Epidemiology Midterm
Which of the following statements about measures to quantify disease prognosis is/are accurate?
B and C are accurate B. The limitation of mortality density rates (incidence density for mortality outcomes) to quantify survival/prognosis is that they assume each person year of follow-up carries the same risk of mortality. C. Life-table methods (i.e. the actuarial life table and Kaplan-Meier) assume that those lost to follow-up have a similar survival experience to those remaining in the study.
In order to calculate an age-adjusted SMR (standardized mortality ratio) using indirect-adjustment, what must you know?
C, D, and E C. The age-specific mortality rates from a standard population D. The age group sizes (# of people) of the study population E. The observed number of deaths in the study population
In order to calculate an age-adjusted SMR (standardized mortality ratio) using indirect-adjustment, what must you know?
C, D, and E only C. The age-specific mortality rates from a standard population D. The age group sizes (# of people) of the study population E. The observed number of deaths in the study population
The percentage of women with heart disease who died from heart disease over the last 5 years.
Case fatality proportion
Referring to Table 2 above, what is the age-specific rate in 45-54 years old individuals (per 100,000/yr) in population A ?
35/170,000 = 20.59 per 100,000 per year
Referring to Table 2 above, what is the expected number of cases in 65-74 years old individuals in Population A?
(21/75,000)*21,000,000 = 5,880
You are interested in studying the frequency of falls leading to hospitalization among an elderly population attending your outpatient clinic. You identify 835 eligible patients over a 4 year period, of whom 215 were followed for 4 years, 331 for 3 years, 104 for 2 years, and 185 for 1 year. During the 4 year period, there were 72 falls requiring hospitalization. What is the incidence density rate of falls leading to hospitalization per 1,000 person-years in this population?
(215*4)+(331*3)+(104*2)+(185*1) = 2246 72 falls requiring hospitalization / 2246 person years = 32.05 falls per 1,000 person years
The table below describes the number of skin cancer cases by age-group, within a study population in which each individual had 100% follow-up for 1 year. What is the total incidence density of skin cancer in the study population per 100,000 person-years?
(284 / 150,000) X 100,000 = 189 per 100,000 person-years
Given the total number of cases you now know about as of January 2014, and the new information that 3 of these patients have died due to diabetes-related complications, what is the case-fatality of T1DM per 1,000 affected residents in Boulder? Hint: First calculate the total number of cases as of January 2014.
(3/414)*1,000 = 7.2 deaths per 1,000
Calculate the value of the measure of frequency you selected in the previous question, per 100,000 residents.
(374 cases / 294,567)*100,000 = 127 per 100,000 residents
What is the age-adjusted rate ratio comparing Population A to Population B?
(Total Expected Cases/Total Standard Population) = [29,683/ (48,000,000+43,000,000+46,000,000+35,000,000+21,000,000)] = [29,683/ 193,000,000] = 0.0001538 or 15.38 per 100,000 p-y Age-adjusted rate ratio = Population A/Population B = 15.38/30.32 = 0.51
What is the correct definition of the incubation period of an infectious disease? (select one best answer)
) The time between exposure and onset of disease signs and symptoms. Incubation period is the time between exposure to the agent and onset of disease signs and symptoms. Refer to Dr. Scallan's lecture on Epidemiology of Infectious Disease
You find out that there were 40 new cases of T1DM reported between January 2012 and January of 2014. Given this information and the estimated mid-year population of 316,342 Boulder residents, what is the rate of T1DM per 100,000 person-years?
[40/(2 years*316,342 mid-year population)]*100,000 = [40/632,684] = 6.3 cases per 100,000 person-years.
The latent period of disease X is 23 days, and the latent period of disease Y is 5 days. Both diseases have an incubation period of 15 days. In which group would quarantine of the infected persons be more effective in preventing the spread of the respective diseases?
a) Disease X Instructor explanation: If the incubation period is shorter than the latent period, isolation of diseased persons can prevent further illnesses. If the incubation is longer than the latent period, those still incubating the disease still be at risk for spreading the disease.
How does the communicable period differ from the latent period? (Choose one best answer.)
a) The latent period comes before the communicable period and is the time delay between exposure and onset of contagiousness. Instructor Explanation: The latent period is the time between exposure and onset of contagiousness. The Communicable period follows the latent period and is the time when the host is actively excreting the agent and can transmit it to others. Refer to Dr. Scallan's lecture on Epidemiology of Infectious Disease (slide 41).
What is the proportion that should appear instead of "A" at day 8 of follow-up in the Kaplan-Meier table above? Please give your answer to two decimal places. Reminder: a proportion is a number bound between 0-1.
6/40 = 0.15
Which of the following is an example of a primary prevention strategy in population health?
. A and B only ( banning smoking in public and fluoridation of water supply) Instructor Explanation: Primary prevention in population health seeks to prevent illness amongst those who are well, before onset of the disease process. Banning smoking in public places and fluoridation of the water supply are strategies that seek to prevent exposures, thus onset of the disease process. Screening (options C and D) are examples of secondary prevention which target detecting disease amongst those with the disease in the subclinical stage. Option E is a tertiary prevention strategy aimed at preventing adverse outcomes amongst those with clinical disease.
What is the proportion that should appear instead of "B" at day 10 of follow-up in the Kaplan-Meier table in the above question?
0.63*0.91 = 0.57
Using the table in the previous question, what was the cumulative mortality of individuals receiving the new treatment at 4 months using the actuarial life table survival analysis method? Please give your answer as a percent rounded to the 10th decimal place.
1-0.8269 = 0.1731 *100 = 17.3%
In Colorado, there were 20 deaths due to smoke, fire and flames in 2019. You examine the coroner's reports on these 20 cases and discover that most of these deaths occurred in residential fires, and that combustible cigarettes or marijuana are involved in many of the cases. 1.You decide to use a case-control study design to evaluate the hypothesis that residential fire deaths are associated with combustible cigarettes or marijuana. Cases are defined as people who died in a residential fire between 2010 and 2019 in Colorado. What is/are the ESSENTIAL characteristic(s) of controls?
1. A and D Instructor explanation: The case-control study design selects cases and controls. Essential considerations for the controls are that they draw from the same population that gave rise to the cases and don't have the outcome at the time of enrollment.
An epidemiologic study was conducted at a high school to determine if participation in sports were associated with methicillin-resistant Staphylococcus aureus (MRSA) skin infections. The investigators identified 127 laboratory-confirmed MRSA skin infection cases and 381 controls from the high school. Based on review of school records it was determined that among the MRSA cases, 12 were on the wrestling team and only 4 controls were on the wrestling team. 1. What type of study design is this? 2. What is the best measure of association, in the above study, to use to examine the exposure disease relationship? 3.In the study presented above, how many more times likely were high school students with a MRSA skin infection to be on the wrestling team than students who did not have MRSA? Report your answer rounded to the 10th decimal place.
1. Case-Control 2. Odds ratio 3.Instructor Explanation: MRSA Cases Controls Member of wrestling team 12 4 Not a member of wrestling team (127-12)=115 (381-4)=377 127 381 OR = (12 * 377) / (4 * 115) = 9.8
A study was conducted to evaluate the risk of stroke associated with tobacco use among 19,843 adults with congenital heart disease (CHD) in Colorado between 2011-2019. Individuals were enrolled into the study in 2011 based on coding in medical records consistent with CHD and were excluded if they had a stroke history. Tobacco use was assessed at study enrollment and each individual was categorized as either having a "significant tobacco use history" or "not a significant tobacco use history" based on pack-years of tobacco use. Of the total enrolled participants, 1,984 were categorized as having a significant tobacco use history and the remaining 17,859 did not. After follow-up of 8 years where no participants were lost to follow-up, 257 individuals with CHD who had a significant tobacco use history experienced a stroke and 1,071 individuals with CHD who did not have a significant tobacco use history experienced a stroke. 1.You wish to evaluate the relationship between tobacco use among individuals with CHD and risk of stroke. What is the answer for the appropriate measure of association for the risk of stroke for individuals with CHD who had a significant tobacco use history compared to individuals with CHD who did not, given the study design? (Choose the one best answer)
1. Instructor Explanation: The appropriate measure of association for this prospective study that obtains cumulative incidence is the risk ratio. Exposure Outcome Total Stroke No stroke Tobacco history 257 1727 1984 No tobacco history 1071 16788 17859 Total 19843 Risk ratio=(257/1984)/(1071/17859)=2.16 2.
Study designs fall into a hierarchy with regard to their ability to generate information about causality of an exposure-disease relationship. Rank the following study designs to match the correct hierarchy with the most powerful design being "1".
1. RCT 2. Cohort/longitutidinal 3. Case-control 4. Cross- sectional 5. Case-series
The following data were collected during an evaluation of a new screening test. For Questions 2-7, calculate the sensitivity, specificity, positive and negative predictive values and accuracy of the test, as well as the prevalence of the disease within the sample population, using data in the table below. Gold Standard (+). Gold Standard (-) total New Test (+) 119 40 159 New Test (-) 9 262 271 Total 128 302 430
1. What is the Sensitivity of the new screening test? Give your answer as a percentage rounded to one decimal place (ex: 15.1%). 1. Recall what each of the table cells represent: A = true positives, B = false positives, C = false negatives, and D = true negatives. Sensitivity is the proportion of people correctly identified by the test as having the disease (A, true positives) out of the total number of individuals who truly have the disease (A+C, true positives plus false negatives). Sensitivity = = TP / (TP+FN) = A / (A+C) = 119 / (119+9) = 0.930 = 93.0% 2. What is the Specificity of the new screening test? Give your answer as a percentage rounded to one decimal place (ex: 15.1%). 2. Specificity is the proportion of people correctly identified by the test as NOT having the disease (D, true negatives) out of the total number of individuals who truly do NOT have the disease (B+D, false positives plus true negatives). Specificity = TN / (FP+TN) = D / (B+D) = 262 / (262+40) = 0.868 = 86.8% 3. What is the positive predictive value (PPV) of the new screening test? Give your answer as a percentage rounded to one decimal place (ex: 15.1%). 3. The positive predictive value is the proportion of people who truly have the disease and who were correctly identified by the test (A, true positives) out of all of the individuals who had a positive test (A+B, true positives plus false positives). PPV = A / (A+B) = 119 / (119+40) = 0.748 = 74.8% 4. What is the negative predictive value (NPV) of the new screening test? Give your answer as a percentage rounded to one decimal place. 4. The negative predictive value is the proportion of people who truly do NOT have the disease and who were correctly identified as negative by the test (D, true negatives) out of all of the individuals who had a negative test (C+D, false negatives plus true negatives). NPV = D / (C+D) = 262 / (9+262) = 0.967 = 96.7% 5. What is the accuracy of the test within the sample described above? Give your answer as a percentage rounded to one decimal place. 5. The accuracy (or validity) of a test is the total proportion of individuals who are correctly classified by the test as diseased or non-diseased. Accuracy = Correctly classified (TP+TN) / Population (TP+FP+FN+TN) = (A+D) / (A+B+C+D)= (119+262)/(119+40+9+262)=0.886 = 88.6%
In order to calculate an age-adjusted mortality rate using direct adjustment, what must you know?
A and B only A. The age group sizes (# of people) of a standard population B. The age-specific mortality rates of the study population
What are essential considerations when selecting controls for a case-control study?
A and D A. Controls should be representative of the population that gave rise to the cases D. Controls should not have the outcome at time of enrollment
Which of the following is the most correct definition of cumulative incidence?
A proportion composed of the number of new cases occurring during a specific time interval per total number of people at risk for the outcome at the beginning of that interval, assuming no losses to follow-up.
Which of the following statements are true about matching of cases and controls in a case-control study?
A, B and D are true A. Each case is matched to one or more controls to improve efficiency of the study B. Matching seeks to provide a similar distribution important study variables between cases and controls (i.e., sex, age-grouping) D. Matching can be used to ensure an adequate number of cases and controls in exposure categories where a disproportionate case/control ratio is expected
Protection offered by a vaccine is an example of which type of immunity? (select one)
Acquired immunity Instructor explanation: Acquired immunity is developed as a person's immune system encounters foreign substances (antigens) such as microbes or those present in a vaccine. Acquired immunity is antigen-specific and takes time to develop, but subsequent responses will be quicker and more effective.
Please read the following journal article excerpt and identify the study design used. "We identified eligible cases and controls from the national cancer register... Case cancers in England and Wales were those of people aged 15-74 years with a diagnosis of mouth, lung and all respiratory system cancers for air ion analysis, and keratinocyte carcinomas for electric field analysis. Cases were all first primary cancers diagnosed between 1974 and 2008 and were found among people living within 600 m and 25 m of a high voltage overhead power line, respectively... Controls were selected from a range of cancers not considered to be associated with electromagnetic fields..."
Case-control
Read the journal excerpt below: BACKGROUND: Rosacea is a common chronic inflammatory dermatosis of unclear origin. It has been associated with systemic comorbidities, but methodical studies addressing this association are lacking. OBJECTIVE: We evaluated: (1) the association between rosacea and systemic comorbidities; and (2) if the severity of rosacea is impacted by comorbidities. METHODS: Patients with rosacea were matched (1:1) to rosacea-free control subjects by age, sex, and race. Systemic comorbidities were determined for each case and control by reviewing electronic medical records. RESULTS: Among 130 participants (65 patients/65 control subjects), we observed a significant association between rosacea and allergies (airborne, food), respiratory diseases, gastroesophageal reflux disease, other gastrointestinal diseases, hypertension, metabolic and urogenital diseases, and female hormone imbalance. Compared with mild rosacea, moderate to severe rosacea was significantly associated with hyperlipidemia, hypertension, metabolic diseases, cardiovascular diseases, and gastroesophageal reflux disease. CONCLUSIONS: Rosacea is associated with numerous systemic comorbid diseases in a skin severity-dependent manner. Physicians should be aware of these associations to provide comprehensive care to patients with rosacea, especially to those with more severe disease. What study design was used?
Case-control
Read the journal excerpt below: Exposure to light at night suppresses the physiologic production of melatonin, a hormone that has anti proliferative effects on intestinal cancers. Although observational studies have associated night-shift work with an increased risk of breast cancer, the effect of night-shift work on the risk of other cancers is not known. We prospectively examined the relationship between working rotating night shifts and the risk of colorectal cancers among female participants in the Nurses' Health Study. We documented 602 incident cases of colorectal cancer among 78 586 women who were followed up from 1988 through 1998. Compared with women who never worked rotating night shifts, women who worked 15 years or more on rotating night shifts were 1.35 times more likely to develop colorectal cancer (95% CI = 1.03 to 1.77), adjusting for confounders. These data suggest that working a rotating night shift at least three nights per month for 15 or more years may increase the risk of colorectal cancer in women. What study design was used?
Cohort
Read the journal excerpt below: An investigator is interested in whether children who live in houses with cats or dogs have lower prevalence of allergic disease than children who live in houses without cats or dogs. To answer this question, she designs a survey and randomly samples 1,100 households with children under 13 years old in the Denver metropolitan area. At each household, she asks the parent to report whether they currently have cats or dogs in the house, and whether their child suffers from asthma, eczema, allergic rhinitis, or other allergic diseases. She then calculates a prevalence ratio for allergic disease among children in households with pets as compared to children in households without pets. What study design was used?
Cross-sectional
The lifetime risk of breast cancer.
Cumulative incidence
The difference between an incidence density rate (IDR) and cumulative incidence is: (check one best answer):
Cumulative incidence is a measure of risk and IDR is a measure of velocity at which new cases of the disease is occurring in a population. Instructor explanation: Cumulative incidence is a measure of risk, it's a proportion (with new cases identified over a specified time interval) and incidence-density-rate is a rate or measure of the velocity at which a disease is occurring in a population (has person-time in the denominator).
What is the cumulative survival at 12 months, calculated using the actuarial life table survival analysis method? Please give your answer as a percent rounded to the 10th decimal place. To obtain a % you need to multiply your answer by 100. For example, if you were reporting the cumulative survival at 2 months from the table you would type in 90.3.
Cumulative survival at 6 months = (0.9524*0..8269) = 0.7875 Cumulative survival at 8 months = (0.9279*0.7875)=0.7307 Cumulative survival at 10 months = (0.8316*0.7307)=0.6077 Cumulative survival at 12 months = (0.8400*0.6077)=0.5105
What type of misclassification would occur if exposure was more frequently misclassified in the disease group versus the non-disease group?
Differential misclassification This is an example of differential misclassification because misclassification is seen in one direction (the diseased group has more misclassification than the non-diseased group); and is related to the exposure or outcome. If misclassification is seen in both directions, and is not related to the exposure or outcome, this would be an example of non-differential misclassification.
What method of standardization have you just implemented? You have strata-specific rates from your population of interest and denominators from a standard population.
Direct standardization
Which causal criterion is supported by this specific statement from the abstract "...the cognitive developmental benefits of breast-feeding increased with duration." (Choose one best answer.)
Dose Response Relationship Stating that the level of effects on the outcome increased with duration of exposure, describes a dose response relationship.
Studies have shown regular exercise to be associated with increased levels of high density lipoprotein cholesterol (HDL-c) levels. The following table was presented from a cross sectional study of 2,906 healthy, nonsmoking middle aged men (mean age 43 years). The subjects were stratified into 6 groups based on the mean number of miles they ran per week. A statistical test was performed to evaluate the presence of a positive linear trend across running categories. Based on this table, which causal guideline could be evoked to support the statement that long distance running increased HDL-c levels (choose one best answer)? Running category. HDL-c levels (mg/dl) Non-runners 47.3 5 miles/ week 50.7 9 miles/ week 52.5 12 miles/ week 53.0 17 miles/ week 56.3 31 miles/ week 60.1 P-value for trend p<0.0001
Dose Response Relationship We see evidence of a dose response relationship - as running distance per week increased HDL levels also increased.
Two podiatrists want to investigate whether a new clinical exam identifies patients with plantar fasciitis. Dr. Corn uses the standard clinical exam, which has a sensitivity of 80% and specificity of 92%. Dr. Bunion uses the new clinical exam method, which has a sensitivity of 90% and a specificity of 92%. If the same sample of patients undergo both exam methods, which of the following are true?
Dr. Corn will correctly identify fewer people with plantar fasciitis compared to Dr. Bunion. Recall that the sensitivity and specificity are properties of the test which do not depend on the prevalence of disease. Also recall that both of these tests are being used in the same sample of patients. The standard clinical exam used by Dr. Corn has a lower sensitivity (80%) than the new clinical exam used by Dr. Bunion (90%) - this means that Dr. Bunion will correctly classify more people with the disease than will Dr. Corn. The specificity of the standard clinical exam and the new clinical exam are the same (92%), meaning that both doctors will correctly classify the same number of people without the disease.
In which of the following epidemiological study designs is the unit of investigation populations or groups of people rather than individuals
Ecologic Study
Please read the following excerpt of a journal article abstract and identify the study design used: Objectives: In the absence of robust direct data on ethnic inequalities in COVID-19-related mortality in the UK, we examine the relationship between ethnic composition of an area and rate of mortality in the area [emphasis added]. Design: Analysis of COVID-19-related mortality rates occurring by 24 April 2020 and ethnic composition of the population... Results: For every 1% rise in proportion of the population who are ethnic minority, COVID-19-related deaths increased by 5·12, 95% CI (4·00 to 6·24), per million...
Ecological
Read the journal excerpt below: According to de Catanzaro's (1981) evolutionary theory of human suicide, threshold intelligence is necessary for suicidality. Thus there is interest in determining if Intelligence and suicide mortality are positively correlated. This theory was investigated in a study that included 85 countries. Average national IQ was found to be significantly positively related to the national male and female suicide rate. The relationship was not attenuated by controlling for the countries' per capita Gross Domestic Product or the type of national IQ estimation. What study design was used?
Ecological
In a study assessing the relationship between stressful life events and heart disease, men with stressful life events were found to have significantly higher risk of developing heart disease compared to women with stressful life events. In this example, what kind of variable is sex (M/F)?
Effect modifier
As an epidemiologist at Jefferson County Health Department, you have been tasked with collecting quantitative data to determine the incidence of Influenza in Jefferson County. Your supervisor has asked you to develop a sentinel reporting system to collect quantitative data. Which of following would be the most efficient and accurate efforts to accomplish this task?
Establish a network of healthcare clinics, hospitals and laboratories within Jefferson county to report all individuals with positive laboratory tests for influenza. In addition, your staff will periodically review laboratory testing results to identify additional missed cases (a combination of passive and active reporting).
It is now January 2014 and as the Boulder County epidemiologist, you want to know the average annual incidence rate of Type 1 diabetes (T1DM) in the city of Boulder for the period between January 2012 and January 2014. You have the Boulder County population size for 2012 and 2013. You also know that there were 374 existing cases with T1DM as of January 2012. What other information, if any, do you need to calculate the average annual incidence density rate?
I can calculate the number of person-years of follow up for my denominator based on a mid-year population (average of 2012 and 2013) and the number of years of follow up (2 years), but I need the number of new cases that were reported between January 2012 and January 2014 for my numerator.
There are multiple epidemiologic study designs. Study designs can be categorized as an experimental or observational design. What is the primary difference between the two?
In an experimental design, the exposure/treatment is assigned
An important distinction between an experimental study (clinical trial) and an observational study is: (choose one best answer)
In experimental studies, the exposure is assigned by the researcher whereas in observational studies, the researcher does not assign the exposure.
A case-control study is characterized by all of the following except
Incidence rates can be computed directly.
What is the correct interpretation of a p-value of 0.7?
Indicates no evidence against the null hypothesis
Which of the following are established methods of preventing the spread of infectious disease? (Select one best answer)
Instructor Explanation: All of the listed prevention strategies are used to prevent the contraction and spread of infectious disease. a) Antimicrobial prophylaxis b) Active immunization c) Avoid exposure d) Passive immunization e) Eliminating the infectious agent
An elementary school with 258 students had a pizza party that 192 kids attended. Starting the next day several families who attended the party reported vomiting among their children. By the next morning, 79 children had fallen ill with vomiting. What is the attack rate of this disease? (report your answer as a percentage rounded to 1 decimal place: Ex: 15.5%)
Instructor Explanation: An attack rate is calculated by dividing the number of people who became ill by the total number of people at risk of becoming ill. There were 258 students who go to the school, but only 192 kids attended the party, so 192 is the at-risk population. Therefore the attack rate is 79/192 = 0.411 or 41.1%.
When a new treatment is developed which prevents death but does not result in the recovery from a disease, which of the following will occur?
Instructor explanation: Prevalence will increase because you have more existing cases.
An epidemiologist is interested in the rate of gonorrhea in a population of 2000 college students who are known to be at risk for this disease. The epidemiologist chose to follow the students for a total of 3 years. A total of 1250 students were followed for the entire 3-year period and among these students there were 90 cases of gonorrhea. 525 students were followed for 2 years and among these there were 30 cases of gonorrhea. 225 students were followed for only 6 months and among these there were 10 cases of gonorrhea. What is the incidence density rate of gonorrhea per 1,000 person years from this study? Please round your answer to the 10th decimal place.
Instructor's Explanation: You have information about how long a certain number of students were at risk of gonorrhea and therefore you can calculate person-years for your denominator. The total person time is (1250*3)+(525*2)+(225*0.5)= 4912.5 Your numerator is the number of students who contracted gonorrhea over the 3-year follow-up period (90+30+10=130). Your rate calculation should look like this: IDR=130/4912.5=0.026463*(1000 person years) =26.463=26.5 per 1000 person years
Study biases and confounding would affect the ___________ validity of a study.
Internal Instructor's Explanation: Internal Validity asks the questions: "How well was the study done?"; "Did it measure what is was supposed to measure?"; "Was the analysis done appropriately?" Threats to internal validity are error (random or systematic), bias (selection or information), misclassification (non-differential or differential) and confounding. Threats to external validity are present when the study population has different characteristics than the general population.
Within a recent medical journal, an advertisement stated: "1,000 subjects with bronchitis were treated with our new medicine. Within three days, 96.4% were asymptomatic." The advertisement claims that their medicine was effective. Based on the evidence given above, what would an epidemiologist think about this claim:
May be incorrect because no control or reference group was involved
The Centers for Disease Control and Prevention and the United Nations have come together to provide a generous amount of funding to investigate the impact of natural disasters leading to environmental radiation leaks on a variety of health outcomes. Due to your large budget, you are able to implement objective radio assays to determine levels of radiation exposure among your subjects. What bias would the use of radio assays likely reduce the most? (Choose one best answer)
Misclassification of exposure Instructor explanation: It would reduce misclassification due to exposure status, since we are confirming if the subjects were exposed via a diagnostic test.
Which of the following statements are true about bias in epidemiological studies? (Choose 1 best answer)
Non-random, systematic deviation of the study results from the truth causes bias Biases arise in epidemiological studies when there are systematic, non-random deviations of the study results from the truth. If deviations between the study results and truth are random and non-systematic, this is not bias, however standard deviation will increase and power will decrease, thus p-values and confidence intervals cannot be used to detect bias.
The results of a case-control studying the association between smoking and risk of Alzheimer's disease are shown below. Which odds ratio is correct?
OR = a*d/c*b OR = 60*74 / 58*32 = 2.39
A study sought to investigate whether underweight (BMI <18.5kg/m2) could be an independent risk factor for cardiovascular disease (CVD). The investigators analyzed data from the cross-sectional Behavioral Risk Factor Surveillance System (BRFSS) administered to 491,773 U.S. adults who self-reported their current BMI and history of CVD events including stroke, heart attack/myocardial infarction, or coronary artery disease. The underweight population had a 19.7% greater likelihood of CVD events than did the normal-weight population, and the overweight and obese population had a 50% and 96% increased likelihood, compared to the normal-weight population, respectively. In the context of results from this cross-sectional study, what measure of association does the statement of "a 19.7% greater likelihood of CVD events in the underweight relative to the normal-weight population relate to"? (select one best answer)
Odds ratio or prevalence ratio Instructor explanation: In cross-sectional studies, exposure and outcome are ascertained at the same time. The measure of disease occurrence that can be calculated is prevalence (in this example prevalence of CVD history). Measures of incidence are not possible as there is no temporal relationship between exposure and disease. The measure of association that can be calculated from cross-sectional studies is a prevalence ratio or an odds ratio.
Select the options below that accurately describe some of the distinctions between an active and passive surveillance system of a health condition.
Options A and B are correct A. An active surveillance system is one where public health professionals actively visit or interacts with health care systems (clinics, laboratories, hospitals) to ascertain individuals with a particular health condition whereas a passive system relies upon the healthcare system to report cases to public health agencies. B. Active surveillance is more time-consuming and expensive than passive surveillance.
A recent clinical trial reported in their "Methods" section, that the study was "double blinded". What did the author mean by this terminology?
Participants and researchers were both unaware of group assignment
Based on data from the Behavior Risk Factor Surveillance System (BRFSS) in Colorado, between 2013 and 2015, 17.4% of adults aged 18+ reported they were physically inactive. Which term best describes this measure of occurrence?
Period Prevalence
The percentage of girls who are positive for a high risk Human Papilloma Virus (HPV) strain during their high school years.
Period prevalence
As of January 1, 2012, there were 374 existing cases of Type I diabetes. The most recent Census reported a population of 294,567 residents living in Boulder on this date. With this information, what measure of Type 1 diabetes frequency can you calculate?
Point Prevalence
We want to determine the proportion of a disease that could be prevented by eliminating its associated exposure in the entire population. Which measure should be calculated?
Population attributable risk percent
Calculate the rate ratio for mortality of colon cancer for males compared to females in patients >65 years of age.
RR = 80.1/53.9 = 1.5
Calculate the rate ratio for mortality of colon cancer for males compared to females in the patients <65 years of age.
RR = 9.3/6.7 = 1.4
What type of bias occurs in a case-control study when cases report their use of a medication (the exposure of interest) more accurately than controls? (Choose one best answer)
Recall bias Instructor Explanation: The correct answer is recall bias: a type of information bias where cases remember and report past exposures differently than controls. Selection bias in case-control studies occurs when the probability of selection of cases or controls is differential on the basis of exposure. The healthy worker effect is a type of bias that frequently occurs in occupational studies when employed individuals are, on average, healthier than a comparison population that includes unemployed individuals. Effect modification is not a type of bias. Berkson's bias applies mostly to studies utilizing hospitalized patients in which higher rates of hospital admissions are seen for individuals with 2 or more conditions, rather than one.
A prospective cohort study was conducted to determine the risk of heart attack among men with varying levels of baldness. Third-year residents in dermatology conducted visual baldness assessments at the start of the study. Four levels of baldness were coded: none, minimal, moderate, and severe. Study participants were followed for 10 years to observe heart attack outcomes. The follow-up rate was close to 100%. Which of the following types of bias were surely avoided in this study? (choose one best answer)
Recall bias of exposure information Differential misclassification of exposure Selection Bias due to loss to follow up Recall bias and differential misclassification of the exposure were avoided because of the prospective nature of the study as the outcome had not yet occurred. Selection bias was avoided because it is a prospective cohort study with nearly 100% follow-up.
A case-control and a prospective cohort study were conducted to assess the association between exposure X and disease Y. The results of the case-control study suggested that the odds of exposure X were 2.05 times higher among those who had disease Y compared to those who did not (Odds Ratio = 2.05, 95% confidence interval 1.50, 3.13). However, the results of the prospective cohort study suggested that the risk of outcome Y was only 3% higher and not statistically significant among those who were exposed to X compared to those who were not (Risk Ratio = 1.03, 95% confidence interval 0.97, 1.86). Assuming the results from the prospective cohort study are reflective of the true association, which of the following responses would most likely explain the differing results? (Choose the one best answer)
Recall bias within the case-control study Instructor explanation: The most likely explanation for the differing results would be recall bias in the case-control study. Recall bias occurs when subjects who experienced the outcome may be more or less likely to report the exposure of interest. This can occur due to the tendency among those with a disease or adverse health outcome to search their history for a potential cause, and therefore are more likely to report being exposed. In this case, recall bias could result in the exposure being over-reported among those who had outcome Y (e.g. cases), and therefore produce a biased estimate of association between exposure X and outcome Y. Recall bias is not as great a concern in cohort studies, because the exposure is generally recorded prior to the occurrence of the disease. Confounding bias is possible in both studies, but is not a sufficient explanation for the difference in results between the two types of studies. The placebo effect is a type of bias in experimental studies in which subjects experience a measurable improvement in health or behavior that is not attributable to treatment. This is one reason why masking (i.e. blinding) is vital within such designs
How do you remove confounding at the study design level in a case/control study (when designing the study, before data collection starts)? (choose one best answer)
Restrict the sampling frame/limit inclusion criteria Match on characteristics that ate thought to be confounders Instructor Explanation: There are two ways to deal with confounding on the study design level in a case-control study. 1) Restrict the sampling frame to a more tightly defined population based on certain demographic characteristics (i.e., males, over the age of 50, non-Hispanic white or Hispanic white). This will reduce the confounding on these factors but will reduce the external validity of your findings. 2) The second way to deal with confounding on the study design level is to match cases and controls on specific demographic factors (i.e., age, sex, residence). Matching will reduce confounding; however, you will not be able to examine the effect of the matching characteristics in your analysis.
Read the journal excerpt below: In 2000, 201,743 women between the ages of 45 and 65 were recruited to participate in a study that assessed whether greater access to preventive care was associated with earlier detection of cervical cancer (cancer diagnosed at stage I). At the start of the study, medical records were obtained documenting the number of pap smears each participant had received in the 10 years before their enrollment in the study. After enrollment, women were followed for 10 years to determine if they had been diagnosed with cervical cancer and at what stage their cancer was diagnosed (stage I, II, etc.). By 2010, 87 cases of stage I cervical cancer and 24 cases of stage II cervical cancer were reported. The researchers found that there was a lower incidence of stage II cervical cancer among women who had received a greater number of gynecological exams where a pap smear was done. What study design was used?
Retrospective cohort
The Center for AIDS Research at the University of North Carolina at Chapel Hill (UNC) initiated a prospective cohort study of HIV-infected patients in 1999. This clinic based cohort study enrolled 2,511 African American and Caucasian individuals and monitored CD4 cell count and HIV RNA levels at the first clinic visit (enrollment visit) and subsequent follow-up visits conducted every 6 months through 2012. During the study, 404 patients died, 1,390 were lost to follow-up and 717 reached the end of the study period alive. Researchers were interested if African American race was causally associated with time of death. Which of the following biases may occur in this study? (select one best answer)
Selection bias due to loss to follow up Instructor's Explanation: The bias of greatest concern is selection bias due to loss to follow-up due to the large numbers of individuals who were lost to follow-up. The other biases are not applicable to this study design.
An investigator wishes to evaluate the effect of tea tree oil on eradiation of toenail fungus in a double-blinded randomized controlled trial. However, she is concerned about comparability of groups by age (>30 years of age or <= 30 years of age) that may influence response to therapy. Which design level strategy could she use to increase the likelihood of comparability between study groups? (Choose the one best answer):
Stratified randomization according to age categories Stratified randomization is the design-level strategy that can be used to increase likelihood of comparability between age groups. Factorial design is used to evaluate the independent and joint effect of two treatments. Blinding (of either study personnel or study participants) and use of a placebo control is used to reduce or eliminate biases.
By assessing JUST the information below, which if Bradford Hill's causal guideline is addressed with this finding: (Choose one best answer) The risk of lung cancer was 18 times greater among smokers compared to non-smokers (p<0.01).
Strength of Association Based solely on the information presented, strength of association evident from the large effect size, is the only guideline for causality that was met. The effect size demonstrates the strength or magnitude of an association between a risk factor/exposure and outcome/disease.
Why is a cohort design more efficient than a case-control design to study a rare exposure?
Subjects can be selected based on their exposure status.
Establishing a causal relationship for potential environmental carcinogens is particularly difficult because the latency period between exposure and illness is often decades long. An epidemiologist supports her claim that environmental agent X has a causal relationship to cancer outcome B by stating that agent X preceded onset of cancer outcome B. Which causal guideline is the epidemiologist more directly addressing with this statement (choose one best answer)?
Temporality The correct answer is temporality. If a factor is believed to cause a disease, then exposure to that factor must occur before disease developmen
Which of the following design elements are needed to have a successful "cross-over" clinical trial (not to be confused with with a case-cross over observational study design).
The effects of each treatment should not be detectable after a washout period
Which of the following are not threats to internal validity (select all that apply)?
The intervention is not feasible in a real-world setting Appropriate calculation of measures of association Internal validity assesses how well the study was done. All are issues of internal validity except "the intervention is not feasible in a real-world setting" which is an issue of external validity (How generalizable is the study?).
The following are standardized mortality ratios (SMRs) for breast cancer mortality by occupation in Australia. The standard population for each SMR is women of the same age group from all occupations in Australia. Note: The SMRs in the table are ratios multiplied by 100 (ex: 0.35 * 100). Based on these SMRs alone, you can make the following conclusion:
The mortality in school teachers during the period 1949 - 1960 was less than what would have been expected for a group of women of the same age in all occupations.
What is the NPV of the test in the new population? Give your answer as a percentage rounded to one decimal place.
The negative predictive value is the proportion of people who truly do NOT have the disease and who were correctly identified as negative by the test (D, true negatives) out of all of the individuals who had a negative test (C+D, false negatives plus true negatives). The NPV will change if the prevalence of the disease changes, but the specificity and sensitivity of the test will remain the same. So given a Sensitivity of 93.0% and a Specificity of 86.8%, create a new population based on a new population with a prevalence of 10% and a total of 10,000 persons total. The new cell values would be: A=930 B=1188 C=70 D=7812 NPV = D / (C+D) = 7812 / (70+7812) = 0.991 = 99.1%
What is the correct definition of attack rate? (select one)
The number of persons at risk for the disease in whom illness develops divided by the total number of persons at risk.
A case-control study investigating the relationship between disease Z and exposure Y was conducted. Researchers calculated an odds ratio of 1.5. What is the best interpretation of this number (you can assume that the prevalence of disease is >10%)?
The odds of exposure Y are 1.5 times higher in those with disease Z compared to those without disease Z.
In which scenario would it be best to select a case-control study design over a retrospective cohort study design?
The outcome being studied is rare
Assuming the same sensitivity and specificity from Questions 2 and 3, calculate PPV and NPV if the prevalence of the disease were only 10% (Assume 10,000 total persons in a new table).
The positive predictive value is the proportion of people who truly have the disease and who were correctly identified by the test (A, true positives) out of all of the individuals who had a positive test (A+B, true positives plus false positives). The PPV will change if the prevalence of the disease changes, but the specificity and sensitivity of the test will remain the same. So given a Sensitivity of 93.0% and a Specificity of 86.8%, create a new population based on a new population with a prevalence of 10% and a total of 10,000 persons total. The new cell values would be: A=930 B=1188 C=70 D=7812 PPV = A / (A+B) = 930 / (930+1188) = 0.439 = 43.9%
In a study on diet and cardiovascular disease (CVD), investigators prospectively follow a group of 200 vegetarians and 400 omnivores. They were interested in the effect of a vegetarian diet on risk of CVD. After 20 years of follow-up, 15 of the vegetarians and 40 of the omnivores developed cardiovascular disease (CVD). The 95% confidence interval on the relative risk of 0.75 ranged from 0.5 to 0.9. Which of the following statements is correct?
Vegetarians were 25% less likely to develop CVD during 20 years of follow-up compared with omnivores.
Researchers conducted a clinical trial to assess whether Treatment X would effectively reduce the risk of a myocardial infarction in patients with significant arterial calcification. However, the trial was stopped due to an observed increase in the risk of death for patients assigned to receive Treatment X compared to the placebo treated group. Furthermore, it is well established that older age further increases an individual's risk of a myocardial infarction therefore the researchers were interested in determining if the increased risk of death due to treatment with Treatment X was only present in the older patients. The table below presents their data. Based on the strata-specific Risk Ratios, what measure(s) of association would you report for this study? Assume that 10% constitutes a meaningful difference to evaluate whether effect modification is present (select one best answer)
The strata-specific RRs for each age group (<55 and > 55)
The following are results of a randomized clinical trial of New Therapy A and Usual Therapy B to reduce new events. Which of the following statements is most correct about the trial results? Assume everyone completed the study after receiving the correct treatment for 6 months (no losses to follow-up) and statistical testing indicated significance of your measure of association.
Therapy A is better than B since the RR of A/B is 0.73
Which of the following is NOT true of cohort studies?
They are relatively inexpensive compared to case-control studies
Referring to Table 2 above, what is the total expected number of cases in Population B?
Total expected = (3,840 + 7,938 + 18,838 + 13,400 + 14,500) = 58,516
A randomized controlled trial tested whether a new anti-hypertension medication to reduce blood pressure was successful at preventing future strokes among subjects with hypertension. The new medication was considered successful. The researchers reported that the "number needed to treat" (or NNT) was 15. The statistic, NNT, can be interpreted as:
Treating 15 hypertensive subjects with the new medication would prevent one stroke case
Randomization is a procedure used for assignment or allocation of subjects to treatment and control groups in experimental studies. Randomization increases the likelihood that: (Choose the one best answer):
Treatment and control groups are similar except for the presence of treatment Instructor Explanation: Randomization makes the two groups equal (except by chance) on all the other factors but the treatment allocation (treatment group versus placebo). It does not guarantee that the results will be significant, generalizable and it will not decrease the likelihood of losses to follow-up.
True or False: In the comparison of rates using direct standardization, the choice of the standard population is arbitrary as long as the rates being compared are being adjusted to the same standard population.
True
A randomized trial comparing two drugs showed a statistically significant difference in symptoms between the two (P value = 0.04). Assume that in reality the two drugs did not actually differ. This is therefore, an example of:
Type I error (α error)
Below are the results of a cohort study. What is the proportion of disease among the exposed that is attributable to the exposure?
Using the formula (Ie - Iu)/Ie *100: Ie=55/83=0.663 Iu=45/78=0.577 AR%=(0.663-0.577)/0.663 * 100 = 12.97 rounded to 13% We can also use the formula [(RR-1)/RR] * 100. RR=(55/83)/(45/78)=1.149 AR%=(1.149-1)/1.149= 0.129 * 100 = 12.9 rounded to 13%
The number of live-born babies who die of sudden infant death syndrome during the first year of life per 100,000 baby years of follow-up.
incidence density rate
