RU2757522C1 - Predicting method for development of chronic kidney disease in patients with type 1 and type 2 diabetes mellitus (options) - Google Patents

Abstract

FIELD: medicine.

SUBSTANCE: group of inventions relates to the field of medicine, especially to prognosis of chronic kidney disease (CKD) development after 5 years in patients with type 1 or type 2 diabetes mellitus (DM). For case of type 1 diabetes a set of diagnostically significant indicators is determined: gender, age at the time of diagnosis, myocardial infarction in anamnesis, diabetic coma in anamnesis, diabetic retinopathy, the value of body mass index (BMI). For case of type 2 diabetes a set of diagnostically significant indicators is determined: gender, age at the time of diagnosis, BMI, GFR, glycated hemoglobin, total cholesterol, neuropathy in anamnesis, stroke in anamnesis, diabetic retinopathy, amputation and oncological diseases in anamnesis. According to the stated formulas probability of developing CKD in 5 years is calculated. Based on the obtained data the presence of a high (more than 50%), medium (25-50%) or low (less than 25%) risk of developing CKD within 5 years is determined.

EFFECT: group of inventions makes it possible to determine risk of developing CKD in patients with type 1 or type 2 diabetes by evaluating a set of the most significant indicators.

2 cl, 8 tbl, 4 ex

Description

Technology area

The group of inventions relates to medicine, namely to predicting the development of chronic kidney disease (CKD) in 5 years in patients with type 1 and 2 diabetes mellitus (DM). The method can be used in therapeutic (endocrinological, nephrological) departments when planning the tactics of managing a patient with diabetes.

State of the art

In conditions of such a complex and multifactorial pathology as CKD, the construction of multivariate models based on a retrospective analysis of data from patients with a known outcome, i.e. who developed and did not develop the disease is the most promising method for assessing risk factors, which makes it possible to establish the aggregate of the most significant of them. This uses the analysis of both population studies and register databases.

From the prior art, a method is known for assessing risk factors for the development of CKD in patients with diabetes in comparison with the general population, which included 5.5 million people from 28 countries, including 781 thousand patients with diabetes [Nelson RG, Grams ME, Ballew SH, et al. Development of Risk Prediction Equations for Incident Chronic Kidney Disease. JAMA. 2019; 322 (21): 2104-2114. doi: 10.1001 / jama.2019.17379 - prototype]. The work analyzed the risk factors for a decrease in the glomerular filtration rate (GFR) <60 ml / min / 1.73 m ² in individuals without carbohydrate metabolism disorders compared with a cohort of diabetes mellitus, while the DM group was analyzed without dividing by type.

Among the disadvantages of this method, a fundamental difference from the proposed method should be noted, which consists in the absence of a separate analysis of the factors for the development of CKD in different types of diabetes. Literature data and clinical practice indicate the pathogenetic features of the course of types of diabetes, which entails the presence of various factors of kidney damage.

So, in type 1 diabetes, the development of CKD is predominantly microvascular in nature with predominant and earlier damage to the glomerulus, its basement membrane (BMC), with the development of increased BMC permeability for proteins - first micro, then macroalbuminuria. Meanwhile, the processes of sclerosis of the interstitial renal tissue and vascular hyalinosis in type 1 diabetes, as a rule, are secondary to the development of increased protein excretion - the so-called classical albuminuric nephropathy. In particular, since type 1 diabetes usually develops at a young age, patients “do not have time to develop” lipid metabolism disorders, therefore the factor of lipid metabolism disorders is less significant in the pathogenesis of CKD in type 1 diabetes. Thus, in our model for type 1 diabetes (calculated from data from real patients with and without CKD), dyslipidemia did not show statistical significance, in contrast to the model for type 2 diabetes, which is characterized by the combined effect of pathogenetic factors leading to kidney damage.

In conditions of the combined effects of hyperglycemia, hypertension, dyslipidemia, obesity, tissue inflammation, etc., which develop in parallel or even precede the development of type 2 diabetes, the mechanism of kidney damage is heterogeneous, damage to the interstitial tissue and non-glomerular structures of the kidneys may prevail over damage to the glomerular filter, and the processes of hyalinosis and sclerosis go parallel to the defeat of the main functional unit - the renal glomerulus. Thus, in type 2 diabetes, nonalbuminuric CKD develops much more often, when the first clinical marker of pathology is a decrease in GFR in the absence of previous micro- and macroalbuminuria, or they develop in parallel.

The inventive method takes into account the difference in the mechanisms of kidney damage in type 1 diabetes and type 2 diabetes and offers a combination of the most significant predictors for each type of diabetes, established on the basis of an analysis of a database of patients who developed and did not develop pathology for 5 years.

It should be noted that the separation of the models in the claimed method is of undoubted interest, since it allows to differentiate the CKD risk phenotype in different types of diabetes.

Thus, the known method for assessing the risk factors for the development of CKD in patients with diabetes (Nelson RG et al.) Is characterized by the following disadvantages.

1. The data obtained describe a fundamentally different format of the risk of pathology - the risk of developing CKD in patients with diabetes compared to the general population without diabetes.

2. In this study, indicators of the predictive value of the predicted result are not presented, which makes it impossible to assess the accuracy of the forecast made by the model.

In contrast to the analogue, as a result of the studies carried out by the authors of the present invention, specific factors of CKD in various types of diabetes were identified, the prognostic value of both negative and positive prognosis was calculated with high indicators of the accuracy of the result.

The technical problem to be solved was the development of an individual accurate prognosis of the development of CKD in diabetes, based on the parameters available in routine clinical practice.

Disclosure of invention

The technical result consists in determining the risk of developing CKD after 5 years in patients with T1DM and T2DM.

The method for predicting the development of CKD after 5 years in patients with diabetes is simple to execute, highly accurate (overall accuracy 94% (95% CI: 93% -95%) in type 1 diabetes; 81% (95% CI: 81% -82% ) with type 2 diabetes) and applicable in the usual conditions of clinical medical institutions, i.e., does not require any indicators that are not evaluated in real clinical practice.

The claimed invention is characterized by the development of a diagnostically significant set of indicators, taking into account the demographic, anthropometric and clinical data of the patient, taking into account the type of diabetes mellitus, which distinguishes this method from known solutions.

The invention makes it possible to predict the development of CKD in 5 years with an overall accuracy of at least 93% in type 1 diabetes and at least 81% in type 2 diabetes. Based on the result obtained, the doctor will be able to plan more frequent monitoring of the patient and preventive measures in case of a positive prognosis of the model (high risk of developing CKD).

The possibility of predicting the risk for this pathology is especially important, since the initial stages of CKD do not cause complaints in the patient and are detected only by the method of active screening, thus, the result of the invention directly affects the tactics of patient management and, thereby, the quality and duration of his life.

The above aspects of the achieved technical result are due to the following sets of essential features:

1. For the case of the presence of type 1 diabetes:

determine a set of diagnostically significant indicators: gender (x _gender ), age at the time of diagnosis (x _age ), a history of myocardial infarction (x _{heart attack} ), a history of diabetic coma (x _coma ), the presence of diabetic retinopathy (x _retinopathy ), body mass index (BMI) value (х _bm ),

Based on the data obtained and in the presence of type 1 diabetes mellitus, the likelihood of developing CKD after 5 years in patients with type 1 diabetes mellitus (p) is calculated using the formula:

) * 100%, p = 1 / (1 +

) * 100%,

+ 0,063*

+ 1,148*

+ 0,617*

+ 0,903*

+ 0,040*

,where z = -7.079 + 1.182 *

+ 0.063 *

+ 1.148 *

+ 0.617 *

+ 0.903 *

+ 0.040 *

,

the indicators are assessed quantitatively as follows:

x _gender, assigned "0 points" if there is a male, "1 point" - female;

x _age , full years;

x _{heart attack,} assign "0 points" in the absence of a history of myocardial infarction, "1 point" - in the presence of a history of myocardial infarction;

x _coma , assign "0 points" in the absence of coma in the anamnesis, "1 point" - in the presence of coma in the anamnesis;

x _retinopathy , assigned "0 points" in the absence of diabetic retinopathy, "1 point" - in the presence of diabetic retinopathy;

x _bmt , kg / m ² ,

at p≤0.25, the probability of developing CKD within 5 years is 25% or less, which corresponds to a low risk,

at 0.25 <p≤0.5, the probability of developing CKD within 5 years is 25-50%, which corresponds to the average risk,

at p> 0.5, the probability of developing CKD within 5 years is more than 50%, which corresponds to a high risk.

2. For the case of the presence of type 2 diabetes:

determine a set of diagnostically significant indicators: gender (x _gender ), age at the time of diagnosis (x _age ), BMI (x _bmi ), GFR (x _skf ), glycated hemoglobin (x _HbA1c ), total cholesterol (x _{total cholesterol} ) , a history of neuropathy (x _neuropathy ), a history of stroke (x _stroke ), diabetic retinopathy (x _retinopathy ), amputation (x _amputation ) and a history of _cancer (x oncology),

Based on the data obtained and in the presence of type 2 diabetes, the probability of developing CKD after 5 years in patients with type 2 diabetes (p) is calculated using the formula:

) * 100%, p = 1 / (1 +

) * 100%,

+ 0,125*

+ 0,061*

+ 0,049*

+ 0,135*

+ 0,012*

+ 0,367*

+ 0,467*

+ 0,476*

+ 0,811*

+ 0,909*

where z = -13.994 + 1.983 *

+ 0.125 *

+ 0.061 *

+ 0.049 *

+ 0.135 *

+ 0.012 *

+ 0.367 *

+ 0.467 *

+ 0.476 *

+ 0.811 *

+ 0.909 *

the indicators are assessed quantitatively as follows:

x _gender, assigned "0 points" if there is a male, "1 point" - female;

x _age , full years;

x _bmt , kg / m ² ;

x _HbA1c ,%;

x _{Total cholesterol,} mmol / l;

x _skf, ml / min / 1.73m ² ;

x _{neuropathy is} assigned "0 points" in the absence of a history of neuropathy, "1 point" - in the presence of a history of neuropathy;

x _stroke, assign "0 points" in the absence of a history of stroke, "1 point" - in the presence of a history of stroke;

x _amputation , assigned "0 points" in the absence of amputation, "1 point" - in the presence of amputation;

x _retinopathy , assigned "0 points" in the absence of diabetic retinopathy, "1 point" - in the presence of diabetic retinopathy;

x _oncology , assign "0 points" in the absence of a history of oncological diseases in the anamnesis, "1 point" - in the presence of oncological diseases in the anamnesis,

at p≤0.25, the probability of developing CKD within 5 years is 25% or less, which corresponds to a low risk,

at 0.25 <p≤0.5, the probability of developing CKD within 5 years is 25-50%, which corresponds to the average risk,

at p> 0.5, the probability of developing CKD within 5 years is more than 50%, which corresponds to a high risk.

The authors of the proposed method revealed that with type 1 and type 2 diabetes, 3 factors were common - these are female sex, age and BMI. These factors have shown high significance in the development of kidney pathology of any genesis and regardless of diabetic status, and can be regarded as general population environmental factors.

Most of the other factors included in the model were associated with associated micro- and macrovascular complications, the development of which significantly increased the risk of developing CKD.

In type 1 diabetes, 3 factors were significant. The first factor is the presence of retinopathy, which emphasizes the prevailing "microvascular" nature of kidney damage with glomerulopathy and increased vascular permeability, the equivalent of which is retinal vascular damage. So, in the classical sense, the diabetic nature of kidney damage and the diagnosis of DN in a patient with type 1 diabetes were even questioned if he did not have retinopathy. The second factor is a history of coma, which can be considered a criterion for extremely unsatisfactory control of carbohydrate metabolism and cardiovascular risk at the same time. And the third factor is the presence of a heart attack, which reflects the high significance of the general mechanisms of kidney damage with the development of CVD in the formation of cardiorenal and nephrocardial syndromes.

In type 2 diabetes mellitus draws attention to the fact that the main predictors of CKD were 2 times more than in type 1 diabetes. They reflected almost all significant mechanisms involved in kidney damage: age-related, sexual, associated with impaired lipid and carbohydrate metabolism, obesity and a wide range of not only micro- and macrovascular complications, but also concomitant diseases with a common nature of systemic inflammation and activation of cytokines and factors. growth, namely, cancer pathology. This once again emphasizes the heterogeneity of kidney damage in this type of diabetes with the involvement of a whole complex of environmental, metabolic, hemodynamic and inflammatory factors in its development, which significantly complicates the possibilities of preventing its development.

Under these conditions, the claimed method makes it possible to assess the individual prognosis of the development of pathology on the basis of the parameters available in normal clinical practice, becomes of paramount importance. Identifying a high-risk group will make it possible to personalize the monitoring algorithm for patients requiring more careful observation.

Implementation of the invention

To predict the development of CKD after 5 years in patients with diabetes, a set of diagnostically significant indicators is determined.

The patient undergoes a complex of clinical examination, including the collection of anamnesis according to medical records and anthropometric indicators.

The collection of the obtained data is processed using a classification model.

The logistic regression model acts as such a classification model. Logistic regression is one of the multiple regression methods, the general purpose of which is to investigate the relationship between a dependent binary trait and / or several quantitative or qualitative independent traits. Logistic regression analysis allows you to build a statistical model to predict the likelihood of an event occurring from the available data. (Rebrova O.Yu. Statistical analysis of medical data. Application of the STATISTICA application package. - M .: MediaSfera, 2003. - 312 p.) Model equation:

), гдеp = 1 / (1 +

), where

+

, гдеZ =

+

, where

, …,

– независимые признаки,

, ...,

- independent signs,

, …,

– коэффициенты.

, ...,

- coefficients.

Development of CKD or maintenance of normal GFR function was used as a binary response. Next, a step-by-step logistic regression analysis was performed using the SPSS 26 software package (IBM, USA). As a result, the step with the best classification matrix was chosen (DC and DS are maximal). The predictors included in this model were chosen as the target population.

The following predictors were analyzed:

1. Gender (male - 0 points / female - 1 point)

2. Age (full years)

3. Duration of SD

4. Laboratory indicators:

- GFR (ml / min / 1.73m ² )

- Total cholesterol (mmol / l)

5. BMI (kg / m ² )

6. Anamnesis data:

Neuropathy (yes - 1 point / no - 0 points)

Myocardial infarction (yes - 1 point / no - 0 points)

Stroke (yes - 1 point / no - 0 points)

Diabetic foot syndrome (SDS) (yes - 1 point / no - 0 points)

Amputation (yes - 1 point / no - 0 points)

Coma (yes - 1 point / no - 0 points)

Diabetic retinopathy (yes - 1 point / no - 0 points)

Angina pectoris (yes - 1 point / no - 0 points)

Chronic heart failure (CHF) (yes - 1 point / no - 0 points)

Albuminuria (mg / day) and not taking angiotensin converting enzyme inhibitors (ACE inhibitors) or angiotensin receptor blockers (ARBs) (yes - 1 point / no - 0 points).

The method was developed as a result of studies carried out on a retrospective sample of data on patients with type 1 and type 2 diabetes, obtained from the Federal Register of Diabetes Mellitus under the auspices of the Federal State Budgetary Institution “National Medical Research Center of Endocrinology” of the Ministry of Health of Russia.

Inclusion criteria:

- Patients with type 1 diabetes and type 2 diabetes;

- Patients whose baseline GFR is ≥ 90 ml / min / 1.73m ² and <130 ml / min / 1.73m ² ;

- Patient data on the presence of CKD after 5 years were available;

- Patients whose GFR after 5 years is either ≥ 90 ml / min / 1.73m ² and <130 ml / min / 1.73m ² , or <60 ml / min / 1.73m ² .

Exclusion criteria:

- Patients without type 1 diabetes and type 2 diabetes;

- Patients whose baseline GFR is <90 ml / min / 1.73m ² or ≥ 130 ml / min / 1.73m ² ;

- Patients with unsatisfactory completed medical documentation;

- Patients whose GFR after 5 years is either <90 ml / min / 1.73m ² and ≥ 60 ml / min / 1.73m ² or ≥ 130 ml / min / 1.73m ² .

After applying the specified inclusion and exclusion criteria, the size of the analyzed sample was:

- 4703 patients with type 1 diabetes, of which 287 (6%) patients with developed CKD after 5 years and 4416 (94%) without CKD after 5 years;

- 12,215 patients with type 2 diabetes, of which 3557 (29%) patients with developed CKD after 5 years and 8658 (71%) without CKD after 5 years.

The survey parameters are used to calculate the classification model, as a result of which the following assessment indicators are determined:

- The likelihood of developing CKD after 5 years in patients with type 1 diabetes and type 2 diabetes;

- The likelihood of no development of CKD after 5 years in patients with type 1 diabetes and type 2 diabetes.

Research carried out, incl. using machine learning algorithms ("logistic regression analysis") made it possible to determine a set of indicators with threshold values for predicting the development of CKD in 5 years. The parameters with the greatest significance were obtained using stepwise logistic regression analysis (SPSS 26 software package (IBM, USA)).

Type 1 SD

As a result of the study, a diagnostically significant complex was obtained with the inclusion of the following parameters for patients with type 1 diabetes:

1. Sex (male - 0 / female - 1), p <0.001

2. Age (full years), p <0.001

3. Myocardial infarction (no-0 / yes-1), p = 0.015

4. Coma (no-0 / yes-1), p = 0.061

5. Diabetic retinopathy (no-0 / yes-1), p <0.001

6. BMI (kg / m ² ), p = 0.002

Not included in the model: duration of diabetes mellitus, GFR, total cholesterol, the presence of neuropathy, stroke, SDS, amputation, coma, angina pectoris and CHF in history, and the presence of albuminuria with no ACE inhibitors and ARBs.

Table 1 lists the determined indicators, which are estimated in absolute values or conventional units.

Table 1. Assessment of indicators

Index Grade Demographic indicators Gender (x _floor ) Male "0 conv. units "
Female "1 conv. units " Age at diagnosis (x _age ) Years Anthropometric indicators BMI (x _BMI ) kg / m2 Anamnesis according to medical records History of myocardial infarction (x _{heart attack} ) No "0 conv. units "
There is “1 conv. units " History of coma (x _coma ) No "0 conv. units "
There is “1 conv. units " History of diabetic retinopathy (x _retinopathy ) No "0 conv. units "
There is “1 conv. units "

The developed method showed a statistically significant result (p <0.001, χ2). Female gender, age, history of myocardial infarction and diabetic retinopathy and BMI are directly related to the likelihood of developing CKD after 5 years (OR 3.260, 95% CI (2.513; 4.229), OR 1.065, 95% CI (1.053; 1.077) OR 3.153, 95% CI (1.254; 7.923), OR 2.468, 95% CI (1.893; 3.217), OR 1.040, 95% CI (1.014; 1.067)).

As a result of the invention, the conclusion about the development of CKD after 5 years in a patient with type 1 diabetes can be made using the formula obtained by using the above logistic regression model:

) * 100%p = 1 / (1 +

) * 100%

+ 0,063*

+ 1,148*

+ 0,617*

+ 0,903*

+ 0,040*

z = -7.079 + 1.182 *

+ 0.063 *

+ 1.148 *

+ 0.617 *

+ 0.903 *

+ 0.040 *

when receiving a value of p> 0.5, a conclusion is made about a high risk of developing CKD after 5 years in a patient with type 1 diabetes, when receiving 0.25 <p ≤ 0.5 - about an average risk, with p ≤ 0.25 - about a low risk.

The main objective of the developed method was to identify patients at risk of developing CKD after 5 years with type 1 diabetes (requiring close monitoring).

The classification matrix is presented in Table 2. The operating characteristics of the logistic regression model are shown in Table 3.

Table 2. Classification matrix as a result of building a model using logistic regression analysis, p <0.001 (χ2) (n = 4703)

SCF GFR <60 ml / min / 1.73m ² GFR> 130 ml / min / 1.73m ² and ≤ 90 ml / min / 1.73m ² Result of the logistic regression model GFR <60 ml / min / 1.73m ² 21 eight GFR> 130 ml / min / 1.73m ² and ≤ 90 ml / min / 1.73m ² 266 4408

Table 3. Operational characteristics of the model

Model DC,% DS,% PCPR,% PCOR,% Logistic regression 7 (5-9) 100 (100-100) 72 (53-87) 94 (94-94)

If, when implementing the proposed method, data on the development of CKD is obtained after 5 years, then with a probability of 72% (53%; 87%) the patient will have CKD. If data is obtained on a low or medium risk of developing CKD after 5 years, then with a probability of 94% (94%; 94%) the patient will not develop CKD, which corresponds to the prognosis of "low risk" and "medium risk".

The cut-off point of the function for identifying the development of CKD after 5 years is 0.5.

SD type 2

In type 2 diabetes mellitus, as a result of the analysis, a diagnostically significant complex was obtained with the inclusion of the following parameters:

1. Sex (male - 0 / female - 1), p <0.001

2. Age (full years), p <0.001

3. BMI (kg / m ² ), p <0.001

4.GFR (ml / min / 1.73m ² ), p <0.001

5.HbA1c (%), p = 0.002

6. Total cholesterol (mmol / L), p <0.001

7. A history of neuropathy (no-0 / yes-1), p <0.001

8. History of stroke (no-0 / yes-1), p <0.001

9. A history of diabetic retinopathy (no-0 / yes-1), p <0.001

10. A history of amputation (no-0 / yes-1), p = 0.004

11. History of oncology (no-0 / yes-1), p = 0.034

Not included in the model: the duration of diabetes, the presence of SDS, angina pectoris and CHF in the anamnesis and the presence of albuminuria with the absence of ACE inhibitors and ARBs.

Table 4 lists the determined indicators, which are estimated in absolute values or conventional units.

Table 4. Assessment of indicators

Index Grade Demographic indicators Gender (x _floor ) Male "0 conv. units "
Female "1 conv. units " Age at diagnosis (x _age ) Years Anthropometric indicators BMI (x _BMI ) kg / m2 Blood chemistry SCF (x _skf ) ml / min / 1.73m ² HbA1c (x _HbA1c ) % Total cholesterol (x _{Total cholesterol} ) mmol / l Anamnesis according to medical records History of neuropathy (x _neuropathy ) No "0 conv. units "
There is “1 conv. units " History of stroke (x _stroke ) No "0 conv. units "
There is “1 conv. units " History of diabetic retinopathy (x _retinopathy ) No "0 conv. units "
There is “1 conv. units " History of amputation (x _amputation ) No "0 conv. units "
There is “1 conv. units " History of oncology (x _oncology ) No "0 conv. units "
There is “1 conv. units "

The inventive method demonstrated a statistically significant result (p <0.001, χ2). Female gender, age, BMI, GFR, total cholesterol, HbA1c, history of neuropathy, stroke, amputation, diabetic retinopathy and oncology have a direct relationship with the likelihood of developing CKD after 5 years (OR 7.266, 95% CI (6.514; 8.106), OR 1.133, 95% CI (1.125; 1.142), OR 1.063, 95% CI (1.054; 1.073), OR 1.013, 95% CI (1.007; 1.018), OR 1.145, 95% CI (1.092; 1.200), OR 1.050 , 95% CI (1.018; 1.084), OR 1.595, 95% CI (1.429; 1.781), OR 1.610, 95% CI (1.274; 2.034), OR 2.250, 95% CI (1.301; 3.890), OR 1.43, 95 % CI (1.263; 1.649), OR 2.483, 95% CI (1.069; 5.764)).

As a result of the invention, the conclusion about the development of CKD after 5 years in a patient with TYPE 2 diabetes can be made using the formula obtained by using the above logistic regression model:

) * 100%p = 1 / (1 +

) * 100%

+ 0,125*

+ 0,061*

+ 0,049*

+ 0,135*

+ 0,012*

+ 0,367*

+ 0,467*

+ 0,476*

+ 0,811*

+ 0,909*

.z = -13.994 + 1.983 *

+ 0.125 *

+ 0.061 *

+ 0.049 *

+ 0.135 *

+ 0.012 *

+ 0.367 *

+ 0.467 *

+ 0.476 *

+ 0.811 *

+ 0.909 *

...

when receiving a value of p> 0.5, a conclusion is made about a high risk of developing CKD after 5 years in a patient with type 2 diabetes, when receiving 0.25 <p ≤ 0.5 - about an average risk, with p ≤ 0.25 - about a low risk.

The main objective of the developed method was to identify patients at risk of developing CKD after 5 years with type 2 diabetes (requiring close monitoring).

The classification matrix is presented in Table 5. The operational characteristics of the logistic regression model are shown in Table 6.

Table 5. Classification matrix as a result of building a model using logistic regression analysis on a training set, p <0.001 (χ2) (n = 12215)

SCF GFR <60 ml / min / 1.73m ² GFR> 130 ml / min / 1.73m ² and ≤ 90 ml / min / 1.73m ² Result of the logistic regression model GFR <60 ml / min / 1.73m ² 2171 917 GFR> 130 ml / min / 1.73m ² and ≤ 90 ml / min / 1.73m ² 1386 7741

Table 6. Operational characteristics of the model built on the training sample

Model DC,% DS,% PCPR,% PCOR,% Logistic regression (training sample) 61 (60-62) 89 (89-90) 70 (69-72) 85 (84-85)

If, when implementing the proposed method, data is obtained on the development of CKD after 5 years, then with a probability of 70% (69%; 72%) the patient will have CKD. If data is obtained on a low or medium risk of developing CKD after 5 years, then with a probability of 85% (84%; 85%) the patient will not develop CKD, which corresponds to the prognosis of "low risk" and "medium risk".

The cut-off point of the function for identifying the development of CKD after 5 years is 0.5.

The inventive method has been successfully validated on a sample of 13,775 patients with type 2 diabetes. The sample was obtained from the Federal Register of Diabetes Mellitus for the period 2015-2019. The classification matrix on the test sample is presented in Table 7. The operational characteristics of the model on the test sample were included in the confidence intervals for the operational characteristics of the model built on the training sample (Table 8).

Table 7. Classification matrix as a result of testing the model (n = 13775)

SCF GFR <60 ml / min / 1.73m ² GFR> 130 ml / min / 1.73m ² and ≤ 90 ml / min / 1.73m ² Result of the logistic regression model GFR <60 ml / min / 1.73m ² 2349 986 GFR> 130 ml / min / 1.73m ² and ≤ 90 ml / min / 1.73m ² 1566 8874

Table 8. Operational characteristics of the model tested on the training sample

Model DC,% DS,% PCPR,% PCOR,% Logistic regression (training sample) 60 90 70 85

The inventive method can be implemented, for example, by software (in the form of a calculator) and be a tool for decision support. The calculator can include fields with a drop-down list (type of diabetes, gender and presence of diabetic retinopathy) and fields with the ability to enter values for indicators (age, BMI). Further, depending on the selected type of SD, the following fields can be displayed:

- With type 1 diabetes:

• History of myocardial infarction (drop-down list)

• A history of coma (drop-down list box)

- With type 2 diabetes:

• A history of neuropathy (drop-down list box)

• History of stroke (drop-down list box)

• Presence of amputation (drop-down list box)

• History of oncological diseases (drop-down list box)

The doctor chooses the gender of the patient, enters the age, BMI, laboratory parameters (with TYPE 2 diabetes) and fills in the anamnesis. As a result, he receives a forecast of the development of CKD in 5 years and its accuracy.

Example 1

L. patient with type 1 diabetes, male, 56 years old, BMI 32.52 kg / ^m2, the initial GFR = 95 ml / min / 1,73m ^2. History of diabetic retinopathy.

According to the model's calculations, this patient has a low risk of developing CKD (21%) after 5 years.

According to the analyzes, after 5 years the patient's GFR = 93 ml / min / 1.73m ² , that is, he did not develop CKD.

Example 2

T. patient with type 1 diabetes, female, 56 years old, BMI 26.23 kg / ^m2, the initial GFR = 97 ml / min / 1,73m ^2. History of coma and retinopathy.

The model predicts that this patient will develop high-risk CKD (55%) after 5 years.

According to the analyzes, after 5 years the patient's GFR = 53 ml / min / 1.73m ² , that is, he developed CKD.

Example 3

C. Patient with type 2 diabetes, female, 51 years old, BMI of 28.4 kg / m ^2, HbA1c = 11,2%, total cholesterol = 7.1 mmol / l, the initial GFR = 94 ml / min / 1 , 73m ² . No history of neuropathy, stroke, amputation or retinopathy.

According to the calculation of the model, this patient has a low risk of developing CKD after 5 years (22%).

According to the analyzes, after 5 years the patient's GFR = 93 ml / min / 1.73m ² , that is, he did not develop CKD.

Example 4

D. The patient with type 2 diabetes, female, 54 years old, BMI 35.14 kg / ^m2, HbA1c = 9,1%, total cholesterol = 5.6 mmol / l, the initial GFR = 91 ml / min / 1 , 73m ² . History of neuropathy and retinopathy.

The model predicts that this patient will develop high-risk CKD (50%) after 5 years.

According to the analyzes, after 5 years the patient's GFR = 54 ml / min / 1.73m ² , that is, he developed CKD.

Claims (31)

1. A method for predicting the development of chronic kidney disease (CKD) in patients with diabetes mellitus (DM), including determining the type of diabetes mellitus and a set of diagnostically significant indicators: gender (x _gender ), age at the time of diagnosis (x _age ), presence of myocardial infarction history (x _{heart attack} ), history of diabetic coma (x _coma ), presence of diabetic retinopathy (x _retinopathy ), body mass index (BMI) value (x _BMI ), based on the data obtained and in the presence of type 1 diabetes, the likelihood of development is calculated CKD after 5 years in patients with type 1 diabetes (p) according to the formula: p = 1 / (1+

)*100%, where z = -7.079 + 1.182 *

+ 0.063 *

+ 1.148 *

+ 0.617 *

+ 0.903 *

+ 0.040 *

, the indicators are assessed quantitatively as follows: x _gender , assigned "0 points" if there is a male, "1 point" - female; x _age , full years; x _{heart attack} , assign "0 points" in the absence of a history of myocardial infarction, "1 point" - in the presence of a history of myocardial infarction; x _coma , assign "0 points" in the absence of coma in the anamnesis, "1 point" - in the presence of coma in the anamnesis; x _retinopathy , assigned "0 points" in the absence of diabetic retinopathy, "1 point" - in the presence of diabetic retinopathy; x _bmt , kg / m ² , at p≤0.25, the probability of developing CKD within 5 years is 25% or less, which corresponds to a low risk, at 0.25 <p≤0.5, the probability of developing CKD within 5 years is 25-50%, which corresponds to the average risk, at p> 0.5, the probability of developing CKD within 5 years is more than 50%, which corresponds to a high risk. 2. A method for predicting the development of chronic kidney disease (CKD) in patients with diabetes, including determining the type of diabetes and a set of diagnostically significant indicators: gender (x _gender ), age at the time of diagnosis (x _age ), BMI value (x _BMI ), GFR (x _SCF ), glycated hemoglobin value (x _HbA1c ), total cholesterol (x _{Total cholesterol} ), history of _neuropathy (x neuropathy), history of stroke (x _stroke ), diabetic retinopathy (x _retinopathy ), amputation (x _amputation ) and oncological diseases in history (x _oncology ), based on the data obtained and in the presence of type 2 diabetes, the probability of developing CKD after 5 years in patients with type 2 diabetes (p) is calculated using the formula: p = 1 / (1+

)*100%, where z = -13.994 + 1.983 *

+ 0.125 *

+ 0.061 *

+ 0.049 *

+ 0.135 *

+ 0.012 *

+ 0.367 *

+ 0.467 *

+ 0.476 *

+ 0.811 *

+ 0.909 *

, the indicators are assessed quantitatively as follows: x _gender , assigned "0 points" if there is a male, "1 point" - female; x _age , full years; x _bmt , kg / m ² ; x _HbA1c ,%; x_{General CS}, mmol / l; x _skf, ml / min / 1.73m ² ; NS_neuropathy, assign "0 points" in the absence of a history of neuropathy, "1 point" - in the presence of a history of neuropathy; x _stroke , assign "0 points" in the absence of a history of stroke, "1 point" - in the presence of a history of stroke; x _amputation , assigned "0 points" in the absence of amputation, "1 point" - in the presence of amputation; x _retinopathy , assigned "0 points" in the absence of diabetic retinopathy, "1 point" - in the presence of diabetic retinopathy; x _oncology , assign "0 points" in the absence of cancer in the anamnesis, "1 point" - in the presence of cancer in the anamnesis, at p≤0.25, the probability of developing CKD within 5 years is 25% or less, which corresponds to a low risk, at 0.25 <p≤0.5, the probability of developing CKD within 5 years is 25-50%, which corresponds to the average risk, at p> 0.5, the probability of developing CKD within 5 years is more than 50%, which corresponds to a high risk.