Publication bias and time-trend bias in meta-analysis of bicycle helmet efficacy – Freestyle Cyclists

Rune Elvik from the Institue of Transport Economics has recently published an examination of publication bias in helmet research meta-analysis.  From the abstract:

This paper shows that the meta-analysis of bicycle helmet efficacy reported by Attewell, Glase, and McFadden (Accident Analysis and Prevention 2001,  45–352) was influenced by publication bias and time-trend bias that was not controlled for. As a result, the analysis reported inflated estimates  of the effects of bicycle helmets. This paper presents a re-analysis of the study. The re-analysis included: (1) detecting and adjusting for publication bias  by means of the trim-and-fill method; (2) ensuring the inclusion of all published studies by means of continuity corrections of estimates of effect rely on  ero counts; (3) detecting and trying to account for a time-trend bias in estimates of the effects of bicycle helmets; (4) updating the study by  including recently published studies evaluating the effects of bicycle helmets. The re-analysis shows smaller safety benefits associated with the use of bicycle helmets than the original study.

The study is a meta-meta-analysis: it analyses earlier analysis used by others who analysed earlier primary analysis.  As such, it doesn’t offer any new empirical evidence into helmet use, efficacy, risk exposure or the effects of helmet laws.  What it does do however is highlight biases in earlier meta-reviews where a number of primary studies that were not conclusive with their findings were excluded.

After controlling for these biases, Elvik concludes:

  • Bike helmets reduce the risk of head injuries, all other things equal, although not by as much as previously reported.
  • Bike helmets increase the risk of neck and rotational injuries.
  • Older studies show a small postive reduction in injuries when considering head, neck and face as a whole.
  • Newer studies of helmet efficacy “summarised by a random-effects model of analysis, indicate no net protective effect.”
  • The results from studies of the effectiveness of helmet laws are not consistent with the claims of helmet efficacy.  Mandatory helmet legislation doesn’t produce the effects that pro-helmet research predicts it should.

So while no new primary data is presented, this study is still interesting in that it highlights the dilemma for promoters of helmet legislation – not a single study anywhere in the world has shown that helmet laws make cycling safer.  Helmet use may reduce the risk of some injuries in the event of an accident, but helmet laws have other unintented consequences that negate any benefit of helmet use and can actually make cycling more dangerous.

Are helmet laws holding back cycling in Australia? You can do something about it. Write to your State MP and ask them to support helmet freedom.

3 Comments

  • […] Publication bias and time-trend bias in meta-analysis of bicycle … Bike helmets increase the risk of neck and rotational injuries. Older studies show a small postive reduction in injuries when considering head, neck and face as a whole. Newer studies of helmet efficacy “summarised by a random-effects . […]

  • The term ‘personal injury’ covers a broad range of injury and accident
    types – from whiplash sustained from a car accident, to a broken ankle
    from slipping on a pavement, to cerebral palsy caused through medical
    negligence. A personal injury can be physical or psychological, and can
    have long or short term effects.

  • […] more recent study (covered here), also focusing on the effect of helmet use in New Zealand, found that with inclusion of neck […]

Leave a Reply

Your email address will not be published.

Take action today and start enjoying the ride! Read more