case-control study
DESCRIPTION
Case-Control StudyTRANSCRIPT
CASE-CONTROL STUDY
ASSOC. PROF. DR. MOHD HASNI JAAFAR
ENVIRONMENTAL PHYSICIAN & TOXICOLOGIST
DEPARTMENT OF COMMUNITY HEALTH
UKMMC
Case-Control Study – intro
To assess association between a disease / condition and an exposure suspected of causing (or preventing) that disease.
Study on a disease / health condition of interest that has already developed.
Case-Control Study - intro
An observational study
–Persons (case) with the disease (or other outcome variable) of interest, and
–Suitable control (comparison, reference) group of persons without the disease.
Case-Control Study - intro
Examine
–The past exposure to a risk factor• Smoking.
• Substances in air, food, water.
RECALL
Case-control = Retrospective study
Study features
Assess only ONE disease & multiple risk factors.
Cannot demonstrate the true risk of developing the disease in individuals with a suspected risk factor.
Study features
Often appropriate in etiologic investigations of relatively rare diseases.–Smaller sample size compared
to cross sectional study & cohort study.
Study question
What happened in the past (risk factors) to give such different results?– Personal characteristics.
• Age, gender, ethnicity. • Genetic, immunological, biochemical
status.– Exposure characteristic.
• Environmental / occupational exposure.
Population-based Case-Control Study
Requires good coverage of the population being studied. –All cases or a sample of
them.
–Controls from population registers or electoral rolls.
Hospital-based Case-Control Study
Majority of case-control studies. Cases are identified in particular
hospital, and controls are selected from the same/similar hospital(s).
Suitable for studies on cancer, liver cirrhosis, SLE, CCF, ESRF, and etc.
Hospital-based Case-Control Study
All hospitals in same geographically defined area should be included.
However, not ideally for studies on chr. bronchitis, DM, HPT etc. (don’t go to hospital - sampling bias).
Selection of controls
Main difficulties.Controls should be as similar
as possible to cases, except they don’t have the disease. –Population or hospital based,
geographical area, age group, gender, behaviour (smoker).
Examples of Case-Control Study
Studies of leukaemia, lymphoma & thyroid cancer incident among Chernobyl workers.
Risk Factors for Maternal Mortality.Smoking and lung cancerRubella and congenital malformation.Antenatal X-ray exposure and neonatal
leukemia.Aspirin and Reye’s syndrome.
Challenges In Case-control Studies
To ensure that the particular cases and controls that are studied are representative.
To ensure unbiased sample / data / measurement / record from respondents.
Biases
Case-control studies are subject to various types of biases that may compromise their interpretation.
Defined as any systematic error in the design, conduct, or analysis of a study that results in a mistaken estimate of an exposure’s effect on the risk of disease.
Biases
Segregation of disease or exposure status influenced by knowledge of the exposure (classification bias).
Cases and controls are not comparable (selection bias).
Memory on past exposure and confounders (human: recall bias; record: recording bias).
Rates of participation and refusal may differ according to the disease or the exposure (response bias).
Selection Bias
Occur in condition like in those–With advanced or multiple
disease(s).–Referred for tertiary care.–Controls that are selected from
patients with other significant disease (as confounder).
Confounding Factors
Potentially subject to the effect of extraneous factors which may distort findings of study (confounding factors).– An extraneous variable that satisfies both
of two conditions: • it is a risk factor for the disease being studied,
and • it is associated with the exposure being
studied but is not a consequence of exposure.
Confounding Factors
LUNG CANCER
Smoking
Alcohol
Matching for case and control
Method used during selection of case and control that ensure similarity between groups.
To pairing individual controls with particular cases according to the matching factors.
To arranging that the distributions of the matching factors among all controls are similar to those found among the cases.
Matching
Cases and controls may be matched on such factors that can be measured equally for case and control candidates such as – age, – gender, – ethnic group,– education level,– socioeconomic status, and others.
However, the factors which are the subject of investigation (including confounders) must never be matched.
Reporting Bias
Retrospective assessment of exposure status poses a particular challenge in case-control studies.
Medical records – A useful source of exposure information, – But are limited to exposures that are
routinely and consistently recorded for all subjects, both cases and controls.
Reporting Bias
– Absence of record or statement of exposure cannot be interpreted as absence of or no exposure.
– Accuracy of medical record information may vary over time or across institutions and it is difficult, if not impossible, to ensure that uniform criteria apply to historical exposure information.
Recall Bias
Method of ascertaining exposure by interview. – Structured interviews provide opportunity for
collection of exposure information, more than from medical records.
– Inaccuracy in reporting of past exposures may arise from
• poor memory• psychosocial or interpersonal factors that may
lead a respondent to exaggerate or understate the exposure.
Information Bias
To reduce bias– Avoid very extreme age groups (memory).– Identification of a group of control subjects who
have a different, but similarly severe condition. – Training of data collectors– Masking for study hypothesis.– Masking case and control status.
• These goals may be hard to achieve in a clinic setting where the clinic staff conducting the interviews or reviewing medical records are likely to have participated in the study design and have knowledge of patients' medical histories.
Data Analysis
Measure association between exposure and occurrence of disease.
Odds are a ratio of events in both exposed and non-exposed groups.
Odds Ratio describes the odds of disease / adverse event among case group relative to control group.
The Odds Ratio is the ratio of the odds of disease for the experimental (case) group relative to the odds of disease in the control group.
Data Analysis
The odds ratio is (a/c)/(b/d), or ad/bc.
Data Analysis
Odds Ratio = (a/c) / (b/d) = (a*d) / (b*c) = 0.08
Data Analysis
A value significantly less than 1.0 indicates that presence of the exposure is associated with absence of disease (protective effect).
A higher value (significantly >1.0) indicates a positive (risk) association between presence of the exposure and the disease.
No association is demonstrated when the odds ratio is not significantly different from 1.0.
Data Analysis
OR is generally a good estimate of relative risk (RR). The terms OR & RR are in fact interchangeable when used in case-control studies (incidence).
95% CI (confidence intervals) on the estimated odds ratio should be calculated for significant level (not include 1).
A p value alone is not sufficiently informative.
Advantages Of Case-control studies
Permit the study of rare diseases.Permit the study of chronic diseases
with long latency between exposure and manifestation.
Can be launched and conducted over relatively short time periods (2-5 years).
Tend to require a smaller sample size than other designs.
Advantages Of Case-control studies
Relatively inexpensive as compared to cohort studies.
Can study multiple potential (risk) causes of a disease.
Very valuable for identifying risk factors and suggesting hypotheses for more rigorous testing.
Disadvantages Of Case-control studies
Information on exposure and past history is primarily based on interview (recall bias).
Validation of information on exposure is difficult, or incomplete, or even impossible (reporting & classification bias).
Generally incomplete control of extraneous variables (confounders).
Choice of appropriate control group may be difficult (selection bias).
Disadvantages Of Case-control studies
By definition, concerned with one disease only.
Risk of disease cannot be estimated directly.
Not efficient for the study of rare exposures.