Statistics

COHORT STUDIES

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

An early cohort in search of favourable outcomes

COHORT STUDIES
These are observational analytic studies. The group(s) of study individuals are defined on the basis of presence or absence of exposure to a suspected risk factor of a disease. The two groups then will be followed for a period of time to assess the occurrence of the disease. All participants should be free from the disease at the start of exposure.

COHORT STUDIES

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

COHORT STUDIES

Strengths:
1. Establish the temporal relationship between exposure and disease 2. Useful for rare exposures 3. Examine multiple outcomes of single exposure 4. Allows direct measurement of incidence of the disease among exposed and unexposed groups.

Limitations:

Expensive Time consuming Problem with follow up

Analysis

Incidence among exposed Relative Risk (RR)=---------------------------------------- Incidence among non-exposed a / a+b RR =------------- c/ c+d

Analysis

95% Confidence Interval around RR in cohort study: ln RR ± 1.96 √ [ (b/a) /a+b] + [ (d/c) /c +d]

Types of Cohort Studies:

1. Retrospective Cohort Study 2. Prospective Cohort Study 3. Ambidirectional Cohort Study


1. Retrospective Cohort Studies
a- Quick, and cheap b- Useful for uncommon exposure and for diseases of long latency period c- Depends on available exposure and disease data BUT these data are usually incomplete and lack information on confounder.

2. Prospective Cohort study

The exposure may or may not occur but the disease is definitely not occurred and the cohort should be followed in time to assess the incidence rate of the disease in the two groups. a- Very reliable b- Direct assessment of exposure c- Available information on confounders

3. Ambidirectional cohort study

Useful for exposures that have short term and long term effects. Data are collected retrospectively and prospectively on the same cohort.

Selection of Exposed Population: Depends on:

1. Nature of research question 2. Frequency of exposure under study 3. The need to have complete and accurate information about the exposure and the follow up

Rare Exposure

As radiation, Specially Exposed individuals are needed as individuals in certain occupation, have certain medical treatment, exposed to specific environment or specific events.

Common Exposure

As smoking, Specially Chosen population not for exposure status, but for the ability to obtain complete information and follow up.

Sources of Exposure Data:

Pre-existing records Information from the study subjects Direct physical examination or an investigation Direct measurement of the environment

Selection of Comparison Group

They should be similar to the study group in all the factors related to the disease EXCEPT the factor under study.

Selection of Comparison Group

Internal Comparison Groups who are not exposed. It is used when a single cohort entered the study and the member are classified into exposure categories.

Selection of Comparison Group

External Comparison Group It is used when specially exposed cohort enters the study

Selection of Comparison Group

Multiple Comparison Groups



Sources of Outcome Data:
Goal: Obtain complete, comparable and reliable information on subsequent disease occurrence in both study groups. Source will depend on: Available resources Nature of the outcome Procedures for identifying outcomes should be applied equally to both groups.

Sources of Outcome Data

Fatal Outcome: Death certificate which is reliable if the goal is total mortality, but it is not reliable if the goal is the cause specific mortality

Sources of Outcome Data

Non Fatal Outcome: Medical records (Physician’s records, discharging records,..)Direct from the participants (should be supported by other sources)Data from periodic medical examination (expensive and time consuming) The method of assessment of outcome should be the same for both groups

Potential Biases in Cohort Studies

Information Bias If the quality and extent of information obtained is different for exposed persons than for non exposed persons

Potential Biases in Cohort Studies

Misclassification Error in categorization of exposed and unexposed individuals, and errors in ascertainment of the outcome of interest. The effect will depend on: Accuracy and completeness of information on which the classification was made, and on the degree to which this error was made differently among the study groups.

Potential Biases in Cohort Studies

Non-Response Bias Non-response can introduce major biases that can complicate the interpretation. Nurse health study: out of 172,000 , only 122,000 accepted (71%) Framingham Study : 6507, only 4469 (69%) accepted) Non response can affect the generalizability of the study but not the validity, unless the non response is related to the exposure under study


Potential Biases in Cohort Studies
Loss to follow up Bias If this ratio is around 30-40% this will affect the validity of the study. This is a particular problem if the loss to follow up is different between the group who developed the disease from the other group To estimate the true relationship we can calculate the relation using the most extreme situations, and have the interval.

Advantages and disadvantages of different observational study designs:

Probability of:
ECOLOGICAL
CROSS SECTIONAL
CASE CONTROL
COHORT
Selection Bias
NA
Medium
High
Low
Recall Bias
NA
High
High
Low
Loss to follow up
NA
NA
Low
High
Confounding
High
Medium
Medium
Low
Time Required
Low
Medium
Medium
High
Cost
Low
Medium
Medium
High