Personalized Medicine Detection Diagnosis Treatment Survival.

Slides:



Advertisements
Similar presentations
Randomized controlled trials
Advertisements

Studying a Study and Testing a Test: Sensitivity Training, “Don’t Make a Good Test Bad”, and “Analyze This” Borrowed Liberally from Riegelman and Hirsch,
Study Designs in Epidemiologic
1 Case-Control Study Design Two groups are selected, one of people with the disease (cases), and the other of people with the same general characteristics.
BIAS AND CONFOUNDING Nigel Paneth. HYPOTHESIS FORMULATION AND ERRORS IN RESEARCH All analytic studies must begin with a clearly formulated hypothesis.
Chance, bias and confounding
Estimation and Reporting of Heterogeneity of Treatment Effects in Observational Comparative Effectiveness Research Prepared for: Agency for Healthcare.
Biostatistics ~ Types of Studies. Research classifications Observational vs. Experimental Observational – researcher collects info on attributes or measurements.
Potential Roles and Limitations of Biomarkers in Alzheimer’s Disease Richard Mayeux, MD, MSc Columbia University.
Epidemiology in Medicine Sandra Rodriguez Internal Medicine TTUHSC.
Model and Variable Selections for Personalized Medicine Lu Tian (Northwestern University) Hajime Uno (Kitasato University) Tianxi Cai, Els Goetghebeur,
Writing a Research Protocol Michael Aronica MD Program Director Internal Medicine-Pediatrics.
Biostatistics. But why? Why do we read scientific litterature? How do we read scientific litterature?
By Dr. Ahmed Mostafa Assist. Prof. of anesthesia & I.C.U. Evidence-based medicine.
Sample Size Determination
Sample Size and Statistical Power Epidemiology 655 Winter 1999 Jennifer Beebe.
EVIDENCE BASED MEDICINE
Cohort Studies Hanna E. Bloomfield, MD, MPH Professor of Medicine Associate Chief of Staff, Research Minneapolis VA Medical Center.
Principles of Epidemiology Lecture 12 Dona Schneider, PhD, MPH, FACE
Community Medicine. The association between low birth weight and maternal smoking during pregnancy can be studied by obtaining smoking histories from.
Statistics in Screening/Diagnosis
BASIC STATISTICS: AN OXYMORON? (With a little EPI thrown in…) URVASHI VAID MD, MS AUG 2012.
Chapter 8 Introduction to Hypothesis Testing
Hypothesis Testing.
Multiple Choice Questions for discussion
 Be familiar with the types of research study designs  Be aware of the advantages, disadvantages, and uses of the various research design types  Recognize.
Lecture 8 Objective 20. Describe the elements of design of observational studies: case reports/series.
Copyright © 2012 Wolters Kluwer Health | Lippincott Williams & Wilkins Chapter 7: Gathering Evidence for Practice.
Study Design. Study Designs Descriptive Studies Record events, observations or activities,documentaries No comparison group or intervention Describe.
Epidemiology The Basics Only… Adapted with permission from a class presentation developed by Dr. Charles Lynch – University of Iowa, Iowa City.
Lecture 4: Assessing Diagnostic and Screening Tests
Lecture 17 (Oct 28,2004)1 Lecture 17: Prevention of bias in RCTs Statistical/analytic issues in RCTs –Measures of effect –Precision/hypothesis testing.
CHP400: Community Health Program- lI Research Methodology STUDY DESIGNS Observational / Analytical Studies Case Control Studies Present: Disease Past:
Chapter 8: Introduction to Hypothesis Testing. 2 Hypothesis Testing An inferential procedure that uses sample data to evaluate the credibility of a hypothesis.
Lecture 6 Objective 16. Describe the elements of design of observational studies: (current) cohort studies (longitudinal studies). Discuss the advantages.
EDRN Approaches to Biomarker Validation DMCC Statisticians Fred Hutchinson Cancer Research Center Margaret Pepe Ziding Feng, Mark Thornquist, Yingye Zheng,
Study Designs in Epidemiologic
Research Process Parts of the research study Parts of the research study Aim: purpose of the study Aim: purpose of the study Target population: group whose.
CHP400: Community Health Program - lI Research Methodology STUDY DESIGNS Observational / Analytical Studies Present: Disease Past: Exposure Cross - section.
Case-control study Chihaya Koriyama August 17 (Lecture 1)
Causal relationships, bias, and research designs Professor Anthony DiGirolamo.
Cost-effectiveness of Screening Tests Mark Hlatky, MD Stanford University.
Evaluating Screening Programs Dr. Jørn Olsen Epi 200B January 19, 2010.
©2010 John Wiley and Sons Chapter 2 Research Methods in Human-Computer Interaction Chapter 2- Experimental Research.
Screening and its Useful Tools Thomas Songer, PhD Basic Epidemiology South Asian Cardiovascular Research Methodology Workshop.
Unit 15: Screening. Unit 15 Learning Objectives: 1.Understand the role of screening in the secondary prevention of disease. 2.Recognize the characteristics.
Organization of statistical research. The role of Biostatisticians Biostatisticians play essential roles in designing studies, analyzing data and.
Case-Control Studies Abdualziz BinSaeed. Case-Control Studies Type of analytic study Unit of observation and analysis: Individual (not group)
EBM --- Journal Reading Presenter :呂宥達 Date : 2005/10/27.
EVALUATING u After retrieving the literature, you have to evaluate or critically appraise the evidence for its validity and applicability to your patient.
Clinical Epidemiology and Evidence-based Medicine Unit FKUI – RSCM
Descriptive study design
BIOSTATISTICS Lecture 2. The role of Biostatisticians Biostatisticians play essential roles in designing studies, analyzing data and creating methods.
European Patients’ Academy on Therapeutic Innovation The Purpose and Fundamentals of Statistics in Clinical Trials.
Types of Studies. Aim of epidemiological studies To determine distribution of disease To examine determinants of a disease To judge whether a given exposure.
Marshall University School of Medicine Department of Biochemistry and Microbiology BMS 617 Lecture 6 –Multiple hypothesis testing Marshall University Genomics.
1 Study Design Imre Janszky Faculty of Medicine, ISM NTNU.
Direct method of standardization of indices. Average Values n Mean:  the average of the data  sensitive to outlying data n Median:  the middle of the.
Sensitivity, Specificity, and Receiver- Operator Characteristic Curves 10/10/2013.
Uses of Diagnostic Tests Screen (mammography for breast cancer) Diagnose (electrocardiogram for acute myocardial infarction) Grade (stage of cancer) Monitor.
Purpose of Epi Studies Discover factors associated with diseases, physical conditions and behaviors Identify the causal factors Show the efficacy of intervening.
Screening Tests: A Review. Learning Objectives: 1.Understand the role of screening in the secondary prevention of disease. 2.Recognize the characteristics.
BIAS AND CONFOUNDING Nigel Paneth.
Diagnostic Test Studies
Principles of Epidemiology E
How to read a paper D. Singh-Ranger.
BIAS AND CONFOUNDING
Review – First Exam Chapters 1 through 5
Evidence Based Diagnosis
Basic statistics.
Presentation transcript:

Personalized Medicine Detection Diagnosis Treatment Survival

Prediction is very difficult, especially about the future. Niels Bohr Danish physicist ( ) Niels Bohr

Biomarkers TestInformationDecision Outcome 1. Discrimination (sensitivity, specificity, predictive value, ROC analysis) 2. Utility (disease free survival, recurrence rates, survival etc)

Diagnostic tests Describing test performance Test Result DiseaseNo disease Total Positiveaba+b Negativecdc+d Totala+cc+da+b+c+d Properties of a test Sensitivity: – a/a+c Specificity: – d/c+d Positive predictive value: – a/a+b Negative predictive value: – d/c+d

The importance of disease prevalence Test result Breast cancer No breast cancer Total Positive3604,9805,340 Negative4094,62094,660 Total40099,600100,000 Screening mammography Properties of the test Sensitivity: 90% a/a+c = 360/400 Specificity: 95% d/c+d = 94,620/99,600 Positive predictive value: a/a+b = 360/5340 = 7% Negative predictive value: d/c+d =94,620/94,660 = 100%

Desiderata for studies of diagnostic tests. “Gold” standard Test result before outcome known “Blind” reading Pre-determined cut-off Sensitivity and specificity. Predictive value. Receiver operating. characteristic curves (ROC).

Diagnostic tests and the spectrum of disease. Spectrum of patients. Clinical spectrum Co-morbid spectrum Pathologic spectrum Potential biases in test evaluation. Exclusion of equivocal cases Work up bias Test review bias Incorporation bias

Clinical value of tests Test Information Decision Outcome

PRINCIPAL AGENT COMPARATIVE AGENT INITIAL STATE, RECIPIENTS OF PRINCIPAL AGENT INITIAL STATE, RECIPIENTS OF COMPARATIVE AGENT SUBSEQUENT EVENTS, RECIPIENTS OF PRINCIPAL AGENT SUBSEQUENT EVENTS, RECIPIENTS OF COMPARATIVE AGENT

Research Designs-General Structure Purpose of research (initial states) Prevention. Prediction of risk in healthy. Treatment response or toxicity in those with disease. Identify factors that influence outcome (prognosis). Types of manoeuver Inherited (eg genetic variant). Acquired –Self selected (smoking, alcohol) –Other (treatment). Imposed (atomic irradiation).

Principal research designs Disease PresentAbsent Presentab Exposure Absentcd Passage of time Relative risk = a/a+b ÷ c/c+d Cohort study

Nested case control studies Screening programs: NBSS, SMPBC, OBSP case 6-8 years follow-up case control Baseline mammogram Risk factors

How many subjects (or samples) do you need? Number of events (eg deaths). Willingness to risk a false positive (Type I) error. Willingness to risk a false negative result (Type II) error. Magnitude of difference worthwhile to detect. Time for accrual and follow-up.

Sample size to detect an improvement in survival (alpha=0.05; 1-beta=0.90) P2-P1 P

Sample size for genetic studies

SUBSEQUENT EVENTS R } PRINCIPAL AGENT COMPARATIVE AGENT { INITIAL STATE

A trial to change diet Vancouver + Surrey Windsor London + Sarnia Hamilton + KW Toronto Funding: Ontario Ministry of Health, Medical Research Council, Canadian Breast Cancer Research Alliance, National Institutes of Health, American Institute for Cancer Research Screening Randomization 4,693 Low-fat diet Usual diet >8 years counseling and follow-up

Association or causation? Not all associations are causal All causal factors show association May be due to bias or confounding Genetic associations –Causal –In linkage disequilibrium with the causal variant –Population stratification

Population stratification Type of confounding Ethnicity –associated with disease –associated with genotype –gives spurious association between genotype and disease Can be controlled in analysis (if recognized) Dispute about importance

Analysis P<0.05 What does this mean?

The meaning of p-values. If the TRUE difference between the compared groups is zero (the null hypothesis), the PROBABILITY of obtaining a difference as large or larger than the one observed by CHANCE is p.

Multiple comparisons The problem. If alpha = comparisons can be expected to generate one p<0.05. (1-(1-alpha) k, where alpha is the level for significance and k=number of tests. What protection? Few, a priori hypotheses Correction for number of tests eg Bonferroni –Alpha/number of tests Stringent alpha eg E Replication/validation

Francis Galton’s ox and the “Winner’s curse”. Country fair in bought tickets and predicted the weight of an ox. Actual weight was 1,198 lbs. None were close to the actual weight. Mean predicted weight (N=787) was 1,197 lbs. At auction, most bids cluster around the “true” value of the object. The winning bid is always higher than the “true” value.

Replication -validation “leave one out” – Applied to “learning set” – Not an independent sample – May help avoid overfitting Independent data set – Preferably also an independent investigator

How to get a statistically significant result. Count or ignore differences in follow-up time. Censor at different time points. Exclude specific causes of death. Exploit sub-group analysis. Use different cut-offs for gene expression (or other test result). Note: all of the above increases the number of statistical tests you can do!

Can you believe the literature? Publication bias (author and editor bias). Multiple statistical testing. The “Winner’s curse”. Bias in the sampling, measurement or analysis of the data. Most published reports are never replicated.

The “ Winners Curse ” False positives more likely : Small studies Small effects Early, hypothesis generating studies Financial interest “ Hot ” field Ioannidis PLos Medicine 2005

How to stay out of trouble Define target population. Standardize sample collection. Collect samples at zero time. Define outcomes at the outset. Random selection of cases and controls. Analyze samples without knowledge of case/control status. Replicate.