RU2765744C2 - Method for determining the likelihood of a lethal outcome based on complaints from a tuberculosis patient with hiv infection upon admission to the hospital of the federal penitentiary service of the russian federation - Google Patents

Abstract

FIELD: medicine.

SUBSTANCE: invention relates to medicine, namely to infectious diseases and phthisiology, and can be used to identify the likelihood of death in a tuberculosis patient with HIV infection. The presence or absence of the following clinical manifestations is determined: complaints of chest pain, headaches, loose stools, absence of cough with phlegm, dry cough, acute onset of the disease, shortness of breath, chills, sweating, weight loss, lack of weakness, loss of appetite, fever. They are evaluated in points. Then, according to the stated formula, the probability of death in a patient with tuberculosis with HIV infection is determined. If the probability value is 50% or higher, the risk of death and the need for hospitalization of the patient in the intensive care unit are predicted. If the probability is below 50%, the patient is admitted to the hospital ward.

EFFECT: method provides for determining the probability of death in a tuberculosis patient with HIV infection upon admission to the hospital of the Federal Penitentiary Service of the Russian Federation (FSIN) by assessing the complex of the most significant clinical signs.

1 cl, 1 dwg, 2 tbl, 1 ex

Description

The invention relates to medicine, namely to infectious diseases and phthisiology.

When tuberculosis is detected in the late stages of HIV infection, there is a high risk of death due to an atypical course, "masks" of other infectious diseases, the appearance of complications (respiratory failure, infectious-toxic shock, hyperthermia, etc.) and / or the addition of secondary infections.

At present, there is a lack of simple, cheap and effective methods to identify a high risk of early mortality in a TB patient with HIV infection on admission to the hospital.

Identification of factors that statistically significantly affect the lethal outcome in a TB patient with HIV infection upon admission to a hospital allows hospitalization in the intensive care unit for round-the-clock observation and treatment.

There are various ways to detect.

Thus, the unfavorable outcome, according to the authors, in the treatment of tuberculosis is significantly associated with the following factors: age over 59 years, relapse of tuberculosis and co-infection tuberculosis + HIV infection (see source: Nanzaluka F. H., Chibuye S., Kasapo C. C. ., Langa N., Nyimbili S., Moonga G., Kapata N., Kumar R., Chongwe G. Factors associated with unfavorable tuberculosis treatment outcomes in Lusaka, Zambia, 2015: a secondary analysis of routine surveillance data // Pan Afr Med J. 2019 Apr 8; 32:159). The disadvantage of this method is that co-infection with tuberculosis + HIV infection itself is a risk factor.

In the following method, risk factors for adverse outcomes are: age, female sex, extrapulmonary tuberculosis, weight loss, HIV infection and duration of illness (see source: Abdullahi O. A., Ngari M. M., Sanga D., Katana G. , Willetts A. Mortality during treatment for tuberculosis; a review of surveillance data in a rural county in Kenya // PLoS One. 2019 Jul 11; 14(7): e0219191). The disadvantage of this method is that co-infection with tuberculosis + HIV infection itself is a risk factor.

In another method, the use of ART with a high risk of side effects and the use of rifampicin-incompatible antiviral drugs (ART) are considered risk factors for adverse outcomes (see source: Demitto F. O., Schmaltz C. AS, Sant'Anna F. M., Arriaga M. V., Andrade V. V., Rolla V. C. Predictors of early mortality and effectiveness of antiretroviral therapy in TB-HIV patients from Brazil // PLoS One.2019 Jun 6;14(6):e0217014). The disadvantage of this method is the lack of use of clinical signs to assess the risk of adverse outcomes.

In the following method, risk factors for poor outcomes are considered to be: age over 60 years, high hospital fees, HIV infection, disease recurrence or withdrawal from previous treatment (see source: Rocha M. S., Oliveira GP, Saraceni V. , Aguiar FP, Coeli C.M., Pinheiro RS Effect of inpatient and outpatient care on treatment outcome in tuberculosis: a cohort study // Rev Panam Salud Publica 2018 Sep 7; 42:e112). The disadvantage of this method is that co-infection with tuberculosis + HIV infection itself is a risk factor.

In another method, a risk factor for an unfavorable outcome is considered to be: extensive drug resistance of Mycobacterium tuberculosis (see source: Frank M., Adamashvili N., Lomtadze N., Kokhreidze E., Avaliani Z., Kempker RR, Blumberg HM Long-term Follow-up Reveals High Posttreatment Mortality Rate Among Patients With Extensively Drug-Resistant Tuberculosis in the Country of Georgia // Open Forum Infect Dis. 2019 Mar 29;6(4):ofz152). The disadvantage of this method is the lack of use of clinical signs to assess the risk of adverse outcomes.

In the following method, the following risk factors for an unfavorable outcome in a prisoner with tuberculosis associated with HIV infection are considered: age, presence and duration of education, place of birth, re-offending, relapse of tuberculosis (see source: Rossetto M., Brand É. M., Teixeira L. B. Factors associated with hospitalization and death among TB/HIV co-infected persons in Porto Alegre, Brazil // PLoS One.2019 Jan 2;14(1):e0209174). The disadvantage of this method is the lack of use of only clinical signs to assess the risk of an unfavorable outcome.

In another method, the detection of lipoarabinomannan in the urine is considered a risk factor for an unfavorable outcome (see source: Gupta-Wright A., Corbett E. L., Wilson D., van Oosterhout JJ, Dheda K., Huerga H., Peter J., Bonnet M., Alufandika-Moyo M., Grint D., Lawn SD, Fielding K. Risk score for predicting mortality including urine lipoarabinomannan detection in hospital inpatients with HIV-associated tuberculosis in sub-Saharan Africa: Derivation and external validation cohort study // PLoS Med. 2019 Apr 5;16(4):e1002776). The disadvantage of this method is high financial costs.

In the following method, the risk factors for an unfavorable outcome of a patient with tuberculosis associated with HIV infection are: age over 54 years, drug addiction, and a low number of CD ₄ lymphocytes less than 50 cells per μl (see source: Chan C.K., Wong K. H, Lee M.P., Chan K.C, Leung C.C, Leung E.C, Chan WK, Mak IK Risk factors associated with 1-year mortality among patients with HIV-associated tuberculosis in areas with intermediate tuberculosis burden and low HIV prevalence // Hong Kong Med J. 2018 Oct;24(5):473-483). The disadvantage of this method is the lack of use of only clinical signs to assess the risk of an unfavorable outcome.

In the following method, the risk factors for an unfavorable outcome of a patient with tuberculosis associated with HIV infection are: the presence of concomitant pathology, high age, late hospitalization (see source: Gamboa-Acuna B., Guillen-Zambrano R., Lizzetti-Mendoza G., Soto A., Lucchetti-Rodriguez A. Factors associated to survival in patients with HIV-TB in the Department of Infectious Diseases of the Arzobispo Loayza National Hospital, Peru, since 2004 to 2012 // Rev Chilena Infectol.2018;35(1): 41-48). The disadvantage of this method is the lack of use of only clinical signs to assess the risk of an unfavorable outcome.

In another method, only low weight is considered a risk factor for poor outcome in a patient with tuberculosis associated with HIV infection (see source: low body mass index) (Naidoo K., Yende-Zuma N., Augustine S. A retrospective cohort study of body mass index and survival in HIV infected patients with and without TB co-infection // Infect Dis Poverty, 2018 Apr 25; 7(1):35). The disadvantage of this method is the use of only one clinical sign to assess the risk of an unfavorable outcome.

In the following method, the risk factors for an unfavorable outcome of a patient with tuberculosis are: age over 60 years, combination with HIV infection, extensive MBT drug resistance, previous refusals of treatment, drug use, drug resistance to ethambutol or streptomycin (see source: Gayoso R., Dalcolmo M., Braga JU, Barreira D. Predictors of mortality in multidrug-resistant tuberculosis patients from Brazilian reference centers, 2005 to 2012 // Braz J Infect Dis. 2018 Jul - Aug; 22(4):305-310). The disadvantage of this method is the lack of use of clinical signs to assess the risk of adverse outcomes and co-infection with tuberculosis + HIV infection itself is a risk factor.

In another method, the risk factors for an unfavorable outcome of a patient with tuberculosis associated with HIV infection are considered to be an increase in the blood serum of the following biomarkers: C-reactive protein, γ-interferon, MCP-3 factor, interleukin-15 and interleukin-17 (see source: Sattler FR, Chelliah D., Wu X., Sanchez A., Kendall MA, Hogg E., Lagat D., Lalloo U., Veloso V., Havlir DV, Landay A. Biomarkers Associated with Death After Initiating Treatment for Tuberculosis and HIV in Patients with Very Low CD ₄ Cells // Pathog Immun 2018;3(1):46-62). The disadvantage of this method is the high financial costs and the availability of a specialized laboratory.

As you can see, we have not found the method closest to the technical solution proposed by us for determining the probability of death only by complaints of a tuberculosis patient with HIV infection.

The technical result of the invention is to establish a method for determining the probability of death in a patient with tuberculosis with HIV infection upon admission to the hospital of the Federal Penitentiary Service of the Russian Federation (FSIN) for a limited number of clinical signs that individually and in combination have high sensitivity and specificity.

The stated technical result is achieved by the fact that, determining the probability of death in a patient with tuberculosis with HIV infection upon admission to the hospital of the Federal Penitentiary Service, which consists in the use of clinical signs, including the presence of a combination of patient complaints, characterized in that the following clinical signs are used as clinical signs: complaints of pain in the chest (B), headaches (B ₁ ), loose stools (G), no cough with sputum (K), dry cough (K ₁ ), acute onset of the disease (N), shortness of breath (O), chills (O ₁ ), sweating (P), weight loss (P ₁ ), lack of weakness (C), loss of appetite (A), fever (T) (body temperature above 37.0 ° C), while in the presence or absence of a sign, it is assigned a value of 1 or 0, followed by a formula calculation. That is, each feature is given an experimentally established weight value in the form of a product of a constant coefficient by 1 or 0:

(0.23354) and in the presence of chest pain (B)=1, no=0;

(0.58072) and in the presence of headaches (B ₁ )=1, absence=0;

(1.38119) and in the presence of loose stools (W)=1, absence=0;

(-0.6816) and in the absence of cough with sputum (K)=1, presence=0;

(0.80112) and in the presence of dry cough (K ₁ )=1, absence=0;

(0.48203) and in the presence of an acute onset of the disease (H)=1, absence=0;

(1.12078) and in the presence of shortness of breath (O)=1, absence=0;

(-0.0187) and in the presence of chills (O ₁ )=1, absence=0;

(0.41297) and in the presence of sweating (P)=1, absence=0;

(0.97654) and in the presence of weight loss (P ₁ )=1, no=0;

(-0.83206) and in the absence of weakness (C)=1, presence=0;

(-0.336) and in the presence of a decrease in appetite (A)=1, no=0;

(0.46756) and in the presence of fever (T)=1, absence=0;

and summed with a constant equal to (-4.14136) and then calculated using the formula:

где е - основание натурального логарифма = 2,71828947,

where e is the base of the natural logarithm = 2.71828947,

z is the sum of the numerical values of each of the features.

z \u003d (-4.14136) + 0.23354x (B) + 0.58072x (B ₁ ) + 1.38119x (W) + (-0.6816) x (K) + 0.80112x (K ₁ ) + 0.48203x(H)+1.12078x(O)+(-0.0187)x(O ₁ )+0.41297x(P)+0.97654x(P ₁ )+(-0.83206)x(C )+(-0.336)x(A)+0.46756x(T),

where "B" - the presence of pain in the chest, "B ₁ " - the presence of headaches, "F" - the presence of loose stools, "K" - the absence of cough with sputum, "K ₁ " - the presence of a dry cough, "N" - the presence of an acute onset of the disease, "O" - the presence of shortness of breath, "O ₁ " - the presence of chills, "P" - the presence of sweating, "P ₁ " - the presence of weight loss, "C" - the absence of weakness, "A" - the presence of a decrease in appetite, "T" - the presence of fever.

Based on the obtained probability value, a conclusion is made about the advisability of further monitoring and treatment of a tuberculosis patient with HIV infection upon admission to the hospital of the Federal Penitentiary Service in the intensive care unit. If the probability value is 50% or more, the risk of death and the need to hospitalize the patient in the intensive care unit are predicted, if the probability value is below 50%, the patient is hospitalized in the hospital ward.

To analyze the relationship between one qualitative feature, which acts as a dependent, resulting indicator, and a subset of quantitative and qualitative features, a logistic regression model is used with a step-by-step algorithm for including and excluding predictors. The results of evaluating the logistic regression equations are represented by a set of regression coefficients, achieved levels of significance for each coefficient. From the resulting logistic regression equations, the selection of equations was carried out that had the highest significance level for the area under the ROC curve (Receiver Operator Characteristic) AUC (AUC - Area Under Curve). The ranking of the selected predictors according to the degree of connection with the dependent variable was performed by sorting the predictors modulo the standardized regression coefficients. For dependent features with two gradations, based on logistic regression equations, an ROC analysis was carried out with the construction of ROC curves. To build a diagnostic rule to assess the risk of an event in patients, the method of statistical modeling was used - simple logistic regression (to identify features with the greatest weight) and multiple logistic regression (for the subsequent construction of a predictive model). The possibility of using this method is conditioned by our compliance with certain conditions, namely: the possibility of taking only two values by the dependent parameter (0 - no, 1 - yes), all other (independent) parameters involved in the analysis can take any values. Binary logistic regression calculates the probability of an event occurring depending on the values of the independent variables. In our work: 1 - a clinical sign was detected in a patient with a fatal outcome due to tuberculosis in combination with HIV infection, 0 - not detected.

The quality of the resulting model was assessed using sensitivity and specificity, as well as the value of the area under the ROC curve (see graph). Comparison of diagnostic signs among themselves was carried out on the basis of the calculation of the area under each ROC-curve. To assess the quality of the model in terms of the area under the ROC curve, we used an expert scale from (see source: Hosmer N. T., Lemeshow S. Applied logistic regression. New York: Wiley, 2000. 397 p). The cut-off criterion was the requirement for maximum sensitivity and specificity of the model.

Our results were used to build a model for predicting the risk of death in a TB patient with HIV infection upon admission to a hospital in a correctional facility of the Federal Penitentiary Service.

The risk probability of an event is estimated by the formula: p=e ^z /1+e ^z , where z=a+b ₁ x ₁ +b ₂ x ₂ +…+b _k x _k , x ₁ is the i-th independent variable (i= 1,2,…k), a - constant estimate, b ₁ , b ₂ , ..., b _k - estimates of logistic regression coefficients (see: Handbook of the Logistic Distribution. Marcel Dekker, Inc.. ISBN 978-0824785871).

If p gets a value less than 0.5, then we can assume that the event will not occur; otherwise, the event is assumed to occur.

To identify the features that have the greatest impact on building a model for predicting the probability of an event, we preliminarily carried out an exploratory analysis on the data array. When building a logistic regression model, the method of stepwise elimination of features was used.

To determine the factors associated with identifying the probability of death in a TB patient with HIV infection and to construct a diagnostic rule, we obtained 63 logistic regression equations that allow us to assess the risk of death in a TB patient with HIV infection. The event of interest to us is the following result: the detection of a lethal outcome of the disease in a patient with tuberculosis with HIV infection during hospitalization. According to the thumb rule, the maximum number of predictors included in the model in our study should not exceed 36. Of the 63 studied signs in patients, 13 predictors were identified (that is, we reduced the number of signs from 63 to 13) with the highest weight:

Our results were used to build a model for predicting the risk of death in a TB patient with HIV infection during hospitalization. When building a multiple logistic regression model, a method with stepwise exclusion of features was used. We ruled out all but thirteen signs: complaints of chest pain, headaches, loose stools, no cough with phlegm, dry cough, acute onset of the disease, shortness of breath, chills, sweating, weight loss, lack of weakness, loss of appetite, fever. The result is presented in table 1.

The quality of the approximation of the regression model is estimated using the similarity function. The measure of likelihood is the negative double value of the logarithm of this function (-2LL) - this is the value that characterizes the compliance of the model with the original data. The lower the value of this indicator, the more adequately the model is formed. The starting value for -2LL is the value that results from a regression model containing only constants. After adding the influence variables (see table above), the value of -2LL is 140.624. This value is 66.418 less than the initial value. Such a decrease in value means an improvement; the difference is referred to as the chi-square value and is very significant. This means that the initial model after the addition of variables has undergone a significant improvement.

Pseudo-coefficients of determination of Cox and Shell R ² and Nagelkirk R ² , obtained on the basis of the ratio of the likelihood functions of models with only a constant and with all coefficients, show the share of the influence of all factor features on the variance of the dependent variable, that is, the part of the variance explained using logistic regression, in this example is 38.47%.

As follows from Table 1, χ2 for the predictor is 66.418, with 13 degrees of freedom (P<0.0001): this means that the predictor is associated with predicting the risk of a fatal outcome of tuberculosis in a patient with HIV infection in a medical institution of the Federal Penitentiary Service. Table 2 below summarizes information about each variable in the model. This is clearly demonstrated in the attached drawing-graph.

Checking the significance of coefficients differing from zero is carried out using Wald statistics using a chi-square distribution, which is the square of the ratio of the corresponding coefficient to its standard error.

In our case, we got: an overly significant coefficient a \u003d -4.14136 and significant coefficients b ₁ \u003d 0.23354, b ₂ \u003d 0.58072, b ₃ \u003d 1.38119, b ₄ \u003d -0.6816, b ₅ \u003d 0 .80112, b ₆ = 0.48203, b ₇ = 1.12078, b ₈ = -0.0187, b ₉ = 0.41297, b ₁₀ = 0.97654, b ₁₁ = -0.83206, b ₁₂ = -0.336, b ₁₃ = 0.46756. With these thirteen coefficient values, we can calculate the probability P for each T-typing value.

As follows from Table 2, according to this model, in the presence of shortness of breath upon admission to the hospital of the Federal Penitentiary Service in a patient with tuberculosis and HIV infection, the chance of predicting a lethal outcome increases by 3.1 times. This is subject to the fixation of other factors. Similarly, in the presence of loose stools - 4.0 times, dry cough - 2.3 times, fever - 1.6 times, etc.

92.01% of cases were correctly classified according to this model, with the area under the ROC curve (AUC) - 0.886 (RMS error - 0.0328, 95% CI - from 0.848 to 0.917). In our model, AUC = 0.886, which indicates a good quality of the model.

This is visually displayed by a graph showing ROC curves for signs: complaints of chest pain, headaches, loose stools, no cough with sputum, dry cough, acute onset of the disease, shortness of breath, chills, sweating, weight loss, lack of weakness , loss of appetite, fever and coefficient β (constant) - predictive probability for the combination of these signs. This graph provides an understanding of the predictive probability of death in a TB patient with HIV infection, as well as sensitivity and specificity for individual clinical signs and their combination.

This method for determining the probability of death in a tuberculosis patient with HIV infection has been successfully used in the clinic, as can be seen from the following example:

From 1999 to 2017, 363 convicts who were admitted to a hospital with suspected tuberculosis were examined in the Tuberculosis Hospital branch of the FKUZ MSCh-43 of the Federal Penitentiary Service of Russia (subsequently confirmed by microbiological: sputum/other liquids cultured on solid or liquid media, X-ray or DNA- diagnosis), with HIV infection 4B, 4B and 5 stages.

Patients are divided into 2 groups:

• the first group - 30 TB patients with HIV infection (8.3%) who subsequently died from these diseases.

• the second group - 333 patients (91.7%) with tuberculosis with HIV infection without death.

As a result, this method made it possible to correctly classify cases according to this model in 92.01%.

As a result, the sensitivity of the probability of a fatal outcome in case of a disease in a tuberculosis patient with HIV infection during hospitalization in a hospital of a medical institution of the Federal Penitentiary Service was 30.0% for a complaint of chest pain, specificity - 79.9%, the presence of headaches 53.3% and 85.6%, the presence of liquid stools 43.3% and 94.3%, the absence of cough with sputum 90.0% and 34.2%, the presence of dry cough 3.3% and 97.3%, the presence of an acute onset of the disease 46.7% and 85.3%, shortness of breath 80.0% and 74.8%, chills 16.7% and 94.3%, sweating 43.3% and 80.2%, weight loss 70, 0% and 82.0%, no weakness 93.3% and 41.7%, loss of appetite 63.3% and 82.6%, fever 86.7% and 60.4%, respectively. The combined combination of this method for detecting the probability of a fatal outcome of the disease in a tuberculosis patient with HIV infection during hospitalization in a hospital of a medical institution of the Federal Penitentiary Service for thirteen clinical manifestations gives a sensitivity of 83.3% with 84.1% specificity (see drawing-graph).

As an example, consider the data of a 35-year-old patient with tuberculosis and HIV infection, K., who, during examination, complains of chest pain, headaches, loose stools, dry cough, shortness of breath, chills, sweating, weight loss, weakness, loss of appetite, fever with acute onset of the disease.

z=(-4.14136)+0.23354x(1)+0.58072x(1)+1.38119x(1)+(-0.6816)x(1)+0.80112x(1)+0, 48203x(1)+1.12078x(1)+(-0.0187)x(1)+0.41297x(1)+0.97654x(1)+(-0.83206)x(0)+(- 0.336)x(1)+0.46756x(1),

where z=1.27879. Hence the probability of death, according to the formula:

(p=2.71828183 ^1.27879 /1+2.71828183 ^1.27879 ), p=78.2%.

The probability is higher than 50%, that is, this patient has a high risk of a fatal outcome of the disease.

An opposite example, patient C., 30 years old, with tuberculosis and HIV infection. On examination, he complains of cough with sputum, weight loss and fever:

z=(-4.14136)+0.23354x(0)+0.58072x(0)+1.38119x(0)+(-0.6816)x(0)+0.80112x(0)+0, 48203x(0)+1.12078x(0)+(-0.0187)x(0)+0.41297x(0)+0.97654x(1)+(-0.83206)x(1)+(- 0.336)x(0)+0.46756x(1),

where z=-3.52932. Hence the probability of death, according to the formula:

(p=2.71828183 ^-3.52932 /1+2.71828183 ^-3.52932 ), p=2.85%.

The probability is below 50%, that is, this patient has a low risk of a fatal outcome of the disease.

Thus, to identify the likelihood of death in a patient with tuberculosis with HIV infection during hospitalization in a hospital of a medical institution of the Federal Penitentiary Service, a detailed clinical examination is carried out. Clarify the presence or absence of the following clinical manifestations: complaints of chest pain, headaches, loose stools, no cough with sputum, dry cough, acute onset of the disease, shortness of breath, chills, sweating, weight loss, lack of weakness, loss of appetite, fever.

Then according to our formula:

z \u003d (-4.14136) + 0.23354x (B) + 0.58072x (B ₁ ) + 1.38119x (W) + (-0.6816) x (K) + 0.80112x (K ₁ ) + 0.48203x(H)+1.12078x(O)+(-0.0187)x(O ₁ )+0.41297x(P)+0.97654x(P ₁ )+(-0.83206)x(C )+(-0.336)x(A)+0.46756x(T),

where "B" - the presence of pain in the chest, "B ₁ " - the presence of headaches, "F" - the presence of loose stools, "K" - the absence of cough with sputum, "K ₁ " - the presence of a dry cough, "N" - the presence of an acute onset of the disease, "O" - the presence of shortness of breath, "O ₁ " - the presence of chills, "P" - the presence of sweating, "P ₁ " - the presence of weight loss, "C" - the absence of weakness, "A" - the presence of a decrease in appetite, "T" - the presence of fever; calculate the sum (z) of the numerical values of the signs (if any) multiplied by the discriminant coefficient of the sign (0.23354 for chest pain, 0.58072 for headaches, 1.38119 for loose stools, -0.6816 for no cough with sputum, 0.80112 for dry cough, 0.48203 for acute onset, 1.12078 for shortness of breath, -0.0187 for chills, 0.41297 for sweating, 0.97654 for weight loss, -0.83206 for no weakness, -0.336 for decreased appetite, 0.46756 for fever) and summed with a constant equal to (-4.14136). Given a logistic function of the form:

We get the opportunity to calculate the probability of death in a tuberculosis patient with HIV infection upon admission to the FSIN hospital using a well-known formula (sources: Agresti, Alan. (2002). Categorical Data Analysis. New York: Wiley-Interscience. ISBN 0-471-36093 -7 Amemiya, T. (1985) Advanced Econometrics Harvard University Press ISBN 0-674-00560-0 Balakrishnan, N. (1991) Handbook of the Logistic Distribution Marcel Dekker, Inc. ISBN 978- 0824785871 Greene, William H. (2003) Econometric Analysis, fifth edition Prentice Hall ISBN 0-13-066189-9 Hosmer, David W. Stanley Lemeshow (2000) Applied Logistic Regression, 2nd ed. New York, Chichester, Wiley, ISBN 0-471-35632-8):

где e - основание натурального логарифма = 2,71828947,

where e is the base of the natural logarithm = 2.71828947,

z is the sum of the numerical values of each of the features.

Thus, the value of the probability of death in a patient with tuberculosis with HIV infection in a medical correctional institution of the Federal Penitentiary Service of the Russian Federation is obtained, according to the value of which the forecast of a lethal outcome is carried out.

If the probability value is 50% or more, the risk of death and the need to hospitalize the patient in the intensive care unit are predicted, if the probability value is below 50%, the patient is hospitalized in the hospital ward.

The method allows to reliably, informatively, free of charge and accurately calculate the probability of death in a tuberculosis patient with HIV infection in a medical correctional institution of the Federal Penitentiary Service of the Russian Federation for hospitalization in the intensive care unit, and also, if necessary, to conduct timely additional examination of the patient to resolve the issue of early release from places imprisonment for health reasons.

Determining the probability of death in a tuberculosis patient with HIV infection in a medical correctional facility of the Federal Penitentiary Service of the Russian Federation without additional financial costs using simple clinical methods is the advantage and advantage of the proposed method compared to known prototypes.

Claims (20)

A method for determining the probability of death in a tuberculosis patient with HIV infection upon admission to a hospital of the Federal Penitentiary Service of the Russian Federation (FSIN), which consists in the use of clinical signs, including complaints of the patient, characterized in that the following are used as clinical signs: complaints of pain in the chest cell (B), headaches (B ₁ ), loose stools (G), absence of cough with sputum (K), dry cough (K ₁ ), acute onset of the disease (N), shortness of breath (O), chills (O1), sweating (P), weight loss (P1), lack of weakness (C), loss of appetite (A), fever - body temperature above 37.0 ° C (T), then evaluate clinical signs: in the presence of pain in the chest (B) assign 1, in the absence of 0; in the presence of headaches (B ₁ ) 1, no 0; in the presence of loose stools (F) 1, in the absence of 0; in the absence of cough with sputum (K) 1, the presence of 0; in the presence of dry cough (K ₁ ) 1, no 0; in the presence of an acute onset of the disease (N) 1, no 0; in the presence of shortness of breath (O) 1, in the absence of 0; in the presence of chills (O ₁ ) 1, absence 0; in the presence of sweating (P) 1, absence 0; in the presence of weight loss (P ₁ ) 1, no 0; in the absence of weakness (C) 1, the presence of 0; in the presence of loss of appetite (A) 1, no 0; in the presence of fever (T) 1, absence 0; and then determine the probability of death by the formula:

where e is the base of the natural logarithm = 2.71828947, z is the sum of the numerical values of each of the clinical signs: z=(-4.14136)+0.23354×(B)+0.58072×(B ₁ )+1.38119×(W)+(-0.6816)×(C)+0.80112×( K ₁ )+0.48203×(N)+1.12078×(O)+(-0.0187)×(O ₁ )+0.41297×(P)+0.97654×(P ₁ )+( -0.83206)×(C)+(-0.336)×(A)+0.46756×(T), if the probability value is 50% or more, the risk of death and the need to hospitalize the patient in the intensive care unit are predicted, if the probability value is below 50%, the patient is hospitalized in the hospital ward.

Images