Wednesday, April 28, 2010

What is there in a name ? Changing the nomenclature of Comm Medicine.

Well, selecting a particular name for a specialty is a very serious issue. Its like we are on the verge of making history. 
Therefore it has to undergo a well structured and organised debate and a highly focussed discussion which involves all the stakeholders (Community Medicine professionals working in all the sectors and past and present students of Community Medicine). This is because we should know that by the nomenclature of Community Medicine, what opportunities and challenges one has faced or is expecting to face in future.
The advantages and disadvantages of a particular nomenclature have to be discussed in depth. For example a MD (Community Medicine) student may face a problem while going abroad where Public Health is only recognized and not Comm Med. This takes away from the student the opportunity just because of the difference in nomenclature. 
The other issue, which I think should be highlighted is; not just the nomenclature; but we should also strive for a standardized curriculum for MD (Comm Med). 
We have a responsibility; that the future generation of Community Medicine professionals should stand up with pride and have a clear sense of direction and understanding and not getting confused as to what and where they belong because of a nomenclature based on no debate.
When the future generation thinks about the name of their specialty they should have clear, logical and evidence based answers (the time to find this is now) rather than coming to know that the nomenclature was decided just by an opinion poll and the name which got the maximum votes got selected. 

We should be driven by evidences and information and not by popular votes.

Suggestions/comments are welcome.

Sunday, April 25, 2010

Signal to Noise Ratio Concept in biostats...

I would like to share this wonderful concept that forms the basis of many theories in biostats.

This originates from the signals received by radio transmitters.If the sound is not clear, we say it is noise. The sound has to cross a certain threshold level to be labelled as a 'signal'. This is analogous to the 'within group variances' (compared to 'noise') and 'between group variances' (compared to 'signal').

Ratio of variances between the groups to variance within the groups is known as F-ratio that we use in comparing more than two groups by ANOVA method.

Now, there emerges four possibilities from this analogy. These are:
1. Signal present and detected
2. Signal present but not detected
3. Signal absent and no signal detected
4. Signal absent but signal detected
The first and third points are detecting the facts. In the second and fourth point, we find that these are errors. Not detecting a signal when it is present is what we call as a type II error (the opposite of which, i.e. detecting a signal when present is power). Similarly, detecting a signal when there is none is the type I error.

I have found my life easier (not in every way but w.r.t. type I and type II errors) after this concept and hope the same for the reader of this post.
Bye for now.....

Sunday, April 11, 2010

Homoscedasticity. Oops !! What's that ?

Homo = same; Scedastic = scatter (which we know by the name of 'variance' in statistics).
Before applying t-test, don't we find a term known as 'equality of variances' between the groups, which is done by the Levene's test. This is for testing homoscedasticity only. This is actually done by dividing variance of one group by variance of another group.
i.e.  .
If this ratio is much different from 1 (for which we always have table already generated by gr8 grandfathers of statistics), we say that the groups are not homoscedasctic.
One more application of this, we commonly find is that in calculating correlations in General Linear Model.
This is homoscedastic, as the scatter is uniform
This is non-homoscedastic (a.k.a. heteroscedastic), as the scatter is non-uniform.
In General Linear Models, to know the correlations, this is one of the assumptions to be met; the other assumptions being that the data should be normally distributed and there should be linearity present in the scatter. bye for now, folks...

Thursday, April 8, 2010

Difference between 'Accuracy' and 'Precision' in statistics

Accuracy is nearing the actual value, where as precision is the degree of reproducibility around the same result.

 High accuracy, but poor precision
 High precision, but poor accuracy

In Epi cluster sampling how does the sample size come to be 210 ?

The formula for sample size calculation for prevalence in community based studies is:
n = (z x z x p x q)/(dxd)
= (1.96)2(0.5)(0.5)/0. 12
= (3.84)(0.25)/0.01 = 96
Taking design effect =2, the total sample size becomes 96 x2 i.e. 192
For a 30 cluster technique, number of subjects to be selected per cluster = 192/30 = 6.4 (rounded up to 7)

That means we have to select 30 clusters, each with 7 units making a total sample size of 30 x 7 = 210 !!!
Interesting, isn't it ??

Wednesday, April 7, 2010

Laverack's domains of Community empowerment

These are the nine organizational areas of influence on community empowerment in a programme context:
(i) participation; (ii) leadership; (iii) problem assessment; (iv) organizational structures; (v) resource mobilization; (vi) links to others; (vii) ‘asking why’; (viii) programme management; and (ix) the role of the outside agents
For more details follow the link:

Indian Diabetes Risk Score developed by Madras Diabetes Research Foundation

CURES also looked at the development of a risk score for diabetes prediction and prevention. One simple way of screening a large population for type 2 diabetes is by development of a simple risk score based on data that can be routinely used at the primary care level. Indian Diabetes Risk Score [IDRS] was developed based on the phase 3 data from CURES. Using statistical analysis, we determined the risk factors, which best predicted diabetes. The IDRS is based on the answers to four simple questions and a simple measurement.
Age [years]< 35 [reference]
35 - 49
> 50
Abdominal obesity Waist <80 cm [female] , <90 [male] [reference]
Waist 80 – 89 cm [female], 90 – 99 cm [male]
Waist > 90 cm [female], > 100 cm [male]
Physical activityVigorous exercise or Strenuous work
Moderate exercise work / home
Mild exercise work / home
No exercise & Sedentary work / home
Family historyNo family history
Either parent
Both parents
Maximum score100
Source : Dr. V. Mohan, et al, J Assoc Physicians India, 2003
An IDRS value > 60 had the optimum sensitivity (72.5%) and specificity (60.1%) and accuracy of 61.3%, for prediction of diabetes in an individual. This simplified Indian Diabetes Risk Score is useful for identifying undiagnosed diabetic subjects in India and could make screening programmes more cost effective as it can reduce the cost by 50% if replaced for screening programmes with blood sugar estimations.

Difference between 'pilot study' and 'pretesting' in research methodology

Most of the times these terms are used interchangeably. The International Development Research Centre Canada site mentions this difference between them:

A PRE-TEST usually refers to a small-scale trial of particular research components.
A PILOT STUDY is the process of carrying out a preliminary study, going through the entire research procedure with a small sample.