The Current Duration Approach to Estimating Time to Pregnancy

Abstract. Time‐to‐pregnancy (TTP) is the duration from the time a couple starts trying to become pregnant until they succeed. It is considered one of the most direct methods to measure natural fecundity in humans. Statistical tools for designing and analysing time to pregnancy studies belong to survival analysis, but several features require special attention. Prospective designs are difficult to carry out and retrospective (pregnancy‐based) designs, being widely used in this area, do not allow efficiently including couples remaining childless. A third possible design starts from a cross‐sectional sample of couples currently trying to become pregnant, using current duration (backward recurrence time) as basis for the estimation of TTP. Regression analysis is then most conveniently carried out in the accelerated failure time model. This paper surveys some practical and technical‐statistical issues in implementing this approach in a large telephone‐based survey, the Epidemiological Observatory of Fecundity in France (Obseff).     

Semiparametric Bayesian classification with longitudinal markers

Summary.  We analyse data from a study involving 173 pregnant women. The data are observed values of the β human chorionic gonadotropin hormone measured during the first 80 days of gestational age, including from one up to six longitudinal responses for each woman. The main objective in this study is to predict normal versus abnormal pregnancy outcomes from data that are available at the early stages of pregnancy. We achieve the desired classification with a semiparametric hierarchical model. Specifically, we consider a Dirichlet process mixture prior for the distribution of the random effects in each group. The unknown random-effects distributions are allowed to vary across groups but are made dependent by using a design vector to select different features of a single underlying random probability measure. The resulting model is an extension of the dependent Dirichlet process model, with an additional probability model for group classification. The model is shown to perform better than an alternative model which is based on independent Dirichlet processes for the groups. Relevant posterior distributions are summarized by using Markov chain Monte Carlo methods.     

Performance evaluation of propensity score methods for estimating average treatment effects with multi-level treatments*

The propensity score (PS) method is widely used to estimate the average treatment effect (TE) in observational studies. However, it is generally confined to the binary treatment assignment. In an extension to the settings of a multi-level treatment, Imbens proposed a generalized propensity score which is the conditional probability of receiving a particular level of the treatment given pre-treatment variables. The average TE can then be estimated by conditioning solely on the generalized PS under the assumption of weak unconfoundedness. In the present work, we adopted this approach and conducted extensive simulations to evaluate the performance of several methods using the generalized PS, including subclassification, matching, inverse probability of treatment weighting (IPTW), and covariate adjustment. Compared with other methods, IPTW had the preferred overall performance. We then applied these methods to a retrospective cohort study of 228,876 pregnant women. The impact of the exposure to different types of the antidepressant medications (no exposure, selective serotonin reuptake inhibitor (SSRI) only, non-SSRI only, and both) during pregnancy on several important infant outcomes (birth weight, gestation age, preterm labor, and respiratory distress) were assessed.     

A multilevel analysis of longitudinal ordinal data: evaluation of the level of physical performance of women receiving adjuvant therapy for breast cancer

Longitudinal health-related quality-of-life (QOL) data are often collected as part of clinical studies. Here two analyses of QOL data from a prospective study of breast cancer patients evaluate how physical performance is related to factors such as age, menopausal status and type of adjuvant treatment. The first analysis uses summary statistic methods. The same questions are then addressed using a multilevel model. Because of the structure of the physical performance response, regression models for the analysis of ordinal data are used. The analyses of base-line and follow-up QOL data at four time points over two years from 257 women show that reported base-line physical performance was consistently associated with later performance and that women who had received chemotherapy in the month before the QOL assessment had a greater physical performance burden. There is a slight power gain of the multilevel model over the summary statistic analysis. The multilevel model also allows relationships with time-dependent covariates to be included, highlighting treatment-related factors affecting physical performance that could not be considered within the summary statistic analysis. Checking of the multilevel model assumptions is exemplified.     

Seasonal patterns of fertility measures: theory and data

The distribution of births by month exhibits a seasonal pattern in most populations. The monthly marital fertility rate for an area of Bangladesh provides a good example of the seasonal periodicity. Seasonal patterns of measures of reproduction in a population of married women are considered. Equations are developed that predict the seasonal patterns of these alternative measures under the assumption that the fertility rate (R) follows a trigonometric curve. This is followed by an empirical analysis of the measures in a Bangladesh population that has a pronounced seasonal fertility. The investigation is intended both to validate the theoretical framework developed in the 1st part of the paper as well as to determine whether seasonal variation in actual populations is sufficiently large to affect the alternative measures significantly. 4 measures are considered: pregnancy prevalence (PP)--the proportion of married women who are pregnant at the survey date; mean open birth interval (MOI)--the time from the last live birth to the date of the survey for parous women and from the time of marriage to the date of the survey for nulliparous women; mean closed interval-birth (MCIB)--the mean interval between the last 2 live births for married women who have a birth in the period immediately preceding the survey date; and mean closed interval-woman (MCIW)--the mean interval between the last 2 live births for women who have had at least 2 children by the time of the survey. It is assumed that the seasonal pattern of the fertility rate of a population follows a cosine curve and that there is no trend in annual fertility from year to year. The lag and relative variability of the other measures are considered in comparison with the fertility rate curve. The predictions from this theoretical effort, when compared with observed patterns and trigonometric regression results for each measure in data from Bangladesh, are shown to be quite accurate. The figure and regression results show that R, PP, and MOI have definite seasonal periodicity, but MCIB and MCIW do not display any seasonal patterns. If there is a secular trend in fertility in addition to seasonality, these relationships between the seasonal patterns of the measures may no longer hold. There is a disadvantage to using closed interval measures, for they are unable to detect effects of limiting of childbearing in a population since they are based only on information from women who have births.     

A frailty modeling approach for parental effects in animal breeding

Survival models involving frailties are commonly applied in studies where correlated event time data arise due to natural or artificial clustering. In this paper we present an application of such models in the animal breeding field. Specifically, a mixed survival model with a multivariate correlated frailty term is proposed for the analysis of data from over 3611 Brazilian Nellore cattle. The primary aim is to evaluate parental genetic effects on the trait length in days that their progeny need to gain a commercially specified standard weight gain. This trait is not measured directly but can be estimated from growth data. Results point to the importance of genetic effects and suggest that these models constitute a valuable data analysis tool for beef cattle breeding.     

Men matter more: the social class identity of married British women, 1985–1991

Summary.  We investigate the class identity of married women as it relates to their own and their husband's class position. Whereas previous workers have attempted to test whether identity depends solely on the husband's position, not at all on the husband's position or equally on the husband's and wife's position, leaving out all intermediate cases, we estimate new diagonal reference models that quantify the relative weight of each partner's class position on their own class identity. In previous literature, it was also argued that women who work full time should be more likely to adopt a sharing model than other women and in some cases these different types of women were compared. We move beyond this simple dichotomy and more systematically formulate hypotheses about the conditions under which women attach more or less weight to their own class position and less or more weight respectively to that of their husbands. To test these hypotheses, we consider models where the weights are allowed to depend on characteristics of each partner and the couple. Using the British Social Attitudes Survey data for 1985–1991, we find that, when the husband's commitment to the labour force exceeds that of the wife, the husband's weight exceeds the wife's weight but, when the wife's commitment exceeds that of the husband's, the weights are approximately equal. We also find (unexpectedly) that women who hold higher positions than their husbands attach more weight to their husband's position than to their own position.     

Weighted log‐rank test for time‐to‐event data in immunotherapy trials with random delayed treatment effect and cure rate

A cancer clinical trial with an immunotherapy often has 2 special features, which are patients being potentially cured from the cancer and the immunotherapy starting to take clinical effect after a certain delay time. Existing testing methods may be inadequate for immunotherapy clinical trials, because they do not appropriately take the 2 features into consideration at the same time, hence have low power to detect the true treatment effect. In this paper, we proposed a piece‐wise proportional hazards cure rate model with a random delay time to fit data, and a new weighted log‐rank test to detect the treatment effect of an immunotherapy over a chemotherapy control. We showed that the proposed weight was nearly optimal under mild conditions. Our simulation study showed a substantial gain of power in the proposed test over the existing tests and robustness of the test with misspecified weight. We also introduced a sample size calculation formula to design the immunotherapy clinical trials using the proposed weighted log‐rank test.     

Estimating the variance of estimated trends in proportions when there is no unique subject identifier

Summary.  Longitudinal population-based surveys are widely used in the health sciences to study patterns of change over time. In many of these data sets unique patient identifiers are not publicly available, making it impossible to link the repeated measures from the same individual directly. This poses a statistical challenge for making inferences about time trends because repeated measures from the same individual are likely to be positively correlated, i.e., although the time trend that is estimated under the naïve assumption of independence is unbiased, an unbiased estimate of the variance cannot be obtained without knowledge of the subject identifiers linking repeated measures over time. We propose a simple method for obtaining a conservative estimate of variability for making inferences about trends in proportions over time, ensuring that the type I error is no greater than the specified level. The method proposed is illustrated by using longitudinal data on diabetes hospitalization proportions in South Carolina.     

A New Lens on High School Dropout: Use of Correspondence Analysis and the Statewide Longitudinal Data System

The combination of log-linear models and correspondence analysis have long been used to decompose contingency tables and aid in their interpretation. Until now, this approach has not been applied to the education Statewide Longitudinal Data System (SLDS), which contains administrative school data at the student level. While some research has been conducted using the SLDS, its primary use is for state education administrative reporting. This article uses the combination of log-linear models and correspondence analysis to gain insight into high school dropouts in two discrete regions in Kentucky, Appalachia and non-Appalachia, defined by the American Community Survey. The individual student records from the SLDS were categorized into one of the two regions and a log-linear model was used to identify the interactions between the demographic characteristics and the dropout categories, push-out and pull-out. Correspondence analysis was then used to visualize the interactions with the expanded push-out categories, boredom, course selection, expulsion, failing grade, teacher conflict, and pull-out categories, employment, family problems, illness, marriage, and pregnancy to provide insights into the regional differences. In this article, we demonstrate that correspondence analysis can extend the insights gained from SDLS data and provide new perspectives on dropouts. Supplementary materials for this article are available online.     

A Class of Multiplicative Estimators of Laspeyres Price Indexes

Publication of indexes measuring changes in prices of retail, wholesale, export, and import items is an important part of many governmental statistics programs. One form of price index that is often used is the fixed-base Laspeyres, in which a fixed market basket of goods is priced over time. This article introduces a new class of multiplicative estimators of Laspeyres indexes. The optimum within the class is derived for long-term price change and compared with two other members of the class when used for estimating both long-term and short-term change. Theoretical properties are derived under a model in which long-term relative price changes for individual items have common within-stratum means and are correlated over time. Theory for long-term and short-term change estimators is tested in a simulation study in which a large number of stratified probability samples is selected from a population extracted from items priced for the U.S. consumer price index.     

Multivariate forests with missing mixed outcomes

In this article, we propose a multivariate random forest method for multiple responses of mixed types with missing responses. Imputation is performed for each bootstrap sample used to build the individual trees that form the forest. The individual trees are built using a weighted splitting rule allowing downweighting of imputed observations. A simulation study shows the benefits of this approach over complete case analysis when missing responses are missing completely at random and missing at random (MAR). In particular, the gain in prediction accuracy of the proposed method is larger in the MAR case and also increases as the proportion of missing increases.     

A progressive rise in stomach cancer-related mortality rate during 1970–1995 in Japanese individuals over 85 years of age

A large number of studies have shown a gradual fall in stomach cancer-related mortality rate during the last decade. Here we analyzed the pattern of stomach cancer-related mortality rates in Japanese aged>85 years from 1970 to 1995. We used data for the entire population of Japan. The magnitude of change was measured by relative risk and cause-elimination life tables to distinguish time trends in mortality rates of stomach cancer for individuals over 85 years of age compared with other age groups (55–84 years). In the over-85 age group, stomach cancer mortality increased from 374 in 1970 to 662 in 1995 per 100,000 (77%) for males and from 232 to 296 per 100,000 (27%) for females. Using the 55–59 years group as the reference category, the relative risk increased from 2.3 to 9.9 and from 2.8 to 11.1 in men and women, respectively. The effects of mortality on life expectancy also increased 1.5 times and 1.1 times, respectively. Our results showed a rise of stomach cancer mortality in Japanese aged over 85 years, which paralleled the increase in relative risk and negative contribution to life expectancy. While the mortality of younger age groups is decreasing, the change over from increase to decrease in the over-85 age group is only just beginning.     

A robust logistic discrimination model

Logistic discrimination is a well documented method for classifying observations to two or more groups. However, estimation of the discriminant rule can be seriously affected by outliers. To overcome this, Cox and Ferry produced a robust logistic discrimination technique. Although their method worked in practice, parameter estimation was sometimes prone to convergence problems. This paper proposes a simplified robust logistic model which does not have any such problems and which takes a generalized linear model form. Misclassification rates calculated in a simulation exercise are used to compare the new method with ordinary logistic discrimination. Model diagnostics are also presented. The newly proposed model is then used on data collected from pregnant women at two district general hospitals. A robust logistic discriminant is calculated which can be used to predict accurately which method of feeding a woman will eventually use: breast feeding or bottle feeding.     

Age at first marriage in Malawi: a Bayesian multilevel analysis using a discrete time-to-event model

Summary.  The paper presents a hierarchical discrete time survival model for the analysis of the 2000 Malawi Demographic and Health Survey data to assess the determinants of transition to marriage among women in Malawi. The model explicitly accounts for the unobserved heterogeneity by using family and community random effects with cross-level correlation structure. A nonparametric technique is used to model the base-line discrete hazard dynamically. Parameters of the model are computed by using a Markov chain Monte Carlo algorithm. The results show that rising age at marriage is a combination of birth cohort and education effects, depends considerably on the family and to some extent on the community in which a woman resides and the correlation between family and community random effects is negative. These results confirm a downward trend in teenage marriage and that raising women's education levels in sub-Saharan Africa has the beneficial effect of increasing age at marriage, and by implication reducing total fertility rates. The negative correlation between family and community random effects has policy implications in that targeting communities with an intervention to increase age at first marriage may not necessarily yield reduced fertility levels in individual families. A campaign that is geared towards individual families would achieve the desired goals. Overall, the findings point to the need for the Government in Malawi to enact public policies which are geared at vastly improving women's education at higher levels. The variation in marriage rates over families poses problems in delivering the policy, since particular policies must be devised for specific groups of families to accomplish the required social and health objectives.     

Bayesian experimental design for a toxicokinetic-toxicodynamic model

The aim of this study is to apply the Bayesian method of identifying optimal experimental designs to a toxicokinetic-toxicodynamic model that describes the response of aquatic organisms to time dependent concentrations of toxicants. As for experimental designs, we restrict ourselves to pulses and constant concentrations. A design of an experiment is called optimal within this set of designs if it maximizes the expected gain of knowledge about the parameters. Focus is on parameters that are associated with the auxiliary damage variable of the model that can only be inferred indirectly from survival time series data. Gain of knowledge through an experiment is quantified both with the ratio of posterior to prior variances of individual parameters and with the entropy of the posterior distribution relative to the prior on the whole parameter space. The numerical methods developed to calculate expected gain of knowledge are expected to be useful beyond this case study, in particular for multinomially distributed data such as survival time series data.     

Estimating the underlying change in unemployment in the UK

By setting up a suitable time series model in state space form, the latest estimate of the underlying current change in a series may be computed by the Kalman filter. This may be done even if the observations are only available in a time-aggregated form subject to survey sampling error. A related series, possibly observed more frequently, may be used to improve the estimate of change further. The paper applies these techniques to the important problem of estimating the underlying monthly change in unemployment in the UK measured according to the definition of the International Labour Organisation by the Labour Force Survey. The fitted models suggest a reduction in root-mean-squared error of around 10% over a simple estimate based on differences if a univariate model is used and a further reduction of 50% if information on claimant counts is taken into account. With seasonally unadjusted data, the bivariate model offers a gain of roughly 40% over the use of annual differences. For both adjusted and unadjusted data, there is a further gain of around 10% if the next month's figure on claimant counts is used. The method preferred is based on a bivariate model with unadjusted data. If the next month's claimant count is known, the root-mean-squared error for the estimate of change is just over 10000.     

Marriage or cohabitation: a competing risks analysis of first-partnership formation among the 1958 British birth cohort

A discrete time competing risks hazards model is used to analyse entry into first partnership among men and women born in Britain in 1958. Using a life-course approach we identify family background and current life experiences which affect the timing and type of first-partnership formation. Education is a key factor influencing the age of entry into first partnership and whether or not the respondent will experience pregnancy before forming the partnership. Religiosity, experience of parental separation and the geographical region of residence are more important in affecting the decision to cohabit rather than to marry directly. The analyses highlight the importance of transitions in other domains such as leaving the parental home in encouraging cohabitation.     

Large sample properties and confidence bands for component-wise varying-coefficient regression with longitudinal dependent variable

Longitudinal studies with repeatedly measured dependent variable (out-come) and time-invariant covariates are common in biomedical and epidemi-ological studies. A useful statistical tool to evaluate the effects of covariates on the outcome variable over time is the varying-coefficient regression, which considers a linear relationship between the covariates and the outcome at a specific time point but assumes the linear coefficients to be smooth curves over time. In order to provide adequate smoothing for each coefficient curve, Wu and Chiang ( 1999 ) proposed a class of component-wise kernel estimators and determined the large sample convergence rates and some of the constant terms of the mean squared errors of their estimators. In this paper we calcu¬late the explicit large sample mean squared errors, including the convergence rates and ail the constant terms, and the asymptotic distributions of the kernel estimators of Wu and Chiang ( 1999 ). These asymptotic distributions are used to construct point-wise confidence intervals and Bonferroni-type confidence bands for the coefficient curves. Through a Monte Carlo simulation, wre show that our confidence regions have adequate coverage probabilities. Applying our procedures to a NIH fetal growth study, we show that our procedures are useful to determine the effects of maternal height, cigarette smoking and al¬cohol consumption on the growth of fetal abdominal circumference over time during pregnancy.     

From waterfall plots to spaghetti plots in early oncology clinical development

Waterfall plots are used to describe changes in tumor size observed in clinical studies. They are frequently used to illustrate the overall drug response in oncology clinical trials because of its simple representation of results. Unfortunately, this visual display suffers a number of limitations including (1) potential misguidance by masking the time dynamics of tumor size, (2) ambiguous labelling of the y‐axis, and (3) low data‐to‐ink ratio. We offer some alternatives to address these shortcomings and recommend moving away from waterfall plots to the benefit of plots showing the individual time profiles of sum of lesion diameters (according to RECIST). The spider plot presents the individual changes in tumor measurements over time relative to baseline tumor burden. Baseline tumor size is a well‐known confounding factor of drug effect which has to be accounted for when analyzing data in early clinical trials. While spider plots are conveniently correct for baseline tumor size, they cannot be presented in isolation. Indeed, percentage change from baseline has suboptimal statistical properties (including skewed distribution) and can be overly optimistic in favor of drug efficacy. We argued that plots of raw data (referred to as spaghetti plots) should always accompany spider plots to provide an equipoised illustration of the drug effect on lesion diameters.