Healthy connections. Public health is concerned with promoting health and preventing illness. To better understand the factors, programs, and initiatives that work best to promote health in a community, practitioners often rely on data from public records. These are often stored in separate databases, making it difficult to correlate relevant data and reveal crucial relationships.
A program to link multiple data sets connected with pregnancy, birth, and subsequent maternal and child health has been developed by a team led by SPH Professor Milton Kotelchuck, chairman of the maternal and child health department. The Pregnancy to Early Life Longitudinal Linkage Project (PELL) is an unusual public-private partnership among SPH, the Massachusetts Department of Public Health (DPH), and the federal Centers for Disease Control (CDC).
The project allows researchers to examine data linked from a variety of sources — collected widely across diverse populations and longitudinally, over time — including birth certificates, records of fetal death, and birth-related hospital discharge data for both mothers and infants. It also includes data from maternal and child health programs, the Birth Defects Registry, child and maternal death certificates, and hospitalization and emergency room records. PELL includes all Massachusetts births since 1998 — more than 320,000 infant and 280,000 maternal records — and it continues to grow as birth records are released every year.
The linked analyses of such data provide valuable information to help evaluate the coverage, cost, and impact of programs such as the Early Intervention Program, designed to provide services to infants and young children at risk for developmental disabilities. Because PELL includes data on the entire population of newborns in the state, it is possible to determine, for example, how many low-birthweight infants were screened by the program, and find common factors among the infants that were missed. PELL can be used in a wide range of other studies — for example, to examine the incidence and health impacts of procedures such as elective C-sections or to better understand the health consequences of multiple births.
In designing this linked data system, the team took great pains to protect the confidentiality of the data. “It exists virtually rather than as a permanent linked database,” says Kotelchuck. “We use generated ID numbers and a series of computer programs that extract linked data from different data sets.” In addition to receiving clearance from the DPH Research and Data Access Review Committee, all projects need to pass scrutiny by the CDC and BU’s Institutional Review Board. The program also ensures that any agency or program contributing data give prior approval before its data is analyzed.
The thoughtful groundwork that underlies PELL not only makes it a valuable tool for better understanding factors that have an impact on maternal and child health, but also makes it a model for other programs nationwide. Having the ability to make important connections between discrete sources of data may point the way toward developing effective interventions for a wide range of public health challenges.
Clearing bottlenecks. No one likes traffic jams. Whether you’re tied up on the Southeast Expressway or trying to retrieve your e-mail, networks can seem to move impossibly slowly. The solution, as reporters in traffic helicopters are fond of saying, is to find alternate routes — but these also may be fraught with unknown perils.
A group of researchers at the Center for Polymer Studies (CPS), under the direction of Eugene Stanley, a UNI professor, CAS professor of physics, and center director, has recently developed a new mathematical model to describe how networks carrying information, electricity, water, even automobiles function — and how they break down.
In 1967, the eminent social scientist Stanley Milgram established that people are all linked to one another through short chains of acquaintances, with a random long-distance link or two in the mix — a phenomenon known as “six degrees of separation” or “small worlds.” Mathematicians have shown that many real-world networks can be described as “small worlds.” On the Internet for example, alumnus Laszlo Barabasi (GRS’94) showed that it takes only 19 clicks to go from any one node on the Web to any other node, even though the World Wide Web contains more than a billion nodes. This is due in part to labels or tags that can be indexed, providing the occasional long-distance link to shorten the route. But it turns out that links between two points may come with a cost — bypassing the traffic jam may mean driving a longer distance.
Stanley and colleagues Sergey Buldyrev, a research associate at CPS, visiting scholar Lidia Braustein from the National University of Mar del Plata in Argentina, and Schlomo Havlin and Reuven Cohen of Israel’s Bar-Ilan University have been collaborating to discover how the costs of various links affect the efficiency of real-world networks. They randomly assigned different costs to links in a network (values along a continuum of high speed to slow speed, for example) and calculated the optimal, lowest cost paths. As long as the costs didn’t vary considerably, they found that the shortest and lowest cost paths were the same. When the cost variation was larger (when there were significant bottlenecks), the lowest cost route become considerably longer, or conversely, the fastest routes became much more costly.
Broadcasting video over the Internet provides a real-life example of this concept. “If you want to watch a video online,” says Havlin, “it is very important that the information arrive in a timely fashion. The speed of the transmission is determined by the slowest link, so low-bandwidth links anywhere in the system disproportionately affect how the video is received. Even if routing software tries to avoid slow links, the optimal path still may not provide optimal reception.
“But if you can identify the key bottlenecks and boost their capacity,” he says, “you may be able to boost the performance of the entire network.”
Briefs" is written by Joan Schwartz in the Office of the Provost. To read
more about BU research, visit http://www.bu.edu/research.