The only other difficulty the organizers encountered, besides the unexpectedly high incidence of hazardous toxicity with busulphan in the early stages of the study, was one that is very common to treatment trials: the problem of ensuring that patients took the right number of tablets, not more and not fewer than they had been prescribed.
Among other things, agency investigators found that Boston Scientific, which halted development of the stent last year after a study showed it frequently fractured, had neglected to tell patients in that trial about the problem's scope.
The resequenced data were then aggregated across 40 problem blocks of the 240 total problems, creating a percentage of correct responses per trial on the problem block.
In subtraction trials, the problems had operands of 9 or below, and thus a result of 8 or below.
Undercover Economist, Tim Harford, has written compellingly in his book Adapt, about how various solutions can be found to difficult problems by doing a number of small scale trials investigating different ideas and using the data gathered from the trials to plan more and better trials until the problem has been overcome.
In a recent study, New Caledonian crows only became sensitive to a causal relation after an average of over 100 trials with the problem [20], [21].
Most observational trials had the problem of confounding or small sample size.
If the animal completed fewer than three trials in the problem, the entire problem was excluded from the analysis.
The experience of this carefully designed trial illustrates the problems of designing long-term, randomized trials.
This trial addresses the problems outlined above with an interactive educational intervention for GPs, which is described in detail below.
