | 0 [0] | 0 [0] | 1 [0.3] | |
| 37–48 | 6 [0.5] | 1 [0.3] | 1 [0.3] | 2 [0.7] | 2 [0.7] | |
| 40–50 | 31 [2.5] | 9 [2.8] | 9 [3.0] | 8 [2.7] | 5 [1.6] | |
| Missing | 929 [75] | 281 [87] | 212 [70] | 220 [73] | 216 [71] | |
| HGB (g/dL) | ||||||
| <11 | 22 [1.8] | 0 [0] | 4 [1.3] | 8 [2.7] | 10 [3.3] | |
| <12 | 154 [13] | 19 [5.9] | 53 [17] | 43 [14] | 39 [13] | |
| >15 | 45 [3.7] | 8 [2.5] | 10 [3.3] | 13 [4.3] | 14 [4.6] | |
| >16 | 2 [0.2] | 0 [0] | 1 [0.3] | 0 [0] | 1 [0.3] | |
| 11–15 | 20 [1.6] | 3 [0.9] | 5 [1.7] | 5 [1.7] | 7 [2.3] | |
| 12–16 | 60 [4.9] | 13 [4.0] | 18 [5.9] | 12 [4.0] | 17 [5.6] | |
| Missing | 929 [75] | 281 [87] | 212 [70] | 220 [73] | 216 [71] | |
| INR | 1.20 (1.10, 1.30) | 1.20 (1.10, 1.30) | 1.20 (1.10, 1.30) | 1.20 (1.10, 1.30) | 1.20 (1.10, 1.40) | 0.055 |
| Lactate (mmol/L) | <0.001 | |||||
| >2 | 353 [29] | 58 [18] | 100 [33] | 81 [27] | 114 [38] | |
| 0.5–2 | 288 [23] | 51 [16] | 80 [26] | 86 [29] | 71 [23] | |
| Missing | 591 [48] | 215 [66] | 123 [41] | 134 [45] | 119 [39] | |
| PT (s) | 13.1 (11.9, 14.6) | 13.0 (11.9, 14.4) | 13.1 (11.9, 14.7) | 13.0 (11.9, 14.5) | 13.3 (12.1, 15.2) | 0.046 |
| PTT (%) | 28 (25, 31) | 28 (26, 31) | 28 (26, 31) | 28 (25, 31) | 28 (25, 32) | 0.60 |
| WBC (109/L) | 11.5 (8.5, 15.3) | 10.4 (7.9, 13.3) | 11.7 (8.6, 16.2) | 12.3 (9.2, 16.4) | 12.2 (9.0, 17.1) | <0.001 |
| Calcium (mmol/L) | 8.40 (7.90, 8.80) | 8.55 (8.10, 8.90) | 8.30 (7.80, 8.80) | 8.40 (7.80, 8.90) | 8.30 (7.75, 8.80) | <0.001 |
| Creatinine (mg/dL) | 0.90 (0.70, 1.10) | 0.80 (0.70, 1.00) | 0.90 (0.70, 1.10) | 0.90 (0.70, 1.10) | 1.00 (0.70, 1.20) | <0.001 |
| Platelets (109/L) | 191 (146, 236) | 201 (162, 232) | 178 (140, 228) | 195 (147, 240) | 191 (142, 242) | 0.02 |
| Hemoglobin (g/dL) | 11.60 (10.20, 12.85) | 11.60 (10.60, 12.80) | 11.70 (10.20, 13.00) | 11.60 (10.10, 12.60) | 11.55 (9.90, 13.00) | 0.20 |
| Hematocrit (%) | 34.9 (30.9, 38.5) | 35.2 (31.9, 38.4) | 35.1 (31.0, 38.9) | 34.6 (30.5, 38.1) | 33.8 (29.8, 38.6) | 0.10 |
| TB | <0.001 | |||||
| ≤1.5 | 361 [29] | 59 [18] | 81 [27] | 100 [33] | 121 [40] | |
| >1.5 | 43 [3.5] | 8 [2.5] | 16 [5.3] | 7 [2.3] | 12 [3.9] | |
| Missing | 828 [67] | 257 [79] | 206 [68] | 194 [64] | 171 [56] | |
| GV | 0.16 (0.12, 0.23) | 0.09 (0.06, 0.10) | 0.14 (0.13, 0.15) | 0.19 (0.18, 0.21) | 0.30 (0.26, 0.39) | <0.001 |
Data are presented as n [%] or median (Q1, Q3). Quartile: Q1, [0.010–0.119]; Q2, (0.119–0.163]; Q3, (0.163–0.227]; Q4, (0.227–1.756]. AKI, acute kidney injury; ALT, alanine aminotransferase; ARDS, acute respiratory distress syndrome; AST, aspartate aminotransferase; BMI, body mass index; CKD, chronic kidney disease; CK, creatine kinase; DBP, diastolic blood pressure; GV, glycaemic variability; HCT, hematocrit; HGB, hemoglobin; HP, hypertension; INR, international normalized ratio; PT, prothrombin time; PTT, partial thromboplastin time; Q, quartile; SBP, systolic blood pressure; SS, septic shock; TB, total bilirubin; WBC, white blood cell.
Subsequently, a CPH model was constructed to examine the association between GV and the risk of IHM. Variables excluded from the final model due to limited contribution included SBP, DBP, albumin, ALT, AST, CK-x, HCT, HGB, INR, lactate, PT, PTT, WBC, calcium, creatinine, PLTs, HGB, HCT, and TB.
As shown in Table 3, when GV was presented as a continuous variable, the results suggested that higher GV was notably linked to an elevated risk of IHM in TLI patients. After adjusting for all confounders, the association remained significant [hazard ratio (HR) =3.588; 95% confidence interval (CI): 1.153, 11.168].
Table 3
| Categories | Model 1 | Model 2 | Model 3 | |||||
|---|---|---|---|---|---|---|---|---|
| HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value | |||
| Continuity | 3.555 (1.602, 7.891) | 0.002 | 4.819 (2.006, 11.578) | <0.001 | 3.588 (1.153, 11.168) | 0.03 | ||
| Quartile | ||||||||
| Q1 (n=324) | Ref. | Ref. | Ref. | |||||
| Q2 (n=303) | 1.538 (0.684, 3.459) | 0.30 | 1.594 (0.708, 3.590) | 0.26 | 1.324 (0.580, 3.025) | 0.51 | ||
| Q3 (n=301) | 1.549 (0.689, 3.484) | 0.29 | 1.394 (0.616, 3.152) | 0.43 | 1.037 (0.451, 2.384) | 0.93 | ||
| Q4 (n=304) | 4.212 (2.057, 8.625) | <0.001 | 3.668 (1.785, 7.540) | <0.001 | 2.685 (1.269, 5.680) | 0.01 | ||
| P of trend | <0.001 | 0.001 | 0.03 | |||||
Model 1 was unadjusted. Model 2 was adjusted for gender, age and race. Model 3 was adjusted for the variables in model 2 and further adjusted for BMI, AKI, ARDS, CKD, hypovolemic, HG, septic, and diabetes. Quartile: Q1, [0.010–0.119]; Q2, (0.119–0.163]; Q3, (0.163–0.227]; Q4, (0.227–1.756]. AKI, acute kidney injury; ARDS, acute respiratory distress syndrome; BMI, body mass index; CI, confidence interval; CKD, chronic kidney disease; CPH, Cox proportional hazards; HG, hyperglycemia; HR, hazard ratio.
In contrast to patients with low GV (Q1), those with high GV (Q4) had a notably higher risk of IHM. This trend was consistent in model 1 (HR =4.212; 95% CI: 2.057, 8.625), model 2 (HR =3.668; 95% CI: 1.785, 7.540), and model 3 (HR =2.685; 95% CI: 1.269, 5.680).
Figure 2 illustrates the potential nonlinear relationship between GV and the risk of IHM in models 1, 2, and 3. The results indicated a significant nonlinear association between GV and risk of IHM in model 1 (P of non-linearity <0.001), model 2 (P of non-linearity =0.008), and model 3 (P of non-linearity =0.04). Furthermore, the threshold value of GV corresponding to an HR of 1 was calculated to be 0.163. The results indicated that when GV was greater than 0.163, the risk of IHM notably increased. Conversely, when GV was below 0.163, the risk of IHM was relatively lower and considered to be in a safe range.
Figure 3 illustrates the results of subgroup analysis based on race, gender, age, and HP. The results revealed a significant positive association between GV and the risk of IHM in patients aged >60 years, males, females, non-White individuals, and those without HP. Additionally, an interaction between GV and race was observed.
Figure 4 illustrates the KM survival analysis. The results indicated notable differences in SR between individuals with different levels of GV during the follow-up period (log-rank P<0.001).
Discussion
Based on the MIMIC-IV database, this retrospective study first identified a significant association between GV and the risk of IHM in TLI patients. This study suggests that clinical attention should be given to glucose homeostasis in TLI patients.
In severely ill individuals, elevated GV can further worsen the patient’s condition. The pathophysiological mechanisms of GV involve a hypermetabolic stress response, where the body secretes thyroid hormones, glucagon, and glucocorticoids. These hormones accelerate the breakdown of nutrients and elevate BG levels, leading to notable BG fluctuations (16-18). A prospective study has confirmed that higher GV is linked to death rates in seriously ill patients (10). Increased GV can induce OS and endothelial injury, exacerbate PE, impair the function of neutrophils and macrophages, disrupt immune function, increase the risk of pulmonary infections, and interfere with the synthesis and secretion of surfactants in the alveoli. This can further reduce pulmonary compliance and exacerbate respiratory failure (19-26), ultimately leading to poorer outcomes in TLI patients.
Our study found a positive association between GV and the risk of IHM in TLI patients, suggesting that monitoring GV in clinical practice can reduce the risk of IHM among TLI patients in the ICU. This study suggests that more precise management of BG in TLI patients is needed in clinical practice. Timely interventions can lower GV fluctuations, thereby decreasing the occurrence of high GV and ultimately improving the prognosis of TLI patients. Nonetheless, studies on the relationship between GV and TLI patients are scarce, and no prospective studies have yet confirmed that reducing GV can improve patient outcomes. Hence, further prospective studies are required to validate our findings. Additionally, effective interventions for reducing GV are lacking. Strict glucose control may increase the risk of hypoglycemia, which in turn may elevate the death rate in seriously ill patients (9,27,28). Effective interventions for reducing GV could be a focus of our future research.
This study revealed an association between GV and the risk of IHM in TLI patients. Nevertheless, several limitations need to be acknowledged. First, since the majority of the included individuals were White, it remains unclear whether these findings apply to African or Asian populations. Second, as the data were all collected from the MIMIC-IV database, information on blood loss or pulmonary function in TLI patients is not included for analysis. Lastly, as a retrospective cohort study, despite efforts to minimize reverse causality in the study design, the findings should be validated through prospective longitudinal cohort studies. Missing data were handled based on the proportion of missingness. However, this approach may introduce bias due to missingness. In addition, BG monitoring frequency varied among patients with varying diabetes risk, and the time intervals of BG measurements in the MIMIC-IV database were inconsistent, potentially influencing the reliability of the findings.
Conclusions
This study suggests that as GV increases, the risk of IHM in TLI patients notably rises. When GV is below 0.163, the risk of IHM in TLI patients is relatively low.
Acknowledgments
None.
Footnote
Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-743/rc
Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-743/prf
Funding: None.
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-743/coif). The authors have no conflicts of interest to declare.
Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.
Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.
References
- ÅRSRAPPORT Swedish trauma register. 2019. Available online: https://swetrau.se
- Vécsei V, Arbes S, Aldrian S, et al. Chest Injuries in Polytrauma. European Journal of Trauma 2005;31:239-43.
- Bardenheuer M, Obertacke U, Waydhas C, et al. Epidemiologie des Schwerverletzten Eine prospektive Erfassung der präklinischen und klinischen Versorgung. Unfallchirurg 2000;103:355-63. [Crossref] [PubMed]
- Bayer J, Lefering R, Reinhardt S, et al. Severity-dependent differences in early management of thoracic trauma in severely injured patients - Analysis based on the TraumaRegister DGU®. Scand J Trauma Resusc Emerg Med 2017;25:10. [Crossref] [PubMed]
- Friberg K, Hofsø K, Ræder J, et al. Prevalence of and predictive factors associated with high levels of post-traumatic stress symptoms 3 months after intensive care unit admission: A prospective study. Aust Crit Care 2024;37:222-9. [Crossref] [PubMed]
- Johnson BK, Comstock RD. Epidemiology of Chest, Rib, Thoracic Spine, and Abdomen Injuries Among United States High School Athletes, 2005/06 to 2013/14. Clin J Sport Med 2017;27:388-93. [Crossref] [PubMed]
- Lheureux O, Prevedello D, Preiser JC. Update on glucose in critical care. Nutrition 2019;59:14-20. [Crossref] [PubMed]
- He W, Mu X, Wu X, et al. The cGAS-STING pathway: a therapeutic target in diabetes and its complications. Burns Trauma 2024;12:tkad050. [Crossref] [PubMed]
- Bohé J, Abidi H, Brunot V, et al. Individualised versus conventional glucose control in critically-ill patients: the CONTROLING study-a randomized clinical trial. Intensive Care Med 2021;47:1271-83. [Crossref] [PubMed]
- Singh M, Upreti V, Singh Y, et al. Effect of Glycemic Variability on Mortality in ICU Settings: A Prospective Observational Study. Indian J Endocrinol Metab 2018;22:632-5. [Crossref] [PubMed]
- Ceriello A, Monnier L, Owens D. Glycaemic variability in diabetes: clinical and therapeutic implications. Lancet Diabetes Endocrinol 2019;7:221-30. [Crossref] [PubMed]
- Hsu JC, Yang YY, Chuang SL, et al. Higher long-term visit-to-visit glycemic variability predicts new-onset atrial fibrillation in patients with diabetes mellitus. Cardiovasc Diabetol 2021;20:148. [Crossref] [PubMed]
- Chun KH, Oh J, Lee CJ, et al. In-hospital glycemic variability and all-cause mortality among patients hospitalized for acute heart failure. Cardiovasc Diabetol 2022;21:291. [Crossref] [PubMed]
- Vandenbroucke JP, von Elm E, Altman DG, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. PLoS Med 2007;4:e297. [Crossref] [PubMed]
- Johnson AEW, Bulgarelli L, Shen L, et al. MIMIC-IV, a freely accessible electronic health record dataset. Sci Data 2023;10:1. [Crossref] [PubMed]
- Ali NA, O'Brien JM Jr, Dungan K, et al. Glucose variability and mortality in patients with sepsis. Crit Care Med 2008;36:2316-21. [Crossref] [PubMed]
- Bagshaw SM, Bellomo R, Jacka MJ, et al. The impact of early hypoglycemia and blood glucose variability on outcome in critical illness. Crit Care 2009;13:R91. [Crossref] [PubMed]
- Marik PE, Raghavan M. Stress-hyperglycemia, insulin and immunomodulation in sepsis. Intensive Care Med 2004;30:748-56. [Crossref] [PubMed]
- Ceriello A, Esposito K, Piconi L, et al. Oscillating glucose is more deleterious to endothelial function and oxidative stress than mean glucose in normal and type 2 diabetic patients. Diabetes 2008;57:1349-54. [Crossref] [PubMed]
- Quagliaro L, Piconi L, Assaloni R, et al. Intermittent high glucose enhances apoptosis related to oxidative stress in human umbilical vein endothelial cells: the role of protein kinase C and NAD(P)H-oxidase activation. Diabetes 2003;52:2795-804. [Crossref] [PubMed]
- Russo VC, Higgins S, Werther GA, et al. Effects of fluctuating glucose levels on neuronal cells in vitro. Neurochem Res 2012;37:1768-82. [Crossref] [PubMed]
- Ceriello A, Novials A, Ortega E, et al. Evidence that hyperglycemia after recovery from hypoglycemia worsens endothelial function and increases oxidative stress and inflammation in healthy control subjects and subjects with type 1 diabetes. Diabetes 2012;61:2993-7. [Crossref] [PubMed]
- Monnier L, Mas E, Ginet C, et al. Activation of oxidative stress by acute glucose fluctuations compared with sustained chronic hyperglycemia in patients with type 2 diabetes. JAMA 2006;295:1681-7. [Crossref] [PubMed]
- Frontoni S, Di Bartolo P, Avogaro A, et al. Glucose variability: An emerging target for the treatment of diabetes mellitus. Diabetes Res Clin Pract 2013;102:86-95. [Crossref] [PubMed]
- Maeda M, Hayashi T, Mizuno N, et al. Intermittent high glucose implements stress-induced senescence in human vascular endothelial cells: role of superoxide production by NADPH oxidase. PLoS One 2015;10:e0123169. [Crossref] [PubMed]
- Wang HW, Lu JY, Wang XH, et al. Implications of lung interstitial dendritic cells of mice in multiple organ dysfunction syndrome. Zhonghua Bing Li Xue Za Zhi 2007;36:307-12.
- Yamada T, Shojima N, Hara K, et al. Glycemic control, mortality, secondary infection, and hypoglycemia in critically ill pediatric patients: a systematic review and network meta-analysis of randomized controlled trials. Intensive Care Med 2017;43:1427-9. [Crossref] [PubMed]
- NICE-SUGAR Study Investigators. Intensive versus conventional glucose control in critically ill patients. N Engl J Med 2009;360:1283-97. [Crossref] [PubMed]
