Analytical Studies

Document Sample
Analytical Studies Powered By Docstoc
					Community                                                         Dr.Salwa Tayel
                            Analytical Studies

Learning Objectives:
   Describe the characteristics of analytic epidemiologic study design.
   Calculate and interpret measures of association.
   Differentiate between attributable risk, relative risk and odds ratio.

Types of epidemiological Studies Designs :
  I- Observational Studies:
    A. Descriptive studies:
   Case report
   Case series
   Correlation studies                      Hypothesis generating
   Cross-sectional studies
    B. Analytic studies:
   Case control
   Cohort study                             Hypothesis testing
       (estimate and/calculate risks(

                                   Analytic Studies

  Basic Question in Analytic Epidemiology:
   Are exposure and disease linked?

                             1. Case-Control Studies

     It is an ―observational‖ design comparing exposures in disease cases vs. healthy
      controls from same population
            Exposure data collected retrospectively
            It is the most feasible design where disease outcomes are rare

                                        Page 1 of 11
   Basic steps in carrying case control study :
1. Identify the study group (cases)
   Define criteria for diagnosis and the criteria for inclusion and exclusion of cases
2. Identify controls: (needed for comparison)
                Controls must be free from the studied disease.
                Controls must be matched with cases for certain characteristics, which are
                   known to influence the outcome of the disease (confounding factors) e.g. age,
                   sex, social class,….
   Sources of controls:
   1. General population
   2. Hospital controls.
   3. Special control series (Friends - neighbours- fellow employees or peers - family members)
           Ratio control to cases 1 : 1 up to 4 : 1
   Why we need control group?
   To compare the exposure rate among both case and controls.

Summarize your data in 2x2 (association) table
    • Summarize frequencies of disease and exposure
    • Calculate association

                                         Page 2 of 11
Proportion of smoking in cases and controls :

   Analysis for risk measurement :
      Exposure rates only can be calculated from the case-control design, but not the
         incidence rates or relative risk.
      For cases, exposure rate will be a/a+c and
      For controls, exposure rate is b/b+d.

Compare percent exposed among cases and controls :

   Issues in Analysis :
        Ascertain whether there is a significant statistical association.
        Calculate chi-square test.

   Odds ratio (OR) :
          Odds ratio (OR) or cross-product ratio (ad/bc) which can estimate the risk (strength of
           the association)
          It is equally useful as RR (when the disease is rare) which can not be derived from
           these studies.

                                           Page 3 of 11
Calculate odds ratio from the following table

Interpretation of (OR) odds ratio:
    >1 means the exposure is a risk factor.
    = 1 means the exposure is not associated with the disease.
    < 1 means the exposure is protective

One of the classic studies was that of eight young women with cancer of the vulva—seen in
one city in a short period of time.
Thirty-two young women were chosen as controls, and both groups were given a series of
questions and what finally separated the two groups was:
a question whether their mother had been given DiEthylStilboestrol (DES) pills while the
young women were in utero 20 years earlier.
The odds ratio OR was very large indeed.

Advantages of case control studies :
  1. Easy to carry.
  2. Quick and cheap.
  3. Useful in early stages of the development of knowledge.
  4. Can be used in rare diseases.
  5. Allows the study of several risk factors.
  6. Useful in the study of disease with a long latency.
  7. Does not require large samples.
  8. Can prove hypothesis ( E & D are related).
  9. Can estimate risk (odds ratio).

Disadvantages of Case Control Study :
   1. Can not calculate rates.
   2. Not useful in rare exposure.
   3. Liable to bias:

                                      Page 4 of 11
Types of bias:
            Selection bias
            Recall bias ( poor memory - selective recall)
            Information bias: poor designed tools for collecting data on disease or

                                   2. Cohort Studies
      An ―observational‖ design where the investigator categorizes individuals on the basis
       of exposure thus comparing individuals with a known risk factor or exposure with
       others without the risk factor or exposure
      Looking for a difference in the risk (incidence) of a disease over time.

What is a cohort?
    Cohort is a group having a common characteristic
Example: A smoker’s cohort means all are smokers in that group
    A cohort, which is exposed to a suspected factor but not yet developed the disease, is
      observed and followed over time.
    Then, the incidence of the disease is measured directly.

It is the best observational design. Why?
      The investigator proceeds from ―E to D‖ i.e. from cause to effect so he will not face a
        chicken egg dilemma and the temporal (time) sequence between E and D can be
        clearly established.
      It uses a control group to accept or reject the hypothesis between E and D.

Types of cohort studies
 According to timeframe of cohort studies:
       Prospective Study starts in the present time and await outcomes in the future.
       It looks forward, follows a condition or disease into the future.

                                       Page 5 of 11
    An example:
          A prospective cohort study began in 1948.
          Included all the men aged 45 to 59
          Giving history of their health habits,
          Examined physically, and giving blood samples.
          The participants were then reviewed every two years until they died, and disease
          events were categorized according to their health behaviors in the histories.

    Retrospective Study The outcomes are measured in the present time on groups of people
     who can be categorized as having had certain characteristics in the past.
    ―to look back‖, looks back in time to study events that have already occurred.

                                          Page 6 of 11
    These studies are especially good for the small groups of people who have had a specific
     exposure (at work).
    Control group of similar people (but without the exposure) can be compared.
    The data at entry need to be good, and all the people included at the start need to be
     followed up till the end for these studies.

    An example of retrospective cohort studies:
              Workers who were exposed to asbestos in the building of US Navy ships
                 during World War II compared to other workers not so exposed.
              In the follow up, having cancer was much more common in those exposed to
                 asbestos than in the others, after controlling for the frequency of smoking in
                 the two groups.

    What are the main differences between the two types?

                     Prospective cohort                        Retrospective cohort
               Exposure occurred, but the disease            Both Exposure and disease
                (outcome) did not occur at the time the        occurred when the study is
                study begins.                                  initiated.
               It begins in the present and continue into    It begins in the past and continues
                the future.                                    to the present.
               Time consuming.                               Quick.
               Expensive.                                    Cheap.

 Issues in analysis:
 The basic analysis involves:
 Calculation of incidence rates among the exposed =(a/a+b)
 Calculation of incidence rates among the non-exposed =(c/c+d )

    Frame work of a cohort:

    Ascertain whether there is a significant statistical association between exposure and disease.
    Calculate chi-square or Z- test.

    Quantify risk:

                                   Incidence among exposed (Ie)
           Relative risk 
                                 Incidence among non exposed (Io)

                                           Page 7 of 11
      RR is measure of association between incidence of disease and certain exposure
      Relative risk (RR) answers the question:
      ―How many times a person who is exposed is at risk of developing disease compared
       to non-exposed?‖

Interpretation of Relative Risk (RR) :
    RR=1: No association between exposure and disease incidence rates are identical
       between groups.
    RR> 1: Positive association (increased risk) exposed group has higher incidence than
       non-exposed group
    RR< 1: Negative association (protective effect) non-exposed group has higher

Relative Risk :

                     Develop       Do not develop        Total       Incidence per
                      CHD              CHD                              1000/yr
   smokers             84              2916              3000             28.0
 Non-smokers           87              4913              5000             17.4

Incidence in smokers = 84/3000 = 28.0/1000
Incidence in non-smokers = 87/5000 = 17.4/1000
Relative risk = 28.0/17.4 = 1.61

                                 Difference Measures

Attributable Risk (AR):
    Number of cases among the exposed that could be eliminated if the exposure were
       removed = Incidence in exposed - Incidence in unexposed
         = a / (a+b) – c / (c+d)

                                     Page 8 of 11
Relative Risk vs. Attributable Risk :

Relative Risk:
            Measure of the strength of association, and indicator used to assess the
               possibility of a causal (etiological) relationship.

Attributable Risk:
            Measure of the potential for prevention of disease if the exposure could be

Attributable Risk (AR)
AR = lexposed – lnonexposed      =      “Risk Difference”

Attributable Risk (AR)
AR = (28.0 – 17.4) / 1000 = 10.6 / 1000

Among SMOKERS, 10.6 of the 28/1000 incident cases of CHD are attributed to the fact that
these people smoke …

Among SMOKERS, 10.6 of the 28/1000 incident cases of CHD that occur could be prevented
if smoking were eliminated.

                                         Page 9 of 11
Advantages of Cohort Studies :

   1.    Valuable in rare exposures (prospective).
   2.    Can examine multiple effects of a single exposure (prospective and retrospective).
   3.    You are sure that exposure happened before outcome (prospective and retrospective).
   4.    Can calculate incidence rates (prospective and retrospective).
   5.    Can quantify risks (Relative risk and Attributable risk) (prospective and retrospective).
   6.    Dose response ratio can be calculated (prospective and retrospective).
   7.    Minimal bias as regards ascertainment of exposure (prospective only).

Disadvantages of Cohort Studies :

   1.    They are inefficient for evaluation of rare disease (both types).
   2.    Take a long time (prospective only).
   3.    Expensive (prospective only).
   4.    Losses to follow up may affect validity of results (prospective only).
   5.    Involve a large sample (both types).
   6.    Retrospective cohort need complete and accurate records.

                                    Now let's Compare :

                       Aspect                       Cohort              Case control
        Logistic of the study
           1. Time                             Long               Quick
           2. Cost                             Expensive          Relatively cheap
           3. Sample size                      Larger             Small
           4. Ethical problem                  Present                        -

        Type of investigation
           1. Rare disease                              -                    Best
           2. Disease with long latency                 -                    Best
           3. Rare exposure                           Best                    -
           4. Multiple outcome                        Best                    -
           5. Several risk factors                     -                     Best

                                        Page 10 of 11
            Aspect                         Cohort            Case control
    1. Selection bias                        -              Always present
    2. Recall bias                           -                 Present
    3. Temporal relationship              Certain             Not certain
    4. Attrition problem                Problematic             Little
    5. Change in environment,           Problematic         Not a problem
  1. Incidence rate                  Can be calculated   Can not be calculated
  2. Relative risk                   Can be calculated    Approximation (OR)
  3. Attributable risk               Can be calculated   Can not be calculated
  4. Dose-response relationship      Can be determined   Can not be determined

                                  Page 11 of 11