Statistics

CASE-CONTROL STUDIES

Ass.Prof. Dr Faris Al-Lami MB,ChB MSc PhD FFPH Dept. of Community Medicine/ College of Medicine/ Baghdad University

CASE-CONTROL STUDIES

It is a type of observational analytic studies. Study subjects are selected on the basis of whether they do (cases) or do not have (controls) the outcome under study. The groups then compared for the proportion of having a history of exposure or characteristic of interest.

CASE-CONTROL STUDIES

Disease
Total
Present
Absent
Exposure
Present
a
b
a+b
Absent
c
d
c+d
Total
a +c
b+d
N

Case-Control Studies – Timing Exposure

? ?
Exposed Unexposed
Disease
Yes (case) No (control)
Investigator

Strengths:

Suitable for diseases of long latency period Quick and inexpensive (as compared to other analytic studies) 3. Suitable for rare diseases 4. Can examine multiple etiologic factors for a single outcome. 5. Requires fewer subjects at entry 6. Few ethical problems

Limitations-

Usually cannot measure disease riskRelies on recall or records for information on past exposures (potential recall bias)May be difficult to determine that ‘cause’ preceded ‘effect’ (Temporal relationship)Unsuitable for rare exposuresSelection biasInformation Bias

Definition of Cases

The definition of the cases should depend on: 1. Homogenous disease entity 2. Strict diagnostic criteria

Depending on certainty of diagnosis, and amount of criteria one can classify the diagnosis into definite, probable, and possible



Sources of Selection of the cases:
1. Hospital-based case control studies The cases will be identified from the hospitals, or other health care facilities. These are common, relatively easy, and inexpensive.

Sources of Selection of the cases:

2. Population based case-control studies. It involves locating and obtaining data from all affected individuals or a random sample from a defined population.

Types of controls:

1. Hospital control: consist of patients at the same hospital with conditions other than the disease under study.

Hospital control: Advantages:

Easily identified insufficient number, with minimal cost, and effort. May come from the same catchments area More willing to know about previous exposures than healthy people (less recall bias). They are exposed to the same factors that make them select this particular hospital (less selection Bias) More willing to cooperate than healthy people, (less non-response).

1. Hospital control: Disadvantage

They are by definition ill and not healthy. Hospitalized controls differ from general population Control disease may be linked to exposure The patient in the control should not have a disease that is related to the same risk factors of the disease under study (CA-lung, MI, Smoking)

2. General population control

Used when the cases are chosen from the general population, and if the hospital control is not desirable or feasible.

2. General population control Difficulties:

More costly and time consuming Population lists are not always available Difficult to contact healthy people with busy work


2. General population control Difficulties:
The quality of information may differ from cases and control (more recall bias) Less motivated to participate (more non-response) Those who accept to participate may systematically differ from those who refused (volunteer Bias).

3. Special Group Control

Friends, relatives, neighbours of the cases . They have the advantage of the general population control, in that they are healthy, but they are more willing to cooperate, and offer a degree of control on certain confounding factors as ethnic background, socioeconomic status, or environment

No. of Controls

When the number of cases and controls is large and the cost of obtaining them is similar we choose one control for every case (1:1 ratio), but if the number of cases is small then the number of controls can be increased and this will increase the power of the study , but this should not exceed 4:1.

Ascertainment of the disease, and exposure status

Outcome status can be made from death certificates, case sheets, discharging cards. Exposure status can be obtained by interview, a questionnaire, or medical records. Data collector should not aware of: who the case is, and who the control is and about the hypothesis under study to minimize the possibility of observational bias.


Step 4 — Measure of Association Odds Ratio (OR) Odds ratio Good estimator of risk or rate ratio, especially for rare disease Odds of exposure among cases divided by odds of exposure among controls

Measure of Association between Exposure and Outcome in Case-Control – 2-by-2 Table Case

Exposed
a
Unexposed
c
V1
Control
b
d
V2
Odds Ratio = (a/c) / (b/d) = ad / bc

Example

Cigarette Smoking
Lung Cancer
Total
Cases
Control
Yes
70
30
100
No
30
70
100
Total
100
100
200



Exercise
ad 70 x 70 Odds Ratio= ------------=-----------------= 5.4 bc 30 x 30 = OR (1± z/√x2) 95% CI =5.4 1± 1.96/ √32 = 3.1- 9.6

Role of Bias in case control studies:

Selection Bias: occur when the inclusion of cases or controls into the study depends on the exposure of interest. 2. Observational Bias: error in obtaining, reporting, or recording of information by the investigator.

Role of Bias in case control studies:

3. Recall Bias: related to difference in the ways the cases and the controls will recall their exposure history. Cases are more likely to remember exposures than healthy controls. 4. Misclassification: refers to errors in the categorization of either the exposure or disease status.