Members Only
Home
Cart
Contact
Workplace injuries and property damage incidents-and the safety and health programs designed to prevent them-are expensive facets of contemporary mining, manufacturing, agricultural and other industries. National Safety Council estimates that
the cost of workplace injuries totaled approximately $156 billion in 2004 (NSC website).
Industries across the world commit significant resources to activities and interventions intended to help prevent these injuries. One very expensive resource extensively used in this battle is human resource hours. In fact, some companies allocate as much as one-third of their available human resource hours to the implementation of safety and health program interventions (Haight, Thomas, Smith, et al, 2001b). If one could determine which activities have the greatest effect on incident rates and to what extent they have this effect, one could optimize the allocation of human resource time to minimize time spent while minimizing incident rates.
Intervention Effectiveness
Organizations whose activities involve the risk of injury or destruction of property commit human and financial resources to intervention activities intended to prevent injuries, fires, spills, unplanned shutdowns, etc. These activities demand a significant human resource commitment. However, little is done to determine whether these activities are correct, effective or of acceptable quality, or are applied in the right quantity. Past intervention effectiveness research has focused on single interventions, return on investment of individual interventions and non-quantitative, subjective survey-based evaluations. Prior to 2001, there were no published studies in refereed journals quantitatively analyzing the effectiveness of a complete, integrated, interactive safety and health intervention program (Haight, et al, 2001b).
Optimizing intervention or injury prevention strategies to decrease the risk of injury and property damage with less costly safety and health programs would contribute to improved productivity and economic vitality in organizations subject to such risks. A problem with this human resource hour allocation is that companies do not adequately or appropriately measure the effectiveness of the prevention activities to which they commit significant resources. Without true measures or appropriate quantification, design and optimization of resource allocation is not possible.
Incident rates long have been the measure of safety and health program effectiveness. These are often referred to as "lagging indicators." In recent years, companies have gravitated to what many call a more "proactive approach" as they develop ways to measure "leading indicators." Various quality aspects of incident prevention activities are determined, such as attendance rates at safety meetings,
at-risk versus safe behavior scores, compliance rates with physical facility inspections, corrective action implementation rates and time to implement corrective measures, etc. These can be potentially good measures of safety and health program effectiveness.
However, both the lagging-indicator and leading-indicator advocates stop short of truly determining the effectiveness of their safety and health programs, and as a result may be potentially wasting resources by allocating them to the wrong activities or overly allocating them to prevention activities that have little effect on incident rates (Haight, et al., 2003).
One key to truly understanding the effectiveness of a safety and health intervention program is to understand the mathematical relationship between the leading indicators (the quantity and quality of the incident prevention activities) and the lagging indicators (the incident rate). Once one has defined this relationship, the safety and health program can be engineered to produce predictable results.
This is not an easy process, particularly because it requires extensive data. The process involves quantifying human activities. The analysis takes time, and many aspects of the program that indicate effectiveness are difficult to measure.
Since it is so difficult to quantify these aspects, research in this area is lean. However, several studies conducted in the last 6 years show positive results, which indicates that once this mathematical relationship is defined, not only can one design a program in terms of allocating specific amounts of resources to particularly effective activities, but also one can even begin to forecast incident/injury rates with a high level of certainty (Haight, et al, 2001a, 2001b; Haight and Thomas, 2003; Iyer, Haight, Del Castillo, et al, 2004; Iyer, Haight, Del Castillo, et al, 2005).
Safety and health intervention programs are complex, integrated and dynamic systems involving implementation of many intervention activities. They involve workers and management team representatives who interact and spend time, energy, thoughts and actions in the name of preventing injuries and other types of incidents.
Many management teams use incident rates as a measure of safety and health program performance. If incident rates are high, intervention activity and effort increases. As the incident rate drops- presumably in response to the increased level of intervention-the level of intervention activity declines as well. This leads to high incident rate variability and creates a setting in which available work hours are at the mercy of an incident rate when most would want the intervention activity to control the incident rate.
This type of program is not designed and in many cases it is not even a planned approach. Many factors influence the outcome of a safety and health program, and it is essential that an organization understand these factors and exactly how they influence the program. It is critical to understand the interrelationships between and among intervention activities. The current approach taken in most organizations does not include provisions for understanding the mathematical relationship between the intervention activities and the incidents those interventions are expected to prevent (Haight, et al, 2003).
An organization must first determine and identify the components of its safety and health intervention or injury prevention program. Then, one must determine how to quantify these intervention activities in terms of both quantity and quality.
An expedient measure to start with is to determine on a weekly basis what percentage of the organization's available human resource hours are committed to implementing each activity. This has been determined to represent a controllable variable that is changed on a regular basis by management with the goal of improving prevention performance or redirecting attention to another aspect of the program. In essence, it is a controllable variable that can be measured and it is a measure of effort expended in the name of preventing injuries.
Quality measures are a little more
difficult to determine and quantify; however, some examples are safety training scores, safety meeting attendance rates, at-risk versus observed safe behavior rates and inspection compliance rates. Each of these varies over time and because of the variation an effect on incident rates is expected, albeit an undefined or undetermined effect at this time (Iyer, et al, 2004).
Once the variables to be measured have been determined (both independent or leading indicators and the dependent or lagging indicators), the organization must begin to collect hourly, quality and incident rate data on some frequency; weekly works well experimentally. The collection must continue for a long enough period to provide enough data to yield statistically significant results (usually 1 year or so if a weekly data collection period has been determined).
Each week, as the leading indicators vary in both quantity and quality, the analysts must determine the amount of variation generated in the lagging indicator due to the variation in the leading indicators. It is this variation that helps to define the mathematical relationship between the two. This variation is measured through various statistical techniques such as multivariate regression analysis, response surface analysis and analysis of variance.
Because intervention activity (leading indicators) are not expected to have an immediate or lasting effect, one can determine the delay and carryover effects through exponential smoothing and moving average techniques to measure the goodness of fit of the curve (equation representing the mathematical relationship between the leading and lagging indicators) over a longer period of time than 1 week only.
With a defined mathematical relationship representing the safety and health program, one can determine which intervention or prevention activities are having the greatest effect on the incident rate and how many of the available human resource hours should be committed to implementing those activities in order to minimize the incident rate.
When the safety and health program has been redesigned in response to the output of this mathematical model, one should continue to collect the same data to confirm that the redesign is actually yielding minimum incident rates as the model predicts. This validation will take another 6 to 12 months to gather enough data. Once this model is working, its performance, in terms of statistical uncertainty, should improve as more data are collected over time.
An organization can also use this mathematical relationship and incident rate history to develop a forecasting equation that will allow the analyst to predict future incident rates with reasonable certainty (Iyer, et al, 2005).
Incidents are costly-perhaps as much as an average of $28,000 per injury, according to NSC. If one considers equipment damage, lost product and production downtime associated with larger catastrophic incidents, the cost is much higher.
The cost of workers spending time on any activity is costly also. In one case, it was determined that the organization studied spent as much as 36% of its available human resource hours on safety and health intervention activity, yet was not experiencing appreciable improvements in its incident rates compared to what the company had experienced when it spent 15 to 17% of its available human resource hours.
If the company could direct 19% of its available human resource hours to other more productive activities, the organization would still achieve acceptable incident rates and experience a significant productivity improvement. For a 400-person organization, assuming a labor and benefit costs of $25/hour (USD), this 19% human resource hour savings translates to approximately $5.3 million in productivity improvement. With this kind of investment, it would seem reasonable to expect that a management team would want to be sure it is spending $5.3 million worth of its organization's time on productive activities.
The basis for this concept is a mathematical approach to designing effective safety and health programs. It is assumed that safety and health intervention activity is implemented to prevent or reduce incidents. It would then be reasonable to assume that these activities have some effect on incident rates. If this effect can be quantified, it is reasonable to expect that there some relationship can be shown between interventions and incident rates.
Let the rate of incidents be I and let Ai, i = 1, 2... N be the expenditure rate of resources committed to implementing intervention activities. The original models developed in the oil and power industries and relating I and Ai produced non-linear relationships in which the p's are parameters controlling the application of the various intervention activities.
After being determined empirically, it is expected that the relationship can be used to design a mathematically optimum safety and health intervention program that minimizes expenditure of available human resource time while concurrently minimizing incidents. With a mathematical relationship defined, it should be possible to use this dynamic function to predict and forecast expected incident rates given specified levels and qualities of intervention activities.
While the original model was applied to a small part of an oil production operation and the forestry operations within a power company with some success, the full value of this model can only be realized through generalizing the model to other industries and expansion of the database over time to help improve accuracy.
Haight, J.M., Thomas, R.E., Smith, L.A., et al. (2001a, May). Evaluating the effectiveness of loss prevention interventions: Developing the mathematical relationship between interventions and incident rates for the design of a loss prevention system (Phase 1). Professional Safety, 46(5), 38-44.
Haight, J.M., Thomas, R.E., Smith, L.A., et al. (2001b, June). An analysis of the effectiveness of loss prevention interventions: Design, optimization and verification of the loss prevention system and analysis model (Phase 2). Professional Safety, 46(6), 33-37.
Haight, J.M., Thomas, R.E. (2003). Intervention effectiveness research: A review of the literature on leading indicators. Chemical Health and Safety, 10(2), 21-25.
Iyer, P., Haight, J.M., del Castillo, E., et al. (2004). Intervention effectiveness research: Understanding and optimizing industrial safety programs using leading indicators. Chemical Health and Safety, 11(2), 9-19.
Iyer, P.S., Haight, J.M., del Castillo, E., et al. (2005). A research model-forecasting incident rate from optimized safety program intervention strategies. Journal of Safety Research, 36(4), 341-351.