Not-so-surprising findings
==========================

* John P.A. Ioannidis
* Panagiotis Papanikolaou

[Two of the authors respond:]

We thank Schneeweiss and Solomon for their interesting comment. We fully agree that nonrandomized studies often include high-risk populations that are excluded from randomized trials. However, this is not an absolute rule. While some of the discrepancies between absolute and relative effect metrics may be explained by differences in baseline risk, this has to be checked on a case-by-case basis. It is also very difficult to reach consensus whether harmful effects are described more appropriately in the absolute or multiplicative scale, so we opted to show both in our evaluation.1 Besides genuine differences in absolute risk, measurement problems and bias should also be considered. Absolute event rates may vary considerably across studies, regardless of design, because of many reasons. These include differences in the definition of the adverse event; the captured range of severity; the threshold of patients and physicians to report (often a reflection to the extent to which they are sensitized); the mode of data collection (in particular, active versus passive surveillance for harms); and whether any efforts at attribution have been made.2,3 In the absence of standardization of collection and reporting of information,4 comparisons of absolute event rates may sometimes remain tenuous. Therefore, while absolute event rates are clinically most meaningful and can be readily translated to numbers needed to harm, relative risks may be somehow more robust.

## REFERENCES

1.  1. Papanikolaou PN, Christidi GD, Ioannidis JPA. Comparison of evidence on harms of medical interventions in randomized and nonrandomized studies. CMAJ 2006;174(5):635-41.
    
    [Abstract/FREE Full Text](http://www.cmaj.ca/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoiY21haiI7czo1OiJyZXNpZCI7czo5OiIxNzQvNS82MzUiO3M6NDoiYXRvbSI7czoyMjoiL2NtYWovMTc1LzIvMTcyLjEuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 

2.  2. Bent S, Padula A, Avins AL. Better ways to question patients about adverse medical events: a randomized, controlled trial. Ann Intern Med 2006;144:257-61.
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=16490911&link_type=MED&atom=%2Fcmaj%2F175%2F2%2F172.1.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000235543100004&link_type=ISI) 

3.  3. Ioannidis JP, Mulrow CD, Goodman SN. Adverse events: the more you search, the more you find. Ann Intern Med 2006;144:298-300.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.7326/0003-4819-144-4-200602210-00013&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=16490917&link_type=MED&atom=%2Fcmaj%2F175%2F2%2F172.1.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000235543100010&link_type=ISI) 

4.  4. Ioannidis JP, Evans SJ, Gotzsche PC, et al; CONSORT Group. Better reporting of harms in randomized trials: an extension of the CONSORT statement. Ann Intern Med 2004;141:781-8.
    
    [CrossRef](http://www.cmaj.ca/lookup/external-ref?access_num=10.7326/0003-4819-141-10-200411160-00009&link_type=DOI) 
    
    [PubMed](http://www.cmaj.ca/lookup/external-ref?access_num=15545678&link_type=MED&atom=%2Fcmaj%2F175%2F2%2F172.1.atom) 
    
    [Web of Science](http://www.cmaj.ca/lookup/external-ref?access_num=000225206900005&link_type=ISI)