MLS 3302 BIOSTATISTICS RESEARCH METHODS LABORATORY PRACTICES Unit
MLS 3302 BIOSTATISTICS, RESEARCH METHODS, & LABORATORY PRACTICES Unit 2 – Research Methods Section 2 – Study Design Matthew Nicholaou, Dr. PH, MT(ASCP)
Unit 2 – Overview Section 1 � The Research Plan Section 2 � Study Section 3 � Bias, Design Fallacy, and Confounding Section 4 � Literature Search & Journal Articles Section 5 � Data Presentation
Unit 2. 2 – Study Design After you have spent time formulating: � Research Question � Hypotheses � Specific Aims You must define and design how you will achieve those Aims to test your Hypothesis and answer the Research Question
Unit 2. 2 – Choosing the Design The optimal study design for a research question is a function of: � The hypothesis under investigation � The information available Consider � Who/What is be studied (sample) � What data are going to be collected (data collection) � How will the data be analyzed (analysis)
Unit 2. 2 – Choosing the Design The Study Design is planned out BEFORE the researcher proceeds � Sample (who/what and how many) � Data collection (what variables are you interested in / concerned about) � Analysis (what statistical tests will you perform to test your hypothesis)
Unit 2. 2 – Major Groups of Studies When planning research there is a ‘hierarchy’ of epidemiology study designs you can choose from Best � Experimental (e. g. randomized control trial) � Observational (e. g. cohort, case-control) � Descriptive (e. g. case series) Worst Each type of study is considered methodologically superior to the types below it
Unit 2. 2 – Descriptive Study Descriptive – make use of available data to examine possible associations (e. g. case series) Advantages: � Cheap & Easy � Good for generating hypotheses Disadvantages: � Can not test a hypothesis � Can not control possible extraneous/confounding variables (high bias)
Unit 2. 2 – Observational Study Observational – Researcher collects detailed information from an observational group(s), does not alter or control independent variables (e. g. cohort) Advantages: Relatively cheap & easy � Can make good estimates of associations (test hypothesis) � Disadvantages: Susceptible to bias (if well designed can control statistically, not experimentally) � Weak on proving causality (except cohort study) �
Unit 2. 2 – Experimental Study Experimental – researches can manipulate the independent variables or groups and observe the effects(e. g. randomized clinical trial) Advantages: Ability to show causal effects (causality) � Can test a hypothesis and control extraneous / confounding variables (reduced bias) � Disadvantages: Costly � Complexity �
Unit 2. 2 – Descriptive Studies Case Reports � Careful evaluation of a single case that may contain important information (e. g. a rare disease case, Rabies [21 US cases from 1996 – 2003]) Case Series � Data from a cluster of [series] of cases are examined (e. g. AIDS epidemic – a cluster of homosexual men with Kaposi’s Sarcoma and/or Pneumocystis was reported in 1984)
Unit 2. 2 – Observational Studies Three main types of observational studies (analytic studies) � Cohort (gold standard) � Case-control � Cross-sectional Exposures and disease status are measured to evaluate associations
Unit 2. 2 – Observational Studies EXPOSURE TIME DISEASE Central tenet of observational studies
Unit 2. 2 – Observational Studies Three ways to differentiate observational studies Temporal Nature � Conducted at a given point in time or conducted over an interval (Prospective = future, Retrospective = past) Characterization of Subjects � Define status individuals according to exposure or disease Measures of Association � Between exposure and disease
Unit 2. 2 – Cross-Sectional Study Measure the occurrence of disease in a population at a single point in time (prevalence) Exposure and disease status are essentially measured at the same time Because a temporal relationship between exposure & disease can not be defined, can not show causality (unless repeated over time!)
Unit 2. 2 – Case-Control Study A group of individuals with a disease (cases) are compared to a group without the disease (controls) Rates of exposure are compared between the groups
Unit 2. 2 – Case-Control Study NOT EXPOSED DISEASE “CASES” NOT EXPOSED NO DISEASE SELECT “CONTROLS”
Unit 2. 2 – Case-Control Study Case-control studies have several advantages: � Possible to study rare disease (selection on disease status) � More than one exposure can be evaluated (selection on disease) � Typical small sample size = low costs � Quick & efficient Still susceptible to many biases (e. g. selection)
Unit 2. 2 – Cohort Study Cohort – is a group of people who have similar characteristics who are followed over time (usually disease free but at risk) � Gold standard of epidemiological studies [Prospective Cohort Studies] These characteristics can be selected on: Exposure � Occupation � Genetic trait � Geographical location (Framingham Study) � Etc. �
Unit 2. 2 – Cohort Study SELECT EXPOSED DEVELOP DISEASE NOT EXPOSED DO NOT DEVELOP DISEASE
Unit 2. 2 – Cohort Study Advantages � Cohort studies, if unbiased, reflect the ‘real-life’ cause-effect sequence…. Causality! � Numerous data and samples are collected over time, creating a bank of information that can be retrospectively mined
Unit 2. 2 – Cohort Study Disadvantages � Expensive � Complex � Difficult to continue to collect data on individuals over time (loss to follow-up)
Unit 2. 2 – Cohort Study Example Multicenter AIDS Cohort Study A prospective cohort study designed to evaluate the risk factors and study the natural history of HIV infection 6, 000 homosexual men were enrolled in 1984 from four urban areas (Los Angeles, Chicago, Baltimore. Washington, Pittsburgh) Selection: men who have sex with men (MSM) over the age of 18
Unit 2. 2 - MACS Longest running and most extensive HIV study in the world, expanded to include: � AIDS Link to Intravenous Experience (ALIVE) � Women’s Interagency HIV Study (WIHS) The data collected allowed for researchers to retrospectively � Associate HIV Viral Load and Progression to AIDS � Identify specific risk associated with behaviors � Effect of therapy on HIV/AIDS survival � Etc. .
Unit 2. 2 – Experimental Studies Randomized Clinical Trial Healthy individuals are selected and randomly assigned to treatment groups Researchers control the exposure (usually a drug and placebo) Then follow the individuals over time for outcome or disease Most powerful experimental study
Unit 2. 2 – Randomized Clinical Trial Advantages: � Control over variables � Can show causality (exposure/treatment -> outcome) � Can be ‘blinded’ to help control biases Individuals are randomly assigned to groups and given treatment/placebo Physicians and participants don’t know what they are getting (double blind) Matching
Unit 2. 2 – Randomized Clinical Trial Disadvantages � Complex � Expensive, expensive, expensive!
Unit 2. 2 – Randomized Clinical Trial TREATMENT GROUP SAMPLE RANDOM ASSIGNMENT PRETEST TREAT POSTEST PLACEBO GROUP
UNIT 2. 2 – Study Design Temporal Nature Subject Selection Cross-Sectional Point in time May collect retrospective data Exposure + disease status measured same time Case-Control Point in time May collect retrospective data Diseased and No Disease Cohort Follow over time Exposed and Not Exposed Clinical Trial Follow over time Randomly assigned to exposure status (treatment)
UNIT 2. 2 – Study Design
- Slides: 30