Observational Research Methods

Ivan J Perry Department of Epidemiology & Public Health University College Cork

Scenario
It is suggested that occupational exposure to volatile anaesthetic agents causes depression. You wish to test this hypothesis.

Learning objectives
At the end of this lecture you should be able to:

List and distinguish between the four major types of observational analytical study designs, including the strengths and limitations of each design Understand the difference between prevalence and incidence
Given a research question you should be able to select the appropriate study design

Observational

Descriptive

Analytical

Experimental

Type of Study
Ecological Cross-sectional

Alternative Name
Correlational Prevalence

Unit of Study
Populations Individuals

Case-control
Cohort

Case-reference
Follow-up

Individuals
Individuals

Ecological studies often provide the first tentative evidence of an association between a causal factor and disease The units of analysis are populations or groups of people rather than individuals

Pope et al., 1995

Advantages
Ecological studies are simple to conduct, often using pre-existing data collected for other purposes
Data can be used from populations with widely differing characteristics

Disadvantages
Suitable exposure and outcome data may not be available as the data is usually pre-existing

Usually not possible to adjust for potential confounders

Links between exposure and disease seen at the aggregate (population) level may be spurious and not seen at the individual level (ecological fallacy)

Objectives
To examine health problem or disease frequency
To examine association between the exposure and health problem or disease frequency Unit of analysis is individual Exposure and disease status of individual is assessed at the same time

Cross-sectional Studies measure prevalence

Prevalence is the number of cases in a defined population at a specified point in time, expressed as a proportion of the total population at risk for the condition. The prevalence rate for a disease is calculated as follows:
P = No. of people with disease at specified time No. of people in population at risk at specified time

60 50 40
% 30

20 10 0 50-54 55-59
Male

60-64

65-69

Female

Cork and Kerry Diabetes & Heart Disease Study, 1998

Advantages
A cross sectional study is short term, easy and economical to conduct

A cross sectional study generally starts with a reference population and is generalisable
Causal inference can be made from cross sectional data, provided it is known that the exposure preceded the effect or disease

Disadvantages
It is not possible to determine in some cases whether the exposure preceded the condition or disease Not suitable for investigation of rare diseases Generally requires large number of subjects The problem of selective survival may be an issue

Type of Study
Ecological Cross-sectional

Alternative Name
Correlational Prevalence

Unit of Study
Populations Individuals

Case-control
Cohort

Case-reference
Follow-up

Individuals
Individuals

Doll and Hills Data

Lung cancer patients Controls Total Smokers 647 622 1269 Non-smokers 2 27 29 Total 649 649 1298

Cases

Controls b

Exposed
Not Exposed

OR = a x d b x c
Note: the odds ratio of the Doll & Hill data shows clearly how much smoking increases the risk of lung cancer.

Using Doll & Hills data:

OR = 647 x 27 = 14.04 622 x 2

Advantages
Well suited to the study of rare disease Relatively quick and inexpensive to conduct

Requires comparatively few subjects

Existing records can be used in some case-control studies

No loss to follow-up.
Allows study of multiple potential causes of disease

May not be able to establish sequence of events

Unsuitable for the study of rare exposure

Disadvantages

Selection of an appropriate control group may be difficult

Relies on record or recall for information on past exposure (potential for recall bias)

Incidence rate
Cumulative Incidence

Relative Risk

Incidence rate (IR)is the number of new cases arising in a given period in a specified population. Incidence can be measured as follows:
IR = No. people who get disease (given time period) Sum of time each person remained under observation and at risk of becoming a case Usually expresses as number of cases per 1000 or per, 100,000 person years of follow-up.

Cumulative Incidence measures the denominator only at the beginning of the study. Cumulative incidence can be measured as follows: No. people who get disease (given time period) CI= No. disease free people in population at risk at the beginning of the period
Usually expresses as % risk during a defined period of follow-up, e.g. over 1 year or 5 years

Ratio Measure
Incidence of disease in exposed Incidence of disease in non-exposed
Note: in case-control studies relative risk is derived indirectly from the odds ratio.

Cigarette smoking and incidence rate of stroke in a cohort of 118,539 women

Smoking category No. stroke cases Person-years observation (>8) Stroke incidence rate (per 100,000 person-years) 17.7 27.9 49.6 30.2

Never smokes Ex-smoker Smoker Total

70 65 139 274

395,594 232,712 280,141 908,447

Colditz et al., 1988

Advantages
Suitable for the study of rare exposure Can assess multiple outcomes (effects) of single exposure Can demonstrate temporal relationship between exposure and disease Allows direct measurement of incidence of disease in the exposed and non-exposed population Changes in exposure over time can be studied Recall and selection bias are unlikely

Disadvantages
Not suitable for the study of a rare disease, unless a large sample size is obtained

If prospective, can be expensive and time consuming

If retrospective, requires the availability of existing record Validity of the result can be affected by loss of follow up