A Warning From 2002: Bernd Franke Questioned CDPHE's Rocky Flats Cancer Study

Tiffany Hansen

Jan 1, 2002

6 min read

More than two decades before Rocky Flats Downwinders launched its Truth & Transparency project, an environmental researcher raised serious concerns about the methodology of a Colorado cancer registry study. His letter reveals how long questions about the state's approach to studying community health have remained unresolved.

A letter from 2002 raises familiar questions

On December 2, 2002, Bernd Franke of the Institute for Energy and Environmental Research (ifeu) in Heidelberg, Germany, wrote a letter to attorney Caron Balkany concerning a Colorado Department of Public Health and Environment (CDPHE) cancer study.

The study was titled Ratios of Cancer Incidence in Ten Areas Around Rocky Flats, Colorado Compared to the Remainder of Metropolitan Denver, 1980–89 with Update for Selected Areas, 1990–95.

Franke had been asked to provide his professional opinion of the report. His response was direct. In a one-page letter, he identified four concerns about the study's methodology and scientific usefulness. More than two decades later, those concerns are strikingly similar to the questions Rocky Flats Downwinders is raising through our Truth & Transparency project.

Concern 1: What could the study actually detect?

Franke's first concern involved the study's detection limits. He argued that without explaining what size of cancer increase the study was capable of detecting as statistically significant, outside reviewers could not properly evaluate its findings. This is a fundamental issue in epidemiological research.

A study may find no statistically significant increase in cancer incidence, but that result must be interpreted in light of the study's ability to detect an increase if one exists. For example, a study might have sufficient statistical power to identify a large increase in a common cancer but insufficient power to detect a smaller increase or an increase in a rare cancer. Without clearly explaining those limitations, readers may incorrectly interpret a statistically nonsignificant finding as proof that no increased risk exists. Franke's concern was not simply about the numbers reported by CDPHE. It was about whether the study provided enough information for others to judge the meaning of those numbers.

Concern 2: An insensitive study design

Franke's second observation was that the study design was insufficiently sensitive.

He wrote:

"It would require substantial increases in cancer incidences to reveal anything using this study design."

This concern relates directly to the first. If a study can detect only relatively large differences in cancer incidence, smaller elevations may go unidentified or fail to reach statistical significance. That does not mean those elevations necessarily exist. It means the absence of a statistically significant finding may not rule them out.

For Rocky Flats Downwinders, this distinction is particularly important when discussing historical environmental exposures. Cancer is not one disease. Different cancers have different risk factors, latency periods, and frequencies within a population. Studies of relatively small geographic populations may face substantial limitations when examining uncommon diagnoses or small changes in incidence. Franke was questioning whether the methodology was capable of answering the health questions being asked.

Concern 3: The comparison population may also have been affected

Franke's third concern addressed how CDPHE selected its comparison population.

The study compared cancer incidence in ten areas around Rocky Flats with incidence in the remainder of metropolitan Denver. Franke questioned the usefulness of that comparison because, in his view, the larger Denver population had also potentially been affected by releases from Rocky Flats. This raises an important methodological question. When researchers compare a potentially exposed population with another population, the interpretation depends partly on whether the comparison population provides a meaningful baseline.

If both populations experienced some exposure, a difference between their cancer rates may be smaller than it would be when compared with an otherwise similar, unexposed population.

That possibility does not establish that the Denver comparison population experienced harmful exposure levels. But it does highlight why researchers must evaluate exposure differences between populations rather than rely on geography alone. Franke concluded that the design did not enable readers to assess cancer risks from Rocky Flats releases.

For RFD, this concern connects directly to our questions about how CDPHE defines study populations and what conclusions can reasonably be drawn from its geographic cancer registry analyses.

Concern 4: Was the study designed to reassure the public?

Franke's final observation was his most pointed.

He wrote:

"It appears that the study design was chosen to calm people down, for public relations purposes, rather than for any real scientific reason."

This is Franke's expressed opinion about the apparent purpose of the study design.

The letter does not provide evidence establishing that CDPHE deliberately selected its methodology to reassure the public, nor does it document the department's internal decision-making. That distinction matters.

But the criticism raises a legitimate question about public-health communication:

When a study has significant methodological limitations, how should its findings be communicated to the people whose health concerns it is supposed to address? Scientific limitations should be explained plainly, whether the findings are reassuring, concerning, or inconclusive. The public deserves to understand both what researchers found and what their methodology could not determine.

The connection to CDPHE's later cancer studies

The letter reviewed a cancer study covering diagnoses from 1980 through 1989, with updates for selected areas through 1995. It did not evaluate CDPHE's later cancer registry analyses.

Nevertheless, the methodological questions have significant parallels. In 2017, CDPHE epidemiologist Mike Van Dyke acknowledged that the state's registry analysis could not conclusively determine whether exposure to Rocky Flats emissions had caused cancer.

He explained that stronger conclusions would require better individual exposure estimates and greater consideration of population migration.

Those limitations are important because geographic cancer registry studies generally associate cases with residence at diagnosis rather than reconstructing where individuals lived during earlier periods of possible exposure. Later, in correspondence with Rocky Flats Downwinders, CDPHE representative Kieu Vu described the department's work as monitoring cancer incidence rates near Rocky Flats since the 1980s and stated that elevated rates had not been found among nearby residents. These statements deserve to be examined together. Van Dyke acknowledged important limits on what the research could establish. Vu's later statement summarized the findings without those same qualifications.

And Franke's 2002 letter shows that questions about the sensitivity and interpretability of geographic cancer studies were being raised much earlier. The fact that similar concerns appear across different decades does not establish that the studies were scientifically invalid.

But it does establish a history of methodological questions that merits a clear, documented response.

Why this letter matters to Truth & Transparency

Rocky Flats Downwinders' Truth and Transparency project seeks greater clarity about the studies used to evaluate public-health concerns surrounding Rocky Flats.

We are particularly interested in three issues that align closely with Franke's observations.

Study methodology: Were the studies sufficiently sensitive to detect meaningful changes in particular cancer rates, and were those limitations clearly disclosed?

Comparison populations: Were the populations selected for comparison appropriate for investigating potential health effects from historical Rocky Flats releases?

Public communication: Have CDPHE's summaries accurately represented the findings, uncertainties, and limitations of its studies?

These are questions about the quality and interpretation of scientific evidence. They do not require presuming that a particular study reached an incorrect conclusion or that researchers acted with improper intentions. Instead, they require examining the original reports, explaining the methods, and making the basis for public statements understandable.

A broader review is needed

The 2002 Franke letter is one more reason Rocky Flats Downwinders believes the historical public-health record warrants independent, multidisciplinary review.

Such a review should include epidemiologists, statisticians, exposure scientists, public-health specialists, and researchers with experience evaluating historical environmental contamination.

It should examine not only published conclusions but also the underlying assumptions, statistical power, geographic boundaries, comparison populations, and data limitations.

Importantly, it should distinguish between detecting cancer incidence patterns and determining whether historical environmental exposure contributed to disease.

Those are related questions, but they are not the same scientific undertaking.