Population-Level Study Design Flashcards
What does the practice of evidence-based medicine integrate?
- Individual clinical expertise 2. The best available external clinical evidence 3. Patient values and expectations
What are the steps taken during evidence-based medicine?
- ASK: Formulate a research question 2. ACCESS: Find and retrieve the ‘best evidence’ 3. APPRAISE: Consider the evidence for its validity and relevance 4. APPLY: Integrate the results into clinical practice 5. ASSESS: Evaluate the effectiveness
Quality of evidence pyramid
What is the highest quality of evidence that can be used?
Systematic reviews
What are the 4 types of evidence? How do they differ?
- Description
- Prediction
- Causal interference
- Qualitative
each require different methods & philosophies – none is ‘better’, they’re just different!
What is causal inference?
the process of drawing a conclusion about a causal connection based on the conditions of the occurrence of an effect:
*
What is the difference between inference of association and causal inference?
Causal inference analyses the response of the effect variable when the cause is changed
What is descriptive evidence?
Used to describe characteristics of a population or phenomenon being studied. It does not answer questions about how/when/why the characteristics occurred … cannot describe what caused a situation.
- What happened?
- Who was affected?
- People with X had Y
What is prediction evidence?
a prediction is made about the outcome of a future event based upon a pattern of evidence
- What will happen?
- Who will be affected?
- People with X are more likely to have Y?
What is qualitative evidence?
Qualitative evidence provides richer, deeper and broader information based on a few individuals or case examples. This type of evidence is valuable for describing how and why.
- What matters…?
- Why does it matter?
- How can we effectively change X…?
- Should we change X?
Causation of infectious disease example
Deterministic vs probabilistic?
A deterministic model does not include elements of randomness. Every time you run the model with the same initial conditions you will get the same results.
A probabilistic model includes elements of randomness. Every time you run the model, you are likely to get different results, even with the same initial conditions (e.g. smoking not always leading to early death)
Example of probabilistic
Why is studying individual people useless with probabilistic events? Why are groups studied instead?
People are very different. Even the same people can respond very differently at different times.
We therefore must work with groups, and use probability and statistics to describe, predict, and make causal inferences.
As the population is impossible or impratical to study, what is done instead?
so we usually study samples and try to generalise
What is clinical researd focused on?
Focused on understanding and helping individuals
What is population health research focused on?
Focused on understanding and helping populations
Examples of populations
1) . The (total) inhabitants of a given geographical area: world, continent, country, county etc
2) . The universe from which a sample is drawn: typically a group of people (UK adults, Scottish children with type 1 diabetes etc) OR can be any collection of units (hospitals in England, births in Europe, total GP consultations in a medical practice over October 2017 etc)
What is a census?
If your sample is the total population
What is a population-based study?
If your study population = a total population of a country or area
Types of studies
Case report vs case series?
A case report is a detailed report of the diagnosis, treatment, response to treatment, and follow-up after treatment of an individual patient.
A case series is group of case reports involving patients who were given similar treatment.
Why are case series beneficial?
Allows for description of natural history & prediction of prognosis:
–Frequency and range of symptoms
–Typical duration
–Typical survival
–Range of prognoses
–Predictors of prognosis
What are register-based studies?
A special type of case-series
- Disease registers count and collect information on people diagnosed with a particular disease
- These are commonly population-based, so you can estimate the disease occurrence per population
What is a cross-sectional study?
A study of a group of people at a single point in time (i.e. studying a cross section of the population in time)
May involve:
- Directly surveying or measuring a group of people
- Gathering routinely-collected data
Example of a cross sectional study
–Outcome: % of moustachioed medical leaders
–Population: Clinical department leaders (n=1018) at the top 50 US medical schools funded by the National Institutes of Health.
Proportion of female department leaders was 13.5% (137/1018)
The proportion of moustachioed department leaders was 18.7% (190/1018)
Descriptive studies usually also report their results broken down (‘stratified’) into subgroups. What are common characteristics for stratification?
What is the purpose of this?
- Age
- Sex
- Socio-economic position
- Self-reported ethnic group
- Marital status
- Occupation
This provides more relevant estimates for these groups and individuals
What are cohort studies?
Cohort studies examine groups of people over time
Example of cohort study
–Population: all adult inpatients with lab-confirmed COVID-19 from two hospitals in Wuhan, China
–Outcome: in-hospital death
- Study: 191 patients were observed until discharge or death (by Jan 31 2020)
- Events: 137 patients were discharged; 54 died in hospital during study period
- Findings: Increased odds of in-hospital death for older age, higher Sequential Organ Failure score, and d-dimer > 1μg/mL
How are hypotheses generated?
Descriptive studies often present multiple stratified results, to generate hypotheses
- Common variables include:
- Individual-level characteristics
–Age, sex, socio-economic position, ethnic group, marital status, occupation
•Area-level characteristics
–Countries, regions, latitude, urban vs rural areas, more vs less deprived areas,
•Temporal characteristics
–Secular trends (i.e. over time)
–Dynamic or seasonal trends (e.g. month, season, day of week)
What are case-control studies?
Examine groups of people over time retrospectively
Example: ‘UK case control study of smoking and risk of amyotrophic lateral sclerosis’
– Population: people diagnosed with ALS between 2008 and 2013 (vs. matched controls)
– Outcome: risk of ALS
- Survey: smoking status collected using environmental questionnaires
- Participants: 388 records with full smoking history
- Findings: Weak association between current smoking and risk of ALS; no increased risk with comprehensive smoking index (CSI) -> no evidence of dose-dependence with higher levels of lifetime smoking
What are the advantages of case-control studies?
–Good to study rare diseases
–Can investigate multiple exposures at a time
–Cheaper and less time consuming than a long-term cohort study
What are the disadvantages of case-control studies?
–Recall bias
–Need for suitable control group
–Can establish correlation but NOT causation
What are ecological studies?
A study that examines variations between geographical areas
- The ‘units of analysis’ are not individual people but areas
- Summary features of these areas are then compared to generate hypotheses
Are ecological studies good for testing hypotheses?
NO - but can be useful for generating hypothesis
Example of ecological study: The association between ultraviolet B irradiance, vitamin D status and incidence rates of type 1 diabetes in 51 regions worldwide
–Population: 51 regions across the world
–Outcome: Type 1 diabetes
–Exposure: Latitude > Ultraviolet irradiance & vitamin D status
