designing epidemiological studies Flashcards
what is descriptive epidemiology?
describe problem at aggregated level
can be used to inform later analytic research
what is analytic epidemiology?
deploy and test hypotheses, often at a person-level through which association can be measured and causation inferred
what are the 5 types of descriptive epidemiological study?
case report
case series
cross-sectional
longitudinal
ecological
what is descriptive epidemiology about?
characterising disease in 1 or more of 3 epidemiological dimensions
- person
- place
- time
what are some examples of demographics that may be investigated in the ‘person’ epidemiological dimension of descriptive epidemiology?
age
gender
occupation
disease status
what are some examples of areas that may be investigated in the ‘place’ epidemiological dimension of descriptive epidemiology?
at a hospital
in a geographical area
among a certain community
what are some examples of areas that may be investigated in the ‘time’ epidemiological dimension of descriptive epidemiology?
a point in time
over a specified period of time
what factors can be considered exposures?
age, gender, occupation (‘person’ dimension)
living in a particular area (‘place’ dimension)
what outcomes are often focused on?
morbidity and mortality
how can exposure and outcomes be measured?
incidence
prevalence
(often point estimates with a confidence interval around them)
what is a statistic?
fixed value derived from a sample that estimates the value in the population
what is a parameter?
fixed, often unknown value, which describes an entire population
why are case reports important?
used to communicate new diseases, new presentations, new findings
can be structured as a bulletin or learning opportunity so-called Continuing Medical Education (CME) or in the UK, Continuing Professional Development (CPD)
what is a case series?
in new diseases (e.g. COVID-19) more than one case reported becomes a case series
what is a cross-sectional study?
describes prevalence of a condition across a population at a single point in time
takes into account both exposure and outcome
e.g. survey
what are the disadvantages of a cross-sectional study?
prevalence measured may be an exposure, outcome or both
lacks follow-up
- risk and temporal relationships (e.g. incidence rate) cannot be easily determined
- only cases of disease present at time of survey assessed (if someone dies of the disease they are not counted) - limits ability to measure disease’s true extent
might be unclear whether the exposure preceded the outcome
what is a longitudinal study?
describe prevalence or incidence of an exposure or outcome over time
may be made up of more than one cross-sectional analysis (aggregated data)
may follow same participants over time (person-level data)
(term can also be applied to any study - descriptive or analytic - that involves measurement at more than one time point)
what is an ecological study?
compare groups, rather than individuals
only analyse aggregate data (not linked to a specific person)
can be descriptive or analytic
what is ecological fallacy/aggregation bias?
assuming that associations between groups hold true between individuals too
why are ecological studies useful?
usually first step in exploring a research question - can generate hypotheses about disease aetiology
suitable when variability of exposure within each group is limited
what are the advantages of an ecological study?
use secondary data sources to complete - therefore relatively inexpensive and quick to complete
what are the disadvantages of an ecological study?
subject to ecological fallacy
relies on secondary data collected for different purposes - may not always be comparable between countries or time periods
might be unclear whether the exposure preceded the outcome
what is primary data?
collected by the researcher first-hand
what is the advantage of primary data?
collected for a pre-specified purpose (test hypotheses or answer the research question(s))
what are the disadvantages of primary data?
significant cost (financial and in terms of time)
collecting and cleaning data will often take a very large proportion of an overall research project’s duration and budget
what is secondary data?
data that have been collected for another purpose
potentially ‘recycled’ for a different purpose
what are the disadvantages of secondary data?
analyses may have to make a series of assumptions because data analysed weren’t intended for the new purpose
introduces critical limitations on how the findings of such a study are interpreted
what is the advantage of secondary data?
usually faster and cheaper to carry out
what is routinely collected data?
form mainstay of day-to-day demography and epidemiology in the field
large administrative datasets that allow us to understand populations and their health (e.g. census)
how can understanding of a local area be improved through routinely collected data?
looking at the number of patients registered at a GP or on the electoral register
(in areas with rapidly changing populations - e.g. London) first language of reception-class school children can often give good understanding of changing ethnic diversity
emergency department attendances, hospital admissions, outpatient attendances
prescribing data from billing data of local pharmacies
what is non-routinely collected data?
corollary to primary data
include surveys, other bespoke datasets
why is the use of non-routinely collected data limited outside research?
prohibitively expensive
time consuming to operate
what is data linkage?
involves joining two or more datasets together
in doing so, finds out more than was possible by analysis of either original dataset alone
e.g. link emergency department attendances with GP records - could look for common characteristics of frequent users of both (identify opportunities for intervention, take pressure off urgent care system)
what are the 2 reasons for which data linkage between primary and secondary care doesn’t exist on a meaningful scale?
technical issues - no technological platform or solution has been developed as yet
privacy concerns - ethical and legal constraints on holding, exchanging and analysing private health information (however, patients are often frustrated at having to provide the same information multiple times to multiple organisations that are both regarded to be part of the NHS)
what is an observational analytic study?
involve investigator observing populations or individuals
do not involve interference in or manipulation of exposure
e.g. case control study
what is a case control study?
compares individuals with a particular condition/disease (cases) to another population with the same general characteristics but without the condition (controls)
how is a case control study carried out?
information on past exposure to possible risk factors (both cases and controls)
frequency and intensity of exposure in cases is compared to controls
what must be done prior to selecting cases in a case control study?
clear eligibility criteria defined based on objectives of study (“case definition”)
- e.g. only people within a certain age bracket
cases should be representative of everyone with the disease under investigation although they are sourced from a variety of places (hospitals, clinics, community setting) (e.g. not just advanced cases that need surgery)
what 2 things must be considered when selecting controls for a case control study?
source of controls
- should come from the same study population as the cases
- should be representative of population at risk
assessment of exposure
- should be measurable with similar accuracy to exposure in cases
what are some limitations of a case control study?
recall bias - self reported recall of behaviour may not be comparable in cases and controls
how are the numbers of cases and controls determined in a case control study?
number of cases can be limited by rarity of disease
statistical confidence can be increased by having more than 1 control per case (studies usually allocate 2 or more controls per case)
what are the advantages of case control studies?
good for studying rare diseases - can identify all of the existing cases that have already built up over many years
relatively cheap
quick to obtain data - can assess exposure and outcome at the same time
what are the disadvantages of case control studies?
bias associated with exposure assessment - presence of a disease may affect how an individual reports past exposure
difficulty selecting good control group
limited to assessing one chosen outcome
no information about temporal relationship between exposure and disease
why can incidents not be calculated in a case control study?
case control study begins with disease status, then explores exposure
already know outcome, so cannot calculate incidents based on exposure
how is an odds ratio in a case control study calculated?
odds of exposure amongst cases (i.e. ratio of exposed cases to unexposed cases) over odds of exposure amongst controls (i.e. ratio of exposed controls to unexposed controls)
case control study estimates likelihood of having exposure in those who have the disease relative to those who do not
what are the features of a cohort study?
involves a group of people without disease who are observed over a period of time to see what happens to them
defining characteristic: track people forward in time from exposure to outcome
what is important when selecting a target population for a cohort study?
appropriate age group (e.g. don’t recruit teenagers to investigate Alzheimer’s)
exposure of interest may be rare, may need to target a specific population
should initially attempt to identify as many subjects as possible without invoking restrictions to make study findings generalisable
what are some examples of factors by which cohorts can be assembled in a cohort study?
geographic region
occupation
disease
risk group
birth cohort
why are multiple exposures often assessed in a cohort study?
need to control for other exposures during analysis phase
what is important when determining exposure in cohort studies?
exposure must be well defined (e.g. binary, all individuals exposed at different levels)
how are exposures assessed in a cohort study?
self report (e.g. questionnaire)
taking physical measurements
using existing records (e.g. medical records, census data)
how can outcomes be ascertained in a cohort study?
routine surveillance
registry data
death certificates
medical records
directly from patients
(method must be identical in both exposed and unexposed groups)