Epidemiology Master Level Quiz

spiderman2017

AnalyticStudiesandClinicalTrials.pdf

Home >Applied Sciences homework help >Epidemiology Master Level Quiz

The material in this video is subject to the copyright of the owners of the material and is being provided for educational purposes under rules of fair use for registered students in this course only. No additional copies of the copyrighted work may be made or distributed.

Descriptive vs. Analytic Epidemiology

n  Epidemiology: the study of the occurrence and distribution of health-related states or events in specified populations, including the study of determinants influencing such states, and the application of this knowledge to control the health problems

Descriptive vs. Analytical Epidemiology

Descriptive epidemiology

Descriptive vs. Analytical Epidemiology

Analytical epidemiology

n  Public health planning: how many people are affected? u  How important is a health problem is to individuals and to society? u  Does the problem merit resources and risks and benefits of intervention?

n  How do the patterns of occurrence of the public health problem vary by different characteristics? u  Helps to generate hypotheses about the causes of the problem

•  Descriptive analytical

Descriptive Epidemiology: Why?

n  Determining risk factors and causes of disease u  Does maternal exposure to asthmagens increase the risk of autism spectrum

disorders in her offspring? u  Is childhood obesity associated with increased incidence of type 2 diabetes?

n  Evaluating preventive and therapeutic interventions that alter the course of disease u  Does beginning HIV treatment earlier (at higher CD4 levels) lead to better health

outcomes (non-detectable viral load)?

Analytical Epidemiology: Why?

n  Can use different study designs to address this research question

Step 1: Is There an Association?

Main Types of Epidemiologic Studies

Study type Characteristics

Experimental • Studies prevention and treatment of disease • Investigator actively manipulates which groups receive the study agent

Observational • Studies causes, prevention and treatment for diseases • Investigator watches as natures takes its course

Cohort • Examines multiple health effects of an exposure • Subjects defined by exposure levels and follow for disease occurrence

Case-control • Typically examines multiple exposures in relation to a disease • Subjects are defined as cases and controls and exposure histories compared

Cross-sectional • Examine relationship between exposure and disease prevalence in a defined population at one point in time

Ecological • Examines relationship between exposure and disease with population-level data rather than individual data

“A controlled experiment … done to evaluate the efficacy [or effectiveness] and safety of a treatment in ameliorating or curing [or preventing] that disease or related health condition.”

Clinical Trial: a Definition

Source: Meinert, CL. (2012). Clinical trials: design, conduct, and analysis. New York: Oxford. 8

Epidemiologic study that does not involve any intervention, experimental or otherwise. Such a study may be one in which nature is allowed to take its course, with changes in one characteristic being studied in relation to changes in other characteristics.

Observational Studies Defined

Source: Last. Dictionary of Epidemiology, 4th edition. 9

n  Example study designs u  Cohort u  Case-control u  Cross-sectional u  Ecological

Observational Studies Defined

Clinical trials

n  Experimentation vs. observation u  Experimentation: “exposure” is directed (or “manipulated”) by the researcher;

exposure is the experimental intervention u  Observation: exposure is observed by the researcher as it plays out over time

n  Controlled trials differ from other epidemiologic studies in that they are experimental rather than observational

Trials vs. Observational Studies

n  Deborah Donnell et al, June 2010, “Heterosexual HIV-1 transmission after initiation of antiretroviral therapy: a prospective cohort analysis,” The Lancet, Volume 375, Issue 9731, 12–18 Pages 2092–2098 u  DOI: 10.1016/S0140-6736(10)60705-2

n  Among 3,381 couples, only 1 of 103 genetically-linked transmissions was among an infected participant starting HAART early u  92% reduction in transmission

Observational Study Findings of ART

n  Cohen et al, 2011, “Prevention of HIV-1 infection with early antiretroviral therapy,” N Engl J Med; 365:493–505 u  DOI: 10.1056/NEJMoa1105243

n  RCT of 1,763 HIV discordant couples in 13 countries, where half started HAART immediately or at CD4 <200 u  96% reduction in HIV transmission

Controlled Trial Results of ART

n  Donnell: June, 2010 u  382 citations through 9/27/13 u  10.05/month

n  Cohen: August, 2011 u  1,331 citations through 9/27/13 u  53.24/month

Which Article Is Cited More Often?

Another Study n  Writing Group for the Women's Health Initiative Investigators, 2002, “Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the women's health initiative randomized controlled trial,” JAMA; 288(3):321–333 u  doi:10.1001/jama.

288.3.321

n  A bias that can arise in observational studies when patients with the worst (or best) prognosis are preferentially allocated to a particular treatment

n  These patients are likely to be systematically different from those not treated, or treated with something else

Confounding by Indication

n  High-risk hypertensive patients are more likely to have adverse outcomes—including MI and death—regardless of treatment u  Patient characteristic associated with outcome

n  High-risk hypertensive patients are more likely to receive calcium channel blockers than ACE inhibitors, beta-blockers,or diuretics u  Patient characteristic associated with treatment

n  Observational studies show that calcium channel blockers are associated with more adverse outcomes u  Due to patient characteristics or true drug effect?

Confounding by Indication: Examples

“Confounding is most simply defined as the mixing of effects between an exposure, an outcome, and a third extraneous variable known as a confounder.”

Confounding

Source: Ashengrau and Seage. (2008). Essentials of Epidemiology in Public Health. Second edition. 19

n  When present, the association between the exposure and outcome is distorted—as the confounder is related to both the exposure and the outcome

Confounding

20 Source: Ashengrau and Seage. (2008). Essentials of Epidemiology in Public Health. Second edition.

X Y

n  Drugs and other treatments of disease u  Versus control (no treatment, placebo or active treatment)

n  Medical and health care technology

n  Comparative effectiveness research

n  Primary prevention u  Community health programs

n  Behavioral interventions (primary and secondary)

Interventions that Can Be Evaluated by RCT

n  Superiority trials: one treatment is better than the other u  Define a priori a clinically important difference

n  Equivalence or non-inferiority trials: two treatments are equal (within some small margin) u  Must be “much” smaller than clinically important difference u  Sometimes larger sample sizes are needed because you are trying to “prove” that

difference between treatments is small

Treatment Comparisons

Design of Controlled Trials

Section B

Design of Controlled Trials Defined population

NEW treatment

Current/NO treatment

Not Improved Improved

!  Ensures that participant assignment to treatment is unbiased !  Treatment given is not influenced by provider bias or patient prognosis: avoids

“confounding by indication” !  Treatment groups are comparable at the start of the study

What Does Randomization Do?

Non-Randomized Observational Study

Exposure n = 1,000

180

No exposure n = 1,000

300 Total deaths

Mortality: 180

1,000 = 18%

300 1,000

= 30%

Non-Randomized Observational Study: Patients with Arrhythmia

Are More Likely to Die, and Proportion of Patients with Arrhythmia in the Two Groups May Differ

Exposure n = 1,000

No exposure n = 1,000

180

800 X(–)

200 X(+)

100 80

300

500 X(–)

500 X(+)

250 50

Total deaths

Mortality: 180

1,000 = 18%

300 1,000

= 30%

Randomized Experimental Study: Proportions of Patients with Arrhythmia in the Two Groups Are Likely to Be Similar

New treatment n = 1,000

240

650 X(–)

350 X(+)

175 65

Current treatment n = 1,000

240

650 X(–)

350 X(+)

175 65

Total deaths

Mortality: 240

1,000 = 24%

240 1,000

= 24%

!  Primary: to remove the potential for bias in the choice of treatment (treatment selection bias, confounding by indication)

!  Secondary: to increase the likelihood of balance between groups for known and unknown risk factors

!  Tertiary: to provide a probability basis for statistical testing

Why Is Randomization Important?

!  Protections from bias !  Major reason that randomized trials are considered the gold standard of study

design

Benefits of Randomization

1.  Selection of subjects !  High-risk population

•  Number of events anticipated, departure from normative levels, e.g., high blood pressure

!  Generalizability, external validity !  Clearly defined in protocol

Elements in the Design of Controlled Trials

1.  Selection of subjects

2.  Allocation of subjects to treatment groups !  Randomization

Elements in the Design of Controlled Trials

!  Define allocation ratio for treatments !  1:1 allocation ratio: 50% receive new treatment, 50% receive control !  Every treatment assignment has a known probability associated with it

•  For example, coin toss, dice throw

!  Makes treatment assignment process unpredictable !  Conceals treatment assignment prior to enrollment of a participant

!  Randomization is distinct from masking, which means you conceal treatment assignment during follow-up !  Commonly referred to as single-blind or double-blind

How Does Randomization Work?

1.  Selection of subjects

2.  Allocation of subjects to treatment groups !  Randomization !  Stratification and randomization

Elements in the Design of Controlled Trials

!  Ensures balance in treatment assignments within subgroups defined before randomization !  Clinic, gender, risk level, arrhythmia status

!  Stratification variable should be related to outcome—prognostic factor that is strongly related to outcome

!  Requires a separate set of treatment assignment schedules for each category of each stratum

Stratified Randomization

1.  Selection of subjects

2.  Allocation of subjects to treatment groups

3.  Uniform data collection !  Treatment !  Outcome !  Prognostic profile at entry (baseline characteristics)

Elements in the Design of Controlled Trials

1.  Selection of subjects

2.  Allocation of subjects to treatment groups

3.  Uniform data collection

4.  Masking (blinding) !  Subjects !  Treaters !  Data collectors !  Data analysts

Elements in the Design of Controlled Trials

Masking by Over- Encapsulation

!  To reduce bias related to prior knowledge or beliefs about treatments

!  For more objective: !  Data collection !  Concomitant treatments !  Outcome assessment !  Interim data interpretation

Why Mask?

Analysis of RCT Studies

Section C

1.  Selection of subjects

2.  Allocation of subjects to treatment groups

3.  Uniform data collection

4.  Masking (blinding)

5.  Data analysis !  Pre-specify primary and secondary endpoints !  Analysis method set a priori

Elements in the Design of Controlled Trials

!  Drop-outs !  Non-adherence to the experimental regimen during follow-up !  Losses to follow-up

!  Drop-ins !  Non-adherence to the control regimen during follow-up

Problems of Non-Compliance

Drop-out and Drop-in Rates: Women’s Health Initiative

Non-Adherence during Follow-up in Alzheimer’s Disease Anti-Inflammatory Prevention Trial (ADAPT)

Refuse or can’t tolerate NSAID

Require NSAID or take on own

Randomized

NSAID Placebo

Non-Adherence during Follow-up in Alzheimer’s Disease Anti-Inflammatory Prevention Trial (ADAPT)

Refuse or can’t tolerate NSAID

Require NSAID or take on own

No treatment

Randomized

NSAID Placebo

Non-Adherence during Follow-up in Alzheimer’s Disease Anti-Inflammatory Prevention Trial (ADAPT)

Refuse or can’t tolerate NSAID

Require NSAID or take on own

NSAID No treatment

Randomized

NSAID Placebo

!  Is this analysis biased?

As Treated Analysis in ADAPT

NSAID No treatment

Randomized

NSAID Placebo

As treated

!  Is this analysis biased?

Intention to Treat Analysis in ADAPT

No treatment NSAID

Randomized

NSAID Placebo

Maintains randomization

!  Not with respect to baseline characteristics

!  May to lead to conservative estimate of the treatment effect

Intention to Treat Analysis in ADAPT

No treatment NSAID

Randomized

NSAID Placebo

5.  Data analysis !  Primary: intention to treat

•  Net effect of non-compliance is to reduce observed differences) !  Secondary: treatment received

•  Observational; no longer have benefits of randomization !  Subgroup analyses

•  Small numbers, hard to determine that treatment effect differs by sub-group

Elements in the Design of Controlled Trials

!  Analyze data based purely on randomization

!  Ignore: !  Ineligibility !  Complete non-adherence !  Treatment terminations !  Treatment switches !  Partial adherence

!  Sounds illogical; in principle, it isn’t

Intention to Treat (ITT)

!  Any exclusion after randomization or adjustment for anything occurring after baseline (e.g., adherence) that is differential by treatment potential to introduce bias and makes the design observational, not experimental

!  Non-adherence is not random !  Adherence may be related to treatment !  Individuals who are not compliant might also have other “factors” which are related

to the outcome

Violation of ITT

!  Most conservative analysis for a superiority design

!  Not necessarily the most conservative analysis for a non-inferiority design !  Analysis by treatment received equally as important

Caveat for ITT

!  Use of pilot studies (run-in periods); effectiveness depends on: !  Precision in predicting compliance !  Costs of recruitment

!  Methods for monitoring compliance !  Interview patients, count pills !  Medication bottle devices !  Blood or urine tests !  Directly observed treatment

Dealing with Non-Compliance

1.  Was the analysis intended at the start of the trial?

2.  Is the analysis likely to be subject to bias?

3.  Is the analysis biologically plausible and therefore justified?

4.  Is the result of the trial significant overall?

Subgroup Analyses: Four Questions to Help Decide if a Subgroup Analysis Is Justified

Meta-Analysis of Statin Dosages and Incident Diabetes

Source: (2011). JAMA;305(24):2556–2564. 17

Meta-Analysis of Statin Dosages and Incident Diabetes

Source: (2011). JAMA;305(24):2556–2564. 18

Crossover Trial

!  Each patient serves as his or her own control, thereby ensuring comparability of treatment groups

!  Feasible only if: !  Outcomes are “recurrent” !  No “carryover” treatment effect after “washout” period

!  Randomize order in which treatments are given as order may influence results

!  Baseline comparability no longer an issue

!  Generalizability remains an issue

Crossover Design

Ethical Issues

!  Is it ethical to randomize?

!  Is it ethical not to randomize?

!  Who should be eligible?

!  Can truly informed consent be obtained?

!  Who should elicit consent?

!  When can placebos be used?

!  When can shams be used?

!  When should a trial be stopped earlier than originally planned?

!  Distinguish between descriptive epidemiology and analytical epidemiology

!  Describe experimental and observational designs

!  Discuss the benefits of randomization

!  State the basic design elements of randomized trials

!  Distinguish between different designs for randomized trials (crossover)

Summary

Main Types of Epidemiologic Studies

Study type Characteristics

Experimental • Studies prevention and treatment of disease • Investigator actively manipulates which groups receive the study agent

Observational • Studies causes, prevention and treatment for diseases • Investigator watches as natures takes its course

Cohort • Examines multiple health effects of an exposure • Subjects defined by exposure levels and follow for disease occurrence

Case-control • Typically examines multiple exposures in relation to a disease • Subjects are defined as cases and controls and exposure histories compared

Cross-sectional • Examine relationship between exposure and disease prevalence in a defined population at one point in time

Ecological • Examines relationship between exposure and disease with population-level data rather than individual data

Your feedback is very important and will be used for future revisions.

The Evaluation link is available on the lecture page.

Lecture Evaluation