Quiz 2 Flashcards
(100 cards)
1
Q
What did the Seychelle child development study want to look for? What did they find instead?
A
Researchers were looking for association between pre-natal mother mercury exposure and children development
Found no association after 6 months
66 months: Benefits were found between mercury and developmental outcomes (probably due to benefits of FISH (omega 3s) outweighing the risks of MERCURY.)
2
Q
Why did populations who eat +++ pilot whales had detrimental effects of mercury?
A
Because selenium is found in lower quantities in pilot whales.
3
Q
What are the 4 types of fish/shellfish most consumed in Canada and US?
A
- Shrimps
- Salmon
- Tuna
- Sardines
Main message: Knowing where the data is coming from is key.
4
Q
Who is at risk of +++ mercury?
A
Pregnant women, children, people who rely on fish as their main source of nutrition (around 1 M people worldwide)
5
Q
Where is most of the research evidence coming from?
A
High-income countries (Canada, US, europe…). Results may not apply to lower income countries
6
Q
Which nutrients in fish make it good for you?
A
Protein, selenium, omega 3s (LCFAs), Vitamin D, iodine
7
Q
What makes fish healthy?
A
Not only the nutrients but the act of fishing (relaxation) and fight against FI. Not just about the molecules inside the fish but what fish means to people
8
Q
What are the main food sources of omega 3s?
A
Nuts and seeds
Fish
Oils
9
Q
What are the causal links between omega 3s and IQ?
A
omega 3s (molecule) first affect the cell which comprise tissues, which comprise organs, which comprise individuals and a group of individuals make up the population
10
Q
What makes fish bad for you?
A
Mercury, bacteria, PCBs, mislabelled fish
Sustainability aspect of overfishing?
11
Q
Name 2 events in which mercury was a problem
A
Japan 1940s to 1950s - industry fed the bay with mercury... belief that "Dilution is a solution to pollution" Grassy Narrows (first nation reserve) was fishing from a water full of mercury from industry
12
Q
What proportion of women of childbearing age have too much mercury in their body?
A
1 in 6 women of childbearing age have mercury levels in their body that might be high enough to cause effect in a child
13
Q
Explain the example of mercury and IQ levels on individuals vs. the population
A
If mercury in fish makes our IG go down by 5 points, we couldn’t be able to tell individually
If it does that for the whole population, we will notice: the 5% that is gifted with a high IQ will half
Why would you be able to tell on a population level? Reduce human potential, will put strain on society. Little changes matter.
14
Q
Differentiate qualitative vs quantitative variables
A
Qualitative variables: Properties that vary in a type of attribute (Ex. Sex, religion, eye color, marital status, type of fish consumed)
Quantitative: Properties that differ in an amount (Ex. height, weight, BP, Number of servings of fish per week)
15
Q
Differentiate discrete vs continuous variables
A
Discrete: Quantitative adjacent variable in which no intermediate values are possible (E.g. number of children)
Continuous: Intermediate values are possible in between two adjacent scale values (E.g. height, weight, BP)
But in practice, these are usually converted into discrete variables
16
Q
Differentiate between independent and dependant variables
A
Independent variable (X): Presumed causal factor in a cause-effect relationship. Exposure, intervention, treatment, diet…
In an experiment, this is what you manipulate (X-axis)
Dependent variable (Y) – outcome; Presumed effect in a cause-effect relationship; “depends” upon the independent variable
In an experiment, this is the outcome you measure (Y axis)
17
Q
What is a variable?
A
A factor or attribute that can assume two or more values
18
Q
What is a domain? What is the smallest possible domain?
A
All possible values a variable can have
Smallest domain is 2 (dichotomous or binary)
19
Q
What is a mediator variable? Provide examples for Fish consumption –> IQ.
A
Variable to explain why X and Y are liked… Provides the CAUSAL link
Fish consumption –> Omega 3 content, protein content, selenium content, iodine content –> IQ (+)
Omega 3 –> neural cells –> increased neural function/brain health etc…
Fish –> omega 3 –> cell –> tissue –> individual –> IQ
20
Q
What is a moderating variable? Provide examples for fish consumption –> IQ.
A
(X –> Y) * moderator (M)
Variable that alters the strength or direction or the relationship (Aka interaction terms)
For fish: E.g. health status, genetics, sex, age, life stage
A moderator can also be a confounder
21
Q
What is a confounding variable? Give an example for fish consumption and IQ
A
Variable (C) that correlates with both X and Y; Not in the causal pathway but causes confusion in the true relationship
E.g. education level (more education = more money to get fish, more education = more knowledge about benefits of fish, …)
22
Q
Suppose there is a statistical relationship between ice cream consumption and drowning
What could be a mediator variable?
What could be a confounding variable?
A
Mediator: none
Confounding: Temperature, age
23
Q
Discuss the transfer of conceptual variables into operational.
A
Scientists need to be explicit on how they define their variables
Conceptual variables are the “ideas”
Must provide operational definitions AND how to measure these (method? Valid?)
Scientists can disagree on how to define certain variables (IQ? CVD?)
Disagreements can be frustrating… Can lead to poor decisions… may lead to debate and progress
24
Q
What are the 4 Stevens Scales of measurement?
A
Nominal, ordinal, interval, ratio
25
Define the nominal scale and give examples.
Qualitative differences in scale value
e.g. sex, politics, college major, food categories, hair color...
Cannot be ordered
26
Discuss the accuracy of food labels study
• Covered under US code of federal regulations
• Scientists measured energy content in 24 snack foods purchased from a single grocery in phoenix on 3 occasions in 2010
• One serving size was based on the label, and then energy content measured via bomb calorimetry
Among a convenience sample of energy‐dense snack foods, caloric content is higher than stated on the nutrition labels, but overall well within FDA limits. This discrepancy may be explained by inaccurate carbohydrate content and serving size.
27
Define the ordinal scale and give examples.
Different scale values represent relative differences and the order matters
e.g. likert scales, apgard score, rank ice cream preferences
> or
28
Define the interval scale and give examples.
Scale values have an equal and exact distance
e.g. temperature, time of day, calendar year
+ or -
29
Define the ratio scale and give examples.
Interval but has an absolute zero
e.g. length, annual income, caloric intake, pregnancies, money
x or ÷
30
Define accuracy, and what bad accuracy can be attributed to.
The degree to which the measurement yields results that agree with the truth. How do you know the measurement is valid? Usually not. Measurements are proxies.
Can be d/t systemic error in the instrument, calibrating error, placement error...
Able to reproduce results?
31
Define reliability.
Precision. The degree to which the measurement yields a consistent value.
Measures what its supposed to measure.
32
Define what a population is.
All cases of observations of interest (universe)
33
Define what a sample is.
A subset of cases or observations
34
Define what a sampling frame is.
A list from which samples will be selected
| How can you get access to the study population?
35
Explain what a parameter is
A characteristic of the population
36
What is a statistic?
Characteristic of the sample
37
What is the Canadian Health Measure Survey (CHMS) and what are its goals?
“aims to collect important health information through ... interview ... blood and urine”
Samples 3-79 yr olds in a nationally representative manner; approx. 5,000- 6,000/cycle
2-yr cycles, started 2007-2009
Led by Statistics Canada, with Health Canada and the Public Health Agency of Canada
Direct physical measures, biospecimens
38
What is the NHANES survey and what are its aims?
• National Health and Nutrition Examination Survey
• “assess health and nutritional status”
• 1+yr old in a nationally representative manner; approx. 5,000-9,000/cycle
• Annual survey since 1971
• Blood Hg since 1999 (in children/female), and
2003-onwards in all participants
• Urine Hg also; Hg speciation also
• Led by National Center for Health Statistics
39
Why does the CHMS biomonitor?
* Establish reference ranges to enable deeper study of sub-populations or across countries
* Establish baseline levels to assess over-time changes
* Help set priorities and take action
* Assess effectiveness of regulatory decisions
* Support research on links between exposure and health effects
40
Define what a bias is in research.
It is a form of systematic error in the research process that consistently shifts results in one direction.
41
What are the 3 main phases of research?
1. Planning
2. Data collection and intervention
3. Data analysis and publication
42
What are the 2 possible types of bias during the research planning phase?
Design bias
| Selection bias
43
What are the 5 possible types of bias during the data collection and intervention phase?
```
Measure bias
Interviewer bias
Recall bias
Compliance bias
Attrition bias
```
44
What are the 2 possible types of bias during the data analysis and publication phase?
Confounding bias
| Publication bias
45
Explain what a design bias is.
| In which phase of research is it?
Using a flawed study design for your research question (e.g. cohort, cross-sectional, case control, case report/series, RCT...)
Part of the planning phase.
46
Explain what a selection bias is.
| In which phase of research is it?
```
Occurs during identification of study population, when some individuals are more likely to be selected for the study than others.
Part of the planning phase
Includes:
Self-selection (volunteer-based)
Convenient sample
Non-response
```
47
Explain what a cross-sectional analysis is.
Observational study; measures the distribution of a characteristic in a population or sample at a certain point in time, it is like a picture of the situation of interest.
o Enroll participants and measure exposures and health simultaneously
o Snapshot in time
o Study at one point in time to determine if there is an association between exposure and health
48
Explain what a cohort study is.
Longitudinal: Follows a defined group of people (the cohort) over time. Outcomes observed in subsets of the cohort who were exposed to a particular factor are compared to outcomes in those not exposed to a particular factor.
o Characterize (observe) exposures in participants and then track them over time
o Compare exposed vs. control
o Exposure --> outcome – start with exposure and then follow group over time; compare between exposed and unexposed
o First establish X and then f/u for Y
49
Explain what a case-control study is.
Longitudinal; compares people with a specific disease or outcome of interest (cases) to people without the disease or outcome (controls) to find associations between the outcome and prior exposure to particular risk factors.
o Enroll participants with disease or outcome and find matched controls (same characteristics but without outcomes)
o Compare previous exposures between groups
o Outcome --> exposure – start with an outcome/disease and then work backwards to identify possible causes
o Retrospective (opposite of cohort)
* Find participants with Y and try to establish what X was
50
What is a RCT?
Experimental; demonstrates if a treatment is beneficial or not, by randomly allocating participants into an experimental group or a control group and follows them over time for the outcomes of interest. A high quality randomized controlled trial exhibits the following characteristics: random allocation, blinding, intention-to-treat analysis, adequate statistical power, adequate follow-up (>80%).
51
What would be the appropriate study design for
- Prevalence
- Incidence
- Cause
- Prognosis
- Treatment effect
1. Cross sectional
2. Cohort
3. Cohort, case-control, cross sectional (in order of reliability)
4. Cohort
5. Controlled trial
52
What are the 3 types of selection bias? Explain them.
1. Self-selection (volunteer participation)
2. Convenient sample (People are referred)
3. Non-response (Those who do not respond/participate often have different characteristics from those who respond/participate)
53
Same one way to avoid self-selection and convenient sample bias.
Random selection of the sample: Each individual is chosen entirely by chance and each member of the study population has an equal chance of being included in the sample
54
Name 3 strategies to avoid non-response bias.
- Provide incentives for completing survey
- Explain why survey is important
- Keep survey short
55
After what % of non-response rate does a study lose validity and reliability?
> 20%
56
Explain what a measure bias is.
| In which phase of research is it?
Occurs when the measurement tool is not sensitive enough to detect the exposure or outcome of interest
Part of data collection and intervention phase
57
Explain what a interviewer bias is.
| In which phase of research is it?
Occurs when there is a systemic difference between how information is asked, recorded, or interpreted. Interviewer may probe for certain risk factors.
Part of data collection and intervention phase
58
Explain what a recall bias is.
| In which phase of research is it?
Occurs when measuring the outcome or the exposure requires individuals to remember past events.
Part of data collection and intervention phase
59
Explain what a compliance bias is.
| In which phase of research is it?
Occurs when some subjects do not adhere to the planned treatment (experimental studies)
Part of data collection and intervention phase
60
Explain what an attrition bias is.
| In which phase of research is it?
Occurs when subjects who leave the study (drop-outs) differ significantly from those who remain (common in cohorts and RCTs)
Loss of > 20% of sample makes study unreliable and not valid.
Part of data collection and intervention phase
61
Name 2 strategies to avoid measure bias
* Using validated measurement tools that have a high sensitivity and specificity to minimize bias
* Look at validation studies (validate certain tools of measurement)
62
Name 3 strategies to avoid interviewer bias
* Standardizing protocols for data collection
* Training interviewers
* Blinding the interviewers to the study hypothesis
63
Name 3 strategies to avoid recall bias
• Using a well-structured questionnaire that has been validated against direct measures
• Giving the participants enough time to reflect before answering
• Verifying the reported data by using a reference criterion (e.g. medical records, data from another family member
Recall bias hard to eliminate in nutrition studies
64
Name 3 strategies to avoid compliance bias.
* Designing protocols that are simple and easy to follow and comply with
* Identify subjects who are unable or unmotivated to comply (with the intervention) during the enrollment process
* Contacting subjects frequently to maintain interest and motivation
65
Name 3 strategies to avoid attrition bias.
* Using protocols to contact participants by email, post, telephone, etc. (keep in touch)
* Provide incentives (monetary, gifts)
* Keeping the interviews short
66
Explain what a confounding bias is.
| In which phase of research is it?
Occurs when an estimated association between exposure and outcome is biased by a third factor associated with both exposure and outcome (confounding variable)
Part of the data analysis and publication phase
67
Explain what a publication bias is.
| In which phase of research is it?
Occurs when researchers and research sponsors are unwilling to publish unfavorable results
Positive results are more likely to be reported and accepted for publication in scientific journals than negative results
Part of the data analysis and publication phase
68
Name 4 strategies to avoid confounding bias.
* Identify confounders prior to the study so confounder can be measured during data collection
* Identified confounders can be controlled with study design (case control or randomization) and during data analysis (regression analysis)
* Unidentified confounders can only be controlled with randomization during participants selection and enrollment
* Confounders that cannot all be controlled for can be minimized by using a very large sample size (effect will be mitigated)
69
Explain the steps of the scientific method
1. Make observations
2. Think of interesting questions
3. Formulate hypotheses
4. Develop testable predictions
5. Gather data to test predictions
6. Refine, alter, expand, or reject hypotheses
7. develop general theories
70
Differentiate experimental from observational studies.
Experimental: Researcher intervenes to change exposure then observes/measures what happens (CAUSATION – more powerful)
Observational: Cannot control the exposure; researcher observes/measures exposure and outcome (does not manipulate)
Environment is not changed
Not a “true” experiment
(ASSOCIATION/CORRELATION – weaker)
71
Differentiate descriptive studies from analytical studies
Both are included in observational studies
```
Descriptive study: Help identify patterns and generate hypothesis. Case report, surveys, fact-finding inquiries
Frequency and distribution of single variables. No comparison group.
(E.g. see what kind of fish class eats with a survey --> gives information)
```
Analytical study
Comparison between groups
Aims to associate exposure/intervention and outcome (E.g. link what kind of fish class eats to BP)
72
What are the 3 types of analytical studies?
Cohort, case-control and cross-sectional
73
What are the pros and cons of cohort studies?
Pros: Can match subjects at baseline; establish timing of events; can standardize exposure
Cons: Associative study (not causal); blinding is difficult; not randomized; rare diseases need large sample size and long follow-up (attrition rate)
74
What are the pros and cons of case-control studies?
Pros: Relatively quick and cheap (do not need to f/u for a long time); most feasible for rare disorders or outcomes with long latency period; fewer subjects needed than other studies
Cons: Associative study (not causal); reliance on recall and records to determine exposures (past diet??); not randomized; control groups hard to find, selection bias (Selection bias: What motivates someone with a rare disease to participate to the study? (self-selection bias); may result in confounding factor or other unexplained variable that drives the observation/association)
75
Explain the MMR vaccine controversy
A controversy surrounding the combined measles, mumps, and rubella (MMR) vaccine started with the 1998 publication of a fraudulent research paper in The Lancet linking the vaccine to colitis and autism spectrum disorders. Following the initial claims in 1998, multiple large epidemiological studies were undertaken, which all found no correlation between vaccination and autism. The author of the paper was found to have multiple undeclared conflicts of interest, had manipulated evidence, and had broken other ethical codes.
Mercury is found in all vaccines for preservation purposes
76
What were the consequences of the MMR vaccines controversy?
Children stopped being vaccinated = diseases that had been eradicated came back (measles and mumps)
$$$ was spent for researching more on this and there was nothing
77
What are the pros and cons of cross-sectional studies?
o Pros: relatively cheap and simple
o Cons: Associative study (not causal); recall bias problems; confounders may be unequally distributed (might be other variables outside of the one you are measuring that may be included)
78
What are case reports?
• An in-depth analysis of a single individual, unit, or event (isolated events)
• Direct observation and questioning
• Provide considerable detail
• Usually concern rare events
Challenges? A) difficult to draw clear causal linkages; B) generalizability of findings; C) potential for observer bias
79
What is a STROBE checklist and why was is made?
STOBE = Strengthening the Reporting of OBservational studies in Epidemiology
A checklist of items that should be included in reports of observational studies. You must report the page number in your manuscript where you consider each of the items listed in this checklist.
Was made d/t the “Incomplete and inadequate reporting of research hampers the assessment of the strengths and weaknesses of the studies reported in the medical literature. Readers need to know what was planned (and what was not), what was done, what was found, and what the results mean. Recommendations on the reporting of studies that are endorsed by leading medical journals can improve the quality of reporting.”
80
What are the pros and cons of RCTs?
o Pros: Unbiased distribution of potential confounders, blinding, statistically robust
o Cons: Expensive (time, $), volunteer bias, ethical concerns (risk that the cure will work; why are some people getting it and others not?)
81
What is a randomized controlled trial?
o “Classic experiment”: one that uses animal model; have control over everything.
o With humans, difficult to manipulate the exposure
o Normally used to test new drugs and treatments
o Participants who meet inclusion criteria are randomly allocated to either receive treatment or nor, and then followed over time
o Random allocation is completely blinded (to participant and researcher)
How do we double-blind a study? Intermediary gives unique identifier to the researchers and will be the only ones to hold the clues
82
What are CONSORT checklists used for, and what are they?
Consolidated Standards of Reporting Trials
Evidence-based, minimum set of recommendations for reporting randomized trials
25-point checklist
83
What is a quasi-experiment?
A non-randomized experiment
Between cohort and RCT
Intervention exists but not planned before starting research, and thus random allocation not involved
Researcher will assign treatment based on a criteria (e.g., cost, feasibility, convenience, cut-off score) and not randomly
Subject to concerns over internal validity (control and exposed groups may not be comparable at baseline)
84
Name the types of studies in order of less important to most important in que evidence pyramid
In vitro < animal < editorials, opinions and ideas < case reports
85
Describe the Bradfort Hill criteria for causation. (9)
A set of nine criteria to provide epidemiologic evidence of a causal relationship between a presumed cause and an observed effect.
1. Strength (effect size)
2. Consistency – is it reproducible
3. Specificity – specific population, specific site, etc.
4. Temporality – cause and then effect
5. Biological gradient – greater exposure = greater incidence
6. Plausibility – a believable mechanism linking
7. Coherence – epid and lab studies
8. Experiment
9. Analogy -- The effect of similar factors may be considered.
86
What are the main uses for biomarkers? (3)
validate dietary survey instruments
Be a surrogate of dietary intake
Integrate nutritional status for an element
87
What are the 4 major things to capture in a survey?
1. Frequency
2. Magnitude
3. Type
4. Duration
88
Describe general steps in developing a survey
1) Reflect upon your research question and convert them into specific topic
- General goal is to characterize fish consumption, then specific topics could be: Amount of fish consumed, Diversity/types of fish consumed, Servings of fish consumed
2) For each topic, identify variables of interest
- YES/NO of fish consumption?
- If yes, # of servings per unit time... etc.
3) Trial your survey
Is it obtaining data for you to help answer your question? Do the items really measure your target variable?
Have you identified practical limitations (length, readability)?
Are instructions clear? Obtain feedback
89
What are the 2 main types of questions?
Open-ended and closed-ended
90
What is an open-ended question?
* People respond on their own terms vs. specific options
* Maximize freedom of response
* May be useful in initial stages of research (background research)
* May be difficult (e.g., how analyze results?) and time-consuming
* Maybe can’t read writing?
91
What is a closed-ended question?
* Responses already coded (data entry and analyses)
* Relevant options given, but these may also be limiting (e.g., a respondent does not agree with the choices)
* Need “don’t know” or “other” category
* Multiple choice (one choice allowed)
* Multiple choice (multiple responses allowed)
* Ranking scale (1-10)
* Forced choice (A or B)
* Likert scale (strongly agree, agree, neutral disagree, strongly disagree)
92
What could be some issues with the wording of questions?
Beware of bias in the wording of questions
Leading questions can bias
Loaded questions can bias
Double-barrel questions (e.g. including ideas of 2 different opinions and make them in one question)
Double negative questions can confuse
93
What are the pros and cons of face-to-face interviews?
Pros: Achieve higher response rates, can establish personal rapport, interviewer can provide quality and interviewer can gauge verbal/facial cues
Cons: Costly (trained interviewers, transportation, etc.), interviewer bias may exist (may not be conscious), and interviewer characteristics (e.g. age, sex, ethnicity) may bias respondent.
94
What is a biomarker and what 3 forms can they take?
“biological specimen that is an indicator of nutritional status with respect to intake or metabolism of dietary constituents.” - may be from urine, saliva, fingernails, blood, hair...
1. Biomarker of exposure
2. Biomarker of health status
3. Biomarker of sensitivity or susceptibility
95
What are biomarkers for...
1. Protein intake?
2. Selenium intake?
3. Vitamin D?
4. DHA/EPA?
1. Urinary N
2. Urinary Se
3. Serum 25(OH)D
4. DHA/EPA in FAs
96
What can you do to overcome a measurement that you know is valid but not reliable?
Do the measurement many times
97
What can you do to overcome a measurement that you know is reliable but not valid?
Adjust for the skewness
98
Differentiate internal validity from external validity
```
External validity: looking at broader knowledge out there. Looking at other research that has had very similar questions and variables to give yourself confidence that this association is legitimate. looking outside. Is your hypothesis/ are your results generalizable based on other studies?
Internal validity: looking inside, did the researchers identify confounders, how do they measure these things, are they accurate, using gold standard proven measurements? Once you have an idea of the quality of the measurements and also whether they controlled for other factors (confounder, mediator, moderator). Basically a way to look at the quality of your own data through different perspectives.
Systematic errors - bias, confounding
Random errors (chance)
```
99
Explain what case reports/ case series are.
• Case-report/series: describes unusual observations among one individual or a small group of individuals that received the intervention or exposure, though there is no control group.
o Case report: Case of one individual
o Case series: Group of people that came with same symptoms
100
What is a longitudinal study?
A longitudinal study is a research design that involves repeated observations of the same individuals over short (days) or long periods of time (decades). It can be a type of observational or experimental study.
