Contact Information

Mombasa Road
Nairobi, Kenya

We Are Available 24/ 7. Call Now.

STUDY DESIGNS

OBSERVATIONAL

ASSIGNMENT # 2

A study was conducted among 3610 children who had been followed-up since birth. The purpose of the study was to determine if exposure to radiation in uterus puts them at risk of developing tumors later in life. Of the 2215 children who had been exposed to radiation, 59 developed tumors, while of the 1395 children who hadn’t been exposed, 13 developed tumors.

1. Which study design was this? Give reason for your choice.

2. Determine and compare the risks of developing tumors between the two groups of children.

3. How many times more is the risk of tumors if a child is exposed to radiation compared to a child not exposed?

4. How much of tumors could be prevented if there was no exposure to radiation?

5. Give three (3) strengths and two (2) limitations of the above study.

ASSIGNMENT # 2

READ ABOUT CASE-CONTROL & EXPERIMENTAL STUDY DESIGNS.

ASSIGNMENT # 2 &SOLUTIONS

A study was conducted among 3610 children who had been followed-up since birth. The purpose of the study was to determine if exposure to radiation in uterus puts them at risk of developing tumors later in life. Of the 2215 children who had been exposed to radiation, 59 developed tumors, while of the 1395 children who hadn’t been exposed, 13 developed tumors.

1. Which study design was this? Give reason for your choice. Cohort

2. Determine and compare the risks of developing tumors between the two groups of children. (26.64 per 1000 vs. 9.32 per 1000)

3. How many times more is the risk of tumors if a child is exposed to radiation compared to a child not exposed? (RR = 2.86)

4. How much of tumors could be prevented if there was no exposure to radiation? (Attributable rate (AR) = 17.3 per 1000)).

5. Give three (3) strengths and two (2) limitations of the above study.

CASE CONTROL STUDIES

DESIGN OF A CASE-CONTROL STUDY

• To examine the possible relation of an exposure to a certain disease or outcome;

Exposure Disease / outcome

DESIGN OF A CASE-CONTROL STUDY

• We first identify a group of individuals: – with that disease (called cases) and

–a group of people without that disease (called controls).

DESIGN OF A CASE-CONTROL STUDY

Cases Have condition or health outcome of interest.

Controls (non-cases) Do not have the health condition or health outcome.

Serves as the comparison group

Selection of Cases and Controls Selection of Cases

Sources include:

•Health facilities (hospitals, clinics). •General population.

Selection of Cases and Controls

Selection of ControlsSources of controls:

•Health facilities, •General population,

–Relatives, –Friends, –Spouses. – Siblings.

MATCHING

Matching is defined as the process of selecting the controls so that they are similar to the cases in certain characteristics, such as age, race, sex, socioeconomic status, and occupation.

– Purpose is to control for confounding.

• Associations on matched on variables/factors cannot be assessed.

DESIGN OF A CASE-CONTROL STUDY

With Disease (Cases)Without disease (Controls)
Exposure status??
Totala + c (=n)b + d (=m)

DESIGN OF CASE-CONTROL STUDIES

After identifying cases and controls:

Ask about history of contact with or exposure to suspected cause(s).

DESIGN OF CASE-CONTROL STUDIES

Sources of exposure data. – Records.

– Interviews (of study participants).

Design stage

Exposure statusCasesControls
Had been Exposed??
Had not been exposed??
Totala + cb + d

Design of case-control studies Then determine:

– What proportion of the cases were exposed and what proportion were not.

– What proportion of the controls were exposed and what proportion were not.

Analysis of data from case-control studies

Comparison:

– Cases vs. controls.

– Proportion exposed in the two groups.

Analysis of data from case-control studies

Exposure statusCasesControls
Had been Exposedab
Had not been exposedcd
Totala + cb + d

Analysis of data from case-control studies (Unmatched)

Exposure statusCasesControls
Had been Exposedab
Had not been exposedcd
Totala + cb + d
Proportion exposeda / (a + c)b / (b + d)

Analysis of data from case-control studies

Measures of AssociationStrength of Association:

Odds ratio.

Impact/Effect

Attributable risk percent (AR%). (can be estimated).

Effect measures in CASE-CONTROL studies

• The odds groups by giving:

ratio is a comparison dividing the odds for

of these two both groups,

The odds ratio is an effect measure that tells us how much larger the odds are for the exposed to be ill than for the nonexposed.

Exposure statusCasesControls
Had been Exposedab
Had not been exposedcd
Totala + cb + d
Proportion exposeda / (a + c)b / (b + d)
Odds of exposurea / cb / d
Odds ratio(Cross-product ratio)(a x d) / (b x c)


= ad bc

Interpretation of Odds ratio (OR):

If OR = 1, suggests no association.

> 1, suggests positive association. < 1, implies negative association

(suggests factor may be protective)

Notes:

Relative risk is estimated by the odds ratio.

• Attributable risk (AR) cannot be determined in case-control studies.

• Attributable rate percent (AR%) can be estimated by:

AR% = OR – 1 OR

Problem:

A case-control study was designed to determine the relationship between use of mobile phones and development of brain tumors. A total of 100 subjects with brain tumors (cases) and a similar number of 100 subjects without brain tumors (controls) were recruited into the study. Past history of use of mobile phones was sought from study participants. Fifty (50) of the cases and 10 of the

controls were found to have used mobile phones in the past. Carry out the analysis of data obtained from this study.

Example: Design of case-control study

First, selection of cases and controls
With Brain Tumor (Cases)Without Brain Tumor (Controls)
Exposure status(Past Mobile phone use)??
Total100100

Example:

Exposure statusWith brain tumor(Cases)Without brain tumor (Controls)
Had used mobile phones in the past5010
Had not used mobile phones in the past5090
Total100100
Proportion exposed to mobile phones50%10%

Example:

Use of mobile phones in the pastCasesControls
Yes5010
No5090
Total100100
Proportion exposed50%90%
Odds of exposure50/5010/90

Example:

Use of mobile phones in the pastCasesControls
Yes5010
No5090
Total100100
Proportion exposed50%10%
Odds of exposure50/5010/90
Odds ratio (OR)9.0

Example:

Use of mobile phones in the pastCasesControls
Yes5010
No5090
Total100100
Proportion exposed50%10%
Statistical test of significanceChi-square (χ²) test

Example: Interpretation:

• The odds of exposure to mobile phones is 9 times more in the case group than the control

group.

• This suggests a positive association between mobile phone use and development of brain tumor.

Estimation of Risk (effect/Impact)

Attributable Risk % (Attributable proportion):

• This measure cannot be directly determined in a case-control study but can be estimated by:

AR % = OR – 1 

OR

• Gives the proportion of disease in the exposed group that is due to that factor.

Notes:

Relative risk is estimated by the odds ratio.

• Attributable risk (AR) cannot be determined in case-control studies.

• Attributable rate percent (AR%) can be estimated by:

AR% = OR – 1 OR

Matched case-control study design

PairCaseControl
1+_
2_+
3__
4++

Paired Analysis Cases

Exposed Unexposed

Exposed Both (r) Mixed (s)

Controls

Unexposed Mixed (t) Neither (u)

r & u – Concordant pairs (agreement)

s & t — Discordant pairs (no agreement)

Paired Analysis

For one control

Case

Exposed Unexposed Exposed r                s

Controls

Unexposed t u McNemar chi2=(t+s)2/(t-s)

Odds ratio: s / t

Advantages of Case-control studies • Relatively inexpensive

• Reasonably rapid , i.e. less time required. • Suited to infrequent conditions

• Permit examining multiple risk factors or exposures.

Disadvantages of Case-Control Studies

– Case and control selection a problem. – Subject to bias (e.g. selection, recall)

— Incidence rates cannot be determined. -Temporal relationship is not clear.

Exposure Outcome -Multiple outcomes cannot be studied.

Principal differences between case-control studies and cohort studies

Case Control StudiesCohort Studies
PopulationStart with affected subjectsStart with healthy subjects
IncidenceNoYes
PrevalenceNoNo
AssociationOdds ratioRelative risk

A fourfold table Retrospective (Case-Control)

E X P O S U R E

present exposed

absent Not exposed

cases present

a

c

controls DISEASE absent

b

d

Mausner, 1985

Total Total

EXPERIMENTAL STUDIES

Experimental Studies…….

• As in a cohort study, individuals start off with exposure status (i.e. Intervention status).

–However, the distinguishing characteristic is that investigators themselves allocate or manipulate the exposure.

42

Both the exposure (intervention) and the outcome must be defined at the time of planning for the

study. And how they will be measured.

43

Experimental Studies

• The subjects are usually randomized to the study groups (or arms).

– The purpose of randomization is to ensure comparability of the groups with respect to baseline characteristics.

• Controlling for ‘confounding’.

44

Experimental studies

Intervention studies can generally be considered either as:

Therapeutic orPreventive.

45

Experimental studies

Therapeutic trials are conducted among patients with a particular disease to determine the ability of an agent or procedure to:

– diminish symptoms,

– prevent recurrence, or

– decrease risk of death from that disease.

46

Experimental studies (cont.)

• A preventive trial involves the evaluation of whether an agent or procedure reduces the risk of developing disease among those free from that condition.

• Thus, preventive trials can be conducted among health individuals at usual risk or those already recognized to be at high risk of developing a disease.

47

Experimental studies (cont.)

• While therapeutic trials are virtually always conducted among individuals, primary prevention measures can be studied among either individuals or entire populations (e.g. community trials).

–Examples of community trials are:

• Assessment of the effect on dental caries of introducing fluoride into the drinking water of some areas and not others, and

• The evaluation of the impact on the incidence of diarrhoeal disease on an improved water supply.

48

RANDOMIZATION outcome

Intervention

no outcome

Study population

outcome

Control

no outcome

baseline

future

time

Study begins here (baseline point

Experimental studies: Masking or Blinding

• It is preferable that neither the participant nor the investigator knows which treatment has been received until after the end of the trial.

– This is called a double-blind design and ensures against biases in the handling or assessment of the groups.

– A single-blind design is when the investigator but not the participant knows the treatment received.

50

Experimental studies: Ethical issues

Although scientifically a randomized controlled trial is the best way to demonstrate the benefit of a

treatment regime, its use raises important ethical issue.

• Is it justifiable to withhold a treatment from the control group that might be beneficial to them?

• Is randomization justified, or should participants be able to choose which group they are in?

• Is informed consent necessary?

51

Experimental studies: Compliance

• An intervention study requires the active participation and cooperation of the study subjects.

– After agreeing to participate, subjects in a trial may deviate from the protocol for a variety of reasons, including:

• developing side effects,

• forgetting to take their medication, or

• simply withdrawing their consent after randomization.

• Non-compliance can affect the validity of a trial.

52

Analysis of Data from Experimental Studies

• More or less as for Cohort Study Designs.

53

Experimental studies: Advantages

• Can examine multiple effects/outcomes of a single intervention.

• Can elucidate temporal relationship between exposure and disease/outcome.

• Allows direst measurement of incidence of disease/outcome.

54

Experimental studies: Disadvantages

• Can be extremely expensive.

• Validity seriously affected by losses to follow-up of study subjects.

• Non-compliance of the study participants.

• Ethical issues when dealing with human subjects. • Take more time to conduct.

55

Hierarchy of Epidemiologic Study Design

Tower & Spector, 2007 (www)

CHOICE OF STUDY DESIGN

Type of study to choose depends on:

• The research question/ objective of the study. • Time available for the study.

• Resources available for the study.

• Disease: Common/rare disease or production problem.

• Type of outcome of interest.

• Quality of data from various sources

Type of study to choose depends on:

⮚Time available for the study

Type of study to choose depends on:

⮚Resources available for the study

Type of study to choose depends on:

⮚Disease: Common/rare disease or production problem.

Type of study to choose depends on:

⮚Type of outcome of interest

Type of study to choose depends on:

⮚Quality of data from various sources

Problem:

• In 1975, there were 1,000 women who worked in a factory painting radium dials on watches. The incidence of bone cancer in these women up to 2005 was compared to that of 1,000 women who worked as telephone operators in 1975. Forty (40) of the radium dial painters and eight (8) of the telephone operators developed bone cancer between 1975 and 2005.

1. What study design is this? 2. What are the:

a) Strengths of this study design? b) Limitations of this study design?

3. How many times more is the risk of bone cancer for women who were exposed to radium dial painting than those women who were not exposed?

4. How much of bone cancer could be prevented if women were not exposed to radium dials?

64

END ON STUDY DESIGNS

65

Share: