Methods of determining the MID Minimal clinically important difference

1 methods of determining mid

1.1 distribution-based methods
1.2 anchor based
1.3 delphi method

methods of determining mid

several techniques calculate mid have been described , can subdivided in 3 categories: distribution-based methods, anchor-based methods , delphi method. there no consensus regarding optimal technique, distribution-based methods have been criticized. example, use of standard error of mean (sem) based on anecdotal observations approximately equal 1/2 sd when reliability 0.75. revicki et al. question why 1 sem should have mid? sem estimated product of sd , square root of 1-reliability of measure. sem used set confidence interval (ci) around individual score, is, observed score plus or minus 1.96 sems constitutes 95% ci. in fact, reliable change index proposed jacobson , truax [12] based on defining change using statistical convention of exceeding 2 standard errors (p. 106).

distribution-based methods

these techniques derived statistical measures of spread of data: standard deviation, standard error of measurement , effect size (which based on standard deviations)

anchor based

the anchor based method compares changes in scores “anchor” reference. anchor establishes if patient better after treatment compared baseline according patients own experience.

a popular anchor anchor question, @ specific point in time after treatment patient might asked: ‘‘do feel improved treatment?’’. answers anchor questions vary simple “yes” or “no”, in between, e.g. “much better”, “slightly better”, “about same”, “somewhat worse” , “much worse”. differences between answered “better” , answered “about same” form basis of anchor method.

an interesting approach anchor based method establishment of anchor before treatment. patient asked minimal outcome necessary undergo proposed treatment. method allows more personal variation 1 patient might require more pain relief, strives towards more functional improvement. different anchor questions , different number of possible answers have been proposed. there no consensus on 1 right question nor on best answers.

delphi method

the delphi method relies on panel of experts reach consensus regarding mcid. expert panel provided information on results of trial , requested provide best estimate of mcid. responses averaged, , summary sent invitation revise estimates. process continued until consensus achieved.