Flexible modeling of longitudinal health-related quality of life data accounting for informative dropout in a cancer clinical trial

PURPOSE

A joint modeling approach is recommended for analysis of longitudinal health-related quality of life (HRQoL) data in the presence of potentially informative dropouts. However, the linear mixed model modeling the longitudinal HRQoL outcome in a joint model often assumes a linear trajectory over time, an oversimplification that can lead to incorrect results. Our aim was to demonstrate that a more flexible model gives more reliable and complete results without complicating their interpretation.

METHODS

Five dimensions of HRQoL in patients with esophageal cancer from the randomized clinical trial PRODIGE 5/ACCORD 17 were analyzed. Joint models assuming linear or spline-based HRQoL trajectories were applied and compared in terms of interpretation of results, graphical representation, and goodness of fit.

RESULTS

Spline-based models allowed arm-by-time interaction effects to be highlighted and led to a more precise and consistent representation of the HRQoL over time; this was supported by the martingale residuals and the Akaike information criterion.

CONCLUSION

Linear relationships between continuous outcomes (such as HRQoL scores) and time are usually the default choice. However, the functional form turns out to be important by affecting both the validity of the model and the statistical significance.

TRIAL REGISTRATION

This study is registered with ClinicalTrials.gov, number NCT00861094.

Modelling the disappearance of coarse woody debris, following a land clearing event

BACKGROUND

Land clearing generates coarse woody debris (CWD), much of which ultimately becomes atmospheric CO₂. Schemes for greenhouse gas accounting must consider the contribution from land clearing, but the timing of the contribution will have large uncertainty, due to a paucity of knowledge about the rate of CWD disappearance. To better understand above-ground CWD disappearance following a land clearing event-through the actions of microorganisms, invertebrates, wildfire, or deliberate burning-we combined statistical modelling with an archive of semi-quantitative observations (units of CWD %), made within Queensland, Australia.

RESULTS

Using a generalised additive mixed-effects model (median absolute error = 14.7%), we found that CWD disappearance was strongly influenced by the: (i) number of years elapsed since clearing; (ii) clearing method; (iii) bioregion (effectively a climate-by-tree species interaction); and (iv) the number of times burned. Years-since-clearing had a strongly non-linear effect on the rate of CWD disappearance. The data suggested that disappearance was reverse-sigmoidal, with little change in CWD apparent for the first three years after clearing. In typical conditions for Queensland, the model predicted that it will take 38 years for 95% of CWD to disappear, following a land clearing event; however, accounting for uncertainty in the data and model, this value could be as few as 5 years, or > 100 years. In contrast, due to an assumption about the propensity of land managers to burn CWD, the official method used to assess Australia's greenhouse gas emissions predicted that 95% of CWD will disappear in < 1 year.

CONCLUSIONS

In Queensland, the CWD generated by land clearing typically takes 38 years to disappear. This ultimately implies that a key assumption of Australia's official greenhouse gas reporting-i.e. that 98% of CWD is burned soon after a clearing event-does not adequately account for delayed CO₂ emissions.

QT interval measurement in ventricular pacing: Implications for assessment of drug effects and pro-arrhythmia risk

QT interval prolongation is a known risk factor for development of malignant ventricular arrhythmias. Measurement of the QT interval is difficult in the setting of ventricular pacing (VP), which can prolong depolarization and increase the QT interval, overestimating repolarization time. VP and cardiac resynchronization therapies have become commonplace in modern cardiac care and may contribute to repolarization heterogeneity and subsequent increased risk for ventricular arrhythmias including Torsades de Pointes. It is imperative for the clinician caring for acutely ill cardiac patients to understand the relationship between QT interval prolongation, both drug-induced and pacing-induced, and repolarization changes with subsequent ventricular arrhythmia risk. In this review, we discuss the components of QT interval assessment for arrhythmogenic risk including arrhythmogenic QT prolongation, methods for adjusting the QT interval to identify repolarization changes, methods to adjust for heart rate, and propose a framework for medication management to assess for drug-induced long QT syndrome in patients with VP.

Associations between perfluoroalkyl substances and lipid profile in a highly exposed young adult population in the Veneto Region

BACKGROUND

Residents of a large area of the Veneto Region (North-Eastern Italy) were exposed for decades to drinking water contaminated by perfluoroalkyl substances (PFAS). PFAS have been consistently associated with raised serum lipids, mainly in cross-sectional studies and in background exposure contexts, but the shape of the dose-response relationships has been poorly investigated. The objectives of our study were to evaluate the association between serum PFAS and serum lipids and their dose-response patterns in a large exposed population.

METHODS

A cross-sectional study was conducted in 16,224 individuals aged 20-39 years recruited in the regional health surveillance program. 15,720 subjects were analysed after excluding pregnant women (n = 327), participants reporting use of cholesterol lowering medications (n = 67) or with missing information on the selected covariates (n = 110). Twelve PFAS were measured by HPLC-MS in serum; three (PFOA, PFOS and PFHxS) were quantifiable in at least 50% of samples. Non-fasting serum total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C) and triglycerides were measured by enzymatic assays in automated analysers and low-density lipoprotein cholesterol (LDL-C), non-HDL cholesterol and total/HDL cholesterol ratio were calculated. The associations between natural log (ln) transformed PFAS and lipids were assessed through generalized additive models using linear regression and smoothing thin plate splines, adjusted for potential confounders.

RESULTS

There were strong positive associations between the ln-transformed PFOA, PFOS, and PFHxS and TC, HDL-C, and LDL-C, and between ln PFOA and PFHxS and triglycerides. Each ln-increase in PFOA was associated with an increase of 1.94 mg/dL (95% CI 1.48-2.41) in TC, with 4.99 mg/dL (CI 4.12-5.86) for PFOS and 2.02 mg/dL (CI 1.45-2.58) for PFHxS.

CONCLUSIONS

Investigation of the shape of exposure-response associations using splines showed a positive association with the largest increases per unit of PFAS in cholesterol levels occurring at the lower range of PFAS concentrations for each compound.

Piecewise latent growth models: beyond modeling linear-linear processes

Piecewise latent growth models (LGMs) for linear-linear processes have been well-documented and studied in recent years. However, in the latent growth modeling literature, advancements to other functional forms as well as to multiple changepoints or knots have been nearly non-existent. This manuscript deals with three extensions. The first is to a piecewise latent growth model incorporating higher-order polynomials. The second is to extend the basic framework to three phases. The last extension is to inherently nonlinear functions. In these extensions, the changepoint(s) is a parameter to be estimated and may be fixed or allowed to vary across subjects as an application warrants. The approaches are developed and two illustrative empirical examples from psychology are used to highlight the methodological nuances. Annotated statistical software is provided to make these elaborations accessible to practitioners and methodologists.

Lipoprotein(a) plasma levels are not associated with incident microvascular complications in type 2 diabetes mellitus

AIMS/HYPOTHESIS

Microvascular disease in type 2 diabetes is a significant cause of end-stage renal disease, blindness and peripheral neuropathy. The strict control of known risk factors, e.g. lifestyle, hyperglycaemia, hypertension and dyslipidaemia, reduces the incidence of microvascular complications, but a residual risk remains. Lipoprotein (a) [Lp(a)] is a strong risk factor for macrovascular disease in the general population. We hypothesised that plasma Lp(a) levels and the LPA gene SNPs rs10455872 and rs3798220 are associated with the incident development of microvascular complications in type 2 diabetes.

METHODS

Analyses were performed of data from the DiaGene study, a prospective study for complications of type 2 diabetes, collected in the city of Eindhoven, the Netherlands (n = 1886 individuals with type 2 diabetes, mean follow-up time = 6.97 years). To assess the relationship between plasma Lp(a) levels and the LPA SNPs with each newly developed microvascular complication (retinopathy n = 223, nephropathy n = 246, neuropathy n = 236), Cox proportional hazards models were applied and adjusted for risk factors for microvascular complications (age, sex, mean arterial pressure, non-HDL-cholesterol, HDL-cholesterol, BMI, duration of type 2 diabetes, HbA_1c and smoking).

RESULTS

No significant associations of Lp(a) plasma levels and the LPA SNPs rs10455872 and rs3798220 with prevalent or incident microvascular complications in type 2 diabetes were found. In line with previous observations the LPA SNPs rs10455872 and rs3798220 did influence the plasma Lp(a) levels.

CONCLUSIONS/INTERPRETATION

Our data show no association between Lp(a) plasma levels and the LPA SNPs with known effect on Lp(a) plasma levels with the development of microvascular complications in type 2 diabetes. This indicates that Lp(a) does not play a major role in the development of microvascular complications. However, larger studies are needed to exclude minimal effects of Lp(a) on the development of microvascular complications.

Analysis of variation in growth and spline-based growth models for Marecha and Lassi dromedary camels

Camel is an important domestic animal that is well adapted to extremely harsh environments. Due to its multi-purpose role, the camel is gaining importance, particularly as a meat producer. Body weights at different ages and growth curves have been extensively studied in several livestock species. However, the information on the camel is very limited. We investigate patterns of variation in birth weight and weaning weight of two prominent breeds of camels in Pakistan, namely Marecha and Lassi. A further objective is not only to develop growth models which will account for systematic differences due to breed and sex but also to allow for individual variation in growth, by the application of appropriate statistical models. Birth weight, weaning, and monthly weights for a period of up to 48 months were recorded for each camel. Growth records of 137 camels consisting of 108 Marecha and 29 Lassi were obtained. Differences were detected between the two breeds, Marecha and Lassi, as well as differences between sexes, but only after 2 years of age. Beyond that, female Marecha camels had slower growth than males, whereas Lassi (mostly females) had faster growth than female Marechas. In addition, there is a sizable variation between individuals within the same breed-sex combination. The results of the present study will be of use to explore the potential of the camel as a meat producer in Pakistan.

Dietary cadmium intake and risk of cutaneous melanoma: An Italian population-based case-control study

INTRODUCTION

Exposure to the heavy metal cadmium has been associated with many adverse health effects, such as atherosclerosis, diabetes, and cancer, possibly melanoma. In non-occupationally exposed individuals, food intake is a major source of cadmium exposure, after smoking. We aimed to assess the risk of melanoma in relation to dietary cadmium intake.

METHODS

Using a population-based case-control study design, we recruited 380 incident cases of newly-diagnosed cutaneous melanoma and 719 matched controls in the Emilia-Romagna Region, Northern Italy in the years 2005-2006. We evaluated dietary intake using a semi-quantitative food frequency questionnaire. We used conditional logistic regression to compute odds ratios (ORs) and 95% confidence intervals (CIs) for melanoma according to quintiles of dietary cadmium intake, adjusting for several potential confounders, and we modeled the association non-parametrically, using restricted cubic splines.

RESULTS

Median energy-adjusted intake of cadmium was 6.11 μg/day (interquartile range 5.38-6.91) among cases and 5.97 μg/day (5.15-6.79) among controls. For each 1 μg/day-increase in cadmium intake, the OR for melanoma was 1.11 (95% CI 1.00-1.24). Melanoma risk generally increased with increasing quintile of cadmium exposure, with ORs of 1.55 (95% CI 0.99-2.42), 1.54 (95% CI 0.99-2-40), 1.75 (95% CI 1.12-2.75), and 1.65 (95% CI 1.05-2.61) for the second through fifth quintiles, compared with the lowest quintile. Sex-stratified analysis showed ORs per 1 μg/day-increase in cadmium intake of 1.10 (95% CI 0.93-1-29) among men and 1.15 (95% CI 0.99-1.33) among women. Using spline regression analysis, we observed a generally linear increase in melanoma risk up to 6 μg/day of cadmium intake, after which the risk appeared to plateau.

CONCLUSIONS

We observed a positive non-linear association between dietary cadmium intake and risk of cutaneous melanoma in a Northern Italy population. However, further studies are needed to elucidate this association, due to concerns about exposure misclassification, unmeasured confounding, and the limited and conflicting evidence from epidemiological findings.

Zero-inflated spatio-temporal models for disease mapping

In this paper, our aim is to analyze geographical and temporal variability of disease incidence when spatio-temporal count data have excess zeros. To that end, we consider random effects in zero-inflated Poisson models to investigate geographical and temporal patterns of disease incidence. Spatio-temporal models that employ conditionally autoregressive smoothing across the spatial dimension and B-spline smoothing over the temporal dimension are proposed. The analysis of these complex models is computationally difficult from the frequentist perspective. On the other hand, the advent of the Markov chain Monte Carlo algorithm has made the Bayesian analysis of complex models computationally convenient. Recently developed data cloning method provides a frequentist approach to mixed models that is also computationally convenient. We propose to use data cloning, which yields to maximum likelihood estimation, to conduct frequentist analysis of zero-inflated spatio-temporal modeling of disease incidence. One of the advantages of the data cloning approach is that the prediction and corresponding standard errors (or prediction intervals) of smoothing disease incidence over space and time is easily obtained. We illustrate our approach using a real dataset of monthly children asthma visits to hospital in the province of Manitoba, Canada, during the period April 2006 to March 2010. Performance of our approach is also evaluated through a simulation study.

A semi-parametric nonlinear model for event-related fMRI

Nonlinearity in evoked hemodynamic responses often presents in event-related fMRI studies. Volterra series, a higher-order extension of linear convolution, has been used in the literature to construct a nonlinear characterization of hemodynamic responses. Estimation of the Volterra kernel coefficients in these models is usually challenging due to the large number of parameters. We propose a new semi-parametric model based on Volterra series for the hemodynamic responses that greatly reduces the number of parameters and enables "information borrowing" among subjects. This model assumes that in the same brain region and under the same stimulus, the hemodynamic responses across subjects share a common but unknown functional shape that can differ in magnitude, latency and degree of interaction. We develop a computationally-efficient strategy based on splines to estimate the model parameters, and a hypothesis test on nonlinearity. The proposed method is compared with several existing methods via extensive simulations, and is applied to a real event-related fMRI study.

Summary of dose-response modeling for developmental toxicity studies

Developmental toxicity studies are an important area in the field of toxicology. Endpoints measured on fetuses include weight and indicators of death and malformation. Binary indicator measures are typically summed over the litter and a discrete distribution is assumed to model the number of adversely affected fetuses. Additionally, there is noticeable variation in the litter responses within dose groups that should be taken into account when modeling. Finally, the dose-response pattern in these studies exhibits a threshold effect. The threshold dose-response model is the default model for non-carcinogenic risk assessment, according to the USEPA, and is encouraged by the agency for the use in the risk assessment process. Two statistical models are proposed to estimate dose-response pattern of data from the developmental toxicity study: the threshold model and the spline model. The models were applied to two data sets. The advantages and disadvantages of these models, potential other models, and future research possibilities will be summarized.