Thursday, May 5, 2011

Assessing the Efficacy of Preventive and Therapeutic Measures – Randomised Trials

Gordis: Chapter 7 and 8

Chapter 7:

 

Allocation of subjects to treatment groups

-          alternatives to randomization

o   studies without comparison

§  case study / case series

o   studies with comparison

§  historical controls

·         comparison group from the past

o   quality of data collection

o   can't be sure difference is due to treatment or other factors

·         can be useful à disease is uniformly fatal

o   decline in case fatality that parallels use of drug – strong support that drug has effect

§  simultaneous non randomized controls

§  randomization

·         can't just match on important variables

o   can only match on variables we know and can measure

o   unknown variables

·         stratified radomisation

o   end up with 2 randomisaed groups, but having initially stratified the groups, we increase the likelihood the two groups will be comparable on stratified variables

-          prognostic profile at entry

-          masking / binding

-          crossover

o   each patient serves as his own control

§  repeated measures

-          factorial design

-          non compliance

o   people who do not comply or who do not participate in studies differ from those who do comply and who do participate.

 

 

Chapter 8 : randomized Trials – Some Further Issues

 

-          sample size

o   difference in response rates to be detected

o   estimate of the response rate in one of the groups

o   level of statistical significance

o   value of power

o   one / two sided

 

 

Ways of Expressing Results of Randomised Trials

 

-          Efficacy =

o   ( rate for thos who received placebo – rate for those who received treatment ) / rate for those who received placebo

-          ratio of the risks  à relative risk

 

 

-          number of patients who would need to be treated

o   1 / ( ratein untreated group – rate in treated group)

 

Generalisability of Results

 

-          internal validity whether study was well done and whether the findings are valid

-          external validity – generalisability

 

 

-          problem of using intermediate measures as endpoints of effectiveness

o   eg, because mortality takes years to manifest

o   vs

o   blood pressure / cholesterol levels

 

 

-          publication bias

 

No comments: