Role of fat, animal protein, and dietary fiber in breast cancer etiology: a case-control study.

A case-control study of 818 breast cancer (BC) patients and 2 matched control groups, surgical controls (SCs) and neighborhood controls (NCs), was undertaken in Israel between 1975 and 1978. The interview schedule included a detailed dietary history based on the frequency of consumption of 250 food items, which were grouped according to their principal nutrient component. The average frequency of consumption of each food item in each nutrient group was computed. Medical, demographic, hormonal, and parity histories were also obtained. Risks associated with fat, animal protein, and fiber consumption were evaluated. Two types of analysis were performed [in 2 age groups (less than 50 yr and greater than or equal to 50 yr)], using the conditional logistic method: evaluating the risk attributable to nutrition only and controlling for nondietary confounding factors as well. When no adjustment for nondietary confounding factors was made, the risk increased with fat intake in both age groups [one-tailed P-value for linear trend = .08 and .07 in age less than 50 and .01 and .10 for the greater than or equal to 50 age category for the BC case (BCC)-SC and BCC-NC comparisons, respectively]. Increased fiber intake decreased the risk in the younger age group (one-tailed P-value for linear trend = .06 and .07 for the BCC-SC and BCC-NC comparisons, respectively), while in the 50-or-over age category the trend was inconsistent. The risk associated with animal protein was much less clear. For women in the highest quartiles of fat and animal protein intake and the lowest quartiles of fiber intake, risk was about twice as high as that for women in the lowest quartiles of fat and animal protein intake and in the highest quartile of fiber intake (one-tailed P-value for linear trend = .04 and .08 for age less than 50 and .08 and .09 for the age category greater than or equal to 50 BCC-SC and BCC-NC comparisons, respectively). When hormonal and demographic confounding factors were controlled for, this pattern persisted but it remained significant for 1 control only. Power increased when cases were analyzed against both controls simultaneously (one-tailed P-value for linear trend = .10 for age less than 50 and .02 for age greater than or equal to 50). Thus a higher fat-animal protein and lower fiber diet is associated with increased cancer risk, but this relationship needs to be studied further.

Coffee and methylxanthines and breast cancer: a case-control study.

A dietary case-control study based on 818 newly diagnosed breast cancer (BC) patients was conducted in Israel between 1975 and 1978. The role of coffee and total methylxanthine intake from coffee, tea, cola, chocolate, and cocoa drinks was evaluated in the BC patients as compared to that in two matched control populations [surgical controls (SC) and neighborhood controls (NC)]. Because it has been suggested that caffeine enhances mammary carcinogenesis in rats fed high polyunsaturated fat diets, analysis was done also in relation to fat consumption. When comparison was done to both matched control groups, a nonsignificant negative association was found between consumption of cups of coffee and BC (odds ratios of greater than or equal to 4 cups of coffee/day vs. less than or equal to 1 per week = 0.6 for BC/NC and 0.7 for BC/SC). This association was observed in all 3 ethnic subgroups studied. The pattern was stronger among the high-fat consumers after controlling for several hormonal confounding factors (two-tailed P-value for linear trend = 0.06 for SC and P = 0.05 for NC). In addition, when the consumption of methylxanthine of BC patients was compared to that of benign breast patients, adjusted by age and ethnic group, a diminished risk was found (odds ratio for BC of the highest level of methylxanthine vs. lowest level = 0.59).

Seasonality in the occurrence of breast cancer.

A nationwide study of all 3183 female patients with breast cancer in Israel diagnosed over a 7-year period (1960 to 1966) was conducted. Monthly series analysis showed a clear seasonal pattern in the symptomatology of the disease most pronounced in patients younger than 55 years, in all ethnic groups, mainly confined to cases with a nonlocalized tumor at diagnosis. Peaks occurred during spring, and troughs appeared during autumn. It is suggested that the pattern is of endogenous, probably hormonal nature, although environmental factors cannot be excluded.

Nutritional factors associated with benign breast disease etiology: a case-control study.

Between 1977 and 1980, 854 biopsied cases of benign breast disease (BBD) and 755 matched surgical and 723 matched neighborhood controls subjects were interviewed in Israel by using a detailed food frequency questionnaire. Cases were classified according to degree of ductal atypia (Black-Chablon grading system). Women with atypic lesions (grades greater than or equal to 3) reported a higher intake of all types of foods compared with both control series. Further analyses indicated that the increased consumption was due primarily to foods containing greater than or equal to 10% fat. Odds ratios associated with the highest fat consumption quartile were close to 3.0. There was a trend for increasing saturated fatty acid consumption with increasing ductal atypica. After adjusting for hormonal and demographic confounders, the association with fat intake was strengthened. Because atypic BBD has been reported to be a precursor of breast cancer, our findings lend support to the hypothesis that dietary fat is a risk factor for breast cancer.

Auditory nerve-brainstem evoked responses in hyperbilirubinemic neonates.

On the basis of the known predilection of the auditory brainstem pathway for bilirubin toxicity, we have examined auditory brainstem responses of neonates during the period of hyperbilirubinemia. The auditory brainstem responses of 24 infants with serum bilirubin values between 15 to 25 mg/dL were compared with the responses of 19 infants without hyperbilirubinemia, who had similar gestational and postnatal ages. Wave IV-V complex was absent in at least one recording of 10/24 jaundiced infants, whereas wave complex IV-V was consistently present in all of the 19 infants without hyperbilirubinemia (P less than .001). Jaundiced infants also had prolonged brainstem transmission time (P less than .01) which reflected increased latency at both lower and upper brainstem levels. The above changes were rapidly reversed in the majority of instances. Neonatal jaundice was associated with significant transient aberrations of auditory brainstem responses, suggestive of a transient brainstem encephalopathy. This evidence of bilirubin entry to the brain at conventionally acceptable serum concentrations raises questions about current concepts of the mechanism of transfer of bilirubin across the blood-brain barrier.

Differential effect of risk factors on early and late wound infections in patients undergoing herniorrhaphies.

In a prospective follow-up of 2846 patients who underwent hernia repair in 22 general surgery departments in Israel, factors affecting early or late infections were explored. Risk factors included inherent patient characteristics such as old age, ethnic group, and type of hernia prior to the surgery. Patient management factors included duration of the operation, use of urinary catheters, and use of drains. Of the 12 variables studied, only three had a constant effect during the entire 30-day follow-up. The other factors affected the occurrence of either early or late infections, but not both. For example, patients undergoing long operations, or from ethnic minorities, had a high rate of early infection, while those with special wound treatment (such as evacuation of hematomas) had high rates of late infection. It is postulated that factors present at the time of the operative incision tend to "cause" early infections, while factors that accumulate over time, or develop after leaving the operating theater, tend to affect late infections.

Discharge from hospital and its effect on surgical wound infections. The Israeli Study of Surgical Infections (ISSI).

A multicenter study of 2846 consecutive hernia operations was conducted to identify risk factors for infections that occurred during the hospitalization and post-discharge from hospital. Operated patients were followed-up for 30 days after surgery, whether at home or still hospitalized, and half the wound infection episodes were found to occur at home. Risk factors for both in-hospital and post-discharge infections seemed to be influenced by; (a) the selective nature of discharge, (b) the differential effect some risk factors had on either early or late infections. On any given day, patients selected by the clinical team to remain in hospital were more "at risk" for infection than those who left. As a result, they had a better chance of being diagnosed as infected during hospitalization. By contrast, those who were discharged home were perceived as low risk for complications. Subsequent infections in these patients occurred either due to factors "causing" late infections, therefore, unappreciated at the time of discharge, or unknown risk factors. More study risk factors were associated with in-hospital than with post-discharge infections, especially those associated with "early" infections. The implications of these findings for future evaluations of medical care in hospitalized patients are discussed.

A criterion for a standardized definition of low birthweight.

A criterion for the definition of low birthweight (LBW) which is tailored to each population group is presented. The suggested criterion is based on an assumption according to which the distribution of birthweight is actually a mixture of the following two components: the predominant (including about 95% of the newborns and is normally distributed) and the residual (assumed to be mainly composed of newborns under extreme risk for perinatal mortality). The LBW category is defined to include a small but equal (for all groups) percentage of the predominant component. Thus, population groups which differ with respect to their specific predominant birthweight distribution are comparable with respect to the perinatal mortality risk. Application of the suggested definition is demonstrated using data from two ethnic groups in Israel. Results show a smaller percentage of female LBW than males in both population groups when using the suggested criterion. The reverse is observed (i.e. more male than female LBW), using the standard criterion (less than 2500 gm). The suggested criterion is shown to reduce the sex and ethnic differences in perinatal mortality among the respective LBW categories.

The Israeli Study of Surgical Infections (ISSI): II. Initial comparisons among hospitals with special focus on hernia operations.

In a study of 5,571 patients from the general surgery departments of 11 Israeli hospitals, the crude overall wound infection rates showed interhospital heterogeneity. The rates ranged from 6.3% to 12.4% (P(chi 2) = 0.039). Controlling for the different distributions of procedures performed in the various institutions did not reduce this variability. None of the hospitals had either consistently high or consistently low infection rates. A hospital could have low rates for one procedure and high rates for another. Therefore, the decision was made to proceed with procedure-specific analyses. This article details results of the analysis of 1,487 hernia operations. Four variables (old age, infection at another site, incarceration, and introduction of drains) accounted for almost all the differences in infection rates among the institutions. Of the four, presence of drains had the strongest association with infection (P derived from the logistic model less than 0.001). The risk was consistent in all hospitals and was unconfounded by other measurable factors. In contrast, the pattern of using drains seemed arbitrary and inconsistent, ranging from 9% of patients in one hospital to 41% in another. These findings were used as a basis for discussion with the surgical teams and for the initiation of a randomized clinical trial on the use of drains in hernia operations.

The Israeli Study of Surgical Infections (ISSI): I. Methods for developing a standardized surveillance system for a multicenter study of surgical infections.

The postoperative wound infection experience in 11 Israeli hospitals was investigated with two objectives: (1) to utilize the variability among hospitals for a better understanding of the determinants of these infections, and (2) to present surgeons and infection control teams with information upon which they can take action. This article summarizes the methods that were applied to maximize the uniformity of information obtained from hospitals. (1) The same kind of patients were sampled in every hospital. (2) The data collection system was standardized, and analyzed for errors. The overall error rate was small (2%) but there were potential problem areas, such as missing information on underlying diagnoses (12%). Few errors (1%) were found in the nurses' observations of the wounds. (3) All wound infections were diagnosed in a uniform manner by convening a panel of four specialists at the central Israeli Study of Surgical Infections (ISSI) office. (4) The definition of infection that was finally chosen for this multicenter study had to be broadened to include, in addition to pus, "discharge other than pus," because of the higher reliability among panel members. This definition was also clinically valid in terms of morbidity. (5) The rate of postdischarge infections was estimated and was not found to be related to the discharge policies in the various hospitals.

Randomized clinical trial of intra-operative antimicrobial prophylaxis of infection after neurosurgical procedures.

A randomized, placebo-controlled, double-blind and sequentially analysed clinical trial to determine the efficacy of intra-operative parenteral gentamicin and vancomycin (with streptomycin in the irrigating solution) in preventing infection at the operative site following neurosurgical procedures is described. Patients receiving prophylaxis had a significantly (P = 0.046) lower operative site infection rate (2/71 = 2.8%) than those receiving placebo (9/77 = 11.7%). This difference was most apparent during an epidemic, the source of which was not evident. Moreover, a total of 13 infections (two operative site, five pneumonia and six urinary tract) occurred among 12 patients receiving prophylaxis, whereas there was a total of 31 infections (nine operative site, nine pneumonia, 10 urinary tract and three septicaemia) among 24 patients receiving placebo. A smaller quantity of antimicrobial drugs was administered postoperatively to patients receiving prophylaxis (3.96 'antibiotic-days' per patient) than to those receiving placebo (6.87 'antibiotic-days' per patient).

Collinearity diagnosis for a relative risk regression analysis: an application to assessment of diet-cancer relationship in epidemiological studies.

In epidemiologic studies, two forms of collinear relationships between the intake of major nutrients, high correlations, and the relative homogeneity of the diet, can yield unstable and not easily interpreted regression estimates for the effect of diet on disease risk. This paper presents tools for assessing the magnitude and source of the corresponding collinear relationships among the estimated coefficients for relative risk regression models. I show how to extend three tools (condition indices, variance decomposition proportions, and standard inflation factors) for diagnosing collinearity in standard regression models to likelihood and partial likelihood estimation for logistic and proportional hazards models. This extension is based on the analogue role of the information matrix in such analyses and the cross-product matrix in the standard linear model. I apply the methodology to relative risk models that relate crude intakes (on the log scale) and nutrient densities to breast cancer cases in the NHANES-I follow-up study. The three diagnostic tools provide complementary evidence of the existence of a strong collinearity in all models that is due largely to homogeneity of the population with respect to our risk scale for the crude intakes. The analysis suggests that the non-significant relative risks for the crude intakes in these models may be due to their involvement in collinear relationships, while the nonsignificant relative risks for the nutrient densities are far less affected by multicollinearity.

The Israeli Study of Surgical Infection of drains and the risk of wound infection in operations for hernia.

In a prospective follow-up study of 5,571 general surgical patients at 11 hospitals (23 departments) in Israel, 1,487 patients underwent operations for hernia (all types) and were screened daily for the development of postsurgical wound infection. Infection developed in 68 (4.6 per cent) at the site of the incision. Fourteen factors were analyzed for the presumed effect on the risk of infection; only four were significant in a multivariate model. Of these factors (old age, incarcerated or recurrent hernia, coexistent infection and drains), the introduction of drains had the strongest effect (relative risk equaled 4.1; p less than 0.001). Drains increased the risk in all the participating hospitals and in any category of patient. They prolonged the period a wound was susceptible to bacterial infection from nine to 16 days. The risk increased linearly with the duration of the drainage. The over-all proportion of patients who had a drain inserted was 19 per cent, with hospitals varying from 9 to 40 per cent of patients operated upon (p less than 0.001). Our findings suggest that the risk of an infection associated with drains may outweight their worth. Furthermore, there is no concensus among surgeons on the need for drains. Some surgeons use drains indiscriminately and others, rarely.

Analysis of geographic differentials in infant mortality rates. The Or Yehuda community.

Comprehensive evaluation of matched infant death certificate data and livebirth certificate data for 1977-1980 was performed for two areas in Israel: Or Yehuda, a small, low socioeconomic status community which had an infant mortality rate of 19.1 per 1,000, and the rest of Ramat Gan district, which had an infant mortality rate of 10.3 per 1,000. A method is presented which illuminates the role of statistical models in analyzing small area data, in evaluating twofold observed differences in crude and factor-specific mortality rates in two areas, in assessing heterogeneity in population stratum-specific mortality rate ratios, and in identifying causes for inter-area differences in infant mortality rate. Identical logistic models were fitted to each of the areas independently, and these were used to investigate effects due to birth weight, sex, parity, maternal age and education, and parental occupation. The differences in the distribution of risk level (number of risk factors) present in each population (or the proportion of multi-problem families) were identified as a single factor that can explain most of the disparity between the areas. The direction and magnitude of the relation between risk level and infant mortality rate were similar in both communities: the greater the number of risk factors, the higher the rate. Identification of a target population for intervention through only one or two specific risk factors would be unprofitable in reducing the overall community infant mortality rate since too many families with multiple risk would be excluded, and too many with single risk factors would be included.

Population variation in tooth, jaw, and root size: a radiographic study of two populations in a high-attrition environment.

Radiographs were taken of the jaws of skeletal remains of two populations of different-phenotype Prehistoric Australians from Roonka and Early New Zealanders (Maoris). On these radiographs crown, root, and corpus size were measured. Corpus height was subdivided into alveolar bone height, defined as the bone superior to the mandibular canal, and basal bone height, defined as that inferior to the mandibular canal. Both between and within the two populations there was a significant and negative correlation between crown size and corpus height. The differences between the two populations in corpus height were associated with differences in alveolar bone height rather than basal bone height and support hypotheses associating continued eruption of adult teeth with growth of the alveolar bone. The findings also support previous studies that have shown only a low correlation between crown size, root size, and corpus height.

Cholesterol uptake capacity of Acholeplasma laidlawii is affected by the composition and content of membrane glycolipids.

The composition of the cell membrane of 20 Acholeplasma laidlawii strains grown under identical conditions was studied and correlated with the capacity of these strains to incorporate cholesterol. Membranes of these strains had similar sodium dodecyl sulfate-polyacrylamide gel electrophoresis patterns and contained the same lipid species, but the relative amounts of the major polar lipids varied. Statistical analyses revealed that the glycolipids, monoglucosyldiglyceride, and an unidentified glycolipid (glycolipid-X) succeeded in explaining 90% (R2 = 0.90) of the cholesterol uptake variations. The regression coefficients for both glycolipids were negative (P less than 0.001), indicating that the capacity of A. laidlawii strains for cholesterol incorporation is inversely proportional to the relative amounts of these glycolipids. Accordingly, an increased capacity for cholesterol incorporation was detected upon aging of A. laidlawii cells. The aged cells contained significantly smaller amounts of both monoglucosyldiglyceride and glycolipid-X, and a higher amount of diglucosyldiglyceride. The change in cholesterol incorporation as a response to glycolipid composition and content can be explained by the low solubility of cholesterol in glycolipids as well as by the induction by the sterol molecule of a nonlamellar phase state that will destabilize a membrane structure containing monoglucosyldiglyceride and glycolipid-X.

A sequential test for comparing two infection rates in a randomized clinical trial, and incorporation of data accumulated after stopping.

A randomized sequential test is considered for deciding whether prophylactic antibiotic is successful in reducing infection rates following neurosurgical operations. The test is of the curtailed "gambler's ruin" type. Its operating characteristics are obtained and tabulated, and the actual medical experiment is described in some detail. The problem of how additional accumulated data, observed after stopping, due to delayed response, should be incorporated in evaluation of sequential P-values and confidence intervals is also considered.

Estimating fetal age by ultrasound measurements: an example of multivariate calibration.

A problem in the estimation of gestational age by ultrasound measurement of fetal bone lengths (the femur length, F, and the biparietal diameter, BPD) is described. By means of multivariate calibration techniques, F and BPD are combined to give an index of gestational age which is significantly more accurate than existing indices based on F or BPD alone. A detailed description of the statistical analysis is given, ranging from the initial steps of data analysis and model choice to the final step of determining the index and its accuracy. A number of interesting aspects of multivariate calibration theory are illustrated.

The impact of dietary measurement error on planning sample size required in a cohort study.

Dietary measurement error has two consequences relevant to epidemiologic studies: first, a proportion of subjects are misclassified into the wrong groups, and second, the distribution of reported intakes is wider than the distribution of true intakes. While the first effect has been dealt with by several other authors, the second effect has not received as much attention. Using a simple errors-in-measurement model, the authors investigate the implications of measurement error for the distribution of fat intake. They then show how the inference of a more narrow distribution of true intakes affects the calculation of sample size for a cohort study. The authors give an example of the calculation for a cohort study investigating dietary fat and colorectal cancer. This shows that measurement error has a profound effect on sample size, requiring a six- to eightfold increase over the number required in the absence of error, if the correlation coefficient between reported and true intakes is 0.65. Reliable detection of a relative risk of 1.36 between a true intake of greater than 47.5% calories from fat and less than 25% calories from fat would require approximately one million subjects.