Health Information and Basic Medical Statistics

Basic Medical Statistics
Dr. Pracheth R.
Statistics in research
• Tool in designing research, analyzing its data,
conclusions
• Classification, tabulation of data
• Descriptive, inferential
• Primary , secondary data
Simple table
Frequency distribution table
Simple bar chart
Multiple bar chart
Component bar chart
Histogram
Frequency polygon
Line diagram
Pie Chart
Normal Distribution
Continued….
Continued…
Continued…..
• Heights, blood pressure, marks
Standard deviation
• Root-mean square deviation
• Measure of how spread out numbers are
• 68% of values are within 1 standard deviation of
the mean
• 95% of values are within 2 standard deviations
of the mean
• 99.7% of values are within 3 standard deviations
of the mean
Sampling
• Procedure to select representative units
• Probability/Non-probability
• Probability:
Simple random
Systematic random
Stratified random
Cluster
Simple random
• Every unit: population-equal chance
• Population: small, homogenous, numbers easily
available: village
• Select 10 subjects randomly from a population of
100,you can write their names, fold them up,
mix them then pick ten.
Systematic Random
• 100 students (N) in a class and we wish to select
a sample of 20 students (n) , first write the
names of 100 students in alphabetical order
• N/n: sampling fraction (k)
• First: randomly, then every kth
• Easier to draw
Stratified Random
• Heterogenous population
• Population: strata: simple random in each strata
• Add all samples
• Example: Socio-economic status- 5 classes
Cluster sampling
• Heterogenous population
• Group units : geographical location
• Select clusters from population-simple random
sampling
• From selected cluster: each and every unit
included
• Assess vaccination coverage
• Completed lists not available: useful
Non-Probability
• Purposive/judgement
• Incidental sampling
• Convenient: telephone directory, registers
Null Hypothesis (H0)
• How far difference between two things –real or
due to chance: testing hypothesis
• No difference between two drugs
• Smoking has no role in cancer
• Two groups: smokers, non-smokers: compare
• Difference more than twice standard error
(variation among sample means): null hypothesis
rejected
• p<0.005= difference real- H0 rejected
Continued….
• Type 1 error: H0 rejected even though it should
be accepted
• Type 2 error: Accepting a false H0
Tests of Significance
• Tests to measure chance of occurrence of
biological variation
• Mean weight gain: 2 kgs – nutritional
supplement, 1 kg: control
• Whether difference of 1 kg due to chance/
supplement
• Measures p value
Continued….
• Quantitative data:
SE of mean
SE of difference between 2 means
Sample size > 30: Z test
Sample size < 30: t test
• Qualitative data:
SE of proportion
SE of difference between two proportions
Chi-square test
SE of mean
• Distribution of sample means about the true
mean
• One sample: SD, how accurate is the sample ?
What about true mean ?
• Calculate SE of mean, set up confidence limits
• 25 males; 20-24 years, mean temp: 98.14 deg F,
SD=0.6
• SE= SD/ √n=0.6/ √25= 0.12
• 95% confidence limits: population mean:
98.14±(2X0.12)= 97.90 to 98.38 degree F
SE of difference between 2
means
Continued….
• Real difference: 370-318-=52 > twice SE (2X7.5)
• Significant
Sample size > 30: Z test
Sample size < 30: t test
SE of proportions
• Proportion of males in village: 52 %
• Random sample: 100 people-40%
• Proportions: p and q
• SE= √(pq/n)= √(52X48/100)= 5
• 2 SE: either side of 52
• 52+2(5)= 62 and 52 -2(5)=42
• Observed proportion: 40%- significant
SE of difference between two
proportions
• Trial of two whooping cough vaccines
Continued….
• Observed difference: 24.4-16.2= 8.2
• 2X SE= 2X6.02= 12
• Not significant
Chi-square test
• Find out the association between 2 attributes
• Categorical variables
Contingency table
Right-handed
Left-handed
Total
Males
43
9
52
Females
44
4
48
Totals
87
13
100
THANKS A LOT