Dr. Alberto Colombi uses JMP's Prediction Profiler to explore how stress, obesity and depression affect the average worksite medical and wage payment per active workers' compensation case.
Becoming proactive on health care
|Challenge||To gain knowledge of the factors that put employees at increased risk of developing chronic disease, how these factors interrelate and what measures will be most effective in preventing disease.|
|Solution||PPG is using principal components analysis in JMP® to examine multiple variables: risk factors, health promotion and outcomes.|
|Results||PPG has launched campaigns aimed at keeping employees healthier, enhancing productivity and making better use of health care investments.|
With an analytical approach to preventive health care using JMP® statistical software, PPG Industries strives to keep workers healthier and more productive, while controlling health care costs.
Dr. Alberto Colombi, PPG Industries’ Worldwide Corporate Medical Director, had some questions about how best to help PPG employees stay healthy and productive.
Among Colombi’s questions was this one: Is there a better approach than the traditional clinical trial for analyzing population health-related data?
“Most advancements in health care knowledge are based on the clinical trial,” Colombi says. “The clinical trial is generally a double-blind process, by which some people are given a treatment and some are not – and the researcher and those treated are blind to that – and we find out if there is a marginal effect of a new drug or device. This paradigm has served medicine well for a long time.”
But Colombi’s objective was to study chronic disease focusing on worksite populations rather than individuals. PPG, a manufacturer of coatings, glass and other specialty products for a number of industries, has about 37,000 employees across North America, Latin America, Europe and Asia-Pacific.
“If you want to look at chronic disease across multiple populations,” Colombi continues, “the traditional clinical trial paradigm is not very useful – because you cannot analyze just one disease or one risk factor at a time.” An employee may be at risk for more than one chronic disease, and each disease has multiple risk factors, while each worksite has a varied burden of chronic diseases affecting its population.
Colombi then pondered the question: In seeking to improve the health of PPG employees, would it be possible to focus on and interpret multi-factorial health-related interactions to reduce uncertainty and support health decisions?
The answer was yes, using JMP statistical analysis software.
And with the success of that research came the second pivotal question: Where should PPG invest its limited health-promotion resources?
Applying process improvement to health care
Colombi had received Sigma Logic® training several years back, and it occurred to him then that the process improvement principles that Six Sigma applied to manufacturing processes might also apply to health care.
When engineers are evaluating a manufacturing process, they simultaneously deal with multiple variables and try to understand which ones most influence certain product outcomes.
“I looked around and thought some of these ideas could definitely be applied to health as well,” Colombi says.
He could conceive of a process that started with good health, but where several factors could then influence the process and result in a defect, which in this case is a chronic disease.
“Now, we’re not in the business of ‘manufacturing’ diseased people,” Colombi says. “But if you can apply this idea that there is a process, there is an input, something happens in between, and then there is an output – if you could find out what influences the appearance of the defects, maybe you could prevent some of those defects.”
Colombi began by lining up the components in the process that can lead to chronic disease. Using 29 worksites as a case study, he asked how much health promotion was conducted at each site, what was the workplace’s population health risk profile (that is, what percentage of employees are obese, have high blood pressure or smoke), what are the outcomes (the number who have heart attacks or strokes) and what is the cost effect (both the direct cost of treatments and the indirect cost in lost productivity)?
“So you take this conceptual process and ask what most influences the final outcome,” Colombi says, “and you begin doing some decision analysis based on many variables.”
Colombi’s process went like this:
- Create a basic model of the business problem.
- Determine how to best represent uncertainties.
- Create an exhaustive set of choices.
- Perform quantitative analysis to determine the optimal decision.
An ‘epiphany’ about JMP®
It was clear, Colombi says, that “we needed a new tool to take on this task of analyzing multi-factorial determinants of health outcomes.
“And we already had it: JMP. We have many, many variables – a number of worksites, a number of risk factors that we want to test – and, lo and behold, we have a tool that allows us to process multiple factors at the same time, one we were already using for manufacturing improvements.”
Colombi now makes extensive use of JMP’s multivariate analysis and modeling capabilities.
“We go through the principal components analysis, then take those components and enter them into models and prediction profilers to see which variables rock the boat the most,” he says.
This process may not point directly to causation, but it does point to high-risk factors and indicates where investment in wellness makes the most sense.
“It was an eye-opener,” Colombi says. “We found a new methodology that allows us to process our information in a way we hadn’t. It was an epiphany. JMP allows us to do all this; it allows us to break new ground.”
Colombi and his team have been using JMP to study diseases related to the circulatory system. For example, researchers have asked: What happens prior to a heart attack? What is the risk profile of the population? What percentage of employees are obese, don’t eat properly, don’t exercise and/or are smokers?
Then, going further upstream with the inquiry, they wanted to know: How much workplace education has taken place? Is information about healthy eating provided? Does the cafeteria offer fresh fruit?
And then, given the risk profiles and the preventive measures, the next question was: How many heart attacks have there been in that particular workplace in the timeframe studied?
From these three domains, Colombi collected more than 70 variables, which he then put into a model in JMP. His team also has been using JMP to examine workers’ compensation to see what upstream variables might influence the frequency and amount of payments.
“So I’ve analyzed two ‘product lines,’ if you wish,” Colombi says, and what he’s learned is that obesity, when interacting with a number of other factors, is a major issue that must be addressed.
“To me, that said that we’ve got to help reduce the weight of our people in several ways. And the cost advantage of doing that is found in both the cardiovascular and the workers’ compensation studies,” he explains. “I’ve now demonstrated that weight reduction is very important, so now I have a stronger, more convincing argument to take action.”
Improving health through fun
Those efforts are now under way. For example, several PPG worksites sponsor a “Biggest Loser” competition, or a “Triple Holiday Challenge” – in which employees try to maintain their weights through Thanksgiving, Christmas and New Year’s Day – host a farmers’ market, have made changes to cafeteria menus and have developed local exercise programs.
Colombi says PPG employees are responding favorably.
“It’s a very serious matter,” he says, “but we’re offering a fun component to it. People like to set up their own teams for competitions; they come up with funny names.”
And he appreciates the role JMP has played in this success.
“It is so visual; it’s so communications-oriented. JMP is above anything else that’s available in allowing us to communicate our statistical data. It makes it graphically consumable, so to speak,” Colombi says.
The end objective, of course, is to prevent illness, enhance the company’s productivity and make health care affordable.
“By keeping people healthy, we can mitigate ever-increasing health care costs,” Colombi says. “And by keeping people healthy, we also can keep them at work and productive.”
It will take time, Colombi acknowledges. But PPG has made a good, positive start.
“We’ve used JMP to demonstrate what matters, convinced people that what matters is close to them, and then taken action. It’s translational research. It’s decision analysis, but then you translate that decision into action,” he says.
We found a new methodology that allows us to process our information in a way we hadn’t. It was an epiphany. JMP allows us to do all this; it allows us to break new ground.
Dr. Alberto Colombi
Worldwide Corporate Medical Director, PPG Industries