The Use of Time to Pregnancy for Estimating and Monitoring Human Fecundity From Demographic and Health Surveys.

BACKGROUND: Available studies on the prevalence of infertility have proved to have certain limitations, with a scarcity of population-based studies and inconsistent reporting from surveys in countries at all income levels. We wanted to test the applicability of the current duration approach to data from the important Demographic and Health Surveys (DHS) program, funded by USAID since its inception in 1985, https://dhsprogram.com/. METHODS: The current duration approach assumes that there is a well-defined time of initiation of attempts to get pregnant and defines the current duration of a still ongoing pregnancy attempt as the time interval from initiation to interview. The DHS interviews do not have an explicit question about initiation. We focused on nullipari and substituted date of "establishment of relationship with current partner" for initiation. Our study used the current duration approach on 15 datasets from DHS during 2002-2016 in eight different countries from sub-Saharan Africa, Asia, and Latin America. RESULTS: Well-established statistical techniques for current duration data yielded results that for some countries postulated surprisingly long median times to pregnancy and surprisingly high estimates of infertility prevalence. Further study of the data structures revealed serious deviations from expected patterns, in contrast to our earlier experience from surveys in France and the United States where participants were asked explicitly about time of initiation of attempts to become pregnant. CONCLUSIONS: Using cohabitation as a proxy for the initiation of attempts to get pregnant is too crude. Using the current duration approach with DHS data will require more explicit questions during the DHS interviews about initiation of pregnancy attempt.

Cohort profile: the Social Inequality in Cancer (SIC) cohort study.

The Social Inequality in Cancer (SIC) cohort study was established to determine pathways through which socioeconomic position affects morbidity and mortality, in particular common subtypes of cancer. Data from seven well-established cohort studies from Denmark were pooled. Combining these cohorts provided a unique opportunity to generate a large study population with long follow-up and sufficient statistical power to develop and apply new methods for quantification of the two basic mechanisms underlying social inequalities in cancer-mediation and interaction. The SIC cohort included 83 006 participants aged 20-98 years at baseline. A wide range of behavioural and biological risk factors such as smoking, physical inactivity, alcohol intake, hormone replacement therapy, body mass index, blood pressure and serum cholesterol were assessed by self-administered questionnaires, physical examinations and blood samples. All participants were followed up in nationwide demographic and healthcare registries. For those interested in collaboration, further details can be obtained by contacting the Steering Committee at the Department of Public Health, University of Copenhagen, at inan@sund.ku.dk.

Nonparametric estimation in an "illness-death" model when all transition times are interval censored.

We develop nonparametric maximum likelihood estimation for the parameters of an irreversible Markov chain on states {0,1,2} from the observations with interval censored times of 0 → 1, 0 → 2 and 1 → 2 transitions. The distinguishing aspect of the data is that, in addition to all transition times being interval censored, the times of two events (0 → 1 and 1 → 2 transitions) can be censored into the same interval. This development was motivated by a common data structure in oral health research, here specifically illustrated by the data from a prospective cohort study on the longevity of dental veneers. Using the self-consistency algorithm we obtain the maximum likelihood estimators of the cumulative incidences of the times to events 1 and 2 and of the intensity of the 1 → 2 transition. This work generalizes previous results on the estimation in an "illness-death" model from interval censored observations.

Interpretability and importance of functionals in competing risks and multistate models.

The basic parameters in both survival analysis and more general multistate models, including the competing risks model and the illness-death model, are the transition hazards. It is often necessary to supplement the analysis of such models with other model parameters, which are all functionals of the transition hazards. Unfortunately, not all such functionals are equally meaningful in practical contexts, even though they may be mathematically well defined. We have found it useful to check whether the functionals satisfy three simple principles, which may be used as criteria for practical interpretability.

Product-limit estimators of the gap time distribution of a renewal process under different sampling patterns.

Nonparametric estimation of the gap time distribution in a simple renewal process may be considered a problem in survival analysis under particular sampling frames corresponding to how the renewal process is observed. This note describes several such situations where simple product limit estimators, though inefficient, may still be useful.

Estimating the effect of current, previous and never use of drugs in studies based on prescription registries.

PURPOSE: Many studies which investigate the effect of drugs categorize the exposure variable into never, current, and previous use of the study drug. When prescription registries are used to make this categorization, the exposure variable possibly gets misclassified since the registries do not carry any information on the time of discontinuation of treatment.In this study, we investigated the amount of misclassification of exposure (never, current, previous use) to hormone therapy (HT) when the exposure variable was based on prescription data. Furthermore, we evaluated the significance of this misclassification for analysing the risk of breast cancer. MATERIALS AND METHODS: Prescription data were obtained from Danish Registry of Medicinal Products Statistics and we applied various methods to approximate treatment episodes. We analysed the duration of HT episodes to study the ability to identify discontinuation of therapy from prescription data. Furthermore, we compared to results based on self-reported duration of HT from the Danish Nurse Cohort.Finally, we analysed the effect of HT exposure on time to breast cancer for the different prescription based exposure variables as well as for self-reported HT use. RESULTS: The results of time to discontinuation varied strongly across the different HT assessments. However, misclassification of HT exposure at baseline was limited and hence analysis of the effect of HT on time to breast cancer showed stability across the different exposure assessments with Hazard Ratios ranging from 1.68 to 1.78 for current use compared to never use. CONCLUSIONS: The findings suggest that it is possible to estimate the effect of never, current and previous use of HT on breast cancer using prescription data.

Effects of sample attrition in a longitudinal study of the association between alcohol intake and all-cause mortality.

BACKGROUND AND AIMS: Longitudinal studies show higher mortality among abstainers and heavy drinkers than among light and moderate alcohol consumers. The influence on this association of missing information on alcohol intake due to attrition (dropout) has not been examined previously. The aims of this study were to characterize participants who dropped out and to evaluate whether the missing information influenced the association between alcohol intake and all-cause mortality. DESIGN AND PARTICIPANTS: Data on the 18 974 participants in the Copenhagen City Heart Study, with four measures of alcohol intake and other life-style factors during 28 years of follow-up, were linked with nation-wide registers on socio-economic covariates, mortality and disease incidence. Logistic regression was used to describe life-style and socio-economic determinants of attrition, and Poisson regression was used to evaluate how attrition affected the association between alcohol intake and mortality. The statistical methods used for dealing with missing values were complete case analysis, carry last observation forward, simple imputations, multiple imputation and weighting. FINDINGS: Abstinence and high alcohol intake, current smoking, physical inactivity and high body mass index increased the odds of dropping out, whereas being married, more years of education, skilled occupation, high income and large residential area decreased the odds. Attrition was associated with increased mortality and incidence rates of heart disease, lung and upper digestive tract cancers and alcoholic liver diseases. Increased mortality among abstainers and heavy drinkers was observed with all methods used for handling missing data on alcohol intake. CONCLUSIONS: Attrition was non-random, and the observed association between alcohol intake and all-cause mortality did not differ by statistical method for handling missing data.

Confounder selection in environmental epidemiology: assessment of health effects of prenatal mercury exposure.

PURPOSE: The purpose of the study is to compare different approaches to the identification of confounders needed for analyzing observational data. Whereas standard analysis usually is conducted as if the confounders were known a priori, selection uncertainty also must be taken into account. METHODS: Confounders were selected by using backward elimination (BE), change in estimate (CIE) method, Akaike information criterion, Bayesian information criterion (BIC), and an empirical approach using a priori information. A modified ridge regression estimator, which shrinks effects of confounders toward zero, also was considered. For each criterion, uncertainty in the estimated exposure effect was assessed by using bootstrap simulations for which confounders were selected in each sample. These methods were illustrated by using data for mercury neurotoxicity in Faroe Islands children. Point estimates and standard errors of mercury effects on confounder-sensitive neurobehavioral outcomes were calculated for each selection procedure. RESULTS: The full model and the empirical a priori model showed approximately the same precision, and these methods were (slightly) inferior to only modified ridge regression. Lower precisions were obtained by using BE with a low cutoff level, BIC, and CIE. CONCLUSIONS: Standard analysis ignores model selection uncertainty and is likely to yield overoptimistic inferences. Thus, the traditional BE procedure with p = 5% should be avoided. If data-dependent procedures are required for confounder identification, we recommend that inferences be based on bootstrap statistics to describe the selection process.

Emergency room utilization in Copenhagen: a comparison of immigrant groups and Danish-born residents.

BACKGROUND: The aim of the study was to investigate whether utilization of the emergency room differed between immigrant groups and Danish-born residents in Copenhagen, Denmark. METHODS: The authors compared the number of emergency room contacts during 1997 among different ethnic groups in the study population. Data were provided by the Statistical Office of the Municipality of Copenhagen. The study population consisted of 183,478 citizens residing in the catchment area of Bispebjerg Hospital in Copenhagen on 1 January 1998. "Contacts" included 22,026 visits made to the emergency room at Bispebjerg Hospital during 1997. Both the study population and "contacts" were characterized by gender, age, income, and country of birth. The immigrants comprised nine ethnic groups according to country of birth. Data were analysed by Poisson regression comparing rate ratios. RESULTS: Persons born in Somalia, Turkey, and ex-Yugoslavia had higher utilization rates of emergency room than Danish-born residents. All other non-Western-born residents had utilization rates similar to Danish-born residents. Persons born in other Western and European countries showed a tendency towards less utilization. CONCLUSION: Higher utilization rates among some immigrant groups may be explained by disparities in health or lack of knowledge about the Danish healthcare system as well as barriers to seeking primary care including language, fear of discrimination, and low satisfaction with primary care. The challenge remains to identify these causal relations, and to find out why utilization patterns vary between immigrant groups.

Effect of ethnic background on Danish hospital utilisation patterns.

The aim of the study is to examine possible ethnic differences in the utilisation patterns of hospitalised immigrants versus patients born in Denmark. Data were obtained from the Register of Prevention at Statistics Denmark. This register includes both clinical and socio-demographic data. All patients discharged as inpatients during 1997 at Bispebjerg Hospital (a major hospital in Copenhagen) were identified through the Register of Prevention and linked to data concerning diagnosis, place of birth, age and gender. To compare immigrants with patients born in Denmark, a study group and a reference group were formed. The final study group consisted of all patients characterised by 22 major diagnostic categories and born outside the five Nordic countries (altogether 858 persons accounting for 976 inpatient contacts). The reference group consisted of 2004 patients accounting for 2432 inpatient contacts characterised by the same diagnostic categories among a random sample of 10,000 patients born in Denmark. The measure of utilisation employed was length of inpatient stay determined by the total number of days that each admission lasted. Data were analysed by a multiple regression analysis controlling for age, gender, diagnosis and place of birth. The results show that for some diagnostic groups, native Danes have longer inpatient stay compared to immigrants, whereas for other diagnostic groups immigrants have longer inpatient stay than native Danes. There was no overall effect of ethnicity on duration of hospital stay and consequently the utilisation patterns of inpatient care seem to reflect equal care for equal needs.

Multi-state models for event history analysis.

An introduction to event history analysis via multi-state models in 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.