Statistical methods as the basis of evidence-based medicine. Their role in the analysis of public health and the activities of health care institutions.
The history of the emergence of preventive disciplines in our country and abroad. The role of N.A. Semashko and Z.P. Solovieva, G.A. Batkis, Yu.P. Lisitsyn and others.
Hosted on ref.rf
in the development and development of prevention.
Social hygiene in the former USSR begins its history with the organization of the Museum "Social Hygiene" of the People's Commissariat of Health of the RSFSR, the director of which was the famous hygienist Prof. A.V. Among the first social hygienists, theoreticians and scientists were such organizers of the health of the people as N.A. Z. P. Solovyov is a doctor, a well-known figure in public medicine. In 1922, N.A. Semashko, with the support of Z.P. Solovyov, A.V. Molkova, N.A. Sysina, S.I. Kaplun and others.
Hosted on ref.rf
authoritative scientists and public figures. Hygiene organized at I Mos.Univer.department of social. hygiene. In 1923, Z.P. Soloviev and his staff created the department of social hygiene at the honey. fak. II Mos. University.
Hosted on ref.rf
In 1941, the department of social sciences. hygiene departments were renamed into the departments of the health organization. This situation had the most detrimental effect on the development of the science of OZiZ. In 1966ᴦ. appointed Minister of Health.
Hosted on ref.rf
USSR B.V. Petrovsky signed an order on the transformation of departments into departments of social sciences. hygiene and health organizations. In October 1999ᴦ. at a meeting of specialized departments with the participation of the Ministry of Health of the Russian Federation, it was decided to rename the discipline into ʼʼPublic Health and Healthcareʼʼaccording to the order of the Minister of Health of the Russian Federation.
Preventive direction in health care in Russia. The system of prevention in medicine and health care. The main types of prevention. Primary prevention as the basis for maintaining public health.
Prevention is one of the most important areas of healthcare, it makes it possible to preserve the health of the population. Prevention (Greek protection, prevention) is a broad and versatile field of activity related to identifying the causes of diseases and injuries, their eradication (weakening) and preventing their manifestation among individuals, groups and the entire population. Types of prevention: 1) individual; 2) public. Given the dependence on the nature of the object of application of preventive measures: 1) primary - preventive measures aimed at the immediate cause of the disease or damage; 2) secondary - measures to influence the conditions and factors that contribute to the development of an already existing disease or injury. By the nature of preventive measures: 1) socio-economic; 2) medical. The history of prevention: begins with the emergence of views on the importance of preventing the causes, conditions, factors of diseases, preventing their severe course, complications. At the dawn of medicine, these were simple, accessible hygienic prescriptions: personal hygiene rules, keeping the body clean, fumigating the sick and their clothes, burning the clothes of the dead, corpses and care items. hygienic recommendations were from ancient physicians. Hippocrates created the doctrine of the role of natural conditions and lifestyle on human health. M.L. spoke about prevention. Mudrov (beginning of the 19th century), I.I. Pirogov, G.A. Zakharyin, but the socio-political system of Russia at the end of the 19-beginning. 20th century prevented the creation and development of preventive health care.
in March 1919, at the 8th Congress of the RCP (b), a program was adopted on the leading direction of health care - preventive care. The role of prevention in health care in our country was fully reflected in the Constitution of the USSR (1977). In the fundamentals of the legislation of the Russian Federation on the protection of the health of citizens (1993), among the basic principles of public health is the "priority of preventive measures."
Status and features of prevention in the context of health care reform. Priority national project ʼʼhealthʼʼ, main directions, its role in the implementation of the preventive direction.
Within the framework of the project, three main areas of work were identified: increasing the priority of primary health care, strengthening the preventive focus of health care and expanding the availability of high-tech medical care. It was planned to focus on strengthening primary health care (municipal clinics, district hospitals) - increasing the salaries of district doctors and nurses, equipping these medical facilities with the necessary equipment, retraining general practitioners, and introducing birth certificates.
The area most in need of support in the healthcare sector is primary health care. Here, 80% of all medical care is provided and the maximum number of units of medical and diagnostic equipment is located; patients most often turn here. Nevertheless, in our country for a long time there was a priority in the development of a stationary link, specialized care, primarily because they were financed from separate sources. The municipalities that fund the outpatient clinic have the most modest capacity today, and therefore the volume of funds received by primary care has not been significant in recent decades. For this reason, the increase in the material base has become an important step in improving the quality of the work of medical institutions. First of all, within the framework of the project, the average level of remuneration of various categories of medical workers has significantly increased (by 80-100% on average).
The quality of primary health care began to improve in other indicators as well. Within the framework of the project, there are programs for the training and retraining of doctors, and there is a positive trend in the change in the number of medical workers. As a result, in 2006, the part-time coefficient decreased from 1.6 to 1.31 in Russia as a whole, and the proportion of persons of retirement age among primary health care workers decreased. And this change is due not only to the arrival of graduates of medical universities and colleges, but also, for example, the creation in some regions of Russia of special conditions for young doctors who have expressed a desire to work in the countryside.
Disease prevention is the most effective remedy for disease. Preventive health care has always been a priority in our country. Within the framework of the priority national project ʼʼHealthʼʼ, state financial support for preventive measures was determined. At the moment, the federal budget has again assumed these obligations.
The main goals were the prevention of HIV and hepatitis B and C, as well as the immunization of the population within the framework of the national vaccination schedule. At the same time, additional medical examinations, medical examinations, screening studies of pregnant women and newborns were introduced. As a result, according to many estimates, in 2006 the incidence of these infections significantly decreased, and the percentage of diseases diagnosed at an early stage increased.
The emergence of a program of birth certificates can also be attributed to the preventive direction. The birth certificate was created with the aim of additional financial support for the activities of health care institutions, providing the right to pay for medical care services. At the same time, its main task is to increase the live birth rate. Infant mortality rates in 2006 (10.21 per 1,000 births) decreased compared to previous years (in 2005 - 11). However, it is planned to achieve international live birth standards while improving the prevention of miscarriage. Of great importance is the increase in the effectiveness of diagnostic and therapeutic measures, the reduction in the number of paid services for the examination and treatment of the pathology of pregnant women.
The main causes of death in Russia are the result of the impact of four to five factors: traffic injuries and road accidents, acute poisoning, oncology, cardiovascular pathology, and the quality of obstetric services. For the next three years, the priority national project "Health" has identified two basic priority areas: the fight against cardiovascular pathologies and road traffic injuries. In 2007 - 2008. it is planned to spend 13.5 billion rubles1 on the prevention of these problems.
Availability of the latest technologies in the field of medicine
As part of a priority national project, the government of the Russian Federation has decided to create high-tech medical centers, and these will probably be one of the most modern centers in the world. To date, 80% of patients receiving high-tech medical care in the federal centers of Moscow and St. Petersburg are residents of these cities and their regions. In the coming years, such centers will be built in Khabarovsk, Krasnoyarsk, Irkutsk and other cities. Οʜᴎ will be distributed evenly and will help bring healthcare to a new level.
There is a point of view that the construction of new centers is not extremely important, it is enough to find additional funds to finance the medical care that is already provided in existing medical institutions. For this reason, for the first time, the federal government has placed an order for 128,000 surgeries1 to be performed in federal specialized medical facilities. Next year this figure will increase to 170,000 operations.
Since 2007, for the first time, medical institutions of the constituent entities of the Russian Federation will be invited to participate in competitions for high-tech operations. By this time, they will have to go through several preparatory stages: complete the approval of clinical standards for the provision of medical care, demonstrate the readiness of the material base and human resources.
In 2006, according to the results of competitive events within the framework of the PNP ʼʼZdorovyeʼʼ, contracts were concluded for the supply of diagnostic equipment with domestic (54%) and foreign (46%) manufacturers1.
4. The importance of studying public health and its conditionality in solving professional problems. Scheme of the study of general health. Key indicators .. It studies the patterns of public health, its conditionality in order to develop evidence-based medical and preventive measures of a strategic, tactical nature to preserve and promote health and improve medical care for the population
1) problems of protecting health and improving the health of various age-gender, social and professional groups and society as a whole
2) Scientifically based optimal methods of healthcare management, forms and methods of work of medical institutions, ways to improve the quality of medical care
historical, expert, budgetary, statistical, sociological methods, organizational experiment͵ economic methods, planning methods, Drawing up an individual program, drawing up prospective programs for social and hygienic monitoring.
Population health indicators: natural movement of the population (demographic indicators, morbidity of the population, disability, physical development; groups: socio-economic, lifestyle, biological, physiographic (natural-climatic).
The use of statistics in preventive medicine. Types of statistical quantities, their use in medicine and health care. Graphic representations of statistical values.
Statistical methods as the basis of evidence-based medicine. Their role in the analysis of public health and the activities of health care institutions.
Statistics-1) is public. science cat. studies the quantitative side of social, mass phenomena in inextricable connection with their qualitative side. 2) is the collection of digital, statistical data that characterizes this or that social phenomenon or process. 3) are the numbers themselves that characterize these phenomena and processes .WITH. has its own methods: m-d mass observation, groupings, tables and graphs. The main task of S. is to establish patterns of the studied phenomena. He studies the quantitative patterns of a continuously changing, developing social life.
Medical statistics - public. science, cat. studies quantity. side of mass phenomena and processes in medicine.
Main sections: \u003d health statistics) sanitary
health statistics) statistics
Health statistics - study. public health in general and separate
its groups and set. depend on health from decomp. social factors. environment.
Health statistics - analyzes data about the network of medical. and
sanit. institution, their activities and personnel, evaluate the effect of decomp. measure-
ty according to the profile and treat diseases.
Statistical population, definitions, types. Unit
observations, records.
Statistical collection - a group or set of relatives
homogeneous elements, i.e., units taken together in concret.
the boundaries of time and space and possessing signs
similarities and differences.
TYPES: 1) GENERAL - comp. from all units of observation;
2) SELECTIVE - part of the general council, which is determined
Xia special methods, possess. signs of similarity and difference (quant.-
expressed as a number, e.g. age; and qualities. - attributive, expressive.
verbally, eg.
Hosted on ref.rf
floor. prof.);
THE UNIT OF OBSERVATION is the primary element of the stat-coy owl-ti, oblad. signs of similarity (sex, age, place of residence; no more than 4x) and differences (quantity and quality.). The sum of the units was research object.
Signs of difference, called. ACCOUNTING SIGNS, yavl. the subject of their analysis:
By nature: \u003d qualities-mi- (attribute) expression. verbally and to them.
def. character (gender, prof).
Quantity, expressed by a number (for example, ascend).
By role: \u003d factorial - affect changes depending on
them signs.
Effective - depend on factorial.
The main properties of statisticians and their statistic. specifications.
PROPERTIES:
one). The distribution of the sign in statistical. owls.
M. b. expressed in absolute numbers and relates. indicators (extensive, intensive, ratios, dynamic series).
2). Average level of features
Har-Xia diff. average values (mode, median, arithmetic mean, weighted mean).
3). Diversity (variability) of traits.
Har-Xia values - limit, amplitude, average. quadratic deviation, coefficient variations).
4). Reliability of signs (representativeness).
Calc. errors of average values, calculation of the boundaries of fluctuations in average. values, comparison of avg. pok-ley.
5). The relationship of m / d signs (correlation)
Har-Xia with pom. coefficient correlations.
Graphic images. Types of graphics images. Rules for constructing graphics. constructions. Application in health care.
Graph. images are used to visualize the statistics. values, allow analyze them in depth.
Called charts. conditional images of numerical values (average and relative) in the form of various geometers.
Hosted on ref.rf
samples (lines, flat, volume. figures)
Distinguish basic. graph types. image:
Charts (linear, radial, bar, intracolumn, sector, curly)
Cartograms
Cartograms
Construction rules:
Presence of a name
Presence of conditional images
Scale
1.LINEAR diagrams approx. to depict the dynamics of a phenomenon or process (for example, the growth of our world, the dynamics of childhood death). In the event that one diagram depicts several phenomena, the lines are applied in different colors. NOT recommended. more than 4 lines.
2. A RADIAL (or polar) chart is based on a system of polar coordinates when depicting a phenomenon in a closed time cycle (day, week, year).
FLAT:
one). POST (rectangular) approx. to depict the dynamics or tactics of a phenomenon. For example, provide us with doctors in the department. countries in defined year. Intra-column - for example, for the row of diseases by class.
2). SECTOR approx. e.g. for pic.
Hosted on ref.rf
Forget the pages or the reasons for the death of us-I, where each reason is occupied. resp. sector in the head of the appanage. weight.
3). FIGURED (volumetric). it has an extra. figures depicted in the form of decomp. figures.
Hosted on ref.rf
For example, an increase in the number of beds in the form of schemes. beds.
4). CARTOGRAMM is an image of statistical. values per geogr.
Hosted on ref.rf
map. To display user data. paint (or shading) of the same color but different intensities.
5). CARTODIAGRAM is an image on geogr.
Hosted on ref.rf
chart map dec. kind.
Statistical methods as the basis of evidence-based medicine. Their role in the analysis of public health and the activities of health care institutions. - concept and types. Classification and features of the category "Statistical methods as the basis of evidence-based medicine. Their role in the analysis of public health and the activities of healthcare institutions." 2017, 2018.
Problems of health and ecology
diobiology, prof. E. B. Burlakova. These data form new ideas about the biological effectiveness of chronic exposure to radiation on humans and unequivocally indicate the incompetence of extrapolating the effects of high doses of ionizing radiation to the region of low doses.
The development of new concepts is important for the formation of balanced plans for the development of nuclear energy and a fair social policy in relation to the liquidators of the Chernobyl disaster and residents of areas contaminated with radionuclides.
When evaluating the effect of radiation on human health, it should be borne in mind that ionizing radiation is a cosmogenic factor in the environment. It is well known that the natural radiation background is necessary for the growth, development and existence of various living beings, including mammals. Understanding radiobiological patterns is associated with insight into the essence of the phenomenon of life, the connection between living things and the cosmos. There are many mysteries in the effects of ionizing radiation, including the positive or negative effect of irradiated biological objects on non-irradiated ones. Of undoubted interest is the idea expressed by A. M. Kuzin in his last note to the staff: “Life, a living body, is a metabolizing system of structures at the molecular level that make up a single whole thanks to information continuously delivered by secondary, biogenic radiation arising under the influence of atomic radiation natural radioactive background of cosmic and terrestrial origin.
REFERENCES
1. Yu. B. Kudryashov, Radiation Biophysics. Ionizing radiation / Yu. B. Kudryashov. - M .: ed. Moscow un-ta, 2004. - 580 p.
2. Yarmonenko, S. P. Radiobiology of man and animals / S. P. Yarmolenko, A. A. Vainson. - M.: Higher. school, 2004. - 550 p.
3. Mothersill, C. Low-dose radiation effects: Experimental hematology and the changing paradigm / C. Mothersill, C. Seymour // Experimental Hematology. - 2003. - No. 31. - S. 437-445.
4. Lee, D.E. The effect of radiation on living cells / D. E. Lee. - M.: Gosatomizdat, 1966. - 288 p.
5. Timofeev-Resovsky, N. V. Application of the hit principle in radiobiology / N. V. Timofeev-Resovsky, V. I. Ivanov, V. I. Korogodin. - M.: Atomizdat, 1968. - 228 p.
6. Goncharenko, E. N. Chemical protection from radiation injury / E. N. Goncharenko. - M.: ed. Moscow un-ta, 1985. - 248 p.
7. National report “20 years after the Chernobyl disaster: consequences in the Republic of Belarus and their overcoming” / Committee on problems of the consequences of the catastrophe at the Chernobyl nuclear power plant under the Council of Ministers of the Republic of Belarus; ed. V. E. Shevchuk, V. L. Guravsky. - 2006. - 112 p.
8. Vozianov, A. Health erects of Chornobyl accident, Eds / A Vozianov, V. Bebeshko, D. Bayka. - Kyiv.: "DIA", 2003. - 508 p.
9. Kuzin, A. M. Structural-metabolic hypothesis in radiobiology / A. M. Kuzin. - M.: Nauka, 1970. - 170 p.
10. Kuzin, A. M. Structural and metabolic theory in radiobiology / A. M. Kuzin. - M.: Nauka, 1986. - 20 p.
11. Knyazeva, E. N. Foundations of synergetics / E. N. Knyazeva, S. P. Kurdimov. - St. Petersburg: Publishing House Aleteyya, 2002. - 31 p.
12. Stepanova, S.I. Biorhythmological aspects of the problem of adaptation / S.I. Stepanova. - M.: Nauka, 1986. - 244 p.
13. Nonmonotonicity of the metabolic response of mammalian cells and tissues to the effect of ionizing radiation / I. K. Kolomiytsev [et al.] // Biophysics. - 2002. - T. 47, Issue. 6. - S. 1106-1115.
14. Kolomiytseva, I. K. Nonmonotonous changes in metabolic parameters of tissues and cells under action ionizing radiation on animals / I. K. Kolomiytseva, T. R. Markevich, L. N. Potekhina // J. Biol. Physics. - 1999. - No. 25. - S. 325-338.
15. E. B. Burlakova, E. B. Burlakova, A. N. Goloshchapov, G. P. Zhizhina, and A. A. Konradov, New aspects of regularities in the action of low-intensity irradiation at low doses, Radiats. biology. Radioecology. - 1999. - T. 39. - S. 26-34.
Received 04/18/2008
USE OF EVIDENCE-BASED MEDICINE DATA IN CLINICAL PRACTICE (literature review)
A. L. Kalinin1, A. A. Litvin2, N. M. Trizna1
1Gomel State Medical University 2Gomel Regional Clinical Hospital
A brief overview of the principles of evidence-based medicine and meta-analysis is given. An important aspect of evidence-based medicine is to determine the degree of reliability of information.
Quantitative pooling of data from different clinical trials using meta-analysis provides results that cannot be obtained from individual clinical trials. Reading and studying systematic reviews and meta-analyses allows you to more effectively navigate a large number of published articles.
Keywords: evidence-based medicine, meta-analysis.
Problems of health and ecology
USE OF DATA OF EVIDENCE BASED MEDICINE IN CLINICAL PRACTICE
(literature review)
A. L. Kalinin1, A. A. Litvin2, N. M. Trizna1
1Gomel State Medical University 2Gomel Regional Clinical Hospital
The purpose of the article is the review of principles of evidence based medicine and the meta-analysis. A prominent aspect of evidence based medicine is a definition of the degree of reliability of the information.
Quantitative association of the given various clinical researches by means of the meta-analysis allows to receive results which cannot be received from separate clinical researches. Reading and studying of systematic reviews and results of the meta-analysis allows to be guided more effectively in a considerable quantity of published articles.
Key words: evidence based medicine, meta-analysis.
No practitioner has sufficient experience to navigate freely in all the variety of clinical situations. It is possible to rely on expert opinions, authoritative guides and reference books, but this is not always reliable due to the so-called lag effect: promising medical methods are introduced into practice after a considerable time after evidence of their effectiveness has been obtained. On the other hand, information in textbooks, manuals and reference books is often outdated even before they are published, and the age of the experienced doctor conducting the treatment is negatively correlated with the effectiveness of the treatment.
The half-life of literature reflects the intensity of progress. For medical literature, this period is 3.5 years. Only 1015% of the information published today in the medical press will have scientific value in the future. After all, if we assume that at least 1% of the 4 million articles published annually have something to do with the medical practice of a doctor, he would have to read about 100 articles every day. It is known that only 10-20% of all medical interventions currently used were based on sound scientific evidence.
The question arises: why do doctors not apply good evidence in practice? It turns out that 75% of doctors do not understand statistics, 70% do not know how to critically evaluate published articles and studies. Currently, in order to practice evidence-based data, a doctor must have the knowledge necessary to assess the reliability of the results of clinical trials, have quick access to various sources of information (primarily international journals), have access to electronic databases (Medline), and be fluent in English.
The purpose of this article is a brief overview of the principles of evidence-based medicine and its component - meta-analysis, which allows you to more quickly navigate the flow of medical information.
The term "Evidence Based Medicine" was first proposed in 1990 by a group of Canadian scientists from McMaster University in Toronto. The term quickly took root in the English-language scientific literature, but at that time there was no clear definition of it. Currently, the following definition is the most common: “Evidence-based medicine is a branch of medicine based on evidence, involving the search, comparison, generalization and wide dissemination of the evidence obtained for use in the interests of patients” .
Today, evidence-based medicine (EBM) is a new approach, direction or technology for collecting, analyzing, summarizing and interpreting scientific information. Evidence-based medicine involves conscientious, explicable and common-sense use of the best modern achievements for the treatment of each patient. The main goal of introducing the principles of evidence-based medicine into healthcare practice is to optimize the quality of medical care in terms of safety, efficiency, cost and other significant factors.
An important aspect of evidence-based medicine is the determination of the degree of reliability of information: the results of studies that are taken as a basis for compiling systematic reviews. The Center for Evidence-Based Medicine at Oxford has developed the following definitions of the degree of reliability of the information provided:
A. High certainty - information based on the results of several independent clinical trials (CTs) with agreement between the results summarized in systematic reviews.
Problems of health and ecology
B. Moderate Reliability - The information is based on the results of at least several independent trials that are similar in purpose.
C. Limited Reliability - The information is based on the results of a single CT.
D. There is no rigorous scientific evidence (CTs have not been conducted) - some statement is based on the opinion of experts.
According to modern estimates, the reliability of evidence from different sources is not the same and decreases in the following order:
1) randomized controlled CT;
2) non-randomized CT with simultaneous control;
3) non-randomized CT with historical control;
4) cohort study;
5) case-control study;
6) cross CI;
7) results of observations;
8) description of individual cases.
Three "pillars" of reliability in clinical medicine are: random blind selection of subjects in comparison groups (blind randomization); sufficient sample size; blind control (ideally - triple). It must be specially emphasized that the incorrect, but widely used term "statistical reliability" with its notorious p<... не имеет к вышеизложенному определению достоверности никакого отношения . Достоверные исследования свободны от так называемых систематических ошибок (возникающих от неправильной организации исследования), тогда как статистика (р <...) позволяет учесть лишь случайные ошибки .
In clinical medicine, randomized controlled trials (RCTs) have become the "gold standard" for testing the effectiveness of interventions and procedures. The process of "blinding" the test participants is designed to eliminate the systematic error of the subjective assessment of the result, because it is natural for a person to see what he wants and not to see what he does not want to see. Randomization should solve the problem of the diversity of the subjects, ensuring the genetic completeness of the "abstract representative of the general population", to which the result can then be transferred. Specially conducted studies have shown that the lack of randomization or its incorrect conduct leads to an overestimation of the effect by up to 150%, or to its underestimation by 90%.
It is extremely important to emphasize that RCT technology allows you to get four answers about the effect of the intervention without any
knowledge of its mechanism. It allows us to reasonably assert from the standpoint of evidence-based medicine that the intervention is 1) effective; 2) useless; 3) harmful; or, in the worst case, that 4) to date, nothing can be said about the effectiveness of this type of intervention. The latter occurs when the intervention of interest to us, due to the small number of participants in the experiment, did not allow us to obtain a statistically significant result in an RCT.
Thus, DM answers the already mentioned questions: it works (harmfully or usefully) / does not work (uselessly) / unknown; but does not answer the questions "how and why it works." Only fundamental research can answer them. In other words, DM for its own purposes can do without fundamental research, while fundamental research cannot do without the procedure for testing the effect according to DM standards in order to implement its results in everyday medical practice.
To optimize the analysis of evidence-based information, special methods of working with information are used, such as a systematic review and meta-analysis. Meta-analysis (metaanalysis) - the use of statistical methods in the creation of a systematic review in order to summarize the results included in the review of studies. Systematic reviews are sometimes called meta-analyses if this method was used in the review. Meta-analysis is carried out in order to summarize the available information and disseminate it in a way that is understandable to readers. It includes the definition of the main goal of the analysis, the choice of methods for evaluating the results, a systematic search for information, the generalization of quantitative information, its analysis using statistical methods, and the interpretation of the results.
There are several varieties of meta-analysis. Cumulative meta-analysis allows you to build a cumulative accumulation curve of estimates as new data become available. A prospective meta-analysis is an attempt to develop a meta-analysis of planned trials. Such an approach may be acceptable in areas of medicine where there is already an established network of information exchange and collaborative programs, such as the Oratel electronic information system developed by WHO for monitoring the quality of dental care for the population. In practice, instead of a prospective meta-analysis, a prospective-retrospective meta-analysis is often used, combining new results with previously published ones. Meta-analysis of individual data is based on the study of the results of treatment of individual patients,
Problems of health and ecology
it requires the cooperation of many researchers and strict adherence to the protocol. In the near future, meta-analysis of individual data is likely to be limited to the study of major diseases, the treatment of which requires large-scale centralized investments.
The main requirement for an informative meta-analysis is to have an adequate systematic review that examines the results of numerous studies on a specific problem according to the algorithm:
Selection of criteria for inclusion of original studies in the meta-analysis;
Assessment of heterogeneity (statistical heterogeneity) of original studies;
Actually meta-analysis (generalized estimate of the effect size);
Analysis of the sensitivity of the conclusions.
The results of a meta-analysis are usually presented as a graph in the form of point estimates with an indication of the confidence interval and the odds ratio (^dds ratio), a summary indicator that reflects the severity of the effect (Figure 1). This allows you to show the contribution of the results of individual studies, the degree of heterogeneity of these results, and a generalized estimate of the effect size. The results of meta-regression analysis can be presented in the form of a graph, along the abscissa axis of which the values of the analyzed indicator are plotted, and along the ordinate axis - the magnitude of the therapeutic effect. In addition, the results of the sensitivity analysis for key parameters should be reported (including a comparison of the results of applying fixed and random effects models, if these results do not match).
Figure 1 - Funnel plot to identify bias associated with predominantly publishing positive study results
The graph shows the data of a meta-analysis on the evaluation of the effectiveness of one of the treatments. The relative risk (RR) in each study is compared to the sample size (study weight). The points on the graph are grouped around the weighted average of the RR (shown by the arrow) in the form of a symmetrical triangle (funnel), within which the data of most studies are placed. Published data from small studies appear to overestimate the treatment effect compared to larger studies. The skewed distribution of points means that some small studies with negative results and significant
variance were not published, i.e., a systematic error associated with the predominant publication of positive results is possible. The graph shows that there are significantly fewer small (10-100 participants) studies with an RR greater than 0.8 than similar studies with an RR less than 0.8, and data from medium and large studies are distributed almost symmetrically. Thus, some small studies with negative results have probably not been published. In addition, the graph makes it easy to identify studies whose results differ significantly from the general trend.
Problems of health and ecology
In most cases, when conducting a meta-analysis, generalized data on the compared groups of patients are used in the form in which they are given in the articles. But sometimes researchers seek to evaluate outcomes and risk factors in individual patients in more detail. These data may be useful in the analysis
survival and multivariate analysis. Meta-analysis of individual patient data is more expensive and time-consuming than meta-analysis of group data; it requires the cooperation of many researchers and strict adherence to the protocol (Figure 2).
A. Graphical representation of standard meta-analysis results. The relative risk of progression in each study and its pooled estimate are presented as dots, and confidence intervals (CI; typically 95% CI) are depicted as horizontal lines. Studies are presented according to the date of publication. Relative risk<1 означает снижение числа исходов в группе лечения по сравнению с группой контроля. Тонкие линии представляют совокупные индивидуальные результаты, нижняя линия - объединенные результаты.
B. Results of a cumulative meta-analysis of data from the same studies. Dots and lines represent, respectively, relative risk values and 95% CI pooled data after each additional study was included in the analysis. If the confidence interval crosses the line OR = 1, then the observed effect is not statistically significant at the chosen significance level of 0.05 (95%). If there is no significant data heterogeneity, the CI narrows when a follow-up study is added.
N is the number of patients in the study; N is the total number of patients.
Figure 2 - Results of standard and cumulative meta-analysis of data from the same studies
In most meta-analysis summary tables, summaries of all trials are presented as a diamond (bottom horizontal line with a dot). The location of the diamond in relation to the vertical line of no effect is fundamental to understanding the effectiveness of the test. If the diamond overlaps the line of no effect, it can be said that there is no difference between the two treatments in the impact on the primary outcome rate.
An important concept for the correct interpretation of the results of a meta-analysis is the definition of the homogeneity of trials. In the language of meta-analysis, homogeneity means that the results of each individual trial are combined with the results of others. Homogeneity can
evaluate at a glance by the location of horizontal lines (Figure 2). If the horizontal lines overlap, these studies can be said to be homogeneous.
To assess the heterogeneity of trials, the numerical value of the criterion %2 is used (in most meta-analysis formats it is referred to as "Chi-squared for homogeneity"). The %2 statistic for group heterogeneity is explained by the following rule of thumb: the x2 criterion has, on average, a value equal to the number of degrees of freedom (number of trials in the meta-analysis minus one). Therefore, an X2 value of 9.0 for a set of 10 trials indicates no evidence of statistical heterogeneity.
Problems of health and ecology
With significant heterogeneity in the results of studies, it is advisable to use a regression meta-analysis, which allows you to take into account several characteristics that affect the results of the studied studies. For example, a detailed assessment of outcomes and risk factors in individual patients is necessary in the analysis of survival and multivariate analysis. The results of the regression meta-analysis are presented as a slope factor with a confidence interval.
Software is available on the Internet for computer meta-analysis.
Free programs:
RevMan (Review Manager) is located at: http://www.cc-ims.net/RevMan;
Meta-Analysis version 5.3: http://www.statistics. com/content/freesoft/mno/metaana53.htm/;
EPIMETA: http://ftp.cdc.gov/pub/Software/epimeta/.
Paid programs:
Comprehensive Meta-Analysis: http://www. meta-analysis.com/;
MetaWin: http://www.metawinsoft.com/;
WEasyma: http://www.weasyma.com/.
Statistical software packages that provide the opportunity to conduct a meta-analysis:
SAS: http://www.sas.com/;
STATA: http://www.stata. com/;
SPSS: http://www.spss.com/.
Thus, the quantitative combination of data from various clinical studies using meta-analysis allows you to get results that cannot be extracted from individual clinical studies. Reading and studying systematic reviews and meta-analyses allows you to more quickly navigate the avalanche of published articles and, from the point of view of evidence-based medicine, select those few that really deserve our time and attention. At the same time, it is necessary to realize that meta-analysis is not a magic wand that solves the problem of scientific evidence, and should not be used as a substitute for clinical reasoning.
REFERENCES
1. Systematic reviews and meta-analysis for the surgeon scientist / S. S. Mahidl // Br. J. Surg. - 2006. - Vol. 93. - P. 1315-1324.
2. A comparison of results of meta-analyses of randomized control trials and recommendations of clinical experts / E. T. Antman // JAMA. - 1992. - Vol. 268, No. 2. - P. 240-248.
3. Evidence based medicine: what it is and what it isn’t / D. L. Sack-ett // BMJ. - 1996. - Vol. 312. - P. 71-72.
4. Egger, M. Meta-analysis: potentials and promises / M. Egger,
S. G. Davey // BMJ. - 1997. - Vol. 315. - P. 1371-1374.
5. Yuriev, K. L. Evidence-based medicine. Cochrane Collaboration / K. L. Yuryev, K. N. Loganovsky // Ukr. honey. cha-sopis. - 2000. - No. 6. - S. 20-25.
6. The Cochrane database of systematic reviews. - London: BMJ Publishing Group and Update Software, 1995. - 260 p.
7. Davies, H. What is a meta-analysis? / H. Davies, I. Crombie // Clinical pharmacology and pharmacotherapy. - 1999. - No. 8. - C. 10-16.
8. Egger, M. Meta-analysis: principles and procedures / M. Egger, S. G. Davey, A. N. Phillips // BMJ. - 1997. - Vol. 315. - P. 1533-1537.
9. Lewis, S. Forest plots: trying to see the wood and the trees / S. Lewis, M. Clarke // BMJ. - 2001. - Vol. 322. - P. 1479-1480.
10. Bero, L. The Cochrane Collaboration. Preparing, maintaining, and disseminating systematic reviews of the effects of health care / L. Bero, D. Rennie // JAMA. - 1995. - Vol. 274. - P. 1935-1938.
11. Does the inclusion of gray literature influence estimates of intervention effectiveness reported in meta-analyses? / L.Mc. Auley // Lancet. - 2000. - Vol. 356. - P. 1228-1231.
12. Fleiss, J. L. The statistical basis of meta-analysis / J. L. Fleiss // Stat. Methods Med. Res. - 1993. - Vol. 2. - P. 121-145.
13. Greenland, S. Invited commentary: a critical look at some popular meta-analytic methods / S. Greenland // Am. J. epidemiol. -
1994. - Vol. 140. - P. 290-296.
14. Guidelines for meta-analyses evaluating diagnostic tests / L. Irwig // Ann. Intern. Med. - 1994. - Vol. 120. - P. 667-676.
15. Stewart, L. A. Practical methodology of meta-analyses (overviews) using updated individual patient data. Cochrane Working Group / L. A. Stewart, M. J. Clarke // Stat. Med. - 1995. - Vol. 14. - P. 2057-2579.
16. Grinkhalkh T. Fundamentals of evidence-based medicine / T. Grinkhalkh; per. from English. - M.: GEOTAR - Media, 2006. - 240 p.
17. Olkin, I. Statistical and theoretical considerations in meta-analysis / I. Olkin // J. Clin. epidemiol. - 1995. - Vol. 48. - P. 133-146.
18. Villar, J. Predictive ability of meta-analyses of randomized controlled trials / J. Villar, G. Carroli, J. M. Belizan // Lancet. -
1995. - Vol. 345. - P. 772-776.
19. Deeks, J.J. Systematic reviews in health care: Systematic reviews of evaluations of diagnostic and screening tests / J. J. Deeks // BMJ. - 2001. - Vol. 323.-P.157-162.
Received 01.02.2008
UDC 616.12-005.8-0.53.8-08
STRUCTURE OF ACUTE MYOCARDIAL INFARCTION, AGE AND GENDER CHARACTERISTICS OF THE COURSE AND MORTALITY AT THE HOSPITAL STAGE OF TREATMENT
N. V. Vasilevich
Gomel State Medical University
The structure, dynamics of development of acute myocardial infarction depending on sex, age, terms of admission to the hospital, severity of myocardial damage at the hospital stage of treatment were traced.
Key words: acute myocardial infarction, gender, age, mortality.
CHAPTER V ANALYSIS OF MEDICAL PUBLICATIONS FROM THE POSITION OF EVIDENTIAL MEDICINECHAPTER V ANALYSIS OF MEDICAL PUBLICATIONS FROM THE POSITION OF EVIDENTIAL MEDICINE
Title of the article. An interesting title grabs attention. If it is of interest, you can proceed to further work on the article. Of particular interest are articles and reviews, the title of which contains information on the principle of "for" and "against", since in addition to the possible interesting position of the author, arguments and counterarguments will be given here. Using the list of recommended literature, it will be easy to get acquainted with the primary sources and form your own opinion about the problem (as
For example, the appendix contains the article “Diuretics: proven and unproven”).
The title is always followed by list of authors and name of institution, in which the work was done. Meeting with a familiar and well-known name and a respected institution allows you to imagine in advance the qualitative level of the research. If the article presents the results of an RCT, it is advisable to take the time to find information on the Roszdrav website about whether this institution has a license to conduct research. The presence of a license, as well as experience in similar work, allow us to treat the information contained in the publication with great confidence.
abstract allows you to get an expanded idea of the essence of the study, the contingent of its participants and conclusions. If the data meets the task of finding information, you can proceed to the analysis of the article. In the absence of an abstract, you should immediately familiarize yourself with the conclusions of the study, published at the end of the article.
The title, abstract and conclusions should give an idea of the possible scientific and methodological level of the study, the category of patients and the possibility of applying its results in real practice (for example, the diagnostic capabilities of a polyclinic and specialized centers differ significantly in favor of the latter).
Research methods- one of the key sections of the publication, since it is he who gives an idea of the quality of the results and conclusions, since a poorly planned and executed study using non-standard methods cannot be the basis for decision-making.
Currently, methodological requirements for high-quality clinical trials have been formed:
Presence of a control group (placebo, conventional therapy, comparison intervention);
Criteria for inclusion and exclusion of patients from the study;
Study design (distribution of patients included in the study before and after randomization);
Description of the randomization method;
Description of the principles of drug use (open, blind, double-blind, triple-blind);
. "blind" and independent evaluation of the results of treatment, not only by end points, but also taking into account laboratory and instrumental indicators;
Presentation of the results (special attention is paid to the clinical and demographic comparability of the control and study groups);
Information about complications and side effects of treatment;
Information on the number of patients who dropped out during the study;
Qualitative and adequate statistical analysis using licensed statistical programs;
Presenting the results in a form that can be cross-checked (only percent and delta changes in the indicator are unacceptable);
Indication of a conflict of interest (with which organizations the author collaborates and who was the sponsor of the study).
Quite a few publications meet all of the above requirements, so when analyzing articles, it is necessary not only to state the existing shortcomings, but to assess their impact on the reliability of the conclusions drawn.
Most experts in the field of evidence-based medicine identify the most important components of a quality medical publication.
Use of patient randomization in the study.
In international peer-reviewed journals, randomization is reported in 90% of articles on clinical trials, but only 30% of them describe a specific method of randomization. At present, the mention of the concept of "randomization", especially in domestic works, has become a sign of a "good" tone. However, the methods used are often not such, and cannot ensure the homogeneity of the compared groups. Sometimes the difference in the number of patients in the comparison groups indicates that randomization was not carried out at all. Can not be attributed to the methods of randomization and "distribution of patients into groups randomly." The use of low-quality randomization methods, obvious flaws in the conduct or its absence make further study of the publication useless and meaningless, since the findings will be unproven. The absence of high-quality information on the problem of interest is better than the use of low-quality information in decision making. Unfortunately, in real practice, low-quality studies prevail over high-quality studies.
The main criteria for evaluating the effectiveness of treatment. It is important that the publication uses commonly accepted hard and surrogate disease-specific endpoints. We cannot agree with the opinion of V.V. Vlasov "Unfortunately, the substitution of" final "results (true evaluation criteria - clinical outcomes) by" intermediate "(indirect evaluation criteria such as lowering glucose or cholesterol levels in the blood, blood pressure) is very common." Today, for each nosology, there are strictly defined surrogate endpoints that affect the prognosis of the disease. In a number of studies, the achievement of "hard" endpoints is impossible in principle, so the evaluation of the effectiveness of an intervention by its impact on surrogate endpoints is quite acceptable. Another thing is that they must be chosen correctly: for example, for arterial hypertension, this is the level of blood pressure, and not the state of lipid peroxidation. In general, work on the study of the next isoenzyme, as a rule, has no clinical significance for two reasons: firstly, no one else determines them except the authors, and secondly, the connection with the end "hard" points is almost never proven.
The significance of the results of the study and their statistical significance. Only what happens with a high probability is statistically significant, and the probability must be set before the start of the study. It is clinically significant that it can be used in a wide range of patients. In terms of its effectiveness, it is significantly superior, and in terms of safety it is not inferior to already existing alternative methods of treatment and diagnostics.
The large sample size (number of patients) in large RCTs makes it possible to statistically reliably detect even small effects from the use of study drugs. The small sample size characteristic of most publications does not allow this, so the small effect in them means that only a small proportion of patients (1-2%) will receive a positive effect from the intervention. Evaluating the safety of an intervention in a small number of patients is considered unethical. Decisions should not be made based on a "pronounced trend", they may be the subject of further scientific research, but not the basis for clinical decision making. In addition, the data of correlation and regression analyzes cannot be used as the basis for clinically significant conclusions, since they reflect the direction
the intensity and severity of the relationship of indicators, and not the change as a result of the intervention.
Recently, certain problems have appeared with large-scale studies. The number of their participants is sometimes so large that even a slight deviation of a trait as a result of an intervention can become statistically significant. For example, the ALLHAT trial included 33,357 patients, of whom 15,255 were treated with chlorathalidon and the remainder were treated with amlodipine or lisinopril. By the end of the study, the chlorthalidone group showed an increase in glucose by 2.8 mg/dL (2.2%), and in the amlodipine group it decreased by 1.8 mg/dL (1.3%). These changes, which in real clinical practice might not have been given any significance, turned out to be statistically significant.
The absence of significant differences in the effectiveness of the compared research methods is most often associated with a small number of patients in the sample. An insufficient sample size makes a negative result insufficient for a negative assessment of the treatment, and if a positive effect is obtained, the intervention does not allow it to be confidently recommended for wide clinical practice.
In addition to evaluating the effectiveness of an intervention in relation to "hard" and "surrogate" endpoints, it is important to know about its impact on quality of life (for example, for a patient with pain, a change in this indicator is more important than the effect on the risk of decompensation of chronic heart failure when using NSAIDs).
Availability of the method in real clinical practice.
The physician must decide how comparable the group of patients included in the study with those patients to whom he intends to apply it (demographic characteristics, severity and duration of the disease, comorbidities, proportion of men and women, existing contraindications to diagnostic and / or therapeutic measures, etc.).
The information presented above mainly concerned studies evaluating the effectiveness of new treatments. Publications on diagnostic problems and fundamental problems of the etiology and pathogenesis of diseases have a number of differences both in their essence and in attributive features, which make them informative from the standpoint of evidence-based medicine.
PUBLICATIONS ON DIAGNOSIS
Diagnostic procedures can be used for different purposes:
As a mandatory standard of examination (for example, measuring blood pressure, determining weight, analyzing blood and urine, etc.) is carried out for all persons who find themselves in a medical institution in connection with any disease to exclude concomitant pathology (case finding);
As a screening tool to identify patients in the healthy population (eg, testing for phenylketonuria in a maternity hospital or measuring blood pressure to identify individuals with hypertension);
To establish and clarify the diagnosis (for example, ECG and esophagogastroendoscopy in the presence of pain in the left side of the chest);
For dynamic monitoring of the effectiveness of treatment (for example, daily monitoring of blood pressure during antihypertensive therapy).
In this regard, it is necessary to have clear information in the article about the purpose of the undertaken diagnostic intervention.
To assess the reliability of information about the benefits of the proposed diagnostic intervention, it is necessary to answer a number of questions:
Has the proposed method been compared with the existing "gold standard" for a specific pathology (for example, echocardiography with ECG in coronary artery disease, pulse wave velocity with ultrasound determination of the thickness of the intima-media complex);
Is the comparison method chosen truly the "gold standard";
Have diagnostic interventions been compared with blinding;
Are the limits of the possible application of the diagnostic method given (for example, the first hours of myocardial infarction for troponins, the level of glycated hemoglobin, etc.);
Is the comorbidity widely represented, affecting the effectiveness of diagnostic intervention;
How reproducible is the diagnostic method, and is it "operator" dependent (for example, morphometry with echocardiography).
Physicians overestimate the reproducibility of imaging studies (ultrasound, x-ray, radioisotope, electrocardiographic, and endoscopic);
On the basis of what tests the norm and pathology were distinguished.
The concept of the norm and the point of separation must be clearly articulated. The separation point is the value of the physiological indicator, which serves as the boundary separating people into healthy and sick. So, the values of 140/90 and 130/80 mm Hg can be taken as a normal level of blood pressure. Naturally, depending on this, significant differences can be obtained, for example, in the frequency of left ventricular hypertrophy using any evaluative diagnostic technique. The split point (x2) allows you to evaluate the sensitivity, specificity and predictive value of the diagnostic intervention. Increasing the split point values reduces sensitivity but increases the specificity and predictive value of a positive diagnostic intervention. Accordingly, with a decrease in the value of the split point to the left (x1), the sensitivity and predictive value of a negative result increase, but the specificity and predictive value of a positive result of a diagnostic test decrease. To describe the changes in the results of the study, depending on the choice of the separation point, the so-called ROC-analysis (Receiver Operating Characteristic analysis) is used, which allows you to assess the risk of false positive results.
When analyzing publications on diagnostic interventions, it is necessary to evaluate:
How convincingly it is proved that the use of a new diagnostic test in combination with other tests standard for this pathology increases the efficiency of diagnosis. An ineffective diagnostic intervention will not improve diagnostic performance when added to the existing “diagnostic test battery”. The criterion for the usefulness of a diagnostic test is the ability to positively influence the outcome of the disease with its help (for example, due to earlier or more reliable detection of pathology);
Is it possible to use a new diagnostic intervention in real everyday clinical practice;
What is the risk of a new diagnostic intervention (even a routine diagnostic intervention has its own risk of complications, such as bicycle ergometry, and even more so coronary angiography in IHD);
What is the cost of a new diagnostic intervention when compared with existing ones, and especially with the “gold standard” (for example, the cost of ECG and EchoCG to determine left ventricular hypertrophy differ significantly, but the latter method is much more accurate);
How detailed is the procedure for conducting a diagnostic intervention (preparation of the patient, technique for conducting a diagnostic intervention, methods of storing the information received).
PUBLICATIONS ABOUT THE COURSE OF THE DISEASE
The most difficult to analyze are publications relating to the course of the disease, since they require knowledge in the field of non-infectious epidemiology from the doctor.
Important questions that the doctor must answer when analyzing the quality of the information provided are:
What principle was put in the formation of the study group of patients (ambulance, general or specialized hospital, polyclinic);
Are there clear diagnostic criteria for assigning patients to the study group? For example, in the medical literature there is no clear definition of the concept of vegetovascular dystonia. Thus, completely different patients can fall into the study group;
Whether the criteria for the outcome of the disease are clearly formulated and whether they correspond to those currently accepted. Only a documented death is obvious, although here the cause of death can be seriously influenced by the place where it is ascertained (at home or in a hospital, an autopsy was performed or not). For all other cases, clear criteria should be developed, it is desirable that the endpoints be evaluated by an independent committee of experts (“streaming committee”);
How was the prospective observation of the course of the disease organized (seeing a doctor, hospitalizations, death).
Completeness of tracking is the key to a qualitative study of the course of the disease. If during the observation more than 10% of patients drop out, then the results of such a study are considered doubtful. With more than 20% of patients dropping out, the results of the study do not represent any scientific value at all, since in groups with a high risk of complications and mortality, they simply cannot be tracked. A dedicated independent committee needs to review the reasons for dropping out of each patient:
Who and how (blindly or not) assessed the outcome of the disease;
Whether the impact of comorbidity on endpoints was taken into account. If not, then the available results are significantly distorted by the clinical and demographic characteristics of the study group;
How and with what accuracy the predictive value of symptoms and events was calculated. The probability of development of the studied events (mortality, survival, development of complications) is the main result. It can be represented as a probability or frequency in fractions of one (0.35), in percent (35%), ppm (35?), odds ratio (3.5:6.5). Be sure to indicate the confidence interval, which will allow you to correctly extrapolate the results to the real contingent of patients. At the same time, it is almost always necessary to standardize the data obtained by sex, age, and other clinical and demographic indicators;
Do the results obtained about the course of the disease influence the choice of diagnostic and therapeutic intervention;
Does the characteristics of the study participants correspond to the contingent of patients that the doctor encounters in real clinical practice.
The above criteria for evaluating disease course studies apply only to prospective follow-ups. Retrospective observations almost never stand up to criticism from the standpoint of non-infectious epidemiology and evidence-based medicine. It is for this reason that the results of such studies (especially domestic ones) carried out in the 70-80s of the last century are of no value.
MEDICAL RESEARCH ON THE STUDY OF THE ETIOLOGY AND PATHOGENESIS OF DISEASES
Such research belongs to the field of fundamental medical knowledge. They are based on the analysis of cause-and-effect relationships and most of the errors in them are associated with ignoring the well-known principle “the appearance of something after an event does not mean that it happened as a result of this event”. A classic example of cause-and-effect relationships is the identification of dose-dependent effects. Any evidence-based relationship should be understandable and explainable from the standpoint of epidemiology and general medical knowledge.
Unlike experimental studies, clinical studies have the only opportunity to obtain data on the etiology and pathogenesis of diseases through epidemiological (prospective and case-control) studies. A key role in their interpretation and assessment of the reliability of the results is played by a systematic error due to underestimation of patient selection bias. The deliberate exclusion of a certain group of patients can lead to completely inexplicable results from the point of view of logic. If this happens, it is necessary to re-analyze the clinical and demographic characteristics of the study population.
Among epidemiological studies, prospective studies are the most reliable, free from many possible errors. However, they are extremely expensive and rarely carried out. Much more often, the genesis of diseases is studied in case-control studies (CSCs). The table shows the main requirements for research on the etiology and pathogenesis of diseases. The main standards for conducting such studies are well known (Horwitz R.I., Feinstein A.R. Methodological standards and contradicting results in case-control research. Am J Med 1979;
. a predetermined method of selecting subjects determined before the start of the study with a clear indication of the criteria for inclusion and exclusion of patients from the study;
. a well-defined causal factor under study and a method for its detection;
. undistorted data collection. Individuals who collect information about patients should not be aware of the purpose for which they are collecting it. Classic
an example of the consequences of targeted information gathering is an almost 5-fold increase in the number of patients with cough while taking ACE inhibitors compared with a group of patients who self-reported its occurrence;
. no differences in the collection of anamnesis in the comparison groups. Formalized and, if necessary, validated questionnaires should be used. If a translation questionnaire is used, it is necessary to confirm the accuracy of the translation by its back translation;
. no unnecessary restrictions in the formation of comparison groups;
. no differences in the diagnostic examination of the comparison groups. The control group is guaranteed not to have the pathology under study. Therefore, a set of highly informative diagnostic tests for each pathology should be developed;
. no differences in the frequency and nature of the examination at the prehospital stage of the comparison groups;
. no differences in the demographic characteristics of the comparison groups;
. no differences in other risk factors, except for the studied one, in the comparison groups.
Ideally, a prospective study is necessary to solve the tasks set. However, this will take years and decades, especially if we are talking about a rare pathology. So, if the disease develops in 10 years in 2 out of 1000, then to identify 10 cases, you need to track at least 5000 people over 10 years. In such cases, studies organized on the principle of "case-control" (CSC) are used. They compare the frequency of a factor (for example, obesity) in patients with the pathology of interest and other diseases. To clarify the role of risk factors, populations in different regions with different severity of the presence of this factor can be compared. The least reliable sources for identifying causal relationships are case studies or descriptions of patient groups.
When identifying shortcomings in the publication, it is necessary to try to understand what caused them: ignorance of the basics of research planning and mathematical statistics, deliberately incorrect interpretation of the data, the author’s enthusiasm (“if the facts interfere with the theory, then they can be discarded”) or the interest of the research sponsor.
Typical mistakes in research are:
Absence of "experimental" (with the analyzed intervention) and "control" (receiving placebo or "traditional", "standard" treatment). In the absence of a control group, the article is useless (sometimes even harmful) and should not be read. Currently, we can talk about the following regularity: using such remedies as homeopathy, acupuncture, liposuction, dietary supplements, the authors get impressive results, but the quality of the study is low;
The absence of exclusion criteria does not provide a full opportunity to compare the homogeneity of the experimental and control groups;
The number and reasons for dropping out of patients during the study are not given. Articles with dropouts of more than 20% of patients may not be read;
Lack of "blinding" of the study;
Lack of static analysis details. Bringing only generally accepted indicators (mean, standard deviation, percentage, delta) is insufficient, especially for small groups. To assess the sufficiency of the number of patients for a negative result of the study, you can use special tables. The cell corresponding to the frequency of events in the treatment group and in the control group represents the number of patients in each group required to detect a decrease in the frequency of 5%, 10%, 25%, 50%, etc. If the number of patients in the material under consideration is less, then the effect could not be detected only because of the small number of patients;
Underestimation of confounding factors, such as gender, age, smoking, alcohol consumption, etc. It is well known that the effectiveness of some β-blockers, such as atenolol, is reduced in smokers, while others (bisoprolol) are not. Statistical analyzes should be adjusted for such factors potentially affecting the parameter being estimated. This procedure is called standardization on one or more indicators.
When making a final decision about the possibility of using published data, the doctor must compare how the study's conclusions correspond to existing ideas. The choice in favor of a new method or approach in the treatment and diagnosis
ke should be based not on the doctor's desire to satisfy his professional interest (in this case, at the expense of the patient's health), but on a coherent and indisputable system of evidence of its benefits and safety.
A critical approach to scientific data is the basis of the power of progress in any field of knowledge, including medicine.
Literature
1. How to Read Clinical Journals: 1. Why to Read them and How to start
Reading them Critically. Can Med Ass J 1981; 124:555-558.
2. Currie B.F. Continuing education from medical periodicals. J Med Educ
3. Statistical Methods in Cancer Research: Part 2. The design and analysis of cohort studies. IARC Sci Publ. N.82. Lyon: WHO, IARC, 1987: 1-406.
4. Bailar J.C., Louis T.A., Lavori P.W., Polansky M. A classification for
biomedical research reports. N Engl J Med 1984; 311:23:1482-1487.
5. Brown G.W., Baca G.M. A classicification of original articles. Am J Dis Child 1986; 140:641-645.
6. How to Read Clinical Journals: 2.To Learn about a Diagnostic Tesr. Can Med Ass J 1981;124:703-710.
7. Der Simonian R., Charette L.J., McPeck B., Mosteller F. Reporting on methods of clinical trials. N Engl J Med 1982; 306:1332-7.
8. Detsky A.S., Sackett D.L. When was a “negative” clinical trial big enough? How many patients you needed depends on what you found. Arch Int
Med 1985; 145:709-12.
9. CONSORT Group. Improving the quality of reporting of randomized
controlled trials: the CONSORT statement. JAMA 1996; 276:637-9.
10. Feinstein A.R. Meta-analysis: Statistical alchemy for the 21st century. J
Clin Epidemiol 1995; 48:71-9.
11. User "s guides to the medical literature: II. How to use an article about therapy or prevention. A. Are the results of the study valid? JAMA 1993; 270: 2598-601.
12. Guyatt G.H., Sackett D.L., Cook D.J. User "s guides to the medical literature: II. How to use an article about therapy or prevention. B. What were the results and will they help me in caring for my patients. JAMA 1994; 271: 59-63.
13. Rosenbaum P.R. Discussing hidden bias in observational studies. Ann
Int Med 1991; 115:901-5.
14. Schultz K.F., Chalmers I., Altman D.G., Grimes D.A., Dore C.J. The methodological quality of randomisation as assessed from reports of trials in specialist and general medical journals. Online J Clin Trails, 1995 (doc N 197).
Problems of health and ecology
12. American Society of Echocardiography minimum standards for the cardiac sonographer: a position paper / S. M. Bierig // J Am Soc Echocardiogr. - 2006. - Vol. 19. - P. 471-474.
13. Antihypertensive drug therapy for mild to moderate hypertension during pregnancy / E. Abalos // The Cochrane Library Syst. Rev. - 2001. - Issue 4.
14. Antihypertensive drugs in pregnancy and fetal growth: evidence for “pharmacological programming” in the first trimester? / H. Bayliss // Hypertens Pregnancy. - 2002. - Vol. 21. - P. 161-174.
15. Antihypertensive therapy in the management of hypertension in pregnancy - a clinical double-blind study of pindolot / G. Bott-Kanner G. // Clin Exp Hypertension Pregnancy. - 1992. - Vol. 11. - P. 207-220.
16. Atenolol and fetal growth in pregnancies complicated by hypertension / C. Lydakis // Am. J. Hypertens. - 1999. - No. 12. - P. 541-547.
17. Australasian Society for the Study of Hypertension in Pregnancy: The detection, investigation and management of hypertension in pregnancy: full consensus statement / M. A. Brown // Am. J. Gynecol. - 2000. - Vol. 40. - P. 139-155.
18. Butters, L. Atenolol in essential hypertension during pregnancy / L. Butters, S. Kennedy, P. C. Rubin // Br. Med. J. - 1990. - Vol. 301.-P. 587-589.
19. Collins, R. Pharmacological prevention and treatment of hypertensive disorders in pregnancy / R. Collins, H.C. S. Wallenburg // Effective Care in Pregnancy and Childbirth / eds. I. Chalmers, M. Enkin, M. J. N. C. Keirse. - Oxford: Oxford University Press, 1989. - P. 512-533.
20. Effect of atenolol on birthweight / G. Y. Lip // Am. J. Cardiol. - 1997. - Vol. 79. - P. 1436-1438.
21. Effects of methyldopa on uteroplacental and fetal hemodynamics in pregnancy-induced hypertension / S. Montan // Am. J. Obstet. Gynecol. - 1993. - Vol. 168. - P. 152-156.
22. Fall in mean arterial pressure and fetal growth restriction in pregnancy hypertension: a meta-analysis / P. von Dadelszen // Lancet. - 2000. - Vol. 355. - P. 87-92.
23. Gallery, E.D.M. Antihypertensive treatment in pregnancy: analysis of different responses to oxprenolol and methyldopa /
E.D.M. Gallery, M. Ross, A. Z. Gyory // Br. Med. J. - 1985. - Vol. 291.-P. 563-566.
24. Gluckman, P. D. Maternal constraint of fetal growth and its consequences / P. D. Gluckman, M. A. Hanson // Semin Fetal Neonatal Med. - 2004. - Vol. 9, No. 5. - P. 419-425.
25. Guidelines Committee. 2003 European Society of Hypertension - European Society of Cardiology guidelines for the management of arterial hypertension // J. Hypertens. - 2003. - Vol. 21, No. 6. - P. 1011-1053.
26. Magee, L. A. Fortnightly review: management of hypertension in pregnancy / L. A. Magee, M. P. Ornstein, P. von Dadelszen // BMJ. - 1999. - Vol. 318, Issue 7194. - P. 1332-1336.
27. Magee, L. A. Oral beta-blockers for mild to moderate hypertension during pregnancy (Cochrane Review) / L. A. Magee, L. Duley // Cochrane Database Syst. Rev. - 2002. - Issue 1.
28. Preeclampsia - a state of sympathetic overactivity / H. P. Schobel // N. Engl. J. Med. - 1996. - Vol. 335. - P. 1480-1485.
29. Prevention of preeclampsia: a randomized trial of atenolol in hyperdynamic patients before onset of hypertension / T. R. Easterling // Obstet. Gynecol. - 1999. - Vol. 93. - P. 725-733.
30. Report of the National High Blood Pressure Education Program Working Group on High Blood Pressure in Pregnancy / R. W. Gifford // Am. J. Obstet. Gynecol. - 2000. - Vol. 183, No. 1. - P. 1-22.
31. The Task Force for the Management of Arterial Hypertension of the European Society of Hypertension and of the European Society of Cardiology / G. Mancia // Eur. Heart J. - 2007. - Vol. 28. - P. 1462-1536.
32. The Task Force on the Management of Cardiovascular Diseases During Pregnancy on the European Society of Cardiology. Expert consensus document on management of cardiovascular diseases during pregnancy // Eur. Heart. J. - 2003. - Vol. 24. - P. 761-781.
33. Use of antihypertensive medications in pregnancy and the risk of adverse perinatal outcomes: McMaster outcome study of hypertension in pregnancy 2 (MOS HIP 2) / J.G. Ray // BMC Pregnancy Childbirth. - 2001. - No. 1. - P.6.
34. World Health Organization - International Society of Hypertension 1999 Guidelines for the Management of Hypertension // High Blood Press. - 1999. - Vol. 8.-P. 1^3.
Received 29.10.2008
USE OF EVIDENCE-BASED MEDICINE DATA IN CLINICAL PRACTICE (message 3 - DIAGNOSTIC STUDIES)
A. A. Litvin2, A. L. Kalinin1, N. M. Trizna3
1Gomel State Medical University 2Gomel Regional Clinical Hospital 3Belarusian State Medical University, Minsk
An important aspect of evidence-based medicine is the completeness and accuracy of data presentation. The purpose of this article is to briefly review the principles of evidence-based medicine in research on the accuracy of diagnostic tests.
Diagnostic tests are used in medicine to establish the diagnosis, severity, and course of a disease. Diagnostic information is obtained from a variety of sources, including subjective, objective, special research methods. This article is based on a description of data on measuring the quality of studies, the advantages of various methods of summary statistics using the method of logistic regression and ROC analysis.
Keywords: evidence-based medicine, diagnostic tests, logistic regression, ROC analysis.
USE OF DATA OF EVIDENCE BASED MEDICINE IN CLINICAL PRACTICE (report 3 - DIAGNOSTIC TESTS)
A. A. Litvin2, A. L. Kalinin1, N. M. Trizna3
1Gomel State Medical University 2Gomel Regional Clinical Hospital 3Belarus State Medical University, Minsk
A prominent aspect of evidence based medicine is completeness and accuracy of data presentation. Article purpose is the short review of principles of evidence based medicine in the researches devoted to accuracy of diagnostic tests.
Problems of health and ecology
Diagnostic tests are used in medicine to screen for diagnosis, grade, and monitor the progression of disease. Diagnostic information is obtained from a multitude of sources, including sings, symptoms and special investigations. This article concentrates on the dimensions of study quality and the advantages of different summary statistics with logistic regression and ROC-analysis.
Key words: evidence based medicine, diagnostic tests, logistic regression, ROC-analysis.
When a doctor makes a judgment about the diagnosis based on the history and examination of the patient, he is rarely completely sure of it. In this regard, it is more appropriate to talk about the diagnosis in terms of its probability. It is still very common to express this probability not in the form of percentages, but with expressions such as "almost always", "usually", "sometimes", "rarely". Since different people invest different degrees of probability in the same terms, this leads to misunderstandings between doctors or between a doctor and a patient. Physicians should be as precise as possible in their conclusions and, if feasible, use quantitative methods to express probabilities.
Although the availability of such quantitative indicators would be very desirable, they are usually not available in clinical practice. Even experienced clinicians are often unable to accurately determine the likelihood of developing certain changes. There is a tendency to overdiagnose relatively rare diseases. It is especially difficult to quantify the probability, which can be very high or very low.
Since the establishment of reliable diagnostic criteria is the cornerstone of clinical thinking, accumulated clinical experience is used to develop statistical approaches to improve diagnostic prediction, which ideally should be presented in the form of computer data banks. In such studies, factors are usually identified
tori, which are in correlation with a particular diagnosis. These data can then be included in a multivariate analysis to determine which are significant independent predictors of the diagnosis. Some types of analysis allow you to identify important factors in predicting the diagnosis and then determine their "weight", which can be transformed into a probability in further mathematical calculations. On the other hand, the analysis allows us to identify a limited number of categories of patients, each of which has its own probability of having a particular diagnosis.
These quantitative approaches to diagnosis, often referred to as "prediction rules", are especially useful if they are presented in a user-friendly way and if their value has been extensively studied in a sufficient number and range of patients. For such prediction rules to be of real help to clinicians, they must be developed on representative patient populations using available reproducible tests so that the results obtained can be applied in medical practice everywhere.
In this regard, it is extremely important to be familiar with several of the most commonly used terms in research analysis and epidemiology, including prevalence, sensitivity, specificity, positive predictive value, and negative predictive value (Table 1) .
Table 1 - Systematic terms most commonly used in diagnostic studies
available absent
Positive a (true positive) b (false positive)
Negatives in (false negatives) r (true negatives)
Distribution (prior probability) = (a + c) / (a + b + c + d) = number of patients / total number of patients examined
Sensitivity \u003d a / (a + b) \u003d number of true positives / total number of patients
Specificity = r / (b+r) = number of true negatives / number of patients without the disease
False-negative rate = b / (a + b) = number of false-negative results / total number of patients
False positive rate = b / (b + d) = number of false positives / number of patients without the disease
Problems of health and ecology
End of table 1
Test results Pathological condition
available absent
Positive predictive value = a / (a + b) = number of true positives / number of all positives
Negative predictive value = r / (c+r) = number of true negatives / number of all negatives
Total accuracy (accuracy) = (a+r) / (a+b+c+d) = number of true positives and true negatives / number of all results
Likelihood ratio of a positive test - = sensitivity / (1 - specificity)
Likelihood ratio of a negative test - = 1 - sensitivity / specificity
Questions answered by these characteristics of the diagnostic test:
1) sensitivity - how good is the test at detecting patients with the condition?
2) specificity - how good is the test at correctly excluding patients who do not have the condition?
3) the predictive value of a positive test result - if a person tests positive, what is the probability that he really has this disease?
4) the predictive value of a negative test result - if a person has a negative test, what is the probability that he really does not have this disease?
5) accuracy index - what proportion of all tests gave correct results (i.e. true positive and true negative results in relation to all)?
6) Likelihood ratio of a positive test - how much more likely is it that the test will be positive in a person with a disease compared to a healthy person?
Since only a minority of the prediction rules meet strict criteria such as the number and range of subjects examined and prospective validation of results, most of them are unsuitable for routine clinical use. Moreover, many prediction rules fail to assess the likelihood of every diagnosis or outcome faced by the clinician. A test with a certain sensitivity and specificity has different positive and negative predictive value when used in groups with different prevalence of the disease. The sensitivity and specificity of any test does not depend on the distribution
The severity of the disease (or the percentage of patients who have the disease out of all examined patients), they depend on the composition of the group of patients among whom this test was used.
In some situations, inaccurate knowledge of the sensitivity and specificity of the test in the studied patient population may limit its clinical value. Since the physician rarely knows (or may know) the patient population on which the test he or she prescribes has been standardized, the results obtained are much less reliable than is commonly thought. Moreover, for any diagnostic test, an increase in sensitivity will be accompanied by a decrease in specificity.
A model with high sensitivity often gives the true result in the presence of a positive outcome (detects positive examples). Conversely, a model with high specificity is more likely to give a true result in the presence of a negative outcome (finds negative examples). If we talk in terms of medicine - the problem of diagnosing a disease, where the model for classifying patients into sick and healthy is called a diagnostic test, then we get the following: 1) a sensitive diagnostic test manifests itself in overdiagnosis - the maximum prevention of missing patients; 2) a specific diagnostic test diagnoses only certain patients. Because no single quantity or derived measure can be expected to have both excellent sensitivity and specificity, it is often necessary to determine which measure is most valuable and necessary for decision making. Graphic image, called the ROC curve
Problems of health and ecology
(Figure 1), linking the discussed characteristics of the test, shows the inevitability of a choice between striving for high sensitivity and specificity. Such a graphic representation indicates that the test results can be defined as normal or pathological, depending on whether
The disease is excluded if the test is highly specific or ruled out if the test is highly sensitive. Different tests may have different sensitivities and specificities. The sensitivity and specificity of more reliable tests are higher than those of invalid tests.
Figure 1 - Graphical representation of the internal discrepancy between sensitivity and specificity
ROC curve (Receiver Operator Characteristic) is the curve that is most commonly used to represent binary classification results in machine learning. The name comes from signal processing systems. Since there are two classes, one of them is called a class with positive outcomes, the second - with negative outcomes. The ROC curve shows the dependence of the number of correctly classified positive examples on the number of incorrectly classified negative examples. In the terminology of ROC analysis, the former are called true positive, the latter are called false negative sets. It is assumed that the classifier has some parameter, varying which we will get one or another breakdown into two classes. This parameter is often called the threshold, or cut-off value.
The ROC curve is obtained as follows. For each cutoff value, which varies from 0 to 1 in increments of, for example, 0.01, sensitivity values Se and specificity Sp are calculated. Alternatively, the threshold may be each successive sample value in the sample. A dependency graph is built: sensitivity Se is plotted along the Y axis, 100% - Sp (one hundred percent minus specificity) is plotted along the X axis. As a result, a certain curve appears (Figure 1). The graph is often supplemented with a straight line y = x.
For an ideal classifier, the plot of the ROC curve passes through the upper left
the angle where the true positive rate is 100% or 1.0 (ideal sensitivity) and the false positive rate is zero. Therefore, the closer the curve is to the upper left corner, the higher the predictive power of the model. Conversely, the smaller the curvature of the curve and the closer it is to the diagonal line, the less efficient the model. The diagonal line corresponds to the "useless" classifier, i.e., the complete indistinguishability of the two classes.
When visually assessing ROC-curves, their location relative to each other indicates their comparative effectiveness. The curve located above and to the left indicates a greater predictive ability of the model. So, in Figure 2, two ROC curves are combined on one graph. It can be seen that model A is better.
Visual comparison of ROC curves does not always reveal the most efficient model. A peculiar method for comparing ROC curves is the estimation of the area under the curves. Theoretically, it changes from 0 to 1.0, but since the model is always characterized by a curve located above the positive diagonal, one usually speaks of changes from 0.5 (a "useless" classifier) to 1.0 (an "ideal" model). This estimate can be obtained directly by calculating the area under the polyhedron, bounded on the right and bottom by the coordinate axes and on the top left - by experimentally obtained points (Figure 3). The numerical indicator of the area under the curve is called AUC (Area Under Curve).
Problems of health and ecology
Figure 2 - Comparison of ROC curves
Figure 3 - Area under the ROC curve
With large assumptions, we can assume that the larger the AUC, the better the predictive power of the model. However, you should be aware that the AUC indicator is intended rather for a comparative analysis of several models; AUC does not contain any
some information about the sensitivity and specificity of the model.
The literature sometimes provides the following expert scale for AUC values, which can be used to judge the quality of the model (table 2).
Table 2 - Expert scale of AUC values
AUC interval Model quality
0.9-1.0 Excellent
0.8-0.9 Very good
0.7-0.8 Good
0.6-0.7 Average
0.5-0.6 Unsatisfactory
The ideal model has 100% sensitivity and specificity. However, this cannot be achieved in practice; moreover, it is impossible to simultaneously increase the sensitivity and specificity of the model.
A compromise is found with the help of the cutoff threshold, since the threshold value affects the ratio of Se and Sp. We can talk about the problem of finding the optimal cut-off value (Figure 4) .
Figure 4 - “Balance point” between sensitivity and specificity
Problems of health and ecology
The cutoff threshold is needed in order to apply the model in practice: to attribute new examples to one of two classes. To determine the optimal threshold, you need to set a criterion for its determination, because different tasks have their own optimal strategy. The criteria for choosing the cut-off threshold can be: 1) the requirement for a minimum value of sensitivity (specificity) of the model. For example, you need to ensure the sensitivity of the test is not less than 80%. In this case, the optimal threshold will be the maximum specificity (sensitivity), which is achieved at 80% (or a value close to
him "on the right" due to the discreteness of the series) sensitivity (specificity).
The given theoretical data are better perceived by examples from clinical practice. The first example we will focus on would be the diagnosis of infected necrotizing pancreatitis (data set taken from the database). The training sample contains 391 records with the selection of 12 independent variables in the following format (Table 3). Dependent variable (1 - presence of the disease, 0 - absence). The distribution of the dependent variable is as follows: 205 cases - no disease, 186 - its presence.
Table 3 - Independent variables for the diagnosis of infected pancreatic necrosis, logistic regression coefficients (example)
Independent variables Data format Coefficient, %
Number of days from onset > 14< 14 2,54
Number of days spent by patients on treatment in the ICU > 7< 7 2,87
Heart rate numerical value 1.76
Respiratory rate numerical value 1.42
Body temperature numerical value 1.47
Blood leukocytes numerical value 1.33
Leukocyte index of intoxication numerical value 1.76
Blood urea numerical value 1.23
Total plasma protein numerical value 1.43
Adequate antibiotic prophylaxis in establishing the diagnosis of severe acute pancreatitis yes / no -1.20
Performing minimally invasive medical and preventive operations yes / no -1.38
Presence of negative dynamics yes/no 2.37
Figure 4 depicts the resulting ROC which can be characterized as a very good curve. The predictive power of the model AUC = 0.839.
Figure 4 - ROC-curve of the diagnostic model of infected pancreatic necrosis
Problems of health and ecology
Consider a fragment of the array of points “feeling of intra-abdominal pressure in patients with severe
validity-specificity” on the example of level acute pancreatitis.
Table 4 - Sensitivity and specificity of different levels of IAP for predicting the development of PPI (example)
IAP, mm Hg Art. Sensitivity, % Specificity, % Se + Sp Se - Sp
13,5 25 100 125 75
14,5 30 95 125 65
15,5 40 95 135 55
16,5 65 95 160 30
17,5 80 90 170 10
18,5 80 80 160 0
19,5 80 70 150 10
20,5 85 65 150 20
21,5 95 55 150 40
23,0 100 45 145 55
24,5 100 40 140 60
25,5 100 25 125 75
As can be seen from the table, the optimal threshold level of IAP in patients with acute destructive pancreatitis, which provides the maximum sensitivity and specificity of the test (or a minimum of type I and II errors), is 17.5 ± 2.3 (M ± SD) mm Hg, at which there is 80% sensitivity and 90% specificity of the method for determining the likelihood of developing infectious complications of pancreatic necrosis. The sensitivity is 80%, which means that 80% of patients with infected necrotizing pancreatitis have a positive diagnostic test. The specificity is 90%, so 90% of patients who do not have infected necrotizing pancreatitis have a negative test result. The balance point at which sensitivity and specificity approximately coincide - 80%, is 18.5. Overall, the positive predictive value of IAP measurement was 86%, and the negative predictive value was 88%.
Carrying out logistic regression and ROC analysis is possible using statistical packages. However, "Statistica" 6 and 7 (http://www.statistica.com) carry out this analysis only using the "Artificial Neural Networks" block. In SPSS (http://www. spss.com) (starting from version 13) ROC analysis is given only in the graphic module and one ROC curve is analyzed. SPSS displays the area under the curve (AUC), significance level, and sensitivity and specificity value at each measurement point. The optimal point (optimal cut-off) must be found by yourself from the table of sensitivity and 1-specificity. The MedCalc program will compare several ROC curves, mark the value of the variable in the table, when
which the ratio of sensitivity and specificity is optimal (optimal cut-off). SAS (http://www.sas.com), as well as R-Commander, has a curve comparison and point finding module, AUC. Logistic regression and ROC analysis are available from the free WINPEPI (PEPI-for-Windows) program (http://www.brixtonhealth.com/winpepi.zip) .
Conclusion
The art of diagnosis is constantly improving. New diagnostic tests appear daily, and the technology of existing methods changes. Overestimation of the accuracy of relevant studies, in particular as a result of bias due to poor research and publication practices, may lead to premature implementation of diagnostic tests and poor clinical decisions. Careful evaluation of diagnostic tests before their widespread use not only reduces the risk of adverse outcomes due to misperceptions about the usefulness of the method, but can also limit the expenditure of health care resources by eliminating unnecessary tests. An integral part of the evaluation of diagnostic tests are studies on the accuracy of diagnostic tests, the most informative of which are the method of logistic regression and ROC analysis.
REFERENCES
1. Greenhalch, T. Fundamentals of evidence-based medicine / T. Greenhalch; per. from English. - M.: GEOTAR-Media, 2006. - 240 p.
Problems of health and ecology
3. Vlasov, V. V. Introduction to evidence-based medicine / V. V. Vlasov. - M. MediaSphere, 2001. - 392 p.
4. Fletcher, R. Clinical epidemiology. Fundamentals of evidence-based medicine / R. Fletcher, S. Fletcher, E. Wagner; per. from English. - M.: MediaSphere, 1998. - 352 p.
5. Banerzhi, A. Medical statistics in plain language: an introductory course / A. Benerzhi; translation from English. - M.: Practical medicine, 2007. - 287 p.
6. Zhizhin, K. S. Medical statistics: textbook. allowance. - Rostov n / D .: Phoenix, 2007. - 160 p.
7. Deeks, J. J. Systematic reviews of evaluations of diagnostic and screening tests / J. J. Deeks // BMJ. - 2001. - Vol. 323. - P. 157-162.
8. Guidelines for meta-analyses evaluating diagnostic tests / L. Irwig // Ann. Intern. Med. - 1994. - Vol. 120. - P. 667-676.
9. Systematic reviews and meta-analysis for the surgeon scientist /
S. S. Mahid // Br. J. Surg. - 2006. - Vol. 93. - P. 1315-1324.
10. Meta-analytical methods for diagnostic test accuracy / L. Irwig // J. Clin. epidemiol. - 1995. - Vol. 48. - P. 119-130.
11. Users" guides to the medical literature. How to use an article about a diagnostic test. A. Are the results of the study valid? / R. Jaeschke // JAMA. - 1994. - Vol. 271. - P. 389 -391.
12. Use of methodological standards in diagnostic test research: getting better but still not good / M. C. Read // JAMA. - 1995. - Vol. 274.-P. 645-651.
13. StAR: a simple tool for the statistical comparison of ROC curves / I. E. Vergara // BMC Bioinformatics. - 2008. - Vol. 9. - P. 265-270.
14. A comparison of parametric and nonparametric approaches to ROC-analysis of quantitative diagnostic tests / K. O. Hajian-Tilaki // Medical Decision Making. - 1997. - Vol. 17, N. 1. - P. 94-102.
15. Receiver operator characteristic (ROC) curves and nonnormal data: An empirical study / M.J. Goddard // Statistics in Medicine. - 1989. - Vol. 9, N. 3. - P. 325-337.
16. Possibilities of predicting infected pancreatic necrosis / A. A. Litvin [et al.] // Problems of health and ecology. - 2007. - T. 12, No. 2. - S. 7-14.
17. Method for monitoring intra-abdominal pressure in patients with severe acute pancreatitis / A. A. Litvin [et al.] // Problems of health and ecology. - 2008. - T. 16, No. 2. - S. 80-85.
18. Comparison of eight computer programs for receiver-operating characteristic analysis / C. Stephan // Clin. Chem. - 2003. - Vol. 49, N. 3. - P. 433-439.
19. Zhu, X. A short preview of free statistical software packages for teaching statistics to industrial technology majors / X. Zxu // J. Ind. technology. - 2005. - Vol. 21, N. 2. - P. 10-20.
20. Borovikov, V. STATISTICA: the art of computer data analysis. For professionals / V. Borovikov. - St. Petersburg: Peter, 2001. - 656 p.
21. Buyul, A. SPSS: the art of information processing. Analysis of statistical data and restoration of hidden patterns / A. Byuyul. - St. Petersburg: DiaSoftYUP, 2002. - 608 p.
22. Abramson, J. H. WINPEPI (PEPI-for-Windows): computer programs for epidemiologists / J. H. Abramson, // Epidemiologic Perspectives & Innovations. - 2004. - Vol. 1, N. 6. - P. 1-10.
Received 24.10.2008
UDC 616.1:616-009.12:616-005.8:616.831-005.1
SOME INDICATORS OF MICROCIRCULATION AND ENDOTHELIAL DAMAGE IN THE ASSESSMENT OF THE RISK OF DEVELOPMENT OF STROKE, MYOCARDIAL INFARCTIONS, FATAL OUTCOMES IN PATIENTS WITH ARTERIAL HYPERTENSION
V. I. Kozlovsky, A. V. Akulyonok Vitebsk State Medical University
The aim of the study was to identify factors associated with an increased risk of myocardial infarction, cerebral stroke, and death in patients with stage II arterial hypertension (AH).
Material and Methods: The study included 220 patients with II degree AH (mean age 57 ± 8.4 years) who were hospitalized due to a hypertensive crisis, and 30 people without AH (mean age
53.7 ± 9 years).
Results: 29 strokes, 18 myocardial infarctions, 26 deaths were recorded in the group of patients with II degree AH during 3.3 ± 1 years of follow-up. An increase in the number of circulating endothelial cells (ECC), leukocyte and platelet aggregation, and leukocyte adhesion in hypertensive patients was associated with an increased risk of myocardial infarction, stroke, and death.
Conclusion: indicators of the number of CECs, aggregation of platelets and leukocytes, and leukocyte adhesion can be used to identify groups of hypertensive patients at an increased risk of developing myocardial infarctions, strokes and deaths, as well as to create complex prognosis models.
Key words: arterial hypertension, risk, myocardial infarction, stroke, death, circulating endotheliocytes.
SOME FINDINGS OF MICROCIRCULATION AND ENDOTHELIAL DAMAGE IN ESTIMATION OF RISK FOR STROKES, MYOCARDIAL INFARCTIONS, LETHAL OUTCOMES IN HYPERTENSIVE PATIENTS
V. I. ^zlovsky, A. V. Akulionak Vitebsk Statel Medical University
Objective: to determine factors associated with increased risk for development of strokes, myocardial infarctions, lethal outcomes in patients with arterial hypertension (AH) II degree.
Methods: 220 patients with AH II degree (mean age 57 ± 8.4 years), complicated by hypertensive crisis, and 30 persons without AH (mean age 53.7 ± 9 years) were followed-up for 3.3±1 years .
Results: elevation of number of circulating endothelial cells (CEC), aggregation of platelets and leukocytes, adhesion of leukocytes in hypertensive patients were associated with increased risk for development of strokes, myocardial infarctions, lethal outcomes.
Lisa is in severe pain after the operation. The clinician must choose between tablets based on external clinical evidence or injections based on personal clinical experience and patient preferences. The doctor knows that according to external clinical evidence, morphine tablets would be the best choice. However, as it turned out during the operation, Lisa suffers from a common side effect of anesthesia - vomiting. This means that if Lisa takes the pill and she vomits, the contents of the pill will come out and have no pain relief. Both the doctor and Lisa know from previous experience that Lisa may vomit within 30 minutes of the anesthetic wears off. Therefore, instead of a pill, the doctor decides to give Liza an injection with morphine.
In this example, the physician, based on personal clinical experience and the patient's preferences, decides to use the morphine injection instead of the morphine tablet, even though the best external clinical evidence is in favor of the latter. The doctor uses the same medical substance (i.e. morphine) as external clinical evidence suggests, but chooses a different dosage form (an injection instead of a tablet).
This is an example of a doctor making a definite decision in the course of treatment, based on supporting evidence, after discussion with the patient.
What is evidence-based medicine?
evidence-based medicine (EBM) is the process of systematically reviewing, evaluating, and using the results of clinical trials in order to provide optimal medical care to patients. Patient awareness of evidence-based medicine is important as it allows them to make more informed decisions about disease management and treatment. It also enables patients to form a more accurate picture of risk, encourages the appropriate use of individual procedures, and enables the clinician and/or patient to make decisions based on supporting evidence.
Evidence-based medicine combines principles and methods. Due to the operation of these principles and methods, decisions, instructions and strategies in medicine are based on current supporting data about the effectiveness of different forms of the course and medical services in general. For medicines, evidence-based medicine relies heavily on information obtained from benefit and risk assessments (efficacy and safety).
The concept of evidence-based medicine emerged in the 1950s. Up to this point, physicians have made decisions largely on the basis of their education, clinical experience, and reading scientific periodicals. However, studies have shown that medical treatment decisions vary significantly among different medical professionals. The basis for the introduction of systematic methods for collecting, evaluating and organizing research data was formed, which became the beginning of evidence-based medicine. The advent of evidence-based medicine has been recognized by physicians, pharmaceutical companies, regulators and the public.
The decision maker needs to rely on their own experience in treating patients, combined with the best supporting evidence from controlled trials and scientific development. It is important to combine clinical experience and controlled trials in the decision-making process. In the absence of clinical experience Risk is the likelihood of harm or injury resulting from treatment in clinical practice or in research. Harm or injury can be physical, but also psychological, social or economic. Risks include developing side effects of treatment or taking a drug that is less effective than standard treatment (as part of a trial). When testing a new medicine, there may be side effects or other risks not anticipated by the researchers. This situation is most typical for the initial stages of clinical trials.
Conducting any clinical trial involves risks. Participants should be informed about possible benefits and risks before deciding to participate (see definition of informed consent).
" target="_blank">The risk associated with certain treatments may result in unwanted effects.
Five-stage model of evidence-based medicine
One approach to evidence-based medicine involves a model of 5 stages:
- formation of a clinically relevant request (search for information by a doctor to make a correct diagnosis),
- search for better supporting data (doctor's search for supporting data in support of the information found in step 1),
- assessment of the quality of supporting data (providing the doctor with high quality and reliability),
- formation of a medical decision based on supporting data (acceptance by the patient and the doctor of an informed decision on treatment based on steps 1-3),
- evaluation of the process (evaluation by the doctor and the patient of the achieved result and the corresponding adjustment of treatment decisions, if necessary).
In the example above, the choice of physician is consistent with both evidence-based medicine and patient feedback. The physician's decision involves the conscious, open and informed use of the best evidence available at the current time, including the patient's experience, to select the best possible care for that patient.
Patient participation in the decision-making process is essential for the development of new treatment principles. Such participation includes reading and understanding treatment information and following recommendations consciously, working with clinical professionals to evaluate and select the best treatment options, and providing feedback on the results. Patients can actively participate in the creation of supporting evidence at any level.
Assessment of supporting data for the needs of evidence-based medicine
The collected information is classified according to the level of supporting evidence it contains in order to assess its quality. The pyramid in the figure below shows different levels of evidence and their ranking.
levels of evidence
Comments or expert opinions
This is data based on the opinions of a panel of experts and aimed at shaping general medical practice.
Study of case series and descriptions of clinical cases
A case series study is a descriptive study of small people. As a rule, it serves as an addition or supplement to the description of a clinical case. A case report is a detailed report of the symptoms, signs, diagnosis, treatment and management of a single patient.
Case-control studies
is an observational retrospective study (with a review of historical data) in which patients suffering from a disease are compared with patients who do not have this disease. Cases such as lung cancer are usually studied in a case-control study. To do this, a group of smokers (group under the influence) and a group of non-smokers (group not under the influence) are recruited, which are monitored for a certain period of time. The difference in incidence of lung cancer is then documented, allowing a variable (the independent variable—in this case, smoking) to be considered as the cause of the dependent variable (in this case, lung cancer).
In this example, a significant increase in lung cancer cases in the smoking group compared to the non-smoking group is taken as evidence of a causal relationship between smoking and the occurrence of lung cancer.
cohort study
The modern definition of a cohort in a clinical trial is a group of individuals with certain characteristics who are monitored for health outcomes.
The Framingham Heart Study is an example of a cohort study conducted to answer an epidemiological question. The Framingham Study began in 1948 and is still ongoing. The purpose of the study is to investigate the impact of a number of factors on the incidence of heart disease. The question researchers are facing is whether high blood pressure, overweight, diabetes, physical activity and other factors are associated with the development of heart disease. For each of the exposure factors (such as smoking), researchers recruit a group of smokers (the exposed group) and a group of non-smokers (the unexposed group). The groups are then observed over a period of time. Then, at the end of the observation period, the difference in the incidence of heart disease in these groups is documented. Groups are compared in terms of many other variables such as
- economic status (for example, education, income and occupation),
- health status (for example, the presence of other diseases).
This means that a variable (the independent variable, in this case, smoking) can be isolated as the cause of the dependent variable (in this case, lung cancer).
In this example, a statistically significant increase in the incidence of heart disease in the smoking group compared to the non-smoking group is taken as evidence of a causal relationship between smoking and the occurrence of heart disease. The results found in the Framingham Study over the years provide compelling evidence that cardiovascular disease is largely the result of measurable and modifiable risk factors, and that a person can control the health of their heart system if they carefully monitor their diet and lifestyle and refuses the consumption of refined fats, cholesterol and smoking, reduces weight or begins to lead an active lifestyle, regulates stress and blood pressure. It is largely due to the Framingham Study that we now have a clear understanding of the association of certain risk factors with heart disease.
Another example of a cohort study that has been running for many years is the National Child Development Study (NCDS), the most studied of all UK newborn cohort studies. The largest study on women is the Nurses Health Study. It began in 1976, the number of accompanied persons is more than 120 thousand people. According to this study, many diseases and outcomes were analyzed.
randomized clinical trials
Clinical trials are called randomized when randomization is used to allocate participants to different treatment groups. This means that treatment groups are randomly populated using a formal system, and there is a chance for each participant to get into each of the study destinations.
Meta-analysis
is a systematic, statistical-based review of data that compares and combines the results of different studies in order to identify patterns, inconsistencies, and other relationships across multiple studies. A meta-analysis can provide support for a stronger conclusion than any single study, but the disadvantages of bias due to the publication preference of positive study results must be kept in mind.
Study results
Outcome research is a broad umbrella term that does not have a fixed definition. Outcome research examines health care outcomes, in other words, the effect of the health care delivery process on the health and well-being of patients. In other words, clinical outcome studies aim to monitor, understand, and optimize the impact of medical treatment on a particular patient or group. Such studies describe scientific research that is related to the effectiveness of health measures and medical services, that is, the results obtained through such services.
Often attention is focused on the person suffering from the disease - in other words, on the clinical (overall results) that are most relevant to this patient or group of patients. These endpoints can be either the degree of pain. However, outcome studies may also focus on the effectiveness of health service delivery, with measures such as , health status, and disease severity (the impact of health problems on the individual).
The difference between evidence-based medicine and outcome research lies in the focus on different issues. While the main goal of evidence-based medicine is to provide the patient with optimal care in accordance with clinical evidence and experience, outcome studies are primarily aimed at predicting endpoints. In a clinical outcome study, these endpoints usually correspond to clinically relevant endpoints.
| Endpoint view | Example |
|---|---|
| Physiological parameter () | Arterial pressure |
| Clinical | Heart failure |
| Symptom In medicine, a symptom is usually a subjective perception of a disease, different from a sign that can be identified and evaluated. Symptoms include, for example, abdominal pain, lumbago, and fatigue, which only the patient feels and can report. A sign may be blood in the stool, a skin rash determined by the doctor, or a high fever. Sometimes the patient may not pay attention to the sign, however, he will give the doctor the information necessary for making a diagnosis. For instance:
A mild headache may only be a symptom.
" target="_blank">Symptoms |
Cough |
| Functional abilities and need for care | Parameter for measuring functional ability, e.g. ability to perform daily activities, quality of life assessments |
In outcome studies, the relevant endpoints are often symptoms or measures of functional ability and care needs that the patient receiving treatment considers important. For example, a patient suffering from an infection who has been injected with penicillin may pay more attention to the fact that he does not have a fever and his general condition has improved than to the effect of penicillin on the actual level of infection. In this case, the symptoms and how he feels are seen as a direct measure of his state of health, and these are the endpoints that are the focus of the outcome study. The patient is also likely to be concerned about the possible side effects associated with penicillin as well as the cost of treatment. In the case of other diseases such as cancer, an important clinical outcome relevant to the patient will be the risk of death.
If the study is long in time, when studying the results of the studies, “ ” can be used. A surrogate endpoint involves the use of a biomarker to measure outcome, acting as a substitute for a clinical endpoint that measures the effect of penicillin by reducing the amount of a protein (C-reactive protein) that is always present in the blood. The amount of this protein in the blood of a healthy person is very small, but with an acute infection it rises rapidly. Thus, measuring the level of C-reactive protein in the blood is an indirect way to determine the presence of infection in the body, therefore, in this case, the protein serves as a "biomarker" of infection. A biomarker is a measurable indicator of a disease state. This parameter is also correlated with the risk of occurrence or progression of the disease, or how the prescribed treatment will affect the disease. Every day, the patient's blood is taken for analysis to measure the amount of the biomarker in the blood.
It must be emphasized that in order to use a surrogate endpoint for the purpose of control and supervision, the token must be validated or verified in advance. It is necessary to demonstrate that changes in the biomarker are correlated (consistent) with the clinical outcome in the case of a specific disease and the effect of treatment.
Additional sources
- World Health Organization (2008). Where are the patients in decision-making about their own care? Retrieved August 31, 2015
Sports message Sports message
Shall we dance Latin American dances?
The main types and styles of swimming: classification, features and characteristics The emergence and development of synchronized swimming in Russia
School them kapitsa. Gobu "phystech-lyceum" named after p.l.kapitsa. Welcoming speech by Director Mashkova M.G.
Various sports sections are open for pensioners in South Ossetia