Indmedica Home | About Indmedica | Medical Jobs | Advertise On Indmedica
Search Indmedica Web
Indmedica - India's premier medical portal

Indian Journal of Community Medicine

Predictors of Nutritional Knowledge Levels of Medical Practitioners - Use of Discriminant Function Analysis

Author(s): Charu Kartik, Tulsi Adhikari, Padam Singh & Mohini Sethi

Vol. 29, No. 3 (2004-07 - 2004-09)

Abstract:

Objectives: To find out the predictors of nutritional knowledge levels of medical practitioners using the descriminant function analysis.

Methods: The study, involved collection of baseline information on existing nutrition knowledge of medical practitioners using multiple choice exercise (MCE). The data collected in the study have been used to determine the predictors of nutritional knowledge using discriminant function analysis. The three situations studied are base line knowledge level, post-intervention knowledge level and change in the knowledge level. For this 2 extreme groups have been considered, using statistical criteria, i.e., poor with the levels less than mean - SD and good with levels more than mean + SD.

Setting: Department of Cardiology, Nephrology, Endocrinology and Medicine in Hospitals (2 Govt. & 2 Private) of Bikaner.

Participants: 240 medical practitioners from 4 hospital of Bikaner.

Results and conclusions: The discriminant analysis showed a very high discriminating power with the cases correctly classified ranging from 80-90% in the three situations. The study findings could be used in targeting those who need such intervention the most.

Introduction:

Nutrition has not received due recognition and attention in medical education. The need for improving nutritional knowledge of medical practitioners in Indian situation had been well recognised. Suneja and Bhat[1] in a study on the knowledge of medical doctors of government and private hospitals in Hissar (Haryana) observed that correct responses in therapeutic nutrition ranged from 38% to 66% in case of government doctors and (18%-49%) in case of private practitioners. The study also showed that most of the doctors were aware of the type of diet to be recommended but they were unaware of the nutritional contributions made to the disease management of the patients. Mahapatra et al[2] in a study on knowledge and functioning of primary health centres (PHC) medical officers (MO) in nutrition and related aspects found that knowledge scores ranged from 33-63%. Knowledge about nutritive value of foods and diet during diseases were assessed by Kapil et al[3], amongst 49 nursing students at the beginning of their training course in human nutrition. Majority of the students had adequate knowledge about dietary advice during antenatal and postnatal period. However, the knowledge of nutritive value of foods and diet in pathological conditions was found to be inadequate. Bhattacharji et al[4] evaluated an educational program on the importance of nutrition and its applications planned for medical and health science students. Pre and post test comparison showed a definite advantage of imparting such education through lectures to medical personnel as there was a significant improvement in their knowledge scores.

As the curriculum of MBBS does not provide adequately for nutritional aspects and hence a need for providing necessary nutritional knowledge to medical practitioners. With this, Charu Kartik[5] has undertaken a survey of 240 medical practitioners in Bikaner district of Rajasthan state (India). These medical practitioners were selected from 4 hospitals, 2 each from government and private. These medical practitioners were from the four O.P.D's covering the specialities of diabetes, cardio-vascular, renal diseases and obesity. The objective of the survey was to assess the nutritional knowledge levels of medical practitioners as well as to plan an intervention to improve the knowledge levels. Baseline knowledge levels were assessed based on 30 questions through a multiple choice exercise. Four dietary reckoners were constructed as intervention tools for improving the physicians' nutritional knowledge for specific dietary application in their clinical procedures and prescriptive practices. The medical practitioners were advised to use these reckoners in their practice for a period of 3 months to enable them to provide proper dietary prescription to the patients. After the intervention the nutritional knowledge of the medical practitioner was reassessed to find out the impact of the intervention. The data collected in this study have been utilised in studying the improvement in the knowledge of medical practitioners with the intervention.

The objectives of this paper is the use of Discriminant Function Analysis[6] for identifying the factors responsible for classification of a medical Practitioner into Poor and Good Categories on the base line knowledge level and the post intervention knowledge level as well as to predict the likely changes due to intervention among medical practitioners with different background characteristics and initial knowledge levels.

Material and Methods :

Sample Selection: The study was undertaken in the following four hospitals, government hospitals i.e. PB Memorial Hospital and Satellite Hospital and private hospitals i.e. challank hospital and M.N. Hospital.

From these, the sample of medical practitioners was taken from the department of cardiology, nephrology, endocrinology and medicine having well established OPDs. Out of the list of 300 medical practitioners (both MBBS & MD) a sample of 240 was taken. The sample size has been worked out using the formula

The assumptions made are:

  1. Nutritional knowledge of 40% = p
  2. Precision of 10% = d
  3. Confidence level of 95%
  4. Design effect of 2.5 = D

The value of n obtained was 240.

Statistical Technique: The response was obtained for 30 questions by multiple choice exercise. The knowledge level was assessed as percentage of correct responses. The estimation of baseline as well as post intervention knowledge levels was made. Based on the baseline knowledge levels, all medical officers were categorised into three groups viz; Poor, Average and Good. These were based on mean + SD classification. The knowledge score less than mean - SD was classified as poor; between mean - SD and mean + SD as average and more than mean+SD as good.

Thereafter the discriminant function analysis has been undertaken for two extreme groups viz; Poor and Good. Similar categorisation has been done for post intervention knowledge levels as well as change in the knowledge levels.

In discriminant analysis, a linear combination of the independent variable is formed which serves as the basis for assigning individuals to groups. The linear discriminant equation is given by

D = B0 + B1X1 + B2X2 + ... + BpXp

4 pq

n =

* D

The X's are the values of the independent variables and the B's are coefficient estimated from the data. If a linear discriminant function is to distinguish between individuals in two groups must differ in their D values. In discriminant analysis, the weight are estimated so that they result in the "best" separation between the groups. Therefore, the B's are chosen so that the values of the discriminant function differ as much as possible between the groups.

Based on the discriminant functions so worked out the likely levels of medical practitioners has been predicted according to background characteristic so as to identify those for whom the intervention should be focused.

Results:

Knowledge Levels: As per the findings of Charu Kartik (2001), the mean and SD of knowledge levels are 36.2 and 3.87 respectivily and after post intervention are 90.6 and at baseline 3.9 respectively. Cross tabulation of the medical practitioners according to baseline knowledge and Post intervention knowledge levels are presented in table-I.

Discriminant Function Analysis: The result of Discriminant Function Analysis for categorising the medical practitioners as poor or good are discussed as under

a) For baseline knowledge level: The cutoff points for classification are the Baseline knowledge < 32.41 as poor and > 40.17 as good.

The independent variables are

  • X1 = Type of hospital = 1 for Government, = 2 for Private
  • X2 = Qualification of the medical practitioner = 1 for MBBS, = 2 for MD
  • X3 = Sex of the practitioner = 1 for Male, = 2 for Female
  • X4 = Age of the practitioner = 1 for 35-40 age group, = 2 otherwise

Of the four variables only X1, X2 and X4 came as significant and discriminant function so obtained is as under

D = -3.846221 + 1.797303 * X1 + 1.158520 * X2 - 0.3048058 * X4

The classification of medical practitioners based on above discriminant function is given in table -2.

b) For post intervention knowledge: The cutoff points for classification are post intervention knowledge < 66.63 as poor and > 74.62 as good.

The independent variables are same as in (a) above except X5 which is the baseline knowledge

Baseline score = X5

Of the five variables X1, X2, X4 and X5 came as significant.

Discriminant function obtained is given by

D=-9.140223+1.419886*X1+.9474066*X2+0.1009737*X4+ 0.1333070*X5

The classification of medical practitioners based on above discriminant function is given in table -2.

c) Change in knowledge level: The cut off points for classification are change in the knowledge level <29.83 as poor and >40.17 is as good.

The independent variables are same as in (b) above.

Of the five variables X1, X2, X4 and X5 came as significant and the discriminant function obtained is given by

D=-11.51706-1.532651*X1-0.4466475*X2+0.02238424*X4+0.4001755*X5

The classification result based on above discriminant function is given in table-2.

Discussion:

Knowledge Level: The baseline knowledge levels were low in general. Those with MBBS qualifications had low levels as compared to those with MD. Further those in government hospitals had low levels as compared to those in private. These findings are in line with others.

Improvement in knowledge Level: It is observed that of those who were in the group of low knowledge 25-30 percent (although only 28 out of 240), more than two third moved in the category of 65-70 percent. Those who were in the category of average 30-35 percent (67 out of 240) more than 70% moved to the category 65-70%. From those who were in the category of good (35-40%) (128 out of 240) over 50% moved to 65 to 75%. Further, those who were in the last category of 40-45, about 2/3 reached to the level 75-80%.

Correct Classification by Discriminant Function Analysis: Based on the results of discriminant analysis;

The percent of "grouped" cases correctly classified for baseline knowledge levels = 82.22%

The percent of "grouped" cases correctly classified for post intervention knowledge level = 80.00%

The percent of "grouped" cases correctly classified for change in knowledge levels = 90.16%

It would appear that the discriminant functions obtained have good discriminating power in the three situations studied. The precent of cases correctly classified are 84%, 80% and 90% respectively, for baseline knowledge level, post intervention knowledge level and change in the knowledge level. For baseline knowledge level, the significant variables are type of hospital, qualification and age. In other cases, the significant variables are type of hospital, qualification, age and baseline knowledge levels.

The study highlights that such intervention should be focused for all specially those in the government hospitals and there may be need for re-orientation to bring them at par with others.

Table - 1 : Cross tabulation of distribution of Medical Practitioners according to base line knowledge and post intervention knowledge.

Base line Knowledge Post Intervention Knowledge
55-60 60-65 65-70 70-75 75-80 N
25-30 3.6 3.6 67.9 21.4 3.6 28
30-35 .0 7.5 71.6 16.4 4.5 67
35-40 .8 5.5 50.8 32.0 10.9 128
40-45 .0 .0 6.3 25.0 68.8 16
45-50 .0 .0 .0 .0 100 1
Total .8 5.4 55.8 25.8 12.5 240

Actual and Predicted Results according to Discriminant analysis.

  Actual Group Predicted Group
    Poor % Good %
a) Baseline knowledge
  Poor 23 (82.1%) 5 (17.9%)
  Good 3 (17.6%) 14 (82.4%)
b) Post Intervention knowledge
  Poor 13 (86.7%) 2 (13.3%)
  Good 7 (23.3%) 23 (76.7%)
c) Change in knowledge
  Poor 20 (100) 0 (0)
  Good 6 (14.6%) 35 (85.4%)

References:

  1. Suneja, N. and Bhat, C.M. Nature and extent of nutritional knowledge of medical practitioner in Hissar district. Ind J Nutr Dietet. 1985; 22(4):104.
  2. Mohapatra, B.,Ramadasmurthy, V., Ramnath, T. and Mohanram, M. Knowledge and functioning of PHC medical officers in nutrition and related aspects. Proc Nutr Soc India. 1987;33:127.
  3. Kapil, U., Manoch, S and Sood, A.K. Knowledge of nutritive value of foods and diet during diseases amongst nursing students. Indian Pediatr. 1990; 27(4): 361-365.
  4. Bhattacharji, S., Joseph A., Abraham S., Mulirjit, S., John, K.R. and Ethirojan, N. Teaching nutrition to medical students. Community based problem-solving approach in medical education. 1990; 24(1): 32-36.
  5. Charu Kartik (2001), Ph.D. Thesis submitted to Delhi University, Nutritional Knowledge of Medical Practitioners In Hospitals; An Assessment.
  6. Snedecor, G.W. and Cochran, W.G. : Statistical Methods, Oxford IBM Publication, London (1967).
Access free medical resources from Wiley-Blackwell now!

About Indmedica - Conditions of Usage - Advertise On Indmedica - Contact Us

Copyright © 2005 Indmedica