Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
380 Cards in this Set
- Front
- Back
Infection
|
Possibility of causing disease
Presence of the microbe in a host to the benefit of the micro-organism. |
|
Colonization
|
Presence of microbes in host but which does NOT result in clinical disease and may be needed for health.
|
|
T/F Bacterial colonization in the body necessarily causes clinical disease
|
False. there are good flora too! think about yogurt
|
|
Latent infection
|
involves persistence of microbe in host with possibility of disease in future without shedding of organism in the interval
|
|
What is an example of a latent infection (this is related to chickenpox)?
|
Example: Shingles from Varicella Zoster
|
|
Some examples of acute disease are ______ Some examples of chronic disease are ______
|
Infections may also be acute (cholera, influenza) or chronic (leprosy, hepatitis B)
|
|
Nonclinical infections
|
inapparent infections which can still cause transmission of a disease.
These can occur at the preclinical (prior to onset of disease) subclinical levels (mild disease, so spreads without being aware of having the disease |
|
Mary Mallon was famous for what disease and how did people respond?
|
She was known as Typhoid Mary. She was a cook who transmitted typhoid fever. She was eventually quarantined
|
|
Describe the typhoid fever carrier state
|
can occur with or without acute symptomatic infection 2-5% of infections More common in women Shed organism for years
|
|
Infectivity
|
The ability of the microbe to cause infection (enter, survive, multiply in host), but not necessarily cause symptoms
|
|
attack rate (for infection) =
|
# infected / # exposed This attack rate (usually based on clinical criteria) is used in epidemic investigation
|
|
Pathogenicity
|
the ability of the microbe to cause disease. …without regard for severity!
|
|
attack rate (for pathogenicity) =
|
# of clinical cases / # infected
|
|
Virulence
|
the degree of severity of disease produced.
|
|
Virulence attack rate =
|
Virulence attack rate = # of fatalities / # of diagnosed cases = to case fatality rate
|
|
Epidemic:
|
a sudden increase in the frequency of infection in a particular region.
|
|
Pandemic:
|
an epidemic on a global scale HIV infection is an example
|
|
Endemic:
|
an infection that occurs regularly at a stable rate in a particular region
|
|
Modes of Transmission Direct
|
Person-to-Person Contact
|
|
Modes of Transmission Indirect
|
Common Vehicle 1.) Single exposure 2.) Multiple exposure 3.) Continuous exposures Vector
|
|
Infections are transmitted from a reservoir
What are 4 common reservoirs? |
Inanimate objects
asymptomatic human symptomatic human Another species |
|
Reservoirs
Two examples of inanimate objects being reservoirs |
Fomite
contaminated water |
|
Reservoirs
Two examples of another species being reservoirs |
Zoonosis
Vector-borne diseases (like mosquitos) |
|
Food borne illnesses.
Two major ones. |
Salmonella
Botulism |
|
Crytosporidiosis is an example of a _______ illness
|
waterbourne
|
|
Airborne illness
Different size aerosols |
Air borne Large particle aerosols
Small particle aerosols or droplet nuclei residual of droplets which have evaporated to < 5 microns - May remain suspended in air for prolonged periods of time |
|
Airborne illnesses
Diseases caused by large particle aerosols |
Bacterial: Hemophilus influenzae, Meningococci Diphtheria, Pertussis, pneumonic plague, Mycoplasma pneumoniae
Viral: - Adenovirus, Mumps, Parvovirus, Rubella |
|
Airborne illnesses diseases caused by small particle aerosols (droplet nuclei)
|
Tuberculosis
Measles Varicella Influenza |
|
Vector borne illnesses
Description |
Vectors are other living organisms which harbor microbe and transmit to susceptible host
|
|
Vector borne illnesses disease examples are:
|
Examples: malaria, dengue, yellow fever
|
|
Fomites. Definition and examples:
|
inanimate objects which are contaminated and spread infection.
Examples: the common cold, cold sores, conjunctivitis, coxsackievirus (hand-foot-mouth disease), croup, E. coli infection, fifth disease (“slap cheek”), Giardia infection, impetigo, influenza, lice, meningitis, pinworms, rotavirus diarrhea, RSV, and strep |
|
Zoönosis
|
infection normally present in vertebrate animals but which can rarely spread and cause disease in humans. (Example: brucellosis)
|
|
Zoonosis
Why aren't we too concerned about zoonoses? |
Person to person transmission is rare. Disease may be mild. Infection usually requires extensive exposure.
|
|
Describe the shape of a point source infection curve.
|
Increase to a peak and then decrease
|
|
Describe the shape of a continuous infection curve
|
Increases to a plateau
|
|
Describe the shape of a propagation infection curve
|
Wavelike appearance with each crest increasing. There is a maximal crest and then a series of decreasing crests
|
|
Statistical inference:
|
drawing conclusions for data using statistical methods to describe and arrange data and to test suitable hypotheses
|
|
Most scientific problems address whether there are __________ or _________ between variables
|
differences, associations
most scientific problems address whether there are differences or associations between variables |
|
"null" hypothesis = ?
"research" hypothesis = ? What are we trying to accomplish in most research studies? |
no difference = “null” hypothesis
a statistically significant difference = “research” hypothesis (also called the alternative hypothesis) - most research studies aim to reject the null hypothesis |
|
null hypothesis:
What does this mean in terms of the predicted effect and the means of the treatment to the placebo? How do represent the association? |
predicted effect does not exist;
that is, the mean response to treatment being tested is equal to the mean response to the placebo in the control group; both responses have a normal distribution with an unknown mean and standard deviation For difference: mean 1 = mean 2 For association: r^2 = 0 |
|
Null hypothesis:
What is the equation in terms of means? Describe what the equation means in words. |
H0: μ1 = μ2
(that is that mean for the intervention group will equal mean for control group) |
|
Research hypothesis:
What is the equation in terms of means? Describe what the equation means in words. |
H1: μ for intervention > μ for control group
|
|
What are the two variables found in studies?
|
Factors (Independent variable)
Outcomes (Dependent variable) |
|
Which of these statistical calculations assume normal distribution?
t-test Chi square ANOVA |
Chi-square and ANOVA
|
|
Type I error:
|
reject the null hypothesis when it’s really true;
goal is to keep Type I error as low as possible |
|
Type II error:
|
say the hypothesis is null when it’s really not
|
|
How is the alpha level (α) related to type I and type II errors?
|
When we pick the alpha level, we set upper limit on probability of making an erroneous decision to reject the null hypothesis (type I error)
|
|
What is the typical alpha value that is chosen for studies?
We are aiming for ______ type I error. |
α = 0.05
We are aiming for low type I error |
|
degrees of freedom (df) =
|
degrees of freedom (df) = total n - 2
total n = total number of participants in the study |
|
Once we have an alpha value, degrees of freedom and a calculated t value,
What do we do to determine whether there is statistical significance? |
use the student’s t distribution table
- look for the t - value that corresponds to the alpha value and the DoF -Statistical Significance if calculated t-value > table t-value |
|
T-Test
What does it mean if: table t-value > calculated t-value? |
we cannot reject the null hypothesis; that is our t is not rare enough and our p value is not small enough
|
|
p-value:
Is the probability that... |
probability that difference between groups during an experiment happened by chance or sampling error;
is the probability (0 to 1.0) of the sample statistics, given the samples size and assuming the sample was derived from a population in which the null hypothesis is true |
|
T/F P-value tells you whether or not the results are important
|
False:
p-value does not tell you whether the null hypothesis is true or not but instead it is a probability- it tells you if your t is rare |
|
What is the equation for power?
What does this means in terms of the two types of errors? |
power = 1-β and is the probability of rejecting the null hypothesis when it’s false
|
|
How can we increase power?
|
- you can increase the power by increasing α (making it less stringent), making more room for error to capture more t’s but at the same time increasing chances of making a Type I error
|
|
Increased power leads to an increased chance of making a _______ error.
|
Type I error
|
|
T/F Virtually any study can be shown to be significant by using a big enough sample size
|
True
|
|
We can increase the power by increasing the ______ because small samples have “flukiness” and more sampling error which equals _______
|
sample size,
larger p value |
|
What is the purpose of a standard error as it relates to the sample statistic?
|
standard error is related to the precision of sample statistic- we can express our confidence in our mean by calculating standard error
|
|
SE gets _____ as the SD increases and _____ as n increases
|
larger
smaller |
|
SE
|
SE = SD/ sq. rt (n)
|
|
What does a large SE of the means indicate?
|
a large SE of means says that there’s a lot of variability between sample means;
that is, we don’t have confidence that our estimate (mean) equals the population |
|
As n increases what happens to the means and the SE?
|
as n is increased, we get more and more stable means (smaller and smaller SE)
|
|
As n increases what happens to the probability that you will get a statistically significant mean?
|
Increases
|
|
What is the value of calculating power prior to the start of the study?
|
Knowing the power allows you to pick the proper n for the study
|
|
Confidence intervals
If the confidence interval for a risk interval crosses one, is there statistical significance? |
No
- if the null hypothesis is that the odds ratio = 1 (no difference), then a 95% CI range that includes 1 means no statistical significance while a 95% CI range that does not include 1 means statistical significance |
|
Confidence Intervals
If the null hypothesis is that the difference in the means is 0, what must be true about the CI if we want to reject the null hypothesis of no difference? |
The CI cannot include 0
|
|
What is the difference between clinical and statistical significance?
|
it’s important to note the difference between statistical significance and clinical significance; to determine clinical significance, we want to know if the effect (shown to be statistically significant) is enough to make a real-world/practical difference
|
|
What measure is typically used to determine clinical significance?
|
effect size:
the observed and standardized measure of the magnitude of the observed effect |
|
Effect size:
|
the observed and standardized measure of the magnitude of the observed effect
|
|
Cohen’s d (an effect size) =
How do we interpret the Cohen's d value? |
(mean1 – mean2)/SD pooled
- an effect size of 0.2 is a small effect - 0.5 a medium effect - 0.8 or higher a large effect |
|
Cohen's d, small effect value
|
0.2
|
|
Cohen's d, medium effect value
|
0.5
|
|
Cohen's d, large effect value
|
0.8 or higher
|
|
Why is the effect size important for studies?
|
effect sizes help address stability of effect across studies: helps scientists collectively move forward with the business of science
|
|
T/F A result that is rare is necessarily significant
|
False
|
|
T/F A p-value above 0.05 means the research is not valuable.
|
False. It is possible that the sample size was not sufficient.
|
|
T/F Scientists are more likely to put more credibility in a lower p-value
|
True
BUT REMEMBER - don’t base your belief in something because of a low p value |
|
T/F Just because results are statistically significant means that they are important
|
False
|
|
_____ or ______ is responsible for ALL mean differences
|
chance or sampling error is responsible for ALL mean differences.
|
|
weak positive correlation-
|
as x ↑, y ↑ but weak correlation
|
|
strong negative correlation
|
as x ↑, y ↓
|
|
perfect positive correlation
|
for every degree x increases, so does y
|
|
Pearson Correlations
Requirements |
1. random selection of sample
2. normality of traits measured 3. at least interval level measurement 4. similar variation in x and y scores (assume that x and y distributions are normal, not skewed) 5. linear relation between x and y |
|
Describe the range of the Pearson's coefficient
|
Pearson: r value ranges between -1 (a perfect negative relationship) to 0 (no relationship) to +1 (a perfect positive relationship)
|
|
Pearsons Coefficient (r)
(+ or -) 0.8-1.0 |
very strong relation
|
|
Pearsons Coefficient (r)
(+ or -) 0.6 -0.8 |
strong relation
|
|
Pearsons Coefficient (r)
(+ or -) 0.4-0.6 |
moderate relation
|
|
Pearsons Coefficient (r)
(+ or -) 0.2-0.4 |
weak relation
|
|
Pearsons Coefficient (r)
(+ or -) 0.0-0.2 |
little or no relation
|
|
What does r^2 (coefficient of determination) tell you
|
r^2 (coefficient of determination) tells you the percentage of shared variability (the strength of the association in proportion of variation in y explained by x)
- the higher the coefficient of correlation, the more you can explain x by y *remember that correlation does not equal causation |
|
T/F Correlation = Causation
|
False
|
|
T/F Pearson's correlations do not allow us to determine causation
|
True.
** it’s important to remember that Pearson correlations help determine a relationship between 2 variables but does not allow us to determine causation** |
|
Ordinal tests require ________ tests
why? |
nonparametric
data that’s ordinal requires nonparametric tests because with ordinal data, we don’t understand how much one scale point is different from another |
|
T/F Nonparametric tests do not assume a normal distribution
|
True
|
|
Spearman Rank
|
a commonly used nonparametric test is the Spearman Rank correlation which is a nonparametric alternative to Pearson’s r that makes no assumptions about the populations mean or standard deviation
|
|
Data is ordinal and does not follow a normal distribution.
Spearman's or Pearsons? |
Spearman's Rank correlation
|
|
Data follows a normal distribution.
Spearman's or Pearsons? |
Pearsons correlation
|
|
Spearman's Rank correlation
What are the requirements |
1. random selection of sample
2. both distributions of scores are in ordinal form (may have been converted by making continuous variables into categories) 3. linear relation between sets of scores |
|
Spearman's Rank Correlation
What are the steps in determining correlation? |
- computed “obtained” test statistic value (rs) is calculated and then the “critical” value of the test statistic (rcrit) is derived from a table
- rs is compared to rcri and: - if absolute value of rs is greater than rcrit, reject the null hypothesis - if absolute value of rs is smaller than rcrit, then fail to reject the null hypothesis |
|
Chi-Square Test
Requirements: |
1. randomly select samples
2. nominal level of measurement 3. independent cell entries (each participant can only be in 1 cell) 4. no expected cell frequency below 5 (assume 5 incidences in each cell) |
|
Chi-Square Test
Cell frequency 3 Can we run the Chi-Square? |
No.
no expected cell frequency below 5 (assume 5 incidences in each cell) |
|
Assumptions of a Chi-Square test:
|
H0: fo = fe
Ha: fo ≠ fe |
|
Steps in performance of a Chi-Square test:
|
1. then determine degrees of frequency: df = (# of rows-1)(# of columns-1)
2. using df and the α level you choose, get critical chi from a table 3. calculate chi for the data 4. compare chi calculated to chi critical |
|
Chi-Square Test:
What is our interpretation when the chi calculated > chi critical? |
- if chi calculated is larger, then reject the null and conclude that the variables are not independent
|
|
the n you choose depends on 4 other values:
|
1. significance level, α (usually 0.05)
2. desired power (usually 80-95%) [power is the probability of rejecting the null when the null is not true] 3. variability of the population(s), σ’s 4. amount of change, Δ, from H0 (null) that we realistically want to detect |
|
sample size required for a study is increased by what changes in the following?
alpha power standard deviation delta |
smaller alpha
larger power larger standard deviation decreases in delta (the changes we want to detect) |
|
randomization:
|
randomly assigning each participant to a treatment group
|
|
How do we use randomization in clinical trials?
|
Used to reduce bias
used in clinical trials for internal validity by reducing bias (ex: having more people with a similar characteristic in 1 group than another group) |
|
How do we use randomization in interventional studies?
|
Used to average out confounding bias
in intervention studies, we have the opportunity to select which subjects get which treatment- by randomly assigning subjects to treatment groups, we hope to average out the effects of confounding variables |
|
T/F Randomization doesn’t guarantee a perfect balance among confounders;
|
True;
however, for reasonably large n’s, it’s unlikely that serious imbalances will occur |
|
Name two situations where randomization isn't possible
|
when studying the difference between men and women
when randomization would be unethical |
|
HIV Pandemic showed us the ______ systems are important warning systems
|
surveillance
|
|
HIV Pandemic
What is the common cause mode of transmission in males? females? |
males - male-male
female - high risk heterosexual activity |
|
How can the prevalence of HIV be increasing while incidence remains stable?
|
- duration of HIV is increasing due to improved treatment (HIV pts are living longer with the disease)
|
|
The upper respiratory tract is capable of removing particles greater than _____
|
4-5 microns
|
|
Epidemic Curves
Latent Period (Definition) |
time between infection and when it can be spread to others
|
|
Epidemic Curves
Incubation Period (Definition) |
time between infection and when symptoms appear
|
|
Epidemic Curves
What is significant about the timing of the latent period and incubation periods? |
Latent period starts earlier than the Incubation period
This allows the infection to spread to individuals prior to when the symptoms start to appear |
|
Epidemic Curves:
Point Source Infections appear _____ skewed |
Positively
|
|
Epidemic Curves:
The gaps between the peaks on the curve of a propagated infection indicate the _______ |
incubation period
|
|
Epidemic Curves:
Propagated infection refers to ________ |
Transmission of infection from person to person
|
|
Multivariable Models
|
have more than 1 independent variable AND 1 dependent variable
|
|
Multivariable Models
multivariate models (MANOVAs) |
are models with more than 1 independent variable and more than 1 dependent variable
|
|
Multivariable Models
Requirements for multivariable models: |
1. random selection of samples
2. normality of traits in population 3. homogeneity of variance 4. at least 1 independent variable (nominal or ordinal) with 3 or more “levels” (which is the only distinction between multivariable models and t-tests) 5. one dependent variable (interval or ratio level) |
|
Multivariable Models
Regression Models: |
are models with an independent variable that’s either interval or ratio (not nominal) and the dependent variable is also interval or ratio
|
|
ANOVA
|
are defined by the number of independent variables (with 3 or more levels)
|
|
ANOVA
a simple or one-way ANOVA has ________ |
1 independent variable
|
|
ANOVA
factorial ANOVA: |
designs with more than 1 independent variable
|
|
two-way ANOVA
|
has 2 independent vaiables
- ex: dosage groups with 3 levels and age groups with 4 levels- this would be called a 3x4 ANOVA |
|
- three-way ANOVA
|
has 3 independent variables
- ex: 3 levels of dosage, 4 levels of age, and 3 levels of weight- this would be called a 3x4x3 ANOVA |
|
repeated measures ANOVA:
|
where dependent variable is measured multiple times (time is an independent variable- longitudinal)
|
|
analysis of covariance (ANCOVA):
|
where dependent variable is corrected based on covariates (control for impact of potential cofounder)
|
|
How are the t-test and ANOVA related?
|
ANOVA is really just a mathematical extension of the t-test
- t-test has 2 levels of independent variables while ANOVA has more than 2 levels of independent variables |
|
ANOVA is an "omnibus" test. What does this mean?
|
it tells you whether there are any significant differences between ANY of the means, but does not tell you which pairs had significant differences
|
|
What kind of test usually follows an ANOVA?
|
another type of test called a “post hoc” test must be done
|
|
in combining probabilities, if the events A and B are mutually exclusive
A = 0.4 B = 0.3 |
P(A or B) = P (A) + P (B)
= 0.4 + 0.3 = 0.7 |
|
say your chance of surviving a year after a diagnosis of prostatic cancer is 80% and that the chance of surviving 2 years is 60%- given that you’ve survived 1-year, what’s your chance of making it to 2 years?
|
A = “survive 2 years”
B = “survive 1 year” P (A|B) = 0.6/0.8 = 0.75 (75%) |
|
Decision Trees
|
can be described as a graphical aid to keep track of probabilities, both conditional and otherwise
|
|
Independence
When A and B are independent then P (B|A) = ??? Meaning |
P (B|A) = P (B|A*) = P(B)
- that means that the probability of B is the same whether you know that A is true, that A is not true, or you don’t know anything about A at all |
|
Multiplication Rule
A box has 5 black balls and 5 white balls. The chance of picking 2 black balls is |
5/10 * 4/9 = 20/90 = 2/9
second number is 4/9 because there is one less black ball to pick from (4) and consequently one less ball to pick from overall (9) |
|
The responsibility of the health of the US population falls under control of the _________
|
state government
|
|
What organization is responsible for the health of the US population when there are interstate implications for a health situation such as an outbreak?
|
Centers for Disease and Prevention (CDC), which is part of the Department of Health and Human Services
|
|
In an outbreak that effects multiple states, what is the relationship between the states and the CDC?
|
The states report data to and request assistance from the CDC
|
|
Cases can be first noted or identified by _______ or ______
|
astute clinicians
surveillance |
|
What do you call the first case identified by surveillance?
|
The "Index case"
|
|
An index case may represent a _______ that is, when investigated, more cases are found and a need for public health action is determined
|
"sentinel" health event
|
|
Surveillance's main purposes
|
As an early warning system
prompt identification of shared exposures in order to prevent additional cases |
|
The Paradox of Public Health:
|
The problem is that we only care about public health when it fails;
When public health infrastructure is working well, it appears like a waste of resources |
|
The following paragraph describes what epidemiological concept?
Tuberculosis was very well-controlled in the US until the 1970s when government decided to cut back on funding; there was a huge surge of TB outbreaks (esp associated with HIV); then the need for public health initiatives was recognized and in the mid 1990s TB cases declined again |
The Paradox of Public Health
|
|
Surveillance is targeted to the level of ____________ and is an _____ _____ prevention method
|
Surveillance is targeted to the level of populations and is an indirect primary prevention method
|
|
Who generally handles surveillance?
|
State public health agencies
|
|
What are the four main components of Surveillance?
|
1. case reporting (physicians report to state agency)
2. data analysis (done by the state public health agency) 3. communication of results (state agency informs physicians about outbreaks so that they are more vigilant) 4. application of findings (usually by physicians) |
|
Surveillance - who is responsible?
Case reporting |
physicians report to state agency
|
|
Surveillance - who is responsible?
Data analysis |
Done by the state public health agency
|
|
Surveillance - who is responsible?
Communication of results |
state agency informs physicians about outbreaks so that they are more vigilant
|
|
Surveillance - who is responsible?
Application of findings |
usually done by physicians
|
|
What is a brief way to define surveillance?
|
information for action
|
|
Life cycle of Disease Prevention and Control
|
Surveillance -> epidemiological research -> applied research -> prevention methods - (back to)> surveillance
Epidemiological research is also linked to prevention methods |
|
Public Health Surveillance Loop
|
Physicians report to the health agencies who analyze the data and return summaries, interpretations and recommendations to physicians and the general public
|
|
Bioterrorism requires that we have a good _______ _______ _____ for preparedness and good ______ _____
|
Bioterrorism requires that we have a good public health system for preparedness and good surveillance systems
|
|
How are cases in bioterrorism usually reported and what does this mean?
|
syndromic case-reporting
- case reports describe a constellation of symptoms (syndrome) This is done because bioterror agents are rare and therefore diagnosis of a specific disease can be difficult |
|
CDC Category A Diseases/Agents for Bioterrorism:
what are the four major features of these diseases? |
- “high-priority agents include organisms that pose a risk to national security because they:
- can be easily disseminated or transmitted from person to person - result in high mortality rates and have potential for major public health impact might cause public panic and social disruption - require special action for public health preparedness” |
|
CDC Category A Diseases/Agents for Bioterrorism:
what are the 6 Cat. A diseases/agents? ***KNOW*** |
1. anthrax
2. botulism 3. plague 4. smallpox 5. tularemia 6. viral hemorrhagic fevers (both filoviruses like Ebola and Marburg and arenaviruses like Lassa and Machupo) |
|
Important Attributes of Surveillance Systems
|
Simplicity
flexibility Data Quality Timeliness Acceptability Sensitivity Positive Predictive Value Stability Representativeness |
|
Important Attributes of Surveillance Systems
Why is simplicity important? |
a simple form that is easy to complete helps to increase compliance
|
|
Important Attributes of Surveillance Systems
Flexibility |
flexibility: able to respond to changes (esp if there’s social disruption)
|
|
Important Attributes of Surveillance Systems
Reliability |
must be valid, reliable, and standardized (so can compare with past data)
|
|
Important Attributes of Surveillance Systems
Timeliness |
must report quickly, analyze quickly, and get information back to physicians quickly
|
|
Important Attributes of Surveillance Systems
Acceptable |
everyone is willing to participate
|
|
Important Attributes of Surveillance Systems
Do we want low or high sensitivity? |
want high sensitivity (like in screening tests) even at expense of specificity
|
|
Important Attributes of Surveillance Systems
Do we want high or low false positives? why? |
don’t want a lot of false positives because would trigger system for unnecessary reasons and waste money
|
|
Important Attributes of Surveillance Systems
Stability |
esp if there’s social disruption
|
|
Important Attributes of Surveillance Systems
- representativeness |
need to report across various groups in a population (rich, poor, all ages, all races, insured and uninsured, homeless, etc.)
|
|
Types of Surveillance
Describe passive surveillance. What are some features and problems of passive surveillance |
passive: physician, infection control personnel (ICP), or lab reports cases as prescribed by state or local law without prompting
- problem: physicians are highly noncompliant - features: provider-initiated, less complete, less labor, cheap |
|
Types of Surveillance
active or stimulated: Define, compare to passive surveillance, features |
active or stimulated: same as passive but also have regular examination and prompting done to ascertain presence or absence of cases
- results in better compliance - features: Health Department-initiated, more complete, labor intensive, expensive |
|
Types of Surveillance
T/F Active or stimulated surveillance is cheaper than passive surveillance |
False.
Since active or stimulated surveillance often is Health Department-initiated, more complete, and labor intensive, the cost is usually higher than in passive surveillance |
|
Who is required by state law to report Disease under penalty of law?
|
Licensed practitioners and labs
|
|
What is the problem with licensed practioners and labs when it comes to reporting disease?
|
penalty laws are rarely enforced
|
|
T/F Most cases of reportable disease are not reported
|
True
|
|
CSTE and CDC list of reportable Diseases
Category 1 - time to report is ____ report to? |
24hrs
Category 1: reportable to local public health department within 24 hours of diagnosis - ex: botulism |
|
CSTE and CDC list of reportable Diseases
Category 2 - time to report is ____ report to? |
1wk
Category 2: reportable to local public health department within 1 week - ex: cryptosporidiosis |
|
CSTE and CDC list of reportable Diseases
Category 3 - time to report is _____ report to? |
1wk
Category 3: reportable to local WV Bureau of Public Health (State Health Department) within 1 week of diagnosis unless otherwise noted - ex: HIV/AIDS must be reported to the state (not locally) within 30 days |
|
What is true reporting of diseases in WV?
|
- in WV, not all reportable diseases are infectious
- ex: elevated lead, birth defects, cancer, hemophilia, etc. - in WV, reporting is required by law and failure to report results in fines (but this is rarely enforced) - in WV, for some diseases, may need to fill out supplemental forms for CDC and WVBPH |
|
What is the general rule in the reporting of diseases?
|
More severe, rarer diseases are more likely to be reported than less severe, more common diseases
|
|
International health regulations require mandatory declaration of what diseases to WHO?
|
Cholera, plague, and yellow fever
|
|
Steps in Epidemic Investigation
|
1. establish the diagnosis** (most important)- only trust your own eyes to get credible information
2. establish a case definition - a very operational definition which can identify cases easily and quickly, not a definition based on serological studies that take long time to get results 3. is an epidemic occurring? 4. characterize by person, place, and time 5. develop hypotheses regarding spread of disease 6. test hypotheses 7. initiate control measures 8. follow-up |
|
Food-caused illnesses often cause this kind of epidemic.
|
Point Source Epidemic
|
|
Attack rates for determining which foods caused the disease.
What two calculations should be done? |
attack rate for eating x = # of people ill that ate x / total # of people who ate x
attack rate for not eating x = # of ill that did not eat x/ total # of people who did not eat x attack rate for eating x – attack rate for not eating x = difference in attack rates |
|
Attack rates for determining which foods caused the disease
The food with the highest positive difference in attack rates is usually the ______ |
cause of the outbreak
there are some exceptions - make sure that the numbers make sense in the context of the whole table Recall in our example that barbecued chicken and cola had the highest attack rates, but cola was a confounder |
|
What is the equation for the basic reproductive number?
|
Ro = c x p x d
c: number of contacts per unit of time p: probability of transmission with contact d: duration of infectiousness (in units of time) |
|
T/F Ro varies with populations, behavior, time, and other factors
|
True
|
|
T/F Ro only accounts for the first spread of disease and does NOT account for secondary transmissions
|
True, subsequent infections are NOT included in Ro
|
|
If we want to control infections we should try to keep Ro ______
|
below 1
if Ro <1, then 1 person is spreading the disease to less than 1 other person on average |
|
T/F Ro gives us an indication of the period between the spread of disease
|
False
Ro tells us nothing about how long this takes (since the units of time cancel out in calculating Ro) |
|
What is a unique situation where Ro falls below one?
|
Ro usually falls below 1 when the pool of susceptible people “runs out” because they all get the disease
|
|
Describe the characteristics of an epidemic in terms of time when Ro is high
|
if Ro is high, then there’s an explosive spread of the disease which depletes S (the pool of susceptible people) just as quickly so it dies out quickly- shorter epidemic
|
|
Describe the characteristics of an epidemic in terms of time when Ro is low
|
- if Ro is low, then have longer epidemic
|
|
T/F Kid 1 doesn't have any symptoms of a disease. He is unable to pass the disease along to one of his friends
|
False
- it’s important to note that the infectious period proceeds the symptomatic period (this is necessary for successful propagation of the organism) |
|
What are the three main things we can do to control infections?
|
Environmental measures
Prompt investigation and management of outbreaks Vaccinations |
|
What is the #1 thing that you should always remember to do to help curb infection control in the Health Care Setting?
***KNOW*** |
hand-washing is the single most important/effective tool for disease control in the healthcare setting
|
|
What are some ways we can help control infection in the hospital setting?
|
Handwashing
Appropriate attire Appropriate use of antibiotics Reduce use of invasive devices Surveillance systems |
|
T/F Doctors have a high compliance when it comes to handwashing
|
False
|
|
Quarantine
Two major features |
:restriction of activities on the basis of exposure
another important feature of quarantine is early detection of disease |
|
Quarantine
Two major forms |
2 forms:
- absolute or complete: no contact whatsoever - modified: partial restriction in activities - ex: a single man brought SARS to Toronto from Singapore and 100’s became ill; many health workers were put in modified quarantine- they were told to stay at home, wear a mask when they were around their families, and to take their temperatures twice a day (early detection of disease) |
|
Absolute Quarantine
|
No contact whatsoever
|
|
Modified Quarantine
|
modified: partial restriction in activities
|
|
Isolation
|
restriction of activities on the basis of infection until infectious period has passed
- usually specific to mode of transmission |
|
The difference between Isolation and Quarantine is?
|
Isolation - person is infected
Quarantine - person may/may not be infected (they may have been exposed) |
|
The purpose of vaccines it to provide herd immunity. What does this mean?
|
With a high proportion of immune individuals the spread of disease is minimized and the individuals who are susceptible are protected by the immune individuals (are “free riders”)
|
|
T/F Vaccinations are individual-level efforts to keep Ro low
|
FALSE
- vaccinations are population-level efforts to keep Ro low |
|
Disease Eradication
Incidence of the disease Types of disease that can be eradicated |
is the permanent reduction of a disease to 0 incidence
- only infectious diseases can be eradicated |
|
WHO definition of eradication.
|
elimination is reducing a condition to lowest feasible levels
preventive measures are no longer needed usually eradication involves a non-sustainable campaign where lots of money is gathered to eradicate a disease |
|
The only disease to have ever been eradicated is _____ and it was the ___ WHO eradication effort
|
smallpox,
5th |
|
What are some factors that must be considered before attempting to eradicate a disease?
|
- non-human reservoirs make eradication almost impossible
- ex: cholera is impossible to eradicate because its reservoir is water - simple measure (vaccination, filtering water, etc) - disease is easily identified |
|
What characterizes a vaccine that can be used for eradication?
|
- safe for all age groups
- effective - single dose - stable in field conditions (don’t need to refrigerate) - inexpensive |
|
Currently the WHO is trying to eradicate ____
|
polio
|
|
Dr. Martin thinks this horrible disease will be the next to be eradicated.
|
Dracunculiasis (Guinea worm disease)
where a worm found in fresh-water crustaceans is taken in by drinking, the crustacean dies but the worm lives and travels down to the human’s foot where it burrows out |
|
How can we prevent the drinking of the worm that causes dracunculiasis?
|
drinking in this worm is easily prevented: can filter water through cheese cloth which removes the crustacean, can boil water which kills the organism, or can use larvicide to kill
|
|
What are some new emerging infectious diseases?
|
- HIV
- SARS - Ebola |
|
What are some old infectious diseases that are reemerging?
|
TB
antibiotic resistant bacteria |
|
What are three host factors that make us more susceptible to emerging infectious diseases?
|
more immuno-compromised people
international travel behaviors |
|
What are three environmental factors that are making us more susceptible to emerging infectious disease?
|
Competition for land which brings us into closer contact with farm animals
Globalization Breakdown of public health systems - such as reduced vaccination |
|
What are two factors in agents that are increasing our risk of emerging infectious diseases?
|
Development of antibiotic resistance
Ability to mutate |
|
Definition of Injury
|
“damage to an individual due to energy exchange or from acute disruption in the normal body energy process”
- excessive transfer of energy can be mechanical (75%), electrical, chemical, thermal, and radiation - mechanical injury includes motor vehicles (31% of all injuries), firearms (22%), falls (8%), and other forms (13%); the remaining 25% of injuries are results of other types of energy transfer |
|
Injuries
Are they accidents? Are they random? |
are not accidents
are predictable (not random) events that are preventable because they have known risk factors |
|
Injuries
2 classifications |
Intentional
Unintentional |
|
What are some examples of injuries resulting from acute disruption of a normal energy process?
|
Drowning, near drowning
asphyxiation, CO poisonings, ingestions |
|
Injuries were the leading cause of death for what age groups in 1945 and 2004?
|
1945 - ages 1-24
2004 - ages 1-44 |
|
What is striking about the incidence of injury and lifetime cost due to injury when we compare the years 1985 and 2000?
|
Incidence of injury dropped 12%
lifetime cost due to injury has increased 158% |
|
What are some ways we can look at injury cost data?
|
incidence counts and rates
fatal, hospitalized, nonhospitalized lifetime productivity losses medical costs total lifetime costs age and sex body region injured mechanism of injury nature of injury |
|
The 3% decrease every 5 years in injury fatality rate since 1910 indicates what?
|
The drop in injury fatality rate parallels the decreases seen in chronic diseases that were more commonly fatal during the earl 20th century (influenza/pneumonia, TB, gastroenteritis)
|
|
T/F New chronic disease are making the deaths due to chronic disease rise
|
True. AIDS is one example
|
|
Who were the "parents of injury epidemiology"?
|
- the parents of injury epidemiology were William Haddon, who worked for highway safety, and Susan Baker, author of Injury Fact Book
|
|
What are the four steps in the public health approach to injury?
|
1. surveillance
2. risk factor identification 3. intervention evaluation 4. implementation |
|
What is the ultimate goal of the public health approach to injury?
|
to change policy to change human behavior
|
|
What are the 3 phases of injury?
|
pre-injury: control of energy source lost
injury event: injury is transferred to people, causing damage post-injury: regain physiological homeostasis and repair damage |
|
Pre-injury
|
control of energy source is lost
|
|
injury event
|
injury is transferred to people, causing damage
|
|
post-injury
|
regain physiological homeostasis and repair damage
|
|
The host agent and environmental factors determine what in an injury?
|
Determine the loss of control which leads to injury, the extent of injury, and the degree of recovery
|
|
Injury as it pertains to the epidemiological triangle
Agent? Vector? |
Agent is the source of energy that caused the injury
Vector can transmit energy from the agent to the host |
|
What can we do to reduce the likelihood of an injury event?
|
prevent or limit energy buildup
control circumstances of energy use to prevent unintended release modifiy energy transfer to limit damage Improve emergency, definitive, and rehabilitative care to affect recovery |
|
What is Haddon's Matrix?
|
- takes the applied temporal factors into account with the epidemiological triangle
- was originally used to create highway safety measures: safer vehicles, safer roads, safer occupants, significant reductions in vehicle deaths per miles driven - since 1972 (the matrix’s creation), Haddson’s matrix has been applied to both intentional and unintentional injury control |
|
What is the practical use of Haddon's matrix?
|
Haddson’s matrix has been applied to both intentional and unintentional injury control
- useful for planning, resource allocation, and preventive strategy identification |
|
injury control:
|
reducing frequency of injury and/or reducing impact from injury
|
|
injury prevention (primary prevention
|
preventing an injury from occurring in the first place
|
|
acute care (secondary prevention):
|
improving the medical response to an injury to help limit the amount of damage
|
|
rehabilitation:
|
improving the outcome from injury and decreasing disability
- often education, law, and technology are used together to help control and prevent injury |
|
Rehabilitation:
Out of education, law, and technology, which one is the most important? |
- of these 3, technology is the most important but it’s also very expensive and time-consuming to develop
|
|
Injury prevention strategies
education |
reduce unsafe behaviors and promote safe behaviors
|
|
Injury prevention strategies
legislation |
laws and policies to promote safety, remove unsafe products or punish unsafe behaviors
|
|
Injury prevention strategies
engineering |
making the environment and products safer through design and manufacturing improvements; minimize human involvement (thus minimize error)
|
|
What are some past injury control successes?
|
- injury control successes include seat belts, auto safety, roadway design, bike helmets, smoke alarm programs, and graduated license programs
|
|
Challenges to Injury Control
ATV |
- even though bike deaths are decreasing due to access to safer places to ride bikes and to use of helmets, ATV deaths are increasing
|
|
Challenges to Injury Control
Gun Control |
gun control: guns are faster, more efficient, and have more rounds; they are also easy to obtain
*death from firearms is considered a U.S. epidemic |
|
Challenges to Injury Control
Impaired driving |
- 30-35% of automobile fatalities are due to DUI’s
- elderly driving also raises issues- should the elderly have to take a license exam every year? |
|
Role of the Practitioner in Injury Control
What can we do? |
- identify injury problems
- design and evaluate interventions - counseling and education - media interactions - legislative advocacy |
|
ATV
Characteristics |
300-600 lbs; oversized, low-pressure tires; straddle seat; operator must be “rider-active”; handle bar for steering; motorized, gas-powered; increasingly larger engines and speeds up to 100 mph;
|
|
ATV
Intended Use Practical use |
intended for use off-road and on non-paved surfaces
are important part of WV culture: farming, mining, other occupations, hunting, competitive riding, and recreation |
|
T/F More than one third of ATV deaths in the last 2 years in WV are due to DUI
|
True. 35% of ATV deaths in the last 2 years in WV are due to DUI
|
|
T/F It is legal to use an ATV on any road with any center line markings in WV
|
False. It is illegal
|
|
ATV
What kind of intervention is described below? Make ATVs safer, educate on wearing helmet and not driving when drunk, and outlawing the use of ATVs on roads and highways |
Pre-incident (primary prevention)
|
|
ATV
What kind of prevention is described below? Don't carry passengers while riding on an ATV |
incident (secondary prevention)
|
|
T/F The typical passenger victim in an ATV accident is a 12-14 year old female
|
True
|
|
T/F People typically do not wear helmets while riding an ATV
|
True
|
|
Why are children more likely than adults to have a head/neck/spine injury in an ATV accident than adults?
|
Because they are thrown out of the vehicle
|
|
Describe the Public Health Burden of ATV injuries in terms of YPLL
|
YPLL (based on 75, not 65) for males is 37 years and for females is 54 years
(this means the average age of death in males from an ATV injury is 38, while it’s 19 in females) |
|
What makes ATV injuries quite costly to taxpayers?
|
- 1/3 of people injured by ATVs are insured by Medicare or Medicaid, a significant burden to taxpayers
|
|
What are the three objectives for Healthy People 2010 pertaining to ATVs?
|
Pass helmet law for youth and adults
Reduce death rates improve injury surveillance |
|
WV 2007 Strategic Highway Safety Plan
|
- reduce fatalities by 20%
- expand helmet law, require liability insurance - require training - develop crash tracking system |
|
What was the only successful legislation passed in WV on ATV safety?
|
in 2004 legislation was passed pertaining to where ATVs can be ridden, passengers, helmets for kids, training, penalties, and local government authority
|
|
ATV
Who can educate individuals about ATV safety? |
- adults and parents can lead by example and use common sense
community provides DMV safety awareness course, school education, healthcare education, and 4H education parents should supervise their children, ensure helmet use, ensure training, understand and explain the risk, and use common sense (which unfortunately can’t be legislated) |
|
ATV
Who is ultimately responsible for the safety of the ATV rider? |
- the responsibility for careful operation rests almost entirely on the ATV rider
|
|
demography:
|
scientific study of the determinants and consequences of population trends (mortality, fertility, migration)
|
|
Our population has reached 6 billion when has most of the population growth occured?
|
In the past 200 years
|
|
Who brought up the issue of the population explosion?
|
Paul Ehrlich,
a population biologist, wrote The Population Bomb, bringing attention to the issue of the population explosion |
|
What is the equation for the natural increase in population?
|
Natural increase = birth rate - death rate
|
|
Why is the population of the world growing despite the fact that both birth rates and death rates are declining?
|
Death rates are lower than birth rates
|
|
Over 90% of the population increase in the 20th and 21st century is occuring in the ______
|
less-developed world
|
|
What is the Law of 70?
|
if a population is growing at a constant rate of 1% per year, it can be expected to double approximately every 70 years
|
|
If the population is growing at a rate of 0.5% per year what is the doubling time?
|
140 years
|
|
What did Malthus propose about populations in his Essay on the Principle of population?
|
population tended to grow geometrically (exponentially) while the means of subsistence (food) grew only arithmetically (this is now known as the “Malthusian Trap”)
|
|
What are Malthus's postulates?
|
1. food is necessary to the existence of man
2. the passion between the sexes is necessary and will remain nearly constant - the power of population is indefinitely greater than the power in the earth to produce subsistence for man |
|
Malthus's postulates
Since there is tension between the growth of the population and the means of subsistence, _____ would serve as a positive check on population growth |
mortality
|
|
Malthus's postulates
Solutions to the population growth/Subsistence problem |
preventive checks via birth control through
1. abstinence from sex outside marriage 2. later age at marriage |
|
What major period of time allowed for transition from linear to exponential growth of subsistence?
|
Industrial revolution.
|
|
Demographic transition framework
|
demographic transition framework illustrates population growth in terms of discrepancies and changes in 2 crude vital rates: mortality and fertility (it ignores migration)
|
|
reasons for measuring health and ill-health:
|
- disease/injury prevention
- health promotion - health services planning - program evaluation |
|
crude death rate (CDR):
|
CDR = (number of deaths/midyear population) x 100,000
|
|
crude birth rate (CBR):
|
CBR = (number of births/midyear population) x 100,000
|
|
Demographers typically express rates per ______
Epidemiologists typically express rates per _____ |
Demographers: rates/1,000
Epidemiologists: rates/100,000 |
|
What is true when we compare Crude Birth Rate and Crude Death Rate?
|
the CBR is more “crude” than the CDR because we are all at risk of dying but we are not all at risk of giving birth
|
|
Demographic Transition
pretransition stage: |
pretransition stage: both birth and death rates are high but birth rate is constant while death rate fluctuates; sometimes there is a net increase, sometimes there is a net loss
|
|
Demographic Transition
transition stage: |
- transition stage: mortality rate drops while there’s a lag in the drop of the birth rate so there are major discrepancies between the 2 (large rate of natural increase)- this is a population explosion
- developed countries have gotten through this stage and into the post-transition stage |
|
Demographic Transition
post-transition stage: |
post-transition stage: both birth and death rates are low but death rate is constant while birth rate fluctuates
- during this stage, perhaps human beings are in a better position to control their fertility |
|
What are the four "perspectives" on Demographic Transition?
|
Description
Classification Explanation Prediction |
|
T/F Demographic Transition is poor at explaining and good at predicting
|
True
|
|
immigration:
|
permanent moves in across natural boundaries
|
|
emigration:
|
permanent moves out across natural boundaries
|
|
Brain Drain
|
describes movement of educated people from less developed countries to more developed countries (like the US)
|
|
T/F In the US immigration exceeds emigration
|
True
|
|
a great paradox: as mortality rates declined over time what happens to the proportion of younger and older people and why?
|
as mortality rates declined over time, this made the population younger, not older because the big beneficiaries of the mortality decline are the young
|
|
infant mortality rate =
|
infant mortality rate = # of infant deaths/# of live births
|
|
Describe family size in a transitional stage such as the Victorian Era.
|
- large families were characteristic of Victorian England not because the culture was concerned with having large families but because it was a transitional stage: kids were being born and surviving unexpectedly (the people were used to 50% or more of children dying before the age of 8 which is what happened in the pretransitional stage)
|
|
Describe family size in a pretransitional stage.
|
pretransitional stage did not have large families (had many children but died in childhood)
|
|
Why was the female infant mortality rate much greater than the male infant mortality great in India (year 2000)?
|
There is a strong preference for male infants -
female infant neglect > male infant neglect |
|
What are the two ways that age-pyramids can be graphed?
|
1. in absolute numerical terms
2. in percentage distribution (better because gives comparability across time and space) |
|
Age Pyramids
What are we comparing? |
1. variation across age groups
2. variations within age group by sex |
|
Describe the shape of an age pyramid for less developed regions.
|
pyramid for less developed regions have a wide base which means high fertility rate so high rate of population growth
|
|
Describe the shape of an age pyramid for more developed regions
|
more developed regions have a narrow base which means low fertility
|
|
Age pyramids are a cross section of data from 1 year and as a result they count _______
|
immigrants
|
|
Bumps in a pyramid can reflect what?
|
Wars can reduce male populations
Bias. such as immigrations of males into the United Arab Emirates |
|
What were the major causes of the death in the early 1900's?
|
communicable, infectious diseases
|
|
What are the major causes of death today?
|
Chronic diseases
|
|
T/F Life expectancies have increased in most parts of the world
|
True
|
|
What are the four stages of the epidemiological transition?
|
1.) Pestilence and Famine
2.) Receding Pandemics 3.) Degenerative and Man-Made Diseases 4.) Delayed Degenerative Diseases and Emerging Infections (Hybristic) (see page 46 of Allison's SG for diagram) |
|
What were some of Thomas McKeown's conclusions about the population explosion?
|
modern medicine had little to do with decrease in mortality rates
showed that decreased death rate trends preceded effective therapies the industrial revolution and increased affluence (not clinical medicine) were responsible for rising life expectancies |
|
Life expectancy is calculated on basis of ____________ but has long-term implications
|
Cross-sectional data
|
|
Public health measures developed in the West have been more important in bringing about decline in mortality rates in _________
|
Developing Countries
|
|
What is the goal of public health in terms of the appearance of the graph of survival?
|
We want the graph to be as rectangular as possible
|
|
Life expectancy refers just to _______
We'd preferably want to ________. |
quantity of life
maximize quality of life (compression of morbidity) |
|
Who developed DALY? What is DALY?
|
Murray and Lopez
Disability Adjusted Life Years which incorporate mortality (quantity of life concerns) and morbidity and other quality of life concerns |
|
Study interpretation: Describe the decision tree that leads from Association to Causal relationship
|
Association present ->Chance absent-> Hypothesis formed beforehand -> no bias -> no confounding -> fits causal criteria -> Causal relationship
|
|
In the decision tree that leads from Association to Causal relationship,
What probably occurred if chance was present? What can we do? |
There is not a statistical significance.
Type II error We need to consider whether the sample size was sufficient. |
|
In the decision tree that leads from Association to Causal relationship,
What happens if a hypothesis was not specified first? |
Hypothesis generating
Study/cluster |
|
What do you need to consider when you are interpreting a "Positive study"?
|
Chance - Random Variability
Is the effect real? Data Dredging Hypothesis testing |
|
What do you need to consider when you are interpreting a "Negative study"?
|
The possibility of:
Random variability, which could have obscured a real association. This relates to sample size and statistical power |
|
The number of the sample size, n is determined by:
|
Magnitude of effect
Proportion of population exposed for case-control studies, or developing the outcome for cohort studies / clinical trials |
|
Error questions:
Ho is True Accept Ho |
Correct
|
|
Error questions:
Ho is True Reject Ho |
Type I error
Reflected in p value cutoff is denoted as alpha |
|
Error questions
Ho is False Reject Ho |
Correct
Power = 1-B |
|
Error questions
Ho is false Accept Ho |
Type II error.
|
|
T/F Increasing Type I error decreases Type II error and vice versa
|
True
|
|
When do we look at Type I and Type II errors?
|
Type I errors: After the study
Type II errors: Before study - this is related to the power! |
|
What is true about most negative studies that are published in the literature?
|
They are underpowered
|
|
Key question is:
How big a sample do I need so that if my study is negative, I know I’m not missing anything? |
This is the statistical power of a study
|
|
What is RR or OR are considered "small effects"
Is it easy to detect small effects? |
It is harder to detect small effects (RR or OR < 1.5)
|
|
Why should we be wary of studies reporting small effects?
|
Studies reporting small effects may easily have arisen because of chance, bias, confounding
|
|
Exposure is difficult to determine and can limit epidemiological efforts. What are two examples of this?
|
EMF, Cancer
|
|
What are two methods to increase statistical power?
|
precise classification of disease and exposure status (minimize random misclassification)
adequate follow-up in cohort studies and clinical trials |
|
Adequate follow up?
97% |
yes
according to Sackett > 95% : no worries |
|
Adequate follow up?
87% |
maybe
according to Sackett >80: gray zone |
|
Adequate follow up?
75% |
no
according to Sackett <80: No |
|
Prospective cohort study looking at mortality in 100 patients with 84% follow-up reported 4 deaths
Is this adequate? |
NO
4 die, 16 lost to follow-up Reported mortality is 4/84 = 4.8% If all those lost to follow-up had died, worst case mortality: 20/100 = 20% If all those lost to follow-up had lived, best case mortality: 4/100 = 4% |
|
Types of Bias
Two Broad Types |
Selection and Information Bias
|
|
Selection Bias
Three types |
The Healthy Worker Effect
Volunteer bias Berkson’s bias |
|
The Healthy Worker Effect
What type of bias? Description? Inventor? |
Selection bias
Worker population may not represent the general population. May be healthier Anthony McMichael. |
|
Volunteer Bias
Type of bias? Description? |
Selection Bias
People who agree to be in a study are different than those in general population |
|
Berkson’s Bias
Type? Description. |
Patients getting care (clinics, hospitals, etc) are systematically different from patients in general populations
|
|
Patients with co-morbid conditions are more likely to seek treatment than those with only disease
What kind of bias is shown here? What was the example given in class? |
Berkson's bias
Eating disorders and substance abuse together are more likely to be reported by family members than either one condition alone |
|
Information Bias
|
Any error in how information is gathered about exposure and/or disease
|
|
What are the types of Information Bias?
|
1. Recall bias
2. Interviewer / Measurement / Expectancy / Systematic or Differential misclassification bias 4. Loss to follow-up 5. Expectancy bias 6. Proficiency bias 7. Neyman bias |
|
Neyman bias
What type of bias? Describe. |
Errors is WHEN we collect data
Length bias will be discussed under Screening Look back bias If we look at point prevalence, those individuals who had a short duration of disease will be underrepresented |
|
Neyman Bias is also known as ________
|
prevalence-incidence bias
|
|
What two types of studies are particularly prone to Loss to Follow-up bias?
|
cohort and intervention studies
|
|
What are ways we can control/limit bias in our studies?
|
Must rely on study design
Prospective studies generally superior Choice of controls in case-control study Rigorous, standardized assessment methods Use of blinding Follow-up |
|
T/F Prospective studies are generally superior in limiting bias than the other types of studies.
|
True
|
|
In a study of prophylactic administration of isoniazid, tuberculin-positive school children were randomly assigned to drug or placebo treatment groups. A third group, consisting of those who elected not to enroll in the study, was also observed. After several years, the third group had a significantly higher rate of tuberculosis than the placebo group.
What kind of bias is shown here? |
Volunteer bias (type of selection bias)
|
|
T/F A confounder can be complex and either known or unknown
|
True
|
|
What can we do if we know what the confounder is?
|
If known, can be measured and statistically controlled
|
|
What is Simpson's Paradox, and what can it determine for us in terms of a variable?
|
If an apparent effect disappears after controlling for variable, it is a confounder
|
|
Methods to Control Confounding
|
Restriction
Randomization Stratification Multivariate analysis Matching |
|
Methods to Control Confounding
Restriction |
Only study people who are homogenous with respect to a known confounder
|
|
Methods to Control Confounding
Early heart studies were primarily focused on men, what kind of confounding reduction method is being employed? |
Restriction
|
|
Methods to Control Confounding
Stratification: Examples |
Indirect, direct, SMR
Mantel Haenszel stratification Chi square, RR for cohort |
|
Mantel Haenszel stratification
|
Can calculate weighted average as OR or RR to obtain summary unconfounded risk estimate
Also works for Chi square |
|
Stratification can only be done on a _____ number of factors
|
limited
|
|
Multivariate analysis
Describe it. Pros? Cons? |
Computer based
Develop a model to explain data Pros: Linear or logistic regression of multiple variables Cons: Not good for complex relationships |
|
What is the effect of matching on a confounder?
|
Match by using same status for potential confounder so effect “cancels out”
|
|
In which kind of study can individuals be matched to themselves?
|
Cross-over study
|
|
Influences on study results
Hawthorne Effect |
Production of workers improved solely because they knew they were being observed
May account for false positive results |
|
Prosecutor’s Fallacy
|
1/50,000 that DNA sample taken from innocent person matches DNA at the crime seen
3/4 that person that has DNA that matches the DNA at crime scene is innocent. |
|
Gambler’s Fallacy
|
A random event is more likely to occur because it has not happened for a period of time
|
|
Regression to the Mean
What is can happen when you take a test and your find out your score (relative to the True score that you really deserve...) |
Be unlucky and score below your ‘true’ grade
Score exactly at your ‘true’ grade Be lucky and score above your ‘true’ grade |
|
Ockham’s Razor
|
"Entities should not be multiplied beyond necessity.”
in other words, the simplest explanations are preferred |
|
Validity (aka internal validity)
|
= is it a good study?
|
|
Generalizability (aka external validity)
|
= does the study mean anything for the patient before me?
|
|
T/F Generalizability is more important than validity.
|
FALSE.
Validity is more important. It has to be a good study before you can generalize the findings to the general population |
|
What factors can threaten generalizability?
|
Volunteer bias
Restriction Invasive intervention studies usually performed on advanced cases Studies usually performed in academic settings Compliance Efficacy (ideal) versus effectiveness (real world) |