Jackson: History of Controlled Trials

Question 1

Why do a RCT?

Answer 1

• Gold standard for ‘if a treatment/intervention work’
• Reduces confounding biases
• Triangle: people in study
• Circle: exposed and the other side a control
• Outcomes over time

Question 2

put this into the GATE framework

Answer 2

• Triangle (participants): clean elective operations, stratified by risk
• Circle:
  
  • exposed group: IV cefotaxime
  • Control group: identical placebo
• Time: 4-6 wks
• Outcomes: Surgical site infection

Tria was ‘double blind’ and randomised

Question 3

How did they do randomisation of this trial? Why?

Answer 3

Low and high-risk patients were randomly assigned separately by the pharmacist in blocks of four to receive IV Cefotaxime (2gm) or an identically labeled placebo. The AB was mixed in 100ml 5% dextrose in water and was infused rapidly on call for operation. No other ABs were permitted.

When you randomised, by chance you should get ~50% of high and low risk. BUT in small groups, this doesn’t always occur, which is why you should stratify first into risk to avoid confounding by this.

Block randomisation: makes sure you get the same amount in the intervention then the comparison (take ‘blocks’ of 4,6,8 people).

Blinding: patient doesn’t know

Double blinding: both patient and those doing treatment/follow up don’t know.

Question 4

Describe the concept of allocation concealment and double blinding

Answer 4

So that the way people are getting into a control or an intervention group is not controlled/influenced at any point by anyone looking after the patient.

Lots of new drug trials used to have patients put into a trial and not randomise it order to ensure their patient got the new treatment. This stuffed up the process of randomisation. NOW we conceal it also from the doctor, and go through an uninvolved 3^rd party.

“concealed allocation or double blind”

Question 5

What can you see from the baseline characteristics of this RCT?

Answer 5

That the two groups (intervention and placebo) have relatively similar characteristics. From this, we can conclude they were effectively randomised.

Question 6

What are the numbers of this study in a GATE framework.

What is EGO and CGO?

Answer 6

• EGO: a/EG
  
  • a= risk of outcome in the exposed group
  • EG: Exposed Group
• CGO: b/CG
  
  • b= risk of outcome in comparison group
  • CG= comparison group

Although the ‘intention to treat’ group of participants was 316 and 317 (t = 633) the actual ‘on-treatment’ numbers were 312 and 313 as EG lost 3 and CG lost 5. Therefore this was not a massive deal.

EG lost 5 to follow up (f/u) and CG lost 4 to f/u. These are removed from the on-treatment numbers → 308 and 309. This isn’t always done, but it doesn’t matter what number you use in the denominator (on-treatment, f/u or intention to treat) as long as they’re all the same!! But COULD be a big deal to the numerators

Often if the loss to f/u is <20% its not a big deal

Question 7

What happens if instead of one rug/placebo right before surgery, the RCT is a drug that needs to be taken daily? How would you modify you RCT?

Answer 7

Often you have a much smaller ‘on-treatment’ group then an ‘intention-to-treat’ group as people have lower compliance/complications.

The Treatment group

Question 8

Summary: Strengths and Weaknesses of RCT

Answer 8

Strengths (only 1)

• Addresses confounding

Weaknesses

• too expensive: why they make them small
• usually too small (lots of random error) : leads to wide confidence intervals
• Usually not ‘real world’
• poor compliance: terrible for long term studies