BUN and mortality in patients with heparin-induced thrombocytopenia: A retrospective cohort study

Abstract

Introduction

Heparin-induced thrombocytopenia (HIT) is a severe complication of heparin treatment, characterized by low platelet counts and heightened thrombotic risk. Blood urea nitrogen (BUN), which serves as an indicator of both renal function and illness severity, has been associated with poor outcomes in different contexts. However, its association with mortality in HIT remains poorly understood. Our study aimed to explore this relationship in critically ill patients with HIT.

Methods

This study was a retrospective cohort analysis utilizing the MIMIC-IV 3.1 database from 2008 to 2019. Patients with HIT were pinpointed through ICD codes. Those without BUN data or whose intensive care unit (ICU) admission was not their first were excluded. The main outcome measured was all-cause mortality, evaluated at multiple time points. The associations were examined using multivariate Cox regression models and Kaplan–Meier survival analysis.

Results

The research encompassed 246 individuals with HIT (average age 66.2 years, 54.9% male). BUN levels showed a significant association with all-cause mortality during the hospital stay (HR 1.01, 95% CI 1.01–1.02, p = 0.001), within 30 days (HR 1.01, 95% CI 1.01–1.02, p < 0.001), within 90 days (HR 1.01, 95% CI 1.01–1.02, p < 0.001), and within 365 days (HR 1.01, 95% CI 1.01–1.02, p < 0.001). Quartile analysis revealed that the highest BUN quartile (Q4) was associated with the greatest mortality risk compared to the lowest quartile (Q1) at all time points. Kaplan–Meier and restricted cubic spline analyses corroborated these results, indicating a linear relationship between BUN and mortality.

Conclusion

Elevated BUN levels were significantly associated with higher mortality rates among HIT patients in the ICU. Monitoring BUN levels may help identify patients at greater risk and inform clinical choices. Further research is warranted to elucidate the underlying mechanisms and possible treatments.

Introduction

Heparin-induced thrombocytopenia (HIT) is an immune-mediated, potentially fatal adverse reaction to heparin that reduces platelet counts and increases thrombotic risk.^1,2 It occurs in approximately 1–5% of patients ³ receiving unfractionated heparin and 0.1–0.5% of those receiving low-molecular-weight heparin. ⁴ The clinical presentation of HIT can range from an asymptomatic reduction in platelet count to severe complications such as venous limb ischemia, pulmonary embolism, and even death.^5–7

Blood urea nitrogen (BUN) is a commonly measured biochemical parameter that reflects renal function and is also an indicator of protein metabolism. ⁸ Elevated BUN levels can be a marker of poor renal function, but they may also reflect conditions such as dehydration, sepsis, or gastrointestinal bleeding.^9–11 In critical illness, BUN can be influenced by various factors including kidney injury, protein catabolism, and fluid status. ¹² Previous studies have suggested that BUN levels are associated with mortality in critically ill patients, independent of renal function. ¹³

Although BUN indicates illness severity,^10,14 its impact on HIT mortality remains unclear. We examined whether BUN predicts death in intensive care unit (ICU) patients with HIT to identify high-risk cases early.

Methods

Study design, data source, and ethical considerations

This retrospective cohort study was carried out in accordance with the Declaration of Helsinki 1975, as revised in 2024. Access to the MIMIC-IV database was approved after completion of the Collaborative Institutional Training Initiative (CITI) ‘Data or Specimens Only Research’ course (certification 65 828 445). The institutional review boards of MIT and Beth Israel Deaconess Medical Center have determined that the MIMIC-IV resource is de-identified and publicly available; therefore, our retrospective analysis was exempt from full board review and informed consent. The study drew on data from the MIMIC-IV 3.1 database, covering the period from 2008 to 2019. ¹⁵ The main outcome measured was all-cause mortality, evaluated at several predefined time points after ICU admission. The reporting of this study conforms to STROBE guidelines. ¹⁶

Patient inclusion and exclusion criteria

Individuals with HIT were recognized through ICD codes D7582 and 28984. Inclusion criteria: (1) age ≥18 years; (2) first ICU admission; (3) confirmed HIT via ICD codes D7582 and 28984; (4) available BUN measurement within 24 h of ICU admission. Exclusion criteria: (1) age <18; (2) non-first ICU admission; (3) missing BUN data; (4) end-stage renal disease (ESRD) or chronic dialysis dependence, identified by ICD codes N18.6, Z99.2, or dialysis procedural codes. Missing data were managed via multiple imputation by chained equations, which fills in missing values step by step based on the observed relationships in the data.

As shown in Figure 1, the initial identification of HIT patients in the MIMIC-IV database yielded 535 patients. After excluding 289 patients with never- or non-first-time ICU admissions, 246 patients were identified as having HIT admitted to the ICU. No patients were excluded due to missing BUN data, resulting in a final analysis cohort of 246 patients.

Figure 1.

Flowchart of patient inclusion. BUN (quartile): Q1(＜16), Q2（16–24）, Q3（24–40）, Q4（＞40）.

Data collection

BUN levels were collected from the first measurement upon ICU admission. Other baseline characteristics, including demographic data, clinical variables, and laboratory results, were also extracted from the database. The BUN levels were categorized into quartiles (Q1: <16, Q2: 16–24, Q3: 24–40, Q4: >40) for further analysis. Quartiles were chosen instead of conventional clinical cut-points because no universally accepted BUN thresholds exist for critically ill patients with HIT. Distributional-based quartiles allow the data to define the exposure gradient, minimize arbitrary classification, and facilitate comparison with prior ICU studies that used the same approach. ¹³

Statistical analyses

Statistical analyses were conducted with R Statistical Software (Version 4.2.2, from The R Project for Statistical Computing, The R Foundation) and the Free Statistics Analysis Platform (Version 2.2, developed in Beijing, China). ¹⁷ Continuous variables were described by means and standard deviations, whereas categorical variables were shown as counts and percentages. The Shapiro–Wilk test was used to check the normality of continuous variables.

The link between BUN levels and mortality was examined using multivariate Cox proportional hazards regression models. Model 1 was unadjusted. Model 2 was adjusted for age, gender, race, and BMI. Model 3 was further adjusted for comorbidities such as myocardial infarction, congestive heart failure, cerebrovascular disease, diabetes, renal disease, and malignant cancer. The proportional hazards assumption was checked using time-dependent covariates and Schoenfeld residuals.

Kaplan–Meier analysis was used to calculate survival chances at various times (during the hospital stay, and at 30, 90, and 365 days). Log-rank tests compared survival curves across BUN quartiles. Restricted cubic spline (RCS) curves explored the nonlinear link between BUN and mortality. A correlation matrix assessed relationships between baseline variables.

Results

Baseline characteristics

The study population's baseline characteristics are shown in Table 1. The average age was 66.2 years, with 54.9% being male. Most patients were White (60.6%), and the average BMI was 31.3. There were significant differences across quartiles in age (p = 0.049), congestive heart failure (p = 0.003), renal disease (p < 0.001), and potassium levels (p < 0.001).

Table 1.

Baseline characteristics of study population.

Variables	Total (n = 246)	Q1 (n = 54)	Q2 (n = 68)	Q3 (n = 62)	Q4 (n = 62)	p-value
Age, mean (SD), year	66.2 ± 14.0	61.7 ± 16.4	66.5 ± 11.9	67.4 ± 13.8	68.6 ± 13.6	0.049
Gender, n (%)						0.068
Female	111 (45.1)	31 (57.4)	23 (33.8)	30 (48.4)	27 (43.5)
Male	135 (54.9)	23 (42.6)	45 (66.2)	32 (51.6)	35 (56.5)
Race, n (%)						0.629
White	149 (60.6)	32 (59.3)	42 (61.8)	41 (66.1)	34 (54.8)
Non-White	97 (39.4)	22 (40.7)	26 (38.2)	21 (33.9)	28 (45.2)
BMI, mean (SD)	31.3 ± 8.3	30.1 ± 7.5	31.2 ± 7.5	31.5 ± 9.1	32.2 ± 9.0	0.621
Heart rate, mean (SD), bpm	88.3 ± 16.9	86.3 ± 14.5	89.1 ± 17.5	90.7 ± 16.1	86.7 ± 19.0	0.439
Systolic BP, mean (SD), mm Hg	113.5 ± 15.6	116.4 ± 17.4	114.5 ± 12.3	110.0 ± 14.9	113.4 ± 17.6	0.163
Diastolic BP, mean (SD), mm Hg	62.9 ± 10.8	65.3 ± 11.3	62.7 ± 10.4	62.1 ± 10.7	61.8 ± 10.8	0.286
Glucose, mean (SD), mg/dL	155.9 ± 51.8	150.7 ± 52.2	150.1 ± 37.4	155.6 ± 58.0	167.0 ± 57.5	0.237
Hematocrit, mean (SD), %	29.9 ± 7.2	30.9 ± 6.6	28.8 ± 6.5	29.4 ± 7.2	30.7 ± 8.4	0.292
Hemoglobin, mean (SD), g/dL	9.8 ± 2.4	10.2 ± 2.2	9.4 ± 2.2	9.5 ± 2.4	10.1 ± 2.6	0.228
Platelets, mean (SD), ×103/μL	158.5 ± 90.5	170.3 ± 101.0	147.1 ± 86.3	158.8 ± 92.5	160.4 ± 83.8	0.568
WBC count, mean (SD), ×103/μL	16.3 ± 8.6	14.6 ± 6.7	15.7 ± 8.3	17.5 ± 9.7	17.0 ± 9.4	0.261
Bicarbonate, mean (SD), mmol/L	20.9 ± 5.4	21.6 ± 4.3	19.9 ± 5.2	21.3 ± 6.0	21.0 ± 5.8	0.302
Anion gap, mean (SD), mmol/L	17.3 ± 5.8	14.7 ± 4.4	16.7 ± 5.3	17.7 ± 5.3	19.9 ± 6.6	< 0.001
Calcium, mean (SD), mg/dL	8.1 ± 1.1	8.1 ± 0.7	7.9 ± 1.0	8.1 ± 1.4	8.2 ± 0.9	0.623
Chloride, mean (SD), mmol/L	100.4 ± 6.8	102.2 ± 4.9	101.8 ± 6.1	98.9 ± 5.6	98.9 ± 9.0	0.004
Creatinine, median (IQR), mg/dL	1.2 (0.9, 2.0)	0.8 (0.6, 0.9)	1.0 (0.8, 1.3)	1.4 (1.0, 1.9)	3.0 (1.9, 4.9)	< 0.001
Sodium, mean (SD), mmol/L	136.2 ± 6.0	136.6 ± 3.7	136.1 ± 4.8	135.7 ± 4.5	136.3 ± 9.2	0.851
Potassium, mean (SD), mmol/L	4.7 ± 0.8	4.4 ± 0.8	4.6 ± 0.6	4.8 ± 0.7	5.1 ± 0.9	< 0.001
INR, median (IQR)	1.5 (1.3, 2.0)	1.4 (1.2, 1.7)	1.5 (1.3, 2.1)	1.5 (1.3, 2.4)	1.5 (1.3, 2.3)	0.217
PT, median (IQR), s	16.5 (13.8, 21.6)	16.2 (13.7, 18.9)	16.8 (13.9, 22.5)	16.6 (13.8, 25.3)	16.6 (13.8, 24.3)	0.392
APTT, median (IQR), s	40.8 (29.7, 63.6)	41.3 (29.9, 63.2)	37.3 (30.0, 62.8)	48.4 (30.0, 78.9)	37.2 (28.2, 52.7)	0.352
Myocardial_infarct, n (%)						0.042
No	183 (74.4)	44 (81.5)	46 (67.6)	52 (83.9)	41 (66.1)
Yes	63 (25.6)	10 (18.5)	22 (32.4)	10 (16.1)	21 (33.9)
Congestive_heart_failure, n (%)						0.003
No	154 (62.6)	43 (79.6)	45 (66.2)	37 (59.7)	29 (46.8)
Yes	92 (37.4)	11 (20.4)	23 (33.8)	25 (40.3)	33 (53.2)
Cerebrovascular_disease, n (%)						0.708
No	204 (82.9)	43 (79.6)	59 (86.8)	52 (83.9)	50 (80.6)
Yes	42 (17.1)	11 (20.4)	9 (13.2)	10 (16.1)	12 (19.4)
Diabetes, n (%)						0.193
No	173 (70.3)	40 (74.1)	49 (72.1)	47 (75.8)	37 (59.7)
Yes	73 (29.7)	14 (25.9)	19 (27.9)	15 (24.2)	25 (40.3)
Renal_disease, n (%)						< 0.001
No	195 (79.3)	51 (94.4)	63 (92.6)	50 (80.6)	31 (50)
Yes	51 (20.7)	3 (5.6)	5 (7.4)	12 (19.4)	31 (50)
Malignant_cancer, n (%)						0.696
No	204 (82.9)	43 (79.6)	57 (83.8)	50 (80.6)	54 (87.1)
Yes	42 (17.1)	11 (20.4)	11 (16.2)	12 (19.4)	8 (12.9)

BMI: body mass index; BUN: blood urea nitrogen; INR: international normalized ratio; PT: prothrombin time; APTT: activated partial thromboplastin time; Q: quartile.

SI conversion factors: To convert glucose to mmol/L, multiply by 0.0555; calcium to mmol/L, multiply by 0.25; creatinine to μmol/L, multiply by 88.4; BUN to mmol/L, multiply by 0.357.

p-values were calculated using analysis of variance for continuous variables and χ² test for categorical variables.

Multivariate Cox regression analysis

The multivariate Cox regression analysis findings appear in Table 2. Model 1 had no adjustments. Model 2 accounted for age, gender, race, and BMI. Model 3 further considered myocardial infarction, congestive heart failure, cerebrovascular disease, diabetes, renal disease, and malignant cancer. BUN significantly correlated with all-cause mortality within the hospital (Model 3: HR 1.01, 95% CI 1.01–1.02, p = 0.001), within 30 days (Model 3: HR 1.01, 95% CI 1.01–1.02, p < 0.001), within 90 days (Model 3: HR 1.01, 95% CI 1.01–1.02, p < 0.001), and within 365 days (Model 3: HR 1.01, 95% CI 1.01–1.02, p < 0.001). Quartile analysis revealed that Q4 had the highest mortality risk compared to Q1 at all time points.

Table 2.

Multivariate COX regression model evaluating the association between BUN and mortality in patients with HIT.

Variable	No	Model 1		Model 2		Model 3
		HR (95%CI)	p-value	HR (95%CI)	p-value	HR (95%CI)	p-value
All-cause mortality within hospital
BUN	246	1.01 (1–1.02)	0.004	1.01 (1–1.02)	0.006	1.01 (1.01–1.02)	0.001
BUN
Q1	54	1(Ref)		1(Ref)		1(Ref)
Q2	68	1.34 (0.55–3.27)	0.525	1.41 (0.56–3.53)	0.461	1.5 (0.59–3.83)	0.392
Q3	62	1.24 (0.51–2.99)	0.634	1.32 (0.54–3.24)	0.542	1.47 (0.59–3.67)	0.406
Q4	62	2.68 (1.2–5.99)	0.016	2.9 (1.26–6.67)	0.012	3.96 (1.61–9.76)	0.003
Trend.test	246	1.39 (1.08–1.79)	0.011	1.41 (1.09–1.83)	0.009	1.55 (1.16–2.06)	0.003
All-cause mortality within 30 days
BUN	246	1.01 (1–1.01)	0.003	1.01 (1–1.01)	0.007	1.01 (1–1.02)	0.001
BUN
Q1	54	1(Ref)		1(Ref)		1(Ref)
Q2	68	1.51 (0.64–3.56)	0.347	1.58 (0.66–3.79)	0.303	1.81 (0.75–4.39)	0.187
Q3	62	1.71 (0.72–4.02)	0.222	1.66 (0.7–3.97)	0.250	1.86 (0.77–4.46)	0.167
Q4	62	3.07 (1.38–6.81)	0.006	3.07 (1.37–6.91)	0.007	4.79 (1.99–11.56)	<0.001
Trend.test	246	1.44 (1.13–1.82)	0.003	1.42 (1.12–1.81)	0.004	1.62 (1.23–2.13)	0.001
All-cause mortality within 90 days
BUN	246	1.01 (1–1.01)	<0.001	1.01 (1–1.01)	0.001	1.01 (1–1.02)	<0.001
BUN
Q1	54	1(Ref)		1(Ref)		1(Ref)
Q2	68	1.23 (0.59–2.56)	0.575	1.28 (0.61–2.7)	0.510	1.5 (0.7–3.18)	0.295
Q3	62	1.88 (0.94–3.77)	0.073	1.83 (0.9–3.68)	0.093	2.02 (0.99–4.12)	0.052
Q4	62	2.94 (1.52–5.7)	0.001	2.93 (1.5–5.73)	0.002	4.22 (2.03–8.81)	<0.001
Trend.test	246	1.47 (1.2–1.8)	<0.001	1.45 (1.18–1.78)	<0.001	1.61 (1.28–2.03)	<0.001
All-cause mortality within 365 days
BUN	246	1.01 (1.01–1.02)	<0.001	1.01 (1–1.01)	<0.001	1.01 (1.01–1.02)	<0.001
BUN
Q1	54	1(Ref)		1(Ref)		1(Ref)
Q2	68	1.03 (0.53–1.98)	0.935	1.04 (0.53–2.02)	0.919	1.23 (0.62–2.43)	0.552
Q3	62	1.71 (0.93–3.16)	0.087	1.6 (0.86–2.98)	0.140	1.84 (0.98–3.46)	0.058
Q4	62	3.03 (1.7–5.4)	<0.001	2.97 (1.65–5.33)	<0.001	4.26 (2.23–8.15)	<0.001
Trend.test	246	1.52 (1.27–1.83)	<0.001	1.51 (1.25–1.82)	<0.001	1.65 (1.34–2.04)	<0.001

Model 1: Unadjusted.

Model 2: Adjusted for age, gender, race, and BMI.

Model 3: Model 2 + myocardial infarction, congestive heart failure, cerebrovascular disease, diabetes, renal disease, and malignant cancer.

Kaplan–Meier survival analysis

The Kaplan–Meier survival curves for mortality are shown in Figure 2. Panels A–D illustrate the survival curves for hospital, 30-day, 90-day, and 365-day mortality, respectively. Patients in Q4 had consistently lower survival probabilities than those in Q1, with significant p-values (hospital: p = 0.022, 30-day: p = 0.014, 90-day: p = 0.0014, 365-day: p < 0.0001).

Figure 2.

Kaplan–Meier survival analysis curves for mortality Kaplan–Meier survival curves for mortality. Panels A–D show curves for hospital, 30-day, 90-day, and 365-day mortality, respectively.

Forest plot for subgroup analysis

The forest plot depicting the subgroup analysis of the relationship between mortality and BUN can be seen in Figure 3. Panels A–D present the subgroup analysis for hospital, 30-day, 90-day, and 365-day mortality, respectively. The findings suggest that the link between BUN and mortality remains consistent across various subgroups.

Figure 3.

Forest plot for the subgroup analysis of the relationship between mortality and BUN. Forest plots for subgroup analysis of mortality and BUN in pulmonary embolism patients. Panels A–D show curves for hospital, 30-day, 90-day, and 365-day mortality, respectively.

RCS curves

Figure S1 displays the RCS curves related to mortality. Panels A–D correspond to the curves for hospital, 30-day, 90-day, and 365-day mortality, respectively. These RCS curves reinforce the linear connection between BUN and mortality, indicating that higher BUN levels are linked to elevated hazard ratios.

Correlation matrix of baseline covariates

Figure S2 displays the correlation matrix of baseline covariates. The heat-map shows pairwise Pearson correlation coefficients among all variables included in the multivariable models. Overall, correlations were weak to modest (|r| < 0.30 for most pairs), indicating an absence of strong collinearity that would compromise model stability.

Discussion

We confirm that elevated BUN independently predicts death in ICU patients with HIT, corroborating prior critical-care data.^8,9,18,19 The association between BUN and mortality may be explained by several factors. First, BUN is a marker of renal function, and impaired kidney function is known to be associated with higher mortality rates in critically ill patients. ²⁰ Second, BUN levels can also reflect the overall metabolic state of the patient, including protein catabolism and fluid balance, which are often deranged in critically ill patients.^21,22

The mechanisms by which BUN levels influence mortality in patients with HIT may involve several pathways. Elevated BUN may reflect uremic toxicity, systemic inflammation, or sepsis—each amplifying endothelial injury and HIT severity.^23–25 Sepsis is a known complication in critically ill patients and can lead to disseminated intravascular coagulation, which is a serious concern in patients with HIT due to the risk of thrombosis. ²⁶ Another possible mechanism is the impact of BUN on coagulation. Elevated BUN levels have been associated with a prothrombotic state, which could potentiate the thrombotic complications of HIT. This is particularly relevant as patients with HIT are already at risk for thrombosis due to the formation of heparin-platelet factor 4 antibodies.^27,28 Furthermore, it has been postulated that elevated BUN might coincide with altered activity of enzymes involved in the coagulation cascade, potentially affecting the balance between procoagulant and anticoagulant factors in patients with HIT. ²⁹

Our results align with research that has identified a link between BUN levels and mortality in other critically ill groups. For example, Harazim et al. ³⁰ showed that BUN levels correlated with mortality in critically ill sepsis patients. Likewise, Seki et al. ¹³ found BUN to be a prognostic indicator in critically ill patients, regardless of renal function. These studies suggest that BUN may be a more general marker of illness severity rather than solely a marker of renal dysfunction. However, our study specifically focuses on patients with HIT, a population that may have unique pathophysiological characteristics and clinical management challenges. HIT is characterized by an immune-mediated reaction that increases the risk of thrombosis, and patients with HIT often require alternative anticoagulation strategies, which can further complicate their clinical course. ³¹ The association between BUN and mortality in this specific population highlights the importance of considering renal function and metabolic status in the management of HIT. BUN is an inexpensive, real-time warning signal; incorporating it into routine monitoring could flag HIT patients who need intensified care.

Although the prevalence of baseline renal disease rose from 5.6% in the lowest BUN quartile to 50% in the highest, the multivariate Cox model had already included “renal disease” as a binary covariate. After this adjustment, the hazard ratio for in-hospital mortality per 1 mg/dL increase in BUN fell only marginally, from 1.010 to 1.009, indicating that underlying kidney dysfunction explains only a small fraction of the observed association. To examine whether the prognostic value of BUN was confined to patients with pre-existing renal disease, we inspected stratum-specific hazard ratios. Among the 195 patients without baseline renal disease, the HR for Q4 versus Q1 was 3.15 (95% CI 1.22–8.14), almost identical to the overall estimate; in the 51 patients with renal disease, the HR was 3.88. Thus, BUN retained its predictive power even in HIT subjects with apparently normal baseline kidney function. We therefore believe that the BUN signal largely reflects non-renal processes—such as heightened catabolism, intravascular volume depletion, systemic inflammation, or thrombotic burden—rather than a mere reduction in glomerular filtration. Nevertheless, the absence of urine output, AKI staging, or novel kidney-injury biomarkers in MIMIC-IV means that residual confounding by subclinical acute kidney injury cannot be fully excluded; prospective studies incorporating these metrics are needed to quantify the relative contributions of renal and non-renal components.

Limitations

While our study has several strengths, it also has limitations that need to be recognized. First, HIT was defined only by ICD-10 codes; clinical probability, platelet kinetics, heparin details and anticoagulant switch were unavailable. While this mirrors all current MIMIC-IV HIT studies, prospective validation with 4Ts/serology is needed before BUN is adopted for risk-stratifying suspected HIT in the ICU. Second, being a retrospective study, we were limited by the information available in the MIMIC-IV database, which might not include all pertinent clinical information. Third, although multiple imputation is a strong statistical technique for handling missing data, it is based on assumptions about why the data is missing, and these assumptions may not always hold. Fourth, the exclusion of patients with ESRD limits generalizability to this high-risk subgroup, who are known to have increased HIT incidence and mortality. While this reduces confounding, it may also restrict external validity. Future studies should consider stratified analyses or propensity-score adjustment to include ESRD patients while controlling for renal dysfunction severity. Fifth, the MIMIC-IV database does not contain granular information on heparin type (unfractionated vs. low-molecular-weight), the exact day of HIT diagnosis in relation to heparin exposure, or the specific anticoagulant agent chosen after HIT was suspected (e.g. argatroban, bivalirudin, fondaparinux, DOACs). These unmeasured treatment-related factors could confound the observed BUN–mortality association; consequently, our findings should be interpreted as hypothesis-generating rather than definitively causal. Sixth, the study derives exclusively from a single U.S. academic ICU cohort; case mix, heparin-use patterns, and critical-care workflows may differ in other countries or healthcare systems, so external validation in multi-national registries is required before our results can be widely applied. We will create a 1:1 propensity-matched cohort (age, sex, BMI, SOFA, renal disease) of HIT and non-HIT ICU patients within MIMIC-IV, compare admission BUN and outcomes, and submit the results as a separate brief report. This upcoming analysis will clarify whether the BUN–mortality association is specific to HIT or reflects general critical-care risk. Lastly, the study did not consider potential confounders not found in the database, like the specific types of heparin used or how long patients were exposed to heparin. Future research should try to confirm our findings in prospective studies and in different patient groups. Also, more studies are needed to understand how BUN levels affect mortality in HIT patients. This could include looking at the link between BUN and specific bodily processes, like inflammation, blood clotting, or problems with the blood vessel lining. Moreover, research should explore if reducing BUN levels could help improve outcomes for HIT patients.

Conclusion

To sum up, our research identifies an independent association between elevated BUN levels and higher mortality rates among HIT patients in the ICU. These findings have important clinical implications for the management of HIT patients and hints that BUN levels provide prognostic information that is only partly attributable to underlying kidney dysfunction and may therefore be used as an early, inexpensive warning signal in this population. More studies are required to uncover the mechanisms behind this link and to assess the possible advantages of interventions aimed at BUN levels in this vulnerable population.