Statistics and Research for the FRCPath Part 1 Dr. Vijay Sharma Consultant Histopathologist and Honorary Senior Lecturer Liverpool Clinical Laboratories Royal Liverpool Hospital.
Topic 1
CLINICAL AUDIT & RESEARCH
Audit and research - similarities Both answer questions about quality of care or service.
Both can be carried out either prospectively or retrospectively retrospectively prospectively
Both involve sampling, data collection and analysis of findings.
Audit or research? Research
Audit
What is the most effective way of treating pressure sores?
How are we treating pressure sores & how does this compare with accepted best practice?
Measure outcomes for the best treatment
The audit would measure process (are we doing the things we should do?) To monitor the success of a treatment which is known to work
Audit and research - differences Audit
Intent
Research
“Research is concerned with discovering the right thing to do:Treatment and Audit is ensuring it isservice done right” (Smith, R. 1992, Audit & Research, BMJ, 305:905-6)
Allocation
Randomisation
Audit and research - differences
Clinical Audit and Effectiveness, Leaflet developed by Joanne Hill from an original idea by UBHT NHS Trust Clinical Audit Department. 2005
EMQ Option A Measures current service against a standard
Option B Designed and conducted to generate new knowledge
Option C Involves an intervention in use only
Answer Question Please assign the above options to describe: -Clinical audit -Research
A, C, E B, D
Option D
Option E
Designed and Addresses conducted to clearly defined produce questions, aims information to and objectives inform delivery of best care
MCQ
Question Answer Clinical audit ultimately is a Qualityimprovement improvementprocess process A. Quality B. Data collection process C. Management improvisation D.Monitoring and reporting process
Why • Improve patient care • Mandatory? – “all healthcare professionals need to understand the principles of clinical audit, and the organisations in which they work must support them in undertaking clinical audit”
What are we trying to achieve?
Have we made things better?
Do something to make things better
Are we achieving it?
Why are we not achieving it?
Ross S, et al. Principles for best practice in clinical audit. Oxon: Radcliffe Medical Press, 2002
Ross S, et al. Principles for best practice in clinical audit. Oxon: Radcliffe Medical Press, 2002
Audit standards •
Clinical Audit is directly related to improving services against a standard that has already been set
“Standard” : •
The level of care to be achieved for any particular criterion –
(Irvine D, Irvine S. Making Sense of Audit. Oxford: Radcliffe Medical Press, 1991)
“Criterion”: •
A systematically developed statement that can be used to assess the appropriateness of specific healthcare decisions, services, and outcomes –
(Institute of Medicine, 1992)
Resources
Ross S, et al. Principles for best practice in clinical audit. Oxon: Radcliffe Medical Press, 2002
Topic 2
STATS
Statistics By chance?
Design and protocol development
Data collection
Data management
Project life span
Data analysis
Reporting results
Other terminologies Incidence
Rate at which new disease diagnosed
Prevalence
Number of people who are alive and who are known to have the disease (i.e. how widespread the disease is)
Sensitivity
Disease status
Specificity
Test result Number
Number of true positives Number of true positives + number of false negatives
of true negatives Number of true negatives + number of false positives https://en.wikipedia.org/wiki/HIV_test
Data distribution 25
Frequency
20
15
10
5
0 138 140 142 144 146 149 151 153 155 157 159 161 163 165 167 169 171 173 175 177 179 181 183 185 187 189 191 193 195
Height
Choose your test Type of data
Normal
Metric
Skewed
Categorical
Ordinal
Nominal
No staining 0 Weak staining 1 Strong staining 2
Female Male
Standard deviation
Paired/unpaired
Paired
Before vs. after treatment
Unpaired
Control population vs. test population
Your data Categorical
Metric Ordinal Unpaired
Paired
Do data have: - normal distribution - similar standard deviations
Paired
Nominal
Unpaired
Unpaired
Paired
Large number of categories?
Differences normally distributed?
no
yes
no
yes
no
Sample size
yes small
Statistical tests T-test (unpaired)
T-test (paired)
Wilcoxon signed rank test
Mann-Whitney U test
Fisher’s exact test
large
Chi-squared test
McNemar’s test
MCQ
Question Answer
If the outcome data are metric, unpaired and show normal distribution with similar standard deviations, the appropriate statistical test to use is:
A. Chi-squared test B. Mann-Whitney test C. Two-sample t test D. Fisher’s exact test E. McNemar’s test
EMQ Option A
Option B
Option C
Option D
Option E
Mann-Whitney U test
Chi-squared test
McNemar’s test
Fisher’s exact test
Two sample ttest (unpaired)
Option F
Option G
Option H
Option I
Option J
Wilcoxon signed rank test
Paired t-test
Question Answer A sample of teenagers are divided into male and female on the one hand, and based on dieting on the other. We hypothesize that the proportion of dieting individuals is higher among the women than among the men, and we want to test whether any difference of proportions that we observe is significant
EMQ Option A
Option B
Option C
Option D
Option E
Mann-Whitney U test
Chi-squared test
McNemar’s test
Fisher’s exact test
Two sample ttest (unpaired)
Option F
Option G
Option H
Option I
Option J
Wilcoxon signed rank test
Paired t-test
Question Answer Subjects are tested prior to a treatment, say for high blood pressure, and the same subjects are tested again after treatment with a blood-pressure lowering medication
EMQ Option A
Option B
Option C
Option D
Option E
Mann-Whitney U test
Chi-squared test
McNemar’s test
Fisher’s exact test
Two sample ttest (unpaired)
Option F
Option G
Option H
Option I
Option J
Wilcoxon signed rank test
Paired t-test
Question Answer The ratio of male to female students in the Science Faculty is exactly 1:1, but in the Pharmacology class over the past ten years there have been 80 females and 40 males. Is this a significant departure from expectation?
Resources • Which test – http://www.youtube.com/watch?v=rulIUAN0U3w (Choosing which statistical test to use - statistics helpCreativeHeuristics - YouTube) – http://www.measuringusability.com/blog/what-test.php – http://www.graphpad.com/support/faqid/1790/
• Statistics – https://www.khanacademy.org/math/probability