drugepi 6 - reverse time order module 4 overview context content area: interpretation of...

DrugEpi 6 - Reverse Time Order

Module 4 OverviewContext

Content Area: Interpretation of Epidemiological EvidenceEssential Question (Generic): Is the association causal? Essential Question (Drug Abuse Specific): Is an association with drug abuse causal?Enduring Epidemiological Understanding: Causation is only one explanation for finding an association between an exposure and a disease. Because observational studies are complicated by factors not controlled by the observer, other explanations must also be considered.

Synopsis: In Module 4, students explore the rationale and methods of interpreting epidemiological studies. Students develop skills to assess possible explanations for an association found in a study, with consideration of explanations of causality, chance, confounding, reversed time order, and bias. Module 4 concludes with consideration of methods for weighing the overall evidence for an association.

Lessons:

Lesson 4-1: Introduction to Interpreting AssociationsLesson 4-2: CausalityLesson 4-3: Chance Lesson 4-4: Bias Lesson 4-5: ConfoundingLesson 4-6: Reverse Time Order Lesson 4-7: Weighing the Evidence


Module 4 - Interpretation of Epidemiological Evidence

Lesson 4-6 - Reverse Time Order

Content

• Definition and explanation of reverse time order• Examples of exposure/outcome scenario that illustrate how reverse time

order can be considered• Example of an improved epidemiologic study design to determine whether

or not there could be reverse time order

Big Ideas

• Association does not necessarily mean causation• When an association is found, several possible explanations must be

considered, including the possibility that the association is due to reverse time order

This project is supported by a Science Education Drug Abuse Partnership Award, Grant Number 1R24DA016357-01, from the National Institute on Drug Abuse,

National Institutes of Health.


Where Are We?

Essential Questions Enduring Understandings

1. How is this disease distributed?

Health-related conditions and behaviors are not distributed uniformly in a population. They have unique distributions that can be described by how they are distributed in terms of person, place, and time.

2. What hypotheses might explain the distribution of disease?

Clues for formulating hypotheses can be found by observing the way a health-related condition or behavior is distributed in a population.

3. Is there an association between the hypothesized cause and the disease?

Causal hypotheses can be tested by observing exposures and diseases of people as they go about their daily lives. Information from these observational studies can be used to make and compare rates and identify associations.

4. Is the association causal?

Causation is only one explanation for an association between an exposure and a disease. Because observational studies are complicated by factors not controlled by the observer, other explanations also must be considered.

5. What should be done when preventable causes of disease are found?

Policy decisions are based on more than the scientific evidence. Because of competing values - social, economic, ethical, environmental, cultural, and political factors may also be considered.


Reverse Time Order

Is the association due to

reverse time order?

1. Cause

2. Chance

3. Chance

Confounding

Reverse Time Order

4.

5.


Which came first?

Egg Chicken

“The Riddle” by Charles Bragg


Scenario 1


Scenario 2


Which came first?

Egg Chicken



Definition

When the hypothesized time order of an exposure and an outcome is

actually reversed and the outcome really comes before the exposure.

Reverse Time Order


Time

Random Assignment

E

E

DZ

DZ

DZ

DZ

Healthy People

Flow Diagram

-

Trial


E

E

DZ

DZ

DZ

DZ

Healthy People

-

Time

Cohort

Study

Flow Diagram


DZ

DZ

E

E

E

E

-

Time

Case-Control

Study

Flow Diagram


DZ

DZ

E

E

-

Time

Cross-

Sectional

Study

Flow Diagram


Which came first? Exposure or Outcome?

Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Random Assignment

-

E

E

DZ

DZ

DZ

DZ

Study Design: Trial

Healthy People

Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

-

E

E

DZ

DZ

DZ

DZ

Study Design: Cohort Study

Healthy People

Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

-

DZ

DZ

E

E

E

E

Study Design: Case-Control Study

Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

-

DZ

DZ

E

E

Study Design: Cross-Sectional Study



Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Random Assignment

-

E

E

DZ

DZ

DZ

DZ

Study Design: Trial

Healthy People

Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

-

E

E

DZ

DZ

DZ

DZ


Healthy People

Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

-

DZ

DZ

E

E

E

E


Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

-

DZ

DZ

E

E


1st 2nd

3rd 4th


Which came first?

Exposure Outcome



Scenario 3

Watching TV

No Watching TV

Overweight

No Overweight

Watching TV Overweight


b

d

a

c

Risks

Count

CompareDivide

Over-weight

No Over-weight

TV

No TV

23 9 or 71.8%32

23

7 15

32

22 or 31.8%22

7

Relative Risk

= 2.3

Total

Count, Divide, and Compare (CDC)

Is there an association?

Did watching TV and being overweight turn up together?


Watching TV

Time Order

Watching TV

No Watching TV

Overweight

No Overweight

Overweight

Watching TV

causes

people to be

overweight.


Watching TV

No Watching TV

Overweight

No Overweight

Watching TV OverweightX

Time Order

Watching TV

causes

people to be

overweight.


Watching TV

No Watching TV

Overweight

No Overweight

Watching TV OverweightX

Watching TV

causes

people to be

overweight.

Time Order


It is possible that people who __________ are more likely to ____________.

But it is also possible that people who are __________

are more likely to __________.

Reversed Time Order

watch TV

be overweight

watch TV

overweight


Which came first?




Watching TV causes weight gain.

Watching TV

No Watching TV

Overweight

No Overweight


Cigarette Smoking

No Cigarette Smoking

Depression

No Depression

Scenario 4

Cigarette Smoking

Depression


b

d

a

c

Count

CompareDivide

38 83

59 196

121

255

121

Risks

or 31.4%38

or 23.1%255

59

Relative Risk

= 1.4

Depress-ion

No

Depress-ion

Cigarette Smoking

No

Cigarette Smoking

Total

Is there an association?

Did cigarette smoking and depression turn up together?

Count, Divide, and Compare (CDC)


Watching TV causes weight gain.

Watching TV

No Watching TV

Overweight

No Overweight


Cigarette Smoking


Depression

No Depression

Time Order

Cigarette Smoking

Depression

Cigarette

smoking

causes

depression.


Watching TV

No Watching TV

Overweight

No Overweight


Cigarette Smoking


Depression

No Depression

Cigarette Smoking

DepressionX

Cigarette

smoking

causes

depression.

Time Order


It is possible that people who _____________

are more likely to ____________.

But it is also possible that people who are __________

are more likely to ______________.

Review - Reversed Time Order

smoke cigarettes

be depressed

smoke cigarettes

depressed


Which came first?

Cigarette Smoking

Depression



Family Meals

No Family Meals

No Substance

Use

Substance Use

Frequent Family Meals and Lack of Substance Use

Family MealsNo

Substance Use


“Having family

meals often

may well

decrease

substance use”

Family MealsNo Substance

Usex

Family Meals

No Family Meals

No Substance

Use

Substance Use



It could be that …

using substances

may lead to

avoidance of family

meals . . .

Family MealsNo Substance

Usex

Family Meals

No Family Meals

No Substance

Use

Substance Use



Which came first?

Family Meals Substance Use



It is possible that people who ________________

are more likely to __________________.

But it is also possible that people who

_______________ are more likely to _______________.

Review - Reversed Time Order

attend family meals

not abuse substances

avoid family meals

use substances


Reverse Time Order

How can I figure out if the association is

due to reverse time order?



Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Random Assignment

-

E

E

DZ

DZ

DZ

DZ

Study Design: Trial

Healthy People

Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

-

E

E

DZ

DZ

DZ

DZ


Healthy People

Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

-

DZ

DZ

E

E

E

E


Time

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

-

DZ

DZ

E

E



ME Eisenberg, D Neumark-Sztainer, JA Fulkerson, M Story. Family meals and substance use: Is

there a long-term protective association? Journal of Adolescent Health 43

(2008) 151-156

A Study to Test Time Order


ME Eisenberg, D Neumark-Sztainer, JA Fulkerson, M Story. Family meals and substance use:

Is there a long-term protective association? Journal of Adolescent Health 43 (2008) 151-156

A Study to Test Time Order

“Most studies regarding health benefits of family meals have used a cross-sectional design . . .”

“Cross-sectional studies . . . lay important groundwork but raise the question of reverse causality . . .”

“The present study therefore builds upon or earlier cross-sectional research . . .”

“We examine family meal patterns over 5 years and their association with initiation of substance use as young people transition from middle school to high school age.”

“We hypothesize that regular family meals during early adolescence will be protective against substance use in middle adolescence, 5 years later.”


E

E

DZ

DZ

DZ

DZ

Healthy People

-

Time

Cohort

Study

Flow Diagram

Yes/No 5 or more

family meals per week

Yes/No Development of substance use

Teenagers substance use at start

of study


Abstract of Journal Report on Study

Purpose: To examine 5-year longitudinal associations between family meal patterns and subsequent substance use in adolescents.

Methods: A total of 806 Minnesota adolescents were surveyed in public schools in 1998–1999 (mean age, 12.8 years) and again by mail in 2003–2004 (mean age, 17.2 years) . . . [We] estimated] the odds of tobacco, alcohol, and marijuana use at follow-up for adolescents reporting regular family meals at baseline compared with those without regular family meals, adjusting for family connectedness and prior substance use.

Results: Family meal frequency at baseline was associated with significantly lower odds of cigarette smoking, alcohol use, and marijuana use at follow-up among female adolescents, even after adjusting for baseline substance use and additional covariates. Family meals were not associated with use of any substance at follow-up for male adolescents after adjusting for baseline use.

Conclusions: Results from this study suggest that regular family meals in adolescence may have a long-term protective association with the development of substance use over 5 years among females. Parents should be encouraged to establish a pattern of regular family meals, as this activity may have long lasting benefits


Detail about Study Findings

Gender / Exposure

Relative Odds (95% Confidence Interval - CI)

* **

Say it in English

Girls

Cigarettes

Alcohol

Marijuana

0.47 (0.29-0.81)

0.49 (0.29 - 0.83)

0.49 (0.26 - 0.93)

The odds of using _____________

are ______ times as

likely for those who have

5 or more family

meals per week

as for those who do not

Boys

Cigarettes

Alcohol

Marijuana

0.67 (0.39 - 1.21)

0.91 (0.53 - 1.55)

0.98 (0.54 - 1.76)

* Analyses account for “family connectedness,” race, socioeconomic status and baseline substance use

** 95% CI means that 95 of 100 studies, this CI contains the true odds ratio


Interpretation

Why is the protective effect seen only in girls?

Eisenberg in article: “This study's findings also are in line with our previous findings suggesting that family meals may be associated with more benefits for females than for males. Although measures were not available to explore reasons behind this difference in the current study, existing theory and research point to differences in the way that males and females interact with family members in a variety of domains. Females may, for example, be more attuned to subtle emotional support offered during family meals, resulting in a more profound protective effect for females than for males.”


Exposure Outcome

If an association was found due to reversed time-order, ….found due to reverse time order, ….

Public Health Significance of Reverse Time Order

X… and you avoided or eliminated the hypothesized

cause, what would happen to the outcome?

?


Big Ideas in this Lesson (4-6)

• Association does not necessarily mean causation

• When an association is found, several possible

explanations must be considered, including the

possibility that the association is due to reverse

time order

Re-Cap


Next Lesson

Amicus Curiae

Friend of the Court

drugepi 6 - reverse time order module 4 overview context content area: interpretation of...

Documents