RU2199941C2 - Method for predicting results of severe craniocerebral trauma - Google Patents

Abstract

FIELD: medicine, clinical neurology, neurosurgery, neurotraumatology. SUBSTANCE: the method deals with coding prognostically valuable symptoms detected during neurological testing, calculating linear- discriminant functions corresponding to two variants in the results of craniocerebral trauma - favorable (LDF1) result and lethal result (LDF2) by the following formulas:

Description

 The invention relates to medicine, namely to clinical neurology (neuropathology), neurosurgery, neurotraumatology, and can be used to determine the prognosis of early outcomes of severe traumatic brain injury (BMI).

 A known method for predicting the outcome of PMI (Zotov Yu.V., Schedrenok VV Surgery of traumatic intracranial hematomas and centers of crushing of the brain. - L., Medicine, 1984. - S. 140-141), based on the unification of estimates of the severity of the injury, allowing to calculate the expected outcomes of treatment for contingents with different TBI structures.

где Е - исход,
f_i - относительная частота больных i-й группы по диагностической шкале или другой унифицированной шкале для оценки тяжести среди всех больных данного контингента,
i=1, 2,..., n - номер группы по шкале тяжести,
p_i - вероятность (относительная частота) исхода для больных i-й группы в генеральной совокупности.To determine the expected mortality (in%) of the total contingent, use the formula

where E is the outcome,
f _i - the relative frequency of patients of the i-th group on a diagnostic scale or other unified scale for assessing severity among all patients of this contingent,
i = 1, 2, ..., n - group number on the severity scale,
p _i - the probability (relative frequency) of the outcome for patients of the i-th group in the general population.

 Thus, the calculation of mortality by the known method has some drawbacks, in particular for the calculation it is necessary to additionally use the severity scale of the head injury; the proposed formula seems cumbersome and complex, because includes many derivatives that require preliminary calculations.

Индекс летальности означает, какое число больных с летальным исходом при данном виде травматического поражения мозга можно ожидать при поступлении 10000 пострадавших.A known method for predicting the outcome of PMTCT (Neurotraumatology. - Edited by Konovalov AN, Likhterman LB, Potapov AA - M .: CPI "Vazar-Ferro", 1994. S. 148), including the calculation " mortality index ", which is a derivative of the incidence and mortality for each type of head injury and calculated by the formula

The mortality index means the number of patients with a fatal outcome with this type of traumatic brain damage can be expected with the arrival of 10,000 victims.

 Similar calculations predict the outcome structure for surviving patients, i.e., determine how many patients with different degrees of disability or good recovery should be expected with different types of head injury for 10,000 victims. The disadvantage of this method is the impossibility of its application for an individual forecast.

A known method for predicting the outcome of PMTCT (Kondakov EN Diagnostics and differentiated tactics of surgical interventions for severe traumatic brain injury. Dis .... Doctor of medical sciences. - L., 1993. - 208 p.), Based on the analysis the prognostic value of 36 possible variants of the syndromes of the three most significant and always noted neurological signs: the background size of the pupils, the presence of pupil photoreaction, the characteristics of the motor reaction to the stimulus; at the same time, the option of combining these signs is evaluated twice - before the start of targeted treatment, as well as in the first 5 days of treatment (at the best indicator). The probability of an adverse outcome was calculated depending on the number of the syndrome. The prognostic significance of each syndrome is set in the "survival" points: the ratio of the number of "discharged" patients to the number of cases of this syndrome in the clinic before ("f1") and 1-5 days after ("f2") operations. The prognosis of outcomes, based on the quantitative assessment of neurological status, expressed in "outcome score" and the magnitude of the displacement of the anterior cerebral artery - ΔPMA - according to carotid angiography, is calculated by the formula
Z = score of the outcome + 0.5xΔPMA
A favorable treatment outcome is predicted at Z> 46, and a fatal outcome at Z <46.

 The known method for predicting the outcome of TMT has several drawbacks, in particular based on the choice of one of a large number of options (36 options), and is also associated with conducting an angiographic study of the victim, the implementation of which is not always possible due to the severity of the general condition of the victim, as well as due to the lack of appropriate equipment; the method cannot be applied upon admission to the admission department.

 A known method for predicting the outcome of PMTCT (Likhterman L.B., Kornienko V.N., Potapov A.A., Kuzmenko V.A., Gorbunov V.I., Troshin V.M. Craniocerebral injury: prognosis of the course and outcomes . - M .: Book LTD, 1993. - S. 195), adopted by us for the prototype, based on the analysis of the results of treatment of 201 victims with PMI. The authors use 26 clinical symptoms to determine their prognostic informational content and to identify the relationship between the signs characterizing the patient’s condition at the time of the operation and the final outcome — favorable and fatal. Using a computer, out of 26 signs, 13 of the most informative were selected, which included: age of patients, hospitalization, state of consciousness, vomiting, type of breathing, blood pressure, body temperature, upward gaze, nystagmus, pupil reaction to light, limb paresis, convulsive syndrome, stiff neck with the number of gradations of each sign from 3 to 8. All 13 of these signs are entered in a table where a prognostic factor is indicated against each gradation of each sign (the decision rule is based on a linear minimax algorithm ) After identifying the victim of all the signs listed in the table, their prognostic factors are summarized. The received amount (D) characterizes the forecast at the time of the operation. With D≥24 there is a risk of death, with D≤16 there is no risk of death; at 16 <D <24, the probability of death and survival is the same. Testing this algorithm on an examination sample showed that the real and predicted outcomes coincided in 75% of cases.

 The disadvantages of the prototype are: low accuracy of the forecast, a large number of signs, which must be assessed when forecasting, a large number of gradations of signs, which makes calculations time-consuming; the prognosis is determined before surgery, and not at the admission office upon admission.

 The invention is aimed at creating a method for predicting the outcome of PMTCT, which allows to increase the accuracy of prediction of early outcomes of PMTCT at the stage of patient admission to the in-patient department of the hospital, simplify the prognosis procedure, and reduce the time required for prognostic calculations.

For this, the graded values of the grades of the 5 most prognostically significant neurological symptoms selected using correlation and stepwise discriminant analyzes are entered into a table containing the coefficients and constants corresponding to each feature included in the model, and two linear discriminant functions are calculated, respectively, with two outcome options - favorable (LDF1) and lethal (LDF2) according to the formulas
LDF1 = 1.90 • X1 + (- 0.19) • X2 + 7.53 • X3 + 13.63 • X4 + 23.97 • X5 + (- 51.41)
LDF2 = 2.51 • X1 + 0.10 • X2 + 8.40 • X3 + 14.41 • X4 + 23.05 • X5 + (- 58.27)
The prognosis of the victim is determined by the greater absolute value of the linearly discriminant function.

 The inventive method differs from the prototype in that the calculated values of the two linear discriminant functions based on the following clinical signs: patient’s age, systolic blood pressure, consciousness, pupil innervation, oculocephalic reflex obtained during the initial neurological examination, the corresponding coefficients and constants, compare and choose the largest of them in absolute value, which determines the prognosis of outcome for this patient. The information ability (infallibility) of the statistically significant (p <0.001) discriminant model is 80%.

The method is as follows. During the initial clinical and neurological examination of a victim with severe traumatic brain injury, gradations of 5 prognostically significant signs are determined in the in-patient department of the hospital, each of which has a digital value:
1) Age:
- 15-24 years - 1 gradation
- 25-34 years - 2 gradation
- 35-44 years - 3 gradation
- 45-54 years - 4 gradation
- 55-64 years - 5 gradation
- 65-74 years - 6 gradation
- 75 years and older - 7th grade
2) Systolic blood pressure
- 110-140 mm Hg - 1 gradation
- 141-180 mm Hg - 2 gradation
- 90-109 mm Hg - 3 gradation
- 181 mmHg and more - 4 gradation
- 89 mmHg and less - 5 gradation
3) Level of consciousness:
- clear - 1 gradation
- light stunning - 2 gradation
- moderate stunning - 3 gradation
- sopor - 4 gradation
- coma I - 5th grade
- coma II - 6th grade
- coma III - 7th grade
4) Innervation of the pupils:
- innervation of the pupils is not broken - 1 gradation
- innervation of the pupils is broken - 2 gradation
5) Oculocephalic reflex:
- oculocephalic reflex is absent - 1 gradation
- oculocephalic reflex is caused - 2 gradation
The numerical values of the gradations of all the signs listed above are recorded in a particular victim and are entered in the table below (see Table 1), where there are prognostic factors and constants used to calculate LDF1 and LDF2, which correspond to favorable and lethal outcomes.

 The next stage, using a calculator or a personal computer, multiplies the digital gradation value of each sign by the prognostic coefficient corresponding to this sign for favorable and fatal outcomes and puts the data in the corresponding cells of the table. Then, in each column, these products are summarized separately, a constant is added to the sum, the final sums obtained are compared and the largest one, to which the particular patient belongs, is selected. In other words, the outcome is determined by the prevailing absolute value of one of the two linearly discriminant functions.

 The proposed method for predicting the outcome of severe traumatic brain injury was developed at the Russian National Scientific Research Institute of Cardiology. prof. A.L. Polenov and passed clinical trials in predicting outcomes of treatment of 21 patients. The results obtained in this case confirm the usefulness of the method: the real and predicted results coincided in 80.95% of cases.

 Example 1. Patient B., 55 years old, and / b 1722-99, was delivered to the institute after receiving a traumatic brain injury. Upon admission to the institute’s admission department, blood pressure was 160/80 mmHg, impaired consciousness was assessed as moderate stunning, anisocoria due to expansion of the right pupil was noted, the pupil’s reaction to light was reduced, and the oculocephalic reflex was caused. Based on the above signs, a forecast table was compiled below (Table 2); the calculation results indicated a favorable outcome, because LDF1> LDF2.

 After a standard examination of the patient, in particular computed tomography, where a subdural hematoma was detected in the area of the right frontal, temporal and parietal lobes, with a volume of 60 cubic meters. cm, which caused the dislocation of the ventricular system from right to left by 10 mm, the patient underwent decompressive trepanation of the skull in the right frontoparietal-temporal region, removal of the subdural hematoma and the center of crush of the pole of the temporal lobe. The postoperative period was uneventful, on the 20th day after the operation, the patient was discharged in a compensated state for outpatient treatment. Thus, the definition of the prognosis of treatment outcome of this patient, presented in the table, based on 5 clinical signs obtained during the initial examination of the victim in the admission department, completely coincided with the real result of treatment.

 Example 2. Patient D., 61 years old, and / b 1637-99, was delivered to the institute after receiving a traumatic brain injury. Upon admission to the ward, blood pressure was 90/60 mm RT. Art., impaired consciousness was evaluated as coma II, anisocoria was noted due to the expansion of the right pupil, the pupil’s reaction to light was reduced, the oculocephalic reflex was absent. Based on the above signs, a prognostic table was compiled below (Table 3); the calculation results indicated a high probability of death, because LDF1 <LDF2.

 After a standard examination of the patient, in particular computed tomography, where a subdural hematoma was detected in the left frontal, temporal and parietal lobes with a volume of 150 cubic meters. cm and dislocation of the ventricular system from left to right by 21 mm, the patient underwent decompressive trepanation of the skull in the left frontoparietal-temporal region with removal of the subdural hematoma. In the postoperative period, despite intensive therapy, the patient's condition progressively worsened and death occurred on the 4th day after the operation due to post-dislocation changes in the brain stem. Thus, the calculation of the forecast by the proposed method coincided with the real result of treatment.

The advantages of the proposed method compared to the prototype are:
1. Improving the accuracy of forecasting outcomes - 80.00% instead of 75% matches of the real outcome with the forecast:
a) through the use of a larger database (300 patients instead of 201 patients);
b) through the use of discriminant analysis, which allows to obtain a forecast in numerical form.

 2. The ability to build a forecast at the stage of admission to the emergency department, and not before the operation, as in the prototype, ie without data of additional instrumental research methods.

 3. The use of the minimum set of the most informative features is 5 instead of 13, which makes the calculations less time-consuming and reduces the duration of their implementation.

 4. The accuracy of the proposed forecast method to a lesser extent depends on the skill level of the doctor than in the prototype, because all 5 informative signs are primary and simple to identify in the in-patient department of the hospital.

 5. The possibility of self-learning system with an increase in the number of observations.

Claims (1)

A method for predicting the outcome of a severe traumatic brain injury, including digital coding of symptoms detected during a neurological examination, using coefficients obtained using the method of mathematical modeling, and calculating for this patient a digital value of the neurological status that determines the prognosis, characterized in that the linear discriminant functions corresponding to two outcomes of traumatic brain injury - favorable (LDF1) and lethal (LDF2) according to the formulas
LDF1 = 1.90 • X1 + (- 0.19) • (X2 + 7.53 • X3 + 13.63 • X4 + 23.97 • X5 + (- 51.41),
LDF2 = 2.51 • X1 + 0.10 • X2 + 8.40 • X3 + 14.41 • X4 + 23.05 • X5 + (- 58.27),
where X1 - age, determined - by 7 gradations;
X2 - systolic blood pressure - 5 gradations;
X3 - consciousness - in 7 gradations;
X4 - innervation of the pupils - 2 gradations;
X5 - oculocephalic reflex - in 2 gradations
and the greater absolute value of LDF1 determines the favorable, and LDF2 - the lethal outcome of severe traumatic brain injury.

Images