study design and analysis in epidemiology - amazon s3 · 2017-05-28 · study design and analysis...
TRANSCRIPT
![Page 1: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/1.jpg)
000
Study Design and Analysis in Epidemiology:
Where does modeling fit?
Clinic on the Meaningful Modeling of Epidemiological Data, 2017
African Institute for Mathematical Sciences
Muizenberg, South Africa
Jim Scott, PhD, MPH, MA
Steve Bellan, PhD, MPH
Slide Set Citation: DOI:10.6084/m9.figshare.5038784
The ICI3D Figshare Collection
![Page 2: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/2.jpg)
Goals• Define Epidemiology
• Discuss metrics of epidemiology• Measures of disease
• Measures of effect
• Describe basic study designs used in epidemiology
• Discuss bias, confounding, and variability
• Compare and contrast statistical models and dynamic models
2
![Page 3: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/3.jpg)
Defining Epidemiology
“The study of the distribution and determinants of health related states and events in populations, and the application of this study to control health problems.”
John M Last Dictionary of Epidemiology
3
![Page 4: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/4.jpg)
Varieties of Infectious Disease Epidemiology
• Risk Factors & Intervention Epidemiology
Risk Factor: A characteristic that is correlated with a measure of disease.
Often used synonymously with • covariate.
Protective factors: Risk factors that are •negatively associated with disease
4
![Page 5: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/5.jpg)
Varieties of Infectious Disease Epidemiology
• Risk Factors & Intervention
• Outbreak
• Clinical
• Molecular & Genetic
• Surveillance 5
Pybus et al. 2015Proc R Soc Lond [B]
Influenza A H3N22002-2007
![Page 6: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/6.jpg)
How does mathematical modeling fit?
• Subfield: Linking pattern with process across scales
BUT ALSO
• Methodologies used in other epi subfields
Importance of knowledge breadth6
![Page 7: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/7.jpg)
What do Introductory Epidemiology courses teach?
• Measures of Disease
• Measures of Effect (of a risk factor)
• Study Designs for Measuring Effects• Dealing with random error• Dealing with confounding• Dealing with bias
• Biostatistical analyses for analyzing data
7
![Page 8: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/8.jpg)
Measures of Disease
Incidence•Cumulative Incidence•
Incidence Density•
Prevalence•Point Prevalence•
Period Prevalence•
Survivorship •(time to event, e.g. death)
Prevalence Study
Pe
op
le0
10
20
30
40
50
60●
●
Healthy
Disease
Prevalence =
4/60 = 7%
24−
05
26−
05
28−
05
30−
05
01−
06
03−
06
05−
06
07−
06
09−
06
11−
06
13−
06
TO
TA
L
0
10
20
30
40
50
60
incid
ence
Cum Incidence = 44 cases
60 people = 73% over 22 days
Inc Density = 44 cases
(60 people * 22 days) = 0.03
cases
person*day
8
![Page 9: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/9.jpg)
Measures of Covariates (risk factors)
• Binary: gender, smoker, circumcised
• Nominal/Categorical: geographic region
• Continuous: birth weight, T-cell count
• Ordinal: education, socioeconomic status (SES)
9
![Page 10: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/10.jpg)
Measures of Effect
How do you measure the effect of a risk factor on a •disease?
Example
How could you measure whether circumcision reduces the risk of HIV infection?
10
![Page 11: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/11.jpg)
Measures of Effect
• Compare measure of disease across levels/values of risk factors
• Relative RiskRatio of rates or proportions
• Prevalence Ratio
• Cum. Incidence Ratio
• Incidence Density Ratio
• Odds Ratio
> 8 hours of sleep < 8 hours of sleep
Relative Risk (RR) & Attributable Risk (AR)
flu
incid
en
ce p
er
100
0 p
eop
le
0
20
40
60
80
100
120
140● Attributable Risk = 10 per 1,000 people
Relative Risk = 110/100 = 1.1
• Attributable RiskSubtract rates or proportions 11
![Page 12: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/12.jpg)
Contingency Tables: Relative Risk (RR)
Disease No Disease Total (Margins)
Exposed a b a+b
Not exposed c d c+d
Total (Margins) a+c b+d a+b+c+d
Cumulative Incidence Ratio (CIR):
cumulative incidence in exposed population
divided by cumulative incidence in unexposed
population.
CIR < 1 exposure correlates with reduced risk of disease
CIR > 1 exposure correlates with increased risk of disease 12
![Page 13: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/13.jpg)
Epidemiologic Studies
• Descriptive Epidemiology• Baseline data on distribution of disease
• Surveillance
• Analytic Epidemiology – Measure Effect• Prospective Cohort Studies
• Cross-sectional Studies
• Retrospective Case-Control Studies
• Ecologic Studies
• Randomized Controlled Trials
Observational
Experimental
13
![Page 14: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/14.jpg)
Cohort Studies• Follow a selected population through time
• Establishes temporal relationships
• Can measure incidence
• Takes lots of resources, money, & time!
• Poor design for rare diseases.
14
![Page 15: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/15.jpg)
Cohort Data and Person-Time
15
![Page 16: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/16.jpg)
Couples Cohort Data
chronic
Bellan et al. (2015)16
![Page 17: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/17.jpg)
Couples Cohort Data
chronic
17
Bellan et al. (2015)
![Page 18: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/18.jpg)
Couples Cohort Data
chronic
18
Bellan et al. (2015)
![Page 19: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/19.jpg)
Couples Cohort Data
acute
chronic
19
Bellan et al. (2015)
![Page 20: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/20.jpg)
Relative Risk: Incidence Density Ratios
Disease No Disease Total (Margins)
Exposed a - PYe
Not exposed c - PY0
Total (Margins) a+c - PYe + PY0
Incidence Density Ratio is the ratio of incidence
density of the exposed population to that of the
unexposed population.
IDR < 1 means exposure correlates with reduced risk of disease
IDR > 1 means exposure correlates with increased risk of disease20
![Page 21: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/21.jpg)
Cross-Sectional Studies• Snapshot of diseases & risk factors.
• Cannot establish temporal relationship.
• Relatively cheap & easy.
• Population must be large to study rare disease.
• Not great for diseases of short duration. Why?
21
![Page 22: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/22.jpg)
Case-Control Studies• Compare diseased individuals to chosen controls.
• Quality of study depends entirely on how controls are chosen.
• Good for rare diseases.
• Relatively cheap & quick.
22
![Page 23: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/23.jpg)
Case Control Studies: Odds Ratios
Disease No Disease Total (Margins)
Exposed a b a+b
Not exposed c d c+d
Total (Margins) a+c b+d a+b+c+d
Odds ratio is the ratio of odds in the diseased
population divided by the odds in the
non-diseased population.
OR < 1 means exposure correlates with reduced risk of disease
OR > 1 means exposure correlates with increased risk of disease
Controls: Number chosen by researcher.
23
![Page 24: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/24.jpg)
Randomized Controlled TrialsExperimental or Intervention Studies•
Establishes temporal relationships•
Addresses confounding (more to come)•
Causal evidence•
24
![Page 25: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/25.jpg)
Ecologic Studies• Measurements made at population rather than
individual level.
• Weaker inference, but easier to gather data.
25
![Page 26: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/26.jpg)
Measures of Covariates (risk factors)
• Binary: gender, smoker, circumcised
• Nominal/Categorical: geographic region
• Continuous: birth weight, T-cell count
• Ordinal: education, socioeconomic status (SES)
26
![Page 27: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/27.jpg)
What do Introductory Epidemiology courses teach?
• Measures of Disease
• Measures of Effect (of a risk factor)
• Study Designs for Measuring Effects• Dealing with random error• Dealing with confounding• Dealing with bias
• Biostatistical analyses for analyzing data
27
![Page 28: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/28.jpg)
Random Error
• How many people must be in a study for the measure of effect to believable?
• Statistical Approach:Assign probabilities to our findings being a product of random error rather than a real phenomenon.
28
![Page 29: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/29.jpg)
Bias
Difference between observed value and true value due to all causes other than random error.
Bias does not go away with greater sample size!
Bias must be dealt with during study design!
29
![Page 30: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/30.jpg)
Selection BiasError due to systematic differences between those
who take part in the study and those who do not.John Last, Dictionary of Epidemiology
Information Bias
A flaw in measuring exposure or outcome data that results in different quality (accuracy) of information between comparison groups.
John Last, Dictionary of Epidemiology30
![Page 31: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/31.jpg)
Confounding
Literacy HIV Status
HIV+ HIV-
Literate 660 340
Illiterate 180 820
What if some of the study population were
much younger than others? 31
![Page 32: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/32.jpg)
ConfoundingPooled HIV+ HIV-
Literate 660 340
Illiterate 180 820
6-15 years old HIV+ HIV-
Literate 30 270
Illiterate 90 810
16-24 years old HIV+ HIV-
Literate 630 70
Illiterate 90 10
6-15 year olds: Literacy = 300/1200 = 25%
16-24 year olds: Literacy = 700/800 = 87.5% 32
![Page 33: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/33.jpg)
Confounding
Literacy HIV Status
HIV+ HIV-
Literate 660 340
Illiterate 180 820
Age
CONFOUNDING
33
![Page 34: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/34.jpg)
Biostatistical Analyses
• Permutation Tests
• Chi Squared Test
• Generalized Linear (Mixed) Models
• Normal Regression
• Logistic Regression
• Poisson Regression
• Negative Binomial Regression
• Survival Analysis
34
![Page 35: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/35.jpg)
• Account for bias and random error to find correlations that may imply causality.
• Often the first step to assessing relationships.
• Systems Approach:Explicitly model multiple mechanisms to understand their interactions.
• Explicitly focuses on dependence of individuals
Statistical Models Dynamic Models
By developing dynamic models in a probabilistic framework we can account for dependence, random error, and bias while
linking patterns at multiple scales.
• Assume independence of individuals (at some scale, i.e. clusters).
• Links observed relationships at different scales.
35
![Page 36: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/36.jpg)
Statistical Models Dynamic Models
• Is HIV status positively associated with the risk of TB infection?
Questions in Epidemiology
• Based on increased TB risk due to HIV, how much should we expect TB notification rate to increase for a given HIV prevalence?
36
![Page 37: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/37.jpg)
Statistical Models Dynamic Models
• Are Insecticide Treated Bednets (ITNs) or Indoor Residual Spraying (IRS) more effective for controlling malaria?
Questions in Epidemiology
• How do we expect the age-distribution of malaria incidence to change after implementing ITNs or IRS?
37
![Page 38: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/38.jpg)
Summary• Epidemiology seeks to measure and understand the
causes of disease
• Epidemiologists conduct observational and experimental studies to obtain data
• Incidence and prevalence• RR’s, OR’s, attributable risk
• Epidemiological studies are subject to bias, confounding, and variability
• Traditionally, focus on discovering correlations that may lead to identification of causal factors
• Dynamic models, informed via epidemiologic studies, can answer questions traditional studies can’t
38
![Page 39: Study Design and Analysis in Epidemiology - Amazon S3 · 2017-05-28 · Study Design and Analysis in Epidemiology: ... and the application of this study to control health problems](https://reader034.vdocument.in/reader034/viewer/2022050406/5f833bbec8d9657fa658da3b/html5/thumbnails/39.jpg)
000
This presentation is made available through a Creative Commons Attribution license. Details of the license and permitted uses are available at
http://creativecommons.org/licenses/by/3.0/
Bellan SE. “Study Design and Analysis in Epidemiology: Where does modeling fit?” Clinic on the Meaningful Modeling of Epidemiological Data. DOI:10.6084/m9.figshare.5038784.
For further information or modifiable slides please contact [email protected].
See the entire ICI3D Figshare Collection. DOI: 10.6084/m9.figshare.c.3788224.
000
000
© 2010 International Clinics on Infectious Disease Dynamics and Data
000