Epidemiology: Overview of Key Concepts and Study Design. Polly Marchbanks

Similar documents
INTRODUCTION TO EPIDEMIOLOGICAL STUDY DESIGNS PHUNLERD PIYARAJ, MD., MHS., PHD.

3. Factors such as race, age, sex, and a person s physiological state are all considered determinants of disease. a. True

Observational Study Designs. Review. Today. Measures of disease occurrence. Cohort Studies

CONCEPTUALIZING A RESEARCH DESIGN

Chapter 2. Epidemiological and Toxicological Studies

Part 1: Epidemiological terminology. Part 2: Epidemiological concepts. Participant s Names:

Epidemiologic Methods and Counting Infections: The Basics of Surveillance

Main objective of Epidemiology. Statistical Inference. Statistical Inference: Example. Statistical Inference: Example

Trial Designs. Professor Peter Cameron

Bias. A systematic error (caused by the investigator or the subjects) that causes an incorrect (overor under-) estimate of an association.

Epidemiological study design. Paul Pharoah Department of Public Health and Primary Care

Overview of Study Designs

Incorporating Clinical Information into the Label

Epidemiologic Study Designs. (RCTs)

Critical Appraisal of Research Protocol

Research Methods: Operational Studies in Tuberculosis. Introduction to Epidemiology. Epidemiology. Epidemiology. The Public Health Cycle

Answer keys for Assignment 4: Measures of disease frequency

Penelitian IKM/Epidemiologi. Contact: Blog: suyatno.blog.undip.ac.id Hp/Telp: /

An Introduction to Epidemiology

Asian Journal of Research in Biological and Pharmaceutical Sciences Journal home page:

Lecture Outline. Biost 590: Statistical Consulting. Stages of Scientific Studies. Scientific Method

Confounding Bias: Stratification

About Reading Scientific Studies

Types of Biomedical Research

Evidence-Based Medicine Journal Club. A Primer in Statistics, Study Design, and Epidemiology. August, 2013

Case-control studies. Hans Wolff. Service d épidémiologie clinique Département de médecine communautaire. WHO- Postgraduate course 2007 CC studies

Study Designs in Epidemiology

Purpose. Study Designs. Objectives. Observational Studies. Analytic Studies

12/26/2013. Types of Biomedical Research. Clinical Research. 7Steps to do research INTRODUCTION & MEASUREMENT IN CLINICAL RESEARCH S T A T I S T I C

Epidemiologic Methods I & II Epidem 201AB Winter & Spring 2002

Causal Association : Cause To Effect. Dr. Akhilesh Bhargava MD, DHA, PGDHRM Prof. Community Medicine & Director-SIHFW, Jaipur

Study Design STUDY DESIGN CASE SERIES AND CROSS-SECTIONAL STUDY DESIGN

Welcome to this third module in a three-part series focused on epidemiologic measures of association and impact.

UNIT 5 - Association Causation, Effect Modification and Validity

Design of Observational and Experimental Clinical Studies Schulz 9/17/2008. The worst-taught taught

Is There An Association?

Rapid appraisal of the literature: Identifying study biases

Chapter 2 Epidemiology

Department of OUTCOMES RESEARCH

General Biostatistics Concepts

Strategies for Data Analysis: Cohort and Case-control Studies

Biases in clinical research. Seungho Ryu, MD, PhD Kanguk Samsung Hospital, Sungkyunkwan University

Epidemiology overview

LA SIERRA UNIVERSITY Department of Health and Exercise Science Fall 2011

In the 1700s patients in charity hospitals sometimes slept two or more to a bed, regardless of diagnosis.

Appraising the Literature Overview of Study Designs

ADENIYI MOFOLUWAKE MPH APPLIED EPIDEMIOLOGY WEEK 5 CASE STUDY ASSIGNMENT APRIL

VACCINE ACTIVE SURVEILLANCE I

CPH601 Chapter 3 Risk Assessment

What is a case control study? Tarani Chandola Social Statistics University of Manchester

Challenges of Observational and Retrospective Studies

Study design. Chapter 64. Research Methodology S.B. MARTINS, A. ZIN, W. ZIN

Chapter 11 Nonexperimental Quantitative Research Steps in Nonexperimental Research

Interpretation of Epidemiologic Studies

Confounding and Bias

Objectives. Lecture 6 July 16, Operating premises of risk assessment. Page 1. Operating premises of risk assessment

Glossary of Practical Epidemiology Concepts

Clinical Research Design and Conduction

Chapter 02. Basic Research Methodology

4/1/2013. Environmental epidemiology: challenges. calculation of RR in cohort study. Research designs & bias in environmental epidemiology

Module 2: Fundamentals of Epidemiology Issues of Interpretation. Slide 1: Introduction. Slide 2: Acknowledgements. Slide 3: Presentation Objectives

Meeting place and time: Biomed D205; Monday and Wednesdays 10:30 p.m. 11:45 p.m.

Using Epidemiology to Identify the Cause of Disease: Cohort study

Confounding and Interaction

Online Supplementary Material

CONSORT 2010 checklist of information to include when reporting a randomised trial*

Handout 1: Introduction to the Research Process and Study Design STAT 335 Fall 2016

1 Case Control Studies

CHECKLIST FOR EVALUATING A RESEARCH REPORT Provided by Dr. Blevins

Clinical problems and choice of study designs

Overview of Clinical Study Design Laura Lee Johnson, Ph.D. Statistician National Center for Complementary and Alternative Medicine US National

Data that can be classified as belonging to a distinct number of categories >>result in categorical responses. And this includes:

Anatomy of Medical Studies: Part I. Emily D Agostino Epidemiology and Biostatistics CUNY School of Public Health at Hunter College

Objective: To describe a new approach to neighborhood effects studies based on residential mobility and demonstrate this approach in the context of

Strategies for data analysis: case-control studies

Health Studies 315. Clinical Epidemiology: Evidence of Risk and Harm

Measures of Association

Understanding Statistics for Research Staff!

Cross-sectional studies

TEACHERS TOPICS. The Role of Matching in Epidemiologic Studies. American Journal of Pharmaceutical Education 2004; 68 (3) Article 83.

Risk Ratio and Odds Ratio

In this second module in the clinical trials series, we will focus on design considerations for Phase III clinical trials. Phase III clinical trials

LA SIERRA UNIVERSITY Department of Health and Exercise Science Winter 2016

Research Methodology Workshop. Study Type

FLHealthCHARTS.com Update List

OHDSI Tutorial: Design and implementation of a comparative cohort study in observational healthcare data

Epidemiologic study designs

PhD Course in Biostatistics

Impact Evaluation Toolbox

The Effects of Maternal Alcohol Use and Smoking on Children s Mental Health: Evidence from the National Longitudinal Survey of Children and Youth

Objectives. Distinguish between primary and secondary studies. Discuss ways to assess methodological quality. Review limitations of all studies.

Recent developments for combining evidence within evidence streams: bias-adjusted meta-analysis

observational studies Descriptive studies

University of Wollongong. Research Online. Australian Health Services Research Institute

VALIDITY OF QUANTITATIVE RESEARCH

Gender Identity and Expression. Week 8

Bias and confounding. Mads Kamper-Jørgensen, associate professor, Section of Social Medicine

Basic Epidemiology: Study Types. Learning Objectives. The Five Ws of Epidemiology. At the end of this presentation, participants will be able to:

Biases in clinical research. Seungho Ryu, MD, PhD Kanguk Samsung Hospital, Sungkyunkwan University

Modeling Binary outcome

Transcription:

Epidemiology: Overview of Key Concepts and Study Design Polly Marchbanks

Lecture Outline (1) Key epidemiologic concepts - Definition - What epi is not - What epi is - Process of epi research

Lecture Outline (2) Study Design general Descriptive studies - Case reports and case series - Descriptive incidence - Descriptive prevalence - Ecologic (correlational)

Lecture Outline (3) Analytic studies - Experimental - Observational - Cohort - Case-Control - Cross-sectional

Lecture Outline (4) Additional epi concepts - Bias or systematic error - Selection bias - Information bias - Confounding Summary, concluding remarks

EPIDEMIOLOGY is not The study of skin diseases Local Atlanta Resident

EPIDEMIOLOGY is (hopefully) The worst-taught course in school not Anonymous Student

EPIDEMIOLOGY is more than The science of long division Statistician I o = (480)(2)/10 6/yr (9.17)(0.953)+5

EPIDEMIOLOGY is more than The science of epidemics and epidemic diseases Medical Dictionary

The Derivation of Epidemiology epi = upon demos = people logy = study of

Fundamental Assumptions Non-randomness Preventability

Definition of Epidemiology The study of the distribution and determinants of disease and health in human populations.

Unique Contribution of Epidemiology Focus is on populations

Epidemiology (Study) The study of the distribution and determinants of disease and health in human populations.

Epidemiology (Distribution) The study of the distribution and determinants of disease and health in human populations.

Epidemiology (Determinants) The study of the distribution and determinants of disease and health in human populations.

Epidemiology (Disease And Health) The study of the distribution and determinants of disease and health in human populations.

Applied Epidemiology Practical, action-oriented, relevant Provides data for decision-making Focused on prevention/intervention

Aims of Epidemiologic Research Describe Explain Predict Control

Levels of Epidemiologic Research Theoretical understanding Prevention/intervention

Cholera in London, 1854

How to Find John Snow Pub London #39 Broadwick Street Corner of Broadwick & Lexington Can walk from theater district Connect these dots on a map: Oxford Circus, Soho Square, Leicester Square, Golden Square Broadwick is near center of the box you have created

Exposure Variable - "E" Characteristic of interest Risk factor variable Predictor variable Independent variable Possible causal factor

Outcome or Disease Variable - "D" Health event of interest Illness, injury Response variable Dependent variable Effect variable

E-D Relationships: Examples Smoking - lung cancer Obesity - heart disease Dietary fat - stroke Income - malnutrition Prenatal care - birth defects Family history - diabetes Alcohol - motor vehicle injury Access to care - maternal mortality

Process of Epidemiologic Research Question Design Conduct Analysis Interpretation Recommendations

Process of Problem-solving Process of Epidemiologic Research 1. What is the problem? 1. Question 2. How should the 2. Design problem be approached? 3. Collect information 3. Conduct 4. Organize and analyze 4. Analysis the information 5. What does the 5. Interpretation information mean? 6. What are the next 6. Recommendations steps?

The process of epidemiologic research is also a method of causal reasoning.

Epidemiology Theoretical Applied Consequential

Epidemiology: A Public Health Practice Summary Definition Quantitative basic science Method of causal reasoning Vehicle for clinical and public health action Can and should make a difference in the lives of people!

Epidemiology is FUN and if it s not FUN, it s not epidemiology! R. Stallones

Epidemiologic Study Design

Study Design: What Is It? Process of planning an investigation Link between concept and operation Bridge from ideas to action

Process of Epidemiologic Research Question Design Conduct Analysis Interpretation Recommendations

!! Let the Epidemiologist Beware!! "...all study designs are potentially flawed... there is no such thing as a perfect study design; therefore, it becomes most important to understand the specific limitations of each design...no type of sophisticated statistical analysis will salvage a poorly designed study." (KKM - pg. 36)

Categories of Study Design Descriptive Analytic

Descriptive Studies: Overview Patterns of occurrence Person, place, time Program planning Generate hypotheses

Descriptive Studies: Examples Trends in serum cholesterol Incidence of cervical cancer Secular trends in heart disease mortality Prevalence of domestic violence Incidence of rape among street youth Epidemiology of low birth weight Prevalence of smoking by age and gender Incidence and prevalence of dementia

Analytic Studies: Overview Determinants of disease Etiologic research Joint distribution of exposure (E) and outcome/disease (D) Test hypotheses Quantify association

Analytic Studies: Examples Diet - hypertension Exercise - depression Alcohol - liver cirrhosis Sun exposure - melanoma Smoking - heart disease Use of long acting reversible contraceptives (LARCs) - unintended pregnancy

Knowledge Continuum Less More Descriptive Search for clues Analytic Clues available

Types of Descriptive Studies Case reports and case series Descriptive incidence studies Descriptive prevalence studies Ecologic (correlational) studies

Case Reports and Case Series Profile of a case or case series Generate new hypotheses Interface: medicine and epidemiology Numerator data only No measure of disease occurrence

Descriptive Incidence Studies (1) Patterns in occurrence of incident cases Defined population Specified period of time Distribution of cases by factors of interest

Descriptive Incidence Studies (2) Case ascertainment: Ongoing reporting systems Medical record review in highly select populations Denominators usually from census Numerator and denominator difficult to obtain for well-defined population

Descriptive Prevalence Studies (1) "Snapshot" of well-defined population Classify D and other variables at same time Captures all existing disease Also known as cross-sectional surveys

Descriptive Prevalence Studies (2) Advantages: Quick, inexpensive, useful Disadvantages: Survivor effect Uncertain temporal relationship

Ecologic (Correlational) Studies (1) D in relation to E at aggregate level Data from groups not individuals Unit of observation is a population

Ecologic (Correlational) Studies (2) Ex: correlation between average cigarette sales and heart disease death rates in two counties Limitation: no individual link of E-D

Ecologic (Correlational) Studies (3) Advantages: Quick, inexpensive, data available Disadvantages: Aggregate may not = individual Inadequate data on co-variables Averages may mask complex relationships

Descriptive Studies - Summary Describe patterns of occurrence Four main types: Case reports and case series Descriptive incidence studies Descriptive prevalence studies Ecologic (correlational) studies Generate hypotheses for analytic study

Analytic Studies Joint distribution of E-D for individual Test hypotheses about E-D relationships Categories: experimental, observational

Experimental Studies (1) Assign E randomly, follow for D or other health outcome Types: Clinical trial Field trial Community trial

Experimental Studies (2) Rolls Royce!! Control of extraneous variables, both known and unknown Limitations: ethical concerns, cost, length, not feasible for rare outcomes, volunteer effect Stopping rules

Observational Studies Nature prevails Three main types: Cohort Case-control Cross-sectional Dimensions: Direction, timing

Observational Studies - Direction Temporal relationship between our observations of E and D Forward: start with E Backward: start with D Nondirectional: E-D simultaneously Essence: start with E or D?

Observational Studies - Timing Chronological relationship between onset of study and occurrence of D Prospective: study onset --> D D occurs after study begins Retrospective: D --> study onset D occurs before study begins Essence: has the D occurred when the study begins?

Cohort Studies - Direction Subjects free of D selected on basis of E, followed forward in time for D Start with persons exposed (YES, NO) Follow forward for disease (YES, NO) Direction always forward E ----> D Timing = prospective or retrospective

Cohort Studies - Prospective D has not yet occurred at study onset E ---> Study Onset ---> D Move from E to D through "real time" D happens in the present, concurrently with calendar time

Cohort Studies - Historical (Retrospective) D has already occurred at study onset E ---> D ---> Study Onset Direction still forward because moving from E to D Move from E to D through "historical time" D happened in the past, nonconcurrently with calendar time

Cohort Studies - Timeline E Study D Begins ----------- -------- ----> Prospective Cohort E D Study Begins ----------- --------- ----> Historical Cohort

Cohort Timeline - Example Maternal Thalassemia (Anemia) And Fetal Death Anemia Study Death Begins E D ----------- -------- ----> Prospective Cohort Anemia Death Study Begins E D ----------- --------- ----> Historical Cohort

Cohort Studies - Review Subjects selected on basis of E Direction always forward E ---> D Timing Prospective: "real time" Retrospective: "historical time"

Cohort Studies Flow Chart Source Population Study Group E+ Study Group E D+ D D+ D

Cohort - Measures of Occurrence THINK INCIDENCE!! Cumulative incidence - risk # new cases at end of follow-up divided by # disease-free persons at start of follow-up Incidence density - rate # new cases at end of follow-up divided by person-time at risk (e.g., person years of disease-free follow-up)

Cohort - Measures of Association (1) Think Relative Risk (RR)!! Risk ratio if cumulative incidence study Rate ratio if incidence density study Incidence of D in exposed divided by incidence of D in unexposed

Cohort - Measures of Association RR = 1 (2) No difference in incidence of D in exposed and unexposed 1.0 = no effect EX: 10/100 // 10/100 = 1.0 No association between E and D was observed - exposed and unexposed persons are equally likely to develop D

Cohort - Measures of Association (3) RR > 1 Suggests that E is a risk factor for D EX: 50/100 // 10/100 = 5.0 Exposed persons are 5 times as likely to develop the D, compared with persons not exposed

Cohort - Measures of Association RR < 1 (4) Suggests that E protects against D EX: 10/100 // 50/100 = 0.2 Exposed persons are 0.2 times as likely to develop the D, compared with persons not exposed

Cohort Studies - Major Advantages Logical temporal sequence Can measure incidence of D Well-suited for rare E Can study many effects of one E

Cohort Studies - Major Disadvantages Many subjects needed for rare D Follow-up: logistics, losses E status can change over time Prospective: time-consuming, costly, observation can influence behaviors Historical: requires suitable records

Case-control Studies - Direction Review direction: start with E or D? Subjects selected on basis of D Start with persons having D (YES, NO) Compare frequency of E in cases (D) with frequency of E in controls (non-d) Look backward for history of E (YES, NO) Direction is backward: D ---> E

Case-control Studies - Timing Review timing: has D happened when study begins? Basic Design: retrospective D has occurred before onset of study Hybrid Design: prospective component Study begins, enroll new cases of D

Control Selection Guidelines Critical design issue No optimal group for all situations Controls should... Represent source population from which the cases were derived Represent persons who, if a case, would have been in the study Be selected independently of E

Possible Sources of Controls Random sample of source population often best way to insure controls selected independent of E Can use other sources, but be cautious Hospitals Friends, relatives, spouses, neighbors

Review: Case-Control Studies Subjects selected on basis of D Start with D Look backward for E Direction always backward Disease ---> Exposure Timing in basic design is retrospective - D already occurred when study begins

Case-Control Studies Flow Chart Source Population Cases D+ Controls D E+ E E+ E

Case-control - Measures of Basic design: Occurrence Usually not possible Hybrid design: Incidence density rates possible if study is population-based

Case-control - Measures of Association THINK ODDS RATIO (OR)!! Odds ratio Estimator of relative risk Odds of E among cases compared with odds of E among controls Also known as exposure odds ratio OR = 1 - no association OR > 1 - suggests E is a risk factor OR < 1 - suggests E is protective

Case-control - Major Advantages Relatively quick and inexpensive Well-suited for rare D and D with long latency Requires fewer subjects at entry Can study multiple E

Case-control - Major Disadvantages Design "backward" Unsuitable for rare E Usually cannot measure D incidence Temporal E-D uncertainty Prone to selection and recall bias

Pop Quiz Breast Cancer Yes No Yes a b H1 OC Use No c d H2 V1 V2 Total

Example of Cohort Study: OC Use and Breast Cancer Start with OC users and nonusers Follow forward in time for breast cancer Compare incidence of breast cancer in OC users with incidence of breast cancer in nonusers

Example of Case-control Study: OC Use and Breast Cancer Start with cases of breast cancer and controls who do not have breast cancer Look backward in time for a history of OC use Compare frequency of OC use in cases with frequency of OC use in controls

Most Fundamental Difference Cohort Start with exposure Case-control Start with disease

Analytic Cross-sectional Studies (1) "Snapshot" in well-defined population Assess E-D (prevalent) at same time Nondirectional design Analytic potential

Analytic Cross-sectional Studies (2) Measures of association Prevalence ratio Prevalence of D among E compared with prevalence of D among non-e Prevalence odds ratio Odds of D among E compared with odds of D among non-e

Analytic Cross-sectional Studies (3) Advantages: Quick, inexpensive, useful, suitable when E cannot change Disadvantages: No measures of incidence Survivor effect Unsuitable for rare E, rare D, and D with short duration Temporal E-D uncertainty

Basic Concepts of Bias

Accuracy Lack of Error / \ Validity Precision Lack of systematic error (bias) Lack of random error (chance) - Function of - Function of study sampling methodology variation

Bias Bias = Systematic error that results in a distortion of the E-D association Validity = lack of bias Validity is a function of study methodology

Bias = Inaccurate Results Due to: Manner in which study subjects are selected - Selection bias Manner in which necessary information is collected or coded (classified) - Information (misclassification) bias Admixture of effect - effect of E on D is mixed up with effects of other variables - Confounding

E E D D D D Selection Bias Problem with who gets into the study Ē Ē

Information Bias D D E Problem with how information is collected or coded (classified) Ē

Confounding Problem with a Third Factor (F) E D - F must be associated with E F - F must be associated with D even in unexposed - F must not be intermediate between E and D

E Ē E Ē D D D D Selection Bias D D E Ē Information Bias E D Confounding F

Ways to Minimize Selection Bias Cohort - Select persons on basis of E independent of D - Strive for complete follow-up Case-control - Select persons on basis of D independent of E

Ways to Minimize Information Bias Obtain and classify all information as correctly as possible Obtain and classify information on E and D independently

Ways to Minimize Confounding Randomization Restriction Matching (partial restriction) Stratification Statistical modeling *Note: Be sure to collect needed information on covariables

Summary (1) Key concepts - Definition of epidemiology - Process of epidemiologic research

Summary (2) Types of study designs Descriptive Case reports and case series Descriptive incidence studies Descriptive prevalence studies Ecologic (correlational) studies Analytic Experimental Observational Cohort Case-control Cross-sectional

Summary (3) Additional epidemiologic concepts Bias or systematic error Selection bias Information bias Confounding

Study Design - Concluding Remarks Different ways to organize designs What design should I select? It depends! Must consider: Objectives of study Current knowledge about E-D Ethical issues Time, money, human resources Flexibility and creativity are KEY!

The End