Optic nerve sheath diameter as a predictor of altered mental status in drug intoxication patients

2.1 Study design

This study was a single-centre, retrospective design using the electronic medical records (EMRs) of patients who presented in the ED with suspected DI patients who underwent brain CT. The study was approved by the Institutional Review Board, and the requirement for written informed consent was waived (IRB approval number: EUMC 2020-02-013).

2.2 Study setting and population

This study included patients who visited authors’ hospital emergency department in tertiary teaching hospital visits from January 2018 through December 2019. The inclusion criteria were (1) patients aged 18 years old or older who visited the ED complaining of DI (within 24 hours of exposure); (2) confirmed DI by laboratory results or patients’ statement of overdose during clinical process; (3) patients whose brain CT was performed simultaneously in the ED due to altered mentality.

The criteria of exclusion were (1) patients with an ophthalmologic disease, such as glaucoma, that could influence ONSD results; (2) patients with brain pathology such as stroke, hemorrhage, or brain tumor; (3) patients who had undergone brain surgery; (4) patients who had traumatic injury on head and (5) patients for whom there was insufficient EMR data.

2.3 Data collection and outcome measurement

The data was collected from EMRs that were stored in a picture archiving and communication system (PACS) (Infinitt PACS, Infinitt, Seoul, Republic of Korea). The data included patients’ demographic characteristics including sex; age; laboratory findings; clinical manifestation including initial vital sign, and clinical outcomes such as admission, hospital length and mortality.

2.4 Image acquisition and ONSD measurement

In the authors’ hospital, brain CT examinations was performed using a 320-slice multidetector CT scanner (Aquilion ONE Dynamic Volume CT, Toshiba Medical Systems Corporation, Otawara, Japan). Brain CTs were reconstructed with a series of 5 mm thicknesses parallel to the tuberculum sellae-occipital protuberance line.

All images were retrospectively and independently evaluated by two board-certified emergency physicians (9 years of experience and 7 years of experience) who were not involved in patient selection and who were blinded to the final diagnosis and outcome. They performed their evaluations in random order with respect to patient numeric code and radiographic image. ONSD was measured at a distance of 3 mm (ONSD3) behind the eyeball immediately below the sclera of a patient’s left and right eyes using axial planes of brain CT images obtained with PACS in stack mode (Fig. 1). The ONSDs obtained for a patient’s left and right eyes were averaged. Window parameters were spine window, middle third; window width: 60; window length: 360; accuracy: 1 pixel with 200% enlarged plane. All measurements were made using the same window, contrast, and brightness. Both reviewers agreed to measurement points and methods prior to image review.

Fig. 1.Use of axial non-contrast computed tomography to measure the optic nerve sheath diameter (ONSD). G, globe.

ONSD (solid line) was defined as the transverse diameter of 3 mm behind the globe (white point).

2.5 Statistical analysis

First, continuous variables are presented as means with SD (standard deviations) and ranges while categorical variables are presented as count (percent). We performed the statistical analysis which was evaluated normally distribution. Baseline clinical and demographic characteristics were summarized using the independent t-test was performed for continuous variables, and Pearson’s chi-squared test was performed for numerical variables.

Second, to evaluate the independent factors for the prognosis of DI patients by mortality and admission, multivariate logistic regression analysis was applied to determine independent factors of prediction of prognosis using data from reviewer 1 (senior emergency physician). Multivariate logistic regression analysis was conducted. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were generated from multivariate analyses. Third, a receiver operating characteristic (ROC) curve analysis [including cut-off values for optimal area under the curve (AUC), sensitivity, and specificity] was performed to identify diagnostic performance of ONSD for predictive power of prognosis of DI patients (mortality, admission). An optimal cut-off value was defined followed by Youden’s index [12]. Fourth, intraclass correlation coefficients (ICCs) with 95% CIs were calculated to determine agreement between reviewers. Significance was set at P $<$ 0.05. Statistical analyses were conducted using SPSS version 18.0 (SPSS Inc., Chicago, IL, USA).

Baseline characteristics

From January 2018 through December 2019, a total of 206 patients were finally enrolled in this study. The baseline characteristics are demonstrated in Table 1. The mean age of enrolled patients was 55.6 years and 56.1% of patients were female. The enrolled patients’ mental status was predominantly verbal response (55.7%) and the most common overdosed drug classes was sleeping drugs (44.3%) following by cardiovascular drug (18%). Among them, 18 (8.7%) were death during clinical process.

3.2 Clinical factors associated with the prognosis of DI patients

Association between death and clinical, laboratory variables are presented in Table 2. In the univariate analysis, body temperature, leukocyte, BUN, creatinine, ONSD, hospital length were significantly higher in patients who died in clinical progress than in surviving patients. In the multivariate logistic regression analysis, ONSD was the only independent variable for prediction of mortality among DI patients.

A comparison analysis between admission and clinical factors are presented in Table 3. Univariate analysis was presented that pulse rate, respiratory rate, decreased alertness, leukocyte, BUN, Creatinine, ONSD, mortality were significant higher in admission group than in discharge group. On the other hand, systolic blood pressure, diastolic blood pressure were significantly low in admission group. In multivariate logistic regression analysis, leukocyte, ONSD, in hospital death were considered independent variables for prediction of admission in ED.

In addition, a comparison analysis mortality and clinical, laboratory variables was conducted among the patients whose level of consciousness below stupor presented in Table 4. In univariate analysis, leukocyte, BUN and ONSD were significantly higher in patients who died in clinical progress and glucose was lower than in surviving patients. In the multivariate logistic regression analysis, glucose and ONSD showed independent variables for prediction of mortality among DI patients.

3.3 Predictive value of ONSD in DI patients

The ROC curves of significant factors for the prediction of death are shown in Fig. 1. The AUC of ONSD for the prediction of death was 0.824 (95% CI, 0.803–0.927) and the cut-off value was 4.35 (Fig. 2). When we set cut-off value 4.35, the sensitivity was 69.05% and the specificity was 53.33% for predictive power of ONSD for prediction death in DI patients (P = 0.001).

Fig. 2.Receiver-operating characteristics (ROC) curves for predicting death in DI patients. AUC, area under the ROC curve; CI, confidence interval; DI, drug intoxication.

The AUC of ONSD for the prediction of death among patients whose level of consciousness below stupor was 0.734 (95% CI, 0.532–0.939) and the cut-off value was 4.4. When we set cut-off value 4.4, the sensitivity was 90.9% and the specificity was 57.1% for predictive power of ONSD for prediction death in DI patients (P = 0.03).

Prediction of admission for DI patients in ED using ONSD are presented in Fig. 2. The AUC of ONSD for the prediction of death was 0.824 (95% CI, 0.722–0.927) and the cut-off value was 3.96 (Fig. 3). When we set cut-off value 3.96, the sensitivity was 100% and the specificity was 51.5% for predictive power of ONSD for prediction admission in DI patients (P = 0.001).

Fig. 3.Receiver-operating characteristics (ROC) curves for predicting admission in DI patients. AUC, area under the ROC curve; CI, confidence interval; DI, drug intoxication.