Folate intake, alcohol and risk of breast cancer among postmenopausal women in Denmark.

OBJECTIVE: There is consistent evidence that alcohol increases the risk of breast cancer. It has been suggested that the increased risk associated with alcohol intake may be reduced by adequate intake of folate. Since many women consume alcohol, detection of a risk-reducing mechanism would have major public health implications. DESIGN: We therefore evaluated the possible interaction between alcohol and folate in a paired nested case-control study among postmenopausal women. SETTING: A total of 24 697 postmenopausal women were included in the 'Diet, Cancer and Health' follow-up study between December 1993 and May 1997. The cohort was followed until December 2000. The study included 388 cases of breast cancer and 388 randomly selected controls were used to estimate the breast cancer incidence rate ratio (IRR) in conditional logistic regression analysis. RESULTS: A previously established association between alcohol intake and risk of breast cancer was present mainly among women with low folate intake. An IRR of 1.19 (95% CI: 0.99-1.42) per 10 g average daily alcohol intake was found for women with a daily folate intake below 300 mug, while among women with a folate intake higher than 350 mug, we could not show an association between the alcohol intake and the breast cancer incidence rate (e.g. folate intake >400 mug; IRR of 1.01 (95% CI: 0.85-1.20)). CONCLUSION: The findings support the evidence that adequate folate intake may attenuate the risk of breast cancer associated with high alcohol intake. SPONSORSHIP: The Danish Cancer Society.

Increased risk of ovarian cancer in integrin beta3 Leu33Pro homozygotes.

We previously demonstrated that integrin beta(3) Leu33Pro homozygotes have an increased risk of cancer, possibly most pronounced for ovarian cancer. We now test the latter hypothesis in case-control and prospective studies. We genotyped 463 Danish women with ovarian cancer, and 4291 women from the Danish general population. Calculation of odds ratios by conditional logistic regression was performed in the case-control study (n = 463 + 3543), and of ovarian cancer incidence, log-rank statistics and hazard ratios by Cox regression in the prospective study (n = 4291) with 9.5-year follow-up. In the case-control study matched for age and marital status, the odds ratio for ovarian cancer in homozygotes versus non-carriers was 1.6 (95% confidence interval: 1.0-2.6). In the prospective study with 28 incident ovarian cancers, non-carriers and homozygotes had incidences of 7 (4-11) and 30 (10-92) per 10 000 person-years (log-rank P = 0.02). The age-adjusted hazard ratio for ovarian cancer in homozygotes versus non-carriers was 3.9 (1.1-13). Risk of ovarian cancer did not differ between heterozygotes and non-carriers in either study. Integrin beta(3) Leu33Pro homozygotes have an increased risk of ovarian cancer.

Waist circumference and body composition in relation to all-cause mortality in middle-aged men and women.

OBJECTIVE: Waist circumference is directly related to all-cause mortality when adjusted for body mass index (BMI). Body fat and fat-free body mass, when mutually adjusted, show with increasing values an increasing and decreasing relation to all-cause mortality. We investigated the association of waist circumference and body composition (body fat and fat-free mass), mutually adjusted, to all-cause mortality. DESIGN: A Danish prospective cohort study with a median follow-up period of 5.8 y. SUBJECTS: In all, 27 178 men and 29 875 women, born in Denmark, aged 50-64 y, and without diagnosis of cancer at the time of invitation. MEASUREMENTS: Waist circumference and body composition estimated from impedance measurements. Cox's regression models were used to estimate the mortality rate ratios (RR). RESULTS: Waist circumference was strongly associated with all-cause mortality after adjustment for body composition; the mortality RR was 1.36 (95% confidence intervals (CI): 1.22-1.52) times higher per 10% larger waist circumference among men and 1.30 (95% CI: 1.17-1.44) times higher among women. Adjustment for waist circumference eliminated the association between high values of the body fat mass index (BFMI) and all-cause mortality. The association between fat-free mass index (FFMI) and mortality remained unaltered. CONCLUSION: Waist circumference accounted for the mortality risk associated with excess body fat and not fat-free mass. Waist circumference remained strongly and directly associated with all-cause mortality when adjusted for total body fat in middle-aged men and women, suggesting that the increased mortality risk related to excess body fat is mainly due to abdominal adiposity.

Waist and hip circumferences and all-cause mortality: usefulness of the waist-to-hip ratio?

OBJECTIVE: To investigate whether waist and hip circumferences, in addition to body mass index (BMI), are related to all-cause mortality. We studied these associations and tested the usefulness of the waist-to-hip ratio for mortality prediction. DESIGN: A Danish prospective cohort study with data collected between 1993 and 1997. SUBJECTS: A total of 27 179 men and 29 875 women born in Denmark and aged 50-64 years were followed for a median of 6.8 years. MEASUREMENTS: BMI, waist and hip circumferences at baseline. RESULTS: The associations between hip circumference and all-cause mortality were inverse for both men and women, but only after adjustment for waist circumference, or BMI, or both. The mortality rate ratios of mutually adjusted waist and hip circumferences were 0.63 (95% CI: 0.56, 0.71), and 0.70 (95% CI: 0.63, 0.79) times higher per 10% larger hip circumference in men and women, respectively, and 1.45 (95% CI: 1.34, 1.57) and 1.22 (95% CI: 1.14, 1.31) times higher per 10% larger waist circumference. The adequacy of the waist-to-hip ratio as a substitute for separate measurements of waist and hip circumferences depended on which other variables the analysis was adjusted for, indicating that the waist-to-hip ratio should be used with precaution. CONCLUSION: When mutually adjusted, waist and hip circumferences show opposite associations with all-cause mortality, probably due to different effects of adipose tissue in the abdominal and gluteofemoral regions. The waist-to-hip ratio cannot always capture these relations adequately.

Sex ratio among offspring of childhood cancer survivors treated with radiotherapy.

It has been postulated that paternal gonadal exposure would increase the sex ratio by inducing X-chromosomal dominant lethals but that maternal gonadal exposure would decrease the sex ratio by inducing recessive sex-linked lethals. We therefore evaluated the sex ratio (male-to-female ratio) of children born to survivors of childhood cancers in Denmark. Children with cancer were identified from the Danish Cancer Registry from 1943 to 1996 and their offspring from the Central Population Registry. Radiation treatments were determined from records within the Cancer Registry and gonadal radiation exposures were estimated based on the cancer being treated and the likely proximity of the radiation fields to the gonads. Overall, 1100 survivors of childhood cancer became the parents of 2130 children. The sex ratio for male (0.99) and female (1.00) cancer survivors was similar and did not differ significantly from the Danish population (1.06). Radiotherapy did not influence the sex ratio of the children of either male or female survivors, and there was no evidence for dose-related changes over categories of estimated dose to parental gonads. We saw no consistent association between the sex ratio and the interval between cancer diagnosis of the parent and birth of the child. This nationwide study provides no support for the hypothesis that radiation exposure to the gonads results in an inherited genetic effect that would be manifested by a change in the sex ratio of children born after exposure. It may be, however, that sex ratio alterations are not a good or even a valid indicator of possible genetic effects in humans.

Does the association between smoking status and selected healthy foods depend on gender? A population-based study of 54 417 middle-aged Danes.

OBJECTIVE: Smoking serves different functions for men and women. Thus, we wanted to investigate the association between smoking behaviour and intakes of selected healthy foods in men and women with special focus on differences and similarities between the two genders. DESIGN: In 1993-1997, a random sample of 80 996 men and 79 729 women aged 50-64 y was invited to participate in the study 'Diet, Cancer and Health'. In all, 27 179 men and 29 876 women attended a health examination and completed a 192-item food-frequency questionnaire (FFQ). The association between smoking status and low, median and high intakes of selected foods was examined among 25 821 men and 28 596 women. SETTING: The greater Copenhagen and Aarhus area, Denmark. RESULTS: For both men and women, smoking status group was associated with diet, such that increasing level of smoking status ranging from never smokers over ex-smokers to currently heavy smokers was associated with a lower intake of the healthy foods: fresh fruit, cooked vegetables, raw vegetables/salad, and olive oil. For wine, increasing level of smoking status category was associated with a higher fraction of abstainers and heavy drinkers. The difference between the extreme smoking status categories was larger than the difference between men and women within smoking status categories such that never smoking men in general had a higher intake of healthy foods than heavy smoking women. Correction for age, educational level, and body mass index (BMI) did not affect the results. CONCLUSION: In this middle-aged population, intake of healthy foods were associated with smoking behaviour with a dose-response type of relationship. The overall pattern was similar for men and women.

The influence of calcium intake and physical activity on bone mineral content and bone size in healthy children and adolescents.

Studies of determinants of bone mineralization during growth are relevant to the attempt to increase peak bone mass. The aim of this study was to examine how calcium intake and physical activity influence bone size (bone area, BA), accretion in BA, whole body bone mineral content (BMC) and accretion in BMC. BA and BMC were examined by dual-energy X-ray absorptiometry (Hologic 1000/W) in healthy girls (n = 192) and boys (n = 140) aged 5-19 years at baseline and 1 year later. Calcium intake was assessed three times by a food frequency questionnaire and physical activity three times by a 24 h recall questionnaire. The influence of calcium intake and physical activity was examined by multiple regression. BA was size-adjusted by including height and weight in all analyses, and BMC was size-adjusted by including BA, height and weight in all analyses. Size-adjusted average BA was associated neither with average calcium intake nor with average physical activity. Size-adjusted accretion in BA was borderline associated with the average calcium intake in boys only (p = 0.07). Size-adjusted average BMC was positively associated with average calcium intake (p) = 0.03 girls; p = 0.07 boys) and borderline associated with average physical activity level in boys (p = 0.07) but not girls (p = 0.7). Size-adjusted accretion in BMC was significantly associated neither with average calcium intake nor with average physical activity level, but was associated with change in calcium intake over the 1 year observation period in boys (p = 0.03) but not girls (p = 0.9). In conclusion, we found that size-adjusted BMC in school-aged children was positively associated with average calcium intake. Size-adjusted accretion in BMC was positively associated with change in dietary calcium intake in boys only. To what degree this is caused by a reduction in remodeling space is unknown.

Connective tissue disease after hip and knee implant surgery.

OBJECTIVE: To increase the knowledge of the long-term effects of artificial hip and knee joint implants. METHODS: The study groups consisted of 24,636 patients with osteoarthritis who underwent hip implant surgery and 5,221 who received knee implants during 1977-89. The post-implant rate of hospitalization for connective tissue disease (CTD) was compared with the rate in the general population of Denmark and with that among osteoarthritis patients without implant surgery. RESULTS: The rates of hospitalization for CTD were higher than the background level among both hip and knee implant patients with osteoarthritis, whereas the comparison with non-implanted osteoarthritis patients revealed that the hospitalization rate for CTD was reduced after hip implant surgery, but increased after knee implant surgery. CONCLUSION: Since the materials used in hip and knee implants in Denmark are not substantially different, these results are unlikely to reflect an implant effect but rather the selection criteria of referral for implant surgery.

Air pollution from traffic at the residence of children with cancer.

The hypothesis that exposure to traffic-related air pollution increases the risk of developing cancer during childhood was investigated. The authors enrolled 1,989 children reported to the Danish Cancer Registry with a diagnosis of leukemia, tumor of the central nervous system, or malignant lymphoma during 1968-1991 and 5,506 control children selected at random from the entire childhood population. The residential histories of the children were traced from 9 months before birth until the time of diagnosis of the cases and a similar period for the controls. For each of the 18,440 identified addresses, information on traffic and the configuration of streets and buildings was collected. Average concentrations of benzene and nitrogen dioxide (indicators of traffic-related air pollution) were calculated for the relevant period, and exposures to air pollution during pregnancy and during childhood were calculated separately. The risks of leukemia, central nervous system tumors, and all selected cancers combined were not linked to exposure to benzene or nitrogen dioxide during either period. The risk of lymphomas increased by 25% (p for trend = 0.06) and 51% (p for trend = 0.05) for a doubling of the concentration of benzene and nitrogen dioxide, respectively, during the pregnancy. The association was restricted to Hodgkin's disease.

Development of the obesity epidemic in Denmark: cohort, time and age effects among boys born 1930-1975.

OBJECTIVE: A global epidemic of obesity is developing, but its causes are still unclear. In Denmark, two periods of steep increases in prevalence of obesity have occurred among young men born in the 1940s and 1960-70s. This study investigated the preceding changes in prevalence of obesity and in the entire body mass index (BMI = weight/height2) distribution by birth cohort, calendar time and age among Danish school boys. METHODS: Children attending Copenhagen schools 1937-1983 had annual health examinations, from which we computerized 1,037,468 measurements of height (m) and weight (kg) of 161,314 boys aged 7-13 y. Obesity was defined as age-specific BMI exceeding the 95.0, the 99.0 and the 99.9 percentile among those born 1930-1934, the latter corresponding to the prevalence of obesity among the young men in these cohorts. The median, standard deviation, skewness, and the 5th, 25th, 75th and 95th percentiles of the age-specific BMI were estimated for each birth cohort. RESULTS: The prevalence of obesity, defined by the 99.9 percentile, increased at all ages during the same birth years as among the young men, and, accordingly, at earlier calendar years. The prevalence of obesity, defined by the 95.0 percentile, showed a distinctly different pattern: a sharp increase, irrespective of age, during the calendar years 1947-1949, and thereafter a stable level until the 1970s, where a further modest increase began. The prevalence defined by the 99.0 percentile showed a mixture of the trends in those defined by the 99.9 and 95.0 percentiles. The median BMI showed small fluctuations, parallel at all ages. The standard deviation and right-sided skewness increased until birth year 1950, but were almost stable thereafter. The pattern of changes in the quartiles mostly reflected those in the median. CONCLUSIONS: The prevalence of obesity defined by the 99.9 or 99.0 percentile has increased in Danish boys born in the 1940s and since the mid 1960s, without corresponding changes in the central part of the BMI distribution. When defining obesity by the 95.0 percentile, there was a sharp distinct age-independent increase in the late 1940s. The development of the obesity epidemic is a heterogeneous phenomenon that has involved changes in environmental influences starting at preschool ages and affecting different subsets of the population, either because of selective exposure or particular susceptibility.

Whole body bone mineral accretion in healthy children and adolescents.

Data on accretion in bone size and bone mineral content (BMC) are needed to evaluate bone mineralisation during childhood. Whole body bone mineral content (BMC) and bone area (BA) were determined by dual energy x ray absorptiometry (Hologic 1000/W) with a one year interval in healthy girls (n = 192) and boys (n = 140) aged 6-19 years. Annual accretion in BMC (DeltaBMC (g/year)) and BA (DeltaBA (cm2/year)) according to sex and pubertal stages were calculated. DeltaBA and DeltaBMC were highly significantly associated with pubertal stages in girls and boys. Centile curves for DeltaBA and DeltaBMC according to sex and age were constructed using the LMS method. Peak DeltaBA and DeltaBMC values were reached earlier in girls (12.3 and 12.5 years, respectively) than in boys (13.4 and 14. 2 years, respectively). The DeltaBA peak was dissociated in time from the DeltaBMC peak, indicating that increase in bone size occurs before increase in bone mineral content. Assuming that 32.2% of BMC consist of calcium, the median (90th centile) annual bone calcium accretion in pubertal stage III was 220 mg/day (302) and 317 mg/day (386) for girls and boys, respectively. To obtain an average bone calcium accretion, a high calcium absorption is needed during puberty. This may have implications for dietary calcium requirements at this time.

Influence of breastfeeding and complementary food on growth between 5 and 10 months.

The aim of this study was to examine the nature of the association between breastfeeding, complementary feeding and growth in a random sample of infants from Denmark, where the prevalence of breastfeeding is high. A semiquantitative food frequency questionnaire and a questionnaire on breastfeeding duration and on weight and length measurements taken at the infant welfare visit at 5 and 10 months were sent to 590 families with 10-month-old infants. A total of 339 infants with complete growth data were included in the analyses. When controlling for mid-parental height and birth weight infants breastfed for > or =7 months gained 198 g less in weight (p < 0.01) and 7 mm less in length (p < 0.01) during the period from 5 to 10 months than infants breastfed for < 7 months. Controlling for these effects, the 10% of the sample with the highest protein intake (i.e. > or =16 energy percentage) gained 262 g more than those with a lower protein intake (p = 0.03). Infants breastfed for > or =7 months received significantly less cow's milk (p < 0.01), and fewer meat-containing dishes (p < 0.05) and sweets or cakes (p < 0.01), which may partly explain the effect of breastfeeding. The long-term consequences of this moderate difference in growth velocity are unknown and the findings should not be used to advocate against breastfeeding during late infancy.

Influence of weight, age and puberty on bone size and bone mineral content in healthy children and adolescents.

The biology of bone mineralization during growth is important for peak bone mass. The aim of the study was to examine how body size, age and puberty influence bone size and bone mineral density. Whole body bone area (BA) and bone mineral content (BMC) were examined by dual-energy X-ray absorptiometry (Hologic 1000/W) in healthy girls (n = 201) and boys (n = 142) aged 5-19 y. The influence of height, weight, age and puberty on bone mineralization was examined by multiple regression. Main determinants of BA were height and weight. Bone width, approximated by BA corrected for height, increased highly significantly with weight and depended weakly significantly on pubertal stage. Main determinants of BMC were BA, height, age and pubertal stages. Bone mineral density, approximated by BMC corrected for BA and height, depended on age and pubertal stage, but not on weight. Thus skeletal size is mainly determined by body size, while bone density is determined by age and pubertal stage.

Analysis of the treatment effect on recurrent bleeding and death in patients with cirrhosis and esophageal varices: multistage competing-risks model compared to conventional methods. The Copenhagen Esophageal Varices Sclerotherapy Project.

BACKGROUND/AIMS: Multiple recurrences of bleeding with high mortality in cirrhosis with esophageal varices have been inadequately analyzed in previous trials. We propose analysis by the multistage competing-risks model, specifying the effect on overall mortality as an effect on mortality during bleeding, rate of cessation of bleeding, mortality rate without bleeding, and rate of rebleeding. METHODS: The Copenhagen Esophageal Varices Project enrolled patients after first bleeding and randomized 94 to usual treatment and 93 to sclerotherapy as supplement. During 9-52 months of follow-up, rebleeding occurred in 49 and 42, and death in 68 and 60 patients, respectively. The proportional hazards regression model (Cox model) was used for reanalysis both by the multistage competing-risks model and by conventional analysis for overall mortality and rate of first rebleeding. In the multistage model, time zero was at entry to any new disease stage, of which the first four were analyzed - two bleeding stages and two bleeding-free stages. RESULTS: The conventional analysis showed a reduction of overall mortality rate in the sclerotherapy group of borderline significance, but no effect on rate of rebleeding. The multistage model indicated that sclerotherapy reduced the rate of rebleeding late in the disease course, and particularly after the first rebleeding. Rate of cessation of bleeding and mortality rates during bleeding and without bleeding were not affected by sclerotherapy. CONCLUSIONS: Conventional analysis may give misleading conclusions, which might be avoided by applying the multistage model. The effect of sclerotherapy on overall mortality may be ascribed entirely to the reduced rate of rebleeding.

Front-door concentrations and personal exposures of Danish children to nitrogen dioxide.

The aims of the study were to evaluate the front-door concentration of traffic exhaust fumes as a surrogate for the personal exposure of children and to study factors in the behavior and the environment of children that affect their personal exposure to nitrogen dioxide (NO(2)). The exposure to NO(2) of 103 children living in Copenhagen and 101 children living in rural areas of Denmark was studied by measuring average concentrations over 1 week with diffusive badge samplers placed outside the front door of the home, inside the child's bedroom, and on each child. Detailed information about the activities of the children involving potential exposure to NO(2) was noted in diaries. The results indicated that the front-door concentration of traffic pollution might be used to classify the personal exposure of urban children, although misclassification would be introduced. Multiple regression analysis showed several factors that affected the personal NO(2) exposure of the children independently, including the front-door concentration, the bedroom concentration, time spent outdoors, gas appliances used at home, passive smoking, and burning candles.

Whole body bone mineral content in healthy children and adolescents.

Data from healthy children are needed to evaluate bone mineralisation during childhood. Whole body bone mineral content (BMC) and bone area were examined by dual energy x ray absorptiometry (Hologic 1000/W) in healthy girls (n = 201) and boys (n = 142) aged 5-19 years. Centile curves for bone area for age, BMC for age, bone area for height, and BMC for bone area were constructed using the LMS method. Bone mineral density calculated as BMC/bone area is not useful in children as it is significantly influenced by bone size. Instead, it is proposed that bone mineralisation is assessed in three steps: height for age, bone area for height, and BMC for bone area. These three steps correspond to three different causes of reduced bone mass: short bones, narrow bones, and light bones.

Ambient air levels and the exposure of children to benzene, toluene, and xylenes in Denmark.

The aims of the study were to evaluate if the front-door concentrations of benzene, toluene, and xylenes can be used to classify the personal exposures of Danish children and to identify factors that affect their personal exposure. Average concentrations were measured over 1 week with diffusive samplers, and the personal exposures of 98 children and the concentrations outside the front doors of their homes were measured simultaneously. Time and activity patterns were noted in diaries. The front-door concentrations were significantly higher in Copenhagen than in rural areas (all P < 0.0001), but the personal exposures were only slightly higher. Even though the personal exposures were highly significantly associated with front-door concentrations in urban areas (all P < 0.004), use of the residential front-door concentration as an exposure surrogate would imply misclassification, as it cannot be used for rural children. Multiple regression analyses brought to light several factors that affect the exposure of children independently, including front-door concentration, riding in cars, and activities involving potential exposure to gasoline vapors like motocross, moped driving, and refueling of cars.

[Duration of breast feeding--which factors are significant?]. / Ammeperiodens laengde--hvilke faktorer har betydning.

Duration of breast-feeding (BF) was studied in 249 randomly chosen healthy, term infants of Danish origin of which 81% participated. BF was initiated by 99.5% of the mothers. At three, six, and nine months, 71%, 52%, and 33%, respectively, were still BF. Only one infant (0.5%) was exclusively BF beyond the age of seven months. In a Cox multiple regression analysis of factors influencing duration of BF, we found a positive association with maternal education (p < 0.001), and age (p = 0.02), and a negative association with the amount of formula given at the maternity ward (p < 0.001). Six months after delivery 79% of the mothers with higher school education (> or = 12 years) were still BF, compared to 29% with a low school education (< or = 9 years). There is still a need for an increased effort to support mothers in BF to be focused on younger mothers with short school education. Formula supplements during the first days of life, given to 73% of the infants, were associated with shorter duration of BF and should be discouraged.

Weight, length, head circumference, and growth velocity in a longitudinal study of Danish infants.

Two longitudinal studies of infant growth, performed from 1985 to 1988 in the Copenhagen area, have been combined to develop an up-to-date growth reference. Percentile curves were constructed and median growth velocities were calculated for monthly intervals, based on individually estimated growth curves. Compared with references based on data from periods when the prevalence of breast-feeding was low, these data support previous suggestions of a new growth pattern, with higher velocities during the first months and slower velocities during the remaining infancy. The growth of breast-fed infants differs from that of infants not being breast-fed, and special growth references for breast-fed infants have been suggested. We therefore examined growth patterns in the present study according to duration of breast-feeding. At 12 months, infants breast-fed > or = 9 months weighed less (400 g (95% CI: 70 g, 740 g) and were 1.0 cm shorter (0.3 cm, 1.8 cm) than infants breast-fed < 9 months. Part of this difference was already present at six months. Despite this we recommend a single growth reference for Danish infants regardless of mode of feeding. The slower growth in infants breast-fed > or = 9 months, which could be due to differences in composition of weaning foods, should be investigated further.

Optimized analysis of recurrent bleeding and death in patients with cirrhosis and esophageal varices. Copenhagen Esophageal Varices Sclerotherapy Project.

The course of cirrhosis after the first episode of variceal bleeding is characterized by recurrent episodes of bleeding and bleeding-free periods, with each new stage being associated with an initially high mortality rate. Analysis of the course of this disease is usually based on traditional survival models measuring observation time from admission to hospital. This approach cannot provide valid answers to the following clinically relevant questions: 1) what is the risk of death while bleeding? 2) what is the risk of continuing bleeding? 3) what is the risk of death after bleeding has stopped? 4) what is the risk of a new bleeding episode among those who remain alive and do not bleed? and 5) do the answers to the former four questions depend on the number of previous bleeding episodes or on the time since the first episode? These questions may be addressed using the multistage, competing risks model presented here. By measuring time from the start of each new bleeding or bleeding-free period, this model synchronizes the patients with respect to disease development and, hence, to level of the rate of new clinical events. This reduces the heterogeneity of risk levels in the patient group at any given time. In an analysis of 94 patients with cirrhosis admitted for their first episode of variceal bleeding, we estimated the mortality rates during bleeding and without current bleeding, the rate of cessation of bleeding and the rate of rebleeding. The influence of number of previous rebleedings and of the time since the first bleeding was examined using a Cox regression model for competing risks.(ABSTRACT TRUNCATED AT 250 WORDS)