In California, it was the perfect breeding ground for a public health crisis: a theme park.

Disneyland not only attracted millions every year from all over the globe, but many of the children who descended on the popular resort in late 2014 were not vaccinated against measles, one of the world’s most contagious diseases.

Within weeks, the illness raged across the state — with many hospitalized — in the largest outbreak of the disease in California in two decades.

With dozens of children sickened and mounting concerns over the state’s winter tourist season, then state Sen. Richard Pan moved quickly to stop the next outbreak.

The longtime pediatrician pushed a controversial bill in early 2015 that would strip away personal exemptions to the state’s vaccination requirements — waivers that allowed children to attend schools without getting the shots.

By then, the illness had spread to at least 125 people, mostly in California., along with six other states and Mexico and Canada in a surge that drew international attention.

At the same time, opponents of Dr. Pan’s bill launched intense protests to pressure state lawmakers to try to save the exemptions.

To gain support, Dr. Pan turned to a team at the University of Pittsburgh that was using a novel approach — computer simulations — to track outbreaks in lawmakers’ districts.

By drawing on reams of data — including vaccination rates, population density, demographics, commuting patterns — the Pitt team simulated the path of the disease in grim detail.

Known as the FRED model — Framework for Reconstructing Epidemiological Dynamics — the program turned out to be instrumental in convincing lawmakers to pass the rare legislation that took away the ability of residents to skip the vaccines because of religious or personal beliefs.

It was the first time in nearly 30 years that a state had removed all the exemptions, except for those set aside for medical reasons.

“FRED made it real,” said Dr. Pan, a Pitt medical graduate who is now running for a U.S. House district in the Sacramento area.

The bill was signed into law, and two years later, California’s vaccination rate for measles jumped more than three percentage points to 96% — putting the state above the level of immunity needed to stem the spread of the disease.

“We haven’t had a measles outbreak that large since,” said Dr. Pan.

The research team from Pitt’s school of public health went on to carry out modeling for measles in other major cities — including Houston, Indianapolis, New York and Miami.

For policy makers, the move to bring in the researchers was an opportunity to help plan for public health emergencies and show people how far the disease could spread.

Deadly outbreak

Three years after the Disneyland outbreak, the Pitt team was called to carry out a similar study in Texas in what would provide an alarming warning — one that signaled the largest and one of the deadliest outbreaks of measles in the United States in more than 30 years.

Over several weeks in 2018, the Pitt team ran hundreds of simulations that showed pockets across the state where school vaccination rates had fallen and where the disease was likely to surge.

To demonstrate the findings, the team members created short videos that showed how a single red dot on a map — one case of measles — could quickly cascade into a sea of red as the disease raced across the state.

The modeling helped convince lawmakers to stop legislation in 2019 that was aimed at making it easier for parents to avoid the shots for their children.

But the warnings were not enough to prompt school or state leaders to take action that would reverse the steady decline in rates.

In 2025, an outbreak of measles spread into nearly every corner of the state, infecting more than 800 people and leading to the deaths of two children and scores of others hospitalized.

Though the majority of the cases took place in west Texas, at least six clusters erupted in the northeast and central parts of the state — the same areas predicted by the Pitt researchers.

The team “basically created a crystal ball,” said Rekha Lakshmanan, then director of advocacy and public policy for The Immunization Partnership in Houston. FRED was “a little more than prescient.”

Dr. Mark Roberts, emeritus professor of health policy and management at Pitt who was part of the 2018 study, said there were enough signs to show the state was vulnerable.

“If we presented the maps we created with a map of what happened in court, we'd be declared guilty of predicting this,” said Dr. Roberts.

The measles epicenter

While modeling software like FRED has provided a tool in planning for health crises, experts caution that it also has limitations in guiding public policy.

Like any model, it relies on assumptions — who comes into contact with an infected person, where they travel and whether the vaccination rates are accurate.

When the researchers used it to try to predict the spread of Covid, those assumptions at times broke down when the modeling couldn’t account for the social upheaval that took place — lockdowns, remote work, school closures and mask mandates.

Even in the Texas measles outbreak, FRED accurately predicted the state’s high vulnerability, but didn’t detect the epicenter.

The worst of the outbreak happened in rural counties — not the more populated metro areas the study focused on.

The development of FRED — named after Pittsburgh’s legendary Fred Rogers — began in 2001 under the leadership of Dr. Donald Burke, the former dean of Pitt’s School of Public Health.

Dr. Burke and fellow FRED researcher John Grefenstette were walking through Pittsburgh International Airport when they were inspired by a Mister Rogers display — featuring the television host’s iconic red sweater and sneakers — and decided to honor one of Pittsburgh’s favorite sons.

“It was based on the notion that we were modeling communities and neighborhoods, and who cared more about neighborhoods and kids than Mister Rogers?” Dr. Burke said.

Then teaching at Johns Hopkins University, Dr. Burke was searching for a way to predict what would happen if there was a smallpox outbreak caused by a terrorist attack in the wake of 9/11.

Government officials at the time believed that they would have to launch a massive vaccination program across the country to immunize everyone against the disease.

But by using the modeling system, researchers showed that an outbreak could be contained by targeting vaccinations in certain areas to cut off the spread of the illness.

“We showed that we didn’t need to inoculate everyone to prevent a pandemic,” Dr. Burke said.

Initially funded with a $13.4 million grant from the National Institutes of Health, the program was bolstered over the years with money from the Gates Foundation and Robert Wood Johnson Foundation.

As funding began to slow seven years ago, the developers formed a private company, Epistemix, to keep the program intact and expand its modeling applications.

Owned by a group of investors and employees of the company, it has yet to turn a profit, but its revenues have grown each year, said the company’s CEO, John Cordier.

He said the company is negotiating to raise $20 million in funding from companies in the health care, pharmaceutical and insurance industries.

Since the formation of the new company, the FRED team has expanded its forecasting beyond viruses, such as predicting the rise of asthma cases in the wake of a large fire.

For more than a decade, the FRED modeling was sought out by public health officials who were planning for potential emergencies.

In 2015, the team created the online U.S. Measles Simulator that continues to allow anyone across the country at no charge to choose a city and then see how an outbreak could play out.

Using detailed information about the people who live in a certain area — how old they are, where they travel, the number of children in each household who go to school every day — the researchers recreate daily life inside the software.

The modeling is often run through 100 simulations — covering a spectrum from the most conservative to the worst-case scenarios — until a pattern emerges that gives researchers a clearer picture of how an outbreak could spread.

“Over the years we’ve made [the models] more and more realistic,” Dr. Burke said.

While it was designed to map the spread of a range of illnesses, including influenza, the rising threat of measles would drive national interest in what FRED was capable of showing.

After the Texas study, the Pitt team went to Florida in 2019 amid a growing national trend: Parents were opting out of vaccine requirements and lawmakers were battling over whether to tighten the rules.

Dr. Karen Liller, the director of the Activist Lab at the University of South Florida College of Public Health, said she reached out to the team because she wanted to show the public the dangerous consequences of falling vaccination rates.

The researchers pulled school immunization levels in pockets across the state and fed them into the model to see how open they were to outbreaks.

“We show in the current simulation that if [the rates] drop by 10% you could get thousands of cases here — not just hundreds,” said Dr. Liller, who grew up in Canonsburg.

In side-by-side windows on the FRED website, people can watch simulated measles outbreaks in the state’s metro areas — with red dots surfacing across the cities when the immunization rates fall.

“The purpose was never to scare people,” she said. “It was to educate them.”

‘A big blow up’

Similar patterns emerged in Pennsylvania when the team carried out modeling of potential outbreaks at the request of the Post-Gazette, using classroom-level data provided by the reporters.

The team looked at schools in a half a dozen counties across the state, including the two largest, Philadelphia and Allegheny, and found that the schools with the lowest rates could trigger outbreaks impacting hundreds of people.

At one private school in Montgomery County where less than two-thirds of kindergarten students are vaccinated against measles, a single case could spread to 430 people within months, simulations show.

In the same school with rates safely above herd immunity, a single case would lead to just seven infections.

Because of the proximity of students to one another in the classrooms, the cases can “cluster,” said Mary Krauland, a research professor who carried out the analysis at Pitt’s Public Health Dynamics Laboratory.

“Those are the places where you can still get a big blow up,” Krauland said.

“This is why doing it on the school vaccination level is better than doing any kind of countywide vaccinations.”

In Texas, the prediction of a measles outbreak was built largely on falling rates in the schools, which can lead to “dramatic increases in the probability of large outbreaks,” the study noted.

Dr. Roberts said Texas leaders never acted on the Pitt team’s findings, even though he briefed the staff members of state lawmakers about the study.

He said the state could have instituted a childhood vaccination campaign or found other ways to head off the outbreak that his model predicted.

What troubled him about the Texas study and other modeling that tracks measles is that some school children — up to 2% — can’t get vaccinated because of medical issues.

He said another small percentage will take vaccines, but their bodies won’t produce the right antibody.

As a result, “when you look at kids who get sick, half of them are people who we’d call ‘innocent bystanders,’” he said.

“I did a whole day presenting to the legislature, and nothing changed. And that’s just politics,” Dr. Roberts said.

“I know I sound like a broken record when I say this, but our job in creating these models is to provide what we think is the best science for what happens under different policy scenarios,” he said.

Based on the team’s simulations of what could happen inside Pennsylvania’s schools, he said he’s growing more concerned about the potential for outbreaks.

“The risks from decreasing vaccination rates are clear,” he said. “More children will be susceptible, there will be more cases, and more children will be at risk for potential serious health consequences — all of which are avoidable.”

Michael D. Sallah and Mike Wereschagin contributed to this story.

Return to main story