A longitudinal impact evaluation of health and safety management in the National Health Service
The research aim was to evaluate whether it is possible to reliably measure change in health and safety performance when a formal health and safety intervention is introduced to the UK healthcare sector. The research methodology and design was informed by a review of relevant literature. Field study data was generated using a six-point design. This included: use of a before-and-after (longitudinal) design, use of comparison groups; use of an intervention that was of interest to participating NHS Trusts, use of a participative style which involved the participating Trusts; use of multiple measurement methods and multiple indicators of effectiveness. Seven NHS Trusts participated; two of which were a control group. The intervention was health and safety management workbook, introduced only to the test group. Evaluation of the impact of the workbook on health and safety performance involved two identical phases, the second approximately twelve months after the first. Each phase consisted of a staff opinion questionnaire survey, based on previously validated work; and a new Health and Safety Executive (HSE) methodology involving management root cause analysis plus cost estimations of incident data. There was wide variation in the numbers of incidents reported within each Trust, although the mean difference between phases was not significant (P<0.1). There was no significant difference between the test and control Trusts (P=0.05). Incident rates were in broad agreement with official HSC data, so far as fatal/major injury rates wee concerned, although there was less agreement for minor incidents. These results suggested that Trust reporting culture may be linked to overall reporting rates but that captured incidents, (which met the project inclusion criteria and therefore represented more serious incidents), were independent of reporting rate, size and type of Trust. The use of incident reporting rates as benchmarking indicators should therefore be used with caution until further work can be carried out to clarify the nature and scope of their limitations. There is also a need to clarify to relationship between major and minor incidents and whether they share similar root causes. This would help to establish whether recording only the more serious incidents with selective root cause analysis would be more cost effective than the processing and analysis of large numbers of incident reports.