Research authors’ reply to Barlow and Barr and Taylor-RobinsonBMJ 2016; 355 doi: https://doi.org/10.1136/bmj.i6639 (Published 16 December 2016) Cite this as: BMJ 2016;355:i6639
- 1School of Health Sciences, City University London, London EC1V 0HB, UK
- 2School of Medicine, Stanford University, Stanford, CA, USA
A valid point that Barr and Taylor-Robinson raise in their editorial accompanying our research article is that one should distinguish between economic crises themselves and policy responses to these crises.1 2 Nothing in our inclusion/exclusion criteria suggests that studies looking at policies would be excluded. None of the omitted studies Barlow refers to were excluded because they focused on policy responses to the crisis.3 They were excluded because they did not meet our criteria: one was published after December 2015,4 another did not clearly refer to health outcomes,5 and the last focused on changes in out-of-work disability benefits.6
We agree with Barr and Taylor-Robinson that better evidence is needed on policy responses to financial crises.7 In fact, we conclude our paper by arguing that we need more evidence on the different mechanisms by which the financial crisis affects health outcomes more generally. Most of the studies we reviewed could not answer this. Some studies clearly state that disentangling the effect of the crisis itself from that of policy responses is very hard. Regidor et al provide a good example.8 Analysing mortality data from Spain, they found that mortality from traffic injuries declined during the financial crisis. However, whether this is because individuals use their cars less, as the literature on financial crisis would suggest and Vetrugno et al argue,9 or whether it can be attributable “to the effectiveness of road safety policies introduced in the years before or during the crisis” is not something their study can answer.
To conclude, our paper aims at offering a methodologically robust appraisal of the evidence on a topic of general interest and at contributing to a wider debate of recessions and health outcomes. It suggests that evidence should be critically assessed for the degree of bias it may be subject to. Furthermore, it presents a tool to assess the risk of bias that may be useful in a wider range of evidence appraisals on similar topics.
Competing interests: See http://www.bmj.com/content/354/bmj.i4588.