RESS flashcards for summative
what is a health related state
Outcome or dependant variable
e.g death, disease
What is a determinant
A definite cause
Exposure
Independent variable - potential cause
e.g smoking, age``
Incidence
Probability that a disease occurs/is contracted in population within a period of time
Equation: the number of new cases in the period / the number at risk over the period
Prevalence
the number of cases of disease at a particular time point
Mortality
Proportion of people dying from the disease compared to the rest of the population
Equation: number dying in period from disease / number in population
Case fatality rate
Proportion of people who die from a specified disease among all individuals diagnosed with the disease over a certain period of time
Equation: number dying in period from disease / number with the disease in period
The scientific method
1 - Observe 2- Propose/modify a hypothesis 3 - Test the hypothesis 4 - Reject/do not reject 5 - If reject, modify hypothesis, If don't reject - Test again
Case series
Describes a sample of cases with the same disease, not doing any analysis just describing
Cross-sectional study
Studies a group of people at a single point in time
Cohort study
Examines disease development in groups of people over time
Case-control study
Examines history of groups of people with / without a disease
Ecological study (our projects)
Examines variations between geographical areas
3 types of health outcomes
- Record based
- Biological/clinical
- Clinician/patient-reported
Validity
- Measures accurately what it is meant to measure
- E.g. using BMI rather than weight as a measure of obesity
Reliability
- Give the same result on retesting
- E.g. Weight on a bathroom scale is reliable within around ½ a pound
Responsiveness
- Can detect real changes when they occur
- Death is unresponsive because once you’re dead you’re dead, it cant change.
- E.g. continuous QoL scale rather than categorical
Person-time
- Used as incidence denominator
- a way of determining how many people are at risk
Considering a common, short duration disease such as influenza, which of the following would you expect to see?
- Low incidence, low prevalence
- High incidence, low prevalence
- High incidence, High prevalence
- Low incidence, high prevalence
High incidence, low prevalence. Lots of people get it, but they’re cured quickly so fewer people have it at a single point in time.
Relative risk (or odds ratio)
risk in exposed group / risk in unexposed group
Interpretation of relative risk
- =1 means risk in exposed = risk in unexposed. No benefit or harm
- <1 means risk in exposed < risk in unexposed. Exposure is protective
- > 1 means risk in exposed > risk in unexposed. Exposure is harmful
Truncation:
E.g. anorexi* = anorexia, anorexic,
Adjacency searching:
E.g. eating adj2 disorder = eating disorder, eating related disorder
Wildcards:
- ? #
- E.g. behavio?ral = behavioural, behavioral (one character, or none)
The threeBoolean operators
AND - both concepts together
OR - both concepts individually
NOT - One concept where the other is not present
Likert scale
a scale used to represent people’s attitudes to a topic.
Continuous data
Numerical value, no limitation on values e.g. weight. Even if recorded in whole units.
Discrete data
Numerical value, limitation on values e.g. number of hospital appointments
Ordinal data
Ordered category e.g. Likert scale, stage of disease. You categorise things and count the number of things in that category.
Nominal data
Non-ordered category e.g. sex, blood group
Normally distributed data analysis
Mean and standard deviation
Skewed data analysis
Median and interquartile range (IQR)
How would you check on the distribution of a continuous variable?
Histogram
Histograms
The area of a bar in a histogram is proportional to the frequency (frequency histogram), so height is proportional to class width. Histograms are plotted for continuous variables.
Bar chart
is a measure of the frequency of each item of a categorical variable
Boxplot
displays the median, interquartile range and outlier values
Scatterplot
looks at how 2 variables are related to each other. So a great way to see if there is a linear relationship between the two. But not to check the distribution
Single categorical exposure & categorical outcome
- calculate risk or odds.
Use risk, risk ratio, odds, odds ratio etc
Single categorical exposure & continuous outcome
Use t-test (or non-parametric equivalent)
Single continuous exposure & continuous outcome
Use correlation (association)
Multiple exposures & continuous outcome
Use linear regression (effect size)
Multiple exposures & categorical outcome
Use logistic regression
What does the p-value measure
Probability of a result
- So p=0.05 means a 1 in 20 (5%) chance of an event happening
Random error
- Error due to random factors
- Measurement error
- Error from chance fluctuations in the profile of our sample
Systematic error
- Error due to non-random factors
- Measurement bias, confounding bias etc.
A distortion of an association between 2 variables due to a common shared cause is called
Confounding bias
A systematic difference between patients chosen vs. not chosen for a study is called:
Selection bias
Confounding bias
Distortion of association due to a common shared cause (confounder)
Accommodate through conditioning
Experimenter bias
Due to the behaviours and actions of the experimenter
Information bias
Due to systematic error in reporting, measurement or recording of information
Selection bias
Occurs through a systematic difference between those selected into a study sample and those not selected
Standard deviation
Measure of spread
Standard error
Measure of precision