Imagine if in your town of 1,500 residents, 20 babies had been born with birth defects over a one-year period. In a larger city, these numbers might go virtually unnoticed, but in a small community, such an apparent “cluster” of a particular health issue could raise a red flag to the residents. But the question is: Should it?
When a particular disease is confirmed to occur in a community at a higher rate than would be expected by chance, it becomes a “disease cluster.”
The Brockovich case and the movie about it catapulted the concept of clusters into public awareness.
This is the type of cluster made famous in the film, Erin Brockovich, a true story about the high percentage of residents of Hinkley, California who were found to suffer from breast cancer, Hodgkin’s disease, miscarriages and spinal deterioration. It turned out the town’s water supply was contaminated with hexavalent chromium, a carcinogen released from the Pacific Gas & Electric Company’s nearby power plant. Although state officials did not determine that cancer rates were any higher than would be expected, the residents filed suit against the company, and won. Pollution of the groundwater continues to be a problem there.
Studying clusters is extraordinarily challenging, and pinning down the “cause” of a cluster even tougher. Interestingly (and luckily), many of the things we fear to be clusters turn out not to be clusters at all.
A disease cluster, according to Centers for Disease Control (CDC), is an “unusual aggregation, real or perceived, of health events that are grouped together in time and space and that is reported to a public health department.” There are two basic types of disease clusters: infectious and non-infectious clusters. Each presents a very different set of problems, for researchers and for the public.
Infectious clusters are communicable diseases caused by pathogens, like the flu, hepatitis, and certain bacterial infections. Because their causes are clearer and their spread more traceable, they present a simpler (though no less serious) diagnostic problem than non-infectious disease clusters.
If a cluster is confirmed to exist in a particular location, finding its cause is generally the next and more complicated problem for the epidemiologists working on the cluster.
Epidemiologists at the CDC and in local governments study the patterns of diseases, both infectious and non-infectious. Those who specialize in clusters try to determine whether increased instances of a particular disease are indeed due to a cluster or due to chance (more on this issue shortly). The government requires healthcare professionals to report occurrences of many diseases, so that patterns can be determined. If a cluster is confirmed to exist in a particular location, finding its cause is generally the next and more complicated problem for the epidemiologists working on the cluster.
Epidemiologists have several ways of tracking infectious diseases, ranging from reports from doctors offices, to monitoring the symptoms rather than the disease itself (considered a more active method or surveillance), to even newer methods like how many people “Google” keywords like “flu symptoms.” This last method actually gives researchers an excellent way to gauge the spread of the flu, since it gives an earlier warning than waiting for patients to visit their doctors. Other infectious diseases have to rely on classic methods of reporting, which often require medical offices and hospitals to report cases to the state government, who track instances of each disease continuously.
The possibility that a non-infectious disease, like cancer or a neurological condition, could spring up in a particular community is undeniably disturbing.
The government and independent organizations like the National Resources Defense Council (NRDC) study non-infectious disease patterns, and have tried to address possible clusters as best they can. Last year, the NRDC released a massive report on the clusters in 13 states. The report outlined the dozens clusters in these state and, when possible, their connections to chemicals in the environments. The NRDC plans to evaluate possible clusters in remaining states in the future, but the process is an elaborate one, and comes with all kinds of complications.
Some of the non-infectious clusters that have gained attention include cancers, birth defects, neurological disorders like multiple sclerosis (MS) and amyitrophic lateral sclerosis (ALS), and childhood diseases like leukemia. For example, cancers have been linked to various carcinogens (for example trichloroethylene, dioxin, tremolite asbestos, and polyaromatic hydrocarbons), which are often used in solvents, building materials, or are the byproducts of fuels.
Neurodegenerative diseases like amyotrophic sclerosis (ALS) and multiple sclerosis (MS) have occasionally been linked to toxins released from lead smelters. Childhood cancers like leukemia have been linked to the presence of high levels of the carcinogens benzene and 1,3-butadiene in the air.
A famous cancer cluster occurred in Camp Lejeune, NC, which houses the army base, and saw increased incidence of several diseases like childhood leukemia, non-Hodgkin’s lymphoma, and certain adult cancers. Male breast cancer, which is exceedingly rare in the general population, apparently struck 60 men living on the base. It was discovered that for some 40 years, the town’s water was contaminated with perchloroethylene from a dry clearer off-base, trichloroethylene from solvents used on base, and with benzene from fuel tank leaks on the Marine Corps Base. The CDC is still evaluating the health effects of the contamination.
Other situations get less attention. In Prairie Grove, Arkansas, state officials identified a group of testicular cancer cases, three of which occurred in young boys. Though the residents were suspicious of certain environmental factors (a closed nuclear reactor, a radioactive landfill, a poultry plant), no cause was found.
There are many cases like this one, in which diseases appear at first glance to be grouped, but investigation reveals nothing, so the cases are stalled or abandoned. Though this may cause residents of the communities significant stress, there is some good evidence that a good portion of these cases are not much more than a matter of chance.
As Wayne LaMorte, an epidemiologist at Emory University puts it, “Clusters are devilishly hard to pin down, mostly because of the small numbers involved.” Determining the existence of a cluster is a statistical nightmare.
For some 40 years, the town’s water was contaminated with perchloroethylene from a dry clearer off-base, trichloroethylene from solvents used on base, and with benzene from fuel tank leaks on the Marine Corps Base.
First, statisticians who must first determine what might be the expected risk for a given disease in the community based on national averages. Then they have to determine whether the instances seen in a given community actually deviate from that average. If the numbers are higher than what you’d expect, there’s always the possibility that this could simply be due to natural variation in a given year – or due to chance. The issue is to determine whether the appearance of a cluster represents an actual phenomenon, or is simply a one-time occurrence.
By definition a cluster is an occurrence of a disease that is more frequent than the norm, but with cancer, the background, so common, and the numbers of potential cluster cases in a community so small, this can be very difficult to determine. Clusters are based on “exquisitely rare events, often in small communities,” says William Harage, an epidemiologist at Harvard. “Identifying these sporadic cases across a lot of noise takes sophisticated stats.” In a country of 350 million, the odds are that multiple instances of cancer or other non-infectious diseases are going to show up at the same time and place.
The issue is to determine whether the appearance of a cluster represents an actual phenomenon, or is simply a one-time occurrence.
Even when the numbers do point to the existence of a cluster, determining the “cause” can be difficult, to say the least. Environmental toxins also exist at a baseline level (just think about the levels of smog and other air pollutants across the country), and finding evidence that that a specific toxin or pollutant might be higher in areas of a disease cluster – and determining how humans could come into contact with them – is quite difficult and connecting the dots is even harder. Look how long it took to "prove" that smoking caused cancer.
The CDC’s Agency for Toxic Substances and Disease Registry (ATSDR) points out that after researchers investigate some of the underlying causes, the appearance clusters can fade away. According to the ATSDR, an apparent cluster of brain cancers “might actually be found to include patients with metastases from distant sites, patients with non-neoplastic infiltrative diseases [diseases in which cells migrate], and even patients who have suffered strokes,” who have all just been lumped into the same category on an epidemiological spreadsheet or hospital database.
Some prospective clusters are even due to reporting errors, such as incorrectly reported laboratory results or clinical findings, or coding problems in hospital discharge summaries. In other words, when we pull apart the situation and address each factor involved, the “cluster” may no longer be there.
The clusters that do occur – and there’s no doubt that they do – are serious issues. The list of harmful chemicals used in manufacture continues to grow, and we are always behind the curve when it comes to understanding how these compounds affect our bodies. For this reason, legislation regulating how the chemicals are used and disposed of is in its infancy. Orders like the Toxic Substances Control Act seek to monitor the life of chemicals, but doing so is a monumental task that has many challenges, and many critics.
Some prospective clusters are even due to reporting errors, such as incorrectly reported laboratory results or clinical findings, or coding problems in hospital discharge summaries.
Controlling these substances is a two-part problem. One part is scientific: pinpointing the environmental link to disease is difficult. The other part is legal and legislative: assigning responsibility, developing regulations to protect the public and enacting those regulations is even harder. Says Emory's LaMorte, “Part of the issue is the government’s uncertainty about how the environment contributes to cancer in the first place. More research is needed first, and more education of the public. This other issue is, quite frankly, Congress. Some politicians are actually focused on removing regulations to promote economic growth.” Along the same lines, businesses themselves may not be eager to change their ways, which are often long-established and costly to overhaul.
Unfortunately, for the people who experience them, disease clusters can tell us a lot about how the environment and the body interact. Changing the ways in which clusters are tracked and evaluated will be up to epidemiologists, environmental specialists, and the government working together to address the issues at hand. A few more Erin Brockovitch-style activists probably wouldn't hurt either.
