Practical methods for competing risks data: a review

Scalable log-ratio lasso regression for enhanced microbial feature selection with FLORAL

Identifying predictive biomarkers of patient outcomes from high-throughput microbiome data is of high interest, while existing computational methods do not satisfactorily account for complex survival endpoints, longitudinal samples, and taxa-specific sequencing biases. We present FLORAL, an open-source tool to perform scalable log-ratio lasso regression and microbial feature selection for continuous, binary, time-to-event, and competing risk outcomes, with compatibility for longitudinal microbiome data as time-dependent covariates. The proposed method adapts the augmented Lagrangian algorithm for a zero-sum constraint optimization problem while enabling a two-stage screening process for enhanced false-positive control. In extensive simulation and real-data analyses, FLORAL achieved consistently better false-positive control compared to other lasso-based approaches and better sensitivity over popular differential abundance testing methods for datasets with smaller sample sizes. In a survival analysis of allogeneic hematopoietic cell transplant recipients, FLORAL demonstrated considerable improvement in microbial feature selection by utilizing longitudinal microbiome data over solely using baseline microbiome data.

Limited impact of colistin resistance on mortality of intensive care patients with carbapenem-resistant bacteraemia

BACKGROUND

Increasing incidence of carbapenem-resistant Gram-negative bacteraemia (CR-GNB) has triggered increased use of polymyxins, likely fuelling the emergence and spread of colistin resistance.

AIM

To estimate the excess clinical burden of colistin resistance in intensive care patients with CR-GNB.

METHODS

A cohort of patients with CR-GNB during their stay in the intensive care unit (ICU) of a university hospital in Greece over a 4-year period (2020-2023) was constructed. Competing risks survival analysis was performed to estimate the burden associated with colistin resistance.

FINDINGS

Of the 177 ICU patients with CR-GNB, 134 (76%) had colistin-resistant isolates, predominantly Acinetobacter baumannii (79%), identified by broth microdilution. Patients with colistin-resistant infection were similar to those with colistin-susceptible infection with respect to age, sex, APACHE II score, Charlson comorbidity index score, Pitt bacteraemia score, prior surgery and the occurrence of polymicrobial cultures. However, patients in the colistin-resistant group had lower risk of mortality compared with those in the colistin-susceptible group (31% vs 44%, P = 0.004 at 14 days, respectively; 46% vs 56% at 28 days, respectively; P = 0.173). Multi-variable regression analysis confirmed that colistin-resistant CR-GNB was associated with significantly lower risk of inpatient death compared with colistin-susceptible CR-GNB within 14 days [cause-specific hazard ratio (csHR) 0.53, 95% CI 0.28-1.01) and 28 days (csHR 0.55, 95% CI 0.31-0.95) of infection onset.

CONCLUSION

Limited impact of colistin resistance on mortality was demonstrated in a large contemporary cohort of ICU patients with CR-GNB, possibly reflecting the recent shift away from colistin-based treatment regimens.

Stress Ulcer Prophylaxis during Invasive Mechanical Ventilation

BACKGROUND

Whether proton-pump inhibitors are beneficial or harmful for stress ulcer prophylaxis in critically ill patients undergoing invasive ventilation is unclear.

METHODS

In this international, randomized trial, we assigned critically ill adults who were undergoing invasive ventilation to receive intravenous pantoprazole (at a dose of 40 mg daily) or matching placebo. The primary efficacy outcome was clinically important upper gastrointestinal bleeding in the intensive care unit (ICU) at 90 days, and the primary safety outcome was death from any cause at 90 days. Multiplicity-adjusted secondary outcomes included ventilator-associated pneumonia, Clostridioides difficile infection, and patient-important bleeding.

RESULTS

A total of 4821 patients underwent randomization in 68 ICUs. Clinically important upper gastrointestinal bleeding occurred in 25 of 2385 patients (1.0%) receiving pantoprazole and in 84 of 2377 patients (3.5%) receiving placebo (hazard ratio, 0.30; 95% confidence interval [CI], 0.19 to 0.47; P<0.001). At 90 days, death was reported in 696 of 2390 patients (29.1%) in the pantoprazole group and in 734 of 2379 patients (30.9%) in the placebo group (hazard ratio, 0.94; 95% CI, 0.85 to 1.04; P = 0.25). Patient-important bleeding was reduced with pantoprazole; all other secondary outcomes were similar in the two groups.

CONCLUSIONS

Among patients undergoing invasive ventilation, pantoprazole resulted in a significantly lower risk of clinically important upper gastrointestinal bleeding than placebo, with no significant effect on mortality. (Funded by the Canadian Institutes of Health Research and others; REVISE ClinicalTrials.gov number, NCT03374800.).

Use of the Win Ratio Analysis in Critical Care Trials

Composite outcomes are commonly used in critical care trials to estimate the treatment effect of an intervention. A significant limitation of classical analytic approaches is that they assign equal statistical importance to each component in a composite, even if these do not have the same clinical importance (i.e., in a composite of death and organ failure, death is clearly more important). The win ratio (WR) method has been proposed as an alternative for trial outcomes evaluation, as it effectively assesses events based on their clinical relevance (i.e., hierarchical order) by comparing each patient in the intervention group with their counterparts in the control group. This statistical approach is increasingly used in cardiovascular outcome trials. However, WR may be useful to unveil treatment effects also in the critical care setting, because these trials are typically moderately sized, thus limiting the statistical power to detect small differences between groups, and often rely on composite outcomes that include several components of different clinical importance. Notably, the advantages of this approach may be offset by several drawbacks (such as ignoring ties and difficulties in selecting and ranking endpoints) and challenges in appropriate clinical interpretation (i.e., establishing clinical meaningfulness of the observed effect size). In this perspective article, we present some key elements to implementing WR statistics in critical care trials, providing an overview of strengths, drawbacks, and potential applications of this method. To illustrate, we conduct a reevaluation of the HYPO-ECMO (Hypothermia during Venoarterial Extracorporeal Membrane Oxygenation) trial using the WR framework as a case example.

Communicating and understanding statistical measures when quantifying the between-group difference in competing risks

Competing risks issues are common in clinical trials and epidemiological studies for patients in follow-up who may experience a variety of possible outcomes. Under such competing risks, two hazard-based statistical methods, cause-specific hazard (CSH) and subdistribution hazard (SDH), are frequently used to assess treatment effects among groups. However, the outcomes of the CSH-based and SDH-based methods have a close connection with the proportional hazards (CSH or SDH) assumption and may have an non-intuitive interpretation. Recently, restricted mean time lost (RMTL) has been used as an alternative summary measure for analysing competing risks, due to its clinical interpretability and robustness to the proportional hazards assumption. Considering the above approaches, we summarize the differences between hazard-based and RMTL-based methods from the aspects of practical interpretation, proportional hazards model assumption and the selection of restricted time points, and propose corresponding suggestions for the analysis of between-group differences under competing risks. Moreover, an R package 'cRMTL' and corresponding step-by-step guidance are available to help users for applying these approaches.

REVISE: re-evaluating the inhibition of stress erosions in the ICU-statistical analysis plan for a randomized trial

BACKGROUND

The REVISE (Re-Evaluating the Inhibition of Stress Erosions in the ICU) trial will evaluate the impact of the proton pump inhibitor pantoprazole compared to placebo in invasively ventilated critically ill patients.

OBJECTIVE

To outline the statistical analysis plan for the REVISE trial.

METHODS

REVISE is a randomized clinical trial ongoing in intensive care units (ICUs) internationally. Patients ≥ 18 years old, receiving invasive mechanical ventilation, and expected to remain ventilated beyond the calendar day after randomization are allocated to either 40 mg pantoprazole intravenously or placebo while mechanically ventilated.

RESULTS

The primary efficacy outcome is clinically important upper GI bleeding; the primary safety outcome is 90-day mortality. Secondary outcomes are ventilator-associated pneumonia, Clostridioides difficile infection, new renal replacement therapy, ICU and hospital mortality, and patient-important GI bleeding. Tertiary outcomes are total red blood cells transfused, peak serum creatinine concentration, and duration of mechanical ventilation, ICU, and hospital length of stay. Following an interim analysis of results from 2400 patients (50% of 4800 target sample size), the data monitoring committee recommended continuing enrolment.

CONCLUSIONS

This statistical analysis plan outlines the statistical analyses of all outcomes, sensitivity analyses, and subgroup analyses. REVISE will inform clinical practice and guidelines worldwide.

TRIAL REGISTRATION

www.

CLINICALTRIALS

gov NCT03374800. November 21, 2017.

Developing prediction models to estimate the risk of two survival outcomes both occurring: A comparison of techniques

INTRODUCTION

This study considers the prediction of the time until two survival outcomes have both occurred. We compared a variety of analytical methods motivated by a typical clinical problem of multimorbidity prognosis.

METHODS

We considered five methods: product (multiply marginal risks), dual-outcome (directly model the time until both events occur), multistate models (msm), and a range of copula and frailty models. We assessed calibration and discrimination under a variety of simulated data scenarios, varying outcome prevalence, and the amount of residual correlation. The simulation focused on model misspecification and statistical power. Using data from the Clinical Practice Research Datalink, we compared model performance when predicting the risk of cardiovascular disease and type 2 diabetes both occurring.

RESULTS

Discrimination was similar for all methods. The product method was poorly calibrated in the presence of residual correlation. The msm and dual-outcome models were the most robust to model misspecification but suffered a drop in performance at small sample sizes due to overfitting, which the copula and frailty model were less susceptible to. The copula and frailty model's performance were highly dependent on the underlying data structure. In the clinical example, the product method was poorly calibrated when adjusting for 8 major cardiovascular risk factors.

DISCUSSION

We recommend the dual-outcome method for predicting the risk of two survival outcomes both occurring. It was the most robust to model misspecification, although was also the most prone to overfitting. The clinical example motivates the use of the methods considered in this study.

Competing risk models to evaluate the factors for time to loss to follow-up among tuberculosis patients at Ambo General Hospital

BACKGROUND

A major challenge for most tuberculosis programs is the inability of tuberculosis patients to complete treatment for one reason or another. Failure to complete the treatment contributes to the emergence of multidrug-resistant TB. This study aimed to evaluate the risk factors for time to loss to follow-up treatment by considering death as a competing risk event among tuberculosis patients admitted to directly observed treatment short course at Ambo General Hospital, Ambo, Ethiopia.

METHODS

Data collected from 457 tuberculosis patients from January 2018 to January 2022 were used for the analysis. The cause-specific hazard and sub-distribution hazard models for competing risks were used to model the outcome of interest and to identify the prognostic factors associated to treatment loss to follow-up. Loss to follow-up was used as an outcome measure and death as a competing event.

RESULTS

Of the 457 tuberculosis patients enrolled, 54 (11.8%) were loss to follow-up their treatment and 33 (7.2%) died during the follow up period. The median time of loss to follow-up starting from the date of treatment initiation was 4.2 months. The cause-specific hazard and sub-distribution hazard models revealed that sex, place of residence, HIV status, contact history, age and baseline weights of patients were significant risk factors associated with time to loss to follow-up treatment. The findings showed that the estimates of the covariates effects were different for the cause specific and sub-distribution hazard models. The maximum relative difference observed for the covariate between the cause specific and sub-distribution hazard ratios was 12.2%.

CONCLUSIONS

Patients who were male, rural residents, HIV positive, and aged 41 years or older were at higher risk of loss to follow-up their treatment. This underlines the need that tuberculosis patients, especially those in risk categories, be made aware of the length of the directly observed treatment short course and the effects of discontinuing treatment.

Fallacies of Using the Win Ratio in Cardiovascular Trials: Challenges and Solutions

The win ratio was introduced into cardiovascular trials as a potentially better way of analyzing composite endpoints to account for the hierarchy of clinical significance of their components and to facilitate the inclusion of recurrent events. The basic concept of the win ratio is to define a hierarchy of clinical importance within the components of the composite outcome, form all possible pairs by comparing every subject in the treatment group with every subject in the control group, and then evaluate each pair for the occurrence of the components of the composite outcome in descending order of importance, starting at the most important and progressing down the hierarchy if the outcome does not result in a win in either pair until pairs are tied for the outcome after exhaustion of all components. Although the win ratio offers a novel method of depiction of outcomes in clinical trials, its advantages may be counterbalanced by several fallacies (such as ignoring ties and weighting each hierarchal component equally) and challenges in appropriate clinical interpretation (establishing clinical meaningfulness of the observed effect size). From this perspective, we discuss these and other fallacies and provide a suggested framework to overcome such limitations to enhance utility of this statistical method across the clinical trial enterprise.

Classifying Polypharmacy According to Pharmacotherapeutic and Clinical Risks in Older Adults: A Latent Class Analysis in Quebec, Canada

INTRODUCTION

The simplistic definition of polypharmacy, often designated as the concomitant use of five medications or more, does not distinguish appropriate from inappropriate polypharmacy. Classifying polypharmacy according to varying levels of health risk would help optimise medication use.

OBJECTIVE

We aimed to characterise different types of polypharmacy among older adults and evaluate their association with mortality and institutionalisation.

METHODS

Using healthcare databases from the Quebec Integrated Chronic Disease Surveillance System, we selected a community-based random sample of the population ≥ 66 years old covered by the public drug plan. Categorical indicators used to describe polypharmacy included number of medications, potentially inappropriate medications (PIMs), drug-drug interactions, enhanced surveillance medications, complex route of administration medications, anticholinergic cognitive burden (ACB) score and use of blister cards. We used a latent class analysis to subdivide participants into distinct groups of polypharmacy. Their association with 3-year mortality and institutionalisation was assessed with adjusted Cox models.

RESULTS

In total, 93,516 individuals were included. A four-class model was selected with groups described as (1) no polypharmacy (46% of our sample), (2) high-medium number of medications, low risk (33%), (3) medium number of medications, PIM use with or without high ACB score (8%) and (4) hyperpolypharmacy, complex use, high risk (13%). Using the class without polypharmacy as the reference, all polypharmacy classes were associated with 3-year mortality and institutionalisation, with the most complex/inappropriate classes denoting the highest risk (hazard ratio [HR] [95% confidence interval]: class 3, 70-year-old point estimate for mortality 1.52 [1.30-1.78] and institutionalisation 1.86 [1.52-2.29]; class 4, 70-year-old point estimate for mortality 2.74 [2.44-3.08] and institutionalisation 3.11 [2.60-3.70]).

CONCLUSIONS

We distinguished three types of polypharmacy with varying pharmacotherapeutic and clinical appropriateness. Our results highlight the value of looking beyond the number of medications to assess polypharmacy.

High-dimensional feature selection in competing risks modeling: A stable approach using a split-and-merge ensemble algorithm

Variable selection is critical in competing risks regression with high-dimensional data. Although penalized variable selection methods and other machine learning-based approaches have been developed, many of these methods often suffer from instability in practice. This paper proposes a novel method named Random Approximate Elastic Net (RAEN). Under the proportional subdistribution hazards model, RAEN provides a stable and generalizable solution to the large-p-small-n variable selection problem for competing risks data. Our general framework allows the proposed algorithm to be applicable to other time-to-event regression models, including competing risks quantile regression and accelerated failure time models. We show that variable selection and parameter estimation improved markedly using the new computationally intensive algorithm through extensive simulations. A user-friendly R package RAEN is developed for public use. We also apply our method to a cancer study to identify influential genes associated with the death or progression from bladder cancer.

The prevalence of hyperglycemia and its impact on mortality among people living with HIV in Georgia

BACKGROUND

Life expectancy and quality of life of people living with HIV have been dramatically improved after introducing antiretroviral therapy, and the prevalence of non-communicable diseases has increased. Several studies have found that hyperglycemia with or without type 2 diabetes was associated with poor outcomes in people living with HIV. The study's objective was to determine the prevalence of hyperglycemia and assess its impact on mortality.

MATERIALS AND METHODS

A retrospective cohort study was conducted among people living with HIV diagnosed in 2012-2018 and followed through 2020 at the Infectious Diseases, AIDS and Clinical Immunology Research Center in Tbilisi, Georgia. Primary outcomes of interest included the prevalence of hyperglycemia and mortality. Causes of death were classified according to the Coding of Death in HIV (CoDe) protocol.

RESULTS

Our study included 2914 people living with HIV. Two hundred and forty-two (8.3%) patients had hyperglycemia, with an increasing prevalence by age. Three hundred one (9.7%) participants died over the median 3.71 (IQR: 2.14-5.37) years of follow-up. Among these, 139 (46.2%) were due to AIDS- related causes, 123 (40.9%)-were due to non-AIDS causes, and in 39 (12.9%) cases, the cause of death could not be determined. Overall, the cohort contributed to 11,148 person-years of follow-up (PYFU), translating into a mortality rate of 2.70 deaths per 100 PYFU. The mortality rate was significantly higher among individuals with hyperglycemia-11.17 deaths per 100 PYFU vs 2.07 deaths per 100 PYFU among normoglycemic patients(p<0.0001).

CONCLUSIONS

Hyperglycemia was associated with increased odds of mortality. Screening and management of hyperglycemia should be integrated into routine HIV clinical services as part of a comprehensive care package.

Survival probability and factors associated with time to loss to follow-up and mortality among patients on antiretroviral treatment in central Kenya

BACKGROUND

Retention of patients who are receiving antiretroviral therapy (ART) remains a challenge especially in the setting of rapid expansion of HIV services. Retention in care remains vital to the HIV care continuum, and has been associated with viral suppression and improved survival. This study aimed to ascertain survival rates, time to loss to follow-up (LTFU) or mortality events and factors associated with time to LTFU or mortality among patients enrolled on antiretroviral therapy at health facilities in central Kenya.

METHODS

This was a retrospective cohort study among patients initiated on ART between 2004 and 2012 in central Kenya. Demographic characteristics, clinical characteristics and outcomes data were analyzed using Stata version 15.1. Competing risks regression analysis and cummulative incidence functions were used to estimate survival.

RESULTS

A total of 31,346 patients were included, of whom 65.6% were female, 76.0% were aged between 20 and 50 years old, and 38.9% were diagnosed at WHO stage III. At 36 months, overall retention was 68.8%, LTFU was 27.1%, and mortality was 4.1%. The total person-years of follow up was 74,986. The incidence rate of LTFU was 9.99 per 100 person years for a total of 9383.25 person-years of follow up. The mortality rate was 1.25 per 100 person years for a total of 875.5 person-years among those who died. The median time to LTFU was 11 months (IQR 3-22) while median time to death was 3 months (IQR 0-13). Men, unmarried patients, patients presenting with advanced HIV, not on TB treatment, and enrolled into the HIV program in later cohorts, had a shorter time to mortality and LTFU.

CONCLUSION

Our study demonstrated evidence of scale-up of HIV treatment programs in central Kenya. While most patients were enrolled at an advanced WHO clinical stage, overall 36-month mortality remained low, but occurred earlier during follow-up. Cohort LTFU at 36-months reduced in later years with the losses occurring within the 1st year of follow-up. Predictors of early mortality and LTFU included being male, single, separated or divorced, advanced WHO clinical stage, and among patients not on TB treatment.

Factors Associated with Treatment Outcomes Among Children and Adolescents Living with HIV Receiving Antiretroviral Therapy in Central Kenya

Expanded access to HIV treatment services has improved outcomes for children and adolescents living with HIV in Kenya. Minimal data are available on these outcomes. We describe temporal trends in outcomes for children and adolescents initiating antiretroviral therapy (ART) from 2004 to 2014 at sites supported by Centre for Health Solutions—Kenya, in central Kenya. We retrospectively analyzed data from children 0–9 years of age (n = 3,519) and adolescents 10–19 years of age (n = 1,663) living with HIV, who newly initiated ART at 47 health facilities in central Kenya. Year cohorts were analyzed from the Comprehensive Patient Application Database (CPAD) and International Quality Care (IQCare) electronic medical databases, including temporal trends in outcomes and associated factors using multivariable competing risk regression analysis. There were more girls (2,453 [52.7%]) than boys, with most enrolled at World Health Organization (WHO) stage II (1,813 [37.7%]) or III disease (1,694 [35.1%]). Most of the children and adolescents (4,431 [96.4%]) did not have tuberculosis (TB) symptoms. Cumulative lost to follow-up (LTFU) incidence at 6, 12, 24, and 36 months were 5.0%, 9.9%, 22.9%, and 33.1%, respectively. Cumulative mortality incidence at 6, 12, 24, and 36 months were 0.7%, 1.0%, 1.2%, and 1.5%, respectively. The incidence of LTFU was higher among female children and adolescents, those initiated on tenofovir-based regimens, and those with presumptive TB symptoms. Mortality risk was higher among those with WHO stage III or IV disease, and children and adolescents on TB treatment or who had presumptive TB. Enrollment occurred at a young age and pediatric-friendly ART regimens were initiated at earlier WHO stages implying effective early infant diagnosis and treatment for all strategies, resulting in improved treatment outcomes. The higher retention rates in recent years as well as the lower retention after many years of follow-up underscore the importance of implementing longitudinal follow-up strategies targeting this population.

Predictors of Recurrence and Survival in Cancer Patients With Pericardial Effusion Requiring Pericardiocentesis

Aim

This study investigated the factors predicting survival and the recurrence of pericardial effusion (PE) requiring pericardiocentesis (PCC) in patients with cancer.

Materials and Methods

We analyzed the data of patients who underwent PCC for large PEs from 2010 to 2020 at The University of Texas MD Anderson Cancer Center. The time to the first recurrent PE requiring PCC was the interval from the index PCC with pericardial drain placement to first recurrent PE requiring drainage (either repeated PCC or a pericardial window). Univariate and multivariate Fine-Gray models accounting for the competing risk of death were used to identify predictors of recurrent PE requiring drainage. Cox regression models were used to identify predictors of death.

Results

The study cohort included 418 patients with index PCC and pericardial drain placement, of whom 65 (16%) had recurrent PEs requiring drainage. The cumulative incidences of recurrent PE requiring drainage at 12 and 60 months were 15.0% and 15.6%, respectively. Younger age, anti-inflammatory medication use, and solid tumors were associated with an increased risk of recurrence of PE requiring drainage, and that echocardiographic evidence of tamponade at presentation and receipt of immunotherapy were associated with a decreased risk of recurrence. Factors predicting poor survival included older age, malignant effusion on cytology, non-use of anti-inflammatory agents, non-lymphoma cancers and primary lung cancer.

Conclusion

Among cancer patients with large PEs requiring drainage, young patients with solid tumors were more likely to experience recurrence, while elderly patients and those with lung cancer, malignant PE cytology, and non-use of anti-inflammatory agents showed worse survival.

Driving Pressure Is Associated With Outcome in Pediatric Acute Respiratory Failure

OBJECTIVES

Driving pressure (ratio of tidal volume over respiratory system compliance) is associated with mortality in acute respiratory distress syndrome. We sought to evaluate if such association could be identified in critically ill children.

DESIGN

We studied the association between driving pressure on day 1 of mechanical ventilation and ventilator-free days at day 28 through secondary analyses of prospectively collected physiology data.

SETTING

Medical-surgical university hospital PICU.

PATIENTS

Children younger than 18 years (stratified by Pediatric Mechanical Ventilation Consensus Conference clinical phenotype definitions) without evidence of spontaneous respiration.

INTERVENTIONS

Inspiratory hold maneuvers.

MEASUREMENTS AND MAIN RESULTS

Data of 222 patients with median age 11 months (2-51 mo) were analyzed. Sixty-five patients (29.3%) met Pediatric Mechanical Ventilation Consensus Conference criteria for restrictive and 78 patients (35.1%) for mixed lung disease, and 10.4% of all patients had acute respiratory distress syndrome. Driving pressure calculated by the ratio of tidal volume over respiratory system compliance for the whole cohort was 16 cm H2O (12-21 cm H2O) and correlated with the static airway pressure gradient (plateau pressure minus positive end-expiratory pressure) (Spearman correlation coefficient = 0.797; p < 0.001). Bland-Altman analysis showed that the dynamic pressure gradient (peak inspiratory pressure minus positive end-expiratory pressure) overestimated driving pressure (levels of agreement -2.295 to 7.268). Rematching the cohort through a double stratification procedure (obtaining subgroups of patients with matched mean levels for one variable but different mean levels for another ranking variable) showed a reduction in ventilator-free days at day 28 with increasing driving pressure in patients ventilated for a direct pulmonary indication. Competing risk regression analysis showed that increasing driving pressure remained independently associated with increased time to extubation (p < 0.001) after adjusting for Pediatric Risk of Mortality III 24-hour score, presence of direct pulmonary indication jury, and oxygenation index.

CONCLUSIONS

Higher driving pressure was independently associated with increased time to extubation in mechanically ventilated children. Dynamic assessments of driving pressure should be cautiously interpreted.

Age at Time of Kidney Transplantation as a Predictor for Mortality, Graft Loss and Self-Rated Health Status: Results From the Swiss Transplant Cohort Study

Introduction: The effect of age on health outcomes in kidney transplantation remains inconclusive. This study aimed to analyze the relationship between age at time of kidney transplantation with mortality, graft loss and self-rated health status in adult kidney transplant recipients.

Methods: This study used data from the Swiss Transplant Cohort Study and included prospective data of kidney transplant recipients between 2008 and 2017. Time-to-event analysis was performed using Cox’ regression analysis, and -in the case of graft loss- competing risk analysis. A random-intercept regression model was applied to analyse self-rated health status.

Results: We included 2,366 kidney transplant recipients. Age at transplantation linearly predicted mortality. It was also predictive for graft loss, though nonlinearly, showing that recipients aged between 35 and 55 years presented with the lowest risk of experiencing graft loss. No relationship of age with self-rated health status was detected.

Conclusion: Higher mortality in older recipients complies with data from the general population. The non-linear relationship between age and graft loss and the higher scored self-rated health status at all follow-up time-points compared to the pre-transplant status -regardless of age- highlight that age alone might not be an accurate measure for risk prediction and clinical decision making in kidney transplantation.

Weighted least-squares regression with competing risks data

The semiparametric accelerated failure time (AFT) model linearly relates the logarithm of the failure time to a set of covariates, while leaving the error distribution unspecified. This model has been widely investigated in survival literature due to its simple interpretation and relationship with linear models. However, there has been much less focus on developing AFT-type linear regression methods for analyzing competing risks data, in which patients can potentially experience one of multiple failure causes. In this article, we propose a simple least-squares (LS) linear regression model for a cause-specific subdistribution function, where the conventional LS equation is modified to account for data incompleteness under competing risks. The proposed estimators are shown to be consistent and asymptotically normal with consistent estimation of the variance-covariance matrix. We further extend the proposed methodology to risk prediction and analysis under clustered competing risks scenario. Simulation studies suggest that the proposed method provides rapid and valid statistical inferences and predictions. Application of our method to two oncology datasets demonstrate its utility in routine clinical data analysis.

Estimating age-specific COVID-19 fatality risk and time to death by comparing population diagnosis and death patterns: Australian data

Background

Mortality is a key component of the natural history of COVID-19 infection. Surveillance data on COVID-19 deaths and case diagnoses are widely available in the public domain, but they are not used to model time to death because they typically do not link diagnosis and death at an individual level. This paper demonstrates that by comparing the unlinked patterns of new diagnoses and deaths over age and time, age-specific mortality and time to death may be estimated using a statistical method called deconvolution.

Methods

Age-specific data were analysed on 816 deaths among 6235 cases over age 50 years in Victoria, Australia, from the period January through December 2020. Deconvolution was applied assuming logistic dependence of case fatality risk (CFR) on age and a gamma time to death distribution. Non-parametric deconvolution analyses stratified into separate age groups were used to assess the model assumptions.

Results

It was found that age-specific CFR rose from 2.9% at age 65 years (95% CI:2.2 – 3.5) to 40.0% at age 95 years (CI: 36.6 – 43.6). The estimated mean time between diagnosis and death was 18.1 days (CI: 16.9 – 19.3) and showed no evidence of varying by age (heterogeneity P = 0.97). The estimated 90% percentile of time to death was 33.3 days (CI: 30.4 – 36.3; heterogeneity P = 0.85). The final age-specific model provided a good fit to the observed age-stratified mortality patterns.

Conclusions

Deconvolution was demonstrated to be a powerful analysis method that could be applied to extensive data sources worldwide. Such analyses can inform transmission dynamics models and CFR assessment in emerging outbreaks. Based on these Australian data it is concluded that death from COVID-19 occurs within three weeks of diagnosis on average but takes five weeks in 10% of fatal cases. Fatality risk is negligible in the young but rises above 40% in the elderly, while time to death does not seem to vary by age.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01314-w.

A tractable Bayesian joint model for longitudinal and survival data

We introduce a numerically tractable formulation of Bayesian joint models for longitudinal and survival data. The longitudinal process is modeled using generalized linear mixed models, while the survival process is modeled using a parametric general hazard structure. The two processes are linked by sharing fixed and random effects, separating the effects that play a role at the time scale from those that affect the hazard scale. This strategy allows for the inclusion of nonlinear and time-dependent effects while avoiding the need for numerical integration, which facilitates the implementation of the proposed joint model. We explore the use of flexible parametric distributions for modeling the baseline hazard function which can capture the basic shapes of interest in practice. We discuss prior elicitation based on the interpretation of the parameters. We present an extensive simulation study, where we analyze the inferential properties of the proposed models, and illustrate the trade-off between flexibility, sample size, and censoring. We also apply our proposal to two real data applications in order to demonstrate the adaptability of our formulation both in univariate time-to-event data and in a competing risks framework. The methodology is implemented in rstan.

Optimal treatment regimes for competing risk data using doubly robust outcome weighted learning with bi-level variable selection

The goal of the optimal treatment regime is maximizing treatment benefits via personalized treatment assignments based on the observed patient and treatment characteristics. Parametric regression-based outcome learning approaches require exploring complex interplay between the outcome and treatment assignments adjusting for the patient and treatment covariates, yet correctly specifying such relationships is challenging. Thus, a robust method against misspecified models is desirable in practice. Parsimonious models are also desired to pursue a concise interpretation and to avoid including spurious predictors of the outcome or treatment benefits. These issues have not been comprehensively addressed in the presence of competing risks. Recognizing that competing risks and group variables are frequently present, we propose a doubly robust estimation with adaptive L 1 penalties to select important variables at both group and within-group levels for competing risks data. The proposed method is applied to hematopoietic cell transplantation data to personalize the graft source choice for treatment-related mortality (TRM). While the existing medical literature attempts to find a uniform solution ignoring the heterogeneity of the graft source effects on TRM, the analysis results show the effect of the graft source on TRM could be different depending on the patient-specific characteristics.

Time-To-Event Data: An Overview and Analysis Considerations

In oncology, overall survival and progression-free survival are common time-to-event end points used to measure treatment efficacy. Analyses of this type of data rely on a complex statistical framework and the analysis results are only valid when the data meet certain assumptions. This article provides an overview of time-to-event data, the basic mechanics of common analysis methods, and issues often encountered when analyzing such data. Our goal is to provide clinicians and other lung cancer researchers with the knowledge to choose the appropriate time-to-event analysis methods and to interpret the outcomes of such analyses appropriately. We strongly encourage investigators to seek out statisticians with expertise in survival analysis when embarking on studies that include time-to-event data to ensure that their data are collected and analyzed using the appropriate methods.

Modeling Survival in Patients With Brain Stroke in the Presence of Competing Risks

Objectives

After heart disease, brain stroke (BS) is the second most common cause of death worldwide, underscoring the importance of understanding preventable and treatable risk factors for the outcomes of BS. This study aimed to model the survival of patients with BS in the presence of competing risks.

Methods

This longitudinal study was conducted on 332 patients with a definitive diagnosis of BS. Demographic characteristics and risk factors were collected by a validated checklist. Patients’ mortality status was investigated by telephone follow-up to identify deaths that may be have been caused by stroke or other factors (heart disease, diabetes, high cholesterol, etc.). Data were analyzed by the Lunn-McNeil approach at alpha=0.1.

Results

Older age at diagnosis (59–68 years: adjusted hazard ratio [aHR], 2.19; 90% confidence interval [CI], 1.38 to 3.48; 69–75 years: aHR, 5.04; 90% CI, 3.25 to 7.80; ≥76 years: aHR, 5.30; 90% CI, 3.40 to 8.44), having heart disease (aHR, 1.65; 90% CI, 1.23 to 2.23), oral contraceptive pill use (women only) (aHR, 0.44; 90% CI, 0.24 to 0.78) and ischemic stroke (aHR, 0.52; 90% CI, 0.36 to 0.74) were directly related to death from BS. Older age at diagnosis (59–68 years: aHR, 21.42; 90% CI, 3.52 to 130.39; 75–69 years: aHR, 16.48; 90% CI, 2.75 to 98.69; ≥76 years: aHR, 26.03; 90% CI, 4.06 to 166.93) and rural residence (aHR, 2.30; 90% CI, 1.15 to 4.60) were directly related to death from other causes. Significant risk factors were found for both causes of death.

Conclusions

BS-specific and non-BS-specific mortality had different risk factors. These findings could be utilized to prescribe optimal and specific treatment.

Semiparametric regression on cumulative incidence function with interval-censored competing risks data and missing event types

Competing risk data are frequently interval-censored, that is, the exact event time is not observed but only known to lie between two examination time points such as clinic visits. In addition to interval censoring, another common complication is that the event type is missing for some study participants. In this article, we propose an augmented inverse probability weighted sieve maximum likelihood estimator for the analysis of interval-censored competing risk data in the presence of missing event types. The estimator imposes weaker than usual missing at random assumptions by allowing for the inclusion of auxiliary variables that are potentially associated with the probability of missingness. The proposed estimator is shown to be doubly robust, in the sense that it is consistent even if either the model for the probability of missingness or the model for the probability of the event type is misspecified. Extensive Monte Carlo simulation studies show good performance of the proposed method even under a large amount of missing event types. The method is illustrated using data from an HIV cohort study in sub-Saharan Africa, where a significant portion of events types is missing. The proposed method can be readily implemented using the new function ciregic_aipw in the R package intccr.

Coronavirus 2019 and People Living With Human Immunodeficiency Virus: Outcomes for Hospitalized Patients in New York City

BACKGROUND

There are limited data regarding the clinical impact of coronavirus disease 2019 (COVID-19) on people living with human immunodeficiency virus (PLWH). In this study, we compared outcomes for PLWH with COVID-19 to a matched comparison group.

METHODS

We identified 88 PLWH hospitalized with laboratory-confirmed COVID-19 in our hospital system in New York City between 12 March and 23 April 2020. We collected data on baseline clinical characteristics, laboratory values, HIV status, treatment, and outcomes from this group and matched comparators (1 PLWH to up to 5 patients by age, sex, race/ethnicity, and calendar week of infection). We compared clinical characteristics and outcomes (death, mechanical ventilation, hospital discharge) for these groups, as well as cumulative incidence of death by HIV status.

RESULTS

Patients did not differ significantly by HIV status by age, sex, or race/ethnicity due to the matching algorithm. PLWH hospitalized with COVID-19 had high proportions of HIV virologic control on antiretroviral therapy. PLWH had greater proportions of smoking (P < .001) and comorbid illness than uninfected comparators. There was no difference in COVID-19 severity on admission by HIV status (P = .15). Poor outcomes for hospitalized PLWH were frequent but similar to proportions in comparators; 18% required mechanical ventilation and 21% died during follow-up (compared with 23% and 20%, respectively). There was similar cumulative incidence of death over time by HIV status (P = .94).

CONCLUSIONS

We found no differences in adverse outcomes associated with HIV infection for hospitalized COVID-19 patients compared with a demographically similar patient group.

A semiparametric method for the analysis of outcomes during a gap in HIV care under incomplete outcome ascertainment

Objectives

Estimation of the cascade of HIV care is essential for evaluating care and treatment programs, informing policy makers and assessing targets such as 90-90-90. A challenge to estimating the cascade based on electronic health record concerns patients "churning" in and out of care. Correctly estimating this dynamic phenomenon in resource-limited settings, such as those found in sub-Saharan Africa, is challenging because of the significant death under-reporting. An approach to partially recover information on the unobserved deaths is a double-sampling design, where a small subset of individuals with a missed clinic visit is intensively outreached in the community to actively ascertain their vital status. This approach has been adopted in several programs within the East Africa regional IeDEA consortium, the context of our motivating study. The objective of this paper is to propose a semiparametric method for the analysis of competing risks data with incomplete outcome ascertainment.

Methods

Based on data from double-sampling designs, we propose a semiparametric inverse probability weighted estimator of key outcomes during a gap in care, which are crucial pieces of the care cascade puzzle.

Results

Simulation studies suggest that the proposed estimators provide valid estimates in settings with incomplete outcome ascertainment under a set of realistic assumptions. These studies also illustrate that a naïve complete-case analysis can provide seriously biased estimates. The methodology is applied to electronic health record data from the East Africa IeDEA Consortium to estimate death and return to care during a gap in care.

Conclusions

The proposed methodology provides a robust approach for valid inferences about return to care and death during a gap in care, in settings with death under-reporting. Ultimately, the resulting estimates will have significant consequences on program construction, resource allocation, policy and decision making at the highest levels.

Association Between Alzheimer Disease and Cancer With Evaluation of Study Biases: A Systematic Review and Meta-analysis

IMPORTANCE

Observational studies consistently report inverse associations between cancer and Alzheimer disease (AD). Shared inverse etiological mechanisms might explain this phenomenon, but a systematic evaluation of methodological biases in existing studies is needed.

OBJECTIVES

To systematically review and meta-analyze evidence on the association between cancer and subsequent AD, systematically identify potential methodological biases in studies, and estimate the influence of these biases on the estimated pooled association between cancer and AD.

DATA SOURCES

All-language publications were identified from PubMed, Embase, and PsycINFO databases through September 2, 2020.

STUDY SELECTION

Longitudinal cohort studies and case-control studies on the risk of AD in older adults with a history of any cancer type, prostate cancer, breast cancer, colorectal cancer, or nonmelanoma skin cancer, relative to those with no cancer history.

DATA EXTRACTION AND SYNTHESIS

Two reviewers independently abstracted the data and evaluated study biases related to confounding, diagnostic bias, competing risks, or survival bias. Random-effects meta-analysis was used to provide pooled estimates of the association between cancer and AD. Metaregressions were used to evaluate whether the observed pooled estimate could be attributable to each bias. The study was designed and conducted according to the Preferring Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline.

MAIN OUTCOMES AND MEASURES

Incidence, hazard, or odds ratios for AD comparing older adults with vs without a previous cancer diagnosis.

RESULTS

In total, 19 cohort studies and 3 case-control studies of the associations between any cancer type (n = 13), prostate cancer (n = 5), breast cancer (n = 1), and nonmelanoma skin cancer (n = 3) with AD were identified, representing 9 630 435 individuals. In all studies combined, cancer was associated with decreased AD incidence (cohort studies: random-effects hazard ratio, 0.89; 95% CI, 0.79-1.00; case-control studies: random-effects odds ratio, 0.75; 95% CI, 0.61-0.93). Studies with insufficient or inappropriate confounder control or greater likelihood of AD diagnostic bias had mean hazard ratios closer to the null value, indicating that these biases could not explain the observed inverse association. Competing risks bias was rare. Studies with greater likelihood of survival bias had mean hazard ratios farther from the null value.

CONCLUSIONS AND RELEVANCE

The weak inverse association between cancer and AD may reflect shared inverse etiological mechanisms or survival bias but is not likely attributable to diagnostic bias, competing risks bias, or insufficient or inappropriate control for potential confounding factors.

Energy transmission in mechanically ventilated children: a translational study

Background

Recurrent delivery of tidal mechanical energy (ME) inflicts ventilator-induced lung injury (VILI) when stress and strain exceed the limits of tissue tolerance. Mechanical power (MP) is the mathematical description of the ME delivered to the respiratory system over time. It is unknown how ME relates to underlying lung pathology and outcome in mechanically ventilated children. We therefore tested the hypothesis that ME per breath with tidal volume (Vt) normalized to bodyweight correlates with underlying lung pathology and to study the effect of resistance on the ME dissipated to the lung.

Methods

We analyzed routinely collected demographic, physiological, and laboratory data from deeply sedated and/or paralyzed children < 18 years with and without lung injury. Patients were stratified into respiratory system mechanic subgroups according to the Pediatric Mechanical Ventilation Consensus Conference (PEMVECC) definition. The association between MP, ME, lung pathology, and duration of mechanical ventilation as a primary outcome measure was analyzed adjusting for confounding variables and effect modifiers. The effect of endotracheal tube diameter (ETT) and airway resistance on energy dissipation to the lung was analyzed in a bench model with different lung compliance settings.

Results

Data of 312 patients with a median age of 7.8 (1.7–44.2) months was analyzed. Age (p <  0.001), RR p <  0.001), and Vt <  0.001) were independently associated with MPrs. ME but not MP correlated significantly (p <  0.001) better with lung pathology. Competing risk regression analysis adjusting for PRISM III 24 h score and PEMVECC stratification showed that ME on day 1 or day 2 of MV but not MP was independently associated with the duration of mechanical ventilation. About 33% of all energy generated by the ventilator was transferred to the lung and highly dependent on lung compliance and airway resistance but not on endotracheal tube size (ETT) during pressure control (PC) ventilation.

Conclusions

ME better related to underlying lung pathology and patient outcome than MP. The delivery of generated energy to the lung was not dependent on ETT size during PC ventilation. Further studies are needed to identify injurious MErs thresholds in ventilated children.

Semiparametric regression and risk prediction with competing risks data under missing cause of failure

The cause of failure in cohort studies that involve competing risks is frequently incompletely observed. To address this, several methods have been proposed for the semiparametric proportional cause-specific hazards model under a missing at random assumption. However, these proposals provide inference for the regression coefficients only, and do not consider the infinite dimensional parameters, such as the covariate-specific cumulative incidence function. Nevertheless, the latter quantity is essential for risk prediction in modern medicine. In this paper we propose a unified framework for inference about both the regression coefficients of the proportional cause-specific hazards model and the covariate-specific cumulative incidence functions under missing at random cause of failure. Our approach is based on a novel computationally efficient maximum pseudo-partial-likelihood estimation method for the semiparametric proportional cause-specific hazards model. Using modern empirical process theory we derive the asymptotic properties of the proposed estimators for the regression coefficients and the covariate-specific cumulative incidence functions, and provide methodology for constructing simultaneous confidence bands for the latter. Simulation studies show that our estimators perform well even in the presence of a large fraction of missing cause of failures, and that the regression coefficient estimator can be substantially more efficient compared to the previously proposed augmented inverse probability weighting estimator. The method is applied using data from an HIV cohort study and a bladder cancer clinical trial.

Excess mortality due to pandrug-resistant Acinetobacter baumannii infections in hospitalized patients

BACKGROUND

Pandrug-resistant Acinetobacter baumannii (PDRAB) is increasingly being reported as a nosocomial pathogen worldwide, but determining its clinical impact is challenging.

AIM

To assess the spectrum of excess mortality attributable to PDRAB infection in acute care settings.

METHODS

This four-year cohort study was conducted in a tertiary-care referral hospital in Greece to estimate excess in-hospital mortality due to PDRAB infection by comparing patients infected to those colonized with PDRAB by means of competing risks survival analysis.

FINDINGS

The study cohort comprised 91 patients (median age: 67 years; 77% men). For most patients, PDRAB was first isolated in the intensive care unit (ICU) (N = 51; 57%) or following ICU discharge (N = 26; 29%). Overall in-hospital mortality was 68% (95% confidence interval (CI): 57.5-77.5%). PDRAB-infected patients (N = 62; 68%) and PDRAB-colonized patients (N = 29; 32%) had similar baseline characteristics, but the absolute excess risk of 30-day mortality in infected patients compared to colonized patients was 34% (95% CI: 14-54%). Multivariable competing risks regression showed that PDRAB infection significantly increased the daily hazard of 30-day in-hospital death (cause-specific hazard ratio (csHR): 3.10; 95% CI: 1.33-7.21) while simultaneously decreasing the daily rate of discharge (csHR: 0.24; 95% CI: 0.08-0.74), thereby leading to longer hospitalization. Stronger effects were observed for bloodstream infections.

CONCLUSION

New effective antimicrobials would be expected to prevent mortality in one of every three patients treated for PDRAB infection and reduce their length of hospitalization. However, available therapeutic options remain extremely limited and emphasis on preventing healthcare-associated transmission of PDRAB is ever more important.

The use of restricted mean time lost under competing risks data

Background

Under competing risks, the commonly used sub-distribution hazard ratio (SHR) is not easy to interpret clinically and is valid only under the proportional sub-distribution hazard (SDH) assumption. This paper introduces an alternative statistical measure: the restricted mean time lost (RMTL).

Methods

First, the definition and estimation methods of the measures are introduced. Second, based on the differences in RMTLs, a basic difference test (Diff) and a supremum difference test (sDiff) are constructed. Then, the corresponding sample size estimation method is proposed. The statistical properties of the methods and the estimated sample size are evaluated using Monte Carlo simulations, and these methods are also applied to two real examples.

Results

The simulation results show that sDiff performs well and has relatively high test efficiency in most situations. Regarding sample size calculation, sDiff exhibits good performance in various situations. The methods are illustrated using two examples.

Conclusions

RMTL can meaningfully summarize treatment effects for clinical decision making, which can then be reported with the SDH ratio for competing risks data. The proposed sDiff test and the two calculated sample size formulas have wide applicability and can be considered in real data analysis and trial design.

A pseudo-likelihood method for estimating misclassification probabilities in competing-risks settings when true-event data are partially observed

Outcome misclassification occurs frequently in binary-outcome studies and can result in biased estimation of quantities such as the incidence, prevalence, cause-specific hazards, cumulative incidence functions, and so forth. A number of remedies have been proposed to address the potential misclassification of the outcomes in such data. The majority of these remedies lie in the estimation of misclassification probabilities, which are in turn used to adjust analyses for outcome misclassification. A number of authors advocate using a gold-standard procedure on a sample internal to the study to learn about the extent of the misclassification. With this type of internal validation, the problem of quantifying the misclassification also becomes a missing data problem as, by design, the true outcomes are only ascertained on a subset of the entire study sample. Although, the process of estimating misclassification probabilities appears simple conceptually, the estimation methods proposed so far have several methodological and practical shortcomings. Most methods rely on missing outcome data to be missing completely at random (MCAR), a rather stringent assumption which is unlikely to hold in practice. Some of the existing methods also tend to be computationally-intensive. To address these issues, we propose a computationally-efficient, easy-to-implement, pseudo-likelihood estimator of the misclassification probabilities under a missing at random (MAR) assumption, in studies with an available internal-validation sample. We present the estimator through the lens of studies with competing-risks outcomes, though the estimator extends beyond this setting. We describe the consistency and asymptotic distributional properties of the resulting estimator, and derive a closed-form estimator of its variance. The finite-sample performance of this estimator is evaluated via simulations. Using data from a real-world study with competing-risks outcomes, we illustrate how the proposed method can be used to estimate misclassification probabilities. We also show how the estimated misclassification probabilities can be used in an external study to adjust for possible misclassification bias when modeling cumulative incidence functions.

Assessment of ST2 for risk of death following graft-versus-host disease in pediatric and adult age groups

Assessment of prognostic biomarkers of nonrelapse mortality (NRM) after allogeneic hematopoietic cell transplantation (HCT) in the pediatric age group is lacking. To address this need, we conducted a prospective cohort study with 415 patients at 6 centers: 170 were children age 10 years or younger and 245 were patients older than age 10 years (both children and adults were accrued from 2013 to 2018). The following 4 plasma biomarkers were assessed pre-HCT and at days +7, +14, and +21 post-HCT: stimulation-2 (ST2), tumor necrosis factor receptor 1 (TNFR1), regenerating islet-derived protein 3α (REG3α), and interleukin-6 (IL-6). We performed landmark analyses for NRM, dichotomizing the cohort at age 10 years or younger and using each biomarker median as a cutoff for high- and low-risk groups. Post-HCT biomarker analysis showed that ST2 (>26 ng/mL), TNFR1 (>3441 pg/mL), and REG3α (>25 ng/mL) are associated with NRM in children age 10 years or younger (ST2: hazard ratio [HR], 9.13; 95% confidence interval [CI], 2.74-30.38; P = .0003; TNFR1: HR, 4.29; 95% CI, 1.48-12.48; P = .0073; REG3α: HR, 7.28; 95% CI, 2.05-25.93; P = .0022); and in children and adults older than age 10 years (ST2: HR, 2.60; 95% CI, 1.15-5.86; P = .021; TNFR1: HR, 2.09; 95% CI, 0.96-4.58; P = .06; and REG3α: HR, 2.57; 95% CI, 1.19-5.55; P = .016). When pre-HCT biomarkers were included, only ST2 remained significant in both cohorts. After adjustment for significant covariates (race/ethnicity, malignant disease, graft, and graft-versus-host-disease prophylaxis), ST2 remained associated with NRM only in recipients age 10 years or younger (HR, 4.82; 95% CI, 1.89-14.66; P = .0056). Assays of ST2, TNFR1, and REG3α in the first 3 weeks after HCT have prognostic value for NRM in both children and adults. The presence of ST2 before HCT is a prognostic biomarker for NRM in children age 10 years or younger allowing for additional stratification. This trial was registered at www.clinicaltrials.gov as #NCT02194439.

Comparison of the marginal hazard model and the sub-distribution hazard model for competing risks under an assumed copula

For the analysis of competing risks data, three different types of hazard functions have been considered in the literature, namely the cause-specific hazard, the sub-distribution hazard, and the marginal hazard function. Accordingly, medical researchers can fit three different types of the Cox model to estimate the effect of covariates on each of the hazard function. While the relationship between the cause-specific hazard and the sub-distribution hazard has been extensively studied, the relationship to the marginal hazard function has not yet been analyzed due to the difficulties related to non-identifiability. In this paper, we adopt an assumed copula model to deal with the model identifiability issue, making it possible to establish a relationship between the sub-distribution hazard and the marginal hazard function. We then compare the two methods of fitting the Cox model to competing risks data. We also extend our comparative analysis to clustered competing risks data that are frequently used in medical studies. To facilitate the numerical comparison, we implement the computing algorithm for marginal Cox regression with clustered competing risks data in the R joint.Cox package and check its performance via simulations. For illustration, we analyze two survival datasets from lung cancer and bladder cancer patients.

Prior stroke and transient ischemic attack as risk factors for subsequent stroke in atrial fibrillation patients: A report from the GARFIELD-AF registry

BACKGROUND

It is not always possible to verify whether a patient complaining of symptoms consistent with transient ischemic attack has had an actual cerebrovascular event.

RESEARCH QUESTION

To characterize the risk of cardiovascular events associated with a history of stroke/transient ischemic attack in patients with atrial fibrillation.

STUDY DESIGN AND METHODS

This study investigated the clinical characteristics and outcomes of patients with a history of stroke/transient ischemic attack among 52,014 patients enrolled prospectively in GARFIELD-AF registry. The diagnosis of stroke or transient ischemic attack was not protocol defined but based on physicians' assessment. Patients' one-year risk of death, stroke/systemic embolism, and major bleeding was assessed by multivariable Cox regression.

RESULTS

At enrollment, 5617 (10.9%) patients were reported to have a history of stroke or transient ischemic attack. Patients with stroke or transient ischemic attack were older and had a greater burden of diabetes, moderate-to-severe kidney disease, and atherothrombosis and higher median CHA₂DS₂-VASc and HAS-BLED scores than those without history of stroke or transient ischemic attack. After adjustment, prior stroke/transient ischemic attack was associated with significantly higher risk for all-cause mortality (hazard ratio (HR), 1.26; 95% confidence interval (CI), 1.12-1.42), cardiovascular death (HR, 1.22; 95% CI, 1.01-1.48), non-cardiovascular death (HR, 1.39; 95% CI, 1.15-1.68), and stroke/systemic embolism (HR, 2.17; 95% CI, 1.80-2.63) than patients without history of stroke/transient ischemic attack. In patients with a prior stroke alone higher risk was observed for all-cause mortality (HR, 1.29; 95% CI, 1.11-1.50), non-cardiovascular death (HR, 1.39; 95% CI, 1.10-1.77), and stroke/systemic embolism (HR, 2.29; 95% CI, 1.83-2.86). No significantly elevated risk of adverse events was seen for patients with history of transient ischemic attack alone.

INTERPRETATION

A history of prior stroke or transient ischemic attack is a strong independent risk factor for mortality and stroke/systemic embolism. This excess risk is mainly attributed to a history of stroke (with or without transient ischemic attack), whereas history of transient ischemic attack is a weaker predictor. Clinical trial registration: NCT01090362.

Nonparametric tests for transition probabilities in nonhomogeneous Markov processes

This paper proposes nonparametric two-sample tests for the direct comparison of the probabilities of a particular transition between states of a continuous time non-homogeneous Markov process with a finite state space. The proposed tests are a linear nonparametric test, an L ²-norm-based test and a Kolmogorov-Smirnov-type test. Significance level assessment is based on rigorous procedures, which are justified through the use of modern empirical process theory. Moreover, the L ²-norm and the Kolmogorov-Smirnov-type tests are shown to be consistent for every fixed alternative hypothesis. The proposed tests are also extended to more complex situations such as cases with incompletely observed absorbing states and non-Markov processes. Simulation studies show that the test statistics perform well even with small sample sizes. Finally, the proposed tests are applied to data on the treatment of early breast cancer from the European Organization for Research and Treatment of Cancer (EORTC) trial 10854, under an illness-death model.

Nonparametric estimation of the cumulative incidence function under outcome misclassification using external validation data

Misclassification of outcomes or event types is common in health sciences research and can lead to serious bias when estimating the cumulative incidence functions in settings with competing risks. Recent work has shown how to estimate nonparametric cumulative incidence functions in the presence of nondifferential outcome misclassification when the misclassification probabilities are known. Here, we extend this approach to account for misclassification that is differential with respect to important predictors of the outcome using misclassification probabilities estimated from external validation data. Moreover, we propose a bootstrap approach in which the observations from both the main study data and the external validation study are resampled to allow the uncertainty in the misclassification probabilities to propagate through the analysis into the final confidence intervals, ensuring appropriate confidence interval coverage probabilities. The proposed estimator is shown to be uniformly consistent and simulation studies indicate that both the estimator and the standard error estimation approach perform well in finite samples. The methodology is applied to estimate the cumulative incidence of death and disengagement from HIV care in a large cohort of HIV infected individuals in sub-Saharan Africa, where a significant death underreporting issue leads to outcome misclassification. This analysis uses external validation data from a separate study conducted in the same country.

Parameter Estimation of Lindley Distribution Based on Progressive Type-II Censored Competing Risks Data with Binomial Removals

The competing risk model based on Lindley distribution is discussed under the progressive type-II censored sample data with binomial removals. The maximum likelihood estimation of the unknown parameters of the distribution is established. Using the Lindley approximation method, we also obtain the Bayesian estimation of the unknown parameters of the distribution under different loss functions. The performance of different estimates is studied in this article. A real practical dataset is analyzed for illustration.

Evaluating probiotics for the prevention of ventilator-associated pneumonia: a randomised placebo-controlled multicentre trial protocol and statistical analysis plan for PROSPECT

INTRODUCTION

Ventilator-associated pneumonia (VAP) is the most common healthcare-associated infection in critically ill patients. Prior studies suggest that probiotics may reduce VAP and other infections in critically ill patients; however, most previous randomised trials were small, single centre studies. The Probiotics: Prevention of Severe Pneumonia and Endotracheal Colonization Trial (PROSPECT) aims to determine the impact of the probiotic Lactobacillus rhamnosus GG on VAP and other clinically important outcomes in critically ill adults.

METHODS

PROSPECT is a multicentre, concealed, randomised, stratified, blinded, controlled trial in patients ≥18 years old, anticipated to be mechanically ventilated ≥72 hours, in intensive care units (ICUs) in Canada, the USA and Saudi Arabia. Patients receive either 1×10¹⁰ colony forming units of L. rhamnosus GG twice daily or an identical appearing placebo. Those at increased risk of probiotic infection are excluded. The primary outcome is VAP. Secondary outcomes are other ICU-acquired infections including Clostridioides difficile infection, diarrhoea (including antibiotic-associated diarrhoea), antimicrobial use, ICU and hospital length of stay and mortality. The planned sample size of 2650 patients is based on an estimated 15% VAP rate and will provide 80% power to detect a 25% relative risk reduction.

ETHICS AND DISSEMINATION

This protocol and statistical analysis plan outlines the methodology, primary and secondary analyses, sensitivity analyses and subgroup analyses. PROSPECT is approved by Health Canada (#9427-M1133-45C), the research ethics boards of all participating hospitals and Public Health Ontario. Results will be disseminated via academic channels (peer reviewed journal publications, professional healthcare fora including international conferences) and conventional and social media. The results of PROSPECT will inform practice guidelines worldwide.

TRIALREGISTRATION NUMBER

NCT02462590; Pre-results.

Semiparametric competing risks regression under interval censoring using the R package intccr

BACKGROUND AND OBJECTIVE

Competing risk data are frequently interval-censored in real-world applications, that is, the exact event time is not precisely observed but is only known to lie between two time points such as clinic visits. This type of data requires special handling because the actual event times are unknown. To deal with this problem we have developed an easy-to-use open-source statistical software.

METHODS

An approach to perform semiparametric regression analysis of the cumulative incidence function with interval-censored competing risks data is the sieve maximum likelihood method based on B-splines. An important feature of this approach is that it does not impose restrictive parametric assumptions. Also, this methodology provides semiparametrically efficient estimates. Implementation of this methodology can be easily performed using our new R package intccr.

RESULTS

The R package intccr performs semiparametric regression analysis of the cumulative incidence function based on interval-censored competing risks data. It supports a large class of models including the proportional odds and the Fine-Gray proportional subdistribution hazards model as special cases. It also provides the estimated cumulative incidence functions for a particular combination of covariate values. The package also provides some data management functionality to handle data sets which are in a long format involving multiple lines of data per subject.

CONCLUSIONS

The R package intccr provides a convenient and flexible software for the analysis of the cumulative incidence function based on interval-censored competing risks data.

Focus on an infrequently used quantity in the context of competing risks: The conditional probability function

In clinical studies of hematologic and oncologic diseases, the outcomes of interest are generally composite time to event endpoints which are usually defined by occurrence of different event types. Nonetheless, clinicians are interested in studying only one event type, which leads to a competing risks situation. In this context, Pepe and Mori presented a quantity directly derived from the cumulative incidence: the conditional probability. This function defines the probability that a given event occurs, conditionally on not having had a competing event by that time. The objective of this paper is to present this conditional cumulative incidence function and to compare its use to the cumulative incidence in different data sets. Different scenarios highlight the importance of the competing event on the interpretation of the conditional probability. Conditional probability needs to be interpreted jointly with the cumulative incidence. This quantity can be of interest especially when the risk of the competing event is large, strongly precludes the risk of the event of interest and provides useful additional information.

Incidence and predictors of loss to follow up among HIV-infected adults at Pawi General Hospital, northwest Ethiopia: competing risk regression model

Objective

This study was aimed at assessing the incidence of lost-to-follow-up and its predictors among HIV-positive adults after initiation into antiretroviral therapy at Pawi General Hospital, northwest Ethiopia.

Results

The overall cumulative incidence of lost-to-follow-up after ART initiation was high, 11.6 (95% CI 9.8–13.7) per 100 adult-years follow-up time. Independent significant predictors of lost to follow up were being aged 15–28 years (aSHR = 0.44; 95% CI 0.24–0.83), being on WHO clinical stage IV (aSHR = 2.09; 95% CI 1.02–3.13); and receiving isoniazid preventive therapy (aSHR = 0.11; 95% CI 0.06–0.18).

Electronic supplementary material

The online version of this article (10.1186/s13104-018-3407-5) contains supplementary material, which is available to authorized users.

Importance of the lipid-related pathways in the association between statins, mortality, and cardiovascular disease risk: The Multi-Ethnic Study of Atherosclerosis

PURPOSE

Estimating how much of the impact of statins on coronary heart diseases (CHD), cardiovascular disease (CVD), and mortality risk is attributable to their effect on low-density lipoprotein cholesterol (LDL), high-density lipoprotein cholesterol (HDL), and triglycerides.

METHODS

A semi-parametric g-formula estimator together with data from the Multi-Ethnic Study of Atherosclerosis (a prospective multi-center cohort study) was utilized to perform a mediation analysis. A total of 5280 participants, men and women of various race/ethnicities from multiple sites across the United States, were considered in the current study.

RESULTS

The adherence adjusted total relative risk reduction (RRR) estimate (95% confidence interval) of statins on CHD was 14% (-16%, 37%), and the indirect component through LDL was 23% (-4%, 58%). For CVD, the total RRR was 23% (2%, 40%), and the indirect component through LDL was 5% (-13%, 25%). The total RRR of mortality was 18% (-1%, 35%), and the indirect component through LDL was -4% (-17%, 12%). The estimated indirect components through HDL and triglycerides were close to zero with narrow confidence intervals for all 3 outcomes.

CONCLUSIONS

The estimated effect of statins on mortality, CVD, and CHD appeared to be independent of their estimated effect on HDL and triglycerides. Our study provides evidence that the preventive effect of statins on CHD could be attributed in large part to their effect on LDL. Our g-formula estimator is a promising approach to elucidate pathways, even if it is hard to make firm conclusions for the LDL pathway on mortality and CVD.