Design and Analysis of Clinical Study

6. Case-control Study

Dr. Tuan V. Nguyen Garvan Institute of Medical Research Sydney, Australia

What is Case-Control ?

• Traditional view: compare - people who

get the disease

- people who

do not get the disease

• “Controls” a misnomer, derived from faulty analogy to controls in experiment • Modern conceptualization: controls are a “window” into the “study base”

Case – Control Study

• Grouping studied: "cases" vs. "Control" group(s).

• Measurements analyzed: past "exposures.“ • Case-selection usually clinic- or hospital-based.

• Controls may also be clinic- or hospital-based, or population-sampled.

• Controls may be unmatched, group-matched, or individually matched.

Yes No Yes No Case-Control Study

Cases Controls Population at risk

Aspirin use No Aspirin use Aspirin use No Aspirin use Case-Control Study

Cases Controls Population at risk

Steps in Designing Case-control Studies

•

Selection of cases

– Precise definition of ‘case’. – Inclusion / Exclusion criteria. – Are cases to be ‘incident’ or ‘prevalent’? – How are cases to be identified? How recruited?

Steps in Designing Case-control Studies

•

Selection of Controls

– Source ( hospital patients without disease; neighbourhood controls; random sample of population; sibs). – Inclusion / exclusion criteria. – Match to cases?

Steps in Designing Case-control Studies

•

Collection of information

– Identify risk factor of interest – Method of collection of information ( questionnaire; medical records; employment records) – Same procedure to be used for cases and controls – Interviewer should be unaware who is a case and who a control.

Two Methods of Selection

•

Select new cases

(i.e.incident) as they come up. Controls are selected from those in the same setting at the same time.

•

Select existing cases

(prevalent) from a defined population. From the same population a larger number of controls are identified.

The Incident type of case-control study is stronger because diagnosis of cases and ascertainment of exposure is being done by the researcher.

Risk factor

Results of a Case-Control Study

Aspirin Use No Aspirin Use Total Yes (cases) Disease No (controls) a b N1 c d N2 N1 and N2 are fixed numbers

Nested Case-Control Study

• Case-control studies within a cohort study • In ARIC (Atherosclerosis Risk in Communities) study, a cohort of 16 thousand men, all men provided serum samples at the outset which were saved.

• The cohort is observed for CHD.

• After 5 years we have 246 cases of CHD.

• We randomly choose 500 participants to be controls.

• We only measure Chlamydia antibody in the stored sera from these 246 + 500 subjects.

• We compare the cases (CHD) to the controls (no CHD) with regard to the presence of exposure (Chlamydia) which preceded the outcome

cases Non-cases cases Non-cases Matched Case-Control Studies Cases Controls Cases Controls All cases or random sample Random sample of non-cases All cases or random sample Matched controls

Effects of Beta-blocker on Hip Fracture

• Select a hip fracture case • Note the patient’s age, sex, weight, bone mineral density (BMD) • Select a sample of controls • Randomly selected

k

controls who have the same age, weight, and BMD as the case

Potential Biases

• A knowledge of the patient's disease status may influence: – Both the intensity and outcome of a search for exposure to the putative cause • A late look at those exposed (or affected) early will miss: – Fatal and other short episodes, plus mild or silent cases and cases in which evidence of exposure disappears with disease onset

COHORT VS. CASE-CONTROL STUDIES OF CHD VS. CHOLESTEREMIA AMONG MEN

UPPER QUARTILE SERUM CHOLESTEROL COHORT STUDY CASE-CONTROL STUDY YES NO CHD BY EXAM 6 CHD BY EXAM 6 YES NO TOTAL YES NO TOTAL 85 116 TOTAL 201 462 1511 1973 547 1627 2174 38 113 151 34 117 151 72 230 302 ODDS RATIO = 2.40

ODDS RATIO = 1.16

Sample Size Calculation

• • • • • •

Power :

probability of detecting a real effect (eg b

Alpha level

= 0.20) : probability of detecting a false effect (eg a = 0.05)

P 0

: probability of exposure in controls

P 1

: probability of exposure in case subjects

R

: odds ratio of exposures between cases and controls

m

: number of control subjects per case subject

Sample Size Calculation

• The estimated sample size is:

Website for Sample Size Calculation

http://www.sph.emory.edu/~cdckms/sample%20size%202%20grps% 20case%20control.html

Advantages and Disadvantages of Case-control Studies Advantages

• Relatively cheap compared to cohort studies • Relatively quick • Useful for study of rare diseases.

• No ethical problems • Useful for diseases with long latent period.

Disadvantages

• Estimate of disease incidence cannot be done • At times difficult to measure exposure accurately • Open to selection bias.

• Difficult to interpret.

Self-evaluation Questions

• Q2: Suppose that in a case-control study using incident cases of colon cancer you found that 80% of the cases were married. Does this demonstrate that being married increases the risk of developing cancer? • Q2: In the same case-control study above, assume that 90% of the control group group are married. If there are 200 cases and 200 controls, estimate the risk of colon cancer for single men. Constuct a 2x2 table and determine and interpret the exposure odds ratio.