Clinical Trials

Question 1

What to look for when looking at a clinical trial paper?

Answer 1

• whether paper addresses your question
  -is study design appropriate for question asked
  -look at study subjects (how they are selected, what population, attention of individuals)
  -look at intervention and comparison
  -is placebo being used

Question 2

External validity

Answer 2

How applicable the results are to the general population of interest
**is the study population similar to the population you are comparing to

Question 3

Types of control groups

Answer 3

1.Historical controls
2.Concurrent controls
3.Cross-over trials

Question 4

Concurrent control groups (parallel arm trials)

Answer 4

Control groups formed at the same time as the treatment group

Question 5

Historical controls

Answer 5

Before and After groups

**less value because there are too many factors that can change over time causing differences between groups

Question 6

Cross-over trials

Answer 6

utilize the same animals as treatment and control groups
»means that order of treatments must be randomized, and often need time period in between to was away any effects

Question 7

3 Key elements of clinical trial design

Answer 7

1.Outcome measures
2. Bias
3. Chance

Question 8

Outcome measures

Answer 8

Check for:
-most clinically relevant
-easiest to measure reliably
-most objective outcome; subjective outcomes would need to be blind and have clear case definitions
-most specific (remove unrelated effects)

Question 9

How many outcomes involved in clinical trials?

Answer 9

-multiple outcomes is risky
-if you look at enough outcomes, one will likely be significant OR mistakes/false conclusions may be made
-author must prioritize them
-should focus on 1 or 2 primary outcome measures

Question 10

Examples of possible outcomes for BRD trial

Answer 10

-mortality (most objective)
-morbidity (need case definition)
-serological conversion (higher antibodies does not always mean immunity)
-Avg daily gain
-Feed efficiency (measured at pen level because can’t feed each individual alone)

Question 11

Experimental unit

Answer 11

-the smallest independent unit to which treatment is allocated

Examples: leg of animal, quarter of udder, individual animal, pen, herd

Question 12

Feeding trial experimental unit

Answer 12

Pen=experimental unit

because can’t feed every individual separately

Question 13

Herd immunity and its role in clinical trial

Answer 13

-vaccinated or treated animals may protect or reduce the challenge to all individuals in the herd or group altering the outcome/results making the differences in outcomes more similar

ex. anthelmintic trials or vaccine trials

Question 14

Bias

Answer 14

-A factor (systemic) other than the treatment causes a difference in the outcome between the treatment and the control group

Question 15

4 key times when bias can occurs

Answer 15

1.Selection/randomization
2. Performance bias (cointervention)
3.Exclusion bias
4.Detection bias

Question 16

Selection/randomization causing bias

Answer 16

-systemic difference in the treatment and control groups attributable to lack of randomization

Question 17

Performance bias (cointervention)

Answer 17

Systemic differences in the care provided apart from the intervention being evaluated

Question 18

Exclusion bias

Answer 18

systemic differences in withdrawals from the trial

Question 19

Detection bias

Answer 19

-systemic differences in outcome assessment or follow-up

ex. detailed post mortem in treatment group but not the control group

Question 20

Sources of selection bias

Answer 20

1.Source Effect- both the treatment and control groups should be uniform. Cannot put all animals from one source into one group and all individuals from another source into another group

2. Non-randomized clinical trials- self selected differences before intervention happens

Question 21

How to get rid of selection bias?

Answer 21

Randomization.
>baseline comparisons helps evaluate effectiveness of randomization

Question 22

Types of randomization

Answer 22

1. Simple- toss a coin, random number tables/generator, Dice
2. Systemic- not true randomization; randomize first animal and then alternate between groups

Question 23

Confounding variables

Answer 23

• factors that may be related to the outcome measure that are not equally distributed between treatment and control groups
  -can lead to a difference in outcome that is not true

Question 24

How are confounding variables accounted for in a clinical trial?

Answer 24

-proper randomization should balance most confounding variables out EXCEPT in small trials

Question 25

Randomization

Answer 25

-produces comparable study groups
-removes investigator bias in participant allocation and guarantees statistical tests

Question 26

Stratification or blocking

Answer 26

-looks to solve confounding variables by dividing the population into different strata according to confounding variables and then use simple randomization in each subgroup

**important for small sizes

Question 27

Matching

Answer 27

-match individuals up based on similarity (age, breed, sex, time of admission) and then on individual is chosen to be control and one to be treatment

Question 28

Performance and exclusion bias

Answer 28

When treatment and control patients receive unequal observation, follow ups and ancillary treatments or their follow ups are lost

Question 29

Problems with ancillary treatment

Answer 29

1. Contamination- control patients accidentally receive the experiment treatment
2. Co-intervention- the performance of additional diagnostic or therapeutic acts on experimental but not control patients

Question 30

Detection bias

Answer 30

-bias occurring with diagnostic decisions because of knowledge of status OR because treatment and control groups process of measurement is different

extremely important in subjective measures

Question 31

Blinding

Answer 31

-Subject and observer are unaware of the treatment group to which subjects have been allocated

Question 32

Single blinding

Answer 32

-only subjects are unaware of which treatment group they are in
-needs to include a placebo

Question 33

Double blinding

Answer 33

both subjects and observers are unaware of treatment groups

Question 34

Triple blinding

Answer 34

subjects, observers, and statistical analysis is blinded

Question 35

What does blinding eliminate?

Answer 35

-performance bias
-exclusion bias
-detection bias