Comments on When Are Cancer Cases More Than Coincidence?

• Posted by Moniqa , librarian on February 17, 2009 at 7:10am EST

• Yes, this kind of things are a bit strange... My father used to work as a teacher in a small school in Sweden. They were only 7 teachers there, but three of them got cancer, one of them was my father. One woman died, one is cured, and my father is still struggling... It happened 12 years ago.

• victims of bad epidemiology
• Posted by boiscience on February 17, 2009 at 7:35am EST

• This is sad. Some one should do some cancer risk perception education here. Like pointing out that 1 in 333 buildings have such "cancer clusters."

• Posted by jcl on February 17, 2009 at 10:10am EST

• No, 1 in 333 buildings do not have a cancer cluster. Rather, out of all buildings that do have such a cluster, the cluster can be explained by chance in 1 of 333 cases.

• ?????
• Posted by m on February 17, 2009 at 10:20am EST

• One certainly sympathizes with cancer victims, but the lack of any causal link makes these fears sound highly questionable.

• Epidemiology
• Posted by JAW on February 17, 2009 at 11:15am EST

• Out of all buildings that have 130 women workers [not "out of all buildings that do have such a cluster"], a cluster like this will arise in 1 of 333 cases, if cancer is distributed by chance (rather than being caused by something in the immediate environment).

• Posted by Keith Johnson , Cancer Cluster Probability on February 17, 2009 at 1:00pm EST

• As the American Cancer Society calculates the odds of a woman contracting breast cancer in her lifetime as one in nine, the odds of a handfull of 130 women getting breast cancer appears to be close to 100 percent, unless the odds are drawn from a series of limiting characteristics, each of which could be countered by another, opposed characteristic. The calculation of "cancer cluster" odds by the public had become so distorted by the late 1980s that the CDC had conducted 1,500 investigations without finding a single case verifying the existence of such a cluster. Consequently the existence of a cancer cluster has to be highly doubtful in any case, and this university building hardly rises to the top of candidates.

  A second consideration not mentioned in the article is the length of time a person takes to develop cancer from precancerous conditions, many years, even decades. So how long have these affected women worked in said building or for the university? If less than many years, their cases have to be removed from the total.

  All in all, the university administration is probably wise in responding to the women's emotional demands. However administrators should not do anything that appears to support the claims that something in the building causes cancer.

• Posted by jcl on February 17, 2009 at 1:00pm EST

• "A UCSD medical report last year found odds of 1 in 333 that chance alone could explain the incidence of breast cancer in the building."

  If this sentence in the article is written clearly, then it does not follow that there is a 1 in 333 chance of this kind of cancer cluster occurring in a given building with 130 women workers. It is a conditional probability -- given that such a cancer cluster exists, what is the probability that it is a coincidence? Say that you choose randomly from two coins, one fair and one with a 3/4 probability of heads, and then you get two heads in a row. The probability that you picked the fair coin given the two heads in a row is 4/13, even though the probability of getting two heads given that you picked the fair coin is 1/4. The probability of no environmental effect given a certain distribution of cancers is not the same as the probability of a certain distribution of cancers given no environmental effect.

• Epidemiology part 2
• Posted by JAW on February 17, 2009 at 2:10pm EST

• To jcl (et al.):
  "If this sentence in the article is written clearly"... then I would agree, but I am pretty sure that the sentence is poorly constructed. I have not read the UCSD medical report (I could not find it on-line), but as a teacher of statistics I've read countless reports like it. I am assuming that the UCSD folks did the usual thing: They fit a model and are now reporting a P-value for a hypothesis test. But they are being a bit inaccurate with their wording (which is _exceedingly_ common, even within the scientific literature). It is possible that they conducted a Bayesian analysis and are now reporting the posterior odds - in which case I would want to know what prior probability they used, and why, and I would prefer that they reported the Bayes factor - but I doubt it.

• Posted by kcs on February 17, 2009 at 7:35pm EST

• I am an employee of this building and have read the report. While the report doesn't state absolute causality, it does identify that the elevator motor emits EMF levels above the standard considered safe within a 60 foot radius. Almost all the cases of cancer have occurred in workers who have offices within 40 feet of this motor, well within the EMF zone. Causality is difficult to prove; likelihood is not. Furthermore, it is actually a 1 in 3,333 chance the high rates of cancer are coincidence.

• Sad
• Posted by concerned on February 17, 2009 at 7:35pm EST

• I have good friends who worked in that building and have become ill. It is sad. I am so glad that finally some notice is being given. The campus is the site of an old army base. Could that have anything to do with it?

• Posted by Laura on February 17, 2009 at 7:35pm EST

• "...the lack of any causal link..."

  May I suggest that we do not know all there is to know about cancer. A cluster like this one, rather than being dismissed out of hand because we don't KNOW of a causal link, should be viewed as an opportunity to look for causal links we don't know about yet so as to avoid them in future. We might not turn up anything, but if we don't look we definitely won't turn up anything.

• stress
• Posted by Math Prof on February 17, 2009 at 11:05pm EST

• Could the stress of believing one is in a cancer hot spot lead to increased risk? There would likely be some health or mental health risks induced by such stress.

  Have they looked into common links outside the building? For example, say employee X brings in homemade cookies once a week and there is some contaminate in his/her home. Or maybe a lot of people in that building volunteer to keep a certain park clean. I would assume they'd looked into smoking rates and other obvious risk factors or household radon levels.

• UCSD Cancer Cluster
• Posted by Mark , Construction Inspector on February 18, 2009 at 9:21am EST

• Below is text from the NIH cancer web page. Please read it before creating a panic about elevators. Let the investigator complete the study before leaping to conclusions.
  "Several studies conducted in the 1980s and early 1990s reported that people who worked in some electrical occupations (such as power station operators and phone line workers) had higher than expected rates of some types of cancer, particularly leukemia, brain tumors, and male breast cancer (2). Some occupational studies showed very small increases in risk for leukemia and brain cancer, but these results were based on job titles and not actual measurements. More recently conducted studies that have included both job titles and individual exposure measurements have no consistent finding of an increasing risk of leukemia, brain tumors, or female breast cancer with increasing exposure to magnetic fields at work..." mjs

• Nay Sayers in this thread.
• Posted by Chris on February 18, 2009 at 9:55am EST

• We need to go beyond one sentence of the article and look at the bigger picture. OK, I have read elsewhere that the EMF emitted by these elevators is equivalent to 100+ houses put together.

  And Keith Johnson's outrageous statement above about "emotional women's demands" is really crass. People's wives, grandmothers, mothers and sisters get cancer. Have some respect, please.

• Breast Cancer Cluster at UCSD
• Posted by Geoffrey Kabat , Senior Epidemiologist at Albert Einstein College of Medicine on February 18, 2009 at 9:20pm EST

• The cluster of eight breast cancer cases in the literature building at UCSD (who were diagnosed since 2000) is most likely due to chance, and the proposed eight-week study by a UCLA epidemiologist is unlikely to turn up a common environmental exposure that could explain the cluster. Here’s why.

  As mentioned by a previous commentator, breast cancer takes perhaps two decades to develop before it is diagnosed. This raises the question of how long the eight women were working in the literature building.

  Furthermore, from over forty years of study of the epidemiology of breast cancer, we know that the main determinants of breast cancer risk are: age at menarche, age at first full-term birth, number of children, age at menopause, and family history of breast cancer. Additionally, long-term hormone replacement therapy is associated with a modest increase in risk, and obesity is a risk factor for postmenopausal breast cancer. Finally, higher socioeconomic status is associated with increased risk. Thus, insofar as we have identified risk factors for breast cancer, these are things a woman carries with her wherever she moves.

  The possibility that environmental exposures play a role in causing breast cancer became a focus of intense interest starting in the early 1990s, and substantial efforts have been devoted to examining such factors, including organochlorine compounds (such as DDT and PCBs); chemicals produced by the combustion of fossil fuels; and electromagnetic fields from power lines and electrical appliances. Environmental exposures, which generally are at very low levels and which vary over time, are difficult to study. Nevertheless, the results of studies that have been done to date do not provide support for a role of environmental factors in the development of breast cancer.

  The only environmental exposure for which there is strong evidence of an effect is exposure to ionizing radiation from the atomic bomb dropped on Hiroshima, which is associated with substantially increased risk of breast cancer, particularly in women who were exposed during adolescence, when the breasts are developing.

  The possible effect of exposure to electromagnetic fields on breast cancer risk has been the focus of three large case-control studies carried out on Long Island, NY, Los Angeles, and Washington state. These studies involved taking detailed measurements of electromagnetic fields in the homes of hundreds of women diagnosed with breast cancer as well as in the homes of similar numbers of comparison women without breast cancer. These studies showed no hint of an association.

  Furthermore, a larger number of studies has examined whether use of electric blankets in the decades preceding a diagnosis of breast cancer is associated with increased risk. These studies also showed no evidence of an effect.

  Electromagnetic fields produced by power lines and appliances have very low frequencies and energies and are not to be confused with “ionizing radiation.’
  Electromagnetic fields encountered in the course of everyday life are far below the strengths at which any biological effects have been detected. Experimental studies have also failed to yield evidence of a mechanism by which ambient electromagnetic fields could affect breast cancer risk.

  If large and carefully-designed studies of electromagnetic fields (and other environmental exposures) and breast cancer have not managed to detect any evidence of a link, it is highly unlikely that the crash survey undertaken by Professor Kheifets will uncover a cause among these eight breast cancer cases.
  _______

  Geoffrey Kabat, Ph.D., a cancer epidemiologist at the Albert Einstein College of Medicine, has conducted studies of breast cancer and environmental factors and is the author of HYPING HEALTH RISKS: ENVIRONMENTAL HAZARDS IN DAILY LIFE AND THE SCIENCE OF EPIDEMIOLOGY (Columbia University Press, 2008).

• 8 cases or more?
• Posted by Chris on February 19, 2009 at 5:45am EST

• When I was at UCSD in the 1990's (before they started keeping track cancer in that building), I recall one or two other women who were graduate students who had cancer or tumors.

  I appreciate the previous post, but I am still quite nervous about this situation. What was the author of the original report referring to when he stated that there higher EMF is associated with elevated risk of cancer?

• Confirmed: At least 4 cases in 1990's; how many more??
• Posted by Chris on February 19, 2009 at 6:50pm EST

• OK, through networking with people I knew at UCSD in the 90's, I can say there are at least 4 more cases of cancers and tumors, beyond the ones that were counted starting in 2000.

• Report
• Posted by Kelly on February 23, 2009 at 4:04pm EST

• Here's a link to the report findings on the cancer cluster.

  http://www.signonsandiego.com/more/lit-building/

• UCSD Breast Cancer Cluster and the Garland Report
• Posted by Geoffrey Kabat, Ph.D. , Dept. of Epidemiology and Population Health at Albert Einstein College of Medicine on March 2, 2009 at 12:00pm EST

• In response to the occurrence of an apparent cluster of 8 cases of breast cancer diagnosed between 2000 and 2008 among women faculty and staff in the Literature Building at the University of California at San Diego, the university administration engaged a professor in the Department of Family and Preventive Medicine, Cedric Garland, to conduct an assessment of potential environmental hazards in the building. Professor Garland submitted a thirty-page report to the chancellor of the university on June 3, 2008 http://www.signonsandiego.com/news/metro/images/090217elevatoreport.pdf .

  The report considers potential hazards from molds, chemical, and radiological exposures, finding no support for these as playing any role in the breast cancer cluster. The final issue, which receives the most attention, is the possibility that exposure to electromagnetic fields (EMF) from the elevator and electrical equipment may have contributed to the cluster.

  Professor Garland concludes that “some possibility exists” that exposure to electromagnetic fields on the first floor of the building in very close proximity to the electrical and elevator equipment rooms could have “contributed modestly” to increased risk, although he states that “this exposure is unlikely to be a principal cause of breast cancer that has been diagnosed in people who have worked in this small area” (pp. 20-21).

  The Garland report is clearly being taken seriously by the university administration, which has hired a UCLA epidemiologist specializing in EMF to conduct a two-month investigation of the cluster. In addition, the author’s identification of EMF as a possible hazard has caused anxiety and anger, and has prompted calls for the administration to take action to remediate the hazard http://www.insidehighered.com/news/2009/02/17/cancer. For these reasons, it is important to take a critical look at the case Professor Garland makes for a role of EMF.

  Professor Garland implies that weak EMF (as low as 2 milliGauss) can play a role in causing cancer, a belief that is contradicted by a massive amount of scientific evidence. His references to the relevant literature are limited and selected. Furthermore, he shows an uncritical acceptance of the few studies he does cite, even though these have clear deficiencies and are at variance with the preponderance of the evidence. Tellingly, he fails to cite some of the major epidemiologic studies on the topic of EMF and breast cancer or to acknowledge the large body of experimental evidence that has failed to find reproducible evidence of biological effects.

  According to the National Research Council (1), ambient levels of 60 Hz magnetic fields “in residences and most workplaces” are typically 0.1-3.0 milliGauss. The NRC concludes that “no conclusive and consistent evidence shows that exposures to residential electric and magnetic fields produce cancer, adverse neurobehavioral effects, or reproductive and developmental effects” (NRC, p. 2).

  A critical reading of the Garland report raises a number of specific issues, questions, and ideas for further study, which are spelled out below.

  Specific Points:

  1. Time period defining the cluster

  Garland defines the time period during which 8 cases of breast cancer were diagnosed in the Literature Building as 2000-2008. However, he notes that an additional case of breast cancer was diagnosed in 1991 (p. 6). If there were no other cases diagnosed between 1991 and 1999, then the occurrence of breast cancer over the period 1991-2008 is half that in the later period. This information is relevant to assessing the existence of a cluster, as well as a possible cause, especially since, presumably, the elevator on which Garland focuses attention was in operation during the earlier period. If there is complete ascertainment of breast cancer cases occurring in faculty and staff working in the building over the entire period, this alone might suggest that the “cluster” of 8 cases between 2000 and 2008 could be due to chance, and additionally that it is not linked to EMF from the elevator.

  2. Age and duration of employment

  Professor Garland gives the ages at diagnosis of the 8 affected women and rightly concludes that the relatively young age (mean age at diagnosis was 53 years) is due to the fact that this is a working population.

  However, a crucial piece of information for assessing the role of any environmental exposure in the building is the length of time each woman worked in the Literature Building – when she started working in the building and if she stopped, when. This information is not provided, although it should be readily available.

  3. Location in the building

  Another basic piece of information necessary to draw any inferences about possible exposures is the office location(s) of each of the 8 women over their period of employment. Furthermore, it would be helpful to have a map, or floor plan, of the building, floor by floor, showing the location of the offices where each case worked and the location(s) of the elevator and equipment room as well as the electrical riser.
  
  4. Clinical characteristics of cases

  It would also have been valuable to obtain clinical information to determine whether there was any unusual pattern among the cases or whether they were similar to breast cancer cases (of comparable age) in terms of tumor size, stage, grade, histology, hormone receptor status, HER-2/neu status, family history, etc. This information would have to be obtained from the patient’s medical chart or the statewide California Cancer Registry. Professor Garland does mention that two of the cases had inflammatory breast cancer and that both died. Inflammatory breast cancer is a rare condition, accounting for approximately 3% of all breast cancer cases.

  The presence or absence of an unusual pattern in the distribution of clinical parameters could shed light on the etiology of these cases. For example, if an unusually high proportion of the cases had a family history of breast cancer in a first degree relative, this would help account for the cluster.

  5. References to the Scientific Literature

  In what purports to be a scientific report attempting to elucidate possible causes of a cluster, one would expect that the relevant literature on EMF and breast cancer would be cited. Three large case-control studies of residential EMF exposure and breast cancer have been published (2-4). These were conducted in Seattle, Washington; on Long Island, New York; and in Los Angeles. None of the 3 studies showed any evidence of an association. (Only one of these studies is cited by Garland). In addition, a larger number of studies have examined the association of electric blanket use and breast cancer (5, 6). Electric blanket use, especially use of older blankets, entailed some of the highest EMF exposures encountered in everyday life. Nevertheless, these studies also show no evidence of an association. (No reference is made to these studies in the report).

  Garland cites 3 references from the literature on EMF and breast cancer (his refs. 7-9). One of these (Wertheimer & Leeper, 1982 [7]) used wiring configurations rather than measured fields to estimate EMF exposure. This paper, which had an elaborate sampling scheme which is difficult to evaluate, reported an association of high wire configuration with increased likelihood of death from cancer in both men and women. No separate analysis of breast cancer was presented. However, a more recent analysis of residential studies (8) indicates that there is no consistent relationship between wire configuration and measured fields, casting doubt on this result.

  A second study cited by Garland (9) reported a very small and mostly not statistically significant increased risk in women with occupational exposure to EMF. This study made use of self-reported job histories, rather than measured fields.

  The last study cited by Garland (the study in Seattle, WA, by Davis, et al., 2002 [2]) found no evidence of a link.

  Thus, Garland’s statement that “Some epidemiological and laboratory studies have linked exposure to residential levels of electromagnetic fields from high electrical current configurations … to breast and other cancers, although the literature on this association is mixed (p. 12)” is misleading. In fact, consideration of the totality of the epidemiologic and experimental literature on EMF and breast cancer and cancer generally indicates that weak fields of the kind encountered in everyday life do not increase the risk of cancer. Numerous comprehensive reports and reviews support this position (1, 10, 11). In addition, the possibility that weak EMF can affect biology has been called into question on theoretical grounds (12, 13).

  A fourth article cited by Garland pertains to an in vitro study indicating that EMF can increase resistance to the effects of tamoxifen in breast cancer cells. And on the basis of this one study, he states in his conclusions that “importantly, such exposure could interfere with treatment using tamoxifen” (p. 21). Has this finding be replicated? How many other relevant studies are there and what do they show? This is another reflection of the sort of lax, ad hoc, and selective citation of the relevant evidence, which could be expected to create enormous anxiety in any woman receiving tamoxifen treatment.

  It needs to be emphasized that there are no replicated effects of weak fields below 50 milliGauss (11-13).

  Another basic fact that one might have expected Professor Garland to mention is that the strength of magnetic fields falls off rapidly with increasing distance from the source.

  6. Use of consultants

  Professor Garland engaged Edward Leeper as a consultant in his evaluation of EMF in the Literature Building. Leeper was co-author on the original report which suggested that exposure to high current configurations from neighborhood power lines and transformers increased the risk of childhood cancer (Wertheimer and Leeper, 1979 [14]) and of adult cancers (Wertheimer and Leeper, 1982 [7]). Although Professor Garland refers to Leeper with reverence as the “co-discoverer of the association of EMF and cancer,” this early work has generally not been confirmed by more detailed and rigorous studies. As mentioned above, Professor Garland is totally uncritical in assessing Wertheimer and Leeper’s work, and this bias toward belief in effects of weak EMF appears to have influenced the conclusion of his report.

  7. “Prudent avoidance”

  The report invokes the principle of “prudent avoidance” to justify shutting down the elevator and changing the location of its motor. While the idea of reducing one’s exposure to a potential hazard sounds reasonable, as stated above, after extensive study there is no convincing evidence of any hazard from weak EMF. In this situation, advocating prudent avoidance does a disservice by misrepresenting what is known and by promoting anxiety.

  8. Another question

  One obvious question is: Is the Literature Building an anomaly or are there other buildings on the UCSD which have similar elevators located on their ground floors? If yes, how many are there, and how many breast cancer cases occurred in each of these buildings?

  9. Conclusion and future study

  Although the report is dated June 3, 2008, it appears to have only been made public in February, 2009, sparking concern in the UCSD community. In response to the Garland report and the concern that it has created, the university administration has hired a UCLA epidemiologist, Professor Leeka Kheifets, who has specialized in studying EMF, to undertake a 2-month study of the cluster, which is currently in progress. One wonders what discussions took place between June 3, 2008 when the Garland report was submitted and February, 2009, and what kind of critical review the report received. It appears that the poorly-justified focus on EMF in the report has been accepted at face value by those who hired Professor Kheifets.

  The conclusion of the Garland report that EMF from the elevator in the Literature Building may have contributed to the cluster of 8 breast cancer cases is based on a selective and biased interpretation of the extensive epidemiological and experimental literature on EMF exposure and the risk of breast cancer. Rather than collecting all of the information relevant to describing the cluster, due to Professor Garland’s preconceived ideas about EMF, his investigation focused attention on EMF, thereby causing unnecessary alarm, distrust, and confusion in members of the UCSD community.

  In order to further evaluate the significance of the cluster, it would be extremely valuable to conduct an historical cohort study of all female UCSD employees who worked in the Literature Building as of 1/1/1990. The cohort could be assembled from UCSD employment records and/or 1990-2008 UCSD telephone directories. Breast cancer incidence in the cohort could be determined from the California Cancer Registry, California Master Death File, direct interview of the surviving subjects, and other related sources. Such a study would measure the actual population-based breast cancer incidence rate in this cohort. This study would have to be undertaken by objective epidemiologists/statisticians with the requisite experience in long-term follow-up of existing cohorts in California.

  References

  1. National Research Council. Possible Health Effects of Exposure to Residential Electric and Magnetic Fields. Washington, D.C.: National Academy Press, 1997.

  2. Davis S, Mirick DK, Stevens RG. Residential magnetic fields and the risk of breast cancer. Am J Epidemiol 2002; 155: 446-454.

  3. Schoenfeld ER, et al. Electromagnetic fields and breast cancer on Long Island: a case-control study. Am J Epidemiol 2003;158:47-58.

  4. London SJ, et al. Residential magnetic field exposure and breast cancer risk: a nested case-control study from a multi-ethnic cohort in Los Angeles County, California. Am J Epidemiol 2003;158:969-980.

  5. Sandler DP. On blankets and breast cancer (editorial). Epidemiology 2003;14:509.

  6. Kabat GC, et al. Electric blanket use and breast cancer on Long Island. Epidemiology 2003;14:514-520.

  7. Wertheimer N, Leeper E. Adult cancer related to electrical wires near the home. Int J Epidemiol 1982;11:345-355.

  8. Ahlbom A, et al. A pooled analysis of magnetic fields and childhood leukemia. Br J Cancer 2000;83:692-698.

  9. McElroy JA, et al. Occupational exposure to electromagnetic field and breast cancer risk in a large population-based, case-control study in the United States. J Occup Environ Med 2007;49:266-274.

  10. National Institute of Environmental Health Sciences. NIEHS Report on Health Effects from Exposure to Power-Line Frequency Electric and Magnetic Fields. NIH Publication no. 99-4493, Washington, D.C., 1999.

  11. Lacy-Hulbert A, Metcalfe JC, Hesketh R. Biological responses to electromagnetic fields. FASEB J 1998;12:395-420.

  12. Adair RK. Constraints on biological effects of weak extremely-low-frequency electromagnetic fields. Physical Rev A 1991;43:1039-1048.

  13. Bennett WR, Jr. Cancer and power lines. Physics Today (April, 1994):23-29.

  14. Wertheimer N, Leeper E. Electrical wiring configurations and childhood cancer. Am J Epidemiol 1979;109:273-284.

  ________

  Geoffrey Kabat, a cancer epidemiologist at the Albert Einstein College of Medicine, has conducted studies of environmental factors, including EMF, in relation to breast cancer risk and is the author of HYPING HEALTH RISKS: ENVIRONMENTAL HAZARDS IN DAILY LIFE AND THE SCIENCE OF EPIDEMIOLOGY (Columbia University Press, 2008).