Hyperlocal Hotspots

Each spring, Rio de Janeiro attracts well-heeled partygoers to its annual bacchanalia, Carnival. The city of 6 million has another, darker side however; one-fifth of its population lives in favelas, or slums—hotbeds for diseases that tend to cluster around the poor, such as tuberculosis and HIV.

Yet, until recently, public health researchers tended to treat Rio and most other large cities with extremely varied populations as single units with homogeneous inhabitants who had the same risk and incidence of disease.

Such thinking is slowly undergoing an evolution, says David Dowdy, MD, PhD, ScM, an assistant professor in Epidemiology. “There’s a growing awareness in tuberculosis and other diseases that we can’t just adopt one-size-fits-all strategies,” he says.

Consequently, Dowdy and other researchers are shifting tactics, focusing on far smaller areas to study diseases. By blocking off such disease “hotspots”—even to the neighborhood level—investigators are gathering fresh data that could eventually lead to novel ways to combat age-old diseases.

Smaller and Smaller

This hyperlocal focus isn’t totally new to public health, Dowdy explains. More than 150 years ago, John Snow famously traced the source of a cholera epidemic to a single London water pump by plotting cases in a neighborhood on a map. Since then, the hotspot approach has been a mainstay for a variety of infectious diseases, notably sexually transmitted diseases (STDs). To combat STDs, health workers focus on tried-and-true reservoirs—sex workers, for example, or people unlikely to take precautions for safe sex, such as teenagers.

“For STDs, parasitic infections and many other infectious diseases, there’s a widely known 80/20 rule,” says Dowdy. “Twenty percent of the population is responsible for 80 percent of transmission.”

In infectious disease outbreaks, those 80/20 diseases have provided natural targets for interventions. Henrik Salje, a doctoral candidate in Epidemiology, explains that it’s been significantly trickier, however, to focus intervention efforts in endemic settings because numerous overlapping transmission chains occur in the same areas and it is unclear who is responsible for the majority of infections. This has discouraged researchers from even bothering to look for disease hotspots or clusters—until now. Recently, he and other investigators have started examining disease transmission and prevention in smaller and smaller areas—tracking dengue transmission within neighborhoods, for example.

Why now? It’s all a matter of technology, explains Salje’s colleague Justin Lessler, PhD ’08, MHS ’08, MS, an assistant professor in Epidemiology. “There’s been this great increase in our ability to collect really fine-scale spatial data,” he says. “There is now wide availability of cheap and accurate GPS systems … most people have them on their phones by default.”

Combining GPS with better software to organize multiple types of data now allows researchers to easily “put a dot on a map,” and combine information in ways that they hadn’t been able to before, says Salje. “We can plot the location of a particular case and the time it happened, and even include genetic information of the culprit organism,” he adds. Genetic information is much more available because of increasingly sophisticated, cheaper technology.

Surprisingly Focal

Such work is exactly what Salje, Lessler, and Derek Cummings, PhD, MPH, MSc, assistant professor in Epidemiology and International Health, are doing with dengue. In a study published in the May 28 Proceedings of the National Academy of Sciences (PNAS), they used geocoding to better understand how dengue, and immunity to this virus, spreads throughout individuals in Bangkok.

The researchers used data gathered over five years from a Bangkok children’s hospital. When patients with dengue-like symptoms came to the hospital for treatment, care providers there drew blood and sent it off for diagnoses—and, if it was positive for dengue, checked which of four viral serotypes caused the infection. Additionally, patients provided basic demographic information, including their addresses.

Once someone has been exposed to a single dengue serotype, they’re immune to that particular serotype for life, Salje explains. Preliminary but decades-old research suggests that they’re also immune to the other three for a stretch of several more months. But dengue is unusual in that if former patients are exposed to any of the other serotypes after this grace period, the resulting disease is much more severe.

To confirm this research and learn just how close cases cluster, the researchers used three basic pieces of information about the patient: the time they became sick, where they lived, and which serotype caused the infection. Using these data and geocoding technology to plot thousands of cases on maps over time, the researchers found that cases of the same serotype—suggesting that they might come from a single lineage of infection, passed from individual to individual—were occurring in areas smaller than a square kilometer. Dengue in these communities followed the same track as previous research suggested. Neighborhoods had a localized outbreak of a single serotype, followed by a period with few or no cases lasting many months, and then they were more likely to have severe disease caused by other serotypes.

Though Bangkok is full of commuters who could easily pass infections all over the city, the disease still clustered around homes, Lessler says. “It’s far more focal than we would have realized without these data,” he adds. The finding eventually could allow researchers better ways to implement prevention efforts or test whether vaccines in development
are working.

Better Targets

Similarly, in Rio de Janeiro, Dowdy’s work on tuberculosis is showing that a small reservoir of individuals could be the key for slowing or stopping this disease’s spread throughout the entire city.

He and his colleagues used past surveillance data to narrow their focus to three areas—comprising about 6 percent of Rio’s population—that appeared to be hotspots for the disease, with TB rates at least double those of the rest of the city. Using additional data on how TB passes through populations from other cities, they constructed computer models to see how tuberculosis transmitted throughout the hotspots and from these hotspots to the rest of the city.

Their findings, also published in the May 28 PNAS, suggested that this mere 6 percent of Rio is responsible for more than 35 percent of TB transmission in the entire city. If prevention and treatment efforts were targeted at just this population, Dowdy says, their models showed the same effect in diminishing tuberculosis over time as targeting the other 94 percent of the city.

However, there is a catch: Targeting this tiny population could prove more difficult and expensive per individual than targeting the majority of the city.

“These hotspots are going to have fewer existing resources, people are less connected to care, and they’re not going to have the same diagnostic and treatment infrastructure,” Dowdy says.

But in the end, it could add up to a better investment. “Even if it’s 10 times more expensive per person, you’re still getting the same effect for 6 percent of the population rather than 94 percent,” Dowdy says. “It’s more bang for your buck.”