13.07.2015 Views

Evaluating non-randomised intervention studies - NIHR Health ...

Evaluating non-randomised intervention studies - NIHR Health ...

Evaluating non-randomised intervention studies - NIHR Health ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Empirical estimates of bias associated with <strong>non</strong>-random allocation5020105Odds ratio210.50.20.10.050.02RCTsCCsFIGURE 10 Comparison of distribution of results of 10,000 concurrently controlled <strong>studies</strong> and 10,000 RCTs resampled from 10 UKcities within the ISTcomparable RCTs. Again, the increase in variabilityof results for concurrently controlled <strong>studies</strong> wasevident, the distribution of results being 1.8 timeswider than for RCTs, slightly less than for theregional IST comparisons (Table 15). Some45% of concurrently controlled results werestatistically significant compared with 13% ofRCTs, the significant results being evenlydistributed between benefit (22%) and harm(23%).1002020105Odds ratio210.50.20.10.050.020.01RCTsCCs54FIGURE 11 Comparison of distribution of results of 8000 concurrently controlled <strong>studies</strong> and 8000 RCTs resampled from eight regionswithin the ECST

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!