Clinical Research

Document Sample
Clinical Research Powered By Docstoc
					                      3-1




Clinical Research:

Essential Knowledge
    for Clinicians
                                               3-2

Goal
   To provide a foundation in order
   to critically review the literature

   v   Objectives
       – To define basic methodology, terms,
         and concepts
       – To review various study designs
                                           3-3

Objective #1


 v   To define basic methodology, terms,
     and concepts
                                3-4

Definitions

         v   Bias
         v   Selection Bias
         v   Measurement Bias
         v   Confounding Bias
                                                                     3-5

Two inferences are involved when drawing
conclusions from the findings of a study and
applying them to the universe outside...


TRUTH IN THE               TRUTH IN THE                 FINDINGS
  UNIVERSE                    STUDY                   IN THE STUDY
               Inference                  Inference
                   #2                         #1

               EXTERNAL                   INTERNAL
               VALIDITY                   VALIDITY
                                    3-6

Definitions (continued)


        v   Underlying Principles
            – p value
            – confidence interval
            – power
                                                                3-7

  Truth in the Population vs. Results
  in the Study Sample: 4 possibilities
                           Association       No association
                        between predictor   between predictor
                          and outcome         and outcome

          Reject null
Results   hypothesis      CORRECT           TYPE 1 ERROR
in the
study     Fail to
sample    reject null   TYPE II ERROR         CORRECT
          hypothesis
                            3-8

Definitions (continued)


          v   Measurement
              – precision
              – accuracy
                                                              3-9

 Difference Between Precision & Accuracy




Good precision   Poor precision   Good precision   Poor precision
Poor accuracy    Good accuracy    Good accuracy    Poor accuracy
                            3-10

Definitions (continued)


          v   Reliability
          v   Validity
                                                  3-11

Combinations - High & Low Reliability & Validity
                                 Validity
                          High              Low


                   High
     Reliability




                   Low
                                     3-12

Definitions (continued)


         v   Diagnostic Tests
             –   sensitivity
             –   specificity
             –   predictive values
                                                                   3-13

   Model for Sensitivity and Specificity
                               Population
                   Diseased                  Classified as diseased



Diseased,                                             Healthy,
classified as                                         classified as
healthy                                               diseased
(false negative)                                      (false positive)



                   Diseased, classified as diseased
                                                                         3-14

Definition of Sensitivity and Specificity

  sensitivity   number of sick people who are classified as sick

                          total number of sick people



  specificity   number of healthy people who are classified as healthy

                          total number of healthy people


 "Sensitivity" refers to the probability that a sick individual will
 be classified as sick.

 "Specificity" refers to the probability that a healthy individual
 will be classified as healthy.
                                           3-15

Objective #2

    v   To review various study designs
        –   Observational
             u Cross sectional

             u Cohort

             u Case control

        –   Experimental
             u Randomized clinical trial
                                                         3-16

    Cross Sectional Study

v   Purpose
    – Measures prevalence of a disease or risk factor
    – Describes variables and their distribution pattern
    – Example: National survey of drinking and liver disease
v   Steps
    – Select sample from population
    – Measure predictor and outcome variables
      simultaneously
                                        3-17

The Principle for Cohort Studies
       Present            Future

      Have
                          Diseased
      characteristic
                          (cases)
      (exposed)


      Do not have
                         Not diseased
      characteristic
                         (controls)
      (unexposed)
                                        3-18

The Principle for Case-Control Studies


       Have
                          Diseased
       characteristic
                          (cases)
       (exposed)


      Do not have
                         Not diseased
      characteristic
                         (controls)
      (unexposed)
                                                           3-19

Effect Measures
Cross Sectional, Cohort, and Case Control Studies
               # of people with disease at one point in time
Prevalence =
                  # of people at risk during that period

                  Incidence rate (exposed)
Relative Risk =
                  Incidence rate (unexposed)

                  Disease risk among exposed
Odds Ratio =
                  Disease risk among unexposed
                                              3-20

Interpreting Observational Studies


  v   Keep in mind:
      – Hill's criteria
      – Advantages and disadvantages of all
        three designs
                                                                                       3-21

Hill's Epidemiologic Criteria for Causal Association
 Criterion                                 Ask
 Strength              What is the relative risk?
 Consistency           Is there agreement among repeated observations in different
                       places, at different times, using different methodologies, by
                       different researchers, under different circumstances?
 Specificity           Is the outcome unique to the exposure?
 Temporality           Does exposure precede the outcome variable?
 Biological gradient   Is there evidence of a dose-response relationship?
 Plausibility          Does the causal relationship make biological sense?
 Coherence             Is the causal association compatible with present knowledge of
                       the disease?
 Experimentation       Does controlled manipulation of the exposure variable change
                       the outcome?
 Analogy               Does the causal relationship conform to a previously described
                       relationship?
                                                                                  3-22

   Major Observational Designs

Design             Advantages                           Disadvantages
COHORT   Establishes sequence of events               Often requires large sample sizes
         Avoids bias in measuring predictors          Not feasible for rare outcomes
         Avoids survival bias
         Can study several outcomes
         # of outcome events grows over time
         Yields incidence, relative risk, excess risk
                                                                                     3-23

    Major Observational Designs

Design                 Advantages                 Disadvantages
CROSS-    May study several outcomes           Does not establish sequence of events
SECTIONAL Control over selection of subjects   Potential bias in measuring predictors
          Control over measurements            Potential survivor bias
          Relatively short duration            Not feasible for rare conditions
          Good first step for cohort study     Does not yield incidence or true relative
          Yields prevalence, relative             risk
              prevalence
                                                                                   3-24

   Major Observational Designs

Design              Advantages                     Disadvantages
CASE-     Useful for studying rare conditions   Potential bias from sampling two
CONTROL   Short duration                           populations
          Relatively inexpensive                Does not establish sequence of events
          Relatively small                      Potential bias in measuring predictors
          Yields odds ratio (usually a good     Potential survivor bias
              approximation of relative risk)   Limited to one outcome variable
                                                Does not yield prevalence, incidence or
                                                   excess risk
                                            3-25

Treatment Efficacy Research

  v   Randomized Clinical Trials
      – The only true experimental design
      – Widely used to assess medication
        effectiveness
      – Controls for most bias
                                                                                   3-26

  Randomized Controlled Trial (RCT)
                                       The Present                   The Future
                                                 Treatment                    No
                                                                  Disease
                                                 #1                           disease
Population


                                                                                No
             Sample                               Placebo           Disease     disease


  Steps: 1. Select a sample from the population
             2.   Measure baseline variables
             3.   Randomize
             4.   Apply interventions (one should be a blinded placebo, if possible)
             3.   Follow-up the cohorts
             4.   Measure outcome variables (blindly, if possible)
                                                  3-27

Pros & Cons of Experimental Design

Advantages
v   Experiments can produce the strongest
    evidence for cause and effect
v   Experiments can be the only possible design
    for some research questions
v   Experiments can sometimes produce a faster
    and cheaper answer to the research question
    than observational studies
                                                  3-28

    Pros & Cons of Experimental Design

Disadvantages
v   Experiments are often costly in time and money
v   Many research questions are not suitable for
    experimental designs
v   Standardized interventions may be different from
    common practice (reducing generalizability)
v   Experiments tend to restrict the scope and narrow
    the study question
                      3-29




Supplemental Slides
                                                        3-30

Determining Sensitivity & Specificity
                            Disease status
  Test result        Present              Absent

  Positive        True-positive        False-positive
                  (TP)                 (FP)

  Negative        False-negative       True-negative
                  (FN)                 (TN)

                  TP + FN              FP + TN
Sensitivity = TP/(TP=FN)          Specificity = TN/(FP=TN)
                                                              3-31

Cross-Sectional Design
                              The Present
                                               Risk
                            Risk factor;       factor;
Population                  Disease            No
                                               disease

                          No risk          No risk
               Sample     factor;          factor;
                          Disease          No disease



       Steps     1. Select a sample from the population
                 2. Measure predictor and outcome variables
                                                                          3-32

   Prospective Cohort Design
                              The Present                   The Future

                                Risk factor              Disease      No
Population                      present                               disease


                                Risk factor                         No
              Sample            absent                  Disease
                                                                    disease


  Steps:     1.   Select a sample from the population
             2.   Measure predictor variables (risk factor present or absent)
             3.   Follow-up the cohort
             4.   Measure outcome variables (disease present or absent)
                                                                                3-33

  Case-Control Design
  The Past or Present                 The Present
   Risk           Risk                                             Population
                                                       Sample
   factor         factor                 Disease                   with disease
                                                       of cases
   present        absent                                             (cases)

Risk          Risk                                               Much larger
factor        factor                No disease      Sample        population
present       absent                                of controls without disease
                                                                   (controls)
Steps:
1. Select a sample from a population of people with the disease (cases)
2. Select a sample from a population at risk that is free of the disease (controls)
2. Measure predictor variables
                                                                                 3-34

    Statistics for Expressing Disease Frequency
    in Observational Studies

Type of Study     Statistic              Definition


Cross-Sectional   Prevalence   # of people who have the disease at one point in time

                                    # of people at risk at that point



Cohort            Incidence    # of new cases of disease over a period of time

                                    # of people at risk during that period

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:4
posted:4/29/2012
language:
pages:34