German Rodriguez

An Assessment of Estimation Procedures for Multilevel Models with Binary Responses

We evaluate two software packages that are available for fitting multilevel models to binary response data, namely VARCL and ML3, by using a Monte Carlo study designed to represent quite closely the actual structure of a data set used in an analysis of health care utilization in Guatemala. We find that the estimates of fixed effects and variance components produced by the software packages are subject to very substantial downward bias when the random effects are sufficiently large to be interesting. In fact, the fixed effect estimates are no better than the estimates obtained by using standard logit models that ignore the hierarchical structure of the data. The estimates of standard errors appear to be reasonably accurate and superior to those obtained by ignoring clustering, although one might question their utility in the presence of large biases. We conclude that alternative estimation procedures need to be developed and implemented for the binary response case.

New estimates of the effectiveness of the Yuzpe regimen of emergency contraception

The purpose of this study was to provide new estimates of the effectiveness of the Yuzpe method of emergency contraception and to offer correctly computed estimates of the confidence intervals for estimated effectiveness rates. Through a literature search, seven studies that present the number of women treated and outcome of treatment by cycle day of unprotected intercourse relative to expected day of ovulation were identified. Probabilities of conception by cycle day of intercourse among women not using contraception and the associated variance-covariance matrix from five other datasets were estimated, and these external estimates were used to assess the effectiveness of the Yuzpe regimen. The 40 estimates of effectiveness, based on seven separate studies and the seven studies combined and five different sets of conception probabilities by cycle day, ranged from a low of 44.2% to a high of 88.7%. The preferred point estimate is that emergency contraceptive pills reduce the risk of pregnancy by 75.4%, with a 95% confidence interval extending from 65.6% to 82.4%. True effectiveness is likely to be at least 75% because treatment failures (observed pregnancies) include women who were already pregnant when treated and women who became pregnant after being treated.

Prenatal and delivery care and childhood immunization in guatemala: Do family and community matter?*

In this paper we investigate family choices about pregnancy-related care and the use of childhood immunization. Estimates obtained from a multilevel logistic model indicate that use of formal (or “modern ”) health services differs substantially by ethnicity, by social and economic factors, and by availability of health services. The results also show that family and community membership are very important determinants of the use of health care, even in the presence of controls for a large number of observed characteristics of individuals, families, and communities.

Improved estimation procedures for multilevel models with binary response: a case-study

Summary. During recent years, analysts have been relying on approximate methods of inference to estimate multilevel models for binary or count data. In an earlier study of random-intercept models for binary outcomes we used simulated data to demonstrate that one such approximation, known as marginal quasi-likelihood, leads to a substantial attenuation bias in the estimates of both fixed and random effects whenever the random effects are non-trivial. In this paper, we fit three-level randomintercept models to actual data for two binary outcomes, to assess whether refined approximation procedures, namely penalized quasi-likelihood and second-order improvements to marginal and penalized quasi-likelihood, also underestimate the underlying parameters. The extent of the bias is assessed by two standards of comparison: exact maximum likelihood estimates, based on a Gauss-Hermite numerical quadrature procedure, and a set of Bayesian estimates, obtained from Gibbs sampling with diffuse priors. We also examine the effectiveness of a parametric bootstrap procedure for reducing the bias. The results indicate that second-order penalized quasi-likelihood estimates provide a considerable improvement over the other approximations, but all the methods of approximate inference result in a substantial underestimation of the fixed and random effects when the random effects are sizable. We also find that the parametric bootstrap method can eliminate the bias but is computationally very intensive.

Estimating a Multivariate Proportional Hazards Model for Clustered Data Using the EM Algorithm, with an Application to Child Survival in Guatemala

Abstract This article discusses a random-effects model for the analysis of clustered survival times, such as those reflecting the mortality experience of children in the same family. We describe parametric and nonparametric approaches to the specification of the random effect and show how the model may be fitted using an accelerated EM algorithm. We then fit two specifications of the model to child survival data from Guatemala. These data had been analyzed before using standard hazard models that ignore cluster effects.

Utilization of care during pregnancy in rural Guatemala: does obstetrical need matter?

This study examines factors associated with the use of biomedical care during pregnancy in Guatemala, focusing on the extent to which complications in an ongoing or previous pregnancy affect a womans decisions to seek care. The findings, based on multilevel models, suggest that obstetrical need, as well as demographic, social, and cultural factors, are important predictors of pregnancy care. In contrast, measures of availability and access to health services have modest effects. The results also suggest the importance of unobserved variables--such as quality of care--in explaining womens decisions about pregnancy care. These results imply that improving proximity to biomedical services is unlikely to have a dramatic impact on utilization in the absence of additional changes that improve the quality of care or reduce barriers to access. Moreover, current efforts aimed at incorporating midwives into the formal health-care system may need to extend their focus beyond the modification of midwife practices to consider the provision of culturally appropriate, high-quality services by traditional and biomedical providers alike.

An approach to the analysis of menstrual patterns in the critical evaluation of contraceptives.

An approach to a standardized analysis of menstrual bleeding patterns is presented. The method utilizes a prospective bleeding record and characterizes the menstrual pattern of a woman according to bleeding and nonbleeding events without imposing definitions of cycle or intermenstrual bleeding. Further, the analysis is done within a time reference period similar to life-table analysis, with the experience of each woman having equal value. The method is thus applicable to the study of menstrual patterns of both normal, untreated women, and women on experimental contraceptive regimens in which expected bleeding patterns are disrupted. The methodology is aided by computer programs to process the data and by the use of standard statistical packages for comparison of output data of different groups.

The Mass Media and Family Planning in Kenya

26 have been posed about a wide range of behavior, including the effects of television on violence, on children’s education, on consumers’ choices of products and on voting behavior, among others. The evidence on the effectiveness of such communications is mixed, 1 but the belief that there is some effect has persisted and is reflected in the emergence of a whole industry devoted to such activities. The application of mass communications to influence fertility is a natural extension of the basic idea that the media can both inform and motivate people, even about such complex subjects as their reproductive means and goals. Communications efforts have become increasingly widespread in the developing world as part of international technical assistance and government programs designed to reduce fertility. A 1986 summary 2 of radio coverage alone cited nearly 250 sources describing

Trends and differentials in breastfeeding behaviour: evidence from the WFS and DHS.

In this paper trends and correlates of breastfeeding behaviour are analysed with data from available surveys conducted in conjunction with the World Fertility Surveys (WFS) and the Demographic and Health Surveys (DHS). Both these series of fertility surveys were designed and carried out in similar fashions, and in both, similar information on breastfeeding behaviour and its correlates was collected. Therefore, these data provide the best comparative information currently available. The first objective is to compare current-status information on breastfeeding with retrospective reports on durations of lactation. The extent to which distortions in retrospectively reported ages at weaning produce biases in summary measures (means and quartiles) of breastfeeding durations is examined, and the loss of precision incurred by the use of current-status data is explored. On the basis of these results, the relative merits of the two types of breastfeeding information are discussed. The second objective is to documen...

Frequency of discontinuation of contraceptive use: results from a French population-based cohort

BACKGROUND Despite the widespread use of highly effective contraceptive methods in France, one in every three pregnancies is unintended. Among women experiencing an unintended pregnancy leading to an abortion, half had changed their contraceptive method in the 6 months preceding the abortion, in most cases switching to a less-effective method or to no method at all. This study provides estimates of method-specific contraceptive discontinuation rates for any reason and for method-related reasons among French women. METHODS The data were drawn from the COCON survey (2000-2004), a population-based French prospective cohort, comprising a representative sample of 2863 women aged 18-44. We estimated Kaplan-Meier life-table probabilities of contraceptive discontinuation during the 4 years of follow-up and tested for differences by intrauterine device (IUD) type and pill composition. RESULTS Probabilities of contraceptive discontinuation for method-related reasons varied widely by method: IUDs were associated with the lowest probabilities of discontinuation (11% within 12 months, 30% within 4 years), followed by the pill (22% and 48%, respectively). Discontinuation rates were significantly higher for all other methods (condoms, withdrawal, fertility awareness methods and spermicides). We found no differences in discontinuation rates by the type of IUD (levonorgestrel-IUD versus copper-IUD) and increasing rates of pill discontinuation with decreasing dosage in estrogen. CONCLUSIONS Contraceptive discontinuation rates among French women are substantially lower than those reported for US women. Comparing the determinants of contraceptive discontinuation and the role of healthcare providers in helping women make these changes would improve our understanding of the reasons for such variation.