Marginal Proportional Hazards Models Research Articles

2799. Evaluating Resistance Patterns of LVAD-related Pseudomonas Infections

Abstract Background Left ventricular assist devices (LVADs) provide a clear benefit to patients with end-stage heart failure. Despite this, device-related infections such as driveline infections, device pocket infections, and LVAD-related bloodstream infections occur in up to 39% of implanted patients. Pseudomonas aeruginosa is a common causative organism in LVAD-associated infections and frequently develops resistance, limiting treatment options. This study aims to analyze the relationships between the frequency of acquired resistance (AR) and single-agent antibiotic courses, as well as the frequency of AR and duration of antibiotic courses; and compare the likelihood of AR between different antibiotic classes. Methods This was a single-center retrospective cohort study involving all patients with LVAD-associated P. aeruginosa infections between 2011-2023. Patients treated with piperacillin-tazobactam, cefepime, ciprofloxacin, or meropenem, were included (n = 37). Each antibiotic course (n = 117) was associated with a pre- and post-antibiotic sensitivity. AR was defined as culture-proven, newly developed resistance to the utilized antibiotic after starting therapy. The incidence and cumulative proportion of AR were estimated using Poisson regression and Kaplan-Meier curve analysis, respectively. A marginal proportional hazards model was used to analyze time to AR between different antibiotic groups. Results There was a significantly lower incidence of AR with cefepime [0.8 (95% CI 0.2-3.3) per 100 person-weeks] compared to piperacillin-tazobactam [3.6 (1.7, 8)], meropenem [7.6 (4.1, 14.5)], and ciprofloxacin [6.4 (3.4, 12.5)] (p=0.04). Similarly, the cumulative proportion of AR was significantly lower with cefepime as compared to other antibiotic classes (Figure 1). The hazard rate of AR was approximately 3.5 to 6.7 times higher with each antibiotic in comparison to cefepime (p = 0.02) (Table 1).Figure 1.Kaplan-Meyer curve demonstrating cumulative proportion of acquired resistance Resistance occurs early and more frequently with piperacillin-tazobactam, meropenem, and ciprofloxacin, but less frequently and late with cefepime. The cumulative proportion of acquired resistance over time differed significantly across antibiotic types (p = 0.02). Cefepime had a statistically significantly lower hazard of AR when compared to the rest of the antibiotic groups as a whole, and when compared individually, had a significantly lower hazard than ciprofloxacin and meropenem. Conclusion This study suggests a lower incidence of AR for LVAD-associated P. aeruginosa infection treated with cefepime. AR also seems to develop later in patients treated with cefepime. There is a 3.5-6.7-fold increase in the risk of acquiring resistance with other antibiotics compared to cefepime. Disclosures Jeffrey Alexis, MD, Abbott: Grant/Research Support

Efficacy of Sotagliflozin in Adults With Type 2 Diabetes in Relation to Baseline Hemoglobin A1c

BackgroundThe SCORED (Effect of Sotagliflozin on Cardiovascular and Renal Events in Patients with Type 2 Diabetes and Moderate Renal Impairment Who Are at Cardiovascular Risk) and SOLOIST-WHF (Effect of Sotagliflozin on Cardiovascular Events in Patients with Type 2 Diabetes Post Worsening Heart Failure) trials demonstrated that sotagliflozin, an SGLT1 and SGLT2 inhibitor, improves outcomes in individuals with type 2 diabetes who have heart failure (HF) or kidney disease. ObjectivesWe assessed the efficacy of sotagliflozin on HF clinical outcomes in individuals with differing baseline glycosylated hemoglobin (HbA1c) levels. MethodsWe included all adults from SCORED and SOLOIST-WHF. The primary outcome was a composite of cardiovascular death, hospitalizations for HF, and urgent visits for HF. The efficacy of sotagliflozin compared with placebo was evaluated by baseline HbA1c using competing-risk marginal proportional hazards models. ResultsWe identified 11,744 adults. Individuals with HbA1c ≤7.5% experienced the primary outcome at a lower rate in the sotagliflozin group (11.2 per 100 person-years) than the placebo group (15.5 per 100 person-years) (HR: 0.73; 95% CI: 0.57-0.93). Similarly, individuals with HbA1c of 7.6% to 9.0% experienced the primary outcome at a lower rate in the sotagliflozin group (7.3 per 100 person-years) than the placebo group (9.4 per 100 person-years) (HR: 0.77; 95% CI: 0.63-0.96). These findings were also consistent among individuals with HbA1c >9.0%, with a primary outcome rate in the sotagliflozin group (7.8 per 100 person-years) that was lower than the placebo group (11.6 per 100 person-years) (HR: 0.65; 95% CI: 0.50-0.84). The efficacy of sotagliflozin was consistent by baseline HbA1c level (P for interaction = 0.58). ConclusionsIn individuals with type 2 diabetes and either HF or kidney disease, sotagliflozin reduced HF outcomes irrespective of baseline HbA1c.

Marginal proportional hazards models for multivariate interval-censored data.

Multivariate interval-censored data arise when there are multiple types of events or clusters of study subjects, such that the event times are potentially correlated and when each event is only known to occur over a particular time interval. We formulate the effects of potentially time-varying covariates on the multivariate event times through marginal proportional hazards models while leaving the dependence structures of the related event times unspecified. We construct the nonparametric pseudolikelihood under the working assumption that all event times are independent, and we provide a simple and stable EM-type algorithm. The resulting nonparametric maximum pseudolikelihood estimators for the regression parameters are shown to be consistent and asymptotically normal, with a limiting covariance matrix that can be consistently estimated by a sandwich estimator under arbitrary dependence structures for the related event times. We evaluate the performance of the proposed methods through extensive simulation studies and present an application to data from the Atherosclerosis Risk in Communities Study.

Marginal proportional hazards models for clustered interval-censored data with time-dependent covariates.

The Botswana Combination Prevention Project was a cluster-randomized HIV prevention trial whose follow-up period coincided with Botswana's national adoption of a universal test and treat strategy for HIV management. Of interest is whether, and to what extent, this change in policy modified the preventative effects of the study intervention. To address such questions, we adopt a stratified proportional hazards model for clustered interval-censored data with time-dependent covariates and develop a composite expectation maximization algorithm that facilitates estimation of model parameters without placing parametric assumptions on either the baseline hazard functions or the within-cluster dependence structure. We show that the resulting estimators for the regression parameters are consistent and asymptotically normal. We also propose and provide theoretical justification for the use of the profile composite likelihood function to construct a robust sandwich estimator for the variance. We characterize the finite-sample performance and robustness of these estimators through extensive simulation studies. Finally, we conclude by applying this stratified proportional hazards model to a re-analysis of the Botswana Combination Prevention Project, with the national adoption of a universal test and treat strategy now modeled as a time-dependentcovariate.

Total hospitalizations after peripheral arterial revascularization in the VOYAGER trial

Abstract Background In the VOYAGER PAD trial, rivaroxaban reduced first and total (first and subsequent) occurrences of major adverse limb and cardiovascular events in patients with peripheral artery disease (PAD) after lower extremity revascularization (LER), but also increased incident bleeding. Purpose Assessment of disease burden by the onset of a particular set of events does not necessarily capture all consequences of a disease process that negatively impact patients' quality of life, motivating the adoption of more patient-centered outcomes. The present analysis of VOYAGER PAD describes the total incidence and duration of hospitalizations reported during the study as a broader measure of total disease burden and net benefit of rivaroxaban therapy. Methods Patients were randomized 1:1 to rivaroxaban 2.5 mg twice daily plus aspirin (n=3286) or aspirin alone (n=3278). Investigators documented the primary reason and duration for each hospitalization occurring after randomization. Total hospitalization rates were estimated by events per 100 patient-years. Treatment group comparisons on total hospitalizations as a time-to-event outcome were by marginal proportional hazards models with death as a competing terminal event, while days in hospital were compared by zero-inflated Poisson regression. All analyses were intention-to-treat. Results A total of 7156 hospitalizations (3265 first, 3891 subsequent) occurred during a median 2.5 years of follow-up, with nearly 40% being hospitalizations for PAD and a small fraction attributed to bleeding events (Figure 1). The rivaroxaban rate was lower for PAD hospitalizations and higher for bleeding event hospitalizations, with more hospitalizations prevented than caused (Table 1). Mean days in hospital among patients hospitalized at least once was significantly lower in the rivaroxaban group for any reason, for PAD, and for bleeding, so that total days in hospital was numerically lower for rivaroxaban. Conclusion Patients with PAD undergoing LER have a high rate of subsequent hospitalizations after an index procedure, driven by re-hospitalizations for PAD. Rivaroxaban decreased both the incidence and duration of PAD hospitalizations. While rivaroxaban increased incident hospitalizations due to bleeding, the mean durations of these hospitalizations were lower relative to placebo, leading to fewer total days in hospital for bleeding (1252 vs. 1531 total days). These findings may be useful for clinicians and patients weighing the risks and benefits of rivaroxaban in PAD after revascularization. Funding Acknowledgement Type of funding sources: Private company. Main funding source(s): BayerJanssen

Longitudinal mediation analysis of time-to-event endpoints in the presence of competing risks.

This proposal is motivated by an analysis of the English Longitudinal Study of Ageing (ELSA), which aims to investigate the role of loneliness in explaining the negative impact of hearing loss on dementia. The methodological challenges that complicate this mediation analysis include the use of a time-to-event endpoint subject to competing risks, as well as the presence of feedback relationships between the mediator and confounders that are both repeatedly measured over time. To account for these challenges, we introduce path-specific effect proportional (cause-specific) hazard models. These extend marginal structural proportional (cause-specific) hazard models to enable effect decomposition on either the cause-specific hazard ratio scale or the cumulative incidence function scale. We show that under certain ignorability assumptions, the path-specific direct and indirect effects indexing this model are identifiable from the observed data. We next propose an inverse probability weighting approach to estimate these effects. On the ELSA data, this approach reveals little evidence that the total effect of hearing loss on dementia is mediated through the feeling of loneliness, with a non-statistically significant indirect effect equal to 1.01 (hazard ratio (HR) scale; 95% confidence interval (CI) 0.99 to 1.05).

Semiparametric marginal regression for clustered competing risks data with missing cause of failure.

Clustered competing risks data are commonly encountered in multicenter studies. The analysis of such data is often complicated due to informative cluster size (ICS), a situation where the outcomes under study are associated with the size of the cluster. In addition, the cause of failure is frequently incompletely observed in real-world settings. To the best of our knowledge, there is no methodology for population-averaged analysis with clustered competing risks data with an ICS and missing causes of failure. To address this problem, we consider the semiparametric marginal proportional cause-specific hazards model and propose a maximum partial pseudolikelihood estimator under a missing at random assumption. To make the latter assumption more plausible in practice, we allow for auxiliary variables that may be related to the probability of missingness. The proposed method does not impose assumptions regarding the within-cluster dependence and allows for ICS. The asymptotic properties of the proposed estimators for both regression coefficients and infinite-dimensional parameters, such as the marginal cumulative incidence functions, are rigorously established. Simulation studies show that the proposed method performs well and that methods that ignore the within-cluster dependence and the ICS lead to invalid inferences. The proposed method is applied to competing risks data from a large multicenter HIV study in sub-Saharan Africa where a significant portion of causes of failure is missing.

Multiple Sclerosis Severity Score (MSSS) improves the accuracy of individualized prediction in MS

Background: The MSBase prediction model of treatment response leverages multiple demographic and clinical characteristics to estimate hazards of relapses, confirmed disability accumulation (CDA), and confirmed disability improvement (CDI). The model did not include Multiple Sclerosis Severity Score (MSSS), a disease duration-adjusted ranked score of disability. Objective: To incorporate MSSS into the MSBase prediction model and compare model accuracy with and without MSSS. Methods: The associations between MSSS and relapse, CDA, and CDI were evaluated with marginal proportional hazards models adjusted for three principal components representative of patients’ demographic and clinical characteristics. The model fit with and without MSSS was assessed with penalized r2 and Harrell C. Results: A total of 5866 MS patients were started on disease-modifying therapy during prospective follow-up (age 38.4 ± 10.6 years; 72% female; disease duration 8.5 ± 7.7 years). Including MSSS into the model improved the accuracy of individual prediction of relapses by 31%, of CDA by 23%, and of CDI by 24% (Harrell C) and increased the amount of variance explained for relapses by 49%, for CDI by 11%, and for CDA by 10% as compared with the original model. Conclusion: Addition of a single, readily available metric, MSSS, to the comprehensive MSBase prediction model considerably improved the individual accuracy of prognostics in MS.

Finite‐sample adjustments in variance estimators for clustered competing risks regression

The marginal Fine-Gray proportional subdistribution hazards model is a popular approach to directly study the association between covariates and the cumulative incidence function with clustered competing risks data, which often arise in multicenter randomized trials or multilevel observational studies. To account for the within-cluster correlations between failure times, the uncertainty of the regression parameters estimators is quantified by the robust sandwich variance estimator, which may have unsatisfactory performance with a limited number of clusters. To overcome this limitation, we propose four bias-corrected variance estimators to reduce the negative bias of the usual sandwich variance estimator, extending the bias-correction techniques from generalized estimating equations with noncensored exponential family outcomes to clustered competing risks outcomes. We further compare their finite-sample operating characteristics through simulations and two real data examples. In particular, we found the Mancl and DeRouen (MD) type sandwich variance estimator generally has the smallest bias. Furthermore, with a small number of clusters, the Wald -confidence interval with the MD sandwich variance estimator carries close to nominal coverage for the cluster-level effect parameter. The -confidence intervals based on the sandwich variance estimator with any one of the three types of multiplicative bias correction or the -confidence interval with the Morel, Bokossa and Neerchal (MBN) type sandwich variance estimator have close to nominal coverage for the individual-level effect parameter. Finally, we develop a user-friendly R package crrcbcv implementing the proposed sandwich variance estimators to assist practical applications.

Total Ischemic Event Reduction With Rivaroxaban After Peripheral Arterial Revascularization in the VOYAGER PAD Trial

BackgroundPatients with peripheral artery disease (PAD) undergoing lower extremity revascularization (LER) are at high risk of major adverse limb and cardiovascular events. The VOYAGER PAD (Efficacy and Safety of Rivaroxaban in Reducing the Risk of Major Thrombotic Vascular Events in Subjects With Symptomatic Peripheral Artery Disease Undergoing Peripheral Revascularization Procedures of the Lower Extremities) trial demonstrated that rivaroxaban 2.5 mg twice daily reduced first events by 15%. The benefit of rivaroxaban on total (first and subsequent) events in this population is unknown. ObjectivesThis study sought to evaluate the total burden of vascular events in patients with PAD after LER and the efficacy of low-dose rivaroxaban on total events. MethodsVOYAGER PAD randomized patients with PAD undergoing LER to rivaroxaban 2.5 mg twice daily plus aspirin or aspirin alone. The primary endpoint was time to first event of acute limb ischemia, major amputation of a vascular cause, myocardial infarction, ischemic stroke, or cardiovascular death. The current analysis considered all events (first and subsequent) for components of the primary endpoint as well as additional vascular events including peripheral revascularizations and venous thromboembolism. HRs were estimated by marginal proportional hazards models. ResultsAmong 6,564 randomized events, there were 4,714 total first and subsequent vascular events including 1,614 primary endpoint events and 3,100 other vascular events. Rivaroxaban reduced total primary endpoint events (HR: 0.86; 95% CI: 0.75-0.98; P = 0.02) and total vascular events (HR: 0.86; 95% CI: 0.79-0.95; P = 0.003). An estimated 4.4 primary and 12.5 vascular events per 100 participants were avoided with rivaroxaban over 3 years. ConclusionsPatients with symptomatic PAD who are undergoing LER have a high total event burden that is significantly reduced with rivaroxaban. Total event reduction may be a useful metric to quantify the efficacy of rivaroxaban in this setting. (Efficacy and Safety of Rivaroxaban in Reducing the Risk of Major Thrombotic Vascular Events in Subjects With Symptomatic Peripheral Artery Disease Undergoing Peripheral Revascularization Procedures of the Lower Extremities [VOYAGER PAD]; NCT02504216)

Multivariate survival analysis in big data: A divide-and-combine approach.

Multivariate failure time data are frequently analyzed using the marginal proportional hazards models and the frailty models. When the sample size is extraordinarily large, using either approach could face computational challenges. In this paper, we focus on the marginal model approach and propose a divide-and-combine method to analyze large-scale multivariate failure time data. Our method is motivated by the Myocardial Infarction Data Acquisition System (MIDAS), a New Jersey statewide database that includes 73,725,160 admissions to nonfederal hospitals and emergency rooms (ERs) from 1995 to 2017. We propose to randomly divide the full data into multiple subsets and propose a weighted method to combine these estimators obtained from individual subsets using three weights. Under mild conditions, we show that the combined estimator is asymptotically equivalent to the estimator obtained from the full data as if the data were analyzed all at once. In addition, to screen out risk factors with weak signals, we propose to perform the regularized estimation on the combined estimator using its combined confidence distribution. Theoretical properties, such as consistency, oracle properties, and asymptotic equivalence between the divide-and-combine approach and the full data approach are studied. Performance of the proposed method is investigated using simulation studies. Our method is applied to the MIDAS data to identify risk factors related to multivariate cardiovascular-related health outcomes.

CKD Progression in Medicare Beneficiaries With Nonvalvular Atrial Fibrillation Treated With Apixaban Versus Warfarin

Comparing kidney disease progression among patients treated with direct oral anticoagulants (DOACs) versus warfarin has not been well studied. We hypothesized that apixaban would be associated with lower risks of progression of chronic kidney disease (CKD) and progression to incident kidney failure than warfarin in patients with atrial fibrillation (AF). Retrospective cohort study. Medicare recipients with stage 3, 4, or 5 CKD and incident AF who received a new prescription for apixaban or warfarin from 2013 through2017. Apixaban or warfarin. Progression to incident kidney failure or, separately, to a more advanced stage of CKD. Marginal structural cause-specific proportional hazards models with inverse probability weighting to estimate marginal hazard ratios (HRs) for each outcome. HRs compared apixaban to warfarin in intention-to-treat and censored-at-drug-switch analyses. 12,816 individuals met inclusion criteria (50.3% received apixaban; 49.7% received warfarin). After weighting, the mean age of the cohort was 80 ± 7 years, 51% were women, and 88% were White. Approximately 84% had stage 3, 15% had stage 4, and 1% had stage 5 CKD. In the intention-to-treat analysis, apixaban, relative to warfarin, was associated with an HR of developing incident kidney failure of 0.98 (95% confidence interval [CI], 0.79-1.22) and of CKD stage progression of 0.90 (95% CI, 0.82-0.99). Corresponding HRs for censored-at-drug-switch analyses were 0.81 (95% CI, 0.56-1.17) and 0.81 (95% CI, 0.70-0.92). Results were similar for a series of subgroup and sensitivity analyses. CKD was defined based on diagnosis codes from claims; findings may not be generalizable to non-Medicare CKD populations. Apixaban, compared with warfarin, was associated with lower risk of CKD stage progression, but not with incident kidney failure.

Impact of pregnancy on antidepressant treatment course: a population-based comparative cohort study in France

The prevention of relapses and the treatment of depression during pregnancy are difficult challenges. The maintenance of antidepressants in pregnancy with its concomitant risks to mother and child needs to be weighed against those associated with not treating the disease. This study aimed at quantifying the impact of the occurrence of pregnancy on the course of antidepressant treatment among newly treated women (< 6months). We performed a comparative observational cohort study using the nationwide French reimbursement healthcare system database. Women who conceived in 2014 and initiated an antidepressant at any time in the 6months before pregnancy were compared with nonpregnant women newly exposed to antidepressants with matching on age, antidepressant exposure, history of psychiatric disorders, and area of residence. The primary outcome was a composite of antidepressant discontinuation, switch to another antidepressant, and concomitant use of antidepressants. The secondary outcome was the resumption of antidepressant during follow-up. We used Cox marginal proportional hazards models to compare time to outcomes between pregnant and nonpregnant women. The pregnant cohort included 6593 women, and the comparison cohort 29,347 nonpregnant women. In the period following the first month of treatment, pregnant women were more likely to experience treatment modification, and especially to stop receiving it, compared with nonpregnant women (adjusted hazard ratio (aHR) 1.58; 95%CI, 1.51-1.62). Pregnant women who discontinued treatment had a 41% decreased incidence of antidepressant resumption compared with nonpregnant women (aHR 0.59; 95%CI, 0.56-0.62). Pregnancy was a determinant of antidepressant treatment modification, and especially of discontinuation.

Atorvastatin Reduces First and Subsequent Vascular Events Across Vascular Territories: The SPARCL Trial

BackgroundIn the SPARCL (Stroke Prevention by Aggressive Reduction in Cholesterol Levels) trial, atorvastatin was compared with placebo in 4,731 participants with recent stroke or transient ischemic attack and no known coronary heart disease. Atorvastatin reduced the first occurrence of stroke and the first occurrence of a composite of vascular events. ObjectivesThe aim of this post hoc analysis was to assess the occurrence of all (first and subsequent) vascular events and the effect of atorvastatin to reduce these events by vascular territory (cerebrovascular, coronary, or peripheral) in SPARCL. MethodsTreatment effects on total adjudicated vascular events, overall and by vascular territory, were summarized by marginal proportional hazards models. Vascular event rates were estimated for each treatment group with cumulative incidence functions. ResultsThe placebo group had an estimated 41.2 first and 62.7 total vascular events per 100 participants over 6 years. There were 164 fewer first and 390 fewer total vascular events in the atorvastatin group (total events hazard ratio: 0.68; 95% confidence interval: 0.60 to 0.77). The total events reduction included 177 fewer cerebrovascular, 170 fewer coronary, and 43 fewer peripheral events. Over 6 years, an estimated 20 vascular events per 100 participants were avoided with atorvastatin treatment. ConclusionsIn participants with recent stroke or transient ischemic attack, the total number of vascular events prevented with atorvastatin was more than twice the number of first events prevented. Total event reduction provides a comprehensive metric to capture the totality of atorvastatin clinical efficacy in reducing disease burden after stroke or transient ischemic attack. (Lipitor in the Prevention of Stroke, for Patients Who Have Had a Previous Stroke [SPARCL]; NCT00147602)

Marginal hazard ratio estimates in joint frailty models for heart failure trials.

This work is motivated by clinical trials in chronic heart failure disease, where treatment has effects both on morbidity (assessed as recurrent non‐fatal hospitalisations) and on mortality (assessed as cardiovascular death, CV death). Recently, a joint frailty proportional hazards model has been proposed for these kind of efficacy outcomes to account for a potential association between the risk rates for hospital admissions and CV death. However, more often clinical trial results are presented by treatment effect estimates that have been derived from marginal proportional hazards models, that is, a Cox model for mortality and an Andersen–Gill model for recurrent hospitalisations. We show how these marginal hazard ratios and their estimates depend on the association between the risk processes, when these are actually linked by shared or dependent frailty terms. First we derive the marginal hazard ratios as a function of time. Then, applying least false parameter theory, we show that the marginal hazard ratio estimate for the hospitalisation rate depends on study duration and on parameters of the underlying joint frailty model. In particular, we identify parameters, for example the treatment effect on mortality, that determine if the marginal hazard ratio estimate for hospitalisations is smaller, equal or larger than the conditional one. How this affects rejection probabilities is further investigated in simulation studies. Our findings can be used to interpret marginal hazard ratio estimates in heart failure trials and are illustrated by the results of the CHARM‐Preserved trial (where CHARM is the ‘Candesartan in Heart failure Assessment of Reduction in Mortality and morbidity’ programme).

Impact of Intravenous Fluid Therapy on Survival Among Patients With Ebola Virus Disease: An International Multisite Retrospective Cohort Study.

Intravenous fluid (IVF) is a frequently recommended intervention in Ebola virus disease (EVD), yet its impact on patient outcomes remains unclear. This retrospective cohort study evaluated patients with EVD admitted to 5 Ebola treatment units (ETUs) in West Africa. The primary outcome was the difference in 28-day survival between cases treated and not treated with IVF. To control for demographic and clinical factors related to both IVF exposure and survival, cases were compared using propensity score matching. To control for time-varying patient and treatment factors over the course of ETU care, a marginal structural proportional hazards model (MSPHM) with inverse probability weighting was used to assess for 28-day survival differences. Among 424 EVD-positive cases with data for analysis, 354 (83.5%) were treated with IVF at some point during their ETU admission. Overall, 146 (41.3%) cases treated with IVF survived, whereas 31 (44.9%) cases not treated with any IVF survived (P = .583). Matched propensity score analysis found no significant difference in 28-day survival between cases treated and not treated with IVF during their first 24 and 48 hours of care. Adjusted MSPHM survival analyses also found no significant difference in 28-day survival for cases treated with IVF (27.3%) compared to those not treated with IVF (26.9%) during their entire ETU admission (P = .893). After adjustment for patient- and treatment-specific time-varying factors, there was no significant difference in survival among patients with EVD treated with IVF as compared to those not treated with IVF.

Dabigatran, Rivaroxaban, and Warfarin in the Oldest Adults with Atrial Fibrillation in Taiwan

To compare the effectiveness and safety of reduced-dose dabigatran, reduced-dose rivaroxaban, and warfarin in individuals aged 85 and older with atrial fibrillation (AF). Retrospective cohort study. Taiwan National Health Insurance claims database, 2011∼2015. Individuals with AF aged 85 and older (mean 88.6) with incident use of oral anticoagulants between June 1, 2012 and May 31, 2015 (N=4,722; dabigatran 110 mg, n=1,489; rivaroxaban 15 mg/10 mg, n=1,736; warfarin, n=1,497). Clinical outcomes included all-cause death, cardiovascular death, ischemic stroke, acute myocardial infarction, arterial embolism or thrombosis, intracranial hemorrhage, and gastrointestinal hemorrhage needing transfusion. Propensity score-matched analysis was performed, and the marginal proportional hazards model was used to estimate the relative risk of various clinical outcomes in a matched dabigatran-warfarin cohort (n=1,180 in each group) and a rivaroxaban-warfarin cohort (n=1,207 in each group) RESULTS: Mean follow-up was 6.6 months for the overall population. Dabigatran group participants had lower risks of all-cause death (hazard ratio (HR)=0.59, 95% confidence interval (CI)=0.45-0.77) and cardiovascular death (HR=0.45, 95% CI=0.30-0.68) than warfarin group participants. Rivaroxaban users also had lower risks of all-cause death (HR=0.61, 95% CI=0.47-0.79) and cardiovascular death (HR=0.52, 95% CI=0.35-0.75) than warfarin users. Dabigatran users also had a lower risk of intracranial hemorrhage than warfarin users (HR=0.31, 95% CI=0.10-0.97). Individuals with AF aged 85 and older who used reduced-dose dabigatran or reduced-dose rivaroxaban had statistically significantly lower all-cause mortality and cardiovascular mortality than those who used warfarin. Reduced-dose dabigatran was also associated with lower risk of intracranial hemorrhage than warfarin.

Regression analysis of bivariate current status data under the proportional hazards model

AbstractThis article discusses the regression analysis of bivariate current status or case I interval‐censored failure time data under the marginal proportional hazards model. Several estimation procedures have been proposed for this problem, but each method either applies only to limited situations or no theoretical justification has been provided. Using Bernstein polynomials and the copula model we develop a sieve maximum likelihood estimation approach that applies to more general situations. In particular this method leaves the underlying copula model completely unspecified and can be easily implemented. The proposed estimators are shown to be strongly consistent and the asymptotic normality and efficiency of the regression parameter estimator are established. Simulation studies are conducted to assess the performance of the proposed method. An illustrative example is also provided. The Canadian Journal of Statistics 45: 410–424; 2017 © 2017 Statistical Society of Canada

Systemic therapies for recurrent or metastatic nasopharyngeal carcinoma: a systematic review

Background:The majority of published studies in recurrent or metastatic nasopharyngeal carcinoma (RM-NPC) are single-arm trials. Reliable modelling of progression-free survival (PFS) and overall survival (OS) outcomes, therefore, is difficult. This study aim to analyse existent literature to estimate the relative efficacy of available systemic regimens in RM-NPC, as well as provide estimates of aggregate OS and PFS.Methods:We conducted a systematic search of MEDLINE, EMBASE and the Cochrane Library to March 2015. Clinical trials (in English only) investigating cytotoxic and molecularly targeted agents in adult patients with RM-NPC were included. All relevant studies were assessed for quality using Downs and Blacks (DB) checklist (maximum quality score of 27). Aggregate data analysis and Student’s t-test were performed for all identified studies (model A). For studies that published analysable Kaplan−Meier curves, survival data were extracted and marginal proportional hazards models were constructed (model B).Results:A total of 56 studies were identified and included in model A, 26 of which had analysable Kaplan−Meier curves and were included in model B. The 26 studies in model B had significantly higher mean DB scores than the remaining 30 (17.3 vs 13.7, P=0.002). For patients receiving first line chemotherapy, the estimated median OS was 15.7 months by model A (95% CI, 12.3–19.1), and 19.3 months by model B (95% CI, 17.6–21.1). For patients undergoing second line or higher therapies (2nd+), the estimated median OS was 11.5 months by model A (95% CI 10.1–12.9), and 12.5 months by model B (95% CI 11.9–13.4). PFS estimates for patients undergoing first-line chemotherapy by model A was 7.6 months (95% CI, 6.2–9.0), and 8.0 months by model B (95% CI, 7.6–8.8). For patients undergoing therapy in the 2nd+ setting, the estimated PFS by model A was 5.4 months (95% CI, 3.8–7.0), and 5.2 months by model B (95% CI, 4.7–5.6).Conclusions:We present the first aggregate estimates of OS and PFS for RM-NPC patients receiving first and second-line or higher treatment settings, which could inform the design of future clinical trials in this disease setting.

Event-Free Survival Is a Surrogate for Overall Survival in Patients Treated for Acute Myeloid Leukemia

Purpose: We evaluated whether event-free survival (EFS) can be used as a surrogate for overall survival (OS) in patients treated for acute myeloid leukemia.Material: We carried out a meta-analysis of individual patient data from four randomized clinical trials carried out under the auspices of the German-Austrian Acute Myeloid Leukemia Study Group (AMLSG): AMLHD 98B (Schlenk et al. Leukemia 2004 18:1798-803; n=254), AMLSG 06-04 (NCT00151255, n=189), AMLSG 07-04 (NCT00151242, n=1,100) and AMLSG 12-09 (NCT01180322, n=268). Some of these trials addressed multiple therapeutic questions, which resulted in a total of 7 independent treatment comparisons.Methods: A two-level modelling approach was used to estimate the association between EFS and OS, and between the treatment effects on EFS and on OS. At the individual level, a copula was fitted to model the joint distribution of EFS and OS, and Spearman's rank correlation coefficient (rho) was used to quantify the association between the endpoints. At the trial-level, a linear regression was fitted through the estimated treatment effects (Weibull-model-based log hazard ratios) on EFS and OS, taking into account the estimation error. The coefficient of determination (R²) was used to quantify the association between the treatment effects. The surrogate threshold effect (STE) was estimated as the treatment effect on EFS that would predict a significant treatment effect on OS.Results: A total of n=1,811 patients were included in the analysis. Spearman's correlation coefficient was equal to 0.76 (standard error, SE, 0.015). The coefficient of determination (R²) of the linear regression between log hazard ratios on EFS and on OS was equal to 0.97 (SE 0.13). The intercept of the regression line was equal to -0.04 (SE 0.04) and the slope was equal to 0.80 (SE 0.21). The surrogate threshold effect was equal to 0.90. Using an alternative method of estimation of treatment effects, marginal proportional hazards models for EFS and OS, the R² was equal to 0.98 (SE 0.21), the intercept of the regression line was equal to -0.02 (SE 0.05), the slope was equal to 0.82 (SE 0.28), and the surrogate threshold effect was equal to 0.89. Further results for different subsets of patients, for example, those with activating FLT3 mutations, will be presented at the meeting.Interpretation: In this population of intensively-treated AML patients, there was a tight association between the treatment effect on EFS and OS, suggesting that the former can be used as a surrogate for the latter in clinical trials assessing the efficacy of new treatments. The surrogate threshold effect of about 0.90 and the regression analysis suggest that a reduction of at least 10% in the risk of an event would reliably predict a reduction of approximately 8% in the risk of death. DisclosuresSchlenk:Celgene: Honoraria, Research Funding; Pfizer: Honoraria, Research Funding; Novartis: Honoraria, Research Funding; Teva: Honoraria, Research Funding; AROG: Honoraria, Research Funding; Amgen: Research Funding; Böhringer Ingelheim: Honoraria; Daiichi Sankyo: Membership on an entity's Board of Directors or advisory committees; Janssen: Honoraria, Membership on an entity's Board of Directors or advisory committees. Buyse:IDDI: Employment; Novartis: Research Funding. Burzykowski:IDDI: Employment; Novartis: Research Funding.