Home page

Main topics
News Headlines

Frequently asked Questions
For Policy Makers

Research evidence
Misleading claims
Helmet laws

Search Engine

New Zealand
Other countries

Full index

Policy statement
Register as a supporter

Download this page

Cyclist and pedestrian casualties in Edinburgh 1980 - 2000

Malcolm Wardlaw BSc MBA


IF cycle helmets are effective in preventing death or serious injury in road crashes, one would expect that the road casualty record would reflect this as helmets become    more    common.    Edinburgh    has    been identified    by    Department    for    Transport    (DfT) surveys (Bryan-Brown and Christie, 2001)    as    being    the    city    with    the    highest cyclehelmet  wearing  rate  in  Great  Britain.  Informal observation shows summer wearing rates to be in the 30-40%  band,  rising  to  60-70%  for  commuters  in winter.

Road casualties

The  'official'  road  casualty  record  is  based  on reports  from  the  police,  using  STATS19  forms,  not hospital  records.  This  means  that  the  police  must know about an incident before it can exist officially. It can be assumed that the police will learn of most road accidents   that   were   attended   by   the   emergency services.  On  the  other  hand,  if  you  fall  off  in  a  back street  and  fracture  your  wrist,  it  is  hard  to  imagine why  you  would  bother  to  tell  the  police  about  it.  So while  it  may  be  taken  that  fatal  casualties  are  well recorded, and serious casualties with a high degree of completeness, lesser injuries will be underreported.

This  is  not  as  much  of  a  problem  as  one  might expect.  The  casualty  record  may  be  compared  to  the record of cycling levels as derived by on-road vehicle counts   also   carried   out   by   the   DfT   (the   Traffic Census) (##10024).     This     comparison     shows     remarkable agreement across time. For instance, in the early 1970s the amount  of cycling  recorded was  similar to  today and the number of cyclist casualties reported was also similar to today, about 4 billion kilometres of cycling and   23,000   casualties   (all   severities)   during   both periods.  However,  in  the  early  1970s  there  were  380 cyclist  deaths  per  year,  while  nowadays  there  are only about 130 per year. 

This  reveals  a  pattern  seen  for  pedestrians  and drivers as well; the number of casualties per distance travelled  has  not  changed  much  over  the  years,  but the  number  of  deaths  within  those  casualties  has fallen  greatly.  That  is,  the  probability  of  death  given injury  has  fallen  over  the  last  three  decades.  The causes of this welcome improvement are not obvious, since the fall seen for cyclists has been about the same as  for  drivers,  despite  great  improvements  in  car design   and   cyclists   only   having   begun   to   wear helmets  significantly  in  the  last  few  years  of  the period.  The  author  suggests  that  better  emergency care,  calming  of  traffic  by  congestion  and  improved tyre design are the major contributors, but he is open to other possibilities.

Returning  to  cyclist  data  in  particular,  it  is  a  fact that  cyclist  casualties  are  underreported,  but  they must   be   consistently   underreported   across   time. Otherwise  it  is  hard  to  understand  the  agreement between  reported  casualties  and  reported  cycling levels noted above. This result is not as surprising as it  seems.  Traffic  casualties  have  been  reported  with increasing detail since the Victorian era, while cycling surveys  began  in  1949,  so  both  procedures  were mature by 1970.

Are helmets saving lives?

If  cyclist  deaths  go  down  as  helmet  use  goes  up, does   that   mean   helmets   are   saving   lives?   Not necessarily. As we saw above, cyclist death rates have fallen  greatly  since  the  early  1970s.  Most  of  the  fall occurred  before  1985,  that  is,  before  helmets  were  in use  in  Britain.  This  illustrates  a  difficulty  –  there  are safety    improvements    that    happen    anyway    for complex   reasons.   How   do   we   distinguish   these natural improvements from any additional effect that might be due to helmets?

A  method  is  here  proposed.  As  hinted  at  above, experience    shows    that    all    road    users    (except motorcyclists)  are  about  equally  affected  by  changes in  road  conditions.  Pedestrians  and  cyclists  follow very well matched injury trends, as the graphs below of  cyclist  and  pedestrian  casualties  in  Edinburgh show.  As  cyclist  casualties  are  rarely  older  than  65 years,  these  have  been  omitted  from  the  pedestrian data.

Take  a  little  time  to  appreciate  how  these  graphs have  been  derived.  Fatal  and  serious  casualties  have been added together and divided by the total number of   casualties.   This   gives   us   a   percentage.   This percentage  indicates  the  risk  of  suffering  at  least serious injury in  a road accident that became  known to  the  police.  So  if  there  were  2  cyclist  deaths,  15 serious injuries and a total of 100 injuries of all kinds that  became  known  to  the  police,  the  Severity  of Injury would be 17%. 

In   both   graphs   there   is   a   high   degree   of correlation  between  cyclist  and  pedestrian  injuries, and especially so for adults. Pedestrians suffer higher severities of injury than cyclists, both for children and for  adults.  For  child  casualties  the  data  are  more variable   because   the   numbers   involved   are   low. During  the  era  before  helmets  were  in  common  use, say before 1993, the injury trends of both cyclists and pedestrians  improved  in  nearly  identical  patterns. That is, although the absolute changes in injuries are large,     the     relationship     between     cyclists     and pedestrians is pretty constant. If we divide the cyclist data  by  the  pedestrian  data,  we'd  get  a  pretty  flat relationship.  This  is  where  we  might  be  able  to observe a helmet effect if one exists. 

After  1993,  helmets  became  progressively  more common in Edinburgh, reaching the quite significant levels  of  use  seen  today.  Can  we  see  any  evidence that  cyclists  have  ‘got  ahead’  of  pedestrians  in  these years?    It  does  not  look  like  it.  If  anything,  cyclists appear  to  be  worse  off  relative  to  pedestrians.  The matter   becomes   clearer   if   we   divide   cyclists   by pedestrians  as  described  above  to  see  the  relative trends (Figure 3).

For adult cyclists, there is a discernible decline in relative  injuries  during  the  1980s,  which  flattens  out after  the  early  1990s,  with  no  further  improvement. There  may  even  have  been  a  slight  decline.  The history  for  children  is  again  more  erratic  because  of the small data number, but there is a decline apparent from  about  1995.  That  is,  cyclist  injuries  became worse relative to pedestrian injuries after 1995.

From these real world results, it would be hard to accept  that  the  use  of  helmets  by  Edinburgh  cyclists has had any beneficial impact on improving reported injury  trends.  For  adults  there  was  no  improvement during  the  time  of  growth  in  helmet  use,  rather, pre-existing improvement came to and end, whilst for children  there  was  deterioration  for  cyclists.  Such  a result is at odds with government policies to promote helmets  to  reduce  reported  serious  casualties  –  and bear  in  mind  that  helmet  use  in  Edinburgh  is  well ahead of the national situation.

One may add this result to others: for London (LRC, 1997) , for Great Britain as a whole (Wardlaw, 2000), for the USA (CPSC, 2001) , Nova Scotia (CMAJ 2002) and  Western  Australia (Robinson, 1996) .  It  is  disturbing  to  find  yet another     instance     of     worsening     injuries     with increasing  helmet  use,  particularly  as  cyclists  were doing better than pedestrians before helmets were an issue.

An analogy

Consider the following situation. You are walking along a fairly wide cliff top path, but come to a stretch only  a  foot  wide,  so  that  to  proceed  you  are  forced right up to the edge. Think about how that would put you  ‘on  edge’.  Then  the  path  widens  out  again  and you  can  relax.  It  is  the  proximity  of  danger  –  or  the perceived   proximity   of   danger   –   that   controls behaviour, to a degree that is impossible to resist for any   length   of   time.   It   would   require   enormous self-control to amble along the very edge of a cliff as nonchalantly as if one had a clear metre of space.

Now    consider    another    situation.    You    are blindfolded  and  led  along  for  a  while  by  someone you trust – imagine you are gripping the back of their belt. After a time, your guide remarks you are close to the  edge  of  a  cliff,  and  indeed,  you  can  hear  the breakers  far  below.  Now  the  pair  of  you  stop.  You remove the blindfold. Imagine your shock to discover that  you  have  been  led  along  the  very  edge  of  the precipice by this person you thought you could trust. 

That  is  illustration  by   exaggeration.  Although blindfolded,  you  were  doing  just  what  you  might have done in full knowledge; walking along the edge of a cliff. The difference was in the mood – the degree of tension or preparedness to react to an emergency.  

An  advocate  of  the  wearing  of  cycle  helmets  has stressed:   “The   argument   that   cyclists   who   wear helmets  take  greater  risks  than  those  who  do  not  is rubbish…  I  ride  safely  at  all  times.  I  now  feel  naked without my helmet…”  (BikeBiz 2003).

I  have  no  reason  to  doubt  that  this  rider  does nothing with a helmet on that he would have avoided bareheaded,  but  by  his  own  admission  wearing  a helmet  alters  his  mood  –  his  choice  of  the  word ‘naked’  suggests  he  feels  less  exposed  with  a  helmet on. And it is very hard to accept that such alteration of mood will not dull reaction times by reducing the degree  of  tension.  The  difference  need  not  be  great. At  20  mph  a  cyclist  will  cover  six  feet  in  a  fifth  of  a second.  Now  add  the  complication  that  this  person clearly believes his helmet will provide protection far in excess of anything seen in the real world… “being led along by someone you thought you could trust”…A false sense of security is a dangerous thing.    

A cycle helmet is designed to prevent injury only in  a  simple,  low-speed  fall  at  12  mph  or  less.  Yet  the  message  is  so  often  put  about  that  'helmets  save lives'. Could it be that this exaggeration has created a faith  in  the  effectiveness  of  helmets  that  cannot  be delivered,  and  the  result  has  been  to  make  matters worse?

It  has  been  observed  that  alteration  of  behaviour has  limited  the  benefits  of  seat  belt  use,  and  put cyclists  in  more  danger  at  the  same  time  (Richens, Imrie and Copas, 2000).  It  is  thus not   implausible   that   too   much   faith   in   helmet effectiveness  could  lead   to  such   an  alteration  of behaviour as to make the situation worse, not better. At the very least, it is clear  that cycle helmets are not improving cyclists' reported injury trends.


BikeBiz 2003

Bicycle Business. 2003.

Bryan-Brown and Christie, 2001

Bryan-Brown K, Christie N, 2001. Cycle helmet wearing in 1999. Transport Research Laboratory Report 487.

CMAJ 2002

Butting heads over bicycle helmets. Canadian Medical Association Journal 2002;167:337-8.2002.

CPSC, 2001

Consumer Product Safety Commission. Reported New York Times, 29 July 2001.

LRC, 1997

Pedal cyclist casualties in Greater London. London Research Centre Factsheet 76, August 1997.


Road Accidents in Great Britain: the casualty report. Department for Transport, London. Various years..

Richens, Imrie and Copas, 2000

Richens J, Imrie J, Copas A, 2000. Condoms and seatbelts, the parallels and lessons. The Lancet 2000;355(9201):400-3.

Robinson, 1996

Robinson DL, 1996. Head injuries and bicycle helmet laws. Accident Analysis & Prevention 1996 Jul;28(4):463-75.

Wardlaw, 2000

Wardlaw MJ, 2000. Three lessons for a better cycling future. BMJ 2000-12-23/30 n321 p1582-5.