Docstoc

Diagnostic Tests

Document Sample
Diagnostic Tests Powered By Docstoc
					Basic Research Designs
      An Overview

  Department of Emergency Medicine
  University of Pennsylvania Health System




 Frances S. Shofer, PhD
      "General Plan”

 Settingup and testing hypotheses
 Design directs how project conducted
     Misconception I

For every research project there is
one single "best design" to answer
the question.
 Compromises

Rigorous scientific integrity
            vs
    Limited resources
      Clinical reality
                  Investigation Sequence
            Case                    Case                   Descriptive
           Reports                  Series                Epidemiology




      Clinical                       Analytic           Animal           Laboratory
       Trials                      Epidemiology       Experiments         Studies


Prevent-         Thera-
  ive            peutic
                                       Case-       Cross-
                          Cohort
                                      Control     sectional
      Study Design Sequence
        Descriptive                              Analytic
   Case                       Case              Observational
  Reports                    Series               Studies

             Hypothesis formation
                                                      Hypothesis
                                                        testing
                SCU                 SCU   SCA


                         New standard           ExperimentalT
Evaluation
                         of health care             rials


                                                    Further
                                                  hypothesis
                                                    testing
     Uncontrolled (Descriptive)


 Case      Case      Descriptive
Reports    Series   Epidemiology
     Case Reports

 Detailed presentations of a single case
  or a handful of cases
 Reports a new finding or is uniquely
  educational
         Types of Case Reports

   Previously undescribed syndrome or disease
   Unexpected association between 2 or more
    diseases or manifestations may represent
    unsuspected causal relationship
   Unexpected evolution suggesting surprising
    new therapeutic effect or adverse drug effect
         Case Report - Example
   Age 15.7 yrs
   Hx of worsening pharyngitis, vaginitis
   Vomiting and watery diarrhea
   Temp 40.9oC
   Injected conjunctiva
   Diffuse erythroderma
   Purulent vaginal discharge

   Throat and vaginal culture staph. aureus +
       Case Reports
   Advantages
    – Surveillance for rare clinical events
    – Rich source of hypothesis about
      disease frequency, risk, prognosis,
      treatment
    – Useful in elucidating mechanisms of
      disease and treatment
   Disadvantages
    – Susceptible to bias
    – Represent isolated events
         Case Series

   Prevalence survey of a group of
    individuals with a particular disease or
    characteristic, performed at a single
    point in time
      Case Series - Characteristics

 Newly recognized condition, unique, or
  highly educational
 Useful in bringing attention to a new
  area of concern for further investigation
 May be only choice of study design for
  uncommon disorders
                 Example: Case Series
    The First 1,000 Cases of Acquired Immune
    Deficiency Syndrome in the United States*

Complication                                       %       Risk Group                        %
Pneumocystis carinii (PCP)                      49.7        Homo- or bisexual men 72.7
Kaposi's sarcoma (KS)                           28.4        IV drug users                   15.5
PCP and KS                                        8.3       Haitian natives                 5.0
Other opportunistic infections 13.6                         Hemophiliacs                    0.7




*ada pted from J affe HW, Bregm an DJ, Se lik RM : J Infec t Dis 1 48: 33 9-3 45 ,, 19 83
*ada pted from J affe HW, Bregm an DJ, Se lik RM : J Infec t Dis 1 48: 33 9-3 45 19 83
         Case Series
   Advantages
    – Delineates the clinical picture of a rare
      disease
    – Characterizes disease in terms of averages
      and variability of aspects such as
      symptomatology
   Disadvantages
    – Cannot study cause and effect relationships
    – Cannot assess disease frequency
      Descriptive Studies

 Describes patterns of disease
  occurrence in populations
 Usually uses routinely collected data
 Along with clinical observations,
  laboratory studies, and other
  information, are used to generate
  hypotheses
                                 Annual Mortality from Asthma
                             3
                             3
Death rate/100,000 persons



                                      England & Wales


                             2
                             2
                                      Sweden


                                      USA

                             1
                             1




                             0
                             0
                                 59   60    61
                                            61   62
                                                 62      63
                                                         63    64   65
                                                                    65   66
                                                                         66   67
                                                                              67

                                                        Year
       Survey/Questionnaires

 Queries research questions directly and
  collates answers
 Not a simple or easy type of design to
  do correctly
 Tremendous potential for bias
      Improving the Quality

 KISS
 Maximize the response rate
   – <40% considered suspicious
   – >80% acceptable
 Use existing validated-measurement
  instruments whenever possible
 Pilot the questionnaire
          Survey: Example

                     The "Hite Report"
                100,000 questionnaires
                      distributed


                         4,500 returned


                                 Bias?

Smith TW: AIDS, Sexual Behavior, and Intravenous Drug Use.
Washington DC, National Academy Press, 1989
Control Group   OUT oF Control Group
                  Controlled Studies
                                  Analytic         Animal          Laboratory
                               Epidemiology      Experiments        Studies




           Experimental                          Observational



      Clinical      Before &                       Case-        Cross-
                                        Cohort
        trial        After                        Control      sectional


Prevent-         Thera-
  ive            peutic
     Observational Studies

 Implies no intervention by the
  investigator
 An analysis of spontaneously occurring
  events
 Group assignments are generally non-
  random in nature
 Considered an intermediate step in
  testing causal hypotheses
     Cross-sectional Studies

 Records observations in a selected
  group, at a single point in time
 Calculate prevalence rates
 Collect preliminary data quickly
 Relatively inexpensive to run
        Cross-sectional Design
Begin          Measure/Classify &
                   Compare


                free of       risk/factor +
               disease/
               outcome        risk/factor -
  Study
population
                 have         risk/factor +
               disease/
               outcome        risk/factor -


               Present
  Domestic Abuse and 911 Calls


       Hypothesis
Female victims of domestic
abuse are more likely to use
the 911 system than do non
victims.
             Domestic Abuse and 911 Calls

                    Measure/Classify &
 Begin                  Compare


                   Domestic        called 911
                     abuse                        77%
  Women             victims       no 911 calls
presenting
to HUP ED
                      Non          called 911
                    victims                       47%
                                   no 911 calls

                    Present
      Disadvantages
 Association does not prove causality
 Temporal relationships not clear




              and                  dilemma
      Cohort Study

 A study 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.
             Cohort Design
                           Measure &      Measure Outcome/
 Begin                      Classify          Compare


                                           disease/outcome +
                            Risk/factor
                                 +
  Study                                    disease/outcome -
                 free of
population
                disease
                                           disease/outcome +
                            Risk/factor
                                 -
               have                        disease/outcome -
             outcome
             already
             (exclude)
             (exclude)

                 Present                     Future
    Cohort Example

          Hypothesis
Risk of occupational exposure to
tuberculosis is likely to be higher
in EDs than inpatient settings
               TB Exposure in the ED

                          Measure &   Measure Outcome/
  Begin                    Classify       Compare


                                             TR+
                             ED
                          employees                 12%
                                             TR-
 Hospital         TR-
employees
                            Other            TR+
                           hospital                  2%
                          employees          TR-
            Already TR+
             (exclude)


                           93-94            95
     Disadvantages

 Prolonged follow-up
 Not a good design when outcome rare
 Inefficient
 Costly
     Case-Control Studies

 Used to test possible causes of a
  disorder
 Also called case-comparison, case-
  referent, or trohoc
 Useful for studying rare diseases
 Susceptible to numerous types of bias
  Case-Control Design

Classify & Compare       Begin


    risk/factor +         CASES
                        people with
    risk/factor -    disease/outcome



    risk/factor +       CONTROLS
                      people without
    risk/factor -    disease/outcome


       Past              Present
   Hypothesis

Aviators who had crashed their
  airplanes might have been
 accident prone as youngsters
     Example: Case Control Study

                          Past            Present

                                           CASES
                                       Flyers who had
                                        crashed their
                                           planes
                      Interviewed on
                         childhood
                         accidents
  Flyers with bad                        CONTROLS
safety records more                    Wreck-less pilots
  accident-prone
         Experimental Studies
   Involves investigator manipulation of events
   Akin to laboratory experiments except living
    populations are the subjects
   Generally involves random assignment to
    groups
   Include clinical trials (preventive and
    therapeutic)
   Considered the ultimate step in testing
    causal hypotheses
             Experimental Design
                                      Measure Outcome/
  Begin                Allocate           Compare


                                      disease/outcome +
                       Experiment/
                       intervention
                                      disease/outcome -
  Study
population
                                      disease/outcome +
                         Control
                                      disease/outcome -



             Present                     Future
         Clinical Trials

   Non-randomized
    – Treatment or group assignment of patients
      is determined subjectively by the
      investigator or patient
   Randomized
    – Assignment of patients to treatment or
      group is done in such a way that the
      probability of assignment to each group is
      equal for all patients
    The "Gold Standard" of Studies

 Prospective randomized, blinded,
  controlled clinical trial
 Most expensive
 Not most appropriate to answer certain
  types of questions
        Types of Blinding/Masking

   Not blinded
   Single blinded
   Double blinded
      ED Intervention to Improve
      Asthma Follow-up Rates

 To improve access to timely primary
  care follow-up after an asthma
  exacerbation requiring ED evaluation
 A simple, inexpensive, three part
  intervention increases the likelihood of
  primary care follow-up in discharged ED
  patients with an asthma exacerbation
                 Asthma Intervention Study
                                               Measure
Identify      Randomize                    Outcome/Compare


                                                        PCP F/U
                       Fanny pack
                                      48 hr call
                       intervention
  Eligible                                            No PCP F/U
  Asthma
  patients                                              PCP F/U
                        Control
                       usual care
                                                      No PCP F/U

Data Collection tool   Pt interview                4 wk PCP F/U call
      Disadvantages

   Focused (narrow research question)
   Expensive $$$$
   Not always possible to randomly assign
    patients to study groups
   Impractical for rare clinical events
      Before and After Designs

 Takes advantage of a change in therapy
  or environment
 Investigator rarely responsible for
  change/intervention
 Sometimes only study design option
  available
Before and After Design

  x
  x x x
      x
          x
          x x x
              x

 Before    After
Before and After Design



  x x x          x x

             x


  Before   After
     Some Examples

 Motorcycle helmet laws
 Seat-belt laws
 Clinical pathways
     A Hypothetical Example

   The Clinical Observation


Patients who suffer from poor visual
acuity and wear glasses seem to
have diets deficient in carrots.
    A Hypothetical Example

          The Science

Carotenes are essential in the
formation of rhodopsin, a retinal
pigment associated with good
night vision
A Hypothetical Example

   The Hypothesis

An association exists
between good vision
and carrot consumption.
Case Control Design
Classify & Compare      Begin


 eat many carrots     CASES
                     patient with
  eat few carrots    poor acuity

 eat many carrots    CONTROLS
                     patients with
  eat few carrots     good vision

      Past              Present
       Cross-sectional Design
Begin          Measure/Classify &
                   Compare


                             multi-carrot diet
               Good
               vision
                             oligo-carrot diet
Patients in
 practice
                             multi-carrot diet
                Poor
               vision
                             oligo-carrot diet


              Present
              Cohort Design
  Begin                    Measure &   Measure Outcome/
                            Classify       Compare


                            Eat many    good visual acuity
                             carrots
                              daily     poor visual acuity
Patients in
                 free of
 practice
                disease
                             Eat few    good visual acuity
                             carrots
                              daily     poor visual acuity
            have poor
          vision already
             (exclude)


                 Present                  Future
              Experimental Design
    Begin                  Allocate      Measure Outcome/
                                             Compare


                                         good visual acuity
                           High carrot
                              diet
                                         poor visual acuity
Patients in
                free of
 practice
               disease
                                         good visual acuity
                           Low carrot
                              diet
            have poor
                                         poor visual acuity
          vision already
             (exclude)


                Present                     Future

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:9
posted:5/27/2012
language:English
pages:56