Epidemiological Study designs

Analytic epidemiology
Etiology 病因
The cause or origin of disease outcomes
Analytic studies分析研究
Evaluate the causal relationship因果關係 between:
:Exposures接觸/ risk factors風險因素/ protective factors保護因素/ causing agent 引起媒介
:Outcomes/ diseases/ effects
Look at disease determinants決定因素 or etiology
Cohort studies are one of the analytic studies


Types of Descriptive Studies
(Observational+Descriptive)
1.Case reports/ Case series 病例報告/病例系列
2.Correlational studies 相關性研究
3.Cross-sectional studies 橫斷面研究
Types of Analytic Epidemiology
(Observational+Etiological)
1.Cohort studies
2.Case-control studies
Experimental Epidemiology
(Etiological+Intervention)
1.Clinical trials
2.Community trials

Case reports/ Case series

->Most basic epidemiological studies
->Account for one-third of publications in medical journal
->Case report: Detailed clinical report by doctors on a single patient
->Case series: Case reports of several patients
->Applications: Discover new diseases, potential causes or side effects

Pros
Constantly monitor potential outbreak
Discover unexpected adverse events of therapies/supplements
Urge regulation/ modifications/ warnings
Cons
No control groups
Very small sample size
Low representativeness
Impossible for any quantitative analysis
Limited generalizability (low)

Correlational studies

Definition::
->examine the correlation between the average exposure in various populations with the overall frequency of diseases within the population
Just a 平均數
->The unit of analyses is a population or aggregate numbers of people 總人數 rather than an individual
->Uses: Search for high/low risk populations to generate hypothesis

 CVD risks over countries
 Correlation coefficient 相關係數
->Examine a linear relationship X vs Y 线性关系
->Max. +1; Min: -1 (r)
->Measure the direction and how the data fit the relationship
->Does not reflect the magnitude of a linear relationship



Pros
Inexpensive and fast
Existing data available and accessible
Useful for preliminary初步 research and hypothesis generation
Cons
Limited control of confounders 混繞因素
Limited generalizability
Ecological fallacy
Cannot link exposure with disease

Cross-sectional studies

3 steps
Sampling
Select a study sample that represents your target population
Determination of exposure and outcome
Simultaneously determine both their exposure status and disease status prevalence at a point of time
Compare the prevalence
Examine the difference in prevalence of disease between exposed and non-exposed groups


 Sampling
->Representation 代表性
->Make the sample as representative of the target population as possible
->Ensure validity
 Precision 準確度
->Take large enough sample
: Precise risk estimates and confidence interval
->Compromise between precision and resource constraint
: Sample size calculation
 Probability sampling 概率抽樣(scientifically preferable)
->Random sampling 隨機抽樣
->Systematic sampling 系統抽樣 (selected according to a random start point but with a fixed, periodic interval )
->Stratified sampling 分層抽樣 (分咗幾group,再係group到抽人)
->Cluster sampling 整群抽樣 (分咗幾group,再揀其中幾group嘅全部人)
->Multi-stage sampling 多階段抽樣(random selection)
 Non-probability sampling 非概率抽樣(convenience and access)
->Convenience sampling 簡單採樣 (最易接觸的對象做採樣)
->Quota sampling 配額抽樣 夠數就收手
->Snowball sampling 雪球取樣


Examples of exposures (can be more than 2 levels)
->Environmental
->Behavioural
->Biomedical
Socio-economic
->Geographic
->Health interventions/ treatments
->Policy
Examples of outcomes (can be more than 2 levels)
->Diseases/ biomarkers 身體特性
->Any health outcomes
Measures to compare the prevalence
->Relative risk 相對風險
: Prevalence in exposed group/ Prevalence in non-exposed group
: Compare outcomes in exposed versus unexposed 有接觸與冇接觸有病情況的分別(病)
->Odds ratio 對比值
: Odds of exposure among diseased/ odds of exposure among non-diseased
: Compare the odds of exposure in diseased versus non-diseases有病與冇病接觸風險(接觸物)
Strength of association
1: No association

1: Greater risk

<1: Lower risk


Cons
->No temporal sequence is shown
: Impossible to tell whether the exposure comes before the disease
: Cannot establish causal relationship 因果關係
->Misclassification
:Beware of remission stage of disease or exposure
->Rely on prevalence
:Only cost-effective for looking at chronic diseases
-。Snapshot
:Provide no clues to time trends or changes in disease prevalence

Cohort Study

click to edit

Features of Descriptive Studies
->Study of the occurrence and distribution of diseases
->Show temporal, demographic, geographic patterns, trends
->Make use of both prevalence and incidence data
->Rely on existing data ->Inexpensive and less time-consuming
->Used when little is known about the disease
->Temporal patterns 時間模式
->Demographic patterns人口模式
->Geographic pattern地理格局
->Trends, target features趨勢
Role of Descriptive Studies
->Applications: Quickly assess resource allocation based on disease distribution
->Formulate potential research hypothesis about diseases determinants: Provide clues for further investigations

click to edit

->Initial selection on the basis of exposure status
:First identify the exposure status of a cohort
:Then compare their disease outcomes

 Cohort隊列
A designated group of individuals that we follow longitudinally through time
 Cohort study群組研究
A follow-up study to measure the incidence of disease in a population
To test the causal relationship between exposures and outcomes


Steps
Select susceptible study subjects 易受影響研究對象
Measure their exposure levels接觸水平
Divide the selected eligible subjects by exposure levels
Follow them for a certain time
Measure the incidence of diseases/outcomes
Statistical data analysis
Result interpretation and recommendation

Prospective cohort study 前瞻性
Follow-up period occurs after the start of study period
Investigator starts the study at present with identification of a cohort
Then, exposure levels are measured
Subjects will be followed for the incidence of disease in the future
Take a relatively long time to complete the study
:Follow-up period can be as long as the study period
-> Exposure: Assessed at start of study & Outcome: followed into the future for outcome


Retrospective cohort study回顧群組研究
Investigator uses existing data collected in the past to identify a cohort
Then, exposure levels in the past cohort are measured
Disease status and records are determined at present
Take a relatively short time to complete the study
:Make use of past records
:No/shorter follow-up period
->Assessed at some point in the past & Outcome: outcome has already occurred


Measuring Exposures and outcomes
Carefully defined in advance of study
Ensure validity有效性, accuracy準確性 and precision精確性 of measurement
Measurements must be comparable可比 for all members of the cohort

Designs
 Similarities
-In both studies, exposure information is measured before the outcome event occurs
-Both compare the difference of disease incidence between exposed and non-exposed groups
 Differences
-Prospective cohort studies have a better control of the quality and quantity of data
-Prospective cohort studies is much more time-consuming
-Prospective cohort studies is very expensive