They started by talking to experts in the Centers for Disease Control and Prevention, the U.S. Department of Health and Human Services, and disaster-planning consultants about the impact a pandemic might have. At the same time, they reached out to other employees to determine just what they would need to do their jobs. But, before they could hold those talks, they realized they had to be clear about their mission, with concise goals that could be understood by anyone in the company. To that end, they formulated two interrelated objectives: to protect employees during a pandemic and to be able to have business continuity.

Then, they approached other Roche staff members. First stop was managers in 17 key functional areas, such as manufacturing and human resources, to explain what the impact of a pandemic could be and how it would affect their areas. As important, they asked for insights into just what the managers needed to start preparing right away and what they would require after a pandemic hit. Manufacturing, for example, used just-in-time inventory management for the production of key solvents. But, once the flu started to spread–and there were not only fewer employees available, but demand started to increase–manufacturing managers figured that approach probably wouldn't work. Result: They'd need to start stocking more of the solvents than they had in the past. In addition, each functional head made a list of business-critical employees, as well as who had to work on site, who could telecommute from home, and the equipment they'd need–laptops, say, or specific software programs.

Once they'd received that information, McGuire and Peterka held a series of meetings with representatives of all functional areas together, to make sure each department knew what the others were doing. That was especially important for cases where functions worked together. Quality assurance, for example, always had to approve products before anything could be shipped. So, it was crucial that specialists in that area work closely with manufacturing to coordinate plans for handling the process during a pandemic.

At the same time, they started to see that the job before them was much more than they could handle on their own. In fact, it required the help of a staff focused on only that task. In December 2005, they got permission from senior management to bring in two full-time people. "You have to dedicate resources to this process in order to be successful," says McGuire. "We needed people without other responsibilities working on the pandemic plan."

By September 2006, the team had a pretty good idea of what it would take to prepare–and they were ready to talk money. The cornerstone of their proposed budget was a timeline developed by the World Health Organization, with probable phases of pandemic progress and key trigger events. Trigger 2, for example, was the development of clusters of human-to-human transmission outside the U.S.; trigger 3, transmission inside the country. With that information, says Peterka, they were able to provide "reasoning and justification for everything we needed and why." They presented it to senior management who accepted most of their requests.

After that, McGuire and Peterka started rolling out their plan. First, early last year, they developed a 610-plan document, detailing step-by-step elements for all functions–everything from key employees who would be allowed into manufacturing plants and back-up personnel for those people to all the functional area scenarios. Then, around the middle of last year, they created a 28-page resource for senior managers, summarizing key actions in their functional areas that would need to take place after each trigger event. In addition, they developed what they called a dashboard, a computerized and hard copy report that would give the CEO crucial updates about such important information as stockpile levels of drugs and the number of employees out sick.

Even the most thorough analysis was sure to have gaps. That's where the second part of the plan came into play. To figure out just what they'd left out, McGuire and Peterka conducted a series of tests. Toward the end of last year, they held the first of three with about 13 senior managers, including the CEO, simulating a scenario in which the pandemic had hit the U.S. and was spreading, the workforce was reduced by 20%-30%, and the government was asking for stepped up production of Tamiflu. The result: a realization that they'd left their own vendors out of the plan. "If our supply chain failed, so would our ability to manufacture," says McGuire. So, a few months later, the company held a program at headquarters for 115 key suppliers to stress the importance of planning for a pandemic.

A third element was related to business recovery. As the crisis started to recede from the area, the company would need to resume normal business operations as quickly as possible. So, McGuire and Peterka mapped out systems for doing that. For example, during the pandemic, the company would stop making all but the critical pharmaceuticals they usually produce. As the epidemic slowed down, the plan included ways to efficiently ramp up production again.

Then, there was the matter of communication. Key employees would get video conferencing equipment for their homes, in case they couldn't get to work. Manufacturing plants would also be outfitted with the technology.

Along the way, the team also drew on lessons learned from the aftermath of 9/11 and, later on, Hurricane Katrina. For example, because cell phones and other communication devices often failed during 9/11, they installed satellite phone systems in manufacturing plants. After Katrina, many people couldn't come to work because they had to take care of family members. With that lesson in mind, McGuire and Peterka realized they had to identify back-ups for key employees, in case those individuals were unable to function. At the same time, part of the plan also focused on a multi-pronged employee protection effort. That included, for example, "social distancing"–permitting only a small percentage of people at headquarters at any one time, seating them more than six feet apart, and delivering food to employees on-site so they wouldn't have to congregate in the cafeteria. In addition, people would be scanned before entering the building for signs of flu. Plus, employees were given a mandatory computer-based learning tool, with information about the illness. The plan also changed the company's prescription drug policy, allowing faster- than-usual prescription renewals after a trigger 2 event happened. And all employees would get a supply of Tamiflu.

Of course, the plan isn't carved in stone. "This is a continual process. You're updating on a regular basis," says McGuire. What's more, the team is constantly taking steps to make sure their plan is up to date and keep senior leadership informed. For example, they meet monthly with the executive team to provide updates and also communicate more frequently via e-mail. Over the next month, they plan to meet with functional area managers and their direct reports to ensure that those employees also understand the plan. And, toward the second half of the year, they intend to hold another simulation, this time company-wide. The exact date won't be announced beforehand.

Ultimately, according to McGuire, the most important lesson is a particularly sobering one. "This isn't something that can be done over a weekend once the pandemic hits," he says. "By then, it will be too late."