VoL 108, No. 3


Copyright © 1978 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved

Printed in


Society for Epidemiologic Research Iowa City, Iowa June 14-16, 1978 CONTRIBUTED PAPERS Infectious Disease Chairmen: W. Barker and B. Visscher Rheumatoid arthritis in Rochester, Minnesota (an incidence and prevalence study). A. Linos,* J. W. Worthington, W. M. O'Fallon and L. T. Kurland (Mayo Clinic, Rochester, MN 65901). All new cases of rheumatoid arthritis (RA) diagnosed in Rochester, MN, residents during 1960-1974 were used to obtain the first known population-based incidence rates in the U.S. Cases were accepted in the study only if they met the American Rheumatism Association (ARA) criteria (morning stiffness, pain, swelling, rheumatoid factor, radiologic, pathologic findings, etc.) for probable and definite RA. ARA criteria were met in 518 residents of Rochester (385 F, 133 M). The average annual age-adjusted (1960 US white) incidence per 100,000 was 32.1 for males and 67.9 for females. In both sexes rates increased with age, rising in males from 3.3 in the 0-9 year age group to 85.4 for those 2 70 years and in females from 15.8 to 193.5. The female:male ratio of incidence rates ranged between two and three to one over all age groups. For adults (>20 years) the incidence rates for probable and definite RA combined were 43.0 for males and 92.6 for females. Rates of definite only RA were 23.5 and 53.9, respectively, for males and females, comparable to the incidence rates among the Hiroshima atomic bomb survivors. As of January 1, 1975, the age-adjusted prevalence of RA was 4.7 per 1000 for males and 9.7 for females. The incidence of stroke and MI among patients on aspirin is compared to the expected based on the rates in the Rochester population. Impact of epidemic influenza among high risk and non-high risk segments of a denned population. W. Barker* and J. Mullooly (University of Rochester, Rochester, NY 14642). Impact of epidemic influenza has been measured among "high risk" and non-high risk segments of the enrolled population of the Kaiser Permanente Health Plan in Portland, Oregon. Utilization of outpatient and inpatient services by this population for flu-related morbidities during two epidemic flu periods (Dec. 1968 * Indicates person who delivered the paper.

- March 1969 and Dec. 1972 - March 1973) has been compared with utilization during a comparable nonepidemic period (Dec. 1970 - March 1971). During epidemic periods there were approximately 14 health service contacts per hundred persons for acute respiratory diseases of all types; during the non-epidemic period, the rate was 11 contacts per 100. Contact rates for upper respiratory disease were nimilur for high risk and non-high risk persons during all periods, while rates for lower respiratory disease requiring hospitalization were two to three-fold greater for high risk persons. These data provide a basis for directly estimating excess health service utilization attributable to epidemic influenza and for refining guidelines for use of influenza vaccine and chemoprophylaxis. Evaluation of an employee health service as a setting for a rubella screening and vaccination program. K. Weiss,* C. Falvo, E. Buimovici-Klein, J. Magill and L. Z. Cooper (The Roosevelt Hospital, New York, NY 10019). In response to a New York State mandate to provide female employees of child-bearing age with rubella screening and vaccination, an urban hospital established a program within its employee health service (EHS). Data collected over a period of two years have been evaluated. The susceptibility rate was 17.7% among 1226 women (age range 17 to 48 years). Susceptibility rates varied greatly (10.9% to 27%) among all ages of women screened. There was no correlation between individual susceptibility and exposure to epidemics during school age. Historical data were unreliable as an indicator of immune status (e.g., 22% of the susceptibles gave a positive history of disease). 87.8% of the susceptible women were contacted. Of these, 60.7% were vaccinated, 17.5% refused vaccine, 12.2% were ineligible because of pregnancy, medical contraindications, age, etc., and 9.6% were awaiting scheduling. Mild reactions (rash, arthralgia, myalgia and fever) were reported by 16.6% of the vacinees. There were no conceptions within the three month period after vaccination. Of those vaccinated, 3/4 returned for follow-up, all had sero-converted. Analysis of the data indicates the value of an EHS-based rubella program. Toward a national rubella immunization policy. C. Falvo* and K. Weiss (The Roosevelt Hospital, New York, NY 10019).


Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018


At the present time, the United States does not have a comprehensive plan to eliminate congenital rubella. There are only recommendations for use of live rubella virus vaccine. Three years' experience with an adolescent rubella screening and vaccination program in an urban hospital indicates many problems. The largest problem was vaccinating the susceptible girls. Out of 1391 teenaged girls screened, 19.1% were susceptible, of which only 28.8% were vaccinated. Susceptibility rates ranged from 11% to 27% depending on age. Compliance in screening was 74.4% for the entire adolescent clinic. Within the clinic's teams of practitioners, the compliance ranged from 60.8 to 92.5%. These data are illustrative of the difficulties of providing rubella immunization to adolescents as well as the variability of outpatient service compliance. The authors have utilized a review of the literature combined with personal experience in order to formulate the following suggestions for a national policy. First, continue the current policy of vaccinating all preschool children; second, either have a national surveillance program or a revaccination program for girls at age 10; and third, during the prenatal period, serologically screen all pregnant women with an organized plan to vaccinate the susceptibles at the time of delivery.


moved more and thus would be expected to have moved to their study residence at a later age. We obtained relevant information from a sample of USbom migrant MS patients (with a definite diagnosis and Los Angeles residence), and US-born migrant controls (matched for age, sex, race, and neighborhood). Using a matched pairs t-test, we determined that cases made significantly (p < .01) more moves prior to age 19 than matched controls. This study also provided an opportunity to examine the association between house pet ownership and the development of MS. Matched pairs did not differ significantly in ownership of house dogs or house cats, duration of pet ownership or age at first exposure. The difference in total moves is consistent with the concept that moving more often early in life increases an individual's probability of coming in contact with geographic foci associated with MS development or a critical level exists for successive exposures to different sets of environmental stimuli, above which MS pathogenesis can be initiated.

Genetic susceptibility to multiple sclerosis. B. Visscher,* R. Detels, J. Dudley, R. Malmgren, P. Terasaki and M. S. Park (UCLA, Los Angeles, CA 90024). Multiple sclerosis (MS) appears to be associated with a factor carried in the major histocompatibility complex on chromosome 6. In the authors' previous study of 81 patients and unrelated controls, 46% and 23% respectively had factor DRW-2. This finding indicates that there may be an MS susceptibility gene situated in this complex, very near to the HLA-DR locus. This hypothesis can be tested in families with more than one case of MS. If a DR locus factor is a marker for the presence of the MS susceptibility gene, each affected person in such a family should have the same DR factor. Preliminary data from a study of 53 families in which at least two living members have MS support the hypothesis that the gene for MS susceptibility is contained within the major histocompatibility locus. DRW-2 was the common factor in seven of nine families for which B-cell typing has been completed. In two of these nine famines, DRW-3 was the common factor. In four other families, for which only HLA-A and HLA-B have been determined, the affected members shared one or two haplotypes identical at these loci. Since these types were also shared by unaffected members of the famines, other factors must necessarily be involved in the causation of MS.

Familial aggregation of transmissible dementia (Creutzfeldt-Jakob Disease). R Neugut,* A. Neugut, E. Kahana, Z. Stein and M. Alter (School of Public Health, Columbia University, New York, NY, and Uri Leibowitz Epidemiology Unit, Hadassah Hospital, Jerusalem, Israel). Creutzfeldt-Jakob Disease (CJD), a rare and fatal degenerative neurologic disorder, shows a strong focus among Libyan-born Israelis. Age-adjusted annual incidence of CJD among them approaches 40 per million, as compared to the more usual rates of about 1 or 2 per million The disease is caused by an unconventional slow virus, yet about 10% of cases of CJD, as reported in the literature, cluster within families and suggest the possibility of a genetic component in the pattern of the disease occurrence. In view of the above and the fact that, prior to mass immigration to Israel in 1949-1951, Libyan Jewry had existed for centuries as a small marriage isolate, the authors undertook to investigate whether increased genetic susceptibility to the CJD agent might account for the excess of the disease among present-day Libyan-born Israelis. Genealogies were obtained for 20 of the 23 identified Libyan-born Israeli CJD victims and among them the following kinships were identified: a mother and daughter, two first cousins, two first cousins once removed, and the deceased mother of a seventh case who was herself retrospectively identified as a CJD victim. These findings, assuredly non-random, are assessed in terms of either purely environmental exposure or with emphasis given to a weakened genetic barrier among the historical marriage isolate.

Multiple sclerosis, migration, and house pets: A casecontrol study. D. Bunnell* and B. Visscher (UCLA, Los Angeles, CA 90024). In a previous study of migrants, Detels et aL observed that people immigrating to the study areas at a later age had a higher prevalence of multiple sclerosis (MS) than those who immigrated at an earner age for each latitude of previous residence and residence at the time of study. One possible explanation for this observation was that MS patients tended to have

Methodology Chairman: P.A. Lachenbruch A comparison of two techniques of gathering historical information concerning cancer cases. E. Zeighami,* A. Bahn, N. Herrmann and L. Houten (U. of Pennsylvania, Philadelphia, PA 19104). It is frequently desirable in epidemiologic studies to utilize information concerning possible risk factors for a disease from respondents other than the actual sam-

Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018



pie member. The validity of this approach is dependent upon the degree of correspondence of such substitute information with that which would have been obtained from the sample member. In order to test the correspondence of such information, a recent casecontrol study of primary colon cancer collected both current and historical information from primary respondent and next-of-kin regarding diet, smoking, occupation, residence, water sources, and medical history of the sample member. The next-of-kin information was then compared with information obtained from primary respondent, both for cases and for controls. Results indicate that information obtained from female secondary respondents has greater reliability than that obtained from males, particularly for dietary habits. Interindividual variation in reliability was greater than intraindividual variation among items. Information from the primary respondent on certain select items might be used to help establish the overall reliability of information from a given secondary respondent by examining the concordance of the results on those select items.

An analytic procedure for investigation of cancer latency in matched case control studies illustrated with occupational data. A. H. Smith, * H. Checkoway, D. F. Goldsmith, P. H. Wolf and H. A. Tyroler (UNC at Chapel Hill, NC 27514). A new method for etiologic investigation of the association between cancer incidence and both latency and secular tune trends of exposure is presented for matched case control studies, with results for prostate cancer, bladder cancer and leukemia from current studies of the U.S. rubber industry. The basis of the method is to use case and control exposure information for each year prior to diagnosis or death of the cases. Estimates are obtained of the odds ratio and the proportional excess exposure among cases compared with their matched controls, a parameter conceptually similar to attributable risk. These are plotted for each year prior to death and for each calendar year, and used for inference concerning hypothesized causal exposures. The latency results suggest that the number of years prior to death when causal exposures may have occurred was 18-33 years for prostate cancer, 10-33 for bladder cancer and 12-38 years for lymphatic leukemia, with peaks in the plots of attributable risk at 29, 20 and 27 years prior to death, respectively. Traditional definitions of cancer latency and reasons for abandoning them in the analysis are discussed. It is concluded that the analytic methods presented enhance the informativeneas of occupational case control studies, and provide an indication for individual case control matching. Mortality patterns in the Old Order Amish: Background and preliminary results. R. F. Hamman* (Dept of Epidemiology, Johns Hopkins School of Hygiene, Baltimore, MD 21205), J. I. Barancik and A. M. Lilienfeld. The Old Order Amish (OOA) are a cultural and religious isolate from VS. society that live in rural settlements mainly in Indiana, Ohio and Pennsylvania. They avoid the use of modern conveniences, alcohol

and cigarettes. Farm life with heavy exercise and a diet high in calories, carbohydrates and animal fats are usual The OOA are highly inbred, do not use contraceptives, have a crude birth rate 1.5 times the US rate, and have an average completed family of 6.3 children. Several of these factors differ from U.S. society, and suggest that altered mortality patterns may exist for the OOA. Death certificates from three states were matched to private Amish censuses to identify OOA deaths, which were verified by OOA contacts. The remainder of county deaths were considered non-Amish and used for comparison. Certificates on a sample of OOA and non-Amish deaths are being validated via physicians to assess comparability. Preliminary analyses of the Indiana OOA settlement (N - 6082; 153 deaths in 1967-1972) showed the following SMR's compared to U.S. white rates for 1969-1971: All causes: M - .69, F ~ .75; ischemic heart disease: M — .50, F «• .84; malignant, neoplasms: M — .78, F - 1.15. Analyses of other settlements and further causes are presented. Standardization of mortality for variation in socioeconomic status. D. Dixon-Davis,* D. Grahn and R. T. Lundy (Argonne National Laboratory, Argonne, IL 60439). Standardization of mortality for variation in regional socioeconomic characteristics as well as for age, sex, and race, is required to improve detection of changes in mortality around point sources of pollution. Mortality by level (ea, e.o, IMR, ASDR) was studied in 191 counties within an 80-km radius of 15 power generation sites that sample the diversity of U.S. socioeconomic, racial, demographic, and geographic features. 1970 NCHS mortality data were evaluated against selected 1970 U.S. Census data. Male mortality was more highly correlated to income, education, occupation, ethnicity, and other social variables than female mortality. Variables highly correlated with adult mortality did not always relate well to infant mortality. The 10 best socioeconomic variables, chosen by multivariate regression, accounted for 37% of the variance in mortality And the best 20 accounted for nearly 60%. Use of dummy variables to test for unparameterized site characteristics increased the explained variance by about 5%. Examination of the residual variances revealed that the highest correlations involved the more populous East Coast counties. This method shows promise for the use of readily available regional socioeconomic statistics for standardization of mortality differentials for varying social factors. Estimation of alcohol consumption from HANES questions. M. Russell,* J. Welte (Research Inst. on Alcoholism, Buffalo, NY 14203), M. Hattwick and W. Hadden (National Center for Health Statistics, Hyattsville, MD 20782). The Health and Nutrition Examination Survey (HANES) is a valuable source of information about the relationship of alcohol use to the health and nutritional status of the U.S. population. However, the four HANES questions on frequency of drinking and quantity of favorite beverage consumed do not permit direct estimation of absolute alcohol consumed per

Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018


average day (AA), a variable of interest in the above relationships. Data from a survey (N = 1041) which «used the 12-item quantity-frequency-variability (QFV) questions of Cahalan et aL were used to construct the HANES alcohol questions, from which a direct estimate of AAHANES was made, based on the necessary assumption that the respondent drank only his favorite beverage. AAHANES, though an underestimate, has a correlation of .93 with AAQFV. To correct the underestimation and to improve the correlation, these data were divided into six groups—high and low frequency for beer, wine, and liquor. Regressions were done in each group, collectively accounting for 89% of the variance in AAQFV. The regression parameters will be used to calculate an adjusted AAHANES, providing an improved estimate of AA for use in further analyses of HANES data on health and nutrition. A multivariate analysis of interstate variation in teenage fertility. E. A. Brann* (Bureau of Epidemiology, CDC, Atlanta, GA 30333). National age-specific fertility rates for females 15-19 years fell dramatically between 1970 and 1974. The percentage decline in fertility was greater among older than younger teenagers and among states which already had lower fertility rates at the beginning of the period. A multiple regression model permitted determination of the correlates of state-to-state variation in teenage fertility. Two dependent variables were examined: 1) the fertility rate of a state's teenagers in 1974, and 2) the percentage fall in that fertility rate between 1970 and 1974. The independent variables included in the model were statewide aggregate data on the abortion-to-live birth ratio, racial composition, ratio of teens receiving contraceptive services to those becoming pregnant (births plus reported abortions), adult personal income level, per cent urban population, and adult educational attainment The model explained 75% of the interstate variation in 1974 teenager fertility. The single most important determining factor was the state's abortion ratio for teenagers. Other significant factors were racial composition (p < .01) and income level (p < .05). With 1970-1974 change in fertility as the dependent variable, the model explained 80% of the variation, with income as the most important determining factor. The contraception-to-conception ratio was also significant (p < .01), as was the racial composition of the state's teenagers (p < .05). Factor analysis of the independent variables shows that the six variables are well accounted for by three factors representing socioeconomic status, ease of access to abortion services, and access to contraception. The analysis shows that the utilization of both contraception and abortion is important in determining the fertility of a state's teenagers, but-fails to answer which is the more important factor. Employing the large-scale-integrated (LSI) motor activity monitor to measure physical activity. R. E. LaPorte,* L. H. Kuller, D. J. Kupfer and R. McPartland (U. of Pittsburgh, Pittsburgh, PA 15261). The authors have developed an electronic unit to


measure physical activity of people during their normal daily routine. The large-scale-integrated (LSI) activity monitor measures transverse or rotational body movement in an internal counter by a mercury switch. The units are slightly larger than a wrist watch and worn at various body locations. The instrument has been evaluated in two ways. Experiment 1: 20 male graduate students wore a unit on their ankle and trunk for two days. The data revealed large individual activity differences and the subjects reported no interference with normal activity. Experiment 2: 10 physical education and 10 non-physical education majors wore two units for two days. The instruments discriminated activity between the two groups and the measurements were highly related to measures of energy expenditure derived from concurrent activity logging (r = + .69). The high relationship to energy analysis, the sensitivity of the units and the ease of implementation indicate that the units may be useful in relating normal activity to acute heart disease and to risk factors of obesity and lipoproteins. The authors are also planning to use the units as an objective guide to screen activity levels in therapeutic programs. Philosophical criteria in cancer epidemiology. J. R. Goldsmith* (California State Dept of Health, Berkeley, CA 94704). K. Popper, T. Kuhn, and I. Lakatos, trained, respectively, as a philosopher, an historian, and a mathematician, have elaborated principles for deciding on scientific stature and process. The present author has sought in their writings to identify criteria for evaluation of the status of epidemiology based on the following questions: 1) Does it add to growth of knowledge in general? 2) Does it provide means of dealing with otherwise intractable problems? 3) Does it possess an integrated set of assumptions, techniques and criteria for assessment of work, and a distinctive set of practitioners? The author concludes, that it ought to have very high status, and that if it were not for the modesty of epidemiologists it might Attempting to overcome this handicap, the author identifies three problems of cancer epidemiology and considers what theoretical and practical consequences might follow from application of teachings by Popper, Kuhn, and Lakatos. These problems are: In populations with a single unusual exposure history, the occurrence of excessive cancer incidence in a variety of different organs; the relationship of cancer etiology to etiology of other disease; and the infrequency of cancer among persons exposed to potent carcinogens. Heart Disease Chairman: M. Farber Aggregation of cholesterol among young families of men with myocardial infarction. R. S. Levine,* C. H. Hennekens, B. Rosner, B. E. Klein, J. Gouriey, M. J. Jesse and H. Gelband (University of Miami School of Medicine, Miami, FL 33152). Plasma cholesterol levels were obtained for 177 children, of which 91 were from 38 families of men who had experienced myocardial infarction (Ml) prior to age 50, and 86 were from 39 families of neighborhood

Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018



controls. Cholesterol levels were higher among case fathers, their wives and their children than among control families. Analysis of variance showed significant sibling aggregation of cholesterol among both groups of children (p < .001). In addition, the intraclass correlation coefficient for children of affected men (n « .71) was higher than that for healthy men (r, = .56). Lastly, for case families only, there were significant father-mother (rp = .42,p = .04) and fathersib (rp ~ .44, p - .03) pairwise correlations. Overall, these findings are consistent with the hypothesis that families of men with premature MI share similar environmental factors, which determine both higher levels and greater similarities for plasma cholesterol. Transient cerebral ischemia in a community of elderly people. T. M. Vogt* (UCLA, Los Angeles, CA 90024), R. I. Pfeffer, A. Heyman, W. E. Wilkinson and J. G. Burch. A mailed self-administered questionnaire was used to detect transient ischemic attacks (TLA) in a community of 17,000 retired white men and women in Southern California. Responses were obtained from 10,880 persons, or 64% of the community. Approximately 7% were considered to have probable TLA during the previous year on the basis of a history of transient motor, sensory, visual or speech disturbances, 14% had dizziness and lightheadedness only (possible TLA) and 79% denied having TLA symptoms. All 651 persons with probable TLA as well as samples of the other groups were interviewed by a neurologist to obtain cross validation of the questionnaire assessment. The one year period prevalence of TLA was estimated to be 1.2% in women and 2.2% in men. The .risk factors for TIA, i.e., hypertension, cervical bruits . -and stroke, were estimated to be present in 32%, 8% and 6%, respectively, of the respondent population. TLA was six times more likely among those with stroke than among persons without this illness and four times more likely in persons with cervical bruits. Although the sensitivity of the TIA questionnaire as a survey instrument appears to be only moderate (detecting only 57% of all probable TLAs), it may prove to be a feasible method for screening TIA in large population groups. Ventricular premature beats and coronary risk factors. B. Lown,* C. H. Hennekens, B. Rosner and S. Grufferman (Harvard Medical School, Peter Bent Brigham Hospital, and Harvard School of Public Health, Boston, MA 02115). Of 10,152 men, aged 35 to 54 years, initially screened for ventricular premature beats (VPB) and several coronary risk factors in a Multiple Risk Factor Intervention Trial (MRFTT) center, 9897 were free from prior myocardial infarction and diabetes. Among these 9897, there was a strikingly positive crude association of VPB with age (p < .001), a weakly positive association with systolic (p •=• .007) but not diastolic blood pressure, and no association with cholesterol. In addition, there was an inverse relation of VPB with Twave height (p < .001) and a decreased frequency of VPB in subjects with heart rates less than 60 per minute. Only two variables, age in years (p < .001) and

number of cigarettes smoked per day (p — .026) significantly discriminated between the presence and absence of VPB in multivariate analysis. These findings may reflect the presence of subclinical ischemic myocardial excitability in subjects at older ages and cigarette smokers. The data suggest that VPB may offer a mechanism for a cigarette smoking effect in coronary heart disease. To anticoagulate or not to anticoagulate in acute myocardial infarction. M. Szklo,* R. Goldberg, J. Tonascia and H. Kennedy (Johns Hopkins School of Hygiene, Baltimore, MD 21205). To examine the effect of anticoagulant therapy on short-term (in-hospital) and long-term prognosis among acute myocardial infarction (MI) patients, a systematic sample of 1401 MI patients discharged from 20 Baltimore hospitals was reviewed. In-hospital casefatality rate (CFR) was 16.6% (122/736) among patients receiving and 40% (266/665) among those not receiving anticoagulant therapy (p < .01). These differences did not disappear when rates were simultaneously adjusted for age, sex, race, type, order and location of MI, and clinical complications experienced during the acute hospital phase. To eliminate the possibility of a selective bias, deaths during the first 24 or 48 hours after admission were excluded, and these differences still persisted. Long-term prognosis was then assessed among 1013 hospital survivors by using a life-table approach. During the ten-year followup period, a significantly higher overall survivorship was demonstrated among patients administered anticoagulant therapy compared to that of patients not receiving anticoagulant therapy during hospitalization (p < .005). The results of this study agree with recent findings, and strengthen the notion that a new clinical trial is important to assess the efficacy of anticoagulant therapy in the secondary prevention of ML A randomized trial of aspirin and sulfinpyrarone in patients with transient ischemic attacks. D. W. Taylor,* D. L. Sackett and H. J. M. Bamett for the Canadian Collaborative Study (McMaster University,

Hamilton, Ontario L8S 4J9, Canada). A randomized controlled trial involving 585 patients in 24 centers across Canada was conducted to determine whether aspirin or sulfinpyrazone, singly or in combination, would reduce the rate of TIA, stroke or death among patients referred to neurologists with TIA judged to be of embolic origin. Patients were entered from January 1, 1972, to June 30, 1976, and followed for an average of 26 months. Complete followup to death or the closure date of June 30, 1977, was achieved for 99% of patients. Aspirin reduced recurrent TIA, stroke or death by 19% (p < .05) and the more important end-points of stroke or death by 31% (p < .05). Further analysis revealed a significant sex difference (p < .003) in which the benefit of aspirin was limited to males (who showed a risk reduction for stroke or death of 48%). This could not be explained by sex differences in entry characteristics or compliance with therapy. Significant interactions were also found between aspirin and both prior MI and diabetes. Males without diabetes or prior MI showed a risk

Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018

SOCIETY FOR EPIDEM1OLOGIC RESEARCH; ABSTRACTS reduction of 59% for stroke or death (p < .002). Sulfinpyrazone was not found to be an effective drug in this trial. Prospective study of psychosocial factors and coronary heart disease in Framingham. S. G. Haynes,* M. Feinleib and W. B. Kannel (National Heart, Lung, and Blood Institute, Bethesda, MD 20014). Between 1965 and 1967, a cohort of 1674 persons aged 45-74 and free of coronary heart disease (CHD) was administered an extensive questionnaire encompassing 20 psychosocial measures. Questions selected by a panel of experts and verified by factor analysis formed scales including Framingham Type A behavior and anger-out Persons scoring above (Type A) and below (Type B) the median on the Framingham Type A scale were followed for the development of CHD over an eight-year period. Among men under 65 years of age, Type A's were two times (p £ .05) as likely to develop angina and myocardial infarction than Type B's. Not showing anger and the frequency of job promotions were also associated with CHD incidence. Among women under 65, the incidence of CHD, in particular angina, was over two times (p S .05) higher among Type A's than Type B's. Not discussing anger was also significantly associated with CHD incidence. In multivariate analysis, all associations remained significant when controlled for the standard coronary risk factors. Among men, the findings were strongest among white-collar workers. The observed associations were similar among employed women and housewives. This study shows that psychosocial factors are independently associated with CHD, in both men and women. Initial intensive intervention in the Multiple Risk Factor Intervention Trial (MRFTT). The MRFTT Research Group.' Prepared by R. Sherwin (Univ. of Maryland School of Medicine, Baltimore, MD 21201). This is an interim report of the experience and early results of intervention on risk factors for coronary heart disease (CHD) among the 6428 middle-aged men randomized to the Special Intervention group of the MRFIT. All participants in the study were selected because of a high risk of CHD on the basis of a multiple logistic function of their serum cholesterol, number of cigarettes smoked daily and diastolic blood pressure (DBP). Intervention was directed toward the same three risk factors, at least two of which were relevant for almost all men. Reduced dietary fat and cholesterol with a moderate increase in polyunsaturated fat (but no drugs) were recommended to reduce cholesterol in all participants. Smoking cessation was supported by a variety of techniques excluding the use of hypnosis, aversion or drugs. Elevated DBP was managed by 1) weight reduction, if appropriate and successful, or 2) a defined program of "stepped care" using well-established drugs. Education about each of the risk factors and application of the principles of changing behavior took place in groups consisting of about ten men and their wives, usually meeting for one and a half hours each week for ten weeks. The staff included an individual trained in the behavioral sciences (usually the group leader), a nutritionist and


a physician. Changes in risk factors, from before randomization to one year after, are presented. Lipoprotein patterns and left main coronary artery disease: Prevalence and predictive value. T. A. Pearson,* B. H. Bulkley, S. C. Achuff, P. O. Kwiterovich and L. Gordis (Johns Hopkins Medical Institutions, Baltimore, MD 21205). Although few studies have correlated anatomicallydefined coronary artery disease (CAD) with epidemiologically-demonstrated risk factors, some reports have suggested that left main coronary disease (LMD) may be associated with hyperlipoproteinemia (HLP). To investigate this association, the authors determined lipoprotein patterns (Types Ila, lib, IV by NTH criteria) in 370 white men and women prior to coronary arteriography for chest pain, and blindly reviewed their angiograms to place them in one of three disease categories: LMD (43 pts), other CAD (228 pts), and normal (99 pts). HLP was present in 44% of pts with LMD, 34% of pts with CAD, and 19% of normals. This trend was stronger among women (5/8,62% with LMD having HLP) and men younger than 50 years (10/14, 71% with LMD having HLP). This trend was accounted for by Type IV but not by Type II HLP. Furthermore, among those with Type IV HLP, LMD was two to three times as common as in the entire population. Thus, Type IV HLP may be a risk factor for LMD and may have clinical value in the prediction of LMD in certain patient groups. Breast Cancer Chairman: D. Thomas Multivariate analysis of breast cancer risk factors among a screened population. L. A. Brinton,* L. Pickle, R. R. Williams, R. N. Hoover, J. F. Fraumeni, Jr. (Environmental Epidemiology Branch, National Cancer Institute, Bethesda, MD 20014). Mailed questionnaire data from participants in the Breast Cancer Detection Demonstration Project (BCDDP), including 405 cancer cases and 1156 normal screenees, were examined to determine whether the usual risk factors for breast cancer occur among individuals participating in screening programs. Various risk factors were assessed by means of stratified analyses, with subsequent testing for confounding and effect modification through a logistic approach. Nearly all of the recognized risk factors appeared to apply for women volunteering for screening. A family history of breast cancer in a mother or grandmother was associated with a five-fold excess risk. In addition, risk was elevated for women reporting early menarche, late menopause, nulliparity, late age at first birth and increased weight Risk was not increased by a prior breast biopsy, possibly because of the selective nature of biopsies in these women. The data also provided an opportunity to evaluate the influence of exogenous hormones on risk. No association with thyroid medications or menopausal estrogens was found, but an increased risk in menopausal women followed the use of oral contraceptives. A synergistic interaction was found between oral contraceptives and several risk factors.

Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018



National Cooperative Diethylstilbestrol-AdenosiB (DESAD) Project Baseline findings. B. Tilley* (U. of Texas School of Public Health, Box 20186, Houston, TX 77025), D. Labarthe, P. O'Brien, E. Adam, K. Noller, S. Robboy, D. Townsend, A. Barnes, D. Decker, R. Kaufman and L. Kurland. Impressions of the relative frequency of vaginal and cervical changes occurring after in utero diethylstilbestrol (DES) exposure vary greatly. Impressions are based on two self-selected groups and one group called through review of medical records. The DESAD Project has examined more than 3000 in utero DES-exposed women. The group selected by review of their mother's prenatal records had the lowest prevalence of vaginal epithelial changes (VEC) per 1000 participants (341). The walk-in and physician referral groups, similar to patients seen in a clinical setting, had much higher rates of VEC (589 and 652, respectively). Neoplasia, including mild dyspLasia, occurred rarely in all groups. The more severe cases were detected only in the strongly self-selected groups. Multdvariate analysis was restricted to the prenatal record review group. This analysis showed DES dose, duration, trimester of exposure and participant's sexual-reproductive history to be associated with VEC. Interpretation of these results is discussed. The replacement hormone use and breast cancer. N. W. Choi,* N. A. Nelson, C. Shalev and H. A. H. Abu Zeid (U. of Manitoba, Winnipeg R3E OW3, Canada). This is the final report of the results from a matched case-control study of effects of replacement hormone therapy on the breast cancer cases occurring during 1975-1977 among Winnipeg women up to 70 years of age. 228 cases and the controls, matched on age and neighborhood, were interviewed. The variables, replacement hormone therapy, hormones for contraception and for menstrual disorders, menopausal status, menstrual histories, and other variables (e.g., reproductive histories, socio-economic etc.) were studied. 11.8% of the young cases (S50) and 10.5% of their controls were under estrogen therapy for menopause (matched pair x8 not statistically significant). 5.3% of the cases and 2.6% of the controls were on estrogen therapy for more than two years. 37.5% of the older cases (>50) and 43.4% of their controls were under estrogen therapy for menopause. 19.1% of the cases and 19.7% of the controls used more than two years. Multivariate analysis involving the fitting of a logit model to the matched pair differences was done. An association between replacement hormone therapy and breast cancer was not found.

from Dr. Sidney Wolfe and the Ralph Nader-financed Health Research Group, stating that the risk of breast cancer in DES-exposed mothers was substantially increased. Because the issue has thus been raised publicly and a large number of women may potentially be affected, the need exists for a prompt re-evaluation of the issue. A careful examination of the original data casts substantial doubt on the suggested association, and this analysis as well as new data from the Rochester, MN, population are presented. The mothers of daughters enrolled in the national DESAD project at the Mayo Clinic Center (Rochester, MN) provide a unique opportunity to quickly identify a cohort of women with known DES exposure, adequate followup time, and ready accessibility of medical records. Population-based breast cancer incidence rates for the local population are also available and can be property adjusted to consider only parous women. Of the cohort of 442 DES-exposed mothers there were six confirmed cases of breast cancer with an expected number of 10.6 cases. The expected number specific for parous women was 9.2. Additional evidence from a previous casecontrol stu,dy done in this same population confirmed that the proportion of DES-exposed women in the control group and the breast cancer study group was similar. Thus, within the scope of this analysis, the investigators' results do not indicate an increased risk of breast cancer in DES-exposed mothers.

Difference in estrogen receptor levels between familial and nonfamilial breast cancer patients. R. Ottman (University of California, Berkeley), P. Hoffman (U. of California, San Francisco) and M. Lagios (Children's Hospital, San Francisco, CA). Levels of estrogen receptor protein (ER) in a group of breast cancer patients were compared to determine whether patients with a first degree family history of breast cancer differed from those without a family history. Each familial case was matched to two nonfamilial cases with respect to race (all white), menopausal status (pre/peri vs. post), origin of assayed specimen (breast/other), prior oophorectomy and exogenous hormone use. Comparison of familial and nonfamilial cases within each menopausal class revealed that the mean log ER in the post-menopausal familial group was significantly lower than in the postmenopausal non-familial group (p < .01). However, no difference was observed either between pre/peri and post-menopausal familial patients or between familial and non-familial premenopausal patients. With respect to this variable, familial breast cancer appears to be a homogeneous disease which is more nirp'lflr to premenopausal than to post-menopausal nonfamilial Evidence against the association of breast cancerrandL an bn breast cancer. •» DES exposure. D. Brian,* B. Tilley, D. Labarthe,e, W. 1 M. O'Fallon and K. Noller (Mayo Clinic, Rochester, MN 55901). Genetic epidemiology of families with high incidence It has been estimated that in the United States, of breast cancer. M.-C. King* (U. of California, Berkeexposure to diethylstilbestrol (DES) may have oc- ley, CA 94720), R. C. Elston (U. of North Carolina, curred in as many as two million live births. Although Chapel Hill, NC) and H. T. Lynch (Creighton U., the effects of DES exposure in utero on female off- Omaha, NE). spring is becoming well documented the effect of nonTwenty-two large families at high risk of breast transplacental exposure is not yet known. Recently, cancer were studied using segregation and linkage reports have appeared in the popular press, pmanating analysis in order to determine 1) if breast cancer in

Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018

SOCIETY FOR EPIDEMIOLOGIC RESEARCH: ABSTRACTS thefte families occurs in a pattern consistent with a disease largely determined by a putative "breast cancer susceptibility" gene, and 2), if so, can the susceptibility gene be physically (chromosomally) linked to a clinically innocuous genetic marker locus? For several of the largest families, an autosomal dominant model for transmission of breast-cancer susceptibility fit the family data significantly better (27- to 2400-fold difference in likelihoods) than an environmental model which assumed the increase in breast-cancer susceptibility to be age- but not genotype-dependent Four families with breast and ovarian cancer were analyzed together. A genetic model for transmission of susceptibility to this "syndrome" fit the data significantly better than the environmental model (15-fold difference in likelihoods). There was no significant difference in proportion of women nulliparous or in age at first pregnancy between breast cancer patients and age-matched sisters or first cousins without breast cancer in the families.


uniform age distribution: 38.0% for ages 0-4 years, 31.5% for ages 6-9 years, and 30.5% for ages 10-15 years. The authors' data estimate the median lethal area of burn (LAM) as 74.9% for total body surface area burned (BSA) and as 61.4% for third degree burn. While females experienced a less favorable mortality than males for BSA (LA«, - 71.9% vs. 74.6%) and third degree burn (LAw - 60.7% vs. 61.8%), the difference was not statistically significant. The 0 through four years age group had an appreciably lower LA» for both third degree and BSA. The LA«, reported here represents significant improvement over earlier data or recent data from non-specialized facilities.

A decision model for bum treatment facilities. G. Feck* (NYS Dept of Health, Albany, NY 12237) and E. Blair (N. Carolina State University). The findings of an epidemiologic survey of bum injuries were used in combination with the applied mathematics techniques of Operations Research to produce objective models by which to determine the Association of prior oophorectomy and estrogen con- number of intensive care bum treatment beds needed sumption with estrogen receptor content of breast in each region of New York State. The populationneoplasms. R. B. Wallace,* B. M. Sherman and J. based survey provided data describing the incidence, Kondo (U. of Iowa College of Medicine, Iowa City, IA severity, and average length of stay of 5791 hospitalized bum patients. The first model, a simple queueing 52242). Estrogen receptor (ER) determination in malignant system, demonstrated that for any population there is breast tissue has prognostic value in choice of therapy: no number of beds which will assure both high average ER positive patients have a 50% chance of responding occupancy and reasonable assurance of meeting peak to endocrine therapy, ER negative patients less than load. The queueing model assumes that Health Ser10%. In .an epidemiologic study of breast cancer risk vice Areas are closed regions with no flow of patients factors, a history of prior total bilateral oophorectomy in or out The second model, a multi-facility system, and menopausal estrogen use was elicited from 45 assumes open regions and a cooperative network in women with primary breast surgery for cancer, 19 of which patients arriving at a bum facility which is full whose neoplasms were estrogen receptor positive and' to capacity are immediately transferred to the next 26 ER negative. In the ER positive group there was a closest facility. The multi-facility model resolves the history of oophorectomy in 0/19 and estrogen use in dilemma of the simple queueing model and demon3/19. In the ER negative group, corresponding values strates that high average occupancy can be realized were 8/26 and 12/26 (p < 0.05). Most of the oophorec- for each Health Service Area along with a high systemtomies occurred prior to natural menopause and were wide assurance of meeting peak demand. The models associated with subsequent hormonal therapy. Sample may be applied to any health problem having the characteristics of rarity and urgency. size did not allow analysis of the separate effects of oophorectomy and estrogen therapy. Manipulation of the female hormonal milieu may be related to the The comparative safety of local versus general anessubsequent ER status of incident breast cancers and thesia for suction curettage abortion. D. A. Grimes, K. possibly alter the prognosis of these neoplasms. F. Schulz, W. Cates, Jr.,* and C. W. Tyler, Jr. (Family Planning Evaluation Division, CDC, Atlanta, GA 30333). Burns, Abortion, and Environment The relative safety of local versus general anestheChairman: G. Spivey sia for suction curettage abortion has not been clearly Mortality in a pediatric bums unit: Probit analysis. P. established. To compare the safety of these two anA. Buffler, L. Suaree, R. Osborne,* H. Carvajal and D. esthetic techniques, the authors studied 36,430 women Larson (U. of Texas Medical Branch, Galveston, TX who received local anesthesia and 17,725 who received 77550). general anesthesia for suction curettage in the United Mortality rates for severe burn injuries differ by States from 1971 through 1975. Data were obtained extent of body surface area (severity), age and insti- through the Joint Program for the Study of Abortion tution. In order to compare the mortality experience under the auspices of the Center for Disease Control, of various institutions, one method of standardization a multi-center, prospective, cohort study. The aggrecommonly used is probit analysis. Using this tech- gated major complication rates for the two groups nique, the authors examined the eight-year mortality were similar, but there were significant differences experience of Shriners Hospital for Crippled Children, between local and general anesthesia for specific comone of three specialized pediatric burn treatment cen- plications and treatments. Local anesthesia was assoters, for the interval 1969-1976. The study population ciated with higher rates of febrile and convulsive morof 812 children had a male excess (66% male) with a bidity; however, general anesthesia was associated

Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018



with higher rates of hemorrhage, cervical injury, and uterine perforation. Both anesthetic techniques appear to be safe, with similar degrees of overall safety, though each is associated with a different spectrum of complications. History of spontaneous abortion and its relation to tobacco and alcohol use in 513 pregnant women. R. Little* (U. of Washington, (GI-80) Seattle, WA 98195). A sample of 513 consecutive women of gravida two or greater who were paying patients at a health maintenance organization were interviewed in their fourth month of pregnancy about their smoking and alcohol use before conception. 80% of the sample were between 20 and 30 years of age, and 92% were white. 26% were smokers when they conceived and 9% reported drinking an average of at least 30 ml (1 oz) absolute alcohol (ethanol) daily in the six months before the current pregnancy. Smoking and drinking before conception were assumed from prior work to be a rough indicator of previous habits. A regression on the number of spontaneous abortions (prior to the current pregnancy) by smoking and alcohol use revealed that current smoking was significantly related to a history of abortion. Alcohol use failed to be significant, even when an equation for nonsmokers only was computed. These results must be interpreted with care, for current drinking and smoking is not always indicative of past behavior. However, the data suggest that smoking may be associated with history of spontaneous abortion, even in women who succeed in carrying a subsequent child into the fourth month of gestation. Preventive potential of low-tar cigarettes. S. D. Stellman,* H. D. Austin and E. L. Wynder (American Health Foundation, New York, NY 10019). During the past 20 years, three major events have occurred which will ultimately influence the rising lung cancer incidence rate: 1) average tar yields per cigarette have fallen, 2) specific condensate activity has declined, and' 3) low tar cigarettes have been smoked by increasing proportions of smokers. In a retrospective study of epidermoid lung cancer, risks based on dose-response curves were derived for specific tar levels and used to compute attributable risks. Distributions of tar levels among non-hospitalized smokers were obtained from various sources used to estimate the number of lung cancer deaths attributable to cigarette smoking at each dosage. A number of plausible schedules of tar reduction suggested by our experience with both hospital interview studies and smoking cessation clinics are used to predict the number of lung cancer deaths which could be prevented at each leveL Cancer mortality in selected Louisiana parishes and drinking water source. M. Gottlieb* (Tulane University, New Orleans, LA 70112). Cancer mortality rates for selected sites are high in Louisiana. Use of the Mississippi River water for drinking has been implicated. A case-control death certificate study of resident deaths and non-cancer deaths from 1960-1975 for selected Louisiana parishes has been undertaken to determine the relationship of

water source and characteristics to cancer mortality. Parishes using surface water were paired to parishes using ground water and grouped to obtain an equal frequency of population using ground as surface water and having similar industrialization characteristics. Information on industry of occupation and industry located near residence will be collected and type of water used and dates of use will be verified. The cancer sites of interest are bladder, kidney, prostate," brain, liver, pancreas, rectum, large intestine, esophagus, stomach, lymphoma, multiple myeloma, leukemia, Hodgkin's disease, lung, breast, and malignant melanoma. Cases from a parish group were matched with controls from the same parish group. Preliminary analysis of the first eight sites indicates a pattern of increased relative odds for living in a surface-water-using parish for cancer of bladder, kidney, and rectum, for at least three out of five parish groups. There are differences in risks among parish groups for cancer mortality by site. Racial differences in smoking behavior and prevalence of chronic bronchitis in two southeastern cities. A. Eahleman' (University of North Carolina at Chapel HUL NC 27514). Previous investigators have shown pronounced racial (black-white) differences in chronic bronchitis prevalence rates. In general these studies have shown that blacks have lower rates than whites and that women have lower rates than men. There is less agreement about smoking behavior. Some studies show greater rates of cigarette smoking among blacks while others do not, but most agree that blacks have proportionately less total exposure to tobacco smoke. The present study, using a data set created by the U.S. Environmental Protection Agency, found striking racial differences in respiratory symptom prevalence and smoking behavior. In summary, there were no significant sex differences in age-, education-, and smokingadjusted chronic bronchitis prevalence rates. There were large racial differences in prevalence, with overall adjusted rates for whites 1.6 to 3.1 times those for blacks. Smoking-specific, age- and education-adjusted rates for whites were twice those for blacks, suggesting that unknown host or environmental factors are influencing the development of the disease. Differential exposure to air pollution probably does not explain the differences. Possible biases from study design are discussed, as are the potentially unique conditions of life in the southeastern United States. Lung function in populations in Los Angeles exposed and unexposed to photochemical/oxidant pollutants. R Detels,' S. Rokaw, A. Coulson, D. Tashkin, J. Sayre and F. Massey (UCLA, Los Angeles, CA 90024). A survey for chronic respiratory disease and impairment was completed in 3465 residents (70% of enumerated) of an area exposed to photochemical/ oiidant pollutants and in 4509 residents (79% of enumerated) of a similar area exposed to low levels of pollutants. Tests included questionnaire, spirometry, plethysmography and the single breath nitrogen test. Cough and cough with sputum were more frequently reported in the low pollution area. Lung function was

Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018

SOCIETY FOR EPIDEMIOLOGIC RESEARCH. ABSTRACTS better in residents of the low pollution area according to FEVi, FVC, maximal expiratory flow rates, closing volume fraction, thoracic gas volume and airway resistance, although maTimal mid-expiratory flow rate, considered to be one of the most sensitive spirometric tests for detection of diseases in small airways, was similar in residents of both areas. The mean value for Ar4>7to.pw was slightly worse among residents of the low pollution area. These finHinga suggest that adverse effects of long-term exposure to photochemical/oxidant pollutants may occur primarily in the large airways. The greatest differences between the two areas were observed in residents aged 18-59 years, suggesting that long-term exposure may be required to cause measurable impairment. Cancer I Chairman: M. Elwood Lung cancer on a nickel smelting island. D. Reed,* R. Lessard and B. Maheux (California State Health Department, Berkeley, CA 94704, and the University of Michigan, Ann Arbor, Ml 48109). New cases of lung cancer for the years 1970-1974 were used to examine the relation of lung cancer and exposure to nickel smelting on the island of New Caledonia. The age-standardized incidence rates were higher for New Caledonia than most other countries in the South Pacific; higher in the city with the smelter than in rural areas, and, within the city, the rates were higher in the residential areas closest to the smelter. A case-comparison group analysis of lung cancer, smoking, and nickel occupation indicated that when age and cigarette smoking were controlled for, the relative risk of lung cancer was three times higher for nickel workers compared to non-nickel workers. Lung cancer and smoking were also significantly associated independently of the effects of age and nickel occupation, and this finding indicated that smoking habits must be taken into consideration when studying other possible causes of lung cancer. Leukemia mortality in an area with high atomic bomb fallout J. Lyon,* M. Klauber, and D. West (U. of Utah, Salt Lake City, UT 84132). Between 1950-1959, approximately 87 aboveground atomic tests were performed in southern Nevada. Approximately 23 of these tests formed clouds which dropped radioactive fallout on parts of the state of Utah. Because of the known leukemogenic effect of atomic fallout, the authors investigated the possibility of an excess of leukemia deaths amofig the children bom between 1950-1958. For comparison, they selected birth cohorts born between 1944-1950 and 1959-1975, and generated the expected number of deaths for all forms of leukemia for each of the three time periods using U.S. 1970 white mortality. For the birth cohorts bom between 1944-1960 and 1959-1975, no excess leukemia mortality was found. For the cohorts born between 1950-1958, an observed to expected ratio of 1.26 was found. Additional analyses, looking at other childhood malignancies, are underway. Unfortunately, radiation exposure data are not available in sufficient detail to quantify exposure, except in crude terms.


Thyroid cancer in Connecticut, 1935-1975; the changing distribution of histologic types over time. L. S. Mendelsohn* and J. W. Meigs (Environmental Epidemiology Branch, National Cancer Institute, Bethesda, MD 20014, and Connecticut Cancer Epidemiology Unit, New Haven, CT 06510). Using incidence data from the Connecticut Tumor Registry, 1935-1975, examination was made of time trends for the major histologic types of thyroid cancer—papillary, follicular, medullary, and anaplastic carcinomas. Over the 40-year period, there was over a three-fold increase in the rates for thyroid cancer, primarily involving the papillary and follicular types. The rise in papillary carcinoma was greatest among females aged 15-44 years, while the increase in follicular carcinoma affected females at all ages. Both cell types showed an upward trend in males at all ages, particularly in later life. A birth cohort analysis for these cell types revealed an increase in age-specific rates with each subsequent cohort This trend seems related, at least partly, to prior irradiation for benign conditions of the head and neck, a commonplace practice from 1920 to the late 1950*8. The rates for anaplastic carcinoma also increased with time, particularly among women aged 65+ years. However, there was a decline in the rates for medullary carcinoma, primarily in women aged 55+ years, perhaps related to the identification of genetic susceptibility through the use of calcitonin radioimmunoassays and removal of thyroid glands in the precursor stage. Radiation- and host-factors in thyroid tumors following thymus irradiation. R. Shore,* L. Hempelmann, B. Pasternack and E. Woodard (New York U. Medical Center, New York, NY 10016, and U. of Rochester School of Medicine, Rochester, NY 14642). This study provides the opportunity to use epidemiologic data to examine basic radiobiologic issues: shape of the dose-response curve, effects of dose-fractionation, the temporal pattern of tumor response, and interactions between host factors and radiation-induction of tumors. Among 2872 persons who were given x-ray treatments in infancy for enlarged thymuses and were followed for an average of 25 years, 24 thyroid cancers (vs. none among 5055 unirradiated sibling controls) and 52 benign thyroid tumors (vs. six among controls) have been found. The dose-response relationship for thyroid cancer showed both a linear and a quadratic component for thyroid cancer induction, but only linearity for thyroid adenomas. Dose-fractionation led to a diminution of cancer induction at lower doses but had no effect on thyroid adenomas. Females were found to have greater thyroid tumor induction than males. Life-table analyses showed that the excess thyroid cancer and thyroid adenoma production continued out to at least 40 years. There was no evidence of an inverse relationship between dose and tumor latency. , A case-control study of intracranial meningiomas in women. S. Preston-Martin* (USC School of Medicine, Los Angeles, CA 90033). A study was designed to test the hypothesis that exposure to ionizing radiation including diagnostic

Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018



medical and dental x-rays was causally related to the development of intracranial meningiomas in women in Los Angeles County, California. Information was also sought on a number of other possible risk factors including head trauma, smoking, consumption of certain foods, and family occurrence of cancers and nervous system tumors. A questionnaire was administered to all incident cages of this tumor occurring from 1972-1975 among Los Angeles County women under age 65 years. *n old friend and a nearest neighbor both matched on age, sex, and race were sought to serve as controls for each case. Questionnaire data were obtained on 189 cases. 1S5 neighbors and 155 friends. Data collected were supportive of the radiation hypothesis. Tumor occurrence was positively associated with a history of head trauma. No excess of cancers or brain tumors was observed among relatives of cases when compared with control relatives. A significant association was observed between meningioma occurrence and consumption of meats such as bacon, ham, sausage and lunchmeat which contain high levels of sodium nitrite. An increase in carcinoma of the uterine corpus in the San Francisco-Oakland SMSA. D. Austin* and K. Roe (Resource for Cancer Epidemiology, California State Department of Hea\th,. Berkeley, CA 94704). A significant rise in the incidence of carcinoma of the uterine corpus in the San Francisco-Oakland SMSA is reported by the Resource for Cancer Epidemiology (RCE). The incidence among white females increased by over 50% in the six years between 1969 and 1975. The increase was limited to the age groups over 50 years and to invasive cancers of the endometrial lining. The increase occurred in all five of the SMSA counties and, in white women in Alameda County, the average annual incidence for the 50-74 year age group tripled in 15 years. The rise is greatest in the areas of most affluence. Possible spurious causes of increase, such as changing diagnostic criteria, better casefinding or reporting are examined and ruled out The data suggest that the increase is due to the recent introduction of a potent endometrial carcinogen to the population group affected. Cancer n Chairman: T. Mason Cancer incidence amongst cosmetologists. J. Walrath* (Connecticut Cancer Epidemiology Unit, New Haven, CT 06510). The purpose of this study was to determine whether cancer incidence rates among cosmetologists differ from those among the general population. The population under study included all cosmetologists in Connecticut who began hairdressing school before December 31, 1965. School teachers in Connecticut were used as a comparison group and were matched individually to the cosmetologists by year of birth and sex. As a second method of attacking the problem cancer incidence rates for the Connecticut population, specific for age, sex, calendar year and cancer site were compared with the observed incidence rates of the cosmetologists. Cancer cases were ascertained through linkage of the files of the study subjects and matched comparison subjects with the files of the Connecticut

Tumor Registry. In each of the two types of analysis, the incidence of cancers of the digestive system, urinary system, brain and endocrine system and female reproductive organs amongst the cosmetologists did not differ from that expected. The cosmetologists appeared to experience an increased incidence of leukemia and a decreased incidence of breast cancer, but further work is needed to determine the significance of these preliminary observations. Cancer incidence and survival by marital status, Third National Cancer Survey. V. L. Ernster,* S. T. Sacks, S. Selvin and N. L. Petrakis (University of California, San Francisco, CA 94143, and UC School of Public Health, Berkeley, CA 94720). Incidence rates for cancers of major sites have been analyzed by sex, age, and marital status, for whites and blacks separately, using population-based incidence data collected in the nine areas of the Third National Cancer Survey (1969-1971) and demographic data for those areas from the 1970 U.S. census. In addition, three-year survival rates by marital status have been calculated, employing the life-table method, and are examined by sex, age, and race. Cancer sites studied include: stomach, colon, rectum, pancreas, bladder, kidney, breast, lung, larynx, esophagus, brain, melanoma, and the lymphosarcomas in both sexes; prostate in males; and ovary and uterine cervix and corpus in females. Patterns of cancer incidence by marital status are compared and consistencies across race, age, and sex groups noted. Among the many interesting findings are: high rates for female breast and ovarian cancers among single women and low rates in separated women; high rates for prostate cancer among widowed men and low rates in separated men; high rates for cancers of the lung and larynx among divorced men and low lung cancer rates among married men and single and married women. These and other results are compared with earlier reports of cancer mortality by marital status and discussed in relation to suspected etiologic factors for the cancers studied. Psychosocial risk factors for cancer of all sites in a nine-year prospective study. A. R. Moss* (University of California, Berkeley, CA 94720). Psychosocial risk factors that have been associated with increased cancer risk in cross-sectional or anecdotal literature are denial as a psychologic coping style and loss of a close personal relationship through bereavement or marital separation. However, little prospective evidence exists to support either hypothesis. Data from the Human Population Laboratory in Alameda County, California, were analyied to test these hypotheses prospectively. Denial, measured by consistent rejection of problems in different social domains, was associated with a two- to threefold relative risk of cancer of all sites in the nine years of followup, and with no excess risk of coronary heart disease. Marital separation was associated with a fourfold excess risk of cancer of all sites but no excess risk of heart disease. Bereavement was associated with different excess risks of both disease outcomes depending on the characteristics of the bereaved person. These associations persisted when controlled for smoking and for health status at the time of the

Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018

SOCIETY FOR EPIDEM1OLOGIC RESEARCH: ABSTRACTS original interview. They suggest that, contrary to the general stress hypothesis, specific psychosocial risk factors may predict heart disease and all site cancer mortality. Malignant melanoma in Utah. D. West,* D. SoodalterToman and J. Lyon (U. of Utah, Salt Lake City, UT 84132). Incidence rates of malignant melanoma in Utah have increased at least 2.5 times for males and females since 1967 (4.2 to 11.4 for males and 3.6 to 10.2 for females). Data were compiled from Utah Cancer Registry records on 827 cases. Cases were separated into four categories: all cases; skin only (90.4% of all cases); non-skin (of which 83.5% were ocular); and skin excluding in situ and lentigo maligna Preliminary findings demonstrate significant variations by religion, sex, location of melanoma on the body, latitude (north vs. south), urban-rural and socioeconomic status. Certain findings were unexpected and do not support findings of others. Rates are significantly lower in the Latter Day Saints (LDS) population for both males and females than in the non-LDS population. Greatest variation occurred between the women with melanoma of the leg; the non-LDS women had higher rates overall, especially in the rural areas. Religious lifestyle will be further studied to show just cause. Unlike most findings, incidence was 28% higher in the north (latitude 42°) than in the south (latitude 37°). Those in lower socioeconomic groups appear to have a rate 11% higher than those in higher groups (5.57 and 6.16, respectively). Additional analysis will be performed according to census tract breakdowns, endocrine involvement and specific location on the body (right vs. left).

Cancer and other mortality among Spanish-sumamed Californians. J. E. Enstrom* and E. A. Operskalski (UCLA, Los Angeles, CA 90024). The approximately three million Spaniah-surnamed Californians, including about 800,000 immigrants, comprise the largest ethnic subgroup in California, and they have been found to be significantly healthier than Californians as a whole. Based on 1970-1975 cancer mortality and 1970 mortality from all causes, Spanish-surnamed Californians, compared with other Californians, have a standardized mortality ratio (SMR) for cancer and total mortality of about 80% for males and 90% for females. Their SMR's for colorectaL lung, breast, and bladder cancer are noticeably low. Also quite interesting is their very low multiple sclerosis death rate, where the SMR is 10% among foreign-born persons and 54% among nativeborn persons. Mortality data for the major diseases as a function of age, sex, and nativity are presented and possible sources of error are discussed. Available information about the health habits, diet, socioeconomic status, and other characteristics of Spanish Americans are reviewed. Comparisons are made with other unusually healthy groups of Californians, such as Mormons and Japanese. Breast cancer in young women: The effect of prior pregnancy. D. T. Janerich,* E. M. Flink and D. Gle-


batis (New York State Department of Health, Albany, NY 12237). Although a first pregnancy at an early age protects against breast cancer, it is not known if the protection materializes immediately, or if some delay occurs. Information on this time relationship would be helpful in studying the mechanism of protection, and in interpreting the results of studies on oral contraceptives and breast cancer. The authors attempted to answer this question by studying pregnancy history and breast cancer in parous young women (50%) of the variance in of blood and 712 non-transfused controls during its BP was explained by Quetelet, HR, and the constelfirst two and a half years. Seventy-five (12.6%) recip- lation of humoral factors: Na, K, UK, PRA, and HCT. ients and 16 (2.2%) controls developed hepatitis. Intra- and inter-individual correlation of BP with Eighty (87.9%) hepatitis cases (65 among recipients these humoral factors were often in opposite direcand 15 among controls) were non-A non-B. The inci- tions, suggesting highly individualistic hemodynamics.

Downloaded from https://academic.oup.com/aje/article-abstract/108/3/224/138458 by Mount Royal College user on 10 August 2018


Maternal and Child Health Chairman: B. Selwyn Illegitimacy as a risk factor in early neonatal mortality. C. Keller* (Epidemiology Branch/NICHD, Bethesda, MD 20014). Marital status of the mother has been reported to be a significant risk factor in infant mortality, and the recent increase in reported illegitimate births suggests that further investigation should be undertaken to understand if and why this is detrimental to the survival of the neonate. The present study is based on over 400,000 white singleton live birth records linked to infant death records for 1967-1973 among residents of Minnesota. The crude early neonatal mortality rate was 10.54 deaths under one week per 1000 live births, and the rate for the 7% reported illegitimate births was 14.83 (45% higher than for legitimate births). Since unwed mothers are considerably younger than the average .woman at delivery, an adjustment for mother's age was made which reduced the difference to 18%. Further adjustment for parity, mother's education and prenatal care reduced this difference to only 15%, which was of borderline statistical significance. Conclusions are that the marital status of the mother at the time of delivery is not a significant independent risk factor for neonatal deaths during the first week of life, particularly for the majority of unmarried women who are young and delivering their first child. However, at least one subgroup, i.e., unwed mothers delivering a second or higher parity infant, are at substantially higher risk of an adverse pregnancy outcome. The association of obesity with infertility and related menstrual abnormalities in 36,081 women. A. J. Hartz,* A. Wong, K. P. Katayama and A. A. Rimm (The Medical College of Wisconsin, Milwaukee, WI 53233). Historical data from 26,638 20- to 40-year-old women were used to study the association between obesity and menstrual abnormalities including evidence of infertility. It was found that women with evidence of anovulatory cycles, Le., irregular cycles greater than 36 days and hirsutism, were more than 13.6 kg (30 lbs.) heavier than women with no menstrual abnormalities after adjusting for height and age. The percentage of women with evidence of anovulatory cycles was 2.6% for women 74% overweight. Women with a single menstrual abnormality including cycles greater than 36 days, irregular cycles, virile hair growth with facial hair, or heavy flow were also significantly heavier than women with normal values for these factors. These findings showing evidence of abnormal ovulation, menstrual abnormalities, and excess hair growth in obese women may be explained by the recent studies of others demonstrating an association between obesity and hormonal imbalances. Maternal age at birth and intelligence in the offspring. L. Belmont,* P. Cohen, J. Dryfoos, Z. Stein, S. Zayac and P. Zybert (Psychiatric Institute, Columbia U.


School of Public Health and Alan Guttmacher Institute, 600 West 168th St., New York, NY 10032). -Three large U.S. data sets have been analyzed to explore the role of maternal age at childbearing on intelligence in the offspring. Young mothers are at a clear disadvantage in terms of those socioeconomic variables (occupation, education and income) which relate to children's I.Q. However, after control for a large number of social and demographic variables, there remains a small, far less important, but consistent relation between maternal age and intelligence in the offspring. The child of an older mother does slightly better on intelligence tests (WISC). This relation is a linear one, and there is no exceptional disadvantage for the child of a teenage mother. Induced abortion and congenital malformations in offspring of subsequent pregnancies. M. Bracken* and T. Holford (Yale University, EPH, New Haven, CT 06510). A case-control {N = 1427 and 3001, respectively) study in Connecticut found delivery of an infant with congenital malformations unrelated to overall previous experience of induced abortion (o — 0.9) or to: 1) abortion of penultimate pregnancy, 2) abortion of first pregnancy when index pregnancy is second, and 3) multiple previous abortions. Non-white women aged 25-29 years who delivered a malformed child were significantly more likely to have aborted (o = 2.6, p < 0.05). This may be due to more frequent histories of illegal and septic abortion, or to other characteristics of these women. Legal abortion performed under safe clinical conditions appears to impose no increased risk for subsequent delivery of a malformed infant Maternal characteristics and birthweight: An etiologic inquiry into higher birthweights among foreign-bom black women in Harlem. B. Mayleas* and D. Rush (CUSPH, Division of Epidemiology, New York, NY 10032). Reports of higher birthweight among foreign-born black and white women in New York City (NYC) have not established a reason for such advantage over native-bom women of the same race. The authors report on a study of birthweight and gestational time among 766 black women, new registrants for prenatal care at a public clinic in Harlem from 1971-1973, and part of a randomized trial of nutritional supplementation during pregnancy. Women weighing

Abstracts of papers presented at the eleventh annual meeting of the Society for Epidemiologic Research.

VoL 108, No. 3 AMERICAN JOURNAL OP EPIDEMIOLOGY Copyright © 1978 by The Johns Hopkins University School of Hygiene and Public Health All rights rese...
2MB Sizes 0 Downloads 0 Views