Introduction to Statistical Methods (SBI203)

10 credits

Aim of this module

The aim of this rotation is to introduce statistical methods and their use in public health and epidemiology.The use of statistical techniques is intrinsically incorporated into the work of epidemiologists. In order to derive the information on which important public health decisions are based, a relevant statistical method has to be applied to data to obtain statistics relevant to the question being answered.

The trainee should develop an understanding of the role of informed statistical advice in how to approach the process of answering questions of public health importance. They will gain an appreciation of the availability and quality of routinely collected data and its suitability to the question at hand. They will also begin to appreciate the many statistical approaches that can be taken to answering the question at hand and how to go about selecting one to use.

They will understand the importance of prior planning of a statistical analysis, how to interpret the results obtained and how to convey these results in the form of a written report to the intended audience.

  1. Identify a pertinent public health question and draft a statistical analysis plan.
  2. Execute the statistical analysis plan, producing a written report to summarise and interpret the findings.
  3. Calculate the sample size required for an analytical study, or the power of a study that has already been conducted, and discuss the impact of this on the study and the interpretation of findings in relation to patients or the public.
Number Work-based learning outcome Title Knowledge
1 1

Identify a relevant public health question.

View
2 1

Identify a relevant data set for analysis and describe the variables.

View
3 1

Identify the relevant statistical methods to answer the question and draft an analysis plan for consultation with a colleague.

View
4 2

Use appropriate software to summarise variables.

View
5 2

Calculate relevant statistics to address the public health question.

View
6 2

Interpret the meaning of the point estimate and confidence interval and make inferences about the population.

View
7 2

Draft a report detailing the data analysis, interpretation and conclusions.

View
8 3

Determine the type I and type II error that is acceptable.

View
9 3

Estimate the proportion exposed among the controls or the attack rate among the unexposed using available data or from the literature.

View
10 3

Determine the appropriate magnitude of effect that you wish to detect.

View
11 3

Use appropriate statistical software to calculate sample size and adjust sample size for non-response.

View
12 3

Using data already available use above steps to calculate the power of the study.

View
You must complete
1 Case-based discussion(s)
1 of the following DOPS / OCEs
Assessment Title Type

Important information

The academic parts of this module will be detailed and communicated to you by your university. Please contact them if you have questions regarding this module and its assessments. The module titles in your MSc may not be exactly identical to the work-based modules shown in the e-portfolio. Your modules will be aligned, however, to ensure that your academic and work-based learning are complimentary.

Learning Outcomes

  1. Summarise and present data appropriately.
  2. Describe statistical techniques to quantify sampling variation.
  3. Select an appropriate statistical method for the analysis of data.
  4. Perform statistical analysis of dummy data.
  5. Interpret the findings of the statistical analysis.
  6. Describe the application and the strengths and weaknesses of different sampling scheme.
  7. Perform sample size and power calculations for a case control or cohort study.

Indicative Content

  • Summarise and present data appropriately
    • Display tables: frequencies, frequency distributions and graphs
    • Proportions, differences, ratios, % change
    • Summary statistics, e.g. mean, mode, median, measures of spread (range, interquartile range, variance, standard deviation)
  • Describe statistical techniques to quantify sampling variation
    • Data distributions
    • P-values and confidence intervals
    • Statistical vs clinical significance
  • Select an appropriate statistical method for the analysis of data
    • Tests for comparing means
    • Tests for comparing proportions
    • Non-parametric methods
    • Correlation and linear regression
  • Stratified analysis: Mantel-Haenzel, standardisation: direct and indirect
    • Introduction to linear and logistic regression
    • Survival analysis (classical approach)
    • Censoring mechanisms
    • Types of time-to-event data
    • Principles of Kaplan-Meier and actuarial survival curves
    • Methods of summarising survival data
    • Methods used to compare groups, e.g.:
    • logrank test for two or more groups, including ordered groups
    • Cox's proportional hazards regression model
    • Hazard ratios and their interpretation
  • Perform statistical analysis of dummy data
    • Overview, strengths and weaknesses of a range of statistical packages for epidemiological analysis
    • Practical computer simulation using dummy data, e.g. using R
    • Overview of the data set
    • Listing observations
    • Frequency tables for categorical variables
    • Description of distributions
    • Test of means
    • Non parametric tests: Kruskal-Wallis, Wilcoxon, Mann-Whitney, Spearman and Kendall’s correlation
    • Stratification
    • Regression techniques
    • Survival analysis
  • Interpret the findings of the statistical analysis
    • Interpret outputs from computer-generated analysis
    • Describe the meaning of the output in terms of public health
  • Describe the application and the strengths and weaknesses of different sampling schemes
    • Simple random
    • Systematic
    • Stratified
    • Multi-stage
    • Cluster sampling
  • Perform sample size and power calculations for a case control or cohort study
    • Type I and II errors
    • Proportion exposed or has outcome
    • Expected magnitude of effect