Document DdXVy8owGmngBYJmZDEkYyYDB

vo/.5, 303-311. April 1996 / Cancer Epidemiology, Biomarkers & Prevention 303 Review The Practice of Causal Inference in Cancer Epidemiology Douglas L. Weed' and Lester S. Gorelic preventive Oncology Branch [D.L. W.] and Comprehensive Minority Biomedical ProgrEim [L.S.G.1, National Cancer Institute. Bethesda. Maryland 20892 Abstract Caw1 inference is an important link between the practice of cancer epidemiology and effective cancer prevention. Although many papers and epidemiology textbooks have vigorously debated theoretical issues in causal inference, almost no attention has been paid to the issue of how causal inference is practiced. In this paper, we review two series of review papers published between 1985 and 1994 to find answers to the following questions: which studies and prior review papers were cited, which causal criteria were used, and what causal conclusions and public health recommendations ensued. Fourteen published reviews on alcohol and breast cancer and 6 published reviews on vasectomy and prostate cancer were examined. For both series of reviews, nearly all available published studies were cited except for ecological studies and prior reviews. Sources of causal criteria were often not provided. When they appeared, all citations were either the 1964 Surgeon General's report or works of Austin Bradford Hill. Reviews often excluded and sometimes altered criteria without giving reasons for these changes. The criteria of consistency and strength of association were almost always used accompanied by dose-response and biological plausibility in a majority of reviews. The criterion of temporality, considered by many methodologists to be a necessary causal condition, was infrequently used. Confounding and bias were often added considerations. Public health recommendations were not discussed in nearly one-half of the reviews. Introduction What causes cancer? How much and what kinds of evidence lead to causal conclusions? When should public health recommendations be made? These are central questions of causal inferencein cancer epidemiology. Causal inference has been an Important methodological focus for at least half a century (1,2). A classic point of reference is the mid-l960s, when two papers, one by a committee appointed by the Surgeon General (3) and the other by Austin Bradford Hill (4),listed what are usually called causal criteria. These criteria are shown in Table 1. At the time, they were used to support the claim that smoking z 5/23/95; revised 12/4/95; accepted 12/5/95. The Costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in y r d a n c e with 18 U.S.C. Section 1734 solely to indicate this fact. To whom requests forreprints should be addressed, at National Cancer Institute, T-41, 9000 Rockville Pike, Bethesda, MD 20892. Phone: (301) 496-8640; Fax: (301) 402-0863, causes lung cancer (3) and to guide causal inference for occupational and environmental diseases (4). From 1965 to 1995, many associations have been examined in terms of the central questions of causal inference. Causal inference is often practiced in review articles and editorials. There, epidemiologists (and others) summarize evidence and consider the issues of causality and public health recommendations for specific exposure-cancer associations. The purpose of this paper is to take a first step toward systematically reviewing the practice of causal inference in cancer epidemiology. Techniques used to assess causation and to make public health recommendations are summarized for two associations: alcohol and breast cancer, and vasectomy and prostate cancer. The alcohol and breast cancer association is timely, controversial, and in the public eye (5). It involves a common exposure and a common cancer and has a large body of empirical evidence; over 50 studies and over a dozen reviews have appeared. The association between vasectomy and prostate cancer also involves a common exposure and a common cancer. It is a relatively recent controversy with a smaller empirical literature and relatively few reviews. The intent of this "review of reviews" is not to criticize individual reviewers nor to make judgments regarding associations. The limitations of making summary statements from small selected samples are recognized. Recommendations for possible improvements to the practice of causal inference are offered. These are based upon the summary findings and upon our understanding of two areas of related research: the methods and theory of causal inference and the scientific quality of review papers. Materials and Methods The MEDLINE and CANCERLIT data bases were searched from 1977 through June 1994,using the keywords "alcohol and breast cancer" and "vasectomy and prostate cancer." Reference lists of primary research studies and reviews identified were perused for mention of relevant articles. Tables of contents from major medical, cancer, public health, and epidemiology journals available at the libraries of the NIH were also examined. For both associations, three types of review papers were found: editorials in which a review of existing literature was included and either cause or public health recommendations were considered, reviews devoted to the specific association, and "mini-reviews" subsumed within reviews of more general issues. Typically, these mini-reviews were relatively brief sections in papers on alcohol and cancer, diet and cancer, breast (and prostate) cancer epidemiology, and male sterilization. Only editorials and reviews devoted solely to the specific associations are included in this paper, with one exception, an IARC monograph on alcohol and cancer with a review of the alcohol and breast cancer literature (6). Editorials and reviews are subsequently referred to as "reviews." For each, the following questions were addressed: .I .r. I' * li 1: `li 1 304 Review: Causal Inference in Cancer Epidemiology Table I Causal criteria as described by A. B. Hill (4) and by the Surgeon General'scommittee (3) Consistency" Strength" Dose-response Biological plausibility Specificity" Temporality" Experimentation Coherence" Analogy a Criteria described by the Surgeon General'scommittee (3). (3); in both, the original five criteria were used. The remaining four reviews cited works of Austin Bradford Hill (4, 97). only one instance was a list of nine criteria used in the assessment (go), although "epidemiological sense" replaced &. terion of coherence of Hill. Chu (76) cited Hill (4) but used only four of his nine criteria: consistency, strength, biological crdibility (Le.,plausibility), and experimentation. Longnecker (63) also cited Hill (4) and used a subset comprising consistency, strength, dose-response, and biological plausibility. Lowenfels et al. (75) cited Hill (97) and used five criteria. In the eight reviews in which no source for criteria is cited, the number of criteria used varied from one (consistency; Ref. 72) to six of fie original nine of Hill (4, 77). I 1 I I I I 1 Overall Use of Causal Criteria and Other Consideratiom. I1.. I; s ( Which studies are included? (Asked as a check for systematic exclusion.) ( Which causal criteria are used and what sources for these criteria are cited? ( Are other considerations (e.g., study design, bias, and confounding) featured? ( What causal conclusions and public health recommendations are made? ( What is the relationship of a conclusion or recommendation to the results of original research studies published by the same author? ( Do reviewers cite the methods, conclusions, or recommendations of earlier reviews? Because the 1964 Surgeon General's list of causal criteria form a subset of the 1965 criteria of Hill, and because reviewem citing no source nevertheless used many of the same criteria, the overall use of criteria can be examined. Table 3 shows that the criterion of consistency was always used and strength of association was nearly always used in this series of reviews. Dose-response and biological plausibility were also used relatively frequently. In only 4 of 14 reviews was the criterion of temporality discussed. The need for a randomized trial and its ethical and practical feasibility was considered in 3 of the 14 reviews. Coherence and analogy were rarely used. Table 2 reveals that bias (7 of 14) and confounding (11 of 14) were the most common additional considerations in this series. Causal Conclusions. The question of cause was addressed in ReSUltS 10 of 14 reviews. None claimed cause. Conclusions included reasonably clear rejections of the idea (78, 80, 81) and those I Alcohol and Breast Cancer that left the issue undecided (70,73,74,77,82). Two reviewers The consumption of alcoholic beverages has been linked with breast cancer since 1974 (7). In the ensuing two decades, ecolog- ical studies (8-1l), case-controlstudies (12-50), and cohort studies (51-60) have been published, representing a total of just over 50 studies. Metaanalyses have also appeared (61-63). Reports from a workshop (64-67) and papers discussing aspects of published cohort studies (68,69) have contributed to the controversy. Twenty-two reviews (63, 70-90) were published between 1985 and 1994, appearing every year except 1986. Fourteen satisfied inclusion criteria (63, 70-82) and are summarized in Table 2; three were published as editorials (70-72). appeared to lean in the direction of concluding cause. Longnecker (63) for example, argued that the "evidence appears to be growing stronger. " Hiatt (79) argued that the data were "not yet sufficient to conclude ... causation." Public Health Recommendations. Public health recommendations were addressed in 8 of 14 reviews. Three of the eight reviews included recommendations for reductions in alcohol consumption, with an emphasis on women at high risk of breast cancer and without cardiovascular disease risk factors (71,72, 79). Recommendations were considered premature in two of the eight reviews (76, 80). Another reviewer argued that total ! Studies Included. The vagaries of the publication process make precise determination of studies available to authors difficult, beyond the obvious condition that studies published less than 1 year before publication of the review were probably not available. There is little evidence of systematicexclusionof studies, although ecologicalstudies were irregularly cited,as were abstracts (91-96). Three reviews published since 1991 used methodological criteria to exclude studies from consideration (63, 80, 82). Occasionally, reports were not included in reviews for several years after publication. An example is a case-control study on diet and breast cancer conducted in Greece (21).Published in 1986,this study was cited in reviews published in 1988 (73), 1989 (77), 1991 (80),and 1994 (63) but not in seven other reviews published between 1988 and 1993. A possible explanation is that authors of reviews may not have been aware that in this study alcohol consumption was measured as one component of the diet. A review published in 1994 (63) identified seven case-control studies (43-49) published between 1978 and 1992that were rarely cited in previous reviews. Six of these seven did not mention "alcohol" in their titles. Use of Causal Criteria and Sources Cited. Sources of causal criteria were cited in six of fourteen reviews (63, 75, 76, 78-80).Two of the six cited the 1964Surgeon General's report abstinence would be required and therefore could not be recommended ( 7 3 , whereas others would not advocate alcohol beverage consumption at any level (78). One reviewer noted that "before any recommendations are made ... a risk benefit analysis is in order" (63). Perhaps the strongest recommenda- tion to reduce alcohol beverage intake (in high-risk women) was included in a 1987 editorial (71) accompanying the publication of two cohort studies in the same issue of the New England Journal of Medicine (53, 54). The author wrote that "whatever can be done...should be done." Four of the eight reviews in which public health recommendations were considered did not discuss cause (71.72, 75, 76). Finally, two papers published within 1 year of each other used identical criteria (3) on almost the same data, and yet One reviewer recommended reductions of alcohol intake for highrisk women (79) and the other did not (78). Authorship of Reviews and Studies. Half of the reviews (7 of 14) were written by individuals who were also authors of studies on the same topic. The author of a positive cohort study argued in a later review that women at high risk should limit their use of alcohol (79) and the authors of two negative case-control studies argued against causation (78) or that rec- I I Cancer Epidemiology, Biomarkers & Prevention 305 ommendations were premature (76). On the other hand, authors of a positive cohort study (74,77) and of a positive case-control study (82) made no public health recommendations nor claimed cause. A clear relationship between causal conclusions or public health recommendations and the results of studies conducted by the reviewers was not apparent. Reviewers Citing Prior Reviews. In only 3 of the 13 reviews was the conclusion of a prior reviewer about causation or public health recommendations mentioned (79-81). Citations of prior reviews were patchy. For example, a review published in 1991 (80) mentioned the opposing interpretations of two reviews published in 1989 (78) and 1990 (79) but did not cite the seven earlier reviews. A review in 1990 (79) mentioned four earlier reviews but conclusions from only two (70, 78). All reviews published in 1989 or earlier made no mention of the conclusions of earlier reviewers about cause or public health. The review by Plant (81) mentioned nearly all prior reviews but included conclusions from only three (71, 77, 78). In some instances, reviewers mentioned biases (82) or possible biological mechanisms (77) discussed in earlier reviews. Summary of Findings. ( Reviewers appear to cite nearly all available published studies except for ecological designs. Reasons for specific exclusions are infrequently given. ( Reviewers irregularly cite sources of causal criteria. All citations are either the 1964 Surgeon General's criteria (3) or works of Austin Bradford Hill, primarily his classic 1965 paper (4). ( In citing Hill, reviewers often exclude and sometimes alter criteria. Reasons for these changes are not provided. ( Independent of citation status, subsets of the criteria of Hill (4) are most frequently used. ( Consistency and strength are almost always invoked, accompanied by dose-response and biological plausibility in a majority of reviews. ( Confounding and bias are often added considerations. ( Public health recommendations (both for and against changes in current policy or practice) appear without claiming or disclaiming cause. ( Reviewers rarely address the methods or conclusions of prior reviews. vasectomy and Prostate Cancer The health effects of vasectomy have been studied since the late 1970s (98). Several prospective studies of morbidity rates in Vasectomized men (99-lOl), case-control studies (102-107), and cohort studies (108-114) have examined the relationship between vasectomy and prostate cancer. Seventeen studies have been published; accompanying them have been 19 publications reviewing or commenting upon the studies (115-133). Six reviews satisfied inclusion criteria (115-120), two of which originally appeared as editorials (115, 116). The findings from the reviews on vasectomy and prostate cancer matched those from alcohol and breast cancer, with these exceptions: ( Dose-response was not used as a causal criterion because of the all-or-none nature of the exposure. ( Because five of the six reviews were published in the Same year, only one earlier review was available for citation. one of the five (118) cited it. ( No reviewers were also authors of original research on a e topic. Discussion To what extent do these findings accurately represent the practice of causal inference in epidemiology? Has a methodological paradigm been described? We turn to the language popularized by the work of the historian and philosopher of science Thomas Kuhn (134) because it is useful in this context, although we are mindful of its many meanings and possible linguistic devaluation in scientific writing (135, 136). The core idea is that a paradigm is shared by members of a scientific discipline and represents the normal practice of that discipline (137). The practice of causal inference in only two exposurecancer associations may not be representative of the practice within cancer epidemiology. Other reviews, summarizing approaches to causal inference for other associations, will help decide the issue. The fact that the findings for alcohol and breast cancer reviews were supported by the vasectomy and prostate cancer series lends support to the idea that a good approximation to a current methodological paradigm has been described. Also important is the fact that no reviewer cited as a source of his or her approach to causal inference any paper after the mid-1960s (the only exception a 1971 version of a 1965 classic by the same author). This situation is consistent with the definition of a paradigm as a classic work [i.e.,as an "exemplar" (134,136)]. Indeed, many papers and textbooks use the original criteria of either the Surgeon General (3) or Hill (4) as the foundation for their treatment of causal inference methodology (138-148). Assuming that a good approximation to the current practice paradigm has been described, we discuss two categories of findings: the use of causal criteria and the "results" that emerge from their use, causal conclusions and public health recommendations. Regarding causal criteria, the most frequently used were consistency, strength, biological plausibility, and (for alcohol and breast cancer) dose-response. Temporality and experimental evidence were much less frequently used, whereas coherence, analogy, and epidemiological sense were almost never used. We take no issue at this time with the emphasis on consistency, strength, biological plausibility, and dose-response (when appropriate). Nor do we question the relative infrequency with which the criterion of experimental evidence was used, given that no such evidence exists for either association. We are intrigued, however, by the lack of emphasis placed on temporality and coherence. In the alcohol and breast cancer series, temporality was considered in only 4 of 14 reviews. In the vasectomy and prostate cancer series, temporality was never used. This is a situation in which practice and theory appear to diverge. Both Hill (4) and the Surgeon General`s committee (3) recognized the importance of demonstrating the temporal priority of exposures over diseases, especiallyfor slowlydevelopingdiseasessuch as cancer. With rare exceptions (141), recent authors consider temporality to be indispensable (149), crucial ( l a ) , a directly deducible (150) sine quu non of causality (142), not a sufficient condition for causality but a necessary one (143, 144). In a 1981 list of ranked criteria, temporality is among the top four, after experimentation,strength, and consistency (151). A 1993 textbook puts temporality first in a ranked list (147). Jnterestingly, HilI (4) argued that "the temporal problem may not arise often." This is a reasonable assertion if temporality refers only to the situationin which a purported cause is really an effect of the disease, so-called "reverse causation" (146). Such a narrow definition of the criterion makes it easily dispensable in some circumstances. In the alcohol and breast cancer series, for example, two reviews claimed that temporality could be considered moot, given cohort evidence demonstrating 0 2 z8 dz s c 8 ,..aW-*"s z08 02 k, 8 V 0 z8 BW z8 z 0 z8 0 z8 Y W0 z8 z8 0 i: M 3 ea 8 V U au -, 0 z8 W 0W s z6 s m z! e.>- d I Cancer Epidemiology, Biomarkers & Prevention 307 WW z6 z8 m L d 0 C0 z" di 5 4=' 308 Review: Causal Inferencein Cancer Epidemiology alcohol consumption before the clinical onset of breast cancer (75, 80). Why was coherence used so infrequently? The simplest answer is that the reviewers may have equated it with plausibility, thus making it redundant. In the two classic sources, the Surgeon General's committee (3) did not carefully distinguish between the two criteria; plausibility was embedded within coherence. Hill (4), on the other hand, distinguished these criteria on logical grounds. For him, an association is plausible if it is consistent with current biological knowledge; an association is coherent if it does "not seriously conflict with the generally known facts of the natural history and biology of the disease." In the more recent methodological literature, this issue remains unresolved (146,152-155). A recent paper emphasizes plausibility but ignores coherence (146). whereas another favors a broad definition for coherence deemphasizing plausibility (152). Perhaps more important than distinguishing between these two criteria is specifying and categorizing the many types of biological evidence used in causal inference (156). With regard to causal conclusions and public health recommendations, it is interesting that reviewers sometimes consider both and sometimes do not mention one or the other. No reviewer in these series offered an explanationfor his or her specificchoice. In the absence of such, we offer some possibilities. Consider, for example, the reviews in which public health recommendationsdid not appear. Perhaps reviewers judged that public health actions should not be taken and therefore made no mention of them. Alternatively,reviewers may not include public health recommendations because they see them as policy decisions and therefore as inappropriate to the professional practice of epidemiology (157). There is no way to easily distinguish between these possibilities. Furthermore, some reviewers include public health recommendations without explicitly considering, much less concluding, cause. The mots of this practice may be found in the seminalpaper of Hill (4). There, he noted that public health action may be recommended in situations in which insufficient evidence exists to firmly con- clude causation. Hill also suggested that different levels of evi- dence may be used for different situations. How to make such judgments, however, remains an elusive methodological problem (2). As a referee pointed out, the summary findings indicate a conservative attitude toward infemng cause. Because causal conclusionsare primarily scientificconcerns, whereas public health recommendationsare primarily ethical concerns (I%), the intended goals of the review may have determined which results were included. Specific goals, however, were not stated in 5 of the 14alcohol and breast cancer reviews (63,70-72, 82); in three,the stated purpose did not match the results (76, 77, 81). Certainly, reviews may be written to summarize the studies and to suggest further research without including causal conclusions or public health recommendations. None of the reviews in these series, however, were so limited. Recommendations for Research. Has epidemiology found the best approach to causal inference? We suspect not. Improvements are important, consistent with the pivotal role of epidemiology in the prevention and control of cancer. Research on several fronts may prove fruitful: examining other series of reviews in light of the techniques of research synthesis (159163), examining the assumptions embedded within the causal criteria and their accompanying inferential rules (164), and proposing theoretical frameworks, both qualitative and quantitative, from which practical approaches to causal inference may be derived (150, 152, 155, 165, 166). Included within these theoretical efforts are further inquiries into the interrelationships of science and ethics in the practice of causal inference (4, Table 3 Frequency of the use of causal criteria in 14 published reviews on alcohol and breast cancer, 1985-1994 - ~~~ Criterion No.of reviews Consistencp Strength", Dose-response" Biological plausibilitya specificitp Temporalitp Experimentationa Coherence', Analogf' Epidemiological sense 14 13 9 8 5 4 3 2 1 1 a Hill, 1965 (4). Surgeon General, 1964 (3) 150, 165). Methods of moral reasoning, for example, may be helpful in making public health recommendations ( 167, 168). Recommendations for Practice. The following rec ommendations do not represent criticisms of the reviews we examined, but rather proposals for future consideration in the multifaceted practice of causal inference. For one component of that practice, namely, citation of studies, a recommendation might seem unnecessary, given that most reviewers cited nearly all available studies. Nevertheless, we recommend that when reviewers exclude studies from consideration, they state their reasons for doing so, using methodological criteria. We recommend that reviewers clearly state their purpose in writing a review. There are at least three possibilities: to summarize the science and make research recommendations, to make causal conclusions, and to make public health recommendations. We recommend that when making causal conclusions and/or public health recommendations, reviewers cite a source for their approach to causal inference. In many reviews summarized above, authors used varying subsets of the criteria of Hill (4) without justification for exclusions and changes. This seems a subjective practice (2). Perhaps the use of different subsets stemmed from differences in training, a factor we can examine indirectly in the epidemiological textbooks (138, 140142,144,145, 147, 148,169-175) published between 1970 and 1995. A remarkable variety of choices regarding the use of criteria are available. Overall, strength (15 of 15), temporaliQ (14 of 15), consistency (13 of 15). biological plausibility (13 of 15), dose-response (13 of 15), and specificity (11 of 15) were the most frequently described. Few lists were ranked. Finally, we recommend that reviewers examine recent reviews on the same topic and cite when appropriate the causal conclusions and public health recommendations from them. Reviews are increasingly viewed as a legitimate part of the scientific literature, serving to summarize research and to provide answers to the central questions of causal inference for specific associations (163). Many other facets of this practice remain to be carefully discussed. At its core, causal inference is a complex matter of: reasoned judgment informed by science and ethics. It is of central importance to epidemiologists involved in public health and to the public at large. Acknowledgments The authors gratefully acknowledge the comments and suggestions from Steven S. Coughlin, Michael S. Kramer, and Arthur Schatzkin. Cancer Epidemiology, Biomarkers & Prevention 309 References 1. Evans, A. S. Causation and disease: a chronological journey. New York: plenum Medical, 1993. 2. Weed, D. L. Causal and preventive inference. In: P. G. Greenwald, B. S. Kramer, and D. L. 'Need (eds.). Cancer Prevention and Control, pp. 285-302. New York: Marcel Dekker, Inc., 1995. 3. U. S. Department of Health, Education, and Welfare. Smoking and health. Public Health Service Publ. No. 1103. Washington D.C.: Government Printing Office, 1964. 4. Hill, A. B. The environment and disease: association or causation? proc. R. SOC.Med.. 58: 295-300, 1965. 5 . Houn, F., Bober, M. A.. Huerta, E. E., Hunting. S . D., Lemon, S., and Weed, D. L. The association between alcohol and breast cancer: popular press coverage of research. Am. J. Public Health, 85: 1082-1086, 1995. 6. World Health Organization, IARC. Monographs on the Evaluation of Carcinogenic Risks to Humans: Alcohol Drinking. Lyon: IARC Vol. 44. 1988. 7. Breslow, N. E., and Enstrom, J. E. Geographic correlations between cancer mortality rates and alcohol-tobacco consumption in the United States. 1. Natl. Cancer Inst., 53: 631-639, 1974. 8. Kono, S., and Ikeda, M. Correlation between cancer mortality and alcoholic beverages in Japan. Br. J. Cancer, 40:449-455, 1979. 9. LaVecchia,C., Franceschi,S., and Cuzik, J. Alcohol and breast cancer.Lancet, I: 621, 1982. 10. Schatzkin, A.. Piantadosi, S.,Micozzi, M.. and Banee, D. Alcohol consump tion and breast cancer: a cross-nationalcorrelation study. Int. J. Epidemiol.. 18: 28-31, 1989. 11. Smith, D. I. Relationship between alcohol consumption and breast cancer morbidity rates in Western Australia: 1971-1984. Drug Alcohol Depend., 24: 61-65, 1989. 12. Williams, R. R., and Horm, J. W. Association of cancer sites with tobacco and alcohol consumption and socioeconomic status of patients: interview study from the Third National Cancer Survey.J. NatI. Cancer Inst., 58: 525-547, 1977. 13. Rosenberg, L., Slone, D., S h a p ~ ,S., Kaufman, D. W., Helmrich, S. P., Mettinen, 0.S., Stolley,P. D., Levy, M., Rosenshein,N. B., and Schoaenfeld,D. A. Breast cancer and alcoholic-bevemgeconsumption. Lancet, 1: 267-271, 1982. 14. Byers, T., and Funch. D. P. Alcohol and breast cancer (Letter).Lancet I: 799-800, 1982. 15. Begg. C. B., Walker, A. M., Wesson, B., and Zelen, M. Alcohol consumption and breast cancer (Letter). Lancet 1: 293-294, 1983. 16. Paganini-Hill, A., and Ross, R. K. Breast cancer and alcohol consumption (Letter).Lancet, 2 626-627, 1983. 17. Webster, L. A., Layde, P. M., Wingo, P. A., and Ory, H. W. Alcohol consumption and risk of breast cancer. Lancet, 2: 724-726, 1983. 18. Talamini, R., LaVecchiq C., Decarli, A., Franceschi. S., Grattoni, E., Grigoletto,E., Liberati,A.. and Tognoni, G. Social factors, diet and breast cancer in a Northern Italian population. Br. J. Cancer, 49: 723-729, 1984. 19. Le,M. G., Hill,C., Kramer, A., and Flamant, R. Alcoholic beverage consumption and breast cancer in a French case-control study. Am. J. Epidemiol., 120: 350-357, 1984. 20. LaVecchia. C.. Decarli, A., Franceschi, S., PampaIlona S., and Tognoni, G. Akohol consumptionand the risk of breast cancer in women. J. Natl. CancerInst., 75: 61-65, 1985. 21. Katsouyanni, K., Trichopoulos, D., Boyle, P., Xirouchaki, E., Trichopoulou, A., Lisseos, B., Vasilaros, S., and McMahon, B. Diet and breast cancer: a case-control study in Greece. lnt. J. Cancer, 38: 815-820, 1986. 22. Le, M. G., Moulton, L. H., Hill, C., and Kramar, A. Consumption of dairy produce and alcohol in a casecontrol study of breast cancer. J. Natl. Cancer Inst., 77: 633-636, 1986. 23. h e y , E. B., Schairer, C., Brinton, L. A., Hoover, R. N., and Fraumeni, J. F., Jr. Alcoholconsumptionand breast cancer. J. Natl. Cancer Inst., 78: 657-661, 1987. 24. O'Connell, D. L., Hulka, B. S., Chambless, L. E., Wilkinson, W. E., and bubner, D. C. Cigarette smoking, alcohol consumpuon. and breast cancer risk. J. Natl. Cancer Inst., 78: 229-234, 1987. 25. Harris, R. E., and Wynder, E. L. Breast cancer and alcohol consumption: a study in weak associations. JAMA, 259: 2867-2871, 1988. 26. Rohan. T. E., and McMichael, A. J. Alcohol consumption and risk of breast Cancer. Int. J. Cancer, 41: 695-699, 1988. 27. Adami, H. O., Lund, E., Bergstrom, R., and Meirik, 0. Cigarette smoking, alcoholconsumption and risk of breast cancer in young women. Br. J. Cancer,58: 832-837, 1988. 28. Richardson, S., devincenzi, I., Pujol, H.. and Gerber, M. Alcohol consumption in a case-controlstudy of breast cancer in southernFrance. Int. J. Cancer,44: 84-89, 1989. 29. Meara. J., McPherson, K.. Roberta. M., Jones, L.. and Vessey. M. Alcohol, cigarette smoking and breast cancer Br I. Cancer, 60: 70-73. 1989. 30. Young. T. A case control study of breast cancer and alcohol consumption habits. Cancer (Phila.), 64;552-558, 1989. 31. Van't Veer, P., Kok, F. J., Hermus, R. J. J., and Sturmans, F. Alcohol dose, frequency and age at first exposure in relation to the risk of breast cancer. Int. J. Epidemiol.. 18: 511-517, 1989. 32. Chu. S.Y.. Lee, N. C.. Wingo, P. A., and Webster, L. A. Alcohol consumption and the risk of breast cancer. Am. J. Epidemiol.. 130: 867-877, 1989. 33. LaVecchia,C.. Negri, E., Parazzhi, F., Boyle, P., Fasoli, M., Gentile, A,, and Franceschi, S. Alcohol and breast cancer: update from an Italian case-control study. Eur. I . Cancer Clin. Oncol., 25: 1711-1717, 1989. 34. Kato, I., Tominaga, S., and Terao, C. Alcohol consumption and cancers of hormone-related organs in females. Jpn. J. Clin. Oncol., 19: 202-207, 1989. 35. Toniolo. P., Riboli, E., PmW F., C h a d , M., and Cappa,A. P. Breast cancer and alcohol consumption: a case-control study in norihem Italy. Cancer Res., 49: 52035206, 1989. 36. Rosenherg, L., Palmer, J. R., Miller, D. R., Clarke, E. A,, and Shapiro, S. A case-control study of alcoholic beverage consumption and breast cancer. Am. J. Epidemiol.. 131: 6-14, 1990. 37. Nasca. P. C., Baptiste, M. S., Field, N. A., Metzger, B. B., Black, M.. Kwon, C. S . , and Jacobson, H. An epidemiological case-control study of breast cancer and alcohol consumption. Int. J. Epidemiol., 19: 532-538, 1990. 38. Graham, S., Hellmann, R., Marshall, J., Freudenheim,J., Vena, J., Swanson, M., Zelgzny, M., Nemoto, T., Stubble, N., and Raimondo, T. Nutritional epidemiology of postmenopausal breast cancer in western New York. Am. J. Epidemiol., 134: 552-566, 1991. 39. Ewertz, M. Alcohol consumptionand breast cancer risk in Denmark. Cancer Causes & Control, 2: 247-252, 1991. 40.Franceschi, S., Serraino, D., Talamini, R., Barra, S., and Bidoli, E. Alcohol and breast cancer in an area with high alcohol consumption. Rev. Epidemiol. Sank hblique, 39: 143-148. 1991. 41. Sneyd, M. J., Paul, C., Spears, G. F.. and Skegg, D. C. Alcohol consumption and risk of breast cancer. Int. J. Cancer, 48: 812-815. 1991. 42. Zaridze, D., Lifanova, Y., Maximovitch, D., Day, N. E., and Duffy, S. W. Diet, alcohol consumption and reproductive factors in a case control study of breast cancer in Moscow. Int. J. Cancer, 48: 493-501, 1991. 43. Miller, A. B., Kelly, A,, Choi, N. W., Matthews, V.. Morgan, R. W., Munan, L., Burch, J. D., Feather, J., Howe, G. R.. and Jain, M. A study of diet and breast cancer. Am. J. Epidemiol., 107: 499-509, 1978. 44.CUismano, R., Dardanoni, G., Dardononi, L., Amendola, P., Greco, G., Spampinato,R., and Gafa, L. Risk factors of female cancers in Ragusa population (Sicily). 2. Breast cancer. Eur. J. Epidemiol., 5: 497-506, 1989. 45. Marubini, E., Decarli. A., Costa, A.. Mazzoltni, C., Andreoli, C., Barbieri, A., Capitelli, E., Carlucci, M., Cavallo, F., M o n f m n i , N., Pastorino, U.,and Salvini, S. The relationship of dietary intake and serum levels of retinol and beta-carotenewith breast cancer. Results of a case-controlstudy. Cancer (Phila.), 61: 173-180, 1988. 46. Dupont, W. D., Page, D. L., Rogers, L. W., and Pal, F. F. Influence of exogenous estrogens. proliferative breast disease, and other variables on breast cancer risk. Cancer (Phila.), 63: 948-957, 1989. 47. Iscovich, J. M., Iscovich, R. B., Howe, G., Shiboski, S., and Kaldor, J. M. A case control study of diet and breast cancer in Argentina. Int. J. Cancer, 44: 770-776, 1989. 48. Ferraroni, M., Decarli. A., Willett, W. C., and Marubini, E. Alcohol and breast cancer risk a case-controlstudy from northern Italy. Int. J. Epidemiol., 20: 859-864, 1991. 49. Pawlega, J. Breast cancer and smoking, vodka drinking and dietary habits. A case-control study. Acta Oncol.. 31: 387-392, 1992. 50. Martin-Moreno, J. M., Boyle. P., Gorgojo, L., Willett, W. C., Gonzalez, J., Villar, F., and Maissonneuve, P. Alcoholic beverage consumption and risk of breast cancer in Spain. Cancer Causes & Control, 4: 345-353, 1993. 51. Seidman, H., Stellman, S. D., and Mushinski, M. H. A different perspective on breast cancer risk factors: some implicationsof the nonattributable risk. CA-a Cancer Journal for Clinicians, 32: 301-313, 1982. 52. Hiatt, R. A,. and Bawol, R. D. Alcoholic beverage consumption and breast cancer incidence. Am. J. Epidemiol., 120: 676-683, 1984. 53. Willett, W. C., Stampfer,M. J., Colditz, G. A., Rosner, B. A,, Hennekens, C. H., and Speizer,F. E. Moderate alcohol consumptionand the risk of breast cancer. N,Engl. I.Med., 316; 1174-1180, 1987. 54. Schatzkin. A.. Jones, D. Y.. Hoover, R. N., Taylor, P. R., Brinton, L. A., Ziegler, R. G., Harvey, E. B., Carter, C. L., Licitra, L. M., Dufour, M. C., and Larson, D. B. Alcohol consumption and breast cancer in the epidemiologic I 310 Review: C a d Inference in Cancer Epidemiology .. 1r &i. follow-upstudy of the f m t National Health and Nutrition Examination Survey.N. Engl. J. Med., 316: 1169-1173. 1987. 55. Hiatt, R. A,, Klatsky, A. L., and Armstrong, M. A. Alcohol consumption and the risk of breast cancer in a prepaid health plan. Cancer Res., 48: 2284-2287, 1988. 56. Garfinkel, L., Boffetta, P., and Stellman, S. D. Alcohol and breast cancer: a cohort study. Pre.v. Med., 17: 686-693, 1988. 57. Schatzkin, A., Carter, C. L., Green, S . B., Kreger, B. E.. Splansky, G. L., Anderson, K. M., Helsel, W. E., and Kannel, W. B. Is alcohol consumption related to breast cancer? Results from the Framingham Heart Study. J. Natl. Cancer Inst., 81: 31-35, 1989. 58. Simon, M. S., Carman, W., Wolfe, R.. and Schottenfeld, D. Alcohol consumption and the risk of breast cancer: a report from the Tecumseh Community Health Study. J. Clin. Epidemiol., 44: 755-761, 1991. 59. Gapstur, S. M., Potter, J. D., Sellers, T. A., and Folsom, A. R. Increased risk of breast cancer with alcohol consumption in postmenopausal women. Am. J. Epidemiol., 136: 1221-1231, 1992. 60. Friedenreich, C. M., Howe, G. R., Miller, A. B., and Jain, M. G. A cohort study of alcohol consumption and risk of breast cancer. Am. J. Epidemiol., 137: 512-520, 1993. 61. Longnecker. M. P., Berlin, J. A,, Orza, M. J., and Chalmers, T. C. A meta-analysis of alcohol consumption in relation to risk of breast cancer. JAMA, 260: 652-656, 1988. 62. Howe, G., Rohan, T.. Decarli, A,, Iscovich, J., Kaldor, J., Katsouyanni, K., Marubini,E., Miller, A,, Riboli. E., Toniolo, P.. Trichopoulos,D. The association between alcohol and breast cancer risk evidence from the combined analysis of six dietary case-control studies. Int. J. Cancer, 47: 707-710, 1991. 63. Longnecker,M. P. Alcoholic beverage consumption in relation to risk of breast cancer meta-analysis and review. Cancer Causes & Control, 5: 73-82, 1994. 64.Wynder, E. L. American Health Foundation workshop on alcohol and breast cancer: introductory remarks. Rev. Med., 17: 667-669, 1988. 65. Harris, R. E., Spritz, N., and Wynder, E. L. Studies of breast cancer and alcohol consumption. Prev. Med., 17: 676-682. 1988. 66. Hiatt, R. A., Klasky, A., and Armstrong. M. A. Alcohol and breast cancer. Prev. Med., 17: 683-685, 1988. 67. American Health Foundation workshop on alcohol and breast cancer, December l , 1987 discussionof future studies and concludingremarks. Prev. Med., 17: 694-699, 1988. 68. Mantel, N. An analysis of two recent epidemiologic reports in The New England Journol ofMedicine associating breast cancer in women with moderate alcohol consumption. Prev. Med., 17: 672-675, 1988. 69. Feinstein, A. R. Scientific standards in epidemiologic studies of the menace of daily life. Science (Washington DC),242: 1257-1263, 1988. 70. Does alcohol cause breast cancer? (Editorial). Lancet, 1: 1311-1312, 1985. 71. Graham, S. Alcohol and breast cancer (Editorial). N. Engl. J. Med., 316: 1211-1213, 1987. 72. Alcohol consumption and breast cancer. Nutr. Rev., 46: 9-10, 1988. 73. WHO, IARC. Monographs on the Evaluation of Carcinogenic Risks to Humans: Alcohol Drinking, Vol. 44, pp. 222-232, 256. WHO, Lyon: IARC, 1988. 74. Stampfer, M. J., Colditz, G. A,, and Willett, W. C. Alcohol intake and risk of breast cancer. Cancer Compr. Ther., 14: 8-15, 1988. 75. Lowenfels, A. B.. and Zevola. S. A. Alcohol and breast cancer: an overview. Alcohol. Clin. Exp. Res., 13: 109-1 11, 1989. 76. Chu, S. Y. Alcohol and breast cancer: review and recommendations. In: V. T. DeVita, S . Hellman. and S. A. Rosenberg (eds.), Cancer Prevention, pp. 1-7. Philadelphia: J. B. Lippincott Co., 1989. 77. Willett, W. C.. Stampfer,M. J., and Colditz,G. A. Does alcohol consumption influence the risk of developing breast cancer? Two views. In: V. T. DeVita. S. Hellman, and S . A. Rosenberg (eds.). Important Advances in Oncology, pp. 267-281. Philadelphia:J. B. Lippincott Co.. 1989. 78. Wynder. E. L., and Harris, R. E. Does alcohol consumption influence the risk of developingbreast cancer? Two views: /it: V. T. DeVita. S. Hellman, and S. A. Rosenberg (eds.). Important Advances in Oncology, pp. 283-293. Philadelphia: J. B. Lippincott Co., 1989. 79. Hiatt. R. A. Alcohol consumption and breast cancer. Med. Oncol. Tumor Pharmacother., 7: 143-151, 1990. 80. Steinberg,J., and Goodwin, P. J. Alcohol and breast cancer risk-putting the current controversy into perspective. Breast Cancer Res. Treat., 19: 221-23 I. 1991. 81. Plant. M. L. Alcohol and breast cancer: a review. Int. J. Addict.. 27: 107-1 28. 1992. 82. Rosenberg, L.. Metzger, L. S, and Palmer. J. R. Alcohol consumption and risk of breast cancer: a review of the epidemiologic evidence. Epidemiol. Rev.. 15: 133-144. 1993. 83. Rohan, T. E., and Bain. C. J. Diet in the etiology of breast cancer. Epidemiol, Rev.. 9: 120-145, 1987. 84. Byers. T. Diet and cancer: any progress in the interim? Cancer (Phila.), 62: 1713-1724, 1988. 85. Kritchevsky,D. Nutrition and breast cancer. Cancer (Phila.), 66: 1321-1325, 1990. 86. Alcohol and cancer. Lancet 335: 634-635. 1990. 87. Stoll, B. A. Breast cancer prevention: approachesand prospects. Dev. &col,, 62: 229-235, 1991. 88. Byers, T. E., and Williamson, D. F. Diet, alcohol, body size and the prevention of breast cancer. Dev. Oncol., 62: 113-131, 1991. 89. Kelsey, J. L., and Gammon, M. D. The epidemiologyof breast cancer. CA-a Cancer Journal for Clinicians, 41: 146-165, 1991. 90.Blot, W. J. Alcohol and cancer. Cancer Res.. 52 (Suppl.): 2119s-2123s. 1992. 91. Klatsky, A. L., Armstrong, M. A., and Friedman, G. D. Alcohol consumption and 17-year cancer monality (Abstract). Am. J. Epidemiol., 126: 770, 1987. 92. Miller, D. R., Rosenberg, L., Clarke, A. E., and Shapiro, S . Breast cancer risk and alcoholic beverage drinking (Abstract). Am. J. Epidemiol., 126: 736, 1987. 93. Chu, S. Y.,Lee, N. C., Wingo, P. A., Olson, D. R., and Webster. L. A. Alcohol consumption and risk of breast cancer (Abstract). Am. J. Epidemiol., 128: 912, 1988. 94. Metzger, L. S., Reif, J. S., and Lopez, L. Diet, alcohol and breast cancer (Abstract). Am. J. Epidemiol., 132: 816, 1990. 95. Reynolds, P., Camacho, T., and Kaplan, G. A. Alcohol consumption and breast cancer: prospective evidence from the Alameda County Study (Abstract). Am. J. Epidemiol., 128: 930, 1988. 96. Longnecker, M. P., Newcomb, P. A., Mittendorf,R., Greenberg,E. R., Clapp, R. W., Bogdan. G., Willett, W. C., and MacMahon, B. Risk of breast cancer in relation to past and recent alcohol consumption (Abstract). Am. J. Epidemiol., 136: 1001, 1992. 97. Hill, A. B. Principles of Medical Statistics. New York: Oxford PES, 1971. 98. Goldacre, M., Vessey, M., Clarke, J., and Heasman, M. Record linkage study of morbidity following vasectomy. In: I. H. Lepow and R. Crozier (eds.), Vasectomy: Immunologic and PathophysiologicEffects in Animals and Man, pp. 567-575. New York Academic Press, 1979. 99. Walker, A. M., Jick, H., Hunter, J. R., Danford, A.. and Rothman, K. J. Hospitalization rates in vasectomized men. JAMA, 245: 2315-2317, 1981. 100. Petitti, D. B., Klein, R., Kipp, H., and Friedman. G. D. Vasectomy and the incidence of hospitalized illness. I. Urol., 129: 760-762, 1983. 101. Massey, F. J., Bernstein, G. S., O'Fallon, W. M., Schuman,L. M., Coulson, A. H., Crozier. R., Mandel, J. S., Benjamin,R. B., Berendes, H. W.. Chang, P. C., Detels, R.. Emslander, R. F., Korelitz, J., Kurland, L. T., Lepow, I H., McGregor, D. D., Nakamura, R. N., QuirDga, J., Schmidt, S.. Spivey, G. H., and Sullivan,T. Vasectomy and health: results from a large cohort study.JAMA, 252: 1023-1029, 1984. 102. Ross, R. K., Pdganini-Hill, A,, and Henderson, B. E. The etiology of prostate cancer: what does the epidemiologysuggest? Prostate,4: 333-344, 1983. 103. Honda, G. D., Bernstein, L., Ross, R. K., Greenland, S . , Gerkins, V., and Henderson, B. E. Vasectomy, cigarette smoking, and age at first sexual intercourse as risk factors for prostate cancer in middle-aged men. Br. J. Cancer, 57: 326-331, 1988. 104. Newell, G. R., Fueger, J. J., Spitz, M. R.. and Babaian. R. J. A case-control study of prostate cancer. Am. J. Epidemiol.. 130: 395-398, 1989. 105. Rosenberg, L., Palmer. J . R., Zauber, A. G.. Warshauer, M. E., Stolley. p. D., and Shapiro, S. Vasectomy and the risk of prostate cancer, Am. J. Epidemiol., 132: 1051-1055. 1990. 106. Mettlin. C.. Natarajan. N.. and Huben. R. Vasectomy and prostate cancer risk. Am. I. Epidemiol., 132: 1056-1061, 1990. 107. Hayes. R. B., Pottern, L. M.. Greenberg, R., Schoenberg, J., Swanson. G. M., Liff, J., Schwartz, A. G., Brown. L. M.. and Hoover. R. N. Vmectomy and prostate cancer in US blacks and whites. Am. I. Epidemiol., 137: 263-269, 1993. 108. Sidney. S . Vasectomy and the risk of prostatic cancer and benign prostatic hypertrophy. J. Urol., 138: 795-797, 1987. 109. Sidney, S., Quesenberry, C. P.. Sadler, M. C.. Guess, H. A,, Lydick. E. G., and Cattolica, E. V. Vasectomy and the risk of prostate cancer in a cohort of multiphasic health-checkupexaminees: second report. Cancer Causes & Control, 2: 113-116, 1991. 110. Nienhuis. H., Goldacre. M.. Seagroatt, V.. Gill, L.. and Vessey. M. Incidence of disease after vasectomy: a record linkage retrospectivecohort study. Br. Med. J.. 304: 743-746, 1992. 1 1 1. Giovannucci, E., Tosteson, T. D.. Speizer, F. E., Vessey, M. P., Colditz, G. A. A long-term study of mortality in men who have undergone vasectomy. N. En@. J. Med.. 326: 1392-1398. 1992. I I 1 I I 1 I j I I I I I Cancer Epidemiology, Biomarkers & Prevention 311 112. Giovannucci, E.. Ascheno. A,, Rimm. E. B.. Colditz, G. A,, Stanipfer. M. I., and Willett, W. C. A prospective cohort study of vasectomy and prostate cancer in U S men. JAMA. 269: 873-877, 1993. 113. Giovannucci. E.. Tosteson. T. D., Speizer.F. E.. Asherio. A,, Vessey, M. P.. and Colditz. G. A. A retrospectivecohort study of vasectomy and prostate cancer in US men. JAMA, 269: 878-882, 1993. 114. Hiatt. R. A.. Armstrong, M. A., Klatsky, A. L.. and Sidney, S. Alcohol consump:ion, smoking and other risk factorsand prostatecancer in a large health plan cohort in California (United States).Cancer Causes & Control. 5: 66-72. 1994. 115. Anonymous.Vasectomy and prostatecancer (Editorial).Lancet, 337: 14451446. 1991. 116. Howards, S. S., and Peterson, H. B. Vasectomy and prostate cancer: chance. bias, or a causal relationship? (Editorial) JAMA, 269: 913-914. 1993. 117. Hcaly. 9. Does vasectomy cause prostate cancer? JAMA, 269: 2620, 1993. 118. Guess, H. A. Is vasectomy a risk factor for prostate cancer? Eur. J. Cancer. 29A: 1055-1060. 1993. 119. Skegg, D. Vasectomy and prostate cancer: is there a link? N. Z. Med. I.. 106: 242-243, 1993. 120. Pollack, A. E. Vasectomy and prostate cancer. Adv. Contracept.. 9: 181186, 1993. 121. Petitti, D. B. Epidemiologic studies of vasectomy. In: G. I. Zatuchni, A. Goldsmith, and I. M. Spieler (eds.), Male Contraception: Advances and Future Prospects, pp. 24-33. Philadelphia: Harper & Row, 1986. 122. Guess, H. A. Invited commentary: vasectomy and prostate cancer. Am. I. Epidemiol.. 132: 1062-1065, 1990. 123. Davis, I. E. Male sterilization. Curr. Opin. Obstet. Gynecol., 4: 522-526, 1992. 124. Howards, S. S. American Urological Association response to two articleson the relationship of vasectomy and prostate cancer. Oncology, 5: 78-80, 1991. 125. DerSimonias. R., Clemens. J., Spirtas. R., and Perlman, I. Vasectomy and prostate cancer risk methodologicalreview of the evidence. J. Clin. Epidemiol., 46: 163-172, 1993. 126. Skegg, D. C. G. Vasectomy and risk of cancers of prostate and testis. Eur. I. Cancer, 29A: 935-936, 1993. 127. Boyle. P., and Zardize, D. G. Risk factors for prostate and testicular cancer. Eur. I. Cancer, 29A: 1048-1055, 1993. 128. Howards. S. S. Possible biological mechanisms for a relationship between vasectomy and prostatic cancer. Eur. J. Cancer, 29A: 1060-1062, 1993. 129. Pienta, K. I, and Esper, P. S. Risk factors for prostate cancer. AM. Intern. Med.. 118: 793-803, 1993. 130. Association for Voluntary Surgical Contraception. AVSC technical statement: vasectomy and prostate cancer, February 17, 1993. Wash., D.C.: Association for Voluntary Surgical Contraception, 1993. 131. Raspa, R. F. Complicationsof vasectomy. Am. Fam. Physician, 48: 12641268, 1993. 132. KIitsch, M. Vasectomy and prostate cancer: more questions than answers. Fam. Plann. Perspect., 25: 133-135, 1993. 133. Noticeboard. Vasectomy and cancer. Lancet, 338: 1586, 1991. 134. Kuhn, T. S. The Structure of Scientific Revolutions. Chicago: University of Chicago Press, 1970. 135. Goodman, N. W. Paradigm, parameter, paralysis of mind. Br. Med. J., 307: 1627-1629, 1993. 136. Masterman, M. The nature of a paradigm. In: I. Lakatos and A. Musgrave (eds.), Criticism and the Growth of Knowledge, pp. 59-89. Cambridge, United Kingdom: University Press, 1970. 137. Kuhn, T. S. Second thoughts on paradigms. In: F. Suppe (ed.), The Structure Of Scientific Theories, pp. 459-482. Urbana, IL: University of Illinois Press, 1974. 138. Susser, M. Causal Thinking in the Health Sciences. New York Oxford University Press, 1973. 139. Evans, A. S. Causation and disease: a chronological journey. Am. J. Epidemiol., 108: 249-258, 1978. IN. Kleinbaum, D. G., Kupper, L. L., and Morgenstem, H. Epidemiologic Research: Principlesand QuantitativeMethods. Belmont, CA: Lifetime Learning, 1982. 141. Ibrahim, M. A. Epidemiology and Health Policy. Rockville, MD: Aspen systems. 1985. 142. Rothman. K. I.Modern Epidemiology,p. 19.Boston: Little,Brown and Co., 1986. 143. Schlesselman,I. J. "Proof` of cause and effect in epidemiological studies: criteria for judgment. Prev. Med., 16: 195-210, 1987. 144. Fletcher, R. H.,Fletcher, S. W., and Wagner, E. H. Clinical Epidemiology: fie Essentials, Ed. 2. Baltimore: Williams & Wilkins, 1988. 145. Elwood. J. M. Causal Relationships in Medicine: A Practical System for Critical Appraisal. New York: Oxford University Press, 1988. 146. Glynn, J. R. A question of attribution. Lancet 342: 530-532. 1993 147. Beaglehole. R.. Bonita, R.. and Kjellstrom. T. Basic Epidemiology.Geneva: WHO, 1993. 148. Page, R. M., Cole, G. E., and Timmreck, T. C. Basic Epidemiological Methods and Biostatistics. Boston: Jones & Banlett, 1995. 149. Last, J. M. Dictionary of Epidemiology, Ed. 2, p. 8. New York: Oxford University Press, 1988. 150. Weed, D. L. Causal criteria and Popperian refutation. In: K. J. Rothman (ed.), Causal Inference, pp. 15-32, Chestnut Hill, MA: Epidemiology Resources lnc.. 1988. 151. Department of Clinical Epidemiology and Biostatistics,McMaster University Health Sciences Centre. How to read clinical journals. IV. To determine etiology or causation. Can. Med. Assoc. J., 124: 985-990, 1981. 152. Susser, M. What is a cause and how do we know one? A grammar for pragmatic epidemiology. Am. J. Epidemiol., 133: 635-648, 1991. 153. Rothman, K. I. (ed). Causal Inference. Chestnut Hill, MA: Epidemiology Resources Inc.. 1988. 154. Susser, M. The logic of Sir Karl Popper and the practice of epidemiology. Am. J. Epidemiol., 124: 711-718, 1986. 155. Renton, A. Epidemiology and causation: a realist view. J. Epidemiol. Community Health, 48: 79-85, 1994. 156. Weed, D. L. Between science and technology: the case of antihistamines and cancer. J. Natl. Cancer Inst., 86: 740-741, 1994. 157. Rothman. K. J., and Poole, C. Science and policy making. Am. I. Public Health, 75: 340-341, 1985. 158. Weed, D. L. Science,ethics guidelines.and advocacyin epidemiology.Ann. Epidemiol., 4: 166-171, 1994. 159. Mulrow, C. D. The medical review article: state of the science. Ann. Intern. Med., 106: 485-488, 1987. 160. Oxman, A. D., and Guyatt, G. H. Guidelines for reading literature reviews. Can. Med. Assoc. J., 138: 697-703, 1988. 161. Milne, R., and Chambers, L. Assessing the scientific quality of review articles. J. Epidemiol. Community Health, 47: 169-170, 1993. 162. Cooper. H., and Hedges, L. V. (eds). Handbook of Research Synthesis.New York: Russel Sage Foundation, 1994. 163. Muir Gray, J. A. Two classes of creativity-improving systematic reviews. J. Epidemiol. Community Health, 48: 4-5, 1994. 164. Weiss. N. S. Inferring causal relationships: elaboration of the criterion of "dose-response. " Am. J. Epidemiol., 113: 487-490, 1981. 165. Greenland, S. Probabilityversus Popper: an elaboration of the insufficiency of current Popperian approaches for epidemiologic analysis. In: K. J. Rothman (ed), Causal Inference, pp. 95-104. Chestnut Hill, MA: EpidemiologyResources Inc., 1988. 166. Gamer, M. S., and Lane, D. A. Causal propositionsin clinical research and practice. 3. Clin. Epidemiol., 45: 639-649, 1992. 167. Weed, D. L. Alcohol, breast cancer. and causal inference: where ethics meets epidemiology. Contemp. Drug probl., 21: 185-204, 1994. 168. Weed, D. L. Epistemology and ethics in epidemiology. In: S. S. Coughlin and T. L. Beauchamp (eds), Ethics and Epidemiology. New York Oxford University Press, in press, 1996. 169. MacMahon, B., and Pugh, T. F. Epidemiology: Principles and Methods. Boston: Little, Brown and Co., 1970. 170. Lilienfeld, A. M., and Lilienfeld, D. E. Foundations of Epidemiology, Ed. 2. New York Oxford University Press, 1980. 17I. Kelsey, J . L., Thompson, D. W., and Evans, A. S. Methods in Observational Epidemiology. New York: Oxford University Press, 1986. 172. Weiss. N. S. Clinical Epidemiology. New York Oxford University Press, 1986. 173. Hennekens, C. H., and Buring. J. E. Epidemiology in Medicine. Boston: Little, Brown and Co., 1987. 174. Friedman, G. D. Primer of Epidemiology,Ed. 4. New York McGraw-Hill, Inc., 1994. 175. Abramson, J. H. Making Sense of Data, Ed. 2. New York: Oxford University Press, 1994.