Randomised Control Trial

Randomised Control Trial

1. Questions It Can Answer

   • Treatment Effects
     • does SMT reduce pain to a greater extend than exercise in nsLBP?
     • comparison groups

2. Design & Method

   • treatment and comparison groups
   • Randomisation
     • should use a computer generated random no. set
     • should not be visible to researches
       • no. given out by a person not involved in rest of study
     • results
       • should give equal characteristics between groups
   • Blinding
     • do the patients know what treatment they are getting
     • do those assessing the outcome know what treatment groups
   • Measuring outcomes
     • must be
       • appropriate
       • valid
       • sensitive
       • reliable
   • Statistical analysis
     • statistically significant p<0.05
     • e.g. no of people improving
       • A = 80 out of 100 people
       • B = 30 out of 100
       • RISK OF IMPROVING
         • A = 80% (0.8)
         • B = 30% (0.3)
         • RISK RATIO
           • 0.8/0.3 = 2.66 times more likely to improve in A compared to B
           • ABSOULTE RISK REDUCTION (ARR)
             • 0.8-0.3 = 0.4
     • Statistical Generalisation
       • 95% Confidence interval
         • e.g.
           • difference from baseline to follow up (A) = 6 B = 2
             • difference in change in pain between groups = 4 in favour of A
               • 4 (-2 to 7) not sig as crosses 0
               • 4 (2-7) sig as does not cross 1
   • Conclusions that represent the results and quality of data collection
   • Sample population
     • inclusion/ exclusion criteria
     • study setting
     • must know in order to generalise
     • the groups
       • details of treatment and comparison group
   • Drop Out
     • loss of no means less statistical strength
     • if particular people drop out then groups become different on average
       • undermines initial randomisation

3. Conclusion

   • should put their results in context
     • if everyone else is different have to provide a strong argument for why they found what they did
     • have to provide a discussion of limitations
       • generalizability
       • effect of dropouts
       • failures to blind

4. Critical Appraisal

   • 1a) Was the assignment of patients to treatment RANDOMISED ?
     • centralised computer randomisation = ideal
       • smaller trials may use independent person
     • Methods
   • 1b) Were the groups SIMILAR at the start of the trial?
     • More similar the better
     • Results
       • should have a table of baseline characteristics comparing randomised groups on a no. of variables that could affect outcome (e.g. age)
         • if not, description of group similarity in first para of results
   • 2a) Aside from allocated treatment were the groups treated equally?
     • Methods
       • follow-up schedule and permitted additional treatments etc
         • Results
           • for actual use
   • 2b) Were all patients who entered the trial accounted for? and were they analysed in the groups to which they were randomised?
     • losses should be <20%
       • and reason
     • Results
       • how many p were randomised and how many actually included
   • 3) Were measures OBJECTIVE or were p and clinicians kept BLIND to which treatments received?
     • ideal if double blinded
     • if outcome is objective (e.g. death) instead of subjective (e.g. symptoms/ function) then blinding is less critical
     • Methods
   • Results
     • Relative Risk Ratio
       • Risk out outcome in treatment group/ risk in control group
         • e.g. RR = 0.10/ 0.15 = 0.67
           • <1 = treatment decreases risk of death
       • how many more times more likely it is that an even will occur in the treatment group relative to control
         • RR of 1 = no difference between 2 groups
           • = treatment has no effect
         • RR <1 = treatment decreases risk of outcome
           • RR >1 = treatment increased risk out outcome
     • Absolute Risk Reduction (ARR)
       • Risk of outcome in control group - risk of outcome in treatment group
         • ABSOLUTE RISK DIFFERENCE
         • e.g. ARR = 0.15 - 0.10 = 0.05 or 5%
           • absolute benefit of treatment is a 5% reduction of death rate
       • Absolute difference in rates of events between 2 groups and gives an indication of baseline risk and treatment effect
         • ARR of 0 = no difference between groups = treatment has no effect
     • Relative Risk Reduction (RRR)
       • ARR / risk of outcome in control group
         • or
           • 1 - RR
             • e.g. 1 - 0.67 = 0.33 or 33%
         • e.g. 0.05/ 0.15 = 0.33 or 33%
       • tells the reduction in the rate of the outcome in the treatment group relative to that in the control
         • e.g. treatment reduced the risk of death by 33% relative to that occurring in the control
     • Number Needed to Treat (NNT)
       • 1 / ARR
         • e.g. 1 / 0.05 = 20
       • represents no. of patients we need to treat with the experimental therapy in order to prevent 1 bad outcome and incorporates the duration of treatment
         • need to treat 20 people for 2 years in order to prevent 1 death
     • Will the results help locally?
       • can they be applied to the local population
       • were all clinically important outcomes considered?
         • if not, does this affect the decision
       • Are the benefits worth the harms and costs?