Medical education and research training in many Russian universities seems to remain largely unchanged since the Soviet time. An evaluation of the current state of epidemiology in Russia concluded that it was still mainly a science addressing the spread of infectious diseases (1). In their analysis of the use of statistical methods in biomedical journals and dissertations, Leonov and Izhevskiy note that Russian researchers primarily use univariate analyses and a limited number of statistical tests (2). These results reflect the contents of research training programs at the universities. International scientific media are often unaccessable and unaffordabe to many researchers. Both high subscription costs of international peer-reviewed journals, limited access to the Internet and insufficient English language skills may contribute to this shortage of information. Widely used research designs such as cohort studies and randomised controlled trials and modern methods of data analysis such as various multivariate techniques and survival analysis, just to name a few, are rarely used by Russian medical professionals. In spite of the fact that medical students and practicing physicians do not have adequate knowledge in research methodology, they are often encouraged to do research. Lack of knowledge in research planning and data processing methods may lead to incorrect data analysis which in turn may result in incorrect conclusions and low quality publications. The worst case scenario is that these incorrect conclusions might be used as a basis for medical, administrative and political decisions that may have a profound influence on the individual and public health.
However, recent progress has to be acknowledged. New textbooks on epidemiology in the international sense of the word were recently published (3-4). In addition, some applied statistical texts were translated into Russian (5-6) as well as some new Russian books on statistics in medicine were issued (7-8). The process of implementing this information into research practice is slow and, with few exceptions (9), the overall quality of biomedical publications in relation to study design, data analysis and interpretation of the results has not improved significantly.
The introduction of teaching both epidemiology and biostatistics in Russian medical schools is warranted. Given that there have been several courses in epidemiology arranged by international teams in Northwest Russia, we attempted to introduce a course in biostatistics. The course was arranged by the Northern Europe group from the Norwegian Institute of Public Health (NIPH) in cooperation with the Northern State Medical University (NSMU) and took place in April 2006 in Arkhangelsk. The aims of the course were to 1) develop skills for critical reading of medical literature, 2) give an overview of basic methods of statistics with special emphasis on the algorithms for the choice of appropriate tests depending on the data at hand and 3) to show how these analyses could be performed using SPSS software.
Information about the course was provided to all departments of the NSMU three months prior to the course. The course was in high demand as indicated by more than 100 applicants. Given the limited number of computers in the computer room, the international office of the NSMU was able to select only 23 participants. Most of the participants had previous experience in research and reported having published in Russian scientific journals. Very few participants were in the planning phase of their first projects.
The course included five modules, one full day each (Table 1). The modules consisted of a theoretical part (four hours) and a practical part that included calculations in SPSS using local Russian material from the Severodvinsk study (10). In addition, educational materials (in Russian) on compact disks were distributed (5, 11). The course was taught by a native Russian speaker.
Table 1. Course contents by modules
Given the non-mathematical background of the participants, the aim of the course was to teach which statistical tests should be used in specific situations and how to interpret the results of the tests. Examples of incorrect use of some statistical tests in publications were also given. The style of teaching was fairly informal encouraging active interaction between the teacher and the participants.
PowerPoint presentations were used during the teaching of the modules 1 and 2. During the rest of the course, lectures were accompanied by writing on a white board, and by showing how to perform data analyses in SPSS directly during the lecture.
At the end of the course an evaluation form (in Russian) was completed by 91% of participants (n=21). The form included eight items and was designed to study the relationship between participants’ background knowledge in biostatistics and how useful they found the different modules of the course.
The participants were asked to estimate 1) their knowledge of biostatistics prior to the course, 2) the usefulness of each of the course modules for critical evaluation of publications, 3) the usefulness of each of the course modules for planning own research/analysing already collected data, 4) the teacher’s performance in making of each of the course modules understandable and 5) the usefulness of the practical examples for the understanding of each of the course modules. The answers were given on a 10-point Lickert scale. In addition, the participants were asked to give their comments after each of these questions. Moreover, we asked 6) whether the participants would recommend this course to their colleagues involved in research, 7) how we could improve the course and 8) which methods they would like to study during the advanced course.
Frequency distribution of the background knowledge in biostatistics of the course participants is presented in Table 2. This finding is of concern considering that most of the participants have scientific degrees and publications.
Table 2. Distribution of the participants’ background knowledge in biostatistics
Generally, the participants’ responses in the evaluation form were positive. The most common score for all questions for all modules was 10 except for the teacher’s performance for the module three where it was 9. Despite the fact that most of the participants seemed to be satisfied with the course, there were few cases with lower scores for some of the questions. Distribution of the scores is presented in Figures 1-4 (boxes represent 50 % of all scores, horizontal lines in boxes represent medians).
Figure 1. Scores for usefulness of each of the course modules for critical evaluation of publications
Figure 2. Scores for usefulness of each of the course modules planning own research/analysing previously collected data
Figure 3. Scores for teacher’s performance to make of each of the course modules understandable
Figure 4. Scores for usefulness of the practical examples for understanding each of the course modules
Interestingly, the first two modules taught with the assistance of PowerPoint presentations got the highest scores for the teacher’s performance indicating that this teaching approach might be more effective than the traditional one with writing on a board. However, this may also indicate that the first two modules included less complicated material than the topics taught during the other days. Moreover, the participants reported higher scores for the usefulness of the first two modules compared with modules 3 and 4 in relation to both assessing publications and analysing own data. This is in line with other publications concluding that multivariable methods are rarely used in Russian medical research (1-2) probably due to lack of teachers and educational literature on multiple methods of analysis written for non-statisticians.
There were positive associations between the participants’ background knowledge in biostatistics and how useful they found modules 3 and 4. This seems to reflect the difficulty of the topics for the beginners and indicates that it might be more appropriate to teach multiple linear regression at a more advanced course. It also suggests that the levels of participants’ background knowledge should be similar to ensure the success of the course. Similar relations were found between the participants’ background knowledge, the teacher’s performance and usefulness of the practical examples (Table 3).
Table 3. Spearmen’s correlation coefficients (with p-values in parentheses) between the level of background knowledge in biostatistics and evaluation scores for the course (only correlations above 0.3 are presented)
All participants reported that they would recommend this course to their colleagues who are involved in research.
The comments on improvements of the course may be summarised in five major points:
- Printed handouts of PowerPoint presentations should be distributed before lectures.
- Individual home assignments should be included, with presentation of results next day in class.
- More assignments with SPSS and more time for practical work in class is needed.
- More articles from Russian biomedical journal for critical evaluation should be given.
- The course was too intensive. Optimal duration of the course should be 2-3 weeks.
Among the methods that the participants wanted to learn during the next course were tests for comparing non-independent groups, survival analysis and agreement studies (Bland-Altman method was mentioned). However, most of the responders simply answered that they would like to participate in a course on a more advanced level.
In summary, based on the analysis of the evaluation forms, the active participation of the students in the discussions during the course and their willingness to continue education in biostatistics on a more advanced level, the first course in biostatistics in Arkhangelsk was a success. Most of the students found the material useful and rated the teacher’s performance and practical examples as adequate or better. Given the high demand for similar courses in Northwest Russia, the participants’ feedback is of a high value to the organisers and will be used for planning the subsequent courses in Russia.
1. Vlassov V. Is there epidemiology in Russia? J Epidemiol Community Health 2000; 54: 740-4.
2. Leonov V, Izhevskiy P. On the usage of applied statistics in the preparation of dissertations for medical and biological degrees. Part I: Methods description (in Russian). Int J of Med Pract 1998; 4: 7-12.
3. Vlassov V. Epidemiology. Moscow, GOETAR-MED, 2004. (in Russian).
4. Fletcher R, Fletcher S, Vagner E. Clinical epidemiology: basics of evidence based medicine. Moscow: MediaSphera, 1998. (in Russian)
5. Glanz S. Primer of biostatistics. Moscow, 1998. (in Russian)
6. Petrie A, Sabin C. Medical statistics at a glance. Moscow, GOETAR-MED, 2003. (in Russian).
7. Sergienko V, Bondareva I. Mathematical statistics in clinical research. Moscow, GOETAR-MED, 2001. (in Russian)
8. Zaitsev V, Liflyandskiy V, Marinkin V. Applied medical statistics St. Petersburg, Foliant, 2003. (in Russian).
9. Siberian Medical Journal. www.medicina.tomsk.ru/russian/smj.htm.
10. Grjibovski A, Bygren L, Svartbo B, Magnus P. Social variations in fetal growth in a Russian setting: an analysis of medical records. Ann Epidemiol 2003; 13: 599-605.
11. Lang T. Twenty statistical errors even you can find in biomedical research articles. Croat Med J 2004; 45: 361-70.