CHANCE News 2.20
        (25 November to 11 December 1993)


Prepared by J. Laurie Snell as part of the
CHANCE Course Project supported by the
National Science Foundation and the
New England Consortium for Undergraduate
Science Education.

Please send suggestions to:

Current and previous issues of CHANCE
News can be found on our chance gopher.
Just point your gopher to:

We thank all those who returned the survey.  Your
reponses were encouraging and helpful.  We would
still be happy to hear from anyone who has not


  "If you need statistics to prove it,
   it isn't true."
            One of Barbara Doyle's Professors




>>>>>==========>> Probability of Surgeons Spreading HIV. JAMA, 6 Oct 1993, letters Ed Gracely pointed out, in the October 6,1993 issue of JAMA, that there are letters from himself and others commenting on articles that deal with estimating the probability that a patient will become infected by a surgeon who has the HIV virus. The letters observe that the authors of these articles made a number of mistakes in calculating such things as the 95% confidence interval for the probability of a transmission, given that n patients have been screened and none found to be infected. The upper limit of the confidence interval can be obtained by solving the equation (1-p)^n =.05 for p. Carrying this out for the example n = 1174, that appears in one of the original papers gives a 95% confidence that the chance of infection does not exceed about 1 per 400, compared to the author's result of about 1 per 23,000. The author of one of the original papers responded that they used the Central Limit Theorem which, they agree, is not very applicable to binomial trials with such a small probability for success. <<<========<<

>>>>>==========>> Proper Use of Percentages. The New York Times We mentioned in the last CHANCE News that the Labor Department planned to use an improved survey for measuring unemployment. Peter Doyle suggested that the following, quoted from the New York Times, would be useful in a Chance course when discussing the proper use of percentages: >> Under the redesigned survey, in the 12 >> months through August, the unemployement >> rate was 7.6 percent, a half-point above >> the 7.1 percent that the department had >> reported. The rate for women was 6.8 >> percent, rather than 6.0 percent; for >> men, the figure went to 6.9 percent >> from 6.7 (Note that the percentages for men and women are both lower than the overall percentage.) <<<========<<

>>>>>==========>> New York Region Expecting to Gain in Jobless Recount. The New York Times, 22 Nov 1993, A1 Robert D. Hershey When the Labor Department tried out their new unemployment survey, they found that New York, New Jersey, and Pennsylvania apparently stand to gain millions of dollars from Federal grants tied to unemployment figures. The Labor Department is baffled why this is the case and caution that their practice survey was only based on a sample of 12,000 as compared to 60,000 for the official survey. One of the causes for the error in the practice survey was the way they asked the first question. If the person who came to the door was an adult women they asked: "What were you doing most last week, keeping house or something else?" Unemployed women who said they were keeping house were then not counted as unemployed. <<<========<<

>>>>>==========>> Does Exposure to Second-Hand Smoke Increase Lung Cancer Risk? Chance, Fall 1993, 11-14 Alan J. Gross In 1986, the National Research Council and the U.S. Surgeon General released reports suggesting an increased risk in lung cancer in nonsmokers due to environmental tobacco smoke (ETS). The data used to make this claim came from 33 epidemiologic studies estimating the risk for lung cancer for a person who does not smoke, but lives with a person who does. The author states that a meta analysis of these 33 studies leads to a relative risk of 1.13 with 95% confidence intervals of 1.00 to 1.28 which he observes is not significant. He gives these relative risks and confidence intervals for the individual studies and points out that the overall picture is not very convincing. He discusses some of the obvious possible biases in these studies -- determination of the cause of death, as well as misclassification of smoking status -- and concludes that these problems could easily account for the apparent reported risk in many of these studies. <<<========<<

>>>>>==========>> What Evidence Is Needed to Link Lung Cancer and Second-Hand Smoke? Chance, Fall 1994, 15-18 Howard E. Rockette This author responds to the above article by suggesting that it is necessary to consider the risk assessment of environmental tobacco smoke (ETS) from a more general point of view. He observes that guidelines for establishing cause and effect proposed by epidemiologists, and used by agencies such as the Environmental Protection Agency, include several different requirements such as biological plausibility, existence of dose-response relation, consistency of response across studies, etc. Evaluations are made on the total weight of evidence and seldom are all requirements established to everyone's satisfaction. From this point of view, he argues that biological plausibility has been established because it has been shown that passive smoking subjects a person to the same carcinogens as active smoking. Animal studies have also shown an association between ETS and cancer. Thus he argues that the role of the epidemiological studies and meta analyses is just to estimate risk from exposure to low doses of the carcinogen. He observes that 27 of the 30 studies included in the 1992 report of the Environmental Protection Agency estimated risks greater than one. He finds these studies as convincing as the previous author found the 33 used by the 1986 reports unconvincing. The author admits, however, that these studies, not being controlled studies, are subject to the standard problems of confounders, bias, etc. Still, as a public policy question, he argues that when a carcinogen has been established, it is reasonable to attempt to control it even when it is not feasible to gather adequate statistical power to detect low risks. <<<========<<

>>>>>==========>> On Cigarettes, Health and Lawyers. The New York Times, 6 Dec 1993, D1 Michael Janofsky A group of flight attendants and heirs of attendants who have died brought a class action suit in Florida against the tobacco companies contending that exposure to cigarette smoke caused the attendants' illnesses and deaths. The lawer for the group took sworn testimony from some the executives of the tobacco companies. The Times article quotes exchanges from this testimony. Here is one sample, but you should read all their answers. They are really amazing! The lawyer is questioning Bennett S. Le Bow, Chairman of the Brooke Group Ltd., owner of the Liggett Group: >> Q. Is it fair to say, then, that since >> you're satisfied that you have a legal >> right to sell cigarettes, you have never >> really explored or studied the issue of >> whether or not cigarettes cause disease? >> A. That is absolutely correct. >> Q. If I asked you, does smoking cause lung cancer... >> A. I don't know. >> Q. O.K., and you really don't care because >> you're selling a legal product? >> A. Correct. One executive had quite a hard time trying to be honest about advice he would give his fifteen year old son on smoking while, at the same time, not saying anything really bad about smoking. Another had to struggle with questions about his wife and children's smoking, especially since they did not even smoke his company's brand of cigarettes. <<<========<<

>>>>>==========>> Clinton vs. Kennedy: Similarities in the Stock Market? Wall Street Journal, 6 Dec 1993, C1 John R. Dorfman A graph of the daily closing Dow Jones Average is compared from the beginning of campaign for the presidency and from the beginning of Clinton's campaign and a "striking similarity" is noted. We are now at a point when both averages were a maximum and if the similarity continues we are due for a serious fall in the market. Stock market experts attempt to explain this similarity but one of them remarks "But history isn't a Xerox machine. It would be interesting to see how many students in a Chance course think this similarity is more than a coincidence.. <<<========<<

>>>>>==========>> An Association between Air Pollution and Mortality in Six U.S. Cities. New England Journal of Medicine, 9 Dec 1993, 1753-1759 Douglas W. Dockery and others Previous studies have reported an association between particulate air pollution and mortality rates, but these studies have been criticized because they often did not control for cigarette smoking and other health risks. The aim of this study is to measure this association while controlling for these risk factors. The study is a prospective cohort study involving adults who participated in the Harvard Six Cities Study. It followed 8,111 adults randomly chosen from the six cities: Watertown, Massachusettes; Harrington, Tennessee; St. Louis, Missouri; Steubenville, Ohio; Portage, Wisconsin; and Topeka, Kansas. Data was obtained on the level of air pollution from various particle types and sizes and on mortality rates of the subjects over a sixteen year period. A significant association was found between mortality and inhalable, fine particles. This was true, also, after controlling for individual differences in other risk factors including age, sex, cigarette smoking, educational level, body-mass index, and occupational exposure. Steubenville and St. Louis had the highest pollution rates and the lowest survival rates for the subjects in the study. The article has nice graphics that display the differences between the cities in pollution and in mortality rates. <<<========<<

>>>>>==========>> Preventing Asthma Epidemics Due to Soybeans by Dust-Control Measures. New England Journal of Medicine, 9 Dec 1993, 1760-1176 Joseph M. Anto and others. The demonstration of cause and effect between air pollution and mortality is marred by the lack of evidence demonstrating biologic plausibility. Nevertheless, this article demonstrates convincingly that soybean dust released into the air during the unloading of ships into a harbor silo, with no dust-control apparatus, was the cause of the asthma epidemics described during the past decade in Barcelona, Spain. <<<========<<

>>>>>==========>> Gallup Takes Its Polls to Huge China Market. Chicago Tribune, 2 Dec 1993, Business 1 Lisa Anderson The Gallup Organization has become the first and only foreign market research firm licensed to operate in the People's Republic of China. A representative of Gallup said "We're not too far away from an era where we will have overnight worldwide polling, the ability to take the pulse of mankind around the globe". An initial Gallup survey carried out in August for the American Chocolate Manufactures Association showed that the average Chinese buys candy every 20 days and that 55.6 percent of them snack at least once daily. Of these snackers, 27 percent prefer chocolate. Gallup currently plans no political-attitude polling in China, although they said that the government did not place restrictions on the firm's activities in terms of location, content or subject matter. <<<========<<

>>>>>==========>> IIT Researchers Using Rats To Test Magnetic Fields. Chicago Sun-Times, 28 November 1003, Sunday News 10 Jim Ritter A five-year, $8.3 million project began in August at the Illinois Institute of Technology to test whether magnetic fields from electricity cause cancer or other health problems in mice and rats. The study uses about 8,000 rats and mice. One group of animals receives a high, continuous exposure 18 1/2 hours a day. Another group gets a high exposure every other hour. Two other groups get a lower exposure on a continuous and an alternating basis. A fifth group receives no exposure. The project includes four studies that will look for cancer, birth defects, reproductive problems and changes in blood and hormones. <<<========<<

>>>>>==========>> Women Now Outnumber Men In College. The Houston Chronicle, 25 Nov 1993, G6 Kenneth Eskey The Census Bureau has issued a new report on school enrollment that shows that 56 percent of college students in 1992 were women, up from 43 percent in 1972. The report pointed out that 17 percent of college students were 35 or older, up from 9 percent in 1972. Of this 17 percent age 35 or older, two-thirds were women indicating that a significant number of older women are going to college. <<<========<<

>>>>>==========>> Flawed Gene Called Link To Colon Cancer. Boston Globe, 3 Dec 1993, National 1 Richard Saltus The race to find the gene that causes colon and rectal cancer is over. It was not won by the Hopkins scientists who last year identified a region of Chromosome 2 that contained the gene, but rather by a much smaller group from the University of Vermont and the Dana-Farber Cancer Institute who said they were just lucky to have noticed a connection between this study and some of their previous work. Everyone carries two identical copies of the gene, but about one in 200 inherit a flawed copy. These individuals have about an 80% chance of developing colon or rectal cancer sometime in their lifetime. Testing for the gene is expected to be possible in about a year. These cancers are treatable when detected early, so this testing should decrease the death rate and families with a history of these cancers could find out by a single test if they are at high risk instead of the present situtation requiring contstant monitoring with expensive test. <<<========<<

>>>>>==========>> Mammogram Guideline is Dropped. The New York Times, 5 Dec 1993, 1 30 Gina Kolata Since 1987, the National Cancer Institute and the American Cancer Society have recommended that women in their 40's have mammograms every one to two years and that women over 50 have them every year. The National Cancer Institute has changed its formal guidelines, dropping its recommendation that women under 50 should have regular mammograms. They say that experts agree that mammograms reduce the death rate for women over 50, but that studies have failed to demonstrate that screening tests for women under 50 save lives. A spokesman for the Institute said that they are trying to get out of the business of providing guidelines and instead, will simply provide the scientific facts as they are. The American Cancer Society is keeping its guidelines that suggest mammograms for women under 40 based on their feeling that there is "presumptive evidence" to recommend them. In addition to the lack of evidence that mammograms for women under 50 save lives, mammograms in younger women have a high false positive rate causing significant emotional stress and sometimes unnecessary treatment. <<<========<<

>>>>>==========>> Researchers Question Under-50 Breast Tests. The Boston Globe, 24 November 1993, p6 byline: Los Angeles Times A study of 32,000 asymptomatic women 30 and older found a high rate of "false positives" in women under 50--that is, women who have abnormal mammograms that do not turn out to be cancer- related. For every 1000 women under 50 having their first mammogram, 53 will have an abnormal finding, resulting in 102 diagnostic procedures. These will lead to the finding of only 2 cancers, one of which will be in invasive. (Over the age of 50, the corresponding figures are 70 abnormal findings, 132 procedures, and 10 cancers, most of which will be invasive.) Noting the cost of the procedures, and the attendant anxiety, the researchers recommend that women under 50 do not undergo routine mammography unless they have a family history of breast cancer. The article notes that this recommendation is sure to be controversial <<<========<<

>>>>>==========>> The news on mammograms tosses women to the wind. Susan Trausch, op/ed column The Boston Globe, 10 December 1993, p27. ...and here is some of that controversy. The argument here is that the new recommendations above are partly politically motivated: President Clinton's proposed health care package won't pay for mammograms for women under 50, so it's quite convenient that the government now doesn't have to recommend those tests. There's a lot to discuss in that statement. But without passing judgment on the thesis, we can at least bemoan the treatment of statistics here. "Statistically insignificant...[means]...some breast cancer deaths are preventable, but not enough to impress the number crunchers. Of course, if the preventable death happened to be yours or mine...we might consider the statistics quite significant, even miraculous." <<<========<<

>>>>>==========>> Mind-body Link in AIDS Studied: Contradictory Findings on Effect of Depression on Immune System. San Francisco Chronicle, 1 Dec 1993, A3 In AIDS therapy, a major effort has been made to encourage a positive outlook toward health and to treat the depression that often accompanies the onset of symptoms. The current issue of the Journal of the American Medical Association reports on two studies that have investigated the effect of depression on the progress of AIDS. The two studies are remarkably similar in the way that they were carried out, but they led to somewhat different results. The study carried out by researchers at the University of California at San Francisco found that, among the 330 HIV-infected men in the study, the decline in CD4 counts was significantly faster among the men who were depressed than it was among the men who were not depressed. They found no significant difference between depression and the time elapsed from HIV diagnosis to AIDS or death. The study carried out at Johns Hopkins involved 1800 HIV positive men. In this study the researchers did not find a significant decline in the CD4 counts nor in the time from HIV diagnosis to AIDS or death between those had a high level of depression and those who did not.. In an accompanying editorial, two psychiatrists comment that both studies found that most HIV patients are not depressed. They recommend that depression continue to be considered as a treatable disorder rather than an expected response to HIV infection. They say that in AIDS "the development of HIV-related symptoms increases the likelihood of depression, but that depressive symptoms do not in themselves increase the progression of HIV disease." <<<========<<

>>>>>==========>> Conference Urges Delaying Use of AZT Until Later in Disease. Houston Chronicle, 1 Dec 1993, A21 Ruth Sorelle A recent conference convened by the U.S. National Institute of Allergy and Infections Disease reports in the current issue of JAMA their recommendations that no anti viral treatment be undertaken if the CD4 cell count is at or above 500. These recommendations are a result of the findings of the European Concorde study that found that AZT did not appear to be effective in slowing the disease in early stages of HIV infection. <<<========<<

>>>>>==========>> International Report Card Shows U.S. Schools Work. The New York Times, 9 Dec 1993, A1 William Celis 3d This article begins: >> In contrast to studies that have >> painted grim portraits of American >> education, a new report finds that, >> compared with other industrialized >> nations, the United States does a >> reasonably good job of educating its >> citizens and preparing them for work. >> The report, issued yesterday by the >> Organization for Economic Cooperation >> and Development, based in Paris, >> indicated that American students lag >> only slightly behind their counterparts >> around the world in math and science and >> that a higher than average percentage of >> American students get college educations. As you read on, however, you wonder about the basis for this positive spin on mathematics and science. The article reports that: >> In math, American students scored >> at the bottom of the scale, just behind >> Spanish students and ahead of only >> Portuguese students. In science, >> American students scored about the >> same as their counterparts in Canada, >> England, France, Scotland and Spain. The United States does have a higher than average percentage of college graduates. It also spends more per student than almost any other country. But the report states that "the extra dollars went not into the classroom, but to workers other than teachers." This article is a good one to have students read carefully to figure out how to make the different statements and statistics consistent. <<<========<<

>>>>>==========>> Heart Attack Risks Shown in Burst of High Activity. The New York Times, 2 Dec 1993, A18 Jane E. Brody Two large studies reported in the current issue of the New England Journal of Medicine provide the strongest evidence to-date that a sudden burst of physical activity can cause a heart attack. The studies also showed that regular physical activity helps prevent heart attacks. One study involved 1,228 heart attack patients in the United States and the other 1,194 patients in Germany. The American study found that the chance of a sedentary person suffering a heart attack during or just after heavy exercise was 50 times greater than that for people who exercise five or more times a week. The American researchers pointed out that, while the relative risk was high, the absolute risk is still low for a sedentary person who shovels snow. They observe that a typical non- smoking 50-year-old has a one-in-a-million chance of suffering a heart attack during a given hour period. They write: >> If this man was habitually sedentary >> but engaged in heavy physical exertion >> during that hour, his risk would >> increase 100 times over the base-line >> value, but his absolute risk during >> that hour would still >> be only one in 10,000. (I wonder if the one in a million is really correct, if the typical 50-year-old person is shoveling snow during that hour.) One researcher also took pains to point out that sexual activity was considerably less strenuous than the types of exercise that precipitated heart attacks in the study. Both research teams allowed the subjects to act as their own controls by asking them what they were doing one hour before the heart attack and at the same time one day before. As a check, they interviewed non-patients of the same age and sex to determine what they were doing during the hours in question. The article also provides a description of the various activities typical of levels of physical activity on a scale from 1 to 8 in terms of MET (metabolic equivalent) units. <<<========<<

>>>>>==========>> Jury Still Out on DNA Evidence. The Recorder, 29 Nov 1993 Richard Barbieri This article outlines the confusion in the California courts as to whether DNA fingerprinting can be admitted as evidence in criminal trials. This confusion has come largely from last year's report from the National Research Council that criticized some of the statistical methods used by the FBI. The Council has put together another panel to reconsider its recommendations about the statistical portions of DNA testimony. Defense lawyers are fearful that this is just an attempt by the FBI to get their way. The lawyers point out that the FBI attacked the first report and are helping to fund this second one. <<<========<<

>>>>>==========>> Claim of Higher Risk for Women Smokers Attacked. Science, 26 Nov 1993, 1375 Gary Taubes It has generally been thought that the only reason that more men than women die of lung cancer from smoking is because women don't smoke as much. As the Oxford statistician Peto has remarked, "If women smoke like men, they'll die like men". A study of Canadian smokers published in the American Journal of Epidemiology concluded that, in fact, smoking may be two to three times more hazardous for women than for men. The researchers identified all lung cancers among women in southern Ontario between 1981 and 1985. For each of the 442 cases, they identified a randomly selected female control matched by age and area of residence. Then they identified 403 male lung cancer cases from the same hospitals and randomly selected male controls. They found that women had a risk 27.9 times as great as non-smokers, while men smokers were only 9.6 times as likely to have lung cancer than the non-smokers. Two other recent studies found similar results. A large, ongoing study by the American Cancer Society and the National Cancer and the National Cancer Institute contradicts the findings of these three recent studies. Other epidemiologists say the results are misinterpretations of epidemiologic data. The editor of the journal Epidmiology remarked that concluding that women have a higher risk by comparing these two ratios is a "meaningless exercise". Among other problems with the comparison, he cited the fact that non-smoking men may have a higher risk for lung cancer then women do becaues of their occupation etc., giving a higher base. The methods employed in this study would then lead to a higher rate for women smokers, even when smoking results in equal absolute risk for men and women. Authors of the study reply that their difference was so large that it could not be accounted for by these considerations, but critics still disagree. <<<========<<

>>>>>==========>> !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! CHANCE News 2.20 (25 November to 11 December 1993) !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Please send suggestions to: jlsnell@dartmouth.edu >>>==========>>|<<==========<<<