Chenxi Li

Bone-Density Testing Interval and Transition to Osteoporosis in Older Women

BACKGROUND Although bone mineral density (BMD) testing to screen for osteoporosis (BMD T score, -2.50 or lower) is recommended for women 65 years of age or older, there are few data to guide decisions about the interval between BMD tests. METHODS We studied 4957 women, 67 years of age or older, with normal BMD (T score at the femoral neck and total hip, -1.00 or higher) or osteopenia (T score, -1.01 to -2.49) and with no history of hip or clinical vertebral fracture or of treatment for osteoporosis, followed prospectively for up to 15 years. The BMD testing interval was defined as the estimated time for 10% of women to make the transition to osteoporosis before having a hip or clinical vertebral fracture, with adjustment for estrogen use and clinical risk factors. Transitions from normal BMD and from three subgroups of osteopenia (mild, moderate, and advanced) were analyzed with the use of parametric cumulative incidence models. Incident hip and clinical vertebral fractures and initiation of treatment with bisphosphonates, calcitonin, or raloxifene were treated as competing risks. RESULTS The estimated BMD testing interval was 16.8 years (95% confidence interval [CI], 11.5 to 24.6) for women with normal BMD, 17.3 years (95% CI, 13.9 to 21.5) for women with mild osteopenia, 4.7 years (95% CI, 4.2 to 5.2) for women with moderate osteopenia, and 1.1 years (95% CI, 1.0 to 1.3) for women with advanced osteopenia. CONCLUSIONS Our data indicate that osteoporosis would develop in less than 10% of older, postmenopausal women during rescreening intervals of approximately 15 years for women with normal bone density or mild osteopenia, 5 years for women with moderate osteopenia, and 1 year for women with advanced osteopenia. (Funded by the National Institutes of Health.).

Follicle-stimulating hormone is independently associated with lean mass but not BMD in younger postmenopausal women

PURPOSE Increased follicle-stimulating hormone (FSH) has been associated with lower bone mineral density (BMD) in animal models and longitudinal studies of women, but a direct effect has not been demonstrated. METHODS We tested associations between FSH, non-bone body composition measures and BMD in 94 younger (aged 50 to 64 years) postmenopausal women without current use of hormone therapy, adjusting for sex hormone concentrations and clinical risk factors for osteoporosis. Lean mass, fat mass and areal BMD (aBMD) at the spine, femoral neck and total hip were measured using dual energy X-ray absorptiometry (DXA). Volumetric BMD (vBMD) was measured at the distal radius using peripheral quantitative computed tomography (pQCT). RESULTS FSH was inversely correlated with lean and fat mass, bioavailable estradiol, spine and hip aBMD, and vBMD at the ultradistal radius. In the multivariable analysis, FSH was independently associated with lean mass (β=-0.099, p=0.005) after adjustment for age, race, years since menopause, bioavailable estradiol, bioavailable testosterone, LH, PTH, SHBG and urine N-telopeptide. FSH showed no statistically significant association with aBMD at any site or pQCT measures at the distal radius in adjusted models. Race was independently associated with aBMD, and race and urine N-telopeptide were independently associated with bone area and vBMD. CONCLUSIONS After adjustment for hormonal measures and osteoporosis risk factors, higher concentrations of FSH were independently associated with lower lean mass, but not with BMD. Previously reported correlations between FSH and BMD might have been due to indirect associations via lean mass or weight.

Parametric likelihood inference for interval censored competing risks data

Parametric estimation of the cumulative incidence function (CIF) is considered for competing risks data subject to interval censoring. Existing parametric models of the CIF for right censored competing risks data are adapted to the general case of interval censoring. Maximum likelihood estimators for the CIF are considered under the assumed models, extending earlier work on nonparametric estimation. A simple naive likelihood estimator is also considered that utilizes only part of the observed data. The naive estimator enables separate estimation of models for each cause, unlike full maximum likelihood in which all models are fit simultaneously. The naive likelihood is shown to be valid under mixed case interval censoring, but not under an independent inspection process model, in contrast with full maximum likelihood which is valid under both interval censoring models. In simulations, the naive estimator is shown to perform well and yield comparable efficiency to the full likelihood estimator in some settings. The methods are applied to data from a large, recent randomized clinical trial for the prevention of mother-to-child transmission of HIV.

Bent Line Quantile Regression with Application to an Allometric Study of Land Mammals' Speed and Mass

Quantile regression, which models the conditional quantiles of the response variable given covariates, usually assumes a linear model. However, this kind of linearity is often unrealistic in real life. One situation where linear quantile regression is not appropriate is when the response variable is piecewise linear but still continuous in covariates. To analyze such data, we propose a bent line quantile regression model. We derive its parameter estimates, prove that they are asymptotically valid given the existence of a change-point, and discuss several methods for testing the existence of a change-point in bent line quantile regression together with a power comparison by simulation. An example of land mammal maximal running speeds is given to illustrate an application of bent line quantile regression in which this model is theoretically justified and its parameters are of direct biological interests.

Quantile regression with a change-point model for longitudinal data: An application to the study of cognitive changes in preclinical alzheimer's disease

Progressive and insidious cognitive decline that interferes with daily life is the defining characteristic of Alzheimers disease (AD). Epidemiological studies have found that the pathological process of AD begins years before a clinical diagnosis is made and can be highly variable within a given population. Characterizing cognitive decline in the preclinical phase of AD is critical for the development of early intervention strategies when disease-modifying therapies may be most effective. In the last decade, there has been an increased interest in the application of change-point models to longitudinal cognitive outcomes prior to and after diagnosis. Most of the proposed statistical methodology for describing decline relies upon distributional assumptions that may not hold. In this article, we introduce a quantile regression with a change-point model for longitudinal data of cognitive function in persons bound to develop AD. A change-point in our model reflects the transition from the cognitive decline due to normal aging to the accelerated decline due to disease progression. Quantile regression avoids common distributional assumptions on cognitive outcomes and allows the covariate effects and the change-point to vary for different quantiles of the response. We provided an approach for estimating the model parameters, including the change-point, and presented inferential procedures based on the asymptotic properties of the estimators. A simulation study showed that the estimation and inferential procedures perform reasonably well in finite samples. The practical use of our model was illustrated by an application to longitudinal episodic memory outcomes from two cohort studies of aging and AD.

The Fine-Gray model under interval censored competing risks data

We consider semiparametric analysis of competing risks data subject to mixed case interval censoring. The Fine-Gray model (Fine & Gray, 1999) is used to model the cumulative incidence function and is coupled with sieve semiparametric maximum likelihood estimation based on univariate or multivariate likelihood. The univariate likelihood of cause-specific data enables separate estimation of cumulative incidence function for each competing risk, in contrast with the multivariate likelihood of full data which estimates cumulative incidence functions for multiple competing risks jointly. Under both likelihoods and certain regularity conditions, we show that the regression parameter estimator is asymptotically normal and semiparametrically efficient, although the spline-based sieve estimator of the baseline cumulative subdistribution hazard converges at a rate slower than root-n. The proposed method is evaluated by simulation studies regarding its finite sample performance and is illustrated by a competing risk analysis of data from an dementia cohort study.

Measurement and control of bias in patient reported outcomes using multidimensional item response theory

BackgroundPatient-reported outcome (PRO) measures play a key role in the advancement of patient-centered care research. The accuracy of inferences, relevance of predictions, and the true nature of the associations made with PRO data depend on the validity of these measures. Errors inherent to self-report measures can seriously bias the estimation of constructs assessed by the scale. A well-documented disadvantage of self-report measures is their sensitivity to response style (RS) effects such as the respondent’s tendency to select the extremes of a rating scale. Although the biasing effect of extreme responding on constructs measured by self-reported tools has been widely acknowledged and studied across disciplines, little attention has been given to the development and systematic application of methodologies to assess and control for this effect in PRO measures.MethodsWe review the methodological approaches that have been proposed to study extreme RS effects (ERS). We applied a multidimensional item response theory model to simultaneously estimate and correct for the impact of ERS on trait estimation in a PRO instrument. Model estimates were used to study the biasing effects of ERS on sum scores for individuals with the same amount of the targeted trait but different levels of ERS. We evaluated the effect of joint estimation of multiple scales and ERS on trait estimates and demonstrated the biasing effects of ERS on these trait estimates when used as explanatory variables.ResultsA four-dimensional model accounting for ERS bias provided a better fit to the response data. Increasing levels of ERS showed bias in total scores as a function of trait estimates. The effect of ERS was greater when the pattern of extreme responding was the same across multiple scales modeled jointly. The estimated item category intercepts provided evidence of content independent category selection. Uncorrected trait estimates used as explanatory variables in prediction models showed downward bias.ConclusionsA comprehensive evaluation of the psychometric quality and soundness of PRO assessment measures should incorporate the study of ERS as a potential nuisance dimension affecting the accuracy and validity of scores and the impact of PRO data in clinical research and decision making.

Assisted reproductive technology and newborn size in singletons resulting from fresh and cryopreserved embryos transfer

Objectives and Study Design The aim of this study was two-fold: to investigate the association of Assisted Reproductive Technology (ART) and small newborn size, using standardized measures; and to examine within strata of fresh and cryopreserved embryos transfer, whether this association is influenced by parental infertility diagnoses. We used a population-based retrospective cohort from Michigan (2000–2009), Florida and Massachusetts (2000–2010). Our sample included 28,946 ART singletons conceived with non-donor oocytes and 4,263,846 non-ART singletons. Methods Regression models were used to examine the association of ART and newborn size, measured as small for gestational age (SGA) and birth-weight-z-score, among four mutually exclusive infertility groups: female infertility only, male infertility only, combined female and male infertility, and unexplained infertility, stratified by fresh and cryopreserved embryos transfer. Results We found increased SGA odds among ART singletons from fresh embryos transfer compared with non-ART singletons, with little difference by infertility source [adjusted odds-ratio for SGA among female infertility only: 1.18 (95% CI 1.10, 1.26), male infertility only: 1.20 (95% CI 1.10, 1.32), male and female infertility: 1.18 (95% CI 1.06, 1.31) and unexplained infertility: 1.24 (95% CI 1.10, 1.38)]. Conversely, ART singletons, born following cryopreserved embryos transfer, had lower SGA odds compared with non-ART singletons, with mild variation by infertility source [adjusted odds-ratio for SGA among female infertility only: 0.56 (95% CI 0.45, 0.71), male infertility only: 0.64 (95% CI 0.47, 0.86), male and female infertility: 0.52 (95% CI 0.36, 0.77) and unexplained infertility: 0.71 (95% CI 0.47, 1.06)]. Birth-weight-z-score was significantly lower for ART singletons born following fresh embryos transfer than non-ART singletons, regardless of infertility diagnoses.

Safety of lumbar puncture in comatose children with clinical features of cerebral malaria

Objective: We assessed the independent association of lumbar puncture (LP) and death in Malawian children admitted to the hospital with the clinical features of cerebral malaria (CM). Methods: This was a retrospective cohort study in Malawian children with clinical features of CM. Allocation to LP was nonrandom and was associated with severity of illness. Propensity score–based analyses were used to adjust for this bias and assess the independent association between LP and mortality. Results: Data were available for 1,075 children: 866 (80.6%) underwent LP and 209 (19.4%) did not. Unadjusted mortality rates were lower in children who underwent LP (15.3% vs 26.7% in the no-LP group) but differences in covariates between the 2 groups suggested bias in LP allocation. After propensity score matching, all covariates were balanced. Propensity score–based analyses showed no change in mortality rate associated with LP: by inverse probability weighting, the average risk reduction was 2.0% at 12 hours (95% confidence interval −1.5% to 5.5%, p = 0.27) and 1.7% during hospital admission (95% confidence interval −4.5% to 7.9%, p = 0.60). Undergoing LP did not change the risk of mortality in subanalyses of children with severe brain swelling on MRI or in those with papilledema. Conclusion: In comatose children with suspected CM who were clinically stable, we found no evidence that LP increases mortality, even in children with objective signs of raised intracranial pressure.

Two-sample tests for survival data from observational studies

When observational data are used to compare treatment-specific survivals, regular two-sample tests, such as the log-rank test, need to be adjusted for the imbalance between treatments with respect to baseline covariate distributions. Besides, the standard assumption that survival time and censoring time are conditionally independent given the treatment, required for the regular two-sample tests, may not be realistic in observational studies. Moreover, treatment-specific hazards are often non-proportional, resulting in small power for the log-rank test. In this paper, we propose a set of adjusted weighted log-rank tests and their supremum versions by inverse probability of treatment and censoring weighting to compare treatment-specific survivals based on data from observational studies. These tests are proven to be asymptotically correct. Simulation studies show that with realistic sample sizes and censoring rates, the proposed tests have the desired Type I error probabilities and are more powerful than the adjusted log-rank test when the treatment-specific hazards differ in non-proportional ways. A real data example illustrates the practical utility of the new methods.