Risk factor analysis and nomogram model establishment for in-hospital death of elderly patients with surgically treated traumatic acute subdural hematoma

Abstract

The purpose of this study was to investigate the risk factors for in-hospital death in elderly patients with surgically treated traumatic acute subdural hematoma (ASDH) and to construct a nomogram model for in-hospital death risk prediction. We analyzed 104 elderly patients who underwent decompressive craniotomy (DC) for isolated traumatic ASDH between May 2013 and May 2021 in our department. Independent factors for in-hospital death were identified via univariate and multivariate logistic regression analyses, and a nomogram model was constructed and validated. The overall in-hospital mortality rate was 58.7% (61/104). Sex (odds ratio [OR] = 11.94), dilated pupils (bilateral vs absent; OR = 194.16) and subarachnoid hemorrhage (SAH) (OR = 16.01) were independent risk factors for in-hospital death. A nomogram model for in-hospital death risk prediction was constructed based on these variables. Statistical analysis revealed that this model had good predictive performance, and receiver operating characteristic (ROC) curve analysis revealed that the area under the curve (AUC) of the nomogram model was 0.916 (95% confidence interval [CI] = 0.861–0.970). The calibration plot, which compared the predicted effective rate and observed effective rate, revealed the accurate predictive ability of the model. The model was internally validated via 200 bootstrap samples to calculate the discrimination accuracy, and the concordance index was 0.891. Decision curve analysis (DCA) demonstrated that the nomogram was clinically beneficial. The most important risk factors for in-hospital death in elderly patients with surgically treated traumatic ASDH were sex, dilated pupils, and SAH. The nomogram constructed from these data could be a promising and convenient tool to predict in-hospital death risk, but further external validation is needed.

1. Introduction

Acute subdural hematoma (ASDH) is a common and serious injury in patients with traumatic brain injury (TBI). Most traumatic ASDHs are in the supratentorial region and are often accompanied by other intracranial injuries. The prognosis of traumatic ASDH is usually poor, and the mortality rate is as high as 24% to 60%.^[1,2] To minimize secondary brain damage, decompressive craniotomy (DC) is recommended when patients have a decreased level of consciousness, a greater hematoma size, a midline shift, or basal cistern obliteration.^[3,4] In recent years, birth rates have declined in many countries, leading to an aging population. Census data estimate that the 65-year-old or older population will increase from 14.9% in 2015 to 22.1% by 2050.^[5] Several studies have shown that advanced age is a risk factor for mortality in surgically treated patients with ASDH, and few patients older than 65 years who undergo surgery are able to regain daily self-care skills, which leads to serious public health, social, and economic concerns.^[6,7] Therefore, the topic of surgical functional outcome and quality of life in this population has become increasingly important. The surgical management of traumatic ASDH in the elderly population is also a controversial topic. According to preoperative clinical and imaging data, predicting the operative mortality of elderly patients with ASDH can help neurosurgeons accurately judge surgical risk and benefits and can provide reasonable advice to patients’ families. Currently, there is no simple and accurate tool to predict in-hospital death of elderly patients with surgically treated traumatic ASDH.

Nomograms are based on multivariate regression analysis, integrating multiple predictive indicators, visualizing complex equations, and making the results more readable. They have been widely used in assessing and calculating the precise risk of both short-term and long-term outcomes in individual patients. This retrospective study investigated the risk factors associated with in-hospital death in elderly patients with surgically treated traumatic ASDH and constructed a nomogram model for in-hospital death risk prediction. We expect that this model can provide useful guidance for neurosurgeons to assess surgical risk and benefits.

2. Methods

2.1. Patient population

This retrospective study was performed on patients older than 65 years of age with traumatic ASDH who underwent DC at our neurosurgical department between May 2013 and May 2021. The inclusion criteria were as follows: ASDH located on the convex surface of the brain; emergency DC performed immediately after admission; and age ≥ 65 years. The exclusion criteria were as follows: patients with penetrating head injuries, acute epidural hematomas, ASDH of the posterior fossa, or intracerebral hematomas; patients with multiple concomitant chest and abdominal injuries or shock; patients receiving long-term anticoagulant or antiplatelet therapy; and patients who had severe diseases of the lungs, liver, kidneys, or hematologic system. Preoperative clinical and imaging data that might be related to in-hospital death were investigated. The study was approved by the Ethics Committee of the Third Affiliated Hospital of Soochow University.

2.2. Data collection

We collected the following clinical information from all patients: age, sex, mechanism of injury, Glasgow Coma Scale (GCS) score, pupil reaction and time from injury to surgery. Preoperative head computed tomography (CT) parameters, such as the site of hematoma, maximum thickness of hematoma, midline shift, subarachnoid hemorrhage (SAH) and status of the basal cistern were also documented. The condition of the basal cistern was evaluated via axial images at the level of the midbrain, which was then divided into 3 branches (1 posterior and 2 lateral) (Fig. 1). Each branch was assessed separately to determine whether it was open or occluded.

Figure 1.

Evaluation of the basal cisterns via computed tomography.

2.3. Statistical analysis

Statistical analyses of the data were carried out using SPSS 22.0 software (IBM Corp., Armonk, New York, USA). Discrete variables were compared using the chi-square test, and continuous variables were compared using the Mann–Whitney U test. All variables were used as candidates and included in a multivariate logistic regression model to identify which variables were independently associated with in-hospital death in elderly patients with surgically treated traumatic ASDH, and all variables with P < .05 were considered statistically significant.

The nomogram model was established with the package “rms” in R software (R version 4.2.3, R Foundation for Statistical Computing, Vienna, Austria). To verify the performance of the nomogram, the area under the curve (AUC) of the receiver operating characteristic (ROC) curve was calculated. In general, an AUC > 0.70 was considered relatively superior for discrimination. Calibration of the nomogram was assessed by plotting the probabilities of the observed outcome and those predicted by the logistic model. The prediction of a well-calibrated model should be mirrored by a 45° diagonal line. The bootstrap resampling method (200 times) was used for internal validation to evaluate the model’s accuracy. Furthermore, decision curve analysis (DCA) was performed to evaluate the clinical benefit and applicability of the generated nomogram.

3. Results

3.1. Patient characteristics

A total of 104 patients were included in the final analysis. The overall in-hospital mortality was 58.7%. The median age of the patients was 71 years (range 65–86 years). Eleven variables contributing to in-hospital death in elderly patients with surgically treated traumatic ASDH were analyzed separately via univariate analyses. Consequently, 7 factors were found to be significantly related to in-hospital death (Table 1).

Table 1.

Variables related to in-hospital death of elderly patients with surgically treated traumatic acute subdural hematoma (n = 104).

Variables	Alive	Exitus	P value
Sex
Male	16	46 (74.2%)	<.001
Female	27	15 (35.7%)
Age (yr)	70	73
Mechanism of injury
Motor vehicle accident	28	42 (60.0%)	.183*
Fall	15	19 (55.9%)	.832
GCS
3 to 5	19	57 (75.0%)	<.001
6 to 8	18	3 (14.3%)
9 to 15	6	1 (14.3%)
Dilated pupils
Absent	19	1 (5.0%)	<.001
Unilateral	12	9 (42.9%)
Bilateral	12	51 (81%)
Site of hematoma
Left	20	30 (60.0%)	.789
Right	23	31 (57.4%)
Thickness of hematoma (mm)
<10	13	6 (31.6%)	.008
≥15	30	55 (64.7%)
Midline shift (mm)
<10	8	6 (42.9%)	.197
≥15	35	55 (61.1%)
SAH
No	33	20 (37.7%)	<.001
Yes	10	41 (80.4%)
Condition of the basal cistern
0	6	2 (25.0%)	<.001
1	10	3 (23.1%)
2	16	12 (42.9%)
3	11	44 (80.0%)
The time from injury to surgery (h)
<4	20	41 (67.2%)	.035
≥4	23	20 (46.5%)

Abbreviations: GCS = Glasgow coma scale, SAH = subarachnoid hemorrhage.

^*Mann–Whitney U test.

The male mortality rate was significantly greater than the female mortality rate (P < .05). Interestingly, this has not been observed in previous studies. We subsequently analyzed the distributions of the GCS score, dilated pupils, hematoma thickness, SAH and condition of the basal cistern in the patients according to sex. Greater proportions of male vs female patients had GCS scores ranging from 3 to 5, bilaterally dilated pupils, hematoma thicknesses ≥ 10 mm, 3 branches of the basal cistern occluded and SAH. This finding indicated that the degree of injury in males was greater than that in females in our study, which led to an increase in the mortality rate. The distribution of the preoperative risk factors according to sex is shown in Table 2.

Table 2.

Distribution of preoperative risk factors in patients based on sex.

	Female	Male
Preoperative GCS score
3 to 5	27	49 (64.5%)
Dilated pupils
Bilateral	22	40 (65.1%)
Thickness of hematoma (mm)
≥10	32	53 (62.4%)
Condition of the basal cistern
3	20	35 (63.6%)
SAH
Yes	19	32 (62.7%)

Abbreviations: GCS = Glasgow coma scale, SAH = subarachnoid hemorrhage.

3.2. Development and validation of the nomogram

Multivariate logistic regression analysis revealed that sex, dilated pupils and SAH were independent risk factors for in-hospital death. The odds ratios (ORs) of these 3 variables and the regression formulas are summarized in Table 3. The nomogram model was developed by assigning a graphic initial score to each of the 3 independent risk factors with a point range from 0 to 100, which was then summed to create a total score, and a vertical line was drawn from the total-point row to indicate the estimated probability of in-hospital death (Fig. 2). It was predicted that a higher total score in the nomogram was associated with a greater probability of in-hospital death. The probability ranged from 0.01 to 0.99.

Table 3.

Independent predictors of the risk factors for in-hospital death of elderly patients with surgically treated traumatic acute subdural hematoma by multivariate logistic analysis.

Factors	OR (95% CI)	P value
Sex
Female	1.00	.001
Male	11.94 (2.88–49.48)
Dilated pupils
Absent	1.00
Unilateral	16.40 (1.26–214.06)	.033
Bilateral	194.16 (14.79–2549.50)	<.001
SAH
No	1.00	<.001
Yes	16.01 (3.79–67.66)

Abbreviations: CI = confidence interval, OR = odds ratio, SAH = subarachnoid hemorrhage.

Figure 2.

Nomogram to determine the probability of in-hospital death from surgically treated traumatic ASDH in elderly patients. ASDH = acute subdural hematoma.

The good predictive performance of the nomogram was confirmed, and the area under the ROC curve was 0.916 (95% CI = 0.861–0.970) (Fig. 3). As shown in Figure 4, the calibration plot, which compares the predicted effective rate and observed effective rate, revealed the accurate predictive ability of the model. The model was internally validated via 200 bootstrap samples to calculate the discrimination accuracy, and the concordance index was 0.891. DCA was employed to evaluate the clinical validity of this model. A wide and practical range of threshold probabilities could be observed in the validation cohort, which indicated that the model performed well for predicting the probability of in-hospital death (Fig. 5).

Figure 3.

Receiver operating characteristic curve of the nomogram.

Figure 4.

Calibration curve for the nomogram.

Figure 5.

Decision curve analyses of the nomogram.

4. Discussion

Traumatic ASDH is the most common condition in patients with severe head trauma. It is associated with high mortality and disability rates, which lead to serious public health, social and economic problems. The surgical management of ASDH in elderly patients remains a controversial topic. In this population, patients who underwent surgery had a relatively high mortality rate, which is consistent with our experience. We observed an overall in-hospital mortality rate of 58.7% in 104 patients. There are limited data concerning prediction models for the in-hospital death risk of elderly patients with surgically treated traumatic ASDH. Based on preoperative clinical and imaging data, we created a practical nomogram based on sex, dilated pupils, and SAH to predict the in-hospital death risk of elderly patients with surgically treated ASDH. Furthermore, the nomogram was assessed and showed satisfactory discrimination, calibration, and clinical utility, indicating good performance.

In this study, multivariate logistic regression analysis revealed that sex was an independent risk factor for in-hospital death, and the in-hospital mortality of male patients was greater than that of female patients, which was not mentioned in previous studies. There are data reporting poorer outcomes in females surviving severe TBI than in males.^[8] However, studies have suggested that male sex is associated with poor outcomes in TBI patients.^[9] We further analyzed the distribution of other risk factors across different sexes and reported that male patients were more seriously injured, leading to higher mortality in our study, which has also been mentioned by other authors.^[10] The cause of the more serious injury in male patients is a focus of future research.

Elderly patients often have multiple chronic comorbidities, and age is considered an important prognostic factor concerning the outcome of TBI patients.^[10,11] However, other authors have suggested that age has no effect on outcomes.^[9] Currently, most studies on ASDH in elderly patients do not address increasing age as a predictor of mortality.^[12] In our case series, we did not divide patients into subgroups, but the median age of patients who died was slightly greater than the average age of patients who survived, although there was no significant difference between them.

Bilateral dilated pupils were an independent risk factor for in-hospital death in this study. Dilated pupils and lower GCS scores were found to be significantly associated with increased mortality and poor patient outcomes.^[13] The presence of dilated pupils in TBI patients not only indicates that the patient has developed brain herniation but also a poorer functional outcome. Some scholars believe that the mortality of patients with bilateral dilated pupils can reach 64% to 88.1%.^[14] The GCS score is the most important factor that directly reflects brain damage and clinical status.^[15] A preoperative GCS score lower than 5 indicates that the patient has a serious primary brain injury. The GCS score was an important risk factor for in-hospital death in this study. According to previous reports, the mortality of patients with a GCS score of 3 and bilateral dilated pupils is 100%.^[16] An analysis of prognostic factors in patients with ASDH revealed that the GCS score and the pupil response are the most powerful independent prognostic factors after TBI.^[17,18]

Patients with traumatic SAH have higher mortality than patients without SAH. SAH was also an independent risk factor for in-hospital death in the present study. Several studies have reported that SAH is an independent predictor for TBI mortality.^[19,20] Other studies have also identified traumatic SAH as an unfavorable prognostic factor.^[21] Blood cells in the subarachnoid space can release substances that cause vasospasm, which can then cause intracranial hypertension, cerebral edema, and even brain herniation.^[22] Some authors have described an association between SAH and an increased incidence of cerebral vasospasm and a higher probability and greater severity in severe TBI patients.^[23] In addition, a recent study revealed that traumatic SAH was an independent predictor of posttraumatic cerebral infarction.^[24]

The thickness of the hematoma was associated with in-hospital death in the univariate analysis, although it fell short of significance in the multivariate analysis. The greater the thickness of the hematoma was, the higher the mortality rate of patients. The thickness of the hematoma and midline shift play important prognostic roles and may be decisive factors for surgical management.^[25] Some studies have shown that a difference between the thickness of the hematoma and a midline shift ≥ 3 mm at the initial CT can predict mortality in ASDH patients.^[26] The condition of the basal cistern is strongly associated with death. Basal cistern occlusion indicates compression of the brainstem by supratentorial brain tissue and is an important prognostic factor associated with mortality and poor functional outcome.^[27] A previous study revealed that patients with absent or compressed basal cisterns with a midline shift >5 mm had a mortality rate of 44%.^[14]

It seems that the association of the time from injury to surgery with mortality was the opposite of what we expected.^[28,29] The mortality rate of patients whose time from injury to surgery was <4 hours was higher than that of patients whose time from injury to surgery was more than 4 hours in the univariate analyses (67.2% vs 46.5%). This finding creates the illusion that a longer time from injury to surgery is a protective factor against death. We further analyzed the relationships between the time from injury to surgery and the GCS score, pupil reaction and condition of the basal cistern and found that the median time from injury to surgery was shorter in patients with GCS scores ranging from 3 to 5, bilateral pupil dilation, and 3 branched occlusions, indicating that patients with shorter times from injury to surgery presented more severe disease. The likely reason for this is that patients who appear to be sicker and have lower GCS scores are sent more quickly to an advanced neurosurgical trauma center for emergency surgery. This selection bias clearly skews the results, necessitating more research for verification.

In the present study, we integrated and internally validated a new nomogram model combining preoperative clinical and imaging data. The nomogram can predict the in-hospital death risk of elderly patients with surgically treated ASDH with high accuracy. Our novel nomogram model may provide a promising and convenient tool for predicting the in-hospital death risk of elderly patients with surgically treated ASDH and can provide reasonable advice to patients’ families. This study has several limitations. First, we included a limited number of cases, which may have affected the credibility of the results. Despite enrolling all eligible patients in our center over the past decade, the incidence rate of isolated traumatic ASDH in elderly patients undergoing emergency surgery is relatively low. An additional concern is further deviation of the data with an extended collection. Moreover, the statistical analysis revealed that this model had good predictive performance and could aid decision making in clinics and hospitals. Second, on the basis of the previous literature, we collected possible factors that could affect patient mortality and divided them into groups. Owing to the limited number of patients, we considered further grouping or the addition of possible confounding factors, which may have introduced greater bias in the results. In future research, we will increase the number of patients to perform additional analyses to explore potential confounding variables. Third, our model included preoperative clinical and imaging data, but the lack of therapeutic data may have resulted in systemic bias that weakened the performance of the nomogram. Fourth, this was a single-center retrospective study. The selection bias caused by single-center data may lead to a lack of broad representation of the results. It is necessary to conduct external validation analyses in other institutions, and prospective multicentre studies are required to validate these findings. Despite these limitations, this is the first attempt to establish and validate a nomogram model to predict the in-hospital death risk of elderly patients with surgically treated ASDH.

5. Conclusion

Our findings revealed a high proportion of in-hospital mortality in elderly patients who were surgically treated for traumatic ASDH. The most important factors affecting the in-hospital rate of surgically treated traumatic ASDH in elderly patients in this study were sex, dilated pupils, and SAH. The nomogram constructed from these data could be a promising and convenient tool to predict in-hospital death risk. In addition, more prospective multicentre studies are needed to confirm our findings.

Author contributions

Conceptualization: Chunbo Liu, Geng Jia.

Investigation: Chunbo Liu, Jia Shi.

Supervision: Huaping Qin.

Writing – original draft: Ruhong Wu.

Writing – review & editing: Ruhong Wu, Chunbo Liu, Jia Shi, Geng Jia, Huaping Qin.