Simulation models for aggregated data meta-analysis: Evaluation of pooling effect sizes and publication biases.

Simulation studies are commonly used to evaluate the performance of newly developed meta-analysis methods. For methodology that is developed for an aggregated data meta-analysis, researchers often resort to simulation of the aggregated data directly, instead of simulating individual participant data from which the aggregated data would be calculated in reality. Clearly, distributional characteristics of the aggregated data statistics may be derived from distributional assumptions of the underlying individual data, but they are often not made explicit in publications. This article provides the distribution of the aggregated data statistics that were derived from a heteroscedastic mixed effects model for continuous individual data and a procedure for directly simulating the aggregated data statistics. We also compare our simulation approach with other simulation approaches used in literature. We describe their theoretical differences and conduct a simulation study for three meta-analysis methods: DerSimonian and Laird method for pooling aggregated study effect sizes and the Trim & Fill and precision-effect test and precision-effect estimate with standard errors method for adjustment of publication bias. We demonstrate that the choice of simulation model for aggregated data may have an impact on (the conclusions of) the performance of the meta-analysis method. We recommend the use of multiple aggregated data simulation models to investigate the sensitivity in the performance of the meta-analysis method. Additionally, we recommend that researchers try to make the individual participant data model explicit and derive from this model the distributional consequences of the aggregated statistics to help select appropriate aggregated data simulation models.

Bias of the additive hazard model in the presence of causal effect heterogeneity.

Hazard ratios are prone to selection bias, compromising their use as causal estimands. On the other hand, if Aalen's additive hazard model applies, the hazard difference has been shown to remain unaffected by the selection of frailty factors over time. Then, in the absence of confounding, observed hazard differences are equal in expectation to the causal hazard differences. However, in the presence of effect (on the hazard) heterogeneity, the observed hazard difference is also affected by selection of survivors. In this work, we formalize how the observed hazard difference (from a randomized controlled trial) evolves by selecting favourable levels of effect modifiers in the exposed group and thus deviates from the causal effect of interest. Such selection may result in a non-linear integrated hazard difference curve even when the individual causal effects are time-invariant. Therefore, a homogeneous time-varying causal additive effect on the hazard cannot be distinguished from a time-invariant but heterogeneous causal effect. We illustrate this causal issue by studying the effect of chemotherapy on the survival time of patients suffering from carcinoma of the oropharynx using data from a clinical trial. The hazard difference can thus not be used as an appropriate measure of the causal effect without making untestable assumptions.

Mortality in patients with Dupuytren's disease in the first 5 years after diagnosis: a population-based survival analysis.

Previous studies suggest that Dupuytren's disease is associated with increased mortality, but most studies failed to account for important confounders. In this population-based cohort study, general practitioners' (GP) data were linked to Statistics Netherlands to register all-cause and disease-specific mortality. Patients with Dupuytren's disease were identified using the corresponding diagnosis code and assessing free-text fields from GP consultations. Multiple imputations were performed to estimate missing values of covariates, followed by 1:7 propensity score matching to balance cases with controls on confounding factors. A frailty proportional hazard model was used to compare mortality between both groups. Out of 209,966 individuals, 2561 patients with Dupuytren's disease were identified and matched to at least four controls. After a median follow-up of 5 years, mortality was found to be actually reduced in patients with Dupuytren's disease. There was no difference in mortality secondary to cancer or cardiovascular disease. Future studies with longer average follow-up using longitudinal data should clarify these associations in the longer term.Level of evidence: III.

The built-in selection bias of hazard ratios formalized using structural causal models.

It is known that the hazard ratio lacks a useful causal interpretation. Even for data from a randomized controlled trial, the hazard ratio suffers from so-called built-in selection bias as, over time, the individuals at risk among the exposed and unexposed are no longer exchangeable. In this paper, we formalize how the expectation of the observed hazard ratio evolves and deviates from the causal effect of interest in the presence of heterogeneity of the hazard rate of unexposed individuals (frailty) and heterogeneity in effect (individual modification). For the case of effect heterogeneity, we define the causal hazard ratio. We show that the expected observed hazard ratio equals the ratio of expectations of the latent variables (frailty and modifier) conditionally on survival in the world with and without exposure, respectively. Examples with gamma, inverse Gaussian and compound Poisson distributed frailty and categorical (harming, beneficial or neutral) distributed effect modifiers are presented for illustration. This set of examples shows that an observed hazard ratio with a particular value can arise for all values of the causal hazard ratio. Therefore, the hazard ratio cannot be used as a measure of the causal effect without making untestable assumptions, stressing the importance of using more appropriate estimands, such as contrasts of the survival probabilities.

Introducing exceptional growth mining-Analyzing the impact of soil characteristics on on-farm crop growth and yield variability.

Sustainable intensification of agriculture requires understanding of the effect of soil characteristics and nutrient supply on crop growth. As farms are increasing in size by acquiring small fields from various farmers, the soil characteristics and nutrient supply might be very different from field to field, while at the same time specific soil properties might limit the nutrient uptake. As a result, there might be a large number of heterogeneous reasons why crop growth varies significantly. New data analysis techniques can help to explain variability in crop growth among fields. This paper introduces Exceptional Growth Mining (EGM) as a first contribution. EGM instantiates the data mining framework Exceptional Model Mining (EMM) such that subgroups of fields can be found that grow exceptionally in terms of three growth parameters (high/low maximum growth, steep/flat linear growth and early/late midpoint of maximum growth). As second contribution, we apply EGM to a case study by analyzing the dataset of a potato farm in the south of the Netherlands. EGM consists of (i) estimating growth curves by applying nonlinear mixed models, (ii) investigating the correlation between the estimated growth parameters, and (iii) applying EMM on these growth curve parameters using a growth curve-specific quality measure. By applying EGM on the data of the potato farm, we obtain the following results: 1) the estimated growth curves represent the variability in potato tuber growth very well (R2 of 0.92), 2) the steepness of the growth curve has a strong correlation with the maximum growth and the midpoint of maximum growth, and the correlation between the midpoint of maximum growth and maximum growth is weak, 3) the subgroup analyses indicate that: high values of K correspond to high maxima; low values of K correspond to low maxima, steep growth curves', and a late midpoint of halfway growth; Mg influences the midpoint of the growth curve; values of B are higher on dry soils with high tuber growth, while low values of B are found on wet soils with high tuber growth; high values of Zn, Mn, and Fe are found in subgroups with low tuber weight, probably related to the soil's low pH. In summary, this paper introduces EGM to obtain understanding in crop response to soil properties and nutrient supply. In addition, EGM provides a way to analyze only small parts of a large dataset, such that the impact of soil factors on growth can be analyzed on a more detailed level than existing methods.

Treatment durability of limited fasciectomy vs. percutaneous needle fasciotomy for Dupuytren's disease.

BACKGROUND: Patients with Dupuytren's disease (DD) are mostly surgically treated by percutaneous needle fasciotomy (PNF) or limited fasciectomy (LF), but data on time intervals to retreatment is lacking. We aimed to estimate the risk of retreatment within certain time periods after treatment with PNF and LF. METHODS: We used data of participants of a cohort study on the course of DD who were treated only with PNF or LF. Our primary outcome measure was time to retreatment of DD. We included sex, age at first treatment and having a first degree relative with DD as confounders in our analysis. We applied a bivariate gamma frailty model to estimate the risk of retreatment within 1,3,5, 10 and 20 years after treatment with PNF and LF. RESULTS: The time to retreatment was significantly shorter after treatment with PNF than after LF (Wald test 7.56, p<0.001). The estimated 10-year risk of retreatment for men who underwent their first treatment at a younger age and with a first degree relative with DD was 97% after PNF and 32% after LF. The estimated 10-year risk for women who underwent their first treatment at an older age without a first degree relative with DD was 20% after PNF and 6% after LF. CONCLUSIONS: Our results show that the patients treated with PNF have a higher risk of retreatment. The results of this study could contribute to individualized information on the treatment durability in the future, which would improve patient counseling about the expected retreatment needs.

Optimizing Catalytic Depolymerization of Lignin in Ethanol with a Day-Clustered Box-Behnken Design.

Lignin is a potential resource for biobased aromatics with applications in the field of fuel additives, resins, and bioplastics. Via a catalytic depolymerization process using supercritical ethanol and a mixed metal oxide catalyst (CuMgAlOx), lignin can be converted into a lignin oil, containing phenolic monomers that are intermediates to the mentioned applications. Herein, we evaluated the viability of this lignin conversion technology through a stage-gate scale-up methodology. Optimization was done with a day-clustered Box-Behnken design to accommodate the large number of experimental runs in which five input factors (temperature, lignin-to-ethanol ratio, catalyst particle size, catalyst concentration, and reaction time) and three output product streams (monomer yield, yield of THF-soluble fragments, and yield of THF-insoluble fragments and char) were considered. Qualitative relationships between the studied process parameters and the product streams were determined based on mass balances and product analyses. Linear mixed models with random intercept were employed to study quantitative relationships between the input factors and the outcomes through maximum likelihood estimation. The response surface methodology study reveals that the selected input factors, together with higher order interactions, are highly significant for the determination of the three response surfaces. The good agreement between the predicted and experimental yield of the three output streams is a validation of the response surface methodology analysis discussed in this contribution.

Can body position be arrhythmogenic?

INTRODUCTION: Palpitations occurring in specific body positions are often reported by patients, but the effect of body position on arrhythmia has received little research attention. We hypothesize that resting body position can exert pro-arrhythmogenic effects in various ways. For example, lateral body position is known to increase change atrial and pulmonary vein dimensions. METHODS: This observational study capitalizes on overnight polysomnography (PSG) recordings from a tertiary sleep clinic. PSGs were retrieved based on any mention of cardiac arrhythmia in the clinical report, irrespective of primary sleep diagnosis or (cardiac) comorbidities. Every instance of atrial ectopy was annotated and subgroups with a homogenous rate of atrial ectopy were created based on the Dunn index. A generalized linear mixed-effects model using age, sex, gender, sleep stage and body position was used to analyse the total amount of atrial ectopy in each combination of sleep stage and body position. Backward elimination was then performed to select the best subset of variables for the model. Presence of a respiratory event was then added to the model for the subgroup with a high atrial ectopy rate. RESULTS: PSGs of 22 patients (14% female, mean age 61y) were clustered and analysed. Body position, sleep stage, age or sex did not have a significant effect on atrial ectopy in the subgroup with a low rate of atrial ectopy (N = 18). However, body position did significantly affect the rate of atrial ectopy in the subgroup with a high rate of atrial ectopy (N = 4; 18%). Respiratory events significantly altered the atrial ectopy rate in only three body positions across two patients. DISCUSSION: In each individual with a high rate of atrial ectopy, the rate of atrial ectopy was significantly higher in either left or right decubital or supine position. Increase in atrial wall stretch in lateral decubital position and obstructive respiratory events in positional sleep apnea are two possible pathophysiological mechanisms, while avoidance of a body position due to symptomatic atrial ectopy in that position is an important limitation. CONCLUSION: In a selected cohort of patients with a high rate of atrial ectopy during overnight polysomnography, the occurrence of atrial ectopy is related to resting body position.

Effect of Multimodal Prehabilitation on Reducing Postoperative Complications and Enhancing Functional Capacity Following Colorectal Cancer Surgery: The PREHAB Randomized Clinical Trial.

Importance: Colorectal surgery is associated with substantial morbidity rates and a lowered functional capacity. Optimization of the patient's condition in the weeks prior to surgery may attenuate these unfavorable sequelae. Objective: To determine whether multimodal prehabilitation before colorectal cancer surgery can reduce postoperative complications and enhance functional recovery. Design, Setting, and Participants: The PREHAB randomized clinical trial was an international, multicenter trial conducted in teaching hospitals with implemented enhanced recovery after surgery programs. Adult patients with nonmetastasized colorectal cancer were assessed for eligibility and randomized to either prehabilitation or standard care. Both arms received standard perioperative care. Patients were enrolled from June 2017 to December 2020, and follow-up was completed in December 2021. However, this trial was prematurely stopped due to the COVID-19 pandemic. Interventions: The 4-week in-hospital supervised multimodal prehabilitation program consisted of a high-intensity exercise program 3 times per week, a nutritional intervention, psychological support, and a smoking cessation program when needed. Main Outcomes and Measures: Comprehensive Complication Index (CCI) score, number of patients with CCI score more than 20, and improved walking capacity expressed as the 6-minute walking distance 4 weeks postoperatively. Results: In the intention-to-treat population of 251 participants (median [IQR] age, 69 [60-76] years; 138 [55%] male), 206 (82%) had tumors located in the colon and 234 (93%) underwent laparoscopic- or robotic-assisted surgery. The number of severe complications (CCI score >20) was significantly lower favoring prehabilitation compared with standard care (21 of 123 [17.1%] vs 38 of 128 [29.7%]; odds ratio, 0.47 [95% CI, 0.26-0.87]; P = .02). Participants in prehabilitation encountered fewer medical complications (eg, respiratory) compared with participants receiving standard care (19 of 123 [15.4%] vs 35 of 128 [27.3%]; odds ratio, 0.48 [95% CI, 0.26-0.89]; P = .02). Four weeks after surgery, 6-minute walking distance did not differ significantly between groups when compared with baseline (mean difference prehabilitation vs standard care 15.6 m [95% CI, -1.4 to 32.6]; P = .07). Secondary parameters of functional capacity in the postoperative period generally favored prehabilitation compared with standard care. Conclusions and Relevance: This PREHAB trial demonstrates the benefit of a multimodal prehabilitation program before colorectal cancer surgery as reflected by fewer severe and medical complications postoperatively and an optimized postoperative recovery compared with standard care. Trial Registration: trialregister.nl Identifier: NTR5947.

Therapeutic drug monitoring-guided treatment versus standard dosing of voriconazole for invasive aspergillosis in haematological patients: a multicentre, prospective, cluster randomised, crossover clinical trial.

OBJECTIVES: Voriconazole therapeutic drug monitoring (TDM) is recommended based on retrospective data and limited prospective studies. This study aimed to investigate whether TDM-guided voriconazole treatment is superior to standard treatment for invasive aspergillosis. METHODS: A multicentre (n = 10), prospective, cluster randomised, crossover clinical trial was performed in haematological patients aged ≥18 years treated with voriconazole. All patients received standard voriconazole dose at the start of treatment. Blood/serum/plasma was periodically collected after treatment initiation of voriconazole and repeated during treatment in both groups. The TDM group had measured voriconazole concentrations reported back, with dose adjustments made as appropriate, while the non-TDM group had voriconazole concentrations measured only after study completion. The composite primary endpoint included response to treatment and voriconazole treatment discontinuation due to an adverse drug reaction related to voriconazole within 28 days after treatment initiation. RESULTS: In total, 189 patients were enrolled in the study. For the composite primary endpoint, 74 patients were included in the non-TDM group and 68 patients in the TDM group. Here, no significant difference was found between both groups (P = 0.678). However, more trough concentrations were found within the generally accepted range of 1-6 mg/L for the TDM group (74.0%) compared with the non-TDM group (64.0%) (P < 0.001). CONCLUSIONS: In this trial, TDM-guided dosing of voriconazole did not show improved treatment outcome compared with standard dosing. We believe that these findings should open up the discussion for an approach to voriconazole TDM that includes drug exposure, pathogen susceptibility and host defence. CLINICAL TRIAL REGISTRATION: ClinicalTrials.gov registration no. NCT00893555.

Comparison of analysis methods and design choices for treatment-by-period interaction in unidirectional switch designs: a simulation study.

BACKGROUND: Due to identifiability problems, statistical inference about treatment-by-period interactions has not been discussed for stepped wedge designs in the literature thus far. Unidirectional switch designs (USDs) generalize the stepped wedge designs and allow for estimation and testing of treatment-by-period interaction in its many flexible design forms. METHODS: Under different forms of the USDs, we simulated binary data at both aggregated and individual levels and studied the performances of the generalized linear mixed model (GLMM) and the marginal model with generalized estimation equations (GEE) for estimating and testing treatment-by-period interactions. RESULTS: The parallel group design had the highest power for detecting the treatment-by-period interactions. While there was no substantial difference between aggregated-level and individual-level analysis, the GLMM had better point estimates than the marginal model with GEE. Furthermore, the optimal USD for estimating the average treatment effect was not efficient for treatment-by-period interaction and the marginal model with GEE required a substantial number of clusters to yield unbiased estimates of the interaction parameters when the correlation structure is autoregressive of order 1 (AR1). On the other hand, marginal model with GEE had better coverages than GLMM under the AR1 correlation structure. CONCLUSION: From the designs and methods evaluated, in general, parallel group design with a GLMM is, preferred for estimation and testing of treatment-by-period interaction in a clustered randomized controlled trial for a binary outcome.

Temporal Trends in Incidence of Premature Cardiovascular Disease Over the Past 7 Decades: The Framingham Heart Study.

Background Premature onset of cardiovascular disease (CVD) imposes a significant societal burden and challenges prevention efforts. Methods and Results Trends in the incidence of premature CVD (before age 55, 60, or 65 years, separate analysis for each threshold) were evaluated in 14 464 Framingham Heart Study participants over 7 decades of observation (1950-2019). The change in the incidence of premature CVD (per decade) in men and women was assessed using overdispersed Poisson regression (accounting for cohort effects), adjusting for age at entry and age at onset of premature CVD within each decade. CVD was defined as a composite of fatal or nonfatal coronary heart disease, stroke or transient ischemic attack, peripheral vascular disease, and heart failure. There were 2223 first CVD events (832 in women) before age 65 years during 282 481 person-years of observations (154 587 in women) between 1950 and 2019. The age-adjusted CVD incidence before age 65 years decreased from 14.8 per 1000 person-years (1950-1959) to 4.69 per 1000 person-years (2010-2019) in men and from 7.23 per 1000 person-years (1950-1959) to 1.73 per 1000 person-years (2010-2019) in women. In adjusted analyses, the incidence of premature CVD decreased per decade in men (18.4% [95% CI, 12.0%-24.0%], for onset before age 55 years; 19.5% [95% CI, 12.0%-27.0%], for onset before age 60 years; 21.3% [95% CI, 16.0%-27.0%], for onset before age 65 years) and women (15.1% [95% CI, 7.0%-22.0%], for onset before age 55 years; 14.0% [95% CI, 6.0%-22.0%], for onset before age 60 years; 18.2% [95% CI, 12.0%-24.0%], for onset before age 65 years). The decline in premature CVD was accompanied by a reduction in smoking and increased use of lipid-lowering treatments across the decades. Incidence of premature coronary heart disease decreased, whereas the contribution of stroke to premature CVD burden increased over time. Conclusions The incidence of premature CVD has decreased among White adults in the Framingham cohort over the past 70 years; the residual burden of premature stroke warrants further study. Additional studies of trends in premature CVD in more racially and geographically diverse populations are warranted to elucidate the generalizability of these findings.

Multicentre paired non-inferiority study of the cardiorespiratory monitoring performance of the wireless and non-adhesive Bambi® belt measuring diaphragm activity in neonates: study protocol.

INTRODUCTION: Cardiorespiratory monitoring is used in the neonatal intensive care unit (NICU) to assess the clinical status of newborn infants and detect critical deteriorations in cardiorespiratory function. Currently, heart rate (HR) is monitored by electrocardiography (ECG) and respiration by chest impedance (CI). Disadvantages of current monitoring techniques are usage of wired adhesive electrodes which may damage the skin and hinder care. The Bambi® belt is a wireless and non-adhesive alternative that enables cardiorespiratory monitoring by measuring electrical activity of the diaphragm via transcutaneous electromyography. A previous study showed feasibility of the Bambi® belt and this study compares the belt performance to ECG and CI. METHODS AND ANALYSIS: This multicentre non-inferiority paired study will be performed in the NICU of the Máxima Medical Center (MMC) in Veldhoven and the Emma Children's Hospital, Amsterdam University Medical Centre (AmsterdamUMC) in Amsterdam, The Netherlands. 39 infants in different postmenstrual age groups (minimally 10 infants<30 weeks, between 30-32 weeks and >32 weeks) will be recruited. These infants will be monitored with the Bambi® belt in addition to standard ECG and CI for 24 hours. The primary outcome is the HR, studied with three criteria: (1) the limits of agreement of the HR measurements in terms of the second-to-second difference in the HR between the belt and standard ECG, (2) the detection of cardiac events consisting of bradycardia and tachycardia and (3) the quality of HR-monitoring. The secondary outcome is the respiratory rate (RR), studied with the criteria (1) agreement in RR-trend monitoring, (2) apnoea and tachypnoea detection and (3) reliable registrations. ETHICS AND DISSEMINATION: This protocol was approved by the Medical Ethical Committee of the MMC and the Central Committee for Human Research. The MMC started patient recruitment in July and the AmsterdamUMC in August 2021. The results will be presented at conferences and published in peer-reviewed journals. TRIAL REGISTRATION NUMBER: NL9480.

Non-inferiority testing for qualitative microbiological methods: Assessing and improving the approach in USP 1223.

The United States Pharmacopoeia (USP) presents two approaches for showing non-inferiority of an alternate qualitative microbiological method versus a compendial method. One approach compares the positive rates for the alternate and compendial methods at one spike level, while the other one compares multiple most probable number (MPN) estimates from a multi-spike design using a t-test. In this paper, we discuss these approaches under certain assumptions and propose a third approach that can be used for both single and multiple dilutions, which we call the generalized MPN (gMPN) approach. Simulations, using Poisson distributed numbers of microorganisms in test samples, confirm that the USP approach based on rates is not suitable, that the USP approach based on MPNs is appropriate for non-inferiority, but the gMPN approach outperforms the MPN-based approach and is therefore recommended.

Disease Course of Primary Dupuytren Disease: 5-Year Results of a Prospective Cohort Study.

BACKGROUND: Predicting progression of Dupuytren disease becomes relevant in an upcoming era with progression-preventing treatment. This study aimed to determine the course of Dupuytren disease and identify factors associated with progression. METHODS: Two hundred fifty-eight patients with Dupuytren disease participated in this prospective cohort study, obtaining 17,645 observations in 5 years. Outcomes were disease extent (surface area) and contracture severity (total passive extension deficit). Demographics, lifestyle, health status, exposure to manual work, and genetic risk scores were gathered as potential predictors. Subject-specific, mixed-effects models were used to estimate disease course, and logistic regression with least absolute shrinkage and selection operator was used to evaluate factors associated with the presence of progression. RESULTS: On average, Dupuytren disease was progressive in all finger rays with regard to area [yearly increase, 0.07 cm2 (95% CI, 0.02 to 0.13 cm2) to 0.25 cm2 (95% CI, 0.11 to 0.39 cm2)]. Progression in total passive extension deficit was only present on the small finger side [yearly increase, 1.75 degrees (95% CI, 0.30 to 3.20 degrees) to 6.25 degrees (95% CI, 2.81 to 9.69 degrees)]. Stability or regression in area and total passive extension deficit was observed in 11 and 13 percent and 16 and 15 percent (dominant and nondominant hands), respectively. Smoking, cancer, genetic risk score, and hand injury were univariate associated with progression in area, but after multivariate variable selection, none of these associations remained. No predictors for progression in total passive extension deficit were found. CONCLUSIONS: Dupuytren disease is progressive, especially with respect to disease extent. Progression in contracture severity is mainly present on the small finger side of the hand. None of the traditional risk and diathesis factors were associated with progression, indicating that new hypotheses about Dupuytren disease progression might be needed. CLINICAL QUESTION/LEVEL OF EVIDENCE: Risk, III.

Inference on latent factor models for informative censoring.

This work discusses the problem of informative censoring in survival studies. A joint model for the time to event and the time to censoring is presented. Their hazard functions include a latent factor in order to identify this joint model without sacrificing the flexibility of the parametric specification. Furthermore, a fully Bayesian formulation with a semi-parametric proportional hazard function is provided. Similar latent variable models have been described in literature, but here the emphasis is on the performance of the inferential task of the resulting mixture model with unknown number of components. The posterior distribution of the parameters is estimated using Hamiltonian Monte Carlo methods implemented in Stan. Simulation studies are provided to study its performance and the methodology is implemented for the analysis of the ACTG175 clinical trial dataset yielding a better fit. The results are also compared to the non-informative censoring case to show that ignoring informative censoring may lead to serious biases.

A latent-class heteroskedastic hurdle trajectory model: patterns of adherence in obstructive sleep apnea patients on CPAP therapy.

BACKGROUND: Sleep apnea patients on CPAP therapy exhibit differences in how they adhere to the therapy. Previous studies have demonstrated the benefit of describing adherence in terms of discernible longitudinal patterns. However, these analyses have been done on a limited number of patients, and did not properly represent the temporal characteristics and heterogeneity of adherence. METHODS: We illustrate the potential of identifying patterns of adherence with a latent-class heteroskedastic hurdle trajectory approach using generalized additive modeling. The model represents the adherence trajectories on three aspects over time: the daily hurdle of using the therapy, the daily time spent on therapy, and the day-to-day variability. The combination of these three characteristics has not been studied before. RESULTS: Applying the proposed model to a dataset of 10,000 patients in their first three months of therapy resulted in nine adherence groups, among which 49% of patients exhibited a change in adherence over time. The identified group trajectories revealed a non-linear association between the change in the daily hurdle of using the therapy, and the average time on therapy. The largest difference between groups was observed in the patient motivation score. The adherence patterns were also associated with different levels of high residual AHI, and day-to-day variability in leakage. CONCLUSION: The inclusion of the hurdle model and the heteroskedastic model into the mixture model enabled the discovery of additional adherence patterns, and a more descriptive representation of patient behavior over time. Therapy adherence was mostly affected by a lack of attempts over time, suggesting that encouraging these patients to attempt therapy on a daily basis, irrespective of the number of hours used, could drive adherence. We believe the methodology is applicable to other domains of therapy or medication adherence.

Tests for publication bias are unreliable in case of heteroscedasticity.

Regression based methods for the detection of publication bias in meta-analysis have been extensively evaluated in literature. When dealing with continuous outcomes, specific hidden factors (e.g., heteroscedasticity) may interfere with the test statistics. In this paper we investigate the influence of residual heteroscedasticity on the performance of four tests for publication bias: the Egger test, the Begg-Mazumdar test and two tests based on weighted regression. In the presence of heteroscedasticity, the Egger test and the weighted regression tests highly inflate the Type I error rate, while the Begg-Mazumdar test deflates the Type I error rate. Although all three tests already have low statistical power, heteroscedasticity typically reduces it further. Our results in combination with earlier discussions on publication bias tests lead us to conclude that application of these tests on continuous treatment effects is not warranted.

Improved One-Class Modeling of High-Dimensional Metabolomics Data via Eigenvalue-Shrinkage.

One-class modelling is a useful approach in metabolomics for the untargeted detection of abnormal metabolite profiles, when information from a set of reference observations is available to model "normal" or baseline metabolite profiles. Such outlying profiles are typically identified by comparing the distance between an observation and the reference class to a critical limit. Often, multivariate distance measures such as the Mahalanobis distance (MD) or principal component-based measures are used. These approaches, however, are either not applicable to untargeted metabolomics data, or their results are unreliable. In this paper, five distance measures for one-class modeling in untargeted metabolites are proposed. They are based on a combination of the MD and five so-called eigenvalue-shrinkage estimators of the covariance matrix of the reference class. A simple cross-validation procedure is proposed to set the critical limit for outlier detection. Simulation studies are used to identify which distance measure provides the best performance for one-class modeling, in terms of type I error and power to identify abnormal metabolite profiles. Empirical evidence demonstrates that this method has better type I error (false positive rate) and improved outlier detection power than the standard (principal component-based) one-class models. The method is illustrated by its application to liquid chromatography coupled to mass spectrometry (LC-MS) and nuclear magnetic response spectroscopy (NMR) untargeted metabolomics data from two studies on food safety assessment and diagnosis of rare diseases, respectively.