Week 3: Measurement, reliability and validity

Question 1

Explain each term regarding exposure measurement:
1. Available dose (cumulative vs current)
2. Administered dose
3. Absorbed dose/uptake
4. Active dose

Answer 1

1. cumulative active dose = total amount given to a patient over time
   current active dose = amount given to a patient at a specific time
2. administered dose = amount that comes in contact; can be administered in different routes
3. absorbed dose/uptake = amount that actually enters the body
4. active dose = how much actually had a biological effect on the target site

Question 2

45 individuals older than 75 admitted to the hospital in Toronto for a shoulder dislocation in the past year. This is an example using the term _________

Answer 2

count

Question 3

1. What is ratio? Give an example
2. What is proportion? Give an example

Answer 3

1. ratio = relationship to COMPARE 2 numbers. The numerator is NOT necessarily included in the denominator. Ex: binary sex ratio –> 5/2 = 5:2 ratio of boys compared to girls
2. proportion = relationship to show something OUT OF the total number. The numerator MUST be included in the denominator. Ex: binary sex proportion –> 2/4 are girls = 50% girls

Question 4

1. What is odd?
2. In a population of 100 people, 25 people are diabetic. What is:
   a) the proportion of diabetes
   b) the odd of being a diabetic

Answer 4

1. the probability of an event occurring relative to it not occurring
2. a) proportion = 25/100 = 0.25
   b) odd = 25/75 = 0.33 –> probability of event occurring (25 people have it) over it not occurring (75 people do not have it)

Question 5

1. What is rate? Explain the numerator and denominator in a rate expression
2. Write a rate for this situation: In 2021, 25 people were diagnosed with chronic bronchitis. The population at risk consisted of 1000 people.

Answer 5

1. Rate = speed of occurrence of an event over time
   numerator = # of events observed for a given time
   denominator = population that is at risk in which the events occur –> incorporates a period of time
2. 25/1000 = 0.025 of pop/year (2021) had chronic bronchitis

Question 6

1. What is prevalence rate?
2. What are the 2 types? Explain (use formula)

Answer 6

1. prevalence rate = proportion of the population that HAS a given disease at a specific TIME
2. Point PR = # with disease at a specific time/ total # of people in pop at same time
   Period PR = # with disease during time period/ total # of people in pop at same time

Question 7

If 3 million people in Canada have diabetes NOW and the current Canadian population is 30 million, what is the point PR of diabetes per 1000 people?

Answer 7

= 3 million / 30 million = 0.1 = 10% / 1 person
per 1000 = 0.1 x 1000 = 100 cases / 1000 people

Question 8

In 2018, Montreal had a population of 152358. There were 6 cases of mononucleosis. What is the period PR of mononucleosis during 2018 per 100,000 persons?

Answer 8

Period PR = 6/152358 = 0.00004 per person x 100,000 = 4 per 100,000 persons

Question 9

1. What is incidence rate? Give the formula
2. What type of studies are incidence rates obtained from:
   a) randomized controlled trials
   b) cohort (longitudinal) studies
   c) case-control studies
   d) clinical trials

Answer 9

1. The proportion of the population AT RISK that develops a given disease or other attribute during a specified time period.
   IR = # of new cases during a specified time period/ population “at risk”
2. b)

Question 10

describe the relationship between prevalence and incidence and what results

Answer 10

incidence (those at risk) —> can become prevalence (those with the disease) —> which leads to the resolution (recovery or death)

Question 11

9,000 residents of Toronto were studied on Jan 1, 2010. Of these residents, 245 had a history of cancer. By Dec 31, 2019 a total of 179 new cases of cancer had been diagnosed in the “at risk” population. What is the 10-year cumulative incidence rate of cancer (first diagnosis) in this cohort per 10,000 people?

Answer 11

IR = 179 / (9000-245) = 0.02 x 10,000 = 204 per 10,000 people.

*the at risk population (denominator) is the total pop - those who have already had cancer since we need to find a first diagnosis IR.

Question 12

2,012 first year university students were sampled in Sept 2021. Of these, 112 had depression. In May 2022, 345 of the sample had depression, including 79 of those who had depression in September.

a) What was the prevalence of depression Sept 2021 and May 2022?
b) How many “resolution” cases were there?
c) What is the IR of new depression cases during the freshman year (Sept 2021 to May 2022) ***don’t understand answer
d) why do we remove those with depression diagnosed at the beginning of the study from the population at risk in part c)?

Answer 12

a) Sept 2021 IR= 112/2012 = 0.056 = 5.6%
May 2022 IR= 345/2012 = 0.17 = 17%

b) 112 - 79 =33

c) IR= (345-79)/ (2012-112) = 266/1900 = 0.14 = 14%

d) because these people have already had depression so they can not be at risk of developing a NEW case of depression. So they are excluded from the population at risk (denominator)

Question 13

Incidence measures frequency of disease _______. Prevalence measures population disease _______

Answer 13

onset
status

Question 14

1. What is relative risk?
2. relative risk tells us the _______ of an association
3. What does relative risk NOT tell us?
4. How is it calculated?

Answer 14

1. tells us how many times as likely it is that someone who is “exposed” to something will experience a specific outcome compared to someone not exposed
2. Strength
3. does NOT tell us anything about how much MORE disease is occurring
4. Make the table, will be incidence in exposed/incidence in unexposed
   RR= [a/(a+b)] / [c/(c+d)]

Question 15

The results of a study evaluating the effects of calling patients and their likelihood to receive a influenza immunization revealed:

• those who received a call and got immunized = 328
• those who received no call and got immunized = 288
• those who received a call and did not get immunized = 332
• those who did received no call and did not get immunized = 370

What is the RR of those who received a call compared to those who did not. What do the results indicate?

Answer 15

exposure = receiving a call
outcome = getting immunized

RR= [328 / (328+332)]/[288 / (288+370)] = (328/660) / (288/658) = 0.49/0.43 = 1.14 or 14% more likely to attend for immunization

Question 16

20,000 men without heart disease were identified in Sept 2015. Of these, 3,614 were smokers and 16,386 were non-smokers. Over a 5 year follow-up period, 552 of the smokers and 1,431 of the non-smokers developed heart disease.
Fill out the table and calculate the RR of heart disease in smokers relative to non-smokers

Answer 16

RR = 1.76 = 76%

Question 17

1. what is random error?
2. What is systematic error?
3. systematic error = ____
4. T/F: it is not possible to have both random and systematic error

Answer 17

1. error due to chance
2. error due to a recognizable source
3. bias
4. F. it is possible

Question 18

A measurement with high precision is ______
A measurement with high accuracy is ______

Answer 18

reliable
valid

Question 19

What are the results of measurements that are classified as…. (explain in terms of reliability and validity)
a) high accuracy, high precision
b) low accuracy, high precision
c) high accuracy, low precision
d) low accuracy, low precision

Answer 19

a) valid, reliable
b) not valid, reliable
c) valid, not reliable
d) not valid, not reliable

Question 20

Why would there be insufficient precision?

Answer 20

• measurement tool is not precise enough
• 2 separate interviewers rate the same person differently using the same scale
• same interviewer rates the same person differently

Question 21

Identify the types of measurement error that could occur from these sources:
1. interviewer or observer
2. participants

Answer 21

1. random error, biased over or underestimation
2. recall bias, random or systemic error

Question 22

Identify the most likely sources of measurement error in the following situations:
a) Asking mothers with preterm birth if they were exposed to pesticide during pregnancy

b) Using 1st year psychology students to administer a questionnaire about depression without any training

c) Using only ‘students with a history of depression’ for b)

d) Asking a group of older adults about the number of their friends in high school

Answer 22

a) recall bias - participant error
b) biased under or over estimation - interviewer error
c) biased over estimation - interviewer error
d) recall bias - participant error

Question 23

1. What is misclassification?
2. what are the 2 types? Explain both and what happens to the association between exposure and outcome

Answer 23

1. classifying participants into the wrong exposure or outcome group
2. non-differential = the same misclassification in all study groups. this weakens associations and can bring the calculation (RR, OR etc.) closer to the null value (no association between outcome and exposure)

differential = different misclassification in the study groups. Associations can change in any directions

Question 24

Non-differential misclassification is often due to _______ measurement error

Differential misclassification is often due to ________ measurement error

Answer 24

random (due to chance)
systemic (locate the source)

Question 25

What are controls vs cases?

Answer 25

controls = no disease
cases = disease

Question 26

Results of a study showed:
exposed + case = 300
unexposed + case = 100
exposed + control = 250
unexposed + control = 150

1. Draw the table, calculate the OR
2. It was found that 10% of all exposed cases and exposed controls were misclassified as unexposed & vice versa (10% of unexposed cases and unexposed controls were misclassified as exposed). Update/fix the data. What type of misclassification occurred? What happened to the OR as a result?
3. It was found that 20% of unexposed cases (but not controls) were misclassified as exposed. What type of misclassification occurred? What happened to the OR as a result?

Answer 26

1. OR = 1.8
2. non-differential misclassification. new OR = 1.6 meaning its closer to the null (weakened association created)
3. differential misclassification. new OR = 2.4 meaning its further from the null in this situation (stronger association created)

Question 27

T/F: Information bias can be fixed once it has occurred

Answer 27

F