What Does “No Elevated Cancer Rates” Near Rocky Flats Actually Mean?

An exchange with Colorado’s cancer registry raises important questions about residential history, latency, migration and the limits of existing cancer surveillance

Rocky Flats Downwinders recently asked the Colorado Central Cancer Registry, part of the Colorado Department of Public Health and Environment (CDPHE), to conduct a small-area cancer-incidence analysis for ZIP code 80007.

The request sought cancer data from 2010 through the most recent complete registry year, including observed and expected cases, age-adjusted rates, standardized incidence ratios and confidence intervals for cancers associated with radiation exposure. We also asked whether results could be examined by age, sex and smaller time periods.

CDPHE explained that it can provide cancer counts by ZIP code but cannot calculate standardized incidence ratios at that level because, in its assessment, reliable population denominators are unavailable.

More importantly, CDPHE’s response revealed a central limitation in using cancer-registry data to investigate historical environmental exposures: cancer cases are generally associated with where a person lives when diagnosed—not necessarily where that person lived when exposure may have occurred decades earlier.

That distinction is especially important at Rocky Flats, where nuclear-weapons production occurred from 1952 until 1989 and radioactive and hazardous materials were released into surrounding communities.

Below is the substance of CDPHE’s response, followed by Rocky Flats Downwinders’ reply.

CDPHE’s Response

Kieu Vu, lead analyst and epidemiologist for the Colorado Central Cancer Registry, thanked Rocky Flats Downwinders for the request and explained that the registry takes community cancer concerns seriously.

Regarding the requested analysis of ZIP code 80007, CDPHE stated:

“We are able to provide counts by zip code; however, standardized incidence ratios cannot be calculated by zip code because there are no reliable resources for population at that geographic level.”

CDPHE explained that cancer is common while confirmed cancer clusters—especially clusters linked to environmental exposure—are rare. According to the agency, adult cancer generally has multiple contributing factors, including genetics, lifestyle, viruses, body weight and exposures accumulated over a lifetime.

CDPHE also explained that the health consequences of exposure depend on:

• What substance a person encountered

• How the exposure occurred

• The amount of exposure

• How long and how frequently the exposure occurred

  
  

For childhood cancers, CDPHE noted that identifying potential causes is especially difficult because the total number of pediatric cancers is small. Although radiation may cause childhood cancer, exposure during childhood may also increase cancer risk later in life.

CDPHE described several characteristics that may make a suspected cancer cluster particularly concerning:

• Many more cases of a particular cancer, or group of related cancers, than expected

• Cancer occurring in an unusual age group

• A rare type of cancer

• Scientific evidence connecting the suspected substance with the cancer

  
  

The registry directed Rocky Flats Downwinders to the state’s Colorado Health Information Dataset, known as CoHID. The system provides cancer counts and crude and age-adjusted incidence rates for several demographic and geographic categories.

Available statistics may be reviewed by factors such as:

• Cancer site

• County

• Year or combination of years

• Sex

• Age

• Race

  
  

CDPHE stated that information not available through CoHID requires a more detailed process for releasing record-level cancer data for research and analysis. The agency also directed us to its Colorado Central Cancer Registry and Rocky Flats information pages.

CDPHE concluded:

“CDPHE has monitored cancer incidence rates near Rocky Flats since the 1980’s and has not found elevated rates of cancer among the residents living near the area.”

It is this statement—and what population it actually describes—that prompted our response.

Rocky Flats Downwinders’ Reply

Dear Ms. Vu,

Thank you for your detailed response and for explaining the limitations of the Colorado Central Cancer Registry and the data available through CoHID.

After reading your response, I do not think it is necessary to pursue my original ZIP-code data request as submitted. Instead, your explanation raises a more fundamental question about what conclusions can reasonably be drawn from the cancer registry regarding people who may have been exposed to contaminants from Rocky Flats.

You noted that cancer risk reflects exposures over a lifetime, while cancer surveillance generally considers where a person lives when diagnosed. That seems particularly important in the case of Rocky Flats.

Many people who lived in Arvada, Westminster and surrounding communities during Rocky Flats’ operating years no longer live there. If someone grew up near Rocky Flats, moved away and developed cancer 10, 20 or 30 years later, it appears that the cancer would be associated with the person’s residence at diagnosis rather than the former residence near Rocky Flats.

Conversely, someone who moved into the area shortly before being diagnosed could be counted as a cancer occurring among current residents near Rocky Flats. If I understand the registry correctly, this creates an important limitation when registry data is used to evaluate a historical environmental exposure.

I am therefore trying to understand how CDPHE reconciles that limitation with the statement in your email that:

“CDPHE has monitored cancer incidence rates near Rocky Flats since the 1980’s and has not found elevated rates of cancer among the residents living near the area.”

What population does “residents living near the area” actually represent? Does it mean people who lived near Rocky Flats during a potentially relevant exposure period—or people whose residence was near Rocky Flats when they were diagnosed?

Those appear to be very different populations.

Geographic boundaries

You explained that standardized incidence ratios cannot be calculated by ZIP code because reliable population denominators are not available at that geographic level. If ZIP codes are not an appropriate unit of analysis, has CDPHE considered using census tracts or groups of census tracts to examine smaller geographic areas around Rocky Flats?

I am also concerned about the effect of using broad geographic areas. Environmental contamination does not necessarily follow county, municipal or ZIP-code boundaries. If a potentially affected population represents only one part of a much larger study area, combining that population with surrounding communities could dilute a localized pattern.

Cancer types and ages

Your response correctly notes that unusual cancer patterns can be particularly important when cancers are rare or occur at unexpectedly young ages. That raises the question of whether analyses of “all cancers” or broad age groups could obscure patterns involving particular cancers, related groups of cancers or younger populations.

Small numbers and statistical uncertainty

Small numbers present another problem. I understand that rare cancers in small populations produce wide confidence intervals and that patient-confidentiality requirements limit what can be publicly reported. However, an absence of statistical significance in a small population is not necessarily evidence that the cancer rate is normal. It can also mean there are too few observations to determine whether an apparent elevation is real.

That distinction is important when communicating results to the public. For that reason, I would appreciate clarification of what CDPHE means when it says it has “not found elevated rates of cancer” near Rocky Flats. Does that mean:

• Observed cancer rates were not higher than expected?

• Some rates were higher than expected but were not statistically significant?

• No statistically significant elevations persisted across multiple study periods?

• Or does it mean something else?

  
  

I would also appreciate knowing whether historical Rocky Flats cancer studies accounted in any way for:

• Latency between potential exposure and cancer diagnosis

• Residential history

• Migration away from communities near Rocky Flats

• Exposure periods during plant operations

• Geographic boundaries based on potential contamination rather than political jurisdictions

  
  

I want to emphasize that I am not asking the Cancer Registry to determine whether Rocky Flats caused an individual’s cancer. I understand why registry data cannot establish that.

My concern is different.

I am trying to determine whether the methods being used are capable of answering the question the community believes is being answered.

If the registry primarily measures cancer among people according to where they live at the time of diagnosis, it may be useful for monitoring current geographic cancer incidence while being much less capable of evaluating a population defined by historical residence and potential exposure decades earlier.

If that is correct, Rocky Flats Downwinders believes this limitation should accompany public statements that cancer monitoring around Rocky Flats has not identified elevated cancer rates.

I would genuinely appreciate your thoughts on this distinction. I would also appreciate references to any Rocky Flats cancer analyses that incorporated residential history, migration, latency or exposure-based geographic boundaries rather than relying primarily on residence at diagnosis.

Thank you again for taking the time to respond and for your willingness to continue the conversation.

Sincerely,

Tiffany Hansen Director, Rocky Flats Downwinders

Why This Exchange Matters

This correspondence does not establish that cancer rates are elevated near Rocky Flats. It does, however, raise a different and essential question: Can studies based largely on residence at diagnosis reliably evaluate health outcomes among people who may have been exposed near Rocky Flats many years or decades earlier?

A finding that a study did not detect a statistically significant elevation is not automatically proof that no elevation exists. The meaning of such a finding depends on the population studied, the geographic boundaries, the cancers examined, the number of cases, the length of follow-up and whether the study adequately addressed latency and migration.

Rocky Flats Downwinders is asking CDPHE to clearly distinguish among:

• No elevation was observed

• An elevation was observed but was not statistically significant

• The available data were insufficient to reach a reliable conclusion

• The study design was not capable of evaluating historically exposed residents who later moved away

  
  

Communities deserve to understand both what a health study found and what its methodology could not determine.

Rocky Flats Downwinders will continue seeking a transparent explanation of these limitations and advocating for research designed around the history of potential exposure—not only a person’s address at the time of diagnosis.