Raised Blood Urea Nitrogen–Creatinine Ratio as a Predictor of Mortality at 30 Days in Spontaneous Intracerebral Hemorrhage: An Experience from a Tertiary Care Center
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.355134
Source of Support: None, Conflict of Interest: None
Keywords: BUNR, dehydration, mortality, spontaneous intracerebral hemorrhage
Spontaneous intracerebral hemorrhage (SICH) is the second common cause of stroke and accounts for 7.5%−30% of all strokes., It generally carries higher morbidity and mortality compared to ischemic stroke.,, Raised blood urea nitrogen and creatinine ratio (BUNR) is a marker of dehydration, and it is related to poor outcome in stroke patients., The ratio varies from 15 to 80 in different studies as a marker of dehydration., We tried to study this ratio as predictor of mortality in the SICH population, the cutoff value of its prediction, and sensitivity and specificity of this ratio in the prediction model.
The study was a part of ongoing hospital-based stroke registry. The registry has the hard copy as well as the electronic records of all the patients who were admitted in the stroke ward of the Department of Neurology, Sir Sunderlal Hospital, Banaras Hindu University. The patients were prospectively followed up in the stroke clinic after discharge, and those who were unable to come for in-person visit were contacted on phone for the modified Rankin Scale (mRS) score. All the relevant clinical, biochemical, and radiological parameters of the patients were recorded.
Inclusion criteria: All patients above the age of 18 years who presented to the Emergency Department between January 1, 2018 and July 31, 2020 with computed tomography (CT) evidence showing SICH within 48 h of the onset were included in the study.
Exclusion criteria: Patients with posttraumatic hematomas, subarachnoid hemorrhage (SAH), arteriovenous malformation (AVM), aneurysms, primary brain malignancies, or requiring surgical interventions were excluded from the study.
Patients' blood samples were collected within 24 h of admission. Laboratory data including complete blood count, blood urea nitrogen, serum creatinine, liver function test, serum protein and albumin, serum triglycerides (TG), cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), and other relevant biochemical parameters were measured. BUNR measured at admission was considered for statistical analysis. Volume of hematoma was measured using the aXbXc/2 method. Patients were admitted and managed as per the current American Heart Association (AHA) guidelines. All patients were admitted in the stroke ward and were started on appropriate antihypertensive along with antiedema measures (wherever required). Antiepileptic drugs were administered in case of clinical suspicion or past history of seizure. Repeat CT was done in case of clinical deterioration. Surgical interventions were done in cases of supratentorial blood volume of more than 30 ml along with midline shift of more than 1 cm and posterior fossa SICH with a diameter of more than 3 cm. However, these patients were further excluded from the study as they were managed in neurosurgical care postoperatively.
Outcome measures: Outcome was determined at 30 days and categorized into two groups: survival and death. Independent predictors of mortality at 30 days were also determined.
Statistical analysis was performed with Statistical Package for Social Sciences (SPSS) version 24.0. Descriptive statistics including mean, median, and standard deviation were computed for baseline characteristics. Chi-square test was used to compare categorical variables, and Student's t-test and Mann–Whitney test were applied to calculate the P value for continuous variables for univariate statistics. Wilcoxon signed-rank test was used to compare mRS 6 at 30 days. Predictors of good outcome (mRS 0–5) and death (mRS 6) at 30 days were analyzed using logistic regression analysis. Variables with a P value less than 0.05 at univariate level were considered significant.
A total of 256 hemorrhagic stroke patients were enrolled in the study. Of these, 21 were excluded due to various reasons (10 were shifted to neurosurgical care for surgery after clinical deterioration, nine were SAH patients and two were AVM patients). The data of each predictor variable was further scrutinized for the possible presence of outliers. Out of the remaining observations, 18 were having a high leverage value and, therefore, were eventually dropped from further analysis. Of the remaining 217 patients, 71 patients did not survive during the initial 30 days. The mean age of the patients who survived (n = 146) was 59.75 ± 13.94 years in comparison to 61.11 ± 13.13 years of those who died (n = 71) because of the hemorrhagic stroke. Of all survivors, 96 were males and 50 were females. Baseline characteristics of the two groups are outlined in [Table 1].
The commonest risk factors were hypertension (n = 155, 71%), smoking (n = 83, 38%), and diabetes (n = 50, 23%). On comparing these risk factors between the two groups, only the presence of diabetes stood significant among those who died versus those who survived (24 vs. 26, P = 0.010).
Among the laboratory parameters, the mean level of urea was significantly lower among survivors than in those who died because of stroke (39.55 ± 20.46 vs. 52.60 ± 27.36, P = 0.001). Creatinine level was insignificantly lower among survivors than in those who died (1.10 ± 0.40 vs. 1.14 ± 0.55, P = 0.417). However, BUNR in the expired group was significantly higher than that of the survivors (47.45 ± 22.25 vs. 37.17 ± 18.04, P = 0.001).
The mean intracerebral hemorrhage (ICH) volume was significantly higher in the expired group than in those who survived (79.78 ± 59.73 vs. 33 ± 40.72 ml, P < 0.001). Intraventricular hemorrhage (IVH) was found in the initial CT of 135 (62%) patients and accompanying SAH in 28 (13%) patients. Mean midline shift was also higher in the expired group than in the survived one (8.6 ± 4.1 vs. 4.9 ± 2.9 mm, P = 0.021). Depth of the hematoma from the cortical surface was ≤10 mm in 75 patients (34.56%) and >10 mm in 142 patients (65.44%). Number of patients with IVH, mean ICH volume, and midline shift were higher in patients who died than in those who survived (P = 0.003, P < 0.001, and P = 0.021, respectively).
Among the clinical variables, there was a significantly lower Glasgow Comma Scale (GCS; dichotomized at 8) and higher ICH scores among the patients who died. The mortality rates observed with individual ICH scores were 0 (0%), 1 (8%), 2 (16%), 3 (70%), and 4 (100%). The mean ICH score among survivors (1.76 ± 0.64) was significantly lower than in their counterparts (3.00 ± 0.81) (P < 0.001). Poor GCS score (<9) was significantly more common in patients who did not survive (22.5% vs. 11%, P = 0.040). On univariate analysis, BUNR of more than 39.17, lower GCS score, higher National Institute of Health Stroke Scale (NIHSS) score, infratentorial location, higher ICH score, and larger hematoma volume were independent predictors of death at 30 days [Table 2].
[Table 2] represents independent predictors of mortality at 30 days of admission. For the analysis of the BUNR variable as a predictor of mortality, it was transformed into a new dichotomous variable where 1 and 0 denoted BUNR greater than and less than/equal to 39.17, respectively. This optimal value of BUNR for its bifurcation was selected using the Youden's method to find the optimal cutoff point that least misclassified the dependent dichotomous output, which is mortality of patients in the present study.
Before proceeding for the multivariable model that contained all the covariates that were significant in univariable analysis at the 25% level, correlation among the selected predictors was checked. [Table 3] presents the coefficient of linear correlation among the selected pair of variables. For ICH and NIHSS scores, the value of correlation index was nearly equal to 0.50. Since the ICH score is correlated with the NIHSS score and both the variables indicate severity. Hence, before proceeding for the multivariable logistics analysis, we will drop the variable NIHSS from our further analysis.
For collecting evidence of whether or not the BUNR was a predictor of mortality, two different multivariable logistic studies were performed. Model A represented the analysis without BUNR as a predictor of mortality, while model B included it [Table 4] and [Table 5].
To establish the credibility of the proposed hypothesis, checking of fitting of the model and verification of its predictive capability were performed. For the former, Akaike Information Criterion (AIC) score was calculated, while the receiver operating characteristic (ROC) curve was plotted using the predicted samples and the area under the curve was obtained as a measure of better predictive performance for both the models [Table 6] and [Table 7] and [Figure 1]. A model with a lower value of AIC was preferred over the one having larger AIC. From [Table 6], it is evident that under the light of given data, model B outperforms model A. From [Figure 1] and [Table 7], one can observe that the model having BUNR as a predictor outperforms the model without it, again validating the probable use of BUNR as a predictor of mortality. Hence, BUNR at the cutoff value of 39.17 was significantly associated with mortality at 30 days with a sensitivity and specificity of 61.97% and 62.33%, respectively (odds ratio [OR] 1.017).
This is the largest study describing the short-term outcome of SICH in the northern part of India. The case fatality rate was roughly 33%, which was high, though within the range of 12%–61% reported in various hospital-based studies., Consistent with the previous published literature, neither gender nor age was a significant outcome predictor. There was also insignificant association of comorbidities like hypertension, smoking, or alcohol abuse with the outcome. However, our study showed diabetes was an independent predictor of mortality at 30 days. This was consistent with the previous reports by Bakshayesh et al. and Tetri et al. Furthermore, we also confirmed that the initial poor neurological status, larger volume of ICH, and higher ICH score were strong predictors of death at 30 days.,,,
The location of the hemorrhage was also a significant risk factor for 30-day mortality. An infratentorial location correlated with high 30-day mortality in our study, which is consistent with the results from other studies. Previous studies, however, revealed conflicting results showing no significant influence in case of an infratentorial bleeding location. Apart from these, in our study, BUNR was significantly associated with mortality at 30 days at the cutoff value of 39.17 with a sensitivity and specificity of 61.97% and 62.33%, respectively, with an OR of 1.017. In most of the published studies, BUNR of more than 15, which was an indicator of dehydration, was associated with poor outcome mostly in ischemic stroke., Few studies looked at the ratio in stroke as a whole, with SICH as a subsection of the study population., However, none of the studies had looked at hemorrhagic stroke individually. Previous studies have reported different ratios for dehydration, varying from 15 to 80. However, our study concludes that BUNR of 39.17 was significantly associated with mortality. Whether this cutoff signifies severe dehydration in our cohort is hard to comment upon and this has to be validated with the clinical features of dehydration in future studies.
Our study has a few strengths. First, this is the first study to look into the BUNR as a predictor of mortality at 30 days in SICH patients. This was a prospective study, and the population was a homogenous cohort of SICH.
The present study had several limitations also. First, all the patients undergoing surgical interventions were excluded from the analysis, which could have biased the results. Secondly, the long-term effect of raised BUNR has to be proven in future studies. Thirdly, the specificity of the ratio is low due to several factors and is a limitation of the model.
This prospective, hospital-based study shows that ICH is associated with significant mortality. Independent predictors of death at 30 days were low GCS on admission, larger hematoma volume, and BUNR of more than 39.17. However, future studies with a larger sample size are needed to confirm the findings of this study and may form the basis of corrective measures of this ratio, which might lead to decreased mortality in the SICH population.
BUNR- blood urea nitrogen creatinine ratio
SICH- symptomatic intra cerebral hemorrhage
SBP- systolic blood pressure
DBP- diastolic blood pressure
NIHSS- National Institute of Health Stroke Scale
mRS- modified Rankin Scale
LDL- low-density lipoprotein
HDL- high-density lipoprotein
VLDL- very low-density lipoprotein
RBS- random blood sugar
ICH- intracerebral hemorrhage
GCS- Glasgow Comma Scale
AST- aspartate aminotransferase
ALT- alanine aminotransferase
Ethical approval and consent to participate
This study was approved by the Institutional Ethics Committee, IMS, BHU, Varanasi, India. Consent for participation has been taken from all the patients/people included in the study.
Availability of supporting data
Raw data is with the first author and the corresponding author, which can be availed on request. Supporting literature for our study is present in references.
We acknowledge the patients' relatives for giving consent for participating in the study.
Abhishek Pathak- study design, statistical analysis, and manuscript writing; Priya Dev- data collection, statistical analysis, and manuscript writing; Varun Kumar Singh- data collection, manuscript writing; Ami Kumar- statistical analysis and manuscript editing; Anand Kumar- data collection, manuscript writing; Vijaya Nath Mishra- manuscript writing and study supervision; Deepika Joshi- manuscript writing and study supervision; Rameshwar Nath Chaurasia- manuscript writing and study supervision
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6], [Table 7]