EPI 504 final

the formula to estimate prevalence odds ratio from a 2x2 table is

(axd)/(bxc)

when there is not misclassification in a study, the values for sensitivity and specificity would be

1 or 100%

the odds of an event can be calculated as

P/(1-P)

describe the healthy worker effect

a certain kind of selection bias that may occur in cohort studies of occupational exposures. the effect occurs when the general population is used as a comparison group, because the general population may not necessarily be as healthy as those who have an occupation

which of the following criteria are used to assess confounding

a priori criteria and data-based criteria

if the crude effect estimate is significantly different from the stratum-specific estimates based on a third variable, but the stratum-specific estimates are similar to each other, which estimate will you report?

adjusted

the numerator in the equation of calculate cumulative incidence includes

all new cases of disease during the period of the study

what is the ecologic fallacy

assuming that individual members of a group exhibit characteristics of the group at large

what type of error cannot be rectified during the analysis phase of the study

bias

what type of error cannot be rectified during the analysis phase of the study?

bias

T/F confidence limits are fixed and do not vary from sample to sample

False

T/F etiologic fraction of a disease in a population is the measure of the population of disease incidence in the exposed that is attributable to the exposure

False

select the appropriate study design. students from a college were surveyed about their sun exposure behaviors. the survey obtained information about their knowledge, attitudes, and behaviors about sun exposure, sun protection, and skin cancer. Information was also collected on their history of sun burns, personal and family history of skin cancer; along with several demographic variables.

cross-sectional

selective survival is a main reason for selection bias in which type of study design

cross-sectional

Name the study design. The investigator observes both the exposure and health outcome simultaneously, so that neither variable may be uniquely identified as occurring first.

cross-sectional study

prevalence odds ratio is a measure of effect used in

cross-sectional study

which study design is useful to describe patterns of disease occurrence

cross-sectional study

the probability that subjects in a study will develop a disease for the first time within a given period of time

cumulative incidence

the branch of epidemiology that allows one to examine the distribution of basic person, place, and time-related aspects of a disease is called

descriptive epidemiology

a process to determine or confirm the presence or absence of a disease based on signs and symptoms

diagnosis

an odds ratio for a matched case control study compares

discordant pairs

which of the following is not a bias

effect modification and interaction

what is the main purpose of selecting controls in a case control study?

estimate the distribution of exposure in the underlying source population

which of the following is not recognized as a step in effect estimation in a case-control study?

estimate the odds of outcome among exposed

which of the following is a synonym of the term "exposure"

factor, predictor, and cause

T/F Randomization makes all characteristics similar between the comparison groups, including exposure status

false

T/F a case-cohort study design is less prone to measurement error compared to cohort study design

false

T/F both risk and rate measures have units

false

T/F comparing ratios effect estimates helps to determine if the interaction is on an additive scale

false

T/F ecologic studies are large studies done in several countries at the same time using data collected from every individual in the country as their unit of analysis

false

T/F incidence is the measure of transition from a diseased state to disease-free state

false

T/F it is not possible to have both additive and multiplicative interaction at the same time

false

T/F measures of impact help us to assess the individual risk perspective on exposure-disease relationships

false

T/F selection bias occurs when individuals have similar probabilities of being included in the study sample according to exposure and/or outcome status, relative to the source population

false

T/F sensitivity, while measuring for information bias, is the probability that a subject who is truly unexposed (or non-diseased) will be correctly classified as unexposed (or non-diseased) in the study

false

T/F study designs that are backward or non-directional have a smaller potential for selection bias than forward designs

false

T/F the incidence rate (or incidence density) is the number of old and new cases divided by the person-time, usually the number of person-years, at risk

false

T/F the null values (representing no association between an exposure and a health outcome) for ratio and difference measures are 0 and 1.

false

T/F there will be a differential misclassification when the rate of misclassification is the same between different study groups

false

T/F time, place, and population are not required components when estimating a rate

false

T/F true population parameter is a random variable

false

T/F true risk ratio of a study is 2.5 and the observed risk ratio is 4.5. the bias here is towards the null.

false

T/F vaccination is an example of tertiary prevention

false

T/F we can estimate incidence in a cross-sectional study

false

T/F we can safely assume that those who volunteer to participate in a study correctly represent the characteristics in a source population

false

T/F when there is proof of effect modification in your study data, you should report adjusted effect estimates

false

T/F you have been informed that 90% of the breast cancer patients are married at the time of diagnosis. this information is sufficient to make a causal association that being married increases the risk for breast cancer

false

an epidemiological study started by examining exposure to a specific drug in teenagers and followed them to study mental health outcomes after 5 years. what is the directionality of the study?

forward

select the appropriate study design. a study examined a new drug for treating hypertension. patients with hypertension were randomly assigned either to receive the new drug or the conventional therapy. blood pressure was monitored at regular intervals for one year in both treatment groups.

randomized clinical trial

what measure of association are you assessing in this statement: the rate of dying within 5 years among workers who were exposed to a radiation accident compared to the unexposed is increased by 70 deaths per 1000 person-years

rate difference

which measure of association provides information on excess risk?

rate difference

an important bias in a case-control study

recall bias

risk ratio and rate ratio provide ____ measure of association

relative

an estimate to assess if your measurements repeatedly and consistently yield the same results under similar conditions

reliability

select the appropriate study design. a study investigated the association between occupational exposure to compound x and skin cancer in a large chemical factory. exposure data was abstracted using employee records and each worker's contact with compound x was determined. not all workers were exposed to compound x. exposure groups were established based on a worker's history of contact with compound x. cases of skin cancer that occurred during the study period were determined using data from the states cancer registry. lung cancer incidence was examined by the worker's exposure status

retrospective cohort study

what measure of association are you assessing in this statement: the risk of developing breast cancer within 10 years among women exposed to hormone replacement therapy is 3 times greater than the risk among the unexposed

risk ratio

which of the following relative measures of association cannot be calculated from cross section study data

risk ratio and rate ratio

which two measures of association would you commonly use to understand a causal association between an exposure and outcome?

risk ratio, rate ratio

precision reflects _______

sampling variability

a systematic difference in the enrollment of participants in a study that leads to an incorrect result or inference---what is this phenomenon called?

selection bias

what is a systematic difference during the time of subject selection in a study that leads to an incorrect result or inference

selection bias

systematic error can occur at which stage of a study

selection, data collection, and data analysis (all of the above)

an estimate of proportion of people with the disease of interest who test positive

sensitivity

the underlying parameters that must be considered when assessing information bias include

sensitivity and specificity

an epidemiologist starts a cohort study by selecting a defined population for study before any of the subjects become exposed, rather than starting by selecting exposed and unexposed subjects. how does this subject selection procedure benefit the epidemiologist?

several exposures can be studied simultaneously

we can measure precision using ____ or _____ around the measured effect estimates.

standard errors, confidence levels

all measurement issues should be addressed prior to data collection process to ensure

standard methodology, cost, time efficiency

if there is an effect modification due to a third/extraneous variable between Exposure and disease, you report

stratum specific estimate

T/F cohort studies begin with exposure and follow participants forward in time to assess disease outcome

true

T/F confidence intervals allow for statistical significance testing

true

T/F cumulative incidence, when measured for a short or brief period as in an outbreak, is referred to as "attack rate"

true

T/F differential probing of one group compared to the other based on prior knowledge can lead to interviewer bias

true

T/F early detection of existing disease to reduce severity and complication is known as secondary prevention

true

T/F ecologic studies allow us to understand a correlation between two factors

true

T/F external validity is most often determined based on expert subjective judgment

true

T/F flaw in measurement methods can lead to information bias

true

T/F follow-up period is the time during which a subject in the study is followed with regard to developing the outcome of interest

true

T/F friends can be a source of control in a case-control design

true

T/F having incident or new cases in a case-control study helps to avoid biases introduced due to death

true

T/F heterogeneity means that the effect estimates are very different in each of the study strata

true

T/F hypothesis testing can be specified in terms of the difference between the two population proportions (RD) being equal to 0 (null), and either different from 0 (2-sided alternative) or greater than 0 (1-sided alternative)

true

T/F if the crude and adjusted estimates are the same, you can proceed to report crude estimate in your study findings

true

T/F in a retrospective study, the health outcome has occurred before the study is initiated

true

T/F in ecologic studies, the units of analysis are populations or groups of people rather than individual people

true

T/F internal validity can be achieved by avoiding selection, measurement, and analytic biases and errors.

true

T/F it is important that unexposed subjects are as healthy as the exposed subjects to avoid selection bias in occupational studies

true

T/F matching helps gain precision in a study

true

T/F misclassification bias occurs when the effect measure for the correctly classified data is meaningfully different from the estimated effect actually observed in the misclassified data

true

T/F odds ratio can be calculated in both case-control and cohort studies

true

T/F one person-year is equivalent to two subjects, each followed for 1/2 year

true

T/F precision is the inverse of random error

true

T/F proportion is a ratio in which the numerator is included in the denominator

true

T/F randomization curtails to a large extent the freedom of study participants to select themselves into different exposure arms of a clinical study

true

T/F randomization helps to account for unmeasured factors

true

T/F rehabilitation for stroke is a type of prevention

true

T/F selection bias occurs in cross-sectional studies when exposed cases can survive longer than unexposed cases or vice versa

true

T/F selection probabilities are hard to estimate

true

T/F statistical inference is the process of drawing conclusions about populations based on sample data

true

T/F the absolute effect provides needed information to make policy decisions on the burden of public health problems due to a given exposure

true

T/F the first step in a case-control study is to identify cases, or diseased persons, to be studied

true

T/F the values for risk (cumulative incidence), when measured as % can range from 0% to 100%

true

T/F true risk ratio of a study is 0.5 and the observed risk ratio is 0.9. the bias here is towards the null

true

T/F we should try to select controls from the same population from which cases are selected

true

a methodology flaw in an epidemiologic study that arises due to systematic error in the study design

validity

an ability of a test to correctly identify persons with or without a disease is called

validity

bias influences the ___ of a study

validity

the degree to which a measurement measures what is it supposed to measure

validity

if you have 2 screening tests for the same condition, one with high sensitivity and another with high specificity, which will you use first?

high sensitivity

what type of allocation procedure assigns subjects into one of the exposure groups being compared so that each subject has the same probability of being in one group as in any other?

randomization

how can we correct selection bias

1. at the time of study design 2. quantitatively using selection probabilities during post-study design phase

what can lead to a misclassified disease status

1. complex diagnostic procedure 2. inadequate access to technology to determine the disease (or testing) 3. laboratory error (all of the above)

list two ethical considerations for conducting a disease screening in a community

1. consent 2. availability for all

list two conditions under which one can approximate odds ratio to risk ratio

1. if population size is small 2. all cases are included

what are the two main differences between experimental and observational epidemiological study designs?

1. in experimental designs, the researcher has control. In observational, the researcher is only an observer. 2. in experimental designs, the researcher can control participants' exposure and can randomly place participants into different exposure groups. in observational designs, the participant controls the exposure status.

a misclassification of disease status may occur due to

1. incorrect diagnosis 2. subject self-report 3. data coding errors (all of the above)

which of the following statements is true about systematic error

1. it can occur from a methodologic flaw in the study design 2. it can occur due to imperfect data collection procedure 3. it can occur by using wrong data analysis methods (all of the above)

what is an important consideration for a good case-definition

1. laboratory confirmation 2. standard diagnostic criteria 3. objective and measurable (all of the above)

List three advantages of a hybrid design

1. less prone to selection bias 2. can be cost effective (nested case control) 3. multiple diseases can be studied from within the same cohort (case-cohort)

for a confounder (or control variable) to meet a priori criteria, which of the following rules must be met

1. must be a risk factor for the disease 2. must be associated with the exposure in the source population 3. cannot be an intervening variable between exposure and disease (all of the above)

2 types of controls

1. population based- selected from same community as cases 2. hospital based

confidence intervals

1. provide the measure of precision of a point estimate 2. are measures of random error in the data 3. provide upper and lower limit numbers between which the point estimate exists (all of the above)

Rank the hierarchy of the five study designs with 1 being the highest strength and 5 being the lowest

1. randomized trials 2. cohort studies 3. case-control studies 4. cross-sectional studies 5. incomplete design

list 3 conditions under which we can approximate odds ratio to risk ratio

1. rare disease/ disease frequency is low 2. number of unexposed to exposed ratio in the controls is the same as in the source population 3. if all cases in the case selection process are included

which of the following statements is true regarding selection bias in cohort studies?

1. selection bias can occur with a. high overall response rate 2. selection bias can occur with very little loss to follow up 3. selection bias may not occur despite small response rate or high loss to follow up (all of the above)

if there is evidence of confounding, which of the following method can be used to report an adjusted effect estimate

1. simple or weighted average 2. mantel-haenszel 3. precision based

statistical test used to examine dose-response

1. test for linear trend 2. mantel-haenszel chi square test for linear trend

to measure cumulative incidence, we need

1. to follow the entire population for a specified time 2. to address and minimize attrition (loss to follow up) 3. that subjects should be disease free at the time of study enrollment 4. to be aware of the population dynamics (all of the above)

which of the following cutoff kappa values measuring inter-rater reliability indicates a poor agreement that is beyond chance?

<0.40

what is the statistical test for interaction

Breslow-Day

which of the following 95% confidence intervals around the odds ratio is more precise?

OR= 2.14 (95% CI= 2.08-2.56)

T/F case-control studies are preferred over cohort studies because they are inexpensive and quick

true

which of the following results is statistically significant

RR=0.45 (95% CI= 0.01-0.67)

4 different types of study designs in which there is potential for selection bias to occur

case control, cohort, clinical trials, cross sectional

select the appropriate study design. a study investigated the association between multivitamin use and anemia in older adults. a total of 1000 patients with anemia in participating hospitals were interviewed about their history of multivitamin use. another group of 1000 patients, who were also receiving care in the same hospitals as cases with anemia, but not anemic, were also interviewed about their history of multivitamin use.

case-control

exposure odds ratio is a measure of effect used in

case-control study

name the study design. investigator first selects subjects on the basis of whether or not they have the health outcome of interest, and then looks back in time to obtain information about their previous exposures.

case-control study

what type of study design would have a selection bias when there is differential withdrawal of participants out from the study based on their exposure status

clinical trial

name the study design. the investigator starts by determining the exposure status for subjects, selected from some population of interest, and follows these subjects over time to determine whether or not they develop the health outcome.

cohort study

risk odds ratio is a measure of effect used in

cohort study

which of the following is the best measure to draw conclusions or inferences about a study

confidence interval

If you are offered three measures of a study results (p-value, point estimate, or confidence interval) which one would you choose to draw conclusions about the study?

confidence interval: provide greater information than p-values

what type of bias occurs when a different factor, other than your main exposure, causes spurious associations in your study findings?

confounding

what type of bis occurs when a different factor, other than your main exposure, causes spurious associations in your study findings

confounding

which of the following is assessed by comparing unadjusted and adjusted effect estimates

confounding

a new diagnosis of breast cancer among participants of the women's health initiative cohort during a 5-year follow-up period (incidence v prevalence)

incidence

for a given duration of disease, the prevalence will ______ as the incidence increases

increase

what is the term used to assess variation in agreement of findings between multiple observers

inter-observer

which one of the following can be assessed using a statistical test

interaction

A disease is call endemic if it

is habitually present in a population or region

expresses the extent to which the observed agreement exceeds that which would be expected by chance alone relative to the most that the observers could hope to improve their agreement.

kappa statistic

What is berkson's bias

occurs when hospital or clinic-based case subjects and control subjects within a case control study are systematically different from each other. this is because the combination of exposure to risk and occurrence of disease increases the participant's likelihood of being admitted to the hospital. the hospital case patients have a systematically higher exposure rate than the controls, causing a bias odds ratio.

what measure of association/effect are you assessing in the following statement: the odds of smoking among mothers who delivered preterm infants are 6 times the odds of smoking among mothers who delivered normal term infants

odds ratio

what is John snow's contribution to epidemiology?

often referred to a the father of modern epidemiology. he tracked cases of cholera in London and was able to determine that drinking water was the source of the contagion, which helped create a foundation for descriptive epidemiology.

statistical testing is the process of drawing conclusions about

population from which a study sample is drawn

an estimate of the probability of truly having the disease, if the screening test result is positive

positive predictive value

random error is mainly associated with the ____ of a study

precision

number of children who have immunity to measles, either because they had the disease or because they received the vaccine (incidence v. prevalence)

prevalence

the proportion of women who reported taking prenatal vitamins during their latest pregnancy in the pregnancy risk assessment monitoring survey (incidence v prevalence)

prevalence

what is prevalence and what is the formula for the two main kinds of prevalence

prevalence is the percentage of the population that has the disease (existing cases) within a given period of time. Two main types: 1. point prevalence (P=C/n) 2. period prevalence (PP=(C+I)/N)

what measure of association are you assessing in the following statement: the prevalence of skin cancer among those exposed to tanning beds is 4 times the prevalence of skin cancer among those who never used tanning beds

prevalence ratio

cross-sectional study is also known as

prevalence study

the odds of an event can be calculated as P/(1-P) where P is the ____ of the event

probability

select the appropriate study design. a study recruited 5000 healthy individuals from town A that had an accidental environmental contamination. the study also enrolled 5000 healthy individuals from a neighboring town B which was otherwise like town A, except that it was not influenced by the environmental accident. Both towns were followed for 10 years to determine several chronic health outcomes among the two study groups (assume very small loss-to-follow up in both groups). an analysis was conducted after 10 years to examine the association between environmental contamination and cancer.

prospective cohort

this type of error indicated that your study results differs from the truth due to chance alone

random error

this type of error indicates that your study results differs from the truth due to chance alone

random error

what is the type of error that can influence the precision of the study

random error

what type of allocation procedure assigns subjects into one of the exposure groups being compared so that each subject has the same probability of being in one group as in any other

randomization

we know that the true relative risk of an association between a selected exposure and outcome is 4.0 in the population X. a study conducted to examine the same association, but in a small sample of participants form population X and with an imperfect study method, yielded a relative risk of 1.5. what type of error in risk estimates would you call this

systematic error or bias

what three factors constitute epidemiologic triad of disease?

the host, the agent, and the environment

what is a p-value

the probability of obtaining a test statistic as extreme as or more extreme than what was observed if the null hypothesis is true. This is assuming that the null hypothesis is true and that there are no sources of bias in the data collection or data analysis

internal validity helps to infer about

the source population

define the term epidemiology

the study of diseases--the ways they are distributed within populations and what influences how they are distributed within populations

which of the following statements is false about cross-sectional studies

they are directional analyses

which of the following is not true for a retrospective cohort study

they are useful to study rare diseases

which of the following is not true for a retrospective cohort study?

they are useful to study rare diseases

for a non-differential misclassification of exposure or outcome, the. bias is ___ (assume no random fluctuations or other errors in measurement of variables being controlled that are misclassified)

towards the null

T/F Person-time is the sum of the periods of time that all persons who are at risk (for the disease or death) contribute to the study

true

T/F a detection bias is when a physician gives more thorough exams to patients who are exposed or have symptoms related to exposure compared to patients who are unexposed

true

T/F a nested case-control study is more prone to measurement error than a cohort study

true

T/F a nested case-control study starts with a defined cohort

true

T/F an observational study is generally helpful to generate and test hypotheses

true

T/F case definitions may evolve over time as additional information is learned about the pathology of the disease

true