Image Credit: Getty Images / LunaKate

Technological advancements have enabled occupational and environmental health and safety professionals to collect increasing amounts of monitoring data, generally to support day-to-day risk management decision-making. Many organizations now store this data electronically in large databases, which OEHS professionals have yet to exploit. Access to huge stores of data in electronic formats creates opportunities to turn these data into more useful information. Through summarizing, communicating, and drawing insights from monitoring data, OEHS professionals can help their organizations better understand risks and set priorities for improvement.

Information scientists distinguish between statistics and data analytics. Statistics are generated to prove hypotheses, but data analytics put data collected for business purposes to secondary use. Off-the-shelf data analytics tools can help OEHS professionals summarize monitoring data by turning it into charts and tables that assist in communication and decision-making.

"Data analytics is where the profession is going," said Steven Jahn, MBA, CIH, FAIHA. "You need technology to empower you to cover so much work and protect so many people."

Data Analytics for Exposure Assessment

When Paul Wambach, CIH, worked for the Department of Energy, DoE staff identified several cases of chronic beryllium disease at multiple sites and realized that the exposure limit for beryllium, which had been in place for years, was not protective enough. Wambach was tasked with writing periodic reports summarizing health and exposure data resulting from the operation of the Department's chronic beryllium disease prevention program. "That put me in the position of having to analyze these exposure monitoring results that were coming from around the complex," Wambach said. "Towards the end there, I had a data set with about 40,000 exposure measurements in it."

Wambach worked with a biostatistician to put together a statistical package to create charts and tables summarizing the data. At about the same time, Wambach read much of the industrial hygiene literature on exposure assessment, which led him to realize that he needed a different approach from most industrial hygiene practice. "What I was doing was creating more of a performance indicator than an exposure assessment," Wambach said. "The numbers I was coming up with really weren't numbers you could tie to any particular person but rather the performance of our exposure management program."

He was trying to determine whether the program was monitoring exposures often enough to be certain that beryllium exposures were being controlled. Although the data were highly censored, not lognormal, and not from a similarly exposed group, applying statistical methods allowed him to figure out whether controls were adequate.

Jahn, who is now a technical advisor for safety and industrial hygiene services at Savannah River Mission Completion, explained the traditional IH approach to evaluating data as follows: "My training back in the '80s, in grad school, was: go get a sample and take that sample for some time interval and send that sample for wet chemistry analysis. And when it came back, it gave me a number. A single data point." That is, OEHS professionals typically take a measurement (or an average of measurements) and compare it to the exposure limit. But the conclusions generated by this approach are often misleading, Jahn explained. "If you make a decision on simply that single data point, it's a 50/50 proposition. You could be making a correct or an incorrect inference."

OEHS professionals "should understand that they need to make a very conservative and biased judgment when they use standard statistics," Jahn said. "They shouldn't be looking at an absolute number against the limit." Instead, they should use monitoring data to create a statistical model of exposures. By assessing the distribution of monitoring data, especially the 95th percentile of measurements, OEHS professionals can infer whether exposure levels likely exceed the limit.

Using Data Analytics in Decision-Making

As Jahn and Wambach have explained, use of data analytics can help OEHS professionals make more accurate exposure assessments. But the ability to model exposures also equips OEHS professionals to make predictions about future exposures, prioritize controls, and plan ahead. Wambach noted that OEHS professionals often review plans for upcoming projects. "You can't go out and monitor the work if it isn't ongoing yet, but you are under the gun to produce a recommendation as to what kind of controls are needed," he said. "If you've got data from similar projects that you can use, that's going to help inform your judgment on what kind of controls are going to be needed."

Data analytics tools may also give OEHS professionals the confidence to recommend controls when collecting more samples will not serve their purposes. Rather than encouraging OEHS professionals to disregard as outliers occasional monitoring results that exceed exposure limits, data analytics tools help them see how often these outliers occur and decide if they must investigate the cause.

Jahn related a story from early in his career, when he'd been an on-site consultant for OSHA. A metals recovery operation had repeatedly called Jahn back to the site to collect samples for lead without changing any controls in the meantime, so the employer could claim they were making an effort to abate lead hazards without spending any money. Like many OEHS professionals, Jahn had felt that more sampling would solve the problem, but it did not, and in the meantime, exposures continued. For repeated exposures such as this, "you can do some modeling in advance of taking a sample," Jahn said. "You can estimate what you think an exposure range is. And then you can validate that initial judgment with some sample sets. It's those sample sets you treat statistically."

Above all, Wambach asserted that the value of data analytics is the ability to confirm whether your control programs are effective. When he worked for the DoE, he could assess the probability that employees were exposed to beryllium. "We can look at how the department as a whole is doing or individual operating elements within the department," he said. "You're giving management a performance indicator they can use to set goals and monitor progress in meeting those goals."

"All the OEHS community needs to recognize the utility of statistics and making informed judgments that they're inferring from limited data back to their clients or to their regulators," Jahn added.

Data Analytics Resources

Learning to use more advanced data analytic methods takes time and effort. "You're not going to go to a one-hour seminar and understand this stuff," Jahn said. "You have to immerse yourself in your data and understand and be probative and very self-aware."

Wambach also acknowledged that "there's a learning curve to climb" when using statistical tools, such as R. "But once you've done it, it's easy to generate these metrics, to pull the data into an analytical file and run it through."

Fortunately, AIHA and other organizations have developed resources to help OEHS professionals learn data analytics. Jahn recommended IHStat, developed by former AIHA President John Mulhausen, PhD, CIH, CSP, FAIHA. Mulhausen is also one of the instructors of a free online course on making accurate exposure risk decisions. A version of the IH Data Analyst program designed for OEHS students and professionals is also available online for free. More tools are available from AIHA’s statistical, risk assessment, and big data, AI, and sensor technologies resources pages.

Another new resource is AIHA's new white paper, "Data Analytics." Wambach explained that the white paper uses an example from the construction industry to walk readers through the process of using data analytics to evaluate large databases. He hopes the document may motivate federal agencies and other large employers to start publishing annual reports summarizing their OEHS data, similar to the way the Bureau of Labor Statistics reports workplace injury statistics.

"In an industrial hygiene utopia, somebody would start analyzing these government-owned data and putting out routine reports summarizing what exposure levels are typically found in workplaces," Wambach said. "This would encourage employers to get the data out of their computers and produce similar sorts of metrics and reports for comparison purposes"

"Data Analytics" may be downloaded for free from AIHA's website.

Related:

The Synergist: "Industrial Hygiene Data Standardization: Past Lessons, Present Challenges, and Future Directions" (December 2020).

The Synergist: "The Promise of Portability: Modernizing the AIHA Exposure Monitoring Data Structure" (June/July 2024).

The Synergist: "The Riddles of Beryllium: A Short History of a Challenging Workplace Hazard" (October 2021).