Please enable JavaScript.

Coggle requires JavaScript to display documents.

CT2: Randomisation (Introduction (2 main issues in RCT design: (Avoidance…

- - - - Sufficient numbers need to be recruited (covered in another session)
    - - Selection, care, evaluation of pts needs to be the same in each group so that comparison is unaffected by anything except the treatment itself
      - Randomisation avoids bias in selection of pts - can also ensure blinding
- - - - e.g. if block size of 10 used, within block there will be 5 in each arm.
      - Treatment allocations ordered randomly within each block
  - - - e.g. RITA-2 varied between 6 and 4 sized blocks.
- - - - Sealed envelopes
        
        Problem: One treatment might not be preferred investigator option, so steaming open/reordering envelopes might occur
        
        Therefore, now rarely used
      - Telephone randomisation
        
        Investigator calls registration office, provides necessary details and treatment allocation given over the phone from randomisation list.
      - Web-based randomisation
        
        Secure website enter required details. Randomisation then revealed to investigator by text message or email.
  - - - Necessary to involve the pharmacist in the randomisation process
        
        Prepare drug packages in advance according to randomisation list and packages treated in similar manner as sealed envelopes.
        
        Pharmacist might be the one responsible for contacting trial office for treatment allocation
  - - - Telephone call assignment by Royal Free Hospital in London where sealed envelopes were held.
    - - Recruited during field visits, needed to provide therapy at time of randomisation. Drug packages pre-prepared by pharmacist of company responsible for providing drugs according to pre-prepared randomisation list.
- - - - Decided in advance that e.g. any imbalance>3 then next allocation uses treatment probability 2/3.
  - - - Before any pts recruited
- - - - Stopped after 12 babies, 11:1 ECMO:standard
        
        Inconconclusive and lots of debate. Potential for bias.
  - - - Sometimes unequal numbers, e.g. to gain more experience of new treatment, increased interest of investigators, economic reasons (e.g. one expensive treatment).
  - - - Pts randomised to 1) don't seek consent, 2) Seek consent - will you accept treatment B - then do analysis of groups 1) and 2) regardless of if accepted treatment B
- - - - Need to recognise difference between subjective belief and objective knowledge
        
        Observational studies often contradicted by RCT results
- - - - 2 key principles:
        
        Each pt has a known (usually equal) chances of each treatment option
        
        Neither pt or investigator can predict in advance the treatment to be allocated
        
        i.e. allocating treatment based on DOB or alternate allocation is not random and this might affect decision to recruit a pt
        
        i.e. could bias results
        
        Allocation concealment different to blinding - allocation concealment is ensuring that the pt and investigator are blinded to potential allocation before randomisation
  - - - Pt eligible for trial inclusion
        
        Clinician willing to accept randomisation
        
        Pt consent obtained
        
        Pt formally recruited
        
        Treatment assigned from randomisation list
        
        CRFs completed
        
        1 more item...
        
        Study no. often assigned now - pt needs to be registered in trial before treatment assignment
        
        Also useful to collect log of people who weren't entered and reasons why
        
        Not always easy - CRASH trial corticosteroids after head injury and impairment of consciousness. In these pts informed consent not easy. Delays not acceptable as emergency situation and treatment shouldn't be delayed. For CRASH Dr in charge took responsibility for entering pts.
        
        Before entering any pts
        
        Clinician can't have strong belief that a treatment option is unacceptable or they will find it difficult to enter pts
        
        Need to avoid inclusion of ineligible pts so eligibility should be checked early
      - Ensures pt entered in ethical and efficient way
      - Important to work out how to identify pts
        
        To ensure that pts are representative of population whom results will be applied to
- - - - Streptomycin and bed rest vs bed rest alone
        
        sealed envelope randomisation
        
        Independent evaluation after 6 months' treatment of pts' x-ray film, blinded to treatment arm
- - - - e.g. allocate treatment A for any numbers 0-4 and B for 5-9
        
        Arbitrary starting point in list should be chosen and then list of numbers produced
        
        In practice this method is rarely used because could lead to real possibility (more than 0.05) of very unequal numbers in each arm when lower pt numbers (e.g. 6:14/ 86:114)
        
        Even in larger trials, if there is interim analysis this issue can arise too, while recruitment still in progress
- - - - No randomisation list prepared but continually updated record of treatment by pt factors kept.
    - - If equal sums then simple randomisation done