2. INGREDIENTS NEEDED TO CALCULATE
SAMPLE SIZE FOR MEDICAL AUDIT
1. Target performance (Proportion)
2. Number of population
3. The degree of confidence wanted in the findings
4. Level of precision
5. Resource constraints (time, access to data, costs)
Target population Sampling Frame
Calculate the
priori sample
size
3. EXAMPLE :
A primary care team has planned an audit of the treatment of hypertensive patients. They are
treating 300 people for the illness, but they do not have time to review all the documents. They
choose one critical criterion and aim for a performance level of 70%. Those receiving therapy
should have measured their blood pressure and the result should have been below 150/90
mmHg on three occasions in the previous 12 months. They are prepared to accept a 5 per cent
sampling-related error . Based on the information given, how much minimum sample size
needed?
Population
Target Performance
Level of precision
6. When the target performance
increases the sample size required
will be lesser.
Effect of accuracy on estimating sample
size
7. Don’t forget to inflate your sample size 10
to 20% from your estimated sample size
Retrospective data has bigger risk of
incompleteness
8. Exercise :
Objective: The study aims to assess the percentage of children with ASD who
engage in occupational therapy services and subsequently receive
intervention for sensory issues.
Target population:
Clinical (patient) records of children with ASD who have received assessment
and intervention for sensory difficulties who have received an ASD diagnosis
based on criteria outlined in the Diagnostic and Statistical Manual of Mental
Disorders, 5th edition (DSM-5) and who were aged 12 years or less when
commencing therapy.
Sampling frame: 500 from the data registry from the hospital
Target of performance: The consensus from the organization report was that
approximately 95% of children with ASD who engage in occupational therapy
services subsequently receive assessment and intervention for sensory
difficulties.
Level of accuracy: 5 %, Confident interval 95%
What is the estimated sample size required?
9. Answer:
To account for the degree of
missing data, inflate another
10%
Estimated sample size = 71
10. Was the standard met?
Sampling is the process of selecting a sub-group of individuals, or
observations that can accurately represent the characteristics of the entire
target population
12. Analyzing
data and
evaluating
information :
Was the
standard
met?
The aim of data
analysis involves
interpreting the
collected data to
discover how the
current practice
compares to the
agreed criteria and the
standards.
The analysis can range
from simple descriptive
statistics to complex
statistical methods.
However, on most
occasions, a simple
and easy to
understood method is
preferred.
Indeed, if the result
requires
implementation the
analysis must be
simple enough for
everyone involved in
the care process able
to comprehend
It identifies the areas
both of
underperformance,
which should be
reviewed in detail to
identify why the care
falls below the desired
levels and how it can
be improved.
13. How to communicate the data:
It is often most effective to begin by deciding what the story (the message) is and to whom (the audience) it will be told. An
unnecessarily complex presentation is an obstacle to effective communication.
For most audiences, graphs and tables are much more easily understood than regression coefficients.
Keep figures simple and neat, label axes, and avoid formats that are likely to confuse readers. The benefits of an appealing
image are not worth the risk of misrepresenting data or message
Descriptive statistics give a summary of the sample being studied without drawing any inferences based on probability
theory.
Descriptive statistics can be broadly put under two categories :
Sorting/grouping and illustration/visual displays Summary statistics.
Frequency table *
Bar chart *
Histogram
Pie chart
Box plot
Central tendency
• Mean – Average of value
• Median – The mid-point of all the value
• Mode - Most commonly occurring value
Measures of dispersion/ how the data spread
• Range
• Standard Deviation
• Variance
14. Example:
An Audit of Type 2 Diabetes Care in a Malaysian Public Community Polyclinic
The researchers set out to study the diabetes care process at one of the public health
clinics via a medical audit exercise focusing on the care process.
References : Chew BH, Palikat JM, Nur Syamimi A, Nor Azillah A, See JK, Hafiz AR, Siti Nazira A, Aienum Y, Muhamad Zaid M, Palaniappan K, Tan YY. An
Audit of Type 2 Diabetes Care in a Malaysian Public Community Polyclinic. Pertanika Journal of Science & Technology 2012;20(1):97-102. 12. Shariff Ghazali
S, Abu Hassan Z, Ali Z, Ian O. A clinical audit on diabetes care in two urban public primary care clinics in Malaysia. Malaysian Journal of Medicine and Health
Sciences 2010;6(1): 101-109.
15. Let’s do some
revision on basic
statistic :
The usual type of data resulting
from a medical audit is
categorical data and would most
commonly be represented as a
bar chart or a pie chart.
Numerical scale and plot data
against it e.g. scatter graphs,
box and whisker plots, etc.
Variable
measurable
quantities that
varies among
individual
Parameter do not
relate to actual
measurement,
(individual scalar but
it quantifies defining
a theoretical model
Reference: Altman D G, Bland J M. Statistics notes Variables and
20. Thank you.
Don’t forget to join
our next hybrid
webinar :
1/ Identifying and
Managing Missing
Data and Outlier in
Clinical and Health
Sciences Research
2/ Data
Management with
REDCap :
INSIGHTS FOR
RESEARCHER