Informa Markets

Author Bio ▼

Safety and Health Practitioner (SHP) is first for independent health and safety news.
April 27, 2010

Get the SHP newsletter

Daily health and safety news, job alerts and resources

Can construction injury rates improve or have we hit a plateau?

The UK construction sector has reduced its injury rate significantly, but can it be sustained? Following on from his article in the March issue of SHP, Andrew Townsend argues that further improvement is unlikely unless we find out what leads to a fall in injury rates.

In 2008/09, 180 people died in UK workplace accidents – 53 of whom worked in construction.1 The Accident Frequency Rate (AFR) for reportable construction injuries was 0.4 per 100,000 hours worked. Both are rather alarming statistics – or are they?

For a construction working group of 10 people and a foreman expending 22,000 hours a year, the expectation will be one reportable injury every 11 years. In that same period there will be approximately 27,000 deaths and 290,000 major injuries as a result of UK road-traffic accidents, according to statistics on the Department for Transport website. The grim reality is that the average construction worker is more likely to know someone who has been killed, or seriously injured in a road accident than someone who has had a work-related reportable injury.

To demonstrate that safety is improving, a minimum number of injuries is required for each point of data to be statistically significant. Based on both probability theory and practical experience, it can be argued that this minimum number is about 10 injuries.
When applied to an AFR of 0.4, an average data point requires 2.5 million man-hours to be valid (10/0.4 x 100,000). However, 2.5 million man-hours a year is equivalent to a workforce of 1250, which is far larger than most construction companies. So, without evidence that small-to-medium enterprises can relate to, how can they be motivated to take relevant action?

In the last 26 years the rate of fatal injuries has reduced threefold,1 with a similar reduction in non-fatal injuries. But what has actually helped bring about this reduction? Is it enforcement, compliance, inspection, safety training, skills training, management systems, better technology, the use of various cultural/climate/behavioural approaches, leadership, or engagement? Or is it a combination of some or all of these?

If it is a combination, what are the relative contributions of these factors? And which ones are specific to each type of construction?

Less is more

In academic writing there has been a dramatic increase in the number of papers that refer to “effective safety”. Yet, more than 90 per cent of these papers do little other than speculate on theories and research methods that might be used but have not been tested.

Similarly, research conducted on behalf of the HSE increasingly features ‘expert’ opinion from a number of organisations and people – and, as a consequence, more words are being generated. Involving people with experience is no bad thing in itself, provided that it leads to clarification and simplification. However, the sheer volume of words and/or the absence of simple, practicable wording would indicate that this has not occurred.

As the situation now stands, health and safety research has produced a bewildering array of different interpretations. The difficulty this creates in determining the relative importance of factors in reducing injury rates is that there are more possible explanations than there are available data to examine these theories. Without the data to identify what works (and, by elimination, discard what does not work), the study of reducing risk will simply lead to what I would term as “explanation overload”.

But is health and safety the only research discipline saddled with the problem of how to access sufficient data, or are there others?

Borrowing ideas from others

A re-examination of source data from process-plant construction shows very strong correlations between labour productivity and injury rates. In my own unpublished internal company research, I observed a similar correlation between injury rates and schedule ratio (i.e. the planned time taken to complete a project, divided by the actual time taken to complete it).

Productivity and schedule data are comparatively easy to access and are more frequently available than injury data. It is then an easy exercise, using linear regression, to convert the various measurements of productivity, or schedule, into a calculated ‘indicated frequency rate’, which can be subjected to validation tests.

Based on initial trials, I estimate that using measures of either productivity or schedule in lieu of injury rates may increase the yield of useable numerical data by 300-400 per cent. This will allow the performance of individual construction projects to be used as data points. Previously, an aggregated company performance was required to produce just one data point.

Multivariate research

The next problem is how to figure out which of the many potential factors behind improving injury rates in construction are really having an effect.

Studies have already been carried out in other industries around the world, including nursing,2 forestry,3 and shipping,4 and have succeeded in relating a number of risk-reduction measures to injury rates. Although small in number, they are indicative of what could be achieved in construction. Translating these successes involves a number of key elements:

  • Statistical skills, methods and tools – those required to interrogate the many numbers involved in ‘multivariate’ research already exist in a number of other disciplines. Arguably, the disciplines that are closest to research into safety risk are medical and epidemiology statistics. These offer a wealth of established techniques5 that can be customised for use in studying injury rates.
  • Input and reviews by experienced experts – statistical procedures process numbers but they neither specify which numbers to process nor interpret them. However, experienced individuals can deduce from statistics what research questions need to be identified and what outputs should be tested.6 The willingness of companies, academics, consultants and workers to provide input into research is evident by the list of contributors to many studies conducted on behalf of the HSE.
  • Recruiting participants – the gains in efficiency of data-gathering enables unique construction types to be studied separately and allow theories of what does and doesn’t work to be validated.7 This will lead to a shift in emphasis from a small number of large studies to a large number of small studies. Persuading organisations to participate in research could prove to be the stiffest challenge. Unlike benchmarking, which looks at ‘best practices’, multivariate analysis requires analysis of patterns of data across an entire range of performance – from the good and the best performers down to the average and poorest.

New research skills

The switch from large studies to a greater number of smaller studies necessitates a different mix of research skills and motivations:

  • Negotiating and good faith – the individual participants must want to get involved. Researchers must have the ability to sell the benefits of research and to establish a working relationship based on integrity.6
  • Strict ethics – researchers will be working with poor as well as good performers, and with companies in competition with each other. Anonymity and preservation of potentially sensitive source data are imperative.8
  • Adaptability – each type of construction manages itself in slightly different ways, and, consequently, the performance indicators that help calculate an ‘indicated injury rate’ will vary. The researchers have to be capable of developing and validating an ‘indicator of injury rate’ unique to each construction discipline. Skill in ensuring that any confidential data from companies are untraceable is also a must.
  • Confidence and courage – researchers need to use their evidence not only to identify what reduces injury rates but also to challenge what does not. People involved will have to be forthright and hands-on,6 capable of collaborating with participants while still maintaining a professional approach.


There is no doubt that the UK’s different construction industries have reduced injury rates significantly. What is uncertain is whether the rate of reduction can be maintained. Unless we can find out what really reduces injury rates, ever-increasing resources will be committed to chasing fewer gains. The result will be a slowing of the process of improvement until it grinds to a halt.

Because of its easier-to-access key performance indicators and unique sub-industries, construction is easy to study. It has much to offer the wider world of safety, both in finding out what is effective safety and in testing different research methods. Rather than being treated as a pariah industry in need of correction, it should be treated as a range of industries from which others can learn.

This leads me to my concluding thought. The disdain that the general public and the media have for the health and safety profession and regulators is palpable. To many of them, quite simply there is too much health and safety. They may have a point, as there are more health and safety measures than can be validated on current numerical evidence. Without adequate ways of demonstrating which actions reduce injury rates, and by how much, there is no mechanism for eliminating what does not work.

Regulation is already under notice from a potential future Conservative government. Unless we get a handle on this issue, there is a risk that occupational health and safety will either grow unchecked and collapse under its own weight, or be deliberately dismantled.


1  HSE (2009): SMT/09/80 ‘An update on the work of HSE’s Construction Division’, a paper by Anthony Lees
2  Mark, B A, et al (2007): ‘Does safety climate moderate the influence of staffing adequacy and work conditions on nurse injuries?’ in Journal of Safety Research, Vol 38, Issue 4, 2007, pp431-446
3  Varonen, U and Mattila, M (2000): ‘The safety climate and its relationship to safety practices, safety of the work environment and occupational accidents in eight wood-processing companies’, in Accident Analysis & Prevention, Volume 32, Issue 6, 2000, pp761-769
4  Lu, C-S and Tsai, C-L (2008): ‘The effects of safety climate on vessel accidents in the container-shipping context’, in Accident Analysis & Prevention, Vol 40, Issue 2, 2008, pp594-601
5  Armitage, P and Berry, G (1974): Statistical Methods in Medical Research, Blackwell Science
6  McNiff, J and Whitehead, J (2006): ‘Action Research’, Sage, pp87-92; and pp186-187
7  Robson, C (1993): Real World Research, Blackwell, p46
8  Gray, D E (2009): ‘Doing Research in the Real World’, Sage, pp438-439

Andrew Townsend has worked in engineering construction for the majority of his career.



The Safety Conversation Podcast: Listen now!

The Safety Conversation with SHP (previously the Safety and Health Podcast) aims to bring you the latest news, insights and legislation updates in the form of interviews, discussions and panel debates from leading figures within the profession.

Find us on Apple Podcasts, Spotify and Google Podcasts, subscribe and join the conversation today!

Related Topics

Notify of

Inline Feedbacks
View all comments