Niels Keiding

Evidence for decreasing quality of semen during past 50 years.

OBJECTIVE--To investigate whether semen quality has changed during the past 50 years. DESIGN--Review of publications on semen quality in men without a history of infertility selected by means of Cumulated Index Medicus and Current List (1930-1965) and MEDLINE Silver Platter database (1966-August 1991). SUBJECTS--14,947 men included in a total of 61 papers published between 1938 and 1991. MAIN OUTCOME MEASURES--Mean sperm density and mean seminal volume. RESULTS--Linear regression of data weighted by number of men in each study showed a significant decrease in mean sperm count from 113 x 10(6)/ml in 1940 to 66 x 10(6)/ml in 1990 (p < 0.0001) and in seminal volume from 3.40 ml to 2.75 ml (p = 0.027), indicating an even more pronounced decrease in sperm production than expressed by the decline in sperm density. CONCLUSIONS--There has been a genuine decline in semen quality over the past 50 years. As male fertility is to some extent correlated with sperm count the results may reflect an overall reduction in male fertility. The biological significance of these changes is emphasised by a concomitant increase in the incidence of genitourinary abnormalities such as testicular cancer and possibly also cryptorchidism and hypospadias, suggesting a growing impact of factors with serious effects on male gonadal function.

Multi-state models for event history analysis

An introduction to event history analysis via multi-state models is given. Examples include the two-state model for survival analysis, the competing risks and illness-death models, and models for bone marrow transplantation. Statistical model specification via transition intensities and likelihood inference is introduced. Consequences of observational patterns are discussed, and a real example concerning mortality and bleeding episodes in a liver cirrhosis trial is discussed.

Neurobehavioral deficits associated with PCB in 7-year-old children prenatally exposed to seafood neurotoxicants.

Prenatal exposure to polychlorinated biphenyls (PCBs) was examined by analysis of cord tissue from 435 children from a Faroese birth cohort. Analysis of 50 paired cord blood samples showed excellent correlation with the cord tissue concentration (r=.90). Among 17 neuropsychological outcomes determined at age 7 years, the cord PCB concentration was associated with deficits on the Boston Naming Test (without cues, two-tailed P=.09 not adjusted for mercury; with cues, P=.03), the Continuous Performance Test reaction time (P=.03), and, possibly, on long-term recall on the California Verbal Learning Test (P=.15). The association between cord PCB and cord-blood mercury (r=.42) suggested possible confounding. While no PCB effects were apparent in children with low mercury exposure, PCB-associated deficits within the highest tertile of mercury exposure indicated a possible interaction between the two neurotoxicants. PCB-associated increased thresholds were seen at two of eight frequencies on audiometry, but only on the left side, and no deficits occurred on evoked potentials or contrast sensitivity. The limited PCB-related neurotoxicity in this cohort appears to be affected by concomitant methylmercury exposure.

Thrombotic Stroke and Myocardial Infarction with Hormonal Contraception

BACKGROUND Although several studies have assessed the risk of venous thromboembolism with newer hormonal contraception, few have examined thrombotic stroke and myocardial infarction, and results have been conflicting. METHODS In this 15-year Danish historical cohort study, we followed nonpregnant women, 15 to 49 years old, with no history of cardiovascular disease or cancer. Data on use of hormonal contraception, clinical end points, and potential confounders were obtained from four national registries. RESULTS A total of 1,626,158 women contributed 14,251,063 person-years of observation, during which 3311 thrombotic strokes (21.4 per 100,000 person-years) and 1725 myocardial infarctions (10.1 per 100,000 person-years) occurred. As compared with nonuse, current use of oral contraceptives that included ethinyl estradiol at a dose of 30 to 40 μg was associated with the following relative risks (and 95% confidence intervals) for thrombotic stroke and myocardial infarction, according to progestin type: norethindrone, 2.2 (1.5 to 3.2) and 2.3 (1.3 to 3.9); levonorgestrel, 1.7 (1.4 to 2.0) and 2.0 (1.6 to 2.5); norgestimate, 1.5 (1.2 to 1.9) and 1.3 (0.9 to 1.9); desogestrel, 2.2 (1.8 to 2.7) and 2.1 (1.5 to 2.8); gestodene, 1.8 (1.6 to 2.0) and 1.9 (1.6 to 2.3); and drospirenone, 1.6 (1.2 to 2.2) and 1.7 (1.0 to 2.6), respectively. With ethinyl estradiol at a dose of 20 μg, the corresponding relative risks according to progestin type were as follows: desogestrel, 1.5 (1.3 to 1.9) and 1.6 (1.1 to 2.1); gestodene, 1.7 (1.4 to 2.1) and 1.2 (0.8 to 1.9); and drospirenone, 0.9 (0.2 to 3.5) and 0.0. For transdermal patches, the corresponding relative risks were 3.2 (0.8 to 12.6) and 0.0, and for a vaginal ring, 2.5 (1.4 to 4.4) and 2.1 (0.7 to 6.5). CONCLUSIONS Although the absolute risks of thrombotic stroke and myocardial infarction associated with the use of hormonal contraception were low, the risk was increased by a factor of 0.9 to 1.7 with oral contraceptives that included ethinyl estradiol at a dose of 20 μg and by a factor of 1.3 to 2.3 with those that included ethinyl estradiol at a dose of 30 to 40 μg, with relatively small differences in risk according to progestin type. (Funded by the Danish Heart Association.).

Age-specific incidence and prevalence : a statistical perspective

In epidemiology incidence denotes the rate of occurrence of new cases (of disease), while prevalence is the frequency in the population (of diseased people). From a statistical point of view it is useful to understand incidence and prevalence in the parameter space, incidence as intensity (hazard) and prevalence as probability, and to relate observable quantities to these via a statistical model. In this paper such a framework is based on modelling each individuals dynamics in the Lexis diagram by a simple three‐state stochastic process in the age direction and recruiting individuals from a Poisson process in the time direction. The resulting distributions in the cross‐sectional population allow a rigorous discussion of the interplay between age‐specific incidence and prevalence as well as of the statistical analysis of epidemiological cross‐sectional data. For the latter, this paper focuses on methods from modern nonparametric continuous time survival analysis, including random censoring and truncation models and estimation under monotonicity constraints. The exposition is illustrated by examples, primarily from the authors epidemiological experience.

Statistical methods for the analysis and presentation of the results of bone marrow transplants. Part I: Unadjusted analysis

In this paper, we describe modern statistical methods for presentation of the results of studies of bone marrow transplantation. We focus here on ‘univariate’ or unadjusted techniques to describe the outcomes of such studies. In another paper we will discuss multivariate methods. We discuss the type of data one may have available to make inference about outcomes. We explain the differences between the Kaplan–Meier estimator of the survival function and the cumulative incidence curve, how these curves should be interpreted and when each is the appropriate summary statistic. We discuss the weighted log rank statistic and show how different weights can be used to put emphasis on detecting differences between groups in different time periods. We also present a simple estimate of current leukemia-free survival which is useful in summarizing post-transplant events. Bone Marrow Transplantation (2001) 28, 909–915.

Risk of acute nonlymphocytic leukemia and preleukemia in patients treated with cyclophosphamide for non-Hodgkin's lymphomas. Comparison with results obtained in patients treated for Hodgkin's disease and ovarian carcinoma with other alkylating agents.

Of 602 patients treated for non-Hodgkins lymphomas, 9 developed overt acute nonlymphocytic leukemia or preleukemia with refractory cytopenia and cytogenetic abnormalities of the bone marrow. A Kaplan-Meier estimate of the cumulative probability of leukemic complications was 6.3 +/- 2.6% (mean +/- SE) 7 years after start of treatment. All 9 patients with leukemic complications belong to a major subgroup of 498 patients treated with alkylating agents, predominantly cyclophosphamide. The risk of leukemic complications in this subgroup was compared with the risk in 312 patients treated with other alkylating agents for Hodgkins disease, and with the risk in 553 patients treated with dihydroxybusulfan for ovarian carcinoma. Cumulative 9-year risks were 8.0 +/- 3.3%, 12.8 +/- 3.5%, and 7.1 +/- 1.9%, respectively. The general risk of secondary leukemia after long-term treatment with alkylating agents ranges from 1% to 1.5% per year from 2 to at least 9 years after start of treatment.

Statistical methods for the analysis and presentation of the results of bone marrow transplants. Part 2: Regression modeling

In this paper, we address methods of multivariate regression. We discuss the value of regression compared to matched pairs analysis, methods of coding variables, basic concepts of the Cox model and interpretation of results of the Cox model. We present methods of handling variables whose effect changes with time. We present methods to check the assumptions of the Cox regression. Finally, and perhaps most importantly, we provide suggestions for presenting the results in clear and thorough tables and graphs.Bone Marrow Transplantation (2001) 28, 1001–1011.

Fecundability in relation to body mass and menstrual cycle patterns

Few studies have investigated the association between body mass index and fecundability, that is, the ability to conceive in a menstrual cycle, among fertile women with normal menstrual cycle pattern. We examined the independent and combined effects of duration and regularity of the menstrual cycle, body mass index, and fecundability from records on pregnant women attending antenatal care at Odense University Hospital, Denmark, between 1972 and 1987. We included only the first birth of each woman who had planned pregnancies and no pre-pregnancy disease (N = 10,903). We estimated the fecundability odds ratio (FR) as the odds of conception in a menstrual cycle. After adjusting for confounders, the fecundability for women with a body mass index >25 kg/m2 was lower than for women with a body mass index of 20-25 kg/m2 [FR = 0.77; 95% confidence interval (CI) = 0.70-0.84]. FR was lower for women with long (>35 days) (FR = 0.74; 95% CI = 0.63-0.87) or irregular cycles (FR = 0.78; 95% CI = 0.70-0.87), even when their body mass index was within the normal range (20-25 kg/m2) and/or their cycles were regular.We examined the association between birth characteristics of offspring and the subsequent maternal risk of breast cancer in a population-based cohort of 998,499 women, 13 to 48 years of age at entry. There were 9,495 incident cases of breast cancer during 12.8 million person-years of follow-up among these women. Compared with mothers of singleton infants, mothers having a multiple birth had an increased risk of breast cancer in the first 5 years after a birth (relative risk (RR) = 1.8; 95% confidence interval (CI) = 1.1-2.8). The risk for mothers having a heavy-weighted child (>3.75 kg), as compared with a child of light weight (< or =3 kg), was also slightly increased (RR = 1.2; 95% CI = 0.9-1.5). This latter effect was primarily due to an increased incidence of tumors larger than 2 cm at diagnosis (RR = 1.4; 95% CI = 0.9-1.9). Our findings are compatible with the hypothesis that the hormonal level during pregnancy influences the risk of breast cancer in the early years after delivery.

THE ROLE OF FRAILTY MODELS AND ACCELERATED FAILURE TIME MODELS IN DESCRIBING HETEROGENEITY DUE TO OMITTED COVARIATES

In survival analysis, deviations from proportional hazards may sometimes be explained by unaccounted random heterogeneity, or frailty. This paper recalls the literature on omitted covariates in survival analysis and shows in a case study how unstably frailty models might behave when asked to account for unobserved heterogeneity in standard survival analysis with no replications per heterogeneity unit. Accelerated failure time modelling seems to avoid these difficulties and also to yield easily interpretable results. We propose that it would be advantageous to upgrade the accelerated failure time approach alongside the hazard modelling approach to survival analysis.