FOUNDATIONS

What is evidence-based health care?

Management of patients using the current best evidence of effectiveness.

What are the 5 steps of practising evidence-based health care?

Ask – Turn uncertainty into a clear question

Search – Find the best available evidence

Evaluate – Check if evidence is valid and useful

Apply – Use it in practice

Review – Assess the results

What are the 3 components of evidence-based healthcare?

Patient choice

Best available evidence

Clinical experience

What is epidemiology?

Study of disease distribution in populations, its determinants, and applying this knowledge to prevent and control health problems.

What are the 3Ds of epidemiology?

Distribution, Disease, Determinants

Name 3 common types of epidemiological comparisons.

• Case-Control Study

• Cohort Study

• Experimental Study

What is biostatistics?

The use of maths and data to study health, diseases, and treatments.

What are the steps in designing a study?

Ask clear, answerable questions

Review existing research

Choose a study design

Select participants

Decide on data to collect and collection method

Collect and clean data

Summarise data

Choose statistical tests

Make conclusions

Share/publish results

Define ‘population’.

The total group of people (or things) being studied.

Define ‘sample’.

A smaller group taken from the population.

Define ‘sampling frame’.

The group from which the sample is selected.

What is a simple random sample?

Each individual has an equal chance of being chosen.

What is a population parameter?

An unknown value in the entire population

What is descriptive statistics?

Statistics that describe a sample.

What is inferential statistics?

Data used to make conclusions about a population.

What is a variable?

Anything that can change or vary.

Define independent variable (IV).

The variable changed by the researcher.

Define dependent variable (DV).

The outcome measured in response to the IV.

What is an extraneous variable?

A variable that could affect results but is controlled.

What is a confounding variable?

A variable that affects the results and is hard to control.

What are demographic variables?

Personal details (e.g. age, sex, etc.).

Define exposure.

Something that might affect the outcome.

What is sampling variation?

Slightly different results from different samples.

What is sampling error?

The difference between the study estimate and the true population value.

What are categorical variables?

Variables that group data by characteristics (e.g., gender).

What are continuous variables?

Variables measured numerically (e.g., height, weight).

What’s the difference between interval and ratio variables?

Interval: Zero doesn’t mean ‘none’

Ratio: Zero means ‘none’

What are the four scales of measurement?

1. Nominal – Categories only (e.g., gender)

2. Ordinal – Ordered but not equally spaced (e.g., rankings)

3. Interval – Numbers with no true zero (e.g., temperature)

4. Ratio – Numbers with true zero (e.g., weight)

What are the two types of statistics?

Descriptive: Summarise sample data

Inferential: Make predictions about a population

What is central tendency?

Typical value – mean, median, mode

What is dispersion?

How spread out values are – range, standard deviation

What are the types of study designs?

Descriptive

Analytic (Observational or Experimental)

What is the difference between observational and experimental studies?

Observational: No intervention.
Experimental: Researcher applies an intervention.

What is PICO used for?

To structure experimental research questions:

P = Population, I = Intervention, C = Comparison, O = Outcome

What is PECO used for?

For observational studies:

P = Population, E = Exposure, C = Comparison, O = Outcome

What is prevalence?

Total cases (old and new) in a population.

How do you calculate prevalence?

Prevalence = (Cases ÷ Population) × 100

What is cumulative incidence?

New cases over time ÷ population at risk at start

What is incidence rate?

New cases ÷ person-time at risk

Why do we use samples in studies?

To represent the population and make inferences.

What are the two types of sampling?

Probability (random)

Non-probability (non-random)

Give examples of probability sampling.

Simple random

Systematic

Stratified

Cluster

Multistage

Give examples of non-probability sampling.

Convenience

Quota

Purposive

Snowba

What is internal validity?

Accuracy of results for those studied.

What is external validity?

Whether results apply beyond the study.

What is reliability?

Consistency of results.

What is random error?

Unpredictable fluctuations in measurement.

What is systematic error (bias)?

Consistent errors due to the method/tool.

What are types of bias?

Selection bias

Information bias

Confounding

What is a normal distribution?

A bell-shaped curve where most data is around the mean.

What does the 68-95-99.7 rule mean?

68% within ±1 SD

95% within ±2 SD

99.7% within ±3 SD

What is a Z-score?

A measure of how far a value is from the mean in SDs.

Formula: Z = (Score − Mean) ÷ SD

What is a descriptive study?

Describes characteristics without looking for cause.

What is an analytic study?

Examines relationships between variables.

What is a case report?

Detailed report on one individual case.

What is a cohort study?

Follows a group over time to see who develops the outcome.

What is statistical inference?

Making generalisations from a sample to a population.

What is estimation?

• Point Estimate: Single best guess

• Interval Estimate: Range with a confidence level

What is NHST (Null Hypothesis Significance Testing)?

Testing if there’s enough evidence to reject a null hypothesis.

What is a p-value?

The probability the results are due to chance.

What is a two-tailed test?

Checks for differences in either direction.

What is a one-tailed test?

Checks for a difference in one direction only.

What is a Chi-Square test?

A non-parametric test for categorical data.

What are Chi-Square assumptions?

• Random sample

• Independent observations

• Adequate expected counts

What is a risk ratio (RR)?

Likelihood of an outcome in exposed vs. unexposed group.

What does an RR of 1 mean?

No association

What is a prospective cohort study?

Starts with healthy people, follows them to see who gets the condition.

What is a retrospective cohort study?

Uses past records to assess exposures and outcomes.

What is a case-control study?

Compares people with and without a condition by looking back at exposures.

What is the odds ratio (OR)?

Measure of association in case-control studies.

What is a one-sample t-test?

Tests if a sample mean is different from a known value.

What is Levene's test?

Checks if variances in groups are equal.

What is Cohen's d?

Measures the size of the difference between two means.

What is a Type I error?

False positive – rejecting a true null hypothesis.

What is a Type II error?

False negative – not rejecting a false null hypothesis

What is study power?

The chance of detecting a real effect if it exists.

What’s the difference between experimental and quasi-experimental designs?

Experimental: Uses randomisation

Quasi-experimental: No randomisation

What improves internal validity?

Random assignment and control of variables.

What is a one-sample t-test?

A statistical test that compares the mean of a single group to a known value (e.g., population mean).

Example: Comparing a group's average IQ to the national average of 100.

What is an independent samples t-test?

A test comparing the means of two separate, unrelated groups.

Example: Comparing average exam scores of males and females

What is a paired samples t-test?

A test comparing means from the same group measured at two different times or under two conditions.

Example: Measuring anxiety levels before and after a course.

What are the steps of Null Hypothesis Significance Testing (NHST)?

1. State the null (H₀) and alternative (Hₐ) hypotheses

2. Set the significance level (α), usually 0.05

3. Choose the appropriate test and calculate the p-value

4. Compare p-value to α and decide to reject or not reject H₀

5. Assess effect size for practical significance

What does the null hypothesis (H₀) represent?

It represents no effect or no difference.

Example: H₀: µ1 = µ2 — The two group means are equal.

What does the alternative hypothesis (Hₐ) represent?

It represents that there is an effect or a difference.
Example: Hₐ: µ1 ≠ µ2 — The two group means are not equal.

What is the significance level (α) and what does it mean?

A threshold for determining statistical significance, often set at 0.05.
If p < α, reject H₀. If p ≥ α, fail to reject H₀.

What is a p-value?

The probability of obtaining results as extreme as those observed, assuming H₀ is true.

A low p-value (e.g., < 0.05) suggests the results are statistically significant.

What is Cohen’s d?

A measure of effect size showing the magnitude of a difference in standard deviation units.
Small = 0.2, Medium = 0.5, Large = 0.8+

Cyclist visibility example: What hypothesis test is appropriate?

Paired samples t-test — same participants measured with and without a reflective vest.

Cyclist visibility: What are the hypotheses?

H₀: µ1 (with vest) = µ2 (without vest) — no difference

Hₐ: µ1 < µ2 — reaction time is faster with the vest

Cyclist visibility: What is the significance level?

α = 0.05 — If p < 0.05, reject the null hypothesis.

What assumptions must be met for a paired samples t-test?

• Data is continuous

• Data is normally distributed

• Differences between paired values are normally distributed

Examples of paired samples t-tests

• Stress scores before and after semester

• Blood glucose levels before and after insulin

• Student scores on two assessments

What is a quasi-experimental design?

A design with no random allocation of participants.
Participants may self-select or be in pre-existing groups.
Sometimes lacks a control group.
Used when RCTs are impractical or unethical.

What is the purpose of quasi-experimental designs?

To study the link between an exposure and an outcome, often in real-world settings like health campaigns.

What is an Uncontrolled Before-After design?

Measures participants before and after an intervention with no control group.
Assumes changes are due to the intervention.

Disadvantages of Uncontrolled Before-After designs?

• No control over external factors

• Maturation effects

• Practice or testing effects

• Instrument/tool changes

What is a Controlled Before-After design?

Uses two groups: one receives the intervention, one does not.

Both groups are measured before and after.

Still has issues with external factors and maturation.

What is a Time Series Design?

Uses repeated measurements over time.

The intervention occurs during the series.

Changes in trend indicate intervention effects.

Advantages and limitations of Time Series Design?

Advantage: more measurements increase validity.

Limitations: still affected by external variables and testing effects.

What is internal validity?

The degree to which observed effects are due to the intervention, not other factors.

Threatened by chance, confounding, and bias.