Relaxing the independent censoring assumption in the Cox proportional hazards model using multiple imputation

Application of the estimand framework for an emulated trial using reference based multiple imputation to investigate informative censoring

BACKGROUND

The ICH E9 (R1) addendum on Estimands and Sensitivity analysis in Clinical trials proposes a framework for the design and analysis of clinical trials aimed at improving clarity around the definition of the targeted treatment effect (the estimand) of a study.

METHODS

We adopt the estimand framework in the context of a study using "trial emulation" to estimate the risk of pneumocystis pneumonia, an opportunistic disease contracted by people living with HIV and AIDS having a weakened immune system, when considering two antibiotic treatment regimes for stopping antibiotic prophylaxis treatment against this disease. A "while on treatment" strategy has been implemented for post-randomisation (intercurrent) events. We then perform a sensitivity analysis using reference based multiple imputation to model a scenario in which patients lost to follow-up stop taking prophylaxis.

RESULTS

The primary analysis indicated a protective effect for the new regime which used viral suppression as prophylaxis stopping criteria (hazard ratio (HR) 0.78, 95% confidence interval [0.69, 0.89], p < 0.001). For the sensitivity analysis, when we apply the "jump to off prophylaxis" approach, the hazard ratio is almost the same compared to that from the primary analysis (HR 0.80 [0.69, 0.95], p = 0.009). The sensitivity analysis confirmed that the new regime exhibits a clear improvement over the existing guidelines for PcP prophylaxis when those lost to follow-up "jump to off prophylaxis".

CONCLUSIONS

Our application using reference based multiple imputation demonstrates the method's flexibility and simplicity for sensitivity analyses in the context of the estimand framework for (emulated) trials.

Competing Risks in Clinical Trials: Do They Matter and How Should We Account for Them?

During patient follow-up in a randomized trial, some deaths may occur. Where death (or noncardiovascular death) is not part of an outcome of interest it is termed a competing risk. Conventional analyses (eg, Cox proportional hazards model) handle death similarly to other censored follow-up. Patients still alive are unrealistically assumed to be representative of those who died. The Fine and Gray model has been used to handle competing risks, but is often used inappropriately and can be misleading. We propose an alternative multiple imputation approach that plausibly accounts for the fact that patients who die tend also to be at high risk for the (unobserved) outcome of interest. This provides a logical framework for exploring the impact of a competing risk, recognizing that there is no unique solution. We illustrate these issues in 3 cardiovascular trials and in simulation studies. We conclude with practical recommendations for handling competing risks in future trials.

Effectiveness and safety of continuous low-molecular-weight heparin versus switching to direct oral anticoagulants in cancer-associated venous thrombosis

Given the existing uncertainty regarding the effectiveness and safety of switching from low-molecular-weight heparin (LMWH) to direct oral anticoagulants (DOACs) in patients with cancer-associated venous thrombosis (CAT), we conducted a comprehensive population-based cohort study utilizing electronic health database in Hong Kong. A total of 4356 patients with CAT between 2010 and 2022 were included, with 1700 (39.0%) patients switching to DOAC treatment. Compared to continuous LMWH treatment, switching to DOACs was associated with a significantly lower risk of hospitalization due to venous thromboembolism (HR: 0.49 [95% CI = 0.35-0.68]) and all-cause mortality (HR: 0.67 [95% CI = 0.61-0.74]), with no significant difference in major bleeding (HR: 1.04 [95% CI = 0.83-1.31]) within six months. These findings provide reassurance regarding the effectiveness and safety of switching from LMWH to DOACs among patients with CAT, including vulnerable patient groups.

An illness-death multistate model to implement delta adjustment and reference-based imputation with time-to-event endpoints

With a treatment policy strategy, therapies are evaluated regardless of the disturbance caused by intercurrent events (ICEs). Implementing this estimand is challenging if subjects are not followed up after the ICE. This circumstance can be dealt with using delta adjustment (DA) or reference-based (RB) imputation. In the survival field, DA and RB imputation have been researched so far using multiple imputation (MI). Here, we present a fully analytical solution. We use the illness-death multistate model with the following transitions: (a) from the initial state to the event of interest, (b) from the initial state to the ICE, and (c) from the ICE to the event. We estimate the intensity function of transitions (a) and (b) using flexible parametric survival models. Transition (c) is assumed unobserved but identifiable using DA or RB imputation assumptions. Various rules have been considered: no ICE effect, DA under proportional hazards (PH) or additive hazards (AH), jump to reference (J2R), and (either PH or AH) copy increment from reference. We obtain the marginal survival curve of interest by calculating, via numerical integration, the probability of transitioning from the initial state to the event of interest regardless of having passed or not by the ICE state. We use the delta method to obtain standard errors (SEs). Finally, we quantify the performance of the proposed estimator through simulations and compare it against MI. Our analytical solution is more efficient than MI and avoids SE misestimation-a known phenomenon associated with Rubin's variance equation.

Long-term Outcomes with Spinal versus General Anesthesia for Hip Fracture Surgery: A Randomized Trial

BACKGROUND

The effects of spinal versus general anesthesia on long-term outcomes have not been well studied. This study tested the hypothesis that spinal anesthesia is associated with better long-term survival and functional recovery than general anesthesia.

METHODS

A prespecified analysis was conducted of long-term outcomes of a completed randomized superiority trial that compared spinal anesthesia versus general anesthesia for hip fracture repair. Participants included previously ambulatory patients 50 yr of age or older at 46 U.S. and Canadian hospitals. Patients were randomized 1:1 to spinal or general anesthesia, stratified by sex, fracture type, and study site. Outcome assessors and investigators involved in the data analysis were masked to the treatment arm. Outcomes included survival at up to 365 days after randomization (primary); recovery of ambulation among 365-day survivors; and composite endpoints for death or new inability to ambulate and death or new nursing home residence at 365 days. Patients were included in the analysis as randomized.

RESULTS

A total of 1,600 patients were enrolled between February 12, 2016, and February 18, 2021; 795 were assigned to spinal anesthesia, and 805 were assigned to general anesthesia. Among 1,599 patients who underwent surgery, vital status information at or beyond the final study interview (conducted at approximately 365 days after randomization) was available for 1,427 (89.2%). Survival did not differ by treatment arm; at 365 days after randomization, there were 98 deaths in patients assigned to spinal anesthesia versus 92 deaths in patients assigned to general anesthesia (hazard ratio, 1.08; 95% CI, 0.81 to 1.44, P = 0.59). Recovery of ambulation among patients who survived a year did not differ by type of anesthesia (adjusted odds ratio for spinal vs. general, 0.87; 95% CI, 0.67 to 1.14; P = 0.31). Other outcomes did not differ by treatment arm.

CONCLUSIONS

Long-term outcomes were similar with spinal versus general anesthesia.

EDITOR’S PERSPECTIVE

Non-inferiority trials with time-to-event data: clarifying the impact of censoring

In non-inferiority (NI) trials with time-to-event data, different types and patterns of censoring may occur, but their impact on trial results is not entirely clear. We investigated the influence of informative and non-informative censoring by conducting extensive simulation studies under the assumption that the NI margin is defined as a maximum acceptable hazard ratio and scenarios typically observed in recent NI trials. We found that while non-informative censoring tends to only affect the power, informative censoring can impact the treatment effect estimates, type I error rate, and power. The magnitude of these effects depends on the between-group differences in the failure and informative censoring risks, as well as the correlation between censoring and failure times, among other factors. The adverse impact of informative censoring was generally decreased with larger NI margins.

Imputation of Missing Data for Time-to-Event Endpoints Using Retrieved Dropouts

We have explored several statistical approaches to impute missing time-to-event data that arise from outcome trials with relatively long follow-up periods. Aligning with the primary estimand, such analyses evaluate the robustness of results by imposing an assumption different from censoring at random (CAR). Although there have been debates over which assumption and which method is more appropriate to be applied to the imputation, we propose to use the collection of retrieved dropouts as the basis of missing data imputation. As retrieved dropouts share a similar disposition, such as treatment discontinuation, with subjects who have missing data, they can reasonably be assumed to characterize the distribution of time-to-event among subjects with missing data. In terms of computational intensity and robustness to violation of underlying distributional assumption, we have compared parametric approaches via MCMC or MLE multivariate sampling procedures to a non-parametric bootstrap approach with respect to baseline hazard function. Each of these approaches follows a process of multiple imputation ("proper imputations"), analysis of complete datasets, and final combination. The type-I error, and power rates are examined under a wide range of scenarios to inform the performance characteristics. A subset of a real unblinded phase III CVOT is used to demonstrate the application of the proposed approaches, compared to the Cox proportional hazards model and jump-to-reference multiple imputation.

Effect of Lipid-Lowering Drugs on Renal and Cardiovascular Outcomes in Patients with Chronic Kidney Disease and Dyslipidemia: A Retrospective Cohort Study

The effects of lipid-lowering drugs (LLDs) on cardiovascular and renal outcomes in patients with advanced chronic kidney disease (CKD) and dyslipidemia are not completely understood. We conducted a retrospective cohort study to evaluate the effect of LLDs on end-stage kidney disease (ESKD), major adverse cardiovascular events (MACEs), and mortality in adult patients with CKD stage 3b, 4, or 5, and dyslipidemia. Participants were recruited between January 1, 2008, and December 31, 2018, and classified as LLD or non-LLD users; the final follow-up date was December 31, 2020. The primary outcome was time to ESKD or death due to renal failure. Sub-distribution hazard regression models adjusted for multivariables, including time-varying lipid profile covariates, were used for the analysis. Among the 6,740 participants, 4,280 patients with CKD and dyslipidemia, including 872 using LLDs and 3,408 not using LLDs, completed the primary analysis. The multivariable analyses showed that LLD users had a significantly lower risk of time to the composite renal outcome (adjusted hazard ratio [aHR], 0.76, 95% confidence interval [CI], 0.65-0.89), and MACE incidence (aHR, 0.75, 95% CI, 0.62-0.93) than did non-LLD users. After adjusting for time-varying covariates of the lipid profile, there was a significant difference in the composite renal outcome (aHR, 0.78, 95% CI, 0.65-0.93) and MACEs (aHR, 0.77, 95% CI, 0.60-0.98). Among adult patients with advanced CKD and dyslipidemia, LLD users had a significantly lower risk of composite renal outcomes and MACEs than non-LLD users. In addition to reducing lipid profile, the use of LLD is associated with renal and cardiovascular protective effects.

SMIM: A unified framework of survival sensitivity analysis using multiple imputation and martingale

Censored survival data are common in clinical trial studies. We propose a unified framework for sensitivity analysis to censoring at random in survival data using multiple imputation and martingale, called SMIM. The proposed framework adopts the δ-adjusted and control-based models, indexed by the sensitivity parameter, entailing censoring at random and a wide collection of censoring not at random assumptions. Also, it targets a broad class of treatment effect estimands defined as functionals of treatment-specific survival functions, taking into account missing data due to censoring. Multiple imputation facilitates the use of simple full-sample estimation; however, the standard Rubin's combining rule may overestimate the variance for inference in the sensitivity analysis framework. We decompose the multiple imputation estimator into a martingale series based on the sequential construction of the estimator and propose the wild bootstrap inference by resampling the martingale series. The new bootstrap inference has a theoretical guarantee for consistency and is computationally efficient compared to the nonparametric bootstrap counterpart. We evaluate the finite-sample performance of the proposed SMIM through simulation and an application on an HIV clinical trial.

Post-covid medical complaints following infection with SARS-CoV-2 Omicron vs Delta variants

The SARS-CoV-2 Omicron (B.1.1.529) variant has been associated with less severe acute disease, however, concerns remain as to whether long-term complaints persist to a similar extent as for earlier variants. Studying 1 323 145 persons aged 18-70 years living in Norway with and without SARS-CoV-2 infection in a prospective cohort study, we found that individuals infected with Omicron had a similar risk of post-covid complaints (fatigue, cough, heart palpitations, shortness of breath and anxiety/depression) as individuals infected with Delta (B.1.617.2), from 14 to up to 126 days after testing positive, both in the acute (14 to 29 days), sub-acute (30 to 89 days) and chronic post-covid (≥90 days) phases. However, at ≥90 days after testing positive, individuals infected with Omicron had a lower risk of having any complaint (43 (95%CI = 14 to 72) fewer per 10,000), as well as a lower risk of musculoskeletal pain (23 (95%CI = 2-43) fewer per 10,000) than individuals infected with Delta. Our findings suggest that the acute and sub-acute burden of post-covid complaints on health services is similar for Omicron and Delta. The chronic burden may be lower for Omicron vs Delta when considering musculoskeletal pain, but not when considering other typical post-covid complaints.

Intravenous edaravone treatment in ALS and survival: An exploratory, retrospective, administrative claims analysis

BACKGROUND

We aimed to evaluate overall survival in US patients with amyotrophic lateral sclerosis (ALS) treated with intravenous (IV) edaravone compared with those not treated with IV edaravone in a real-world setting.

METHODS

This exploratory retrospective comparative effectiveness observational analysis included patients with ALS who were enrolled in an administrative claims database from 8 August 2017 to 31 March 2020. Propensity score matching identified IV edaravone-treated patients (cases) and non-edaravone-treated patients (controls) matched for covariates: age, race, geographic region, sex, pre-index disease duration, insurance, history of cardiovascular disease, riluzole prescription, gastrostomy tube placement, artificial nutrition, noninvasive ventilation, and all-cause hospitalisation. For cases, the index date was the date of the first claim for IV edaravone. For controls, it was the date IV edaravone was available (8 August 2017). The effect of IV edaravone on all-cause mortality was estimated with shared frailty Cox regression analysis.

FINDINGS

318 cases were matched to 318 controls. In both groups, 208 patients (65.4%) had a history of riluzole prescription. As of 31 March 2021, there were 155 deaths (48.7%) among the cases and 196 among the controls (61.6%). Median overall survival time was 29.5 months with edaravone and 23.5 months without, respectively, and the risk of death was 27% lower in cases than in controls (HR, 0.73; 95% CI, 0.59-0.91; p=0.005).

INTERPRETATION

In this real-world analysis, IV edaravone treatment in a large predominantly riluzole-treated US cohort was associated with prolonged overall survival compared with not using IV edaravone. Data from adequately powered RCTs are needed to support this finding.

FUNDING

Funded by Mitsubishi Tanabe Pharma America.

Considerations of Competing Risks Analysis in Cardio-Oncology Studies: JACC: CardioOncology State-of-the-Art Review

Cardio-oncology research studies often require consideration of potential competing risks, as the occurrence of other events (eg, cancer-related death) may preclude the primary event of interest (eg, cardiovascular outcome). However, the decision to conduct competing risks analysis is not always straightforward, and even when deemed necessary, misconceptions exist about the appropriate choice of analytical methods to address the competing risks. R researchers are encouraged to consider competing risks at the study design stage and are provided provide an assessment tool to guide decisions on analytical approach on the basis of study objectives. The existing statistical methods for competing risks analysis, including cumulative incidence estimations and regression modeling are also reviewed. Cardio-oncology-specific examples are used to illustrate these concepts and highlight potential pitfalls and misinterpretations. R code is also provided for these analyses.

Personalizing treatment in end-stage kidney disease: deciding between hemodiafiltration and hemodialysis based on individualized treatment effect prediction

Background

Previous studies suggest that hemodiafiltration reduces mortality compared to hemodialysis in patients with end-stage kidney disease (ESKD), but controversy surrounding its benefits remain and it is unclear to what extent individual patients benefit from hemodiafiltration. This study aimed to develop and validate a treatment effect prediction model to determine which patients would benefit most from hemodiafiltration compared to hemodialysis in terms of all-cause mortality.

Methods

Individual participant data from four randomized controlled trials comparing hemodiafiltration with hemodialysis on mortality were used to derive a Royston-Parmar model for prediction of absolute treatment effect of hemodiafiltration based on pre-specified patient and disease characteristics. Validation of the model was performed using internal-external cross validation.

Results

The median predicted survival benefit was 44 (Q1-Q3: 44–46) days for every year of treatment with hemodiafiltration compared to hemodialysis. The median survival benefit with hemodiafiltration ranged from 2 to 48 months. Patients who benefited most from hemodiafiltration were younger, less likely to have diabetes or a cardiovascular history and had higher serum creatinine and albumin levels. Internal-external cross validation showed adequate discrimination and calibration.

Conclusion

Although overall mortality is reduced by hemodiafiltration compared to hemodialysis in ESKD patients, the absolute survival benefit can vary greatly between individuals. Our results indicate that the effects of hemodiafiltration on survival can be predicted using a combination of readily available patient and disease characteristics, which could guide shared decision-making.

Comment on 'statistical consideration and challenges in bridging study of personalized medicine': a modified variance for sensitivity analysis

A pivotal clinical trial is often necessary to assess drug efficacy in the intended to use (IU) population. Ideally, patients should be enrolled based on a positive test result from a well-characterized companion diagnostic (CDx). However, the central challenge is that patients are instead recruited on the basis of a clinical trial assay (CTA) result. This challenge arises because, CTA is available at all local labs; the time delay to enable enrollment based on CDx could result in a significant proportion of patients being unable to participate, adversely affecting precision and/or bias. The difficulty is therefore that patients are recruited on the basis that their CTA result is positive (CTA+) but the goal is to assess the drug efficacy in patients positive by the companion diagnostic (CDx+). In this commentary, we will examine an apparent weakness of a variance formula that is proposed in the context of a sensitivity analysis. We will develop an alternative formula, and argue that this should be used instead.

Estimation and inference of predictive discrimination for survival outcome risk prediction models

Accurate risk prediction has been the central goal in many studies of survival outcomes. In the presence of multiple risk factors, a censored regression model can be employed to estimate a risk prediction rule. Before the prediction tool can be popularized for practical use, it is crucial to rigorously assess its prediction performance. In our motivating example, researchers are interested in developing and validating a risk prediction tool to identify future lung cancer cases by integrating demographic information, disease characteristics and smoking-related data. Considering the long latency period of cancer, it is desirable for a prediction tool to achieve discriminative performance that does not weaken over time. We propose estimation and inferential procedures to comprehensively assess both the overall predictive discrimination and the temporal pattern of an estimated prediction rule. The proposed methods readily accommodate commonly used censored regression models, including the Cox proportional hazards model and the accelerated failure time model. The estimators are consistent and asymptotically normal, and reliable variance estimators are also developed. The proposed methods offer an informative tool for inferring time-dependent predictive discrimination, as well as for comparing the discrimination performance between candidate models. Applications of the proposed methods demonstrate enduring performance of the risk prediction tool in the PLCO study and detected decaying performance in a study of liver disease.

A Bayesian Approach for Imputation of Censored Survival Data

A common feature of much survival data is censoring due to incompletely observed lifetimes. Survival analysis methods and models have been designed to take account of this and provide appropriate relevant summaries, such as the Kaplan–Meier plot and the commonly quoted median survival time of the group under consideration. However, a single summary is not really a relevant quantity for communication to an individual patient, as it conveys no notion of variability and uncertainty, and the Kaplan–Meier plot can be difficult for the patient to understand and also is often mis-interpreted, even by some physicians. This paper considers an alternative approach of treating the censored data as a form of missing, incomplete data and proposes an imputation scheme to construct a completed dataset. This allows the use of standard descriptive statistics and graphical displays to convey both typical outcomes and the associated variability. We propose a Bayesian approach to impute any censored observations, making use of other information in the dataset, and provide a completed dataset. This can then be used for standard displays, summaries, and even, in theory, analysis and model fitting. We particularly focus on the data visualisation advantages of the completed data, allowing displays such as density plots, boxplots, etc, to complement the usual Kaplan–Meier display of the original dataset. We study the performance of this approach through a simulation study and consider its application to two clinical examples.

Inference on latent factor models for informative censoring

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

Estimation of a treatment policy estimand for time to event data using data collected post discontinuation of randomised treatment

Discontinuation from randomised treatment is a common intercurrent event in clinical trials. When the target estimand uses a treatment policy strategy to deal with this intercurrent event, data after cessation of treatment is relevant to estimate the estimand and all efforts should be made to collect such data. Missing data may nevertheless occur due to participants withdrawing from the study and assumptions regarding the values for data that are missing are required for estimation. A missing-at-random assumption is commonly made in this setting, but it may not always be viewed as appropriate. Another potential approach is to assume missing values are similar to data collected after treatment discontinuation. This idea has been previously proposed in the context of recurrent event data. Here we extend this approach to time-to-event outcomes using the hazard function. We propose imputation models that allow for different hazard rates before and after treatment discontinuation and use the posttreatment discontinuation hazard to impute events for participants with missing follow-up periods due to study withdrawal. The imputation models are fitted as Andersen-Gill models. We illustrate the proposed methods with an example of a clinical trial in patients with chronic obstructive pulmonary disease.

Editor's Choice - Risk of Major Amputation Following Application of Paclitaxel Coated Balloons in the Lower Limb Arteries: A Systematic Review and Meta-Analysis of Randomised Controlled Trials

OBJECTIVE

There have been concerns about the long term safety of paclitaxel coated devices in the lower limbs. A formal systematic review and meta-analysis of randomised controlled trials (RCTs) was performed to examine the long term risk of major amputation using paclitaxel coated balloons in peripheral arterial disease (PAD).

METHOD

This systematic review was registered with PROSPERO (ID 227761). A broad bibliographic search was performed for RCTs investigating paclitaxel coated balloons in the peripheral arteries (femoropopliteal and infrapopliteal) for treatment of intermittent claudication or critical limb ischaemia (CLI). The literature search was last updated on 20 February 2021 without any restrictions on publication language, date, or status. Major amputations were analysed with time to event methods employing one and two stage models. Sensitivity and subgroup analyses, combinatorial meta-analysis, and a multivariable dose response meta-analysis to examine presence of a biological gradient were also performed.

RESULTS

In all, 21 RCTs with 3 760 lower limbs were analysed (52% intermittent claudication and 48% CLI; median follow up two years). There were 87 major amputations of 2 216 limbs in the paclitaxel arms (4.0% crude risk) compared with 41 major amputations in 1 544 limbs in the control arms (2.7% crude risk). The risk of major amputation was significantly higher for paclitaxel coated balloons with a hazard ratio (HR) of 1.66 (95% CI 1.14 - 2.42; p = .008, one stage stratified Cox model). The prediction interval was 95% CI 1.10 - 2.46 (two stage model). The observed amputation risk was consistent for both femoropopliteal (p = .055) and infrapopliteal (p = .055) vessels. Number needed to harm was 35 for CLI. There was good evidence of a significant non-linear dose response relationship with accelerated risk per cumulative paclitaxel dose (chi square model p = .007). There was no evidence of publication bias (p = .80) and no significant statistical heterogeneity between studies (I2 = 0%, p = .77). Results were stable across sensitivity analyses (different models and subgroups based on anatomy and clinical indication and excluding unpublished trials). There were no influential single trials. Level of certainty in evidence was downrated from high to moderate because of sparse events in some studies.

CONCLUSION

There appears to be heightened risk of major amputation after use of paclitaxel coated balloons in the peripheral arteries. Further investigations are warranted urgently.

Midodrine for the Prevention of Vasovagal Syncope : A Randomized Clinical Trial

BACKGROUND

Recurrent vasovagal syncope is common, responds poorly to treatment, and causes physical trauma and poor quality of life. Midodrine prevents hypotension and syncope during tilt tests in patients with vasovagal syncope.

OBJECTIVE

To determine whether midodrine can prevent vasovagal syncope in usual clinical conditions.

DESIGN

Randomized, double-blind, placebo-controlled clinical trial. (ClinicalTrials.gov: NCT01456481).

SETTING

25 university hospitals in Canada, the United States, Mexico, and the United Kingdom.

PATIENTS

Patients with recurrent vasovagal syncope and no serious comorbid conditions.

INTERVENTION

Patients were randomly assigned 1:1 to placebo or midodrine and followed for 12 months.

MEASUREMENTS

The primary outcome measure was the proportion of patients with at least 1 syncope episode during follow-up.

RESULTS

The study included 133 patients who had had a median of 6 syncope episodes in the prior year (median age, 32 years; 73% female). Compared with patients receiving placebo, fewer patients receiving midodrine had at least 1 syncope episode (28 of 66 [42%] vs. 41 of 67 [61%]). The relative risk was 0.69 (95% CI, 0.49 to 0.97; P = 0.035). The absolute risk reduction was 19 percentage points (CI, 2 to 36 percentage points), and the number needed to treat to prevent 1 patient from having syncope was 5.3 (CI, 2.8 to 47.6). The time to first syncope was longer with midodrine (hazard ratio, 0.59 [CI, 0.37 to 0.96]; P = 0.035; log-rank P = 0.031). Adverse effects were similar in both groups.

LIMITATION

Small study size, young and healthy patients, relatively short observation period, and high proportion of patients from 1 center.

CONCLUSION

Midodrine can reduce the recurrence of syncope in healthy, younger patients with a high syncope burden.

PRIMARY FUNDING SOURCE

The Canadian Institutes of Health Research.

A review of the use of controlled multiple imputation in randomised controlled trials with missing outcome data

Background

Missing data are common in randomised controlled trials (RCTs) and can bias results if not handled appropriately. A statistically valid analysis under the primary missing-data assumptions should be conducted, followed by sensitivity analysis under alternative justified assumptions to assess the robustness of results. Controlled Multiple Imputation (MI) procedures, including delta-based and reference-based approaches, have been developed for analysis under missing-not-at-random assumptions. However, it is unclear how often these methods are used, how they are reported, and what their impact is on trial results. This review evaluates the current use and reporting of MI and controlled MI in RCTs.

Methods

A targeted review of phase II-IV RCTs (non-cluster randomised) published in two leading general medical journals (The Lancet and New England Journal of Medicine) between January 2014 and December 2019 using MI. Data was extracted on imputation methods, analysis status, and reporting of results. Results of primary and sensitivity analyses for trials using controlled MI analyses were compared.

Results

A total of 118 RCTs (9% of published RCTs) used some form of MI. MI under missing-at-random was used in 110 trials; this was for primary analysis in 43/118 (36%), and in sensitivity analysis for 70/118 (59%) (3 used in both). Sixteen studies performed controlled MI (1.3% of published RCTs), either with a delta-based (n = 9) or reference-based approach (n = 7). Controlled MI was mostly used in sensitivity analysis (n = 14/16). Two trials used controlled MI for primary analysis, including one reporting no sensitivity analysis whilst the other reported similar results without imputation. Of the 14 trials using controlled MI in sensitivity analysis, 12 yielded comparable results to the primary analysis whereas 2 demonstrated contradicting results. Only 5/110 (5%) trials using missing-at-random MI and 5/16 (31%) trials using controlled MI reported complete details on MI methods.

Conclusions

Controlled MI enabled the impact of accessible contextually relevant missing data assumptions to be examined on trial results. The use of controlled MI is increasing but is still infrequent and poorly reported where used. There is a need for improved reporting on the implementation of MI analyses and choice of controlled MI parameters.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01261-6.

Investigation of the efficacy of the short regimen for rifampicin-resistant TB from the STREAM trial

BACKGROUND

The STREAM trial demonstrated that a 9-11-month "short" regimen had non-inferior efficacy and comparable safety to a 20+ month "long" regimen for the treatment of rifampicin-resistant tuberculosis. Imbalance in the components of the composite primary outcome merited further investigation.

METHODS

Firstly, the STREAM primary outcomes were mapped to alternatives in current use, including WHO programmatic outcome definitions and other recently proposed modifications for programmatic or research purposes. Secondly, the outcomes were re-classified according to the likelihood that it was a Failure or Relapse (FoR) event on a 5-point Likert scale: Definite, Probable, Possible, Unlikely, and Highly Unlikely. Sensitivity analyses were employed to explore the impact of informative censoring. The protocol-defined modified intention-to-treat (MITT) analysis population was used for all analyses.

RESULTS

Cure on the short regimen ranged from 75.1 to 84.2% across five alternative outcomes. However, between-regimens results did not exceed 1.3% in favor of the long regimen (95% CI upper bound 10.1%), similar to the primary efficacy results from the trial. Considering only Definite or Probable FoR events, there was weak evidence of a higher risk of FoR in the short regimen, HR 2.19 (95%CI 0.90, 5.35), p = 0.076; considering only Definite FoR events, the evidence was stronger, HR 3.53 (95%CI 1.05, 11.87), p = 0.030. Cumulative number of grade 3-4 AEs was the strongest predictor of censoring. Considering a larger effect of informative censoring attenuated treatment differences, although 95% CI were very wide.

CONCLUSION

Five alternative outcome definitions gave similar overall results. The risk of failure or relapse (FoR) may be higher in the short regimen than in the long regimen, highlighting the importance of how loss to follow-up and other censoring is accounted for in analyses. The outcome of time to FoR should be considered as a primary outcome for future drug-sensitive and drug-resistant TB treatment trials, provided sensitivity analyses exploring the impact of departures from independent censoring are also included.

Comparison of censoring assumptions to reduce bias in tuberculosis treatment cohort analyses

Objective

Observational tuberculosis cohort studies are often limited by a lack of long-term data characterizing survival beyond the initial treatment outcome. Though Cox proportional hazards models are often applied to these data, differential risk of long-term survival, dependent on the initial treatment outcome, can lead to violations of model assumptions. We evaluate the performance of two alternate censoring approaches on reducing bias in treatment effect estimates.

Design

We simulate a typical multidrug-resistant tuberculosis cohort study and use Cox proportional hazards models to assess the relationship of an aggressive treatment regimen with hazard of death. We compare three assumptions regarding censored observations to determine which produces least biased treatment effect estimates: conventional non-informative censoring, an extension of short-term survival informed by literature, and incorporation of predicted long-term vital status.

Results

The treatment regimen’s protective effect on death is consistently underestimated by the conventional censoring method, up to 7.6%. Models using the two alternative censoring techniques produce treatment effect estimates consistently stronger and less biased than the conventional method, underestimating the treatment effect by less than 2.4% across all scenarios.

Conclusions

When model assumptions are violated, alternative censoring techniques can more accurately estimate associations between treatment and long-term survival. In multidrug-resistant tuberculosis cohort analyses, this bias reduction may yield more accurate and, larger effect estimates. This bias reduction can be achieved through use of standard statistical procedures with a simple re-coding of the censoring indicator.

GRADE Guidelines: 29. Rating the certainty in time-to-event outcomes-Study limitations due to censoring of participants with missing data in intervention studies

OBJECTIVES

To provide Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) guidance for the consideration of study limitations (risk of bias) due to missing participant outcome data for time-to-event outcomes in intervention studies.

STUDY DESIGN AND SETTING

We developed this guidance through an iterative process that included membership consultation, feedback, presentation, and iterative discussion at meetings of the GRADE working group.

RESULTS

The GRADE working group has published guidance on how to account for missing participant outcome data in binary and continuous outcomes. When analyzing time-to-event outcomes (e.g., overall survival and time-to-treatment failure) data of participants for whom the outcome of interest (e.g., death and relapse) has not been observed are dealt with through censoring. To do so, standard methods require that censored individuals are representative for those remaining in the study. Two types of censoring can be distinguished, end of study censoring and censoring because of missing data, commonly named loss to follow-up censoring. However, both types are not distinguishable with the usual information on censoring available to review authors. Dealing with individuals for whom data are missing during follow-up in the same way as individuals for whom full follow-up is available at the end of the study increases the risk of bias. Considerable differences in the treatment arms in the distribution of censoring over time (early versus late censoring), the overall degree of missing follow-up data, and the reasons why individuals were lost to follow-up may reduce the certainty in the study results. With often only very limited data available, review and guideline authors are required to make transparent and well-considered judgments when judging risk of bias of individual studies and then come to an overall grading decision for the entire body of evidence.

CONCLUSION

Concern for risk of bias resulting from censoring of participants for whom follow-up data are missing in the underlying studies of a body of evidence can be expressed in the study limitations (risk of bias) domain of the GRADE approach.

A four-step strategy for handling missing outcome data in randomised trials affected by a pandemic

BACKGROUND

The coronavirus pandemic (Covid-19) presents a variety of challenges for ongoing clinical trials, including an inevitably higher rate of missing outcome data, with new and non-standard reasons for missingness. International drug trial guidelines recommend trialists review plans for handling missing data in the conduct and statistical analysis, but clear recommendations are lacking.

METHODS

We present a four-step strategy for handling missing outcome data in the analysis of randomised trials that are ongoing during a pandemic. We consider handling missing data arising due to (i) participant infection, (ii) treatment disruptions and (iii) loss to follow-up. We consider both settings where treatment effects for a 'pandemic-free world' and 'world including a pandemic' are of interest.

RESULTS

In any trial, investigators should; (1) Clarify the treatment estimand of interest with respect to the occurrence of the pandemic; (2) Establish what data are missing for the chosen estimand; (3) Perform primary analysis under the most plausible missing data assumptions followed by; (4) Sensitivity analysis under alternative plausible assumptions. To obtain an estimate of the treatment effect in a 'pandemic-free world', participant data that are clinically affected by the pandemic (directly due to infection or indirectly via treatment disruptions) are not relevant and can be set to missing. For primary analysis, a missing-at-random assumption that conditions on all observed data that are expected to be associated with both the outcome and missingness may be most plausible. For the treatment effect in the 'world including a pandemic', all participant data is relevant and should be included in the analysis. For primary analysis, a missing-at-random assumption - potentially incorporating a pandemic time-period indicator and participant infection status - or a missing-not-at-random assumption with a poorer response may be most relevant, depending on the setting. In all scenarios, sensitivity analysis under credible missing-not-at-random assumptions should be used to evaluate the robustness of results. We highlight controlled multiple imputation as an accessible tool for conducting sensitivity analyses.

CONCLUSIONS

Missing data problems will be exacerbated for trials active during the Covid-19 pandemic. This four-step strategy will facilitate clear thinking about the appropriate analysis for relevant questions of interest.

Dismantling the Fragility Index: A demonstration of statistical reasoning

The Fragility Index has been introduced as a complement to the P-value to summarize the statistical strength of evidence for a trial's result. The Fragility Index (FI) is defined in trials with two equal treatment group sizes, with a dichotomous or time-to-event outcome, and is calculated as the minimum number of conversions from nonevent to event in the treatment group needed to shift the P-value from Fisher's exact test over the .05 threshold. As the index lacks a well-defined probability motivation, its interpretation is challenging for consumers. We clarify what the FI may be capturing by separately considering two scenarios: (a) what the FI is capturing mathematically when the probability model is correct and (b) how well the FI captures violations of probability model assumptions. By calculating the posterior probability of a treatment effect, we show that when the probability model is correct, the FI inappropriately penalizes small trials for using fewer events than larger trials to achieve the same significance level. The analysis shows that for experiments conducted without bias, the FI promotes an incorrect intuition of probability, which has not been noted elsewhere and must be dispelled. We illustrate shortcomings of the FI's ability to quantify departures from model assumptions and contextualize the FI concept within current debate around the null hypothesis significance testing paradigm. Altogether, the FI creates more confusion than it resolves and does not promote statistical thinking. We recommend against its use. Instead, sensitivity analyses are recommended to quantify and communicate robustness of trial results.

Transcatheter InterAtrial Shunt Device for the treatment of heart failure: Rationale and design of the pivotal randomized trial to REDUCE Elevated Left Atrial Pressure in Patients with Heart Failure II (REDUCE LAP-HF II)

BACKGROUND

A randomized, sham-controlled trial in patients with heart failure (HF) and left ventricular ejection fraction (LVEF) ≥40% demonstrated reductions in pulmonary capillary wedge pressure (PCWP) with a novel transcatheter InterAtrial Shunt Device (IASD). Whether this hemodynamic effect will translate to an improvement in cardiovascular outcomes and symptoms requires additional study.

STUDY DESIGN AND OBJECTIVES

REDUCE Elevated Left Atrial Pressure in Patients with Heart Failure II (REDUCE LAP HF-II) is a multicenter, prospective, randomized, sham-controlled, blinded trial designed to evaluate the clinical efficacy of the IASD in symptomatic HF and elevated left atrial pressures. Up to 608 HF patients age ≥ 40 years with LVEF ≥40%, PCWP ≥25 mm Hg during supine ergometer exercise, and PCWP ≥5 mm Hg higher than right atrial pressure will be randomized 1:1 to the IASD versus sham control. Key exclusion criteria include hemodynamically significant valvular disease, evidence of pulmonary arterial hypertension, and right heart dysfunction. The primary endpoint is a hierarchical composite, analyzed by the Finkelstein-Schoenfeld methodology, that includes (1) cardiovascular mortality or first nonfatal ischemic stroke through 12 months; (2) total (first plus recurrent) HF hospitalizations or healthcare facility visits for intravenous diuretics up to 24 months, analyzed when the last randomized patient completes 12 months of follow-up; and (3) change in Kansas City Cardiomyopathy Questionnaire overall summary score from baseline to 12 months. Follow-up echocardiography will be performed at 6, 12, and 24 months to evaluate shunt flow and cardiac chamber size/function. Patients will be followed for a total of 5 years after the index procedure.

CONCLUSIONS

REDUCE LAP-HF II is designed to evaluate the clinical efficacy of the IASD device in patients with symptomatic HF with elevated left atrial pressure and LVEF ≥40%.

Strengthening the interpretability of clinical trial results by assessing the effect of informative censoring on the primary estimand in PRECISION

BACKGROUND

The ICH E9(R1) addendum states that the strategy to account for intercurrent events should be included when defining an estimand, the treatment effect to be estimated based on the study objective. The estimator used to assess the treatment effect needs to be aligned with the estimand that accounted for intercurrent events. Regardless of the strategy, missing data resulting from patient premature withdrawal could undermine the robustness of the study results. Informative censoring due to dropouts in an events-based study is one such example. Sensitivity analyses using imputation methods are useful to examine the uncertainty due to informative censoring and address the robustness and strength of the study results.

METHODS

We assessed the effect of premature patient withdrawal in the PRECISION study, a randomized non-inferiority clinical trial of patients with chronic arthritic pain that compared the cardiovascular safety of three nonsteroidal anti-inflammatory drugs-based treatment policies or paradigms. The protocol-defined use of concomitant or rescue medications was permitted since changes in pain medications due to insufficient analgesia were expected in patients in this long-term study. Anticipating that premature study discontinuations could potentially lead to informative censoring, a supplementary analysis was pre-specified in which censored outcomes due to the premature study discontinuation were imputed based on adverse events that were clinically associated with the primary endpoint (cardiovascular outcome based on the Antiplatelet Trialists Collaboration composite endpoint). Furthermore, tipping point analyses were conducted to test the robustness of the primary analysis results by assuming data censored not at random. The level of increase at which the primary study conclusion would change was estimated.

RESULTS

For the analysis of time to first primary endpoint event through 30 months, 4065 out of the 24,081 enrolled patients were lost to follow-up, withdrew consent, or were no longer willing to participate in the study. These withdrawals occurred gradually and resulted in a cumulative total of 5893 censored patient-years of observation (10.2%). The rate of discontinuation and the baseline characteristics of the discontinued patients were similar across the three treatment groups. The non-inferiority conclusion from the primary analysis was confirmed in the supplementary analysis incorporating relevant adverse events. Furthermore, tipping point analyses demonstrated that in order to lose non-inferiority in the primary analysis, the risk of primary endpoint events during the censored observation time would have to increase by more than 2.7-fold in the celecoxib group while remaining constant in the other nonsteroidal anti-inflammatory drugs groups, demonstrating that the scenarios where the study results are invalid appear not plausible.

CONCLUSIONS

Supplementary and sensitivity analyses presented to address informative censoring in PRECISION helped to further interpret and strengthen the study results.

Prediction meets causal inference: the role of treatment in clinical prediction models

In this paper we study approaches for dealing with treatment when developing a clinical prediction model. Analogous to the estimand framework recently proposed by the European Medicines Agency for clinical trials, we propose a 'predictimand' framework of different questions that may be of interest when predicting risk in relation to treatment started after baseline. We provide a formal definition of the estimands matching these questions, give examples of settings in which each is useful and discuss appropriate estimators including their assumptions. We illustrate the impact of the predictimand choice in a dataset of patients with end-stage kidney disease. We argue that clearly defining the estimand is equally important in prediction research as in causal inference.

Reduction in All-Cause Mortality with Fluticasone Furoate/Umeclidinium/Vilanterol in Patients with Chronic Obstructive Pulmonary Disease

Rationale: The IMPACT (Informing the Pathway of Chronic Obstructive Pulmonary Disease Treatment) trial demonstrated a significant reduction in all-cause mortality (ACM) risk with fluticasone furoate/umeclidinium/vilanterol (FF/UMEC/VI) versus UMEC/VI in patients with chronic obstructive pulmonary disease (COPD) at risk of future exacerbations. Five hundred seventy-four patients were censored in the original analysis owing to incomplete vital status information.Objectives: Report ACM and impact of stepping down therapy, following collection of additional vital status data.Methods: Patients were randomized 2:2:1 to FF/UMEC/VI 100/62.5/25 μg, FF/VI 100/25 μg, or UMEC/VI 62.5/25 μg following a run-in on their COPD therapies. Time to ACM was prespecified. Additional vital status data collection and subsequent analyses were performed post hoc.Measurements and Main Results: We report vital status data for 99.6% of the intention-to-treat population (n = 10,355), documenting 98 (2.36%) deaths on FF/UMEC/VI, 109 (2.64%) on FF/VI, and 66 (3.19%) on UMEC/VI. For FF/UMEC/VI, the hazard ratio for death was 0.72 (95% confidence interval, 0.53-0.99; P = 0.042) versus UMEC/VI and 0.89 (95% confidence interval, 0.67-1.16; P = 0.387) versus FF/VI. Independent adjudication confirmed lower rates of cardiovascular and respiratory death and death associated with the patient's COPD.Conclusions: In this secondary analysis of an efficacy outcome from the IMPACT trial, once-daily single-inhaler FF/UMEC/VI triple therapy reduced the risk of ACM versus UMEC/VI in patients with symptomatic COPD and a history of exacerbations.

Propensity score methods for time-dependent cluster confounding

In observational studies, subjects are often nested within clusters. In medical studies, patients are often treated by doctors and therefore patients are regarded as nested or clustered within doctors. A concern that arises with clustered data is that cluster-level characteristics (e.g., characteristics of the doctor) are associated with both treatment selection and patient outcomes, resulting in cluster-level confounding. Measuring and modeling cluster attributes can be difficult and statistical methods exist to control for all unmeasured cluster characteristics. An assumption of these methods however is that characteristics of the cluster and the effects of those characteristics on the outcome (as well as probability of treatment assignment when using covariate balancing methods) are constant over time. In this paper, we consider methods that relax this assumption and allow for estimation of treatment effects in the presence of unmeasured time-dependent cluster confounding. The methods are based on matching with the propensity score and incorporate unmeasured time-specific cluster effects by performing matching within clusters or using fixed- or random-cluster effects in the propensity score model. The methods are illustrated using data to compare the effectiveness of two total hip devices with respect to survival of the device and a simulation study is performed that compares the proposed methods. One method that was found to perform well is matching within surgeon clusters partitioned by time. Considerations in implementing the proposed methods are discussed.

Sensitivity analysis for clinical trials with missing continuous outcome data using controlled multiple imputation: A practical guide

Missing data due to loss to follow-up or intercurrent events are unintended, but unfortunately inevitable in clinical trials. Since the true values of missing data are never known, it is necessary to assess the impact of untestable and unavoidable assumptions about any unobserved data in sensitivity analysis. This tutorial provides an overview of controlled multiple imputation (MI) techniques and a practical guide to their use for sensitivity analysis of trials with missing continuous outcome data. These include δ- and reference-based MI procedures. In δ-based imputation, an offset term, δ, is typically added to the expected value of the missing data to assess the impact of unobserved participants having a worse or better response than those observed. Reference-based imputation draws imputed values with some reference to observed data in other groups of the trial, typically in other treatment arms. We illustrate the accessibility of these methods using data from a pediatric eczema trial and a chronic headache trial and provide Stata code to facilitate adoption. We discuss issues surrounding the choice of δ in δ-based sensitivity analysis. We also review the debate on variance estimation within reference-based analysis and justify the use of Rubin's variance estimator in this setting, since as we further elaborate on within, it provides information anchored inference.

Individual participant data meta-analysis of intervention studies with time-to-event outcomes: A review of the methodology and an applied example

Many randomized trials evaluate an intervention effect on time-to-event outcomes. Individual participant data (IPD) from such trials can be obtained and combined in a so-called IPD meta-analysis (IPD-MA), to summarize the overall intervention effect. We performed a narrative literature review to provide an overview of methods for conducting an IPD-MA of randomized intervention studies with a time-to-event outcome. We focused on identifying good methodological practice for modeling frailty of trial participants across trials, modeling heterogeneity of intervention effects, choosing appropriate association measures, dealing with (trial differences in) censoring and follow-up times, and addressing time-varying intervention effects and effect modification (interactions).We discuss how to achieve this using parametric and semi-parametric methods, and describe how to implement these in a one-stage or two-stage IPD-MA framework. We recommend exploring heterogeneity of the effect(s) through interaction and non-linear effects. Random effects should be applied to account for residual heterogeneity of the intervention effect. We provide further recommendations, many of which specific to IPD-MA of time-to-event data from randomized trials examining an intervention effect.We illustrate several key methods in a real IPD-MA, where IPD of 1225 participants from 5 randomized clinical trials were combined to compare the effects of Carbamazepine and Valproate on the incidence of epileptic seizures.

Sex-independent senescence in a cooperatively breeding mammal

Researchers studying mammals have frequently interpreted earlier or faster rates of ageing in males as resulting from polygyny and the associated higher costs of reproductive competition. Yet, few studies conducted on wild populations have compared sex-specific senescence trajectories outside of polygynous species, making it difficult to make generalized inferences on the role of reproductive competition in driving senescence, particularly when other differences between males and females might also contribute to sex-specific changes in performance across lifespan. Here, we examine age-related variation in body mass, reproductive output and survival in dominant male and female meerkats, Suricata suricatta. Meerkats are socially monogamous cooperative breeders where a single dominant pair virtually monopolizes reproduction in each group and subordinate group members help to rear offspring produced by breeders. In contrast to many polygynous societies, we find that neither the onset nor the rate of senescence in body mass or reproductive output shows clear differences between males and females. Both sexes also display similar patterns of age-related survival across lifespan, but unlike most wild vertebrates, survival senescence (increases in annual mortality with rising age) was absent in dominants of both sexes, and as a result, the fitness costs of senescence were entirely attributable to declines in reproductive output from mid- to late-life. We suggest that the potential for intrasexual competition to increase rates of senescence in females-who are hormonally masculinized and frequently aggressive-is offset by their ability to maintain longer tenures of dominance than males, and that these processes when combined lead to similar patterns of senescence in both sexes. Our results stress the need to consider the form and intensity of sexual competition as well as other sex-specific features of life history when investigating the operation of senescence in wild populations.

Long-term effects of intensive multifactorial therapy in individuals with screen-detected type 2 diabetes in primary care: 10-year follow-up of the ADDITION-Europe cluster-randomised trial

BACKGROUND

The multicentre, international ADDITION-Europe study investigated the effect of promoting intensive treatment of multiple risk factors among people with screen-detected type 2 diabetes over 5 years. Here we report the results of a post-hoc 10-year follow-up analysis of ADDITION-Europe to establish whether differences in treatment and cardiovascular risk factors have been maintained and to assess effects on cardiovascular outcomes.

METHODS

As previously described, general practices from four centres (Denmark, Cambridge [UK], Leicester [UK], and the Netherlands) were randomly assigned by computer-generated list to provide screening followed by routine care of diabetes, or screening followed by intensive multifactorial treatment. Population-based stepwise screening programmes among people aged 40-69 years (50-69 years in the Netherlands), between April, 2001, and December, 2006, identified patients with type 2 diabetes. Allocation was concealed from patients. Following the 5-year follow-up, no attempts were made to maintain differences in treatment between study groups. In this report, we did a post-hoc analysis of cardiovascular and renal outcomes over 10 years following randomisation, including a 5 years post-intervention follow-up. As in the original trial, the primary endpoint was a composite of first cardiovascular event, including cardiovascular mortality, cardiovascular morbidity (non-fatal myocardial infarction and non-fatal stroke), revascularisation, and non-traumatic amputation, up to Dec 31, 2014. Analyses were based on the intention-to-treat principle. ADDITION-Europe is registered with ClinicalTrials.gov, NCT00237549.

FINDINGS

343 general practices were randomly assigned to routine diabetes care (n=176) or intensive multifactorial treatment (n=167). 317 of these general practices (157 in the routine care group, 161 in the intensive treatment group) included eligible patients between April, 2001, and December, 2006. Of the 3233 individuals with screen-detected diabetes, 3057 agreed to participate (1379 in the routine care group, 1678 in the intensive treatment group), but at the 10-year follow-up 14 were lost to follow-up and 12 withdrew, leaving 3031 to enter 10-year follow-up analysis. Mean duration of follow-up was 9·61 years (SD 2·99). Sustained reductions over 10 years following diagnosis were apparent for bodyweight, HbA_1c, blood pressure, and cholesterol in both study groups, but between-group differences identified at 1 and 5 years were attenuated at the 10-year follow-up. By 10 years, 443 participants had a first cardiovascular event and 465 died. There was no significant difference between groups in the incidence of the primary composite outcome (16·1 per 1000 person-years in the routine care group vs 14·3 per 1000 person-years in the intensive treatment group; hazard ratio [HR] 0·87, 95% CI 0·73-1·04; p=0·14) or all-cause mortality (15·6 vs 14·3 per 1000 person-years; HR 0·90, 0·76-1·07).

INTERPRETATION

Sustained reductions in glycaemia and related cardiovascular risk factors over 10 years among people with screen-detected diabetes managed in primary care are achievable. The differences in prescribed treatment and cardiovascular risk factors in the 5 years following diagnosis were not maintained at 10 years, and the difference in cardiovascular events and mortality remained non-significant.

FUNDING

National Health Service Denmark, Danish Council for Strategic Research, Danish Research Foundation for General Practice, Novo Nordisk, Novo Nordisk Foundation, Danish Centre for Evaluation and Health Technology Assessment, Danish National Board of Health, Danish Medical Research Council, Aarhus University Research Foundation, Astra, Pfizer, GlaxoSmithKline, Servier, HemoCue, Wellcome Trust, UK Medical Research Council, UK National Institute for Health Research, UK National Health Service, Merck, Julius Center for Health Sciences and Primary Care, UK Department of Health, and Nuts-OHRA.

Trial of Satralizumab in Neuromyelitis Optica Spectrum Disorder

BACKGROUND

Neuromyelitis optica spectrum disorder (NMOSD) is an autoimmune disease of the central nervous system and is associated with autoantibodies to anti-aquaporin-4 (AQP4-IgG) in approximately two thirds of patients. Interleukin-6 is involved in the pathogenesis of the disorder. Satralizumab is a humanized monoclonal antibody targeting the interleukin-6 receptor. The efficacy of satralizumab added to immunosuppressant treatment in patients with NMOSD is unclear.

METHODS

In a phase 3, randomized, double-blind, placebo-controlled trial, we randomly assigned, in a 1:1 ratio, patients with NMOSD who were seropositive or seronegative for AQP4-IgG to receive either satralizumab, at a dose of 120 mg, or placebo, administered subcutaneously at weeks 0, 2, and 4 and every 4 weeks thereafter, added to stable immunosuppressant treatment. The primary end point was the first protocol-defined relapse in a time-to-event analysis. Key secondary end points were the change from baseline to week 24 in the visual-analogue scale (VAS) pain score (range, 0 to 100, with higher scores indicating more pain) and the Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-F) score (range, 0 to 52, with lower scores indicating more fatigue). Safety was also assessed.

RESULTS

A total of 83 patients were enrolled, with 41 assigned to the satralizumab group and 42 to the placebo group. The median treatment duration with satralizumab in the double-blind period was 107.4 weeks. Relapse occurred in 8 patients (20%) receiving satralizumab and in 18 (43%) receiving placebo (hazard ratio, 0.38; 95% confidence interval [CI], 0.16 to 0.88). Multiple imputation for censored data resulted in hazard ratios ranging from 0.34 to 0.44 (with corresponding P values of 0.01 to 0.04). Among 55 AQP4-IgG-seropositive patients, relapse occurred in 11% of those in the satralizumab group and in 43% of those in the placebo group (hazard ratio, 0.21; 95% CI, 0.06 to 0.75); among 28 AQP4-IgG-seronegative patients, relapse occurred in 36% and 43%, respectively (hazard ratio, 0.66; 95% CI, 0.20 to 2.24). The between-group difference in the change in the mean VAS pain score was 4.08 (95% CI, -8.44 to 16.61); the between-group difference in the change in the mean FACIT-F score was -3.10 (95% CI, -8.38 to 2.18). The rates of serious adverse events and infections did not differ between groups.

CONCLUSIONS

Among patients with NMOSD, satralizumab added to immunosuppressant treatment led to a lower risk of relapse than placebo but did not differ from placebo in its effect on pain or fatigue. (Funded by Chugai Pharmaceutical; ClinicalTrials.gov number, NCT02028884.).

Flexible parametric model for survival data subject to dependent censoring

When modeling survival data, it is common to assume that the (log-transformed) survival time (T) is conditionally independent of the (log-transformed) censoring time (C) given a set of covariates. There are numerous situations in which this assumption is not realistic, and a number of correction procedures have been developed for different models. However, in most cases, either some prior knowledge about the association between T and C is required, or some auxiliary information or data is/are supposed to be available. When this is not the case, the application of many existing methods turns out to be limited. The goal of this paper is to overcome this problem by developing a flexible parametric model, that is a type of transformed linear model. We show that the association between T and C is identifiable in this model. The performance of the proposed method is investigated both in an asymptotic way and through finite sample simulations. We also develop a formal goodness-of-fit test approach to assess the quality of the fitted model. Finally, the approach is applied to data coming from a study on liver transplants.

Reference-based sensitivity analysis for time-to-event data

The analysis of time‐to‐event data typically makes the censoring at random assumption, ie, that—conditional on covariates in the model—the distribution of event times is the same, whether they are observed or unobserved (ie, right censored). When patients who remain in follow‐up stay on their assigned treatment, then analysis under this assumption broadly addresses the de jure, or “while on treatment strategy” estimand. In such cases, we may well wish to explore the robustness of our inference to more pragmatic, de facto or “treatment policy strategy,” assumptions about the behaviour of patients post‐censoring.

This is particularly the case when censoring occurs because patients change, or revert, to the usual (ie, reference) standard of care. Recent work has shown how such questions can be addressed for trials with continuous outcome data and longitudinal follow‐up, using reference‐based multiple imputation. For example, patients in the active arm may have their missing data imputed assuming they reverted to the control (ie, reference) intervention on withdrawal. Reference‐based imputation has two advantages: (a) it avoids the user specifying numerous parameters describing the distribution of patients' postwithdrawal data and (b) it is, to a good approximation, information anchored, so that the proportion of information lost due to missing data under the primary analysis is held constant across the sensitivity analyses. In this article, we build on recent work in the survival context, proposing a class of reference‐based assumptions appropriate for time‐to‐event data. We report a simulation study exploring the extent to which the multiple imputation estimator (using Rubin's variance formula) is information anchored in this setting and then illustrate the approach by reanalysing data from a randomized trial, which compared medical therapy with angioplasty for patients presenting with angina.

Sexual Risk Outcomes of an Emotion Regulation Intervention for At-Risk Early Adolescents

OBJECTIVE

With this study, we examined the efficacy of a health intervention program that was focused on emotion regulation (ER) skills in reducing sexual risk behaviors among early adolescents with suspected mental health symptoms.

METHODS

Seventh grade adolescents with suspected mental health symptoms participated in a 6-week, after-school sexual risk prevention trial in which a counterbalanced, within-school design comparing an ER focused program to a time- and attention-matched comparison group was used. Adolescents completed a computer-based survey regarding their sexual behavior at 6-month intervals for 2.5 years.

RESULTS

Adolescents who received ER skills training exhibited a delay in the transition to vaginal sex over 30 months compared with those in the comparison condition (adjusted hazard ratio = 0.61; 95% confidence interval [0.42 to 0.89]). They also reported fewer instances of condomless sex over the follow-up period (adjusted rate ratio = 0.36; 95% confidence interval [0.14 to 0.90]). Among those who were sexually active, those in the ER condition reported fewer instances of vaginal or anal sex (adjusted rate ratio = 0.57; 95% confidence interval [0.32 to 0.99]).

CONCLUSIONS

An intervention used to teach ER skills for the context of health decision-making resulted in lower risk among young adolescents with suspected mental health symptoms by delaying the onset of vaginal sex as well as reducing penetrative acts without a condom. Incorporating emotion education into health education may have important health implications for this age group.

Valid and efficient subgroup analyses using nested case-control data

BACKGROUND

It is not uncommon for investigators to conduct further analyses of subgroups, using data collected in a nested case-control design. Since the sampling of the participants is related to the outcome of interest, the data at hand are not a representative sample of the population, and subgroup analyses need to be carefully considered for their validity and interpretation.

METHODS

We performed simulation studies, generating cohorts within the proportional hazards model framework and with covariate coefficients chosen to mimic realistic data and more extreme situations. From the cohorts we sampled nested case-control data and analysed the effect of a binary exposure on a time-to-event outcome in subgroups defined by a covariate (an independent risk factor, a confounder or an effect modifier) and compared the estimates with the corresponding subcohort estimates. Cohort analyses were performed with Cox regression, and nested case-control samples or restricted subsamples were analysed with both conditional logistic regression and weighted Cox regression.

RESULTS

For all studied scenarios, the subgroup analyses provided unbiased estimates of the exposure coefficients, with conditional logistic regression being less efficient than the weighted Cox regression.

CONCLUSIONS

For the study of a subpopulation, analysis of the corresponding subgroup of individuals sampled in a nested case-control design provides an unbiased estimate of the effect of exposure, regardless of whether the variable used to define the subgroup is a confounder, effect modifier or independent risk factor. Weighted Cox regression provides more efficient estimates than conditional logistic regression.

On the Use of Summary Comorbidity Measures for Prognosis and Survival Treatment Effect Estimation

Prognostic scores have been proposed as outcome based confounder adjustment scores akin to propensity scores. However, prognostic scores have not been widely used in the substantive literature. Instead, comorbidity scores, which are limited versions of prognostic scores, have been used extensively by clinical and health services researchers. A comorbidity is an existing disease an individual has in addition to a primary condition of interest, such as cancer. Comorbidity scores are used to reduce the dimension of a vector of comorbidity variables into a single scalar variable. Such scores are often added to regression models with other non-comorbidity variables such as age and sex, both for analyzing prognosis and for confounder adjustment when analyzing treatment effects. Despite their widespread use, the properties of and conditions under which comorbidity scores are valid dimension reduction tools in statistical models is largely unknown. In this article, we show that under relatively standard assumptions, comorbidity scores can have equal prognostic and confounder-adjustment abilities as the individual comorbidity variables, but that biases can occur if there are additional effects, such as interactions, of covariates beyond that captured by the comorbidity score. Simulations were performed to illustrate empirical properties and a data example using breast cancer data from the SEER Medicare Database demonstrates the application of these results.

Transplant as a competing risk in the analysis of dialysis patients

Time-to-event analyses are frequently used in nephrology research, for instance, when recording time to death or time to peritonitis in dialysis patients. Many papers have pointed out the important issue of competing events (or competing risks) in such analyses. For example, when studying one particular cause of death it can be noted that patients also die from other causes. Such competing events preclude the event of interest from occurring and thereby complicate the statistical analysis. The Kaplan-Meier approach to calculating the cumulative probability of the event of interest yields invalid results in the presence of competing risks, thus the alternative cumulative incidence competing risk (CICR) approach has become the standard. However, when kidney transplant is the competing event that prevents observing the outcome of interest, CICR may not always be the matter of interest. We discuss situations where both the Kaplan-Meier and the CICR approach are not suitable for the purpose and point out alternative analysis methods for such situations. We also look at the suitability and interpretation of different estimators for relative risks. In the presence of transplant as a competing risk, one should very clearly state the research question and use an analysis method that targets this question.

How serious is bias in effect estimation in randomised trials with survival data given risk heterogeneity and informative censoring?

It is often assumed that randomisation will prevent bias in estimation of treatment effects from clinical trials, but this is not true of the semiparametric Proportional Hazards model for survival data when there is underlying risk heterogeneity. Here, a new formula is proposed for estimation of this bias, improving on a previous formula through ease of use and clarity regarding the role of the mid-study cumulative hazard rate, shown to be an important factor for the bias magnitude. Informative censoring (IC) is recognised as a source of bias. Here, work on selection effects among survivors due to risk heterogeneity is extended to include IC. A new formula shows that bias in causal effect estimation under IC has two sources: one consequent on heterogeneity and one from the additional impact of IC. The formula provides new insights not previously shown: there may less bias under IC than when there is no IC and even, in principle, zero bias. When tested against simulated data, the new formulae are shown to be very accurate for prediction of bias in Proportional Hazards and accelerated failure time analyses which ignore heterogeneity. These data are also used to investigate the performance of accelerated failure time models which explicitly model risk heterogeneity ('frailty models') and G estimation. These methods remove bias when there is simple censoring but not with informative censoring when they may lead to overestimation of treatment effects. The new formulae may be used to help researchers judge the possible extent of bias in past studies. Copyright © 2017 John Wiley & Sons, Ltd.

A Retrospective Cohort Study of Acute Kidney Injury Risk Associated with Antipsychotics

BACKGROUND

A recent large database analysis raised concerns of potential acute kidney injury (AKI) risk associated with antipsychotics. However, whether individual atypical and typical antipsychotics are associated with differential AKI risks has not been investigated.

OBJECTIVE

The current study compared the risks of AKI and known causes of AKI associated with a broad range of atypical and typical antipsychotics.

METHOD

This retrospective cohort analysis used January 2007-June 2013 US nationwide Humana claims data to define episodes of antipsychotic drug therapy for patients with schizophrenia and bipolar disorder. Study drugs were aripiprazole, fluphenazine, haloperidol, olanzapine, quetiapine, risperidone, and ziprasidone. Study outcomes were hospitalizations with AKI, and known causes of AKI, i.e., hypotension, acute urinary retention, neuroleptic malignant syndrome/rhabdomyolysis, and pneumonia. AKI was the primary outcome of the study. Cox regressions using haloperidol as the baseline comparator were used to estimate the impact of alternative antipsychotics on the risks of study adverse events following the initiation of treatment. The Cox models controlled for treatment history, comorbidities, and concomitant drug use in the prior 6 months. They also controlled for patient demographics and dose of current treatment.

RESULTS

The overall incidence of AKI was 25.0 per 1000 person-years. According to our multivariate regression results, the risk of AKI was significantly increased in patients taking olanzapine [hazard ratio (HR) 1.344, 95% confidence interval (CI) 1.057-1.708], quetiapine (HR 1.350, 95% CI 1.082-1.685), and ziprasidone (HR 1.338, 95% CI 1.035-1.729) relative to haloperidol. Aripiprazole (HR 1.152, 95% CI 0.908-1.462) and risperidone (HR 1.147, 95% CI 0.923-1.426) had insignificantly higher risks of AKI compared with haloperidol, whereas fluphenazine (HR 0.729, 95% CI 0.483-1.102) had an insignificantly lower risk of AKI. When compared between drug classes, atypical antipsychotics had a significantly higher risk of AKI (HR 1.313, 95% CI 1.083-1.591) than typical antipsychotics.

CONCLUSIONS

Antipsychotics are associated with differential AKI risks, with several atypical antipsychotics having higher risks than haloperidol. However, the overall incidence of AKI was moderate, and AKI risk should only raise concern for clinicians with elderly patients or patients who are vulnerable to kidney disease.