Integrating Biostatistics and Experimental Physiology : Students

Indian J Physiol Pharmacol 1998; 42 (4) : 453-459

INTEGRATING BIOSTATISTICS AND EXPERIMENTAL PHYSIOLOGY: STUDENTS' PERCEPTIONS AND FUTURE SCOPE

M. VAZ*, R. PADMAVATHI AND G. VEENA

Department of Physiology, St John's Medical College, Bangalore - 560 034

( Received on November 22, 1997) Abstract: The use of biostatistics in experimental physiology is recognised by researchers. However, there has been no concerted attempt to integrate biostatistics into the undergraduate experimental physiology programme. This paper describes one such initiative. The student's response to the exercise was largely positive; it enabled them to describe and interpret data more effectively and understand the experiments more completely. The attitudes of the students to the exercise and their performance at a statistics examination at the end of the exercise was determined by a number of parameters, including prior statistical knowledge, general academic performance and the extent to which they liked mathematics. However, even those students who disliked mathematics indicated that they appreciated the value of the exercise. The results indicate the need to integrate biostatistics into the undergraduate physiology course. Key words: medical education physiology statistics

INTRODUCTION undergraduates. This paper describes the attitudes of students to an integrated The interpretation of experimental and biostatistics programme that we recently investigative data is one of the objectives introduced, outlines the determinants of of the physiology curriculum. In a wider their attitudes and explores the scope of context, the knowledge of statistics is such an integrated programme. important to medical graduates in ituations where they perform research, METHODS keep abreast of current literature (1,2) and undertake public health initiatives (3). Sixty students of the first term of Despite this, however, there has been no I MBBS form the basis of this study. There concerted effort to ensure that medical were an equal number of males and females graduates in India are proficient in the in the sample. At the time of the study the application of biostatistics. Biostatistics and students had completed a set of practicals experimental physiology, inevitably go relating to nerve and muscle physiology together. This provides a unique using the gastrocnemius muscle and sciatic opportunity to integrate biostatistics into nerve preparation of the frog. The present the experimental physiology course of exercise was carried out during a set of *Corresponding Author 454 Vaz et al Indian J Physiol Pharmacol 1998; 42(4) human experiments designed to record normal distribution, coefficient of variation, muscle strength and its determinants, and quartiles, centiles, correlation, P value, null whole body and muscle endurance during hypothesis and 't' test. The form thus sought exercise. Muscle strength was determined to determine the level of knowledge of using a handgrip dynamometer (Harpenden, statistics for each stu.dent, whether they eMS Weighing Equipment Ltd. London). received any prior formal training in Anthropometric determinants of muscle statistics, and their attitudes towards strength assessed, included body size (from statistics in terms of its benefits and body mass index: weight in kgfheight2 in disadvantages. m), fat-free mass (from the BMI derived age and gender specific equation of Deurenberg) During the course of the experiments, (4), muscle mass (5) and forearm muscle students were provided with specific area (6). Isolated muscle endurance in the practical sheets which outlined the aims and forearm was assessed using a handgrip procedures of the experiment. The dynamometer, coupled to a load cell and worksheets incorporated tables and blank linked to a polygraph. Whole body fitness spaces into which students fiLed their data was assessed using a standard 2 km walking and calculated simple statistical parameters test (7). such as mean, standard deviation and coefficient of variation. The worksheets also The sequence that we followed is contained the formulae for students to depicted in Table 1. Prior to the statistical calculate derived anthropometric indices exercise, a pre-assessment form was given such as body mass index, fat-free mass, to the students. This form consisted offixed muscle mass and forearm muscle area. response questions (Yes/No), statements Students were instructed on the need to with which level of agreement was repeat measurements, as a means of determined using a Likert-type scale, and determining measurement error and open-ended questions. Students were also understanding the concept of intraindividual required to indicate from a standard list biological variability. After the experiment, which statistical terms they had a working the data sheets were collected from the knowledge of. This list consisted of the students, entered on computer and analysed following terms: mean, standard deviation, by a staff member.

TABLE I: Sequence of the present exercise.

1. Administration of a pre-assessment questionnaire to determine prior attitudes and knowledge towards statistics (15 min) 2. Physiology practical: collection of data 3. Collation of data and statistical analysis on computer 4. Theory class on biostatistics (60 min) with minimal mathematics 5. Filling up of statistics worksheets for the physiology practical under faculty guidance (60 min) 6. Administration of a post-assessment questionnaire 7. Statistics test 1 week after 5, based on a subsequent [lractIcal (20 min) Indian J Physiol Pharmacol 1998; 42(4) Biostatistics and Physiology 455

Prior to the presentation of the results worksheets. Interspersed in the worksheets to students a single class was held covering were also questions which required students very basic elements of descriptive and to interpret the data. The faculty member inferential statistics. Concepts that were discussed each of these interpretative covered included mean, standard deviation, questions and guided the students on how range, centiles and quartiles, normal to answer them. distribution, statistical significance and correlations. Care was taken to avoid the After the first statistical exercise a post use of mathematical formulae, since this can assessment form was administered. This be counter productive (8). The excessive use form included a table of statements which of statistical "jargon" was avoided and incorporated a Likert type scale (Table II). emphasis placed on the interpretation of The standard list of statistical terms data. administered as part of the pre-assessment form was repeated as part of the post In order to facilitate the understanding assessment form in order to determine of statistics as a descriptive and whether there was any increase in their interpretative tool in experimental working knowledge of statistics. physiology, we prepared special statistical worksheets for the students. In the first In order to assess whether the student's worksheet, we reinforced some of the perceptions of their understanding of statistical concepts introduced in the statistics determined from the post lecture, by graphically depicting some of the assessment form, matched their actual statistical parameters (normal distribution, performance, students were required to fill mean, SD, positive and negative in a statistical worksheet on a subsequent correlations). The statistical worksheet was practical, without any guidance. This divided into a section on descriptive constituted the "statistics test". Following statistics and a section on inferential the test, a worked out example of the test statistics. In the descriptive section students was put on the notice board so that all were required to compare their own data in students could verify their answers. We relation to the data of the entire class. have earlier demonstrated that such a Statistical terms that were computed in this "formative assessment" exercise with rapid section included mean, SD, range and feedback can be a powerful tool In quartiles. In the inferential section students facilitating learning (9). were required to note and then comment on data that had been subjected to correlations, The statistical analysis of the data was and the 't' test. On the first occasion, the done depending on the nature of the data. student's were instructed on how to fill For categorical data, non-parametric tests these worksheets. A faculty member then were applied; Wilcoxon test for paired data presented the data sequentially to the entire and the Mann Whitney test for independent class using an overhead projector, and the data. For the remainder of the data students filled these data into their parametric tests were used. 456 Vaz et al Indian J Physiol Pharmacol 1998; 42(4)

TABLE II: Distribution of student's perceptions of the statistics exercise, using a Likert-type scale.

Not at all A little Moderately Quite a lot Very much so 1 2 3 4 5 To what extent has this exercise (a) Increased your 2 15 29 11 knowledge of the use of statistics (b) Been enjoyable 1 2 7 24 23 (c) Complicated 29 19 5 2 5 things* (d) Helped you to 1 13 20 23 understand the experiment (e) Helped you to 2 11 25 19 describe the data (0 Helped you to 1 15 22 18 interpret the data**

The numbers indicate the number of reponses obtained. Three students were absent for this Qluestionnare. 'Two and one** student present for the questionnare did not fill up this response.

RESULTS a doctor to have a working knowledge of statistics. 73% (44/60) of the students were When asked to rate the subjects that aware of the importance of statistics in they had done in school in terms of describing data, and a marginally smaller "likability", 43.3% (26/60) rated percentage (68%; 41160) of the importance mathematics among the less likable subjects. in interpreting data. 15% (9/60) felt that Only 38% (22/60) of the students indicated statistics complicated things. that they had received formal training in statistics prior to this exercise. In keeping Following the statistical exercise, the with this - 52% (31/60) rated their own attitudes of the students were evaluated in knowledge of statistics as poor to very poor. terms of a questionnare with Likert-type The remainder indicated that their scaling. This is represented in Table II. It knowledge was adequate. No student rated is clear that a vast majority of students their knowledge of statistics as being either found the exercise enjoyable, and indicated good or excellent. When students were asked that they had increased their knowledge. to tick those statistical terms that they had They also indicated that the exercise helped a working knowledge of from a standard list them to understand the experiment, of 10 Items (normal distribution, mean, de. cribe and interpret data. Despite the fact standard deviation, coefficient of variation, that there were some students who centile, quartile, null hypothesis, continued to feel that statistics complicated correlation, P value, t test) - 87% (52/60) things, an analysis of scores in the pre ticked 4 or less than 4 items. Despite this assessment form revealed that the class as - 83% (50/60) felt that it was necessary for a whole found statistics less complicated Indian J Physiol Pharmacol 1998; 42(4) Biostatistics and Physiology 457

after the presen t statistics exerCIse of knowledge of statistics prior to the (Wilcoxon test; P=0.02). The number of statistical exercise was not a determinant statistical items that they felt familiar with, of performance in the statistics test. The increased from before the exercise (pre; 2.3 50th percentile of marks achieved at the most ± 1.9 vs. Post; 7.6 ± 2.0 items, paired t-test, recent physiology internal assessment P < 0.01). 90% (54/60) of the students felt examination was used to divide students that such an exercise should be contiuned into low academic performers and high during their course and for future batches. academic performers. High academic Despite this very positive response, performers derived greater benefits from the however, only 63% (38/60) felt that statistics statistical exercise in terms of knowledge should be integrated into the examinations. acquired (P< 0.01), and ability to interpret The primary reasons cited for not including data (P < 0.02) than low academic it in examinations was the study load the performers. They also found the exercises students already had and that statistics was more enjoyable (P=O.l), were able to "difficult". describe data better (P=0.07), and were able to understand the experiments more Since there was a wide scatter of (P=O.09), although these differences were attitudes, knowledge and performance not statistically different from the responses during this exercise, we attempted to of low achievers. High academic achievers delineate the determinants of these also performed better in the statistical test variables. Students who listed their although again the difference was not knowledge of statistics as poor before the significant. Students who listed exercise were less likely to appreciate the mathematics among the most likable value of statistics in promoting subjects performed better than students who understanding of experiments as compared indicated a relative dislike for mathematics to students who evaluated their knowledge (Table III). Prior exposure to statistics was as being adequate (3.9 ± 0.9 vs. 4.4 ± 0.7 on not a determinant of performance in the the Likert scale, P < 0.05). However, level statistics test or of attitudes.

TABLE III : Determinants of student performance in the statistics examination.

Low performers High performers P value

Marks obtained in the 7.4±2.5 14.5±2.3 0.000 statistics test (m; x=17) No. of statistical terms 7.1±2.4 8.0± 1.5 0.1 familiar with, post-exercise (max = 10) Degree of "positivity" about 18.1±3.9 19.0±3.4 0.39 the exercise (a+b+d+e+f-c, from Table II) Academic performance 15.7±3.0 16.8±2.1 0.13 (internal assessment marks, max = 25) "Likability" of mathematics 4.1± 1.8 2.7±1.5 0.004 (from 6 subjects at school, 1 = most likable 458 Vaz et al Indian J Physiol Pharmacol 1998; 42(4)

The marks obtained in the statistics test considerably enhanced. For instance, in the was positively correlated with the students experiment of muscle strength, we were able own evaluation of their statistical to test various hypotheses; whether there knowledge, obtained from the number of were gender differences, whether muscle choices they made out of the standard strength was related to muscle mass, and statistical item list r=O.38, P < 0.01). whether gender differences were explained Performance in the test, was also correlated by differences in muscle mass. These with the extent of positivity that the inferences were only possible through the students had about the entire exercise application of statistics. (r=0.28, P < 0.05). This was obtained as a composite figure from Table II (calculated Our study demonstrates that there are as the sum of scores of a, b, d, e and f minus a large number of students who dislike c). There was also a positive correlation mathematics, and that this was a between the students own evaluation of determinant of their performance in the their statistical knowledge and their feelings statistics test. The level of academic of positivity towards the entire exercise (r performance is a determinant of attitudes, = 0.39, P < 0.01l. These analyses reflect the as is a prior knowledge of statistics. As a internal consistency of the data. group, there were several encouraging outcomes of this exercise. Students indicated DISCUSSION a significantly higher number of statistical terms that they had become familiar with, The importance of understanding and results of the statistics test indicated statistics and applying it has been that they were able to apply this new adequately recognised in the medical field. learning, to the extent of their acquired Medical education h s moved away from knowledge. Students also felt that statistics providing information to developing critical was less complicating than they initially thinkers, physicians who can integrate thought. This may in part be because we advances in science and medicine into the refrained from making the exercise too practice of medicine throughout their mathematical; an approach that has been careers (10). The training of students to recommended for beginners (8). Students interpret data is in keeping with this overall also indicated a high level of enjoyment, objective. Traditionally, interpretation of although this may, in part be because we data in student experiments in physiology applied the exercise to humr 1 experiments has been largely qualitative. This paper (11). demonstrates that collation of data and application of simple statistical methods as a routine part of the experimental The integration of statistics into physiology course can contribute towards an experimental physiology will only succeed increased knowledge and the building up of if all Physiology teachers are familiar with positive attitudes towards statistics among the application of statistics. There is a medical students. In addition, the possibility that without adequate inferential scope of experiments is preparation, this integration can evolve into Indian J Physiol Pharmacol 1998; 42(4) Biostatistics and Physiology 459 a meaningless mathematical exercise which The present paper demonstrates that it might actually prejudice students against IS possible to integrate effective statistical learning and applying biostatics. A computer learning into an experimental physiology helps in the exercise that we have devised, course. The vast opportunities during the becau~e it allows for statistics to be physiology practical course allow students generated in close proximity to the data to apply their newly acquired statistics collection. Statistical software in the public knowledge on data that they themselves domain (EPIINFO for example) allows for have generated. This is likely to enhance the application of a wide range of statistical their involvement in the learning process, tests. Students can also be assigned the duty and is distinct from traditional bio-statistics of data entry on rotation, and this will courses which are largely theoretical and facilitate their understanding of applied use examples which may often be far computer skills. removed from the interests of the student.

REFERENCES

1. Gehlback SH. Interpreting the Medical Literature. adult chronic undernutrition. Br J Nulr 1996; 76: Practical Epidemiology for Clinicians. 2'''' Edition. 9-15. New York, Macmillan Publishing Co. 1988. 7. The International Union of Physiological Sciences 2. Linzer M, Brown JT, Frazier LM, DeLong ER, (IUPS) Commission on teaching Physiology. In A iegel we. Impact of a Medical Journal Club on source book of Practical experiments in Physiology Houfle-staff Reading Habits, Knowledge, and requiring minimal equipment. Singapore. critical Appraisal Skills. JAMA 1988; 260: 2537 World Scientific Publishing Co. Pvt. Ltd. 1991: 2541. p.164-165. 3. Guilb rt JJ. Educational Handbook for Health 8. Health Research Methodology. A Guide for Personnel. Revised Edition. Geneva. WHO Offset Training in Research Methods-WHO. New Delhi, Publicati n No. 35 p 1.13. Oxford University Press 1992: p 241. 4. Deurenberg P, Westrate JA, Seidell JC. Body mass 9. Vaz M, Avadhany ST, Rao BS. Student index as a measure of body fatness: age-and sex Perspectives on the role of formative assessment specific prediction formulae. Br J Nulr 1991; 65: in physiology. Medical Teacher 1996; 18: 324-326. 105-114. 10. Ralston III HS, Beattie OS, Bderns KJ et al. 5. Heymsfield SB, McManus C, Smith J, Stevens V, Capturing in the promise of science in Nixon OW. Anthropometric measurements of medical schools. Academic Medicine 1996; 71: 1314 muscle mass: revised equations for calculating 1323. bone-free arm muscle area. Am J Clin Nulr 1982; 11. Padmavathi R, Maruthy KN, Borghona S, Vaz M. 36: 680-690. The perceptions of first year medical students on 6. Vaz M, Thangam S, Prabhu A, Shetty PS. Maximal animal and human experiments in physiology. voluntary contraction as a functional indicator of Indian J Physiol Pharmacol 1998; 42: 127-130.