Method for Incomplete and Imbalanced Data Based on Multivariate Imputation by Chained Equations and Ensemble Learning

The classification analysis of incomplete and imbalanced data is still a challenging task since these issues could negatively impact the training of classifiers, which were also found in our study on the physical fitness assessments of patients. And in fields such as healthcare, there are higher requirements for the accuracy of the generated imputation values. To train a high-performance classifier and pursue high accuracy, we attempted to resolve any potential negative impact by using a novel algorithmic approach based on the combination of multivariate imputation by chained equations and the ensemble learning method (MICEEN), which can solve the two problems simultaneously. We used multivariate imputation by chained equations to generate more accurate imputation values for the training set passed to ensemble learning to build a predictor. On the other hand, missing values were introduced into minority classes and used them to generate new samples belonging to the minority classes in order to balance the distribution of classes. On real-world datasets, we perform extensive experiments to assess our method and compare it to other state-of-the-art approaches. The advantages of the proposed method are demonstrated by experimental results for the benchmark datasets and self-collected datasets of physical fitness assessment of tumor patients with varying missing rates.

Reflections on estimands for patient-reported outcomes in cancer clinical trials

It is common and important to include the patient's perspective of the impact of treatment on health-related quality of life (HRQoL) outcomes. In this commentary, we focus on applying the new addendum to ICH E9 guideline E9 (R1) relating to the estimand framework to Patient Reported Outcomes (PROs) collected in cancer clinical trials, from a statistician's viewpoint. Currently, common practice for statistical analysis of PRO endpoints of published cancer clinical trials demonstrates ambiguity, leaving critical questions unspecified, hindering conclusions about the effect of treatment on PRO endpoints as well as comparability between clinical trials. To avoid this scenario, we advocate the systematic use of the estimand framework which requires the prospective definition of clear PRO research questions. Among the five attributes of the estimands framework, the definition of the endpoint (what is the right PRO measure and timeframe to target and why?), the intercurrent event identification and management (what happens with PRO data post-disease progression, what is the impact of death?) and the population-level summary (what is an acceptable statistical summary for PRO data?) require the most attention for PRO estimands. We identify good practice and highlight discussion points including the challenges of statistical analysis in the presence of missing and/or unobservable data and in relation to death. Through this discussion we highlight that there is no "statistical magic", but that the estimand framework will help you find out what you really want to know when quantifying the benefit of treatments from the patients' perspective.

Mepolizumab Reduces Systemic Corticosteroid Use in Chronic Rhinosinusitis With Nasal Polyps

BACKGROUND

Systemic corticosteroids (SCSs) are associated with short- and long-term adverse effects.

OBJECTIVE

To assess mepolizumab efficacy according to prior SCS use and characterize mepolizumab's SCS-sparing capabilities, in patients with severe chronic rhinosinusitis with nasal polyps.

METHODS

In the randomized, double-blind, phase III SYNAPSE trial (NCT03085797), adults with severe chronic rhinosinusitis with nasal polyps eligible for repeat sinus surgery despite standard of care treatment received mepolizumab (100 mg subcutaneously) or placebo every 4 weeks for 52 weeks. The impact of prior SCS courses (0/1/>1) on mepolizumab versus placebo treatment responses (changes from baseline in total endoscopic nasal polyp [week 52], nasal obstruction visual analog scale [weeks 49-52], and 22-item Sino-Nasal Outcome Test total [week 52] scores) was analyzed post hoc. To characterize mepolizumab's SCS-sparing capabilities, time-to-first SCS course for nasal polyps (prespecified) and total prednisolone-equivalent oral corticosteroid dose by patient baseline characteristics (post hoc, in patients with ≥1 SCS course during SYNAPSE) were assessed up to week 52.

RESULTS

Mepolizumab versus placebo improved treatment responses, irrespective of prior SCS use. By week 52, the probability of requiring SCSs for nasal polyps (Kaplan-Meier estimate [95% CI]) was lower with mepolizumab (25.4% [20.0-32.1]) versus placebo (37.5% [31.1-44.6]). In patients requiring 1 or more dose of SCSs, total (mean ± SD mg/y) prednisolone-equivalent oral corticosteroid dose was lower with mepolizumab (438.9 ± 350.40) versus placebo (505.2 ± 455.091), overall and irrespective of prior sinus surgeries, blood eosinophil count, or comorbidities.

CONCLUSIONS

Mepolizumab is associated with clinical benefits in patients with severe chronic rhinosinusitis with nasal polyps regardless of prior SCS use and has an SCS-sparing effect.

Type-I-error rate inflation in mixed models for repeated measures caused by ambiguous or incomplete model specifications

Pre-specification of the primary analysis model is a pre-requisite to control the family-wise type-I-error rate (T1E) at the intended level in confirmatory clinical trials. However, mixed models for repeated measures (MMRM) have been shown to be poorly specified in study protocols. The magnitude of a resulting T1E rate inflation is still unknown. This investigation aims to quantify the magnitude of the T1E rate inflation depending on the type and number of unspecified model items as well as different trial characteristics. We simulated a randomized, double-blind, parallel group, phase III clinical trial under the assumption that there is no treatment effect at any time point. The simulated data was analysed using different clusters, each including several MMRMs that are compatible with the imprecise pre-specification of the MMRM. T1E rates for each cluster were estimated. A significant T1E rate inflation could be shown for ambiguous model specifications with a maximum T1E rate of 7.6% [7.1%; 8.1%]. The results show that the magnitude of the T1E rate inflation depends on the type and number of unspecified model items as well as the sample size and allocation ratio. The imprecise specification of nuisance parameters may not lead to a significant T1E rate inflation. However, the results of this simulation study rather underestimate the true T1E rate inflation. In conclusion, imprecise MMRM specifications may lead to a substantial inflation of the T1E rate and can damage the ability to generate confirmatory evidence in pivotal clinical trials.

Factors associated with hearing outcomes after a middle fossa approach in 131 consecutive patients with vestibular schwannomas

OBJECTIVE

The middle fossa approach is an effective option for the treatment of small (Koos grade I and II) vestibular schwannomas (VSs) when the goal is hearing preservation. The authors evaluated the rates of hearing preservation and examined the factors associated with improved hearing outcomes after the middle fossa approach for VSs.

METHODS

In this retrospective, single-center cohort study evaluating the clinical outcomes after resection of small VSs using the middle fossa approach, consecutive adult patients (> 18 years) who underwent surgery between January 2000 and December 2021 were included. Clinical and imaging characteristics were analyzed, including baseline hearing status, duration of surgery, anesthetic parameters, and imaging characteristics of the surgically treated tumors.

RESULTS

Among the 131 included patients, 102 had valid and discoverable pre- and postoperative audiology assessments. The mean follow-up was 26 months (range 1-180 months). There were 85 patients with serviceable hearing preoperatively, defined as American Academy of Otolaryngology-Head and Neck Surgery (AAO-HNS) class A or B, of whom 78% retained class A or B hearing at the last follow-up. Binary logistic regression demonstrated that preoperative hearing AAO-HNS class (OR 0.19, 95% CI 0.05-0.77; p = 0.02), overlap between fundus and cochlea (OR 0.32, 95% CI 0.11-0.96; p = 0.04), and duration of anesthesia (OR 0.98, 95% CI 0.97-0.99; p = 0.03) were independent predictors of hearing outcomes. Additionally, 75% of patients with high diffusion-weighted imaging signal in the tumor (p = 0.009) and 67% of patients with the tumor originating at the modiolus of the cochlea (p = 0.004) had poor hearing outcomes.

CONCLUSIONS

The hearing preservation rates after microsurgical resection of small VSs using the middle fossa approach are high, with 78% of patients maintaining AAO-HNS class A or B hearing. Poor hearing status at baseline, longer duration of anesthesia, and large overlap between the fundus of the internal auditory canal and the cochlea were independently associated with unfavorable hearing outcomes. Imaging characteristics can be used to stratify patients' risk of hearing loss.

Sensitivity analyses in longitudinal clinical trials via distributional imputation

Missing data is inevitable in longitudinal clinical trials. Conventionally, the missing at random assumption is assumed to handle missingness, which however is unverifiable empirically. Thus, sensitivity analyses are critically important to assess the robustness of the study conclusions against untestable assumptions. Toward this end, regulatory agencies and the pharmaceutical industry use sensitivity models such as return-to-baseline, control-based, and washout imputation, following the ICH E9(R1) guidance. Multiple imputation is popular in sensitivity analyses; however, it may be inefficient and result in an unsatisfying interval estimation by Rubin's combining rule. We propose distributional imputation in sensitivity analysis, which imputes each missing value by samples from its target imputation model given the observed data. Drawn on the idea of Monte Carlo integration, the distributional imputation estimator solves the mean estimating equations of the imputed dataset. It is fully efficient with theoretical guarantees. Moreover, we propose weighted bootstrap to obtain a consistent variance estimator, taking into account the variabilities due to model parameter estimation and target parameter estimation. The superiority of the distributional imputation framework is validated in the simulation study and an antidepressant longitudinal clinical trial.

Pexidartinib Provides Modest Pain Relief in Patients With Tenosynovial Giant Cell Tumor: Results From ENLIVEN

BACKGROUND

The double-blind, randomized, placebo-controlled phase 3 study of orally administered PLX3397 in patients with pigmented villonodular synovitis or giant cell tumor of the tendon sheath (ENLIVEN) showed that pexidartinib provides a robust objective tumor response in adults with tenosynovial giant cell tumors (TGCT) not amenable to improvement with surgery. Based on these results, in 2019, pexidartinib received accelerated approval in the United States in this population as a breakthrough therapy under an orphan drug designation. However, the ability of pexidartinib to relieve pain in ENLIVEN was not fully detailed, and the relationship between pain relief and objective tumor response was not described.

QUESTIONS/PURPOSES

(1) What level of pain relief was achieved by pexidartinib treatment in ENLIVEN? (2) How was pain relief related to objective tumor responses? (3) How durable was pain relief?

METHODS

The current study included planned primary and exploratory assessments of patient-assessed worst pain at the site of the tumor in the ENLIVEN trial. ENLIVEN was a phase 3 randomized, placebo-controlled clinical trial in which adults with TGCT not amenable to improvement with surgery received pexidartinib or placebo for 24 weeks, after which eligible patients could receive open-label pexidartinib. Of 174 patients assessed for eligibility, 121 were randomized (50% [60] to placebo, 50% [61] to pexidartinib), and 120 were given either placebo or pexidartinib (59 received placebo and 61 received pexidartinib) and were included in an intent-to-treat analysis. Fifty-nine percent (71 of 120) of the overall treated population was female, and 88% (106 of 120) were White. Mean age was 45 ± 13 years. Tumors were mostly in the lower extremities (92% [110 of 120]), most commonly in the knee (61% [73 of 120]) and ankle (18% [21 of 120]). As a secondary outcome, patients scored worst pain at the site of the tumor in the past 24 hours on an 11-point numeric rating scale (NRS). The primary definition of a pain response was a decrease of at least 30% in the weekly mean worst-pain NRS score and increase of less than 30% in narcotic analgesic use between baseline and week 25. Planned exploratory assessments of pain included the frequency of a pain response using alternative thresholds, including a decrease in worst-pain NRS score of 50% or more and a decrease of at least 2 points (minimum clinically important difference [MCID]), the magnitude of pain reduction between baseline and week 25, correlation between worst-pain NRS score and tumor shrinkage by RECIST 1.1 criteria, and the durability of the pain response during the open-label extension. Pain responses during the randomized portion of the trial were compared according to intention-to-treat analysis, with a one-sided threshold of p < 0.025 to reduce the risk of false-positive results. Pain assessment was complete for 59% (35 of 59) of patients in the placebo group and 54% (33 of 61) of patients in the pexidartinib group. Demographic and disease characteristics did not differ between the two treatment groups.

RESULTS

A difference in the primary assessment of a pain response was not detected between pexidartinib and placebo (response percentage 31% [19 of 61] [95% CI 21% to 44%] versus 15% [9 of 59] [95% CI 8% to 27%]; one-sided p = 0.03). In the exploratory analyses, pexidartinib provided a modest improvement in pain (response percentage 26% [16 of 61] [95% CI 17% to 38%] versus 10% [6 of 59] [95% CI 5% to 20%]; one-sided p = 0.02 using the 50% threshold and 31% [19 of 61] [95% CI 21% to 44%] versus 14% [8 of 59] [95% CI 7% to 25%]; one-sided p = 0.02 using the MCID threshold). The least-squares mean change in the weekly mean worst-pain NRS score between baseline and week 25 was larger in patients treated with pexidartinib than placebo (-2.5 [95% CI -3.0 to -1.9] versus -0.3 [95% CI -0.9 to 0.3]; p < 0.001), although the mean difference between the two groups (-2.2 [95% CI -3.0 to -1.4]) was just over the MCID. Improvement in the weekly mean worst-pain NRS score correlated with the reduction in tumor size (r = 0.44; p < 0.001) and tumor volume score (r = 0.61; p < 0.001). For patients in the open-label extension, the change in the worst-pain NRS score from baseline was similar to the change at the end of the randomized portion and just above the MCID (mean -2.7 ± 2.2 after 25 weeks and -3.3 ± 1.7 after 50 weeks of receiving pexidartinib).

CONCLUSION

Based on the current study, a modest reduction in pain, just larger than the MCID, may be an added benefit of pexidartinib in these patients, although the findings are insufficient to justify the routine use of pexidartinib for pain relief.

LEVEL OF EVIDENCE

Level II, therapeutic study.

Impact of a magic recreation program on older adults with minor depressive symptoms in a long-term care facility: A pilot randomized controlled trial

OBJECTIVES

To evaluate a magic recreation program to reduce depressive symptoms in institutionalized older adults.

METHODS

We conducted a pilot randomized controlled trial in which participants were assigned to either a magic group (n = 6) or a control group with usual activities (n = 6). The magic group received a 6-week magic recreation program. The data were analyzed by generalized estimating equations in terms of intention-to-treat analysis. A sensitivity analysis was conducted by examining the complete case analysis.

RESULTS

The magic recreation program significantly improved the scores of Patient Health Questionnaire-9 in the magic group (Wald χ 2 = 8.816, p = 0.004, Cohen's  d  = 1.51, power = 0.9968). The results of the sensitivity analysis were consistent with the results of primary analysis.

CONCLUSIONS

The 6-week magic recreation program reduced depressive symptoms among institutionalized older adults with minor depressive symptoms.

A multi-task learning-based generative adversarial network for red tide multivariate time series imputation

Red tide data are typical multivariate time series (MTS) and complete data help analyze red tide more conveniently. However, missing values due to artificial or accidental events hinder further analysis of red tide phenomenon. Generative adversarial network (GAN) is effective in capturing distribution of MTS while the imputation performance is far from satisfactory, especially in conditions of high missing rate. One of the remaining open challenges is that common GAN-based imputation methods usually lack the ability to excavate implicit correlations between different attributions and downstream tasks, from which advanced latent information about missing values can be mined to improve imputation performance. To deal with the problem, a novel multi-task learning-based generative adversarial imputation network (MTGAIN) is proposed by introducing the prediction task into GAN to unearth more detailed information about missing values to better model distribution of red tide MTS. Furthermore, the homoscedastic uncertainty of multiple tasks is exploited to balance the weights of losses between generation and prediction tasks. The experiments conducted on a real-world dataset demonstrate that MTGAIN outperforms existing methods in terms of imputation and post-imputation performances, especially in conditions of high missing rate.

Fast tipping point sensitivity analyses in clinical trials with missing continuous outcomes under multiple imputation

When dealing with missing data in clinical trials, it is often convenient to work under simplifying assumptions, such as missing at random (MAR), and follow up with sensitivity analyses to address unverifiable missing data assumptions. One such sensitivity analysis, routinely requested by regulatory agencies, is the so-called tipping point analysis, in which the treatment effect is re-evaluated after adding a successively more extreme shift parameter to the predicted values among subjects with missing data. If the shift parameter needed to overturn the conclusion is so extreme that it is considered clinically implausible, then this indicates robustness to missing data assumptions. Tipping point analyses are frequently used in the context of continuous outcome data under multiple imputation. While simple to implement, computation can be cumbersome in the two-way setting where both comparator and active arms are shifted, essentially requiring the evaluation of a two-dimensional grid of models. We describe a computationally efficient approach to performing two-way tipping point analysis in the setting of continuous outcome data with multiple imputation. We show how geometric properties can lead to further simplification when exploring the impact of missing data. Lastly, we propose a novel extension to a multi-way setting which yields simple and general sufficient conditions for robustness to missing data assumptions.

Atrial fibrillation, quality of life and distress: a cluster analysis of cognitive and behavioural responses

PURPOSE

Few studies have examined specific cognitive and behavioural responses to symptoms, which may impact health-related outcomes, in conjunction with illness representations, as outlined by the Common-Sense-Model. Patients with atrial fibrillation (AF) report poor quality-of-life (QoL) and high distress. This cross-sectional study investigated patterns/clusters of cognitive and behavioural responses to illness, and illness perceptions, and relationships with QoL, depression and anxiety.

METHODS

AF patients (N = 198) recruited at cardiology clinics completed the AF-Revised Illness Perception Questionnaire, Atrial-Fibrillation-Effect-on-Quality-of-Life Questionnaire, Patient Health Questionnaire-8 and Generalized Anxiety Disorder Questionnaire. Cluster analysis used Ward's and K-means methods. Hierarchical regressions examined relationships between clusters with QoL, depression and anxiety.

RESULTS

Two clusters of cognitive and behavioural responses to symptoms were outlined; (1) 'high avoidance'; (2) 'low symptom-focussing'. Patients in Cluster 1 had lower QoL (M = 40.36, SD = 18.40), greater symptoms of depression (M = 7.20, SD = 5.71) and greater symptoms of anxiety (M = 5.70, SD = 5.90) compared to patients in Cluster 2 who had higher QoL (M = 59.03, SD = 20.12), fewer symptoms of depression (M = 3.53, SD = 3.56) and fewer symptoms of anxiety (M = 2.56, SD = 3.56). Two illness representation clusters were outlined; (1) 'high coherence and treatment control', (2) 'negative illness and emotional representations'. Patients in Cluster 2 had significantly lower QoL (M = 46.57, SD = 19.94), greater symptoms of depression (M = 6.12, SD = 5.31) and greater symptoms of anxiety (M = 4.70, SD = 5.27), compared with patients in Cluster 1 who had higher QoL (M = 61.52, SD = 21.38), fewer symptoms of depression (M = 2.85, SD = 2.97) and fewer symptoms of anxiety (M = 2.16, SD = 3.63). Overall, clusters of cognitive and behavioural responses to symptoms, and illness perceptions significantly explained between 14 and 29% of the variance in QoL, depression and anxiety.

CONCLUSION

Patterns of cognitive and behavioural responses to symptoms, and illness perceptions are important correlates of health-related outcomes in AF patients.

The role of family support and conflict in cyberbullying and subjective well-being among Chilean Adolescents during the Covid-19 period

Introduction

Life satisfaction plays a crucial role in integral development and mental health during childhood and adolescence. Recently, it has been shown that cyberbullying has severe consequences for the mental health and wellbeing of victims such as increased anxiety, depressive symptoms and even suicide risk. Although the role of the family in life satisfaction and cyberbullying behaviors has been studied, there is limited information on its impacts during the current pandemic period.

Objective

The aim of this study is to determine the role of family variables regarding students’ levels of life satisfaction and cyberbullying victimization during the pandemic period.

Method

Structural equation modeling was done using data from a cross-sectional study (n = 287; age ranged 14–18) conducted in six schools in Santiago, Chile during 2020.

Results

The tested model has a good fit and parsimonious adjustment. It explained 25.9% of the life satisfaction and 9% of the variance of cyberbullying victimization. Family support was positively associated with life satisfaction (p < 0.001) and negatively associated with cyberbullying victimization (p < 0.05). Likewise, family conflict was positively associated with levels of cyberbullying victimization (p < 0.05) and negatively associated with life satisfaction levels (p < 0.001). Finally, family visits were only positively associated with life satisfaction (p < 0.01).

Conclusions

Generating interventions on several levels focused on positive family bonds has become essential and urgent. This is especially important considering their protective impacts on cyberbullying victimization in promotion of adolescent well-being and quality of life.

•

Level of family support decreased cyberbullying victimization and increased life satisfaction during the Pandemic.

•

Level of family conflict increased the risk of cyberbullying victimization and decreased levels of life satisfaction.

•

Interventions and policies are urgently needed to promote positive family bonds and work-life balance.

Impute the missing data using retrieved dropouts

Background

In the past few decades various methods have been proposed to handle missing data of clinical studies, so as to assess the robustness of primary results. Some of the methods are based on the assumption of missing at random (MAR) which assumes subjects who discontinue the treatment will maintain the treatment effect after discontinuation. The agency, however, has expressed concern over methods based on this overly optimistic assumption, because it hardly holds for subjects discontinuing the investigational drug. Although in recent years a good number of sensitivity analyses based on missing not at random (MNAR) assumptions have been proposed, some use very conservative assumption on which it might be hard for sponsors and regulators to reach common ground.

Methods

Here we propose a multiple imputation method targeting at “treatment policy” estimand based on the MNAR assumption. This method can be used as the primary analysis, in addition to serving as a sensitivity analysis. It imputes missing data using information from retrieved dropouts defined as subjects who remain in the study despite occurrence of intercurrent events. Then imputed data long with completers and retrieved dropouts are analyzed altogether and finally multiple results are summarized into a single estimate. According to definition in ICH E9 (R1), this proposed approach fully aligns with the treatment policy estimand but its assumption is much more realistic and reasonable.

Results

Our approach has well controlled type I error rate with no loss of power. As expected, the effect size estimates take into account any dilution effect contributed by retrieved dropouts, conforming to the MNAR assumption.

Conclusions

Although multiple imputation approaches are always used as sensitivity analyses, this multiple imputation approach can be used as primary analysis for trials with sufficient retrieved dropouts or trials designed to collect retrieved dropouts.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-022-01509-9.

Standard and reference-based conditional mean imputation

Clinical trials with longitudinal outcomes typically include missing data due to missed assessments or structural missingness of outcomes after intercurrent events handled with a hypothetical strategy. Approaches based on Bayesian random multiple imputation and Rubin's rules for pooling results across multiple imputed data sets are increasingly used in order to align the analysis of these trials with the targeted estimand. We propose and justify deterministic conditional mean imputation combined with the jackknife for inference as an alternative approach. The method is applicable to imputations under a missing-at-random assumption as well as for reference-based imputation approaches. In an application and a simulation study, we demonstrate that it provides consistent treatment effect estimates with the Bayesian approach and reliable frequentist inference with accurate standard error estimation and type I error control. A further advantage of the method is that it does not rely on random sampling and is therefore replicable and unaffected by Monte Carlo error.

Effects of Internet-Based Cognitive Behavioral Therapy for Harmful Alcohol Use and Alcohol Dependence as Self-help or With Therapist Guidance: Three-Armed Randomized Trial

Background

Alcohol use is a major contributor to health loss. Many persons with harmful use or alcohol dependence do not obtain treatment because of limited availability or stigma. They may use internet-based interventions as an alternative way of obtaining support. Internet-based interventions have previously been shown to be effective in reducing alcohol consumption in studies that included hazardous use; however, few studies have been conducted with a specific focus on harmful use or alcohol dependence. The importance of therapist guidance in internet-based cognitive behavioral therapy (ICBT) programs is still unclear.

Objective

This trial aims to investigate the effects of a web-based alcohol program with or without therapist guidance among anonymous adult help-seekers.

Methods

A three-armed randomized controlled trial was conducted to compare therapist-guided ICBT and self-help ICBT with an information-only control condition. Swedish-speaking adult internet users with alcohol dependence (3 or more International Classification of Diseases, Tenth Revision criteria) or harmful alcohol use (alcohol use disorder identification test>15) were included in the study. Participants in the therapist-guided ICBT and self-help ICBT groups had 12-week access to a program consisting of 5 main modules, as well as a drinking calendar with automatic feedback. Guidance was given by experienced therapists trained in motivational interviewing. The primary outcome measure was weekly alcohol consumption in standard drinks (12 g of ethanol). Secondary outcomes were alcohol-related problems measured using the total alcohol use disorder identification test-score, diagnostic criteria for alcohol dependence and alcohol use disorder, depression, anxiety, health, readiness to change, and access to other treatments or support. Follow-up was conducted 3 (posttreatment) and 6 months after recruitment.

Results

During the recruitment period, from March 2015 to March 2017, 1169 participants were included. Participants had a mean age of 45 (SD 13) years, and 56.72% (663/1169) were women. At the 3-month follow-up, the therapist-guided ICBT and control groups differed significantly in weekly alcohol consumption (−3.84, 95% Cl −6.53 to −1.16; t₄₁₇=2.81; P=.005; Cohen d=0.27). No significant differences were found in weekly alcohol consumption between the self-help ICBT group and the therapist-guided ICBT at 3 months, between the self-help ICBT and the control group at 3 months, or between any of the groups at the 6-month follow-up. A limitation of the study was the large number of participants who were completely lost to follow-up (477/1169, 40.8%).

Conclusions

In this study, a therapist-guided ICBT program was not found to be more effective than the same program in a self-help ICBT version for reducing alcohol consumption or other alcohol-related outcomes. In the short run, therapist-guided ICBT was more effective than information. Only some internet help-seekers may need a multisession program and therapist guidance to change their drinking when they use internet-based interventions.

Trial Registration

ClinicalTrials.gov NCT02377726; https://clinicaltrials.gov/ct2/show/NCT02377726

Pexidartinib improves physical functioning and stiffness in patients with tenosynovial giant cell tumor: results from the ENLIVEN randomized clinical trial

Background and purpose - The ENLIVEN trial showed that, after 25 weeks, pexidartinib statistically significantly reduced tumor size more than placebo in patients with symptomatic, advanced tenosynovial giant cell tumor (TGCT) for whom surgery was not recommended. Here, we detail the effect of pexidartinib on patient-reported physical function and stiffness in ENLIVEN.Patients and methods - This was a planned analysis of patient-reported outcome data from ENLIVEN, a double-blinded, randomized phase 3 trial of adults with symptomatic, advanced TGCT treated with pexidartinib or placebo. Physical function was assessed using the Patient-Reported Outcomes Measurement Information System (PROMIS)-physical function (PF), and worst stiffness was assessed using a numerical rating scale (NRS). A mixed model for repeated measures was used to compare changes in PROMIS-PF and worst stiffness NRS scores from baseline to week 25 between treatment groups. Response rates for the PROMIS-PF and worst stiffness NRS at week 25 were calculated based on threshold estimates from reliable change index and anchor-based methods.Results - Between baseline and week 25, greater improvements in physical function and stiffness were experienced by patients receiving pexidartinib than patients receiving placebo (change in PROMIS-PF = 4.1 [95% confidence interval (CI) 1.8-6.3] vs. -0.9 [CI -3.0 to 1.2]; change in worst stiffness NRS = -2.5 [CI -3.0 to -1.9] vs. -0.3 [CI -0.9 to 0.3]). Patients receiving pexidartinib had higher response rates than patients receiving placebo for meaningful improvements in physical function and stiffness. Improvements were sustained after 50 weeks of pexidartinib treatment.Interpretation - Pexidartinib treatment provided sustained, meaningful improvements in physical function and stiffness for patients with symptomatic, advanced TGCT.

Identifying treatment effects using trimmed means when data are missing not at random

Patients often discontinue from a clinical trial because their health condition is not improving or they cannot tolerate the assigned treatment. Consequently, the observed clinical outcomes in the trial are likely better on average than if every patient had completed the trial. If these differences between trial completers and non-completers cannot be explained by the observed data, then the study outcomes are missing not at random (MNAR). One way to overcome this problem-the trimmed means approach for missing data due to study discontinuation-sets missing values as the worst observed outcome and then trims away a fraction of the distribution from each treatment arm before calculating differences in treatment efficacy (Permutt T, Li F. Trimmed means for symptom trials with dropouts. Pharm Stat. 2017;16(1):20-28). In this paper, we derive sufficient and necessary conditions for when this approach can identify the average population treatment effect. Simulation studies show the trimmed means approach's ability to effectively estimate treatment efficacy when data are MNAR and missingness due to study discontinuation is strongly associated with an unfavorable outcome, but trimmed means fail when data are missing at random. If the reasons for study discontinuation in a clinical trial are known, analysts can improve estimates with a combination of multiple imputation and the trimmed means approach when the assumptions of each hold. We compare the methodology to existing approaches using data from a clinical trial for chronic pain. An R package trim implements the method. When the assumptions are justifiable, using trimmed means can help identify treatment effects notwithstanding MNAR data.

Chronic obstructive pulmonary disease: moving from symptom relief to mortality reduction

Chronic obstructive pulmonary disease (COPD) has a 3-year mortality rate up to 37%, 2-6 times higher than the general population. We present evidence supporting pharmacological therapies to improve patient life expectancy, focusing on inhaled corticosteroids (ICSs) combined with long-acting bronchodilators (LABDs). A reduction in 3-year all-cause mortality (ACM) has been shown in patients with severe COPD treated with fluticasone propionate (an ICS) and salmeterol [long-acting beta-agonist (LABA)], compared with placebo. An observational study of elderly patients with severe COPD and multiple comorbidities suggested ICS+LABD reduce ACM compared with LABD monotherapy. Patients with symptomatic COPD at risk of exacerbations saw a mortality benefit with the ICS/long-acting muscarinic antagonist (LAMA)/LABA combinations fluticasone furoate/umeclidinium/vilanterol (FF/UMEC/VI) or budesonide/glycopyrrolate/formoterol (BUD/GLY/FOR) versus UMEC/VI or GLY/FOR (LAMA/LABA combinations) in the IMPACT and ETHOS trials, respectively. Reduced risk of mortality may be due to modulation of airway inflammation, thereby reducing activation of proinflammatory mediators in the peripheral circulation. Importantly, estimated annual risk reduction for ACM with ICS/LAMA/LABA combinations in patients with COPD is of the same order of magnitude as for statins (patients with coronary disease) and angiotensin-converting enzyme inhibitors (patients with vascular disease). Based on the current data, the pharmacological treatment of COPD appears not only able to improve symptoms and reduce the frequency of exacerbations but is also very promising in improving patient prognosis in the long term.

Mepolizumab for chronic rhinosinusitis with nasal polyps (SYNAPSE): a randomised, double-blind, placebo-controlled, phase 3 trial

BACKGROUND

Chronic rhinosinusitis with nasal polyps affects approximately 2-4% of the general population, and long-term use of systemic corticosteroids is associated with adverse effects. The aim of this study was to assess the efficacy and safety of mepolizumab in adults with recurrent, refractory severe bilateral chronic rhinosinusitis with nasal polyps.

METHODS

SYNAPSE was a randomised, double-blind, placebo-controlled, parallel-group, phase 3 trial done at 93 centres, mainly hospitals, in 11 countries. Eligible patients were aged 18 years or older with recurrent, refractory, severe, bilateral nasal polyp symptoms (nasal obstruction symptom visual analogue scale [VAS] score of >5), were eligible for repeat nasal surgery (overall symptoms VAS score >7 and endoscopic nasal polyps score of ≥5, with a minimum score of 2 in each nasal cavity) despite standard of care treatment, and had to have at least one nasal surgery in the past 10 years. Patients were randomly assigned (1:1), using permuted block design, to receive either 100 mg mepolizumab subcutaneously or placebo once every 4 weeks, in addition to standard of care (mometasone furoate intranasal spray for at least 8 weeks before screening and during the study, saline nasal irrigations, systemic corticosteroids or antibiotics, or both), as required, for 52 weeks. Site staff, the central study team, and patients were masked to study treatment and absolute blood eosinophil counts. The coprimary endpoints were change from baseline in total endoscopic nasal polyp score at week 52 and in mean nasal obstruction VAS score during weeks 49-52, assessed in the intention-to-treat population (ITT). This study is registered with ClinicalTrials.gov, NCT03085797.

FINDINGS

From May 25, 2017, to Dec 12, 2018, 854 patients were screened for eligibility. 414 patients were randomly assigned with 407 included in the ITT population; 206 received mepolizumab and 201 received placebo. Total endoscopic nasal polyp score significantly improved at week 52 from baseline with mepolizumab versus placebo (adjusted difference in medians -0·73, 95% CI -1·11 to -0·34; p<0·0001) and nasal obstruction VAS score during weeks 49-52 also significantly improved (-3·14, -4·09 to -2·18; p<0·0001). Adverse events considered related to study treatment were reported in 30 (15%) of 206 patients receiving mepolizumab and 19 (9%) of 201 receiving placebo. On-treatment serious adverse events occurred in 12 (6%) patients receiving mepolizumab and 13 (6%) receiving placebo; none were considered related to treatment in those receiving mepolizumab. One death was reported in the placebo group (myocardial infarction; death occurred 99 days after the last dose) and was considered unrelated to the treatment.

INTERPRETATION

Mepolizumab treatment improved nasal polyp size and nasal obstruction compared with placebo, with no new safety indications, in patients with recurrent, refractory severe chronic rhinosinusitis with nasal polyps. These findings suggest that mepolizumab provides an effective add-on treatment option to standard of care in this population.

FUNDING

GlaxoSmithKline.

Estimands and missing data in clinical trials of chronic pain treatments: advances in design and analysis

In clinical trials of treatments for chronic pain, the percentage of participants who withdraw early can be as high as 50%. Major reasons for early withdrawal in these studies include perceived lack of efficacy and adverse events. Commonly used strategies for accommodating missing data include last observation carried forward, baseline observation carried forward, and more principled methods such as mixed-model repeated-measures and multiple imputation. All these methods require strong and untestable assumptions concerning the conditional distribution of outcomes after dropout, given the observed data. We review recent developments in statistical methods for handling missing data in clinical trials, including implications of the increased emphasis being placed on precise formulation of the study objectives and the estimand (treatment effect to be estimated) of interest. A flexible method that seems to be well suited for the analysis of chronic pain clinical trials is control-based imputation, which allows a variety of assumptions to be made concerning the conditional distribution of postdropout outcomes that can be tailored to the estimand of interest. These assumptions can depend, for example, on the stated reasons for dropout. We illustrate these methods using data from 4 clinical trials of pregabalin for the treatment of painful diabetic peripheral neuropathy and postherpetic neuralgia. When planning chronic pain clinical trials, careful consideration of the trial objectives should determine the definition of the trial estimand, which in turn should inform methods used to accommodate missing data in the statistical analysis.

Quality of Life With Pembrolizumab for Recurrent and/or Metastatic Head and Neck Squamous Cell Carcinoma: KEYNOTE-040

BACKGROUND

Head and neck squamous cell carcinoma (HNSCC) affects health-related quality of life (HRQoL); few treatments have demonstrated clinically meaningful HRQoL benefit. KEYNOTE-040 evaluated pembrolizumab vs standard of care (SOC) in patients with recurrent and/or metastatic HNSCC whose disease recurred or progressed after platinum-containing regimen.

METHODS

Patients received pembrolizumab 200 mg or SOC (methotrexate, docetaxel, or cetuximab). Exploratory HRQoL analyses used European Organisation for Research and Treatment of Cancer (EORTC) 30 quality-of-life, EORTC 35-question quality-of-life head and neck cancer-specific module, and EuroQoL 5-dimensions questionnaires.

RESULTS

The HRQoL population comprised 469 patients (pembrolizumab = 241, SOC = 228). HRQoL compliance for patients in the study at week 15 was 75.3% (116 of 154) for pembrolizumab and 74.6% (85 of 114) for SOC. The median time to deterioration in global health status (GHS) and QoL scores were 4.8 months with pembrolizumab and 2.8 months with SOC (hazard ratio = 0.79, 95% confidence interval [CI] = 0.59 to 1.05). At week 15, GHS / QoL scores were stable for pembrolizumab (least squares mean [LSM] = 0.39, 95% CI = -3.00 to 3.78) but worsened for SOC (LSM = -5.86, 95% CI = -9.68 to -2.04); the LSM between-group difference was 6.25 points (95% CI = 1.32 to 11.18; nominal 2-sided P = .01). A greater difference in the LSM for GHS / QoL score occurred with pembrolizumab vs docetaxel (10.23, 95% CI = 3.15 to 17.30) compared with pembrolizumab vs methotrexate (6.21, 95% CI = -4.57 to 16.99) or pembrolizumab vs cetuximab (-1.44, 95% CI = -11.43 to 8.56). Pembrolizumab-treated patients had stable functioning and symptoms at week 15, with no notable differences from SOC.

CONCLUSIONS

GHS / QoL scores were stable with pembrolizumab but declined with SOC in patients at week 15, supporting the clinically meaningful benefit of pembrolizumab in recurrent and/or metastatic HNSCC.

Essential statistical principles of clinical trials of pain treatments

This article presents an overview of fundamental statistical principles of clinical trials of pain treatments. Statistical considerations relevant to phase 2 proof of concept and phase 3 confirmatory randomized trials investigating efficacy and safety are discussed, including (1) research design; (2) endpoints and analyses; (3) sample size determination and statistical power; (4) missing data and trial estimands; (5) data monitoring and interim analyses; and (6) interpretation of results. Although clinical trials of pharmacologic treatments are emphasized, the key issues raised by these trials are also directly applicable to clinical trials of other types of treatments, including biologics, devices, nonpharmacologic therapies (eg, physical therapy and cognitive-behavior therapy), and complementary and integrative health interventions.

Efficacy of Once-Weekly Semaglutide vs Empagliflozin Added to Metformin in Type 2 Diabetes: Patient-Level Meta-analysis

CONTEXT

No head-to-head trials have directly compared once-weekly (OW) semaglutide, a human glucagon-like peptide-1 analog, with empagliflozin, a sodium-glucose co-transporter-2 inhibitor, in type 2 diabetes (T2D).

OBJECTIVE

We indirectly compared the efficacy of OW semaglutide 1 mg vs once-daily (OD) empagliflozin 25 mg in patients with T2D inadequately controlled on metformin monotherapy, using individual patient data (IPD) and meta-regression methodology.

DESIGN, SETTING, PARTICIPANTS, AND INTERVENTIONS

IPD for patients with T2D receiving metformin monotherapy and randomized to OW semaglutide 1 mg (SUSTAIN 2, 3, 8 trials), or to OD empagliflozin 25 mg (PIONEER 2 trial) were included. Meta-regression analyses were adjusted for potential prognostic factors and effect modifiers.

MAIN OUTCOME MEASURES

The primary efficacy outcomes were change from baseline to end-of-treatment (~1 year) in HbA1c (%-point) and body weight (kg). Responder outcomes and other clinically relevant efficacy measures were analyzed.

RESULTS

Baseline characteristics were similar between OW semaglutide (n = 995) and empagliflozin (n = 410). Our analyses showed that OW semaglutide significantly reduced mean HbA1c and body weight vs empagliflozin (estimated treatment difference: -0.61%-point [95% confidence interval (CI): -0.72; -0.49] and -1.65 kg [95% CI: -2.22; -1.08], respectively; both P < 0.0001). Complementary analyses supported the robustness of these results. A significantly greater proportion of patients on OW semaglutide vs empagliflozin also achieved HbA1c targets and weight-loss responses.

CONCLUSIONS

This indirect comparison suggests that OW semaglutide 1 mg provides superior reductions in HbA1c and body weight vs OD empagliflozin 25 mg in patients with T2D when added to metformin monotherapy.

Defining estimands using a mix of strategies to handle intercurrent events in clinical trials

Randomized controlled trials (RCTs) are the gold standard for evaluation of the efficacy and safety of investigational interventions. If every patient in an RCT were to adhere to the randomized treatment, one could simply analyze the complete data to infer the treatment effect. However, intercurrent events (ICEs) including the use of concomitant medication for unsatisfactory efficacy, treatment discontinuation due to adverse events, or lack of efficacy may lead to interventions that deviate from the original treatment assignment. Therefore, defining the appropriate estimand (the appropriate parameter to be estimated) based on the primary objective of the study is critical prior to determining the statistical analysis method and analyzing the data. The International Council for Harmonisation (ICH) E9 (R1), adopted on November 20, 2019, provided five strategies to define the estimand: treatment policy, hypothetical, composite variable, while on treatment, and principal stratum. In this article, we propose an estimand using a mix of strategies in handling ICEs. This estimand is an average of the "null" treatment difference for those with ICEs potentially related to safety and the treatment difference for the other patients if they would complete the assigned treatments. Two examples from clinical trials evaluating antidiabetes treatments are provided to illustrate the estimation of this proposed estimand and to compare it with the estimates for estimands using hypothetical and treatment policy strategies in handling ICEs.

Additive integer-valued data envelopment analysis with missing data: A multi-criteria evaluation approach

Traditional data envelopment analysis (DEA) models assume that all the inputs and outputs data are available. However, missing data is a common problem in data analysis. Although several scholars have developed techniques to conduct DEA with missing data, these techniques have some disadvantages. A multi-criteria evaluation approach is proposed to measure the efficiency of decision making units (DMUs) with missing data. In this approach, analysts first estimate the upper and lower bounds of DMUs’ efficiency using the proposed I-addIDEA-U models (interval additive integer-valued DEA models with undesirable outputs) that can be applied to address integer-valued variables and undesirable outputs. Then, DMUs’ “relative” efficiency is evaluated using the proposed “Halo + Hot deck” DEA method (if there is no correlation between variables) or regression DEA techniques (if there is a correlation between variables). Finally, the multi-index comprehensive evaluation method is applied to determine which scenario (the lower bound of efficiency, the “relative” efficiency, or the upper bound of efficiency) should be selected. With a case study, it is shown that the proposed multi-criteria evaluation approach is more effective than traditional approaches such as the mean imputation DEA method, the deletion DEA method, and the dummy entries DEA method.

Empirical evaluation of the implementation of the EMA guideline on missing data in confirmatory clinical trials: Specification of mixed models for longitudinal data in study protocols

In confirmatory clinical trials, the prespecification of the primary analysis model is a universally accepted scientific principle to allow strict control of the type I error. Consequently, both the ICH E9 guideline and the European Medicines Agency (EMA) guideline on missing data in confirmatory clinical trials require that the primary analysis model is defined unambiguously. This requirement applies to mixed models for longitudinal data handling missing data implicitly. To evaluate the compliance with the EMA guideline, we evaluated the model specifications in those clinical study protocols from development phases II and III submitted between 2015 and 2018 to the Ethics Committee at Hannover Medical School under the German Medicinal Products Act, which planned to use a mixed model for longitudinal data in the confirmatory testing strategy. Overall, 39 trials from different types of sponsors and a wide range of therapeutic areas were evaluated. While nearly all protocols specify the fixed and random effects of the analysis model (95%), only 77% give the structure of the covariance matrix used for modeling the repeated measurements. Moreover, the testing method (36%), the estimation method (28%), the computation method (3%), and the fallback strategy (18%) are given by less than half the study protocols. Subgroup analyses indicate that these findings are universal and not specific to clinical trial phases or size of company. Altogether, our results show that guideline compliance is to various degrees poor and consequently, strict type I error rate control at the intended level is not guaranteed.

Incorporating and interpreting regulatory guidance on estimands in diabetes clinical trials: The PIONEER 1 randomized clinical trial as an example

Regulatory guidelines describe the use of estimands in designing and conducting clinical trials. Estimands ensure alignment of the objectives with the design, conduct and analysis of a trial. An estimand is defined by four inter-related attributes: the population of interest, the variable (endpoint) of interest, the way intercurrent events are handled and the population level summary. A trial may employ multiple estimands to evaluate treatment effects from different perspectives in order to address different scientific questions. As estimands may be an unfamiliar concept for many clinicians treating diabetes, this paper reviews the estimand concept and uses the PIONEER 1 phase 3a clinical trial, which investigated the efficacy and safety of oral semaglutide vs placebo, as an example of the way in which estimands can be implemented and interpreted. In the PIONEER 1 trial, two estimands were employed for each efficacy endpoint and were labelled as: (a) the treatment policy estimand, used to assess the treatment effect regardless of use of rescue medication or discontinuation of trial product, and provides a broad perspective of the treatment effect in the population of patients with type 2 diabetes in clinical practice; and (b) the trial product estimand, used to assess the treatment effect if all patients had continued to use trial product for the planned duration of the trial without rescue medication, thereby providing information on the anticipated treatment effect of the medication. Both approaches are complementary to understanding the effect of the studied treatments.

Assessment of Usability and Task Load Demand Using a Robot-Assisted Transfer Device Compared With a Hoyer Advance for Dependent Wheelchair Transfers

OBJECTIVE

Manual lifting can be burdensome for people who care for power wheelchair users. Although technologies used for dependent transfers are helpful, they have shortcomings of their own. This study compares the usability and task load demand of a novel robot-assisted transfer device to a clinical standard when performing dependent transfers.

DESIGN

A cross-sectional study was conducted to assess caregivers (N = 21) transferring a 56-kg mannequin with the Strong Arm and Hoyer Advance at three transfer locations. Feedback was gathered through qualitative surveys.

RESULTS

Usability was significant in multiple areas important for transfers. Caregiver fatigue and discomfort intensity were reduced, and the Strong Arm was preferred at the three transfer locations. Device ease and efficiency favored Strong Arm at two stations as was discomfort frequency. In addition, physical demand, frustration, and effort were significantly lower using Strong Arm compared with the Hoyer Advance.

CONCLUSIONS

Compared with the Hoyer, participants favored Strong Arm for transfer usability and task load demand. However, further Strong Arm developments are needed.

The relative efficiency of time-to-progression and continuous measures of cognition in presymptomatic Alzheimer's disease

Introduction

Clinical trials on preclinical Alzheimer’s disease are challenging because of the slow rate of disease progression. We use a simulation study to demonstrate that models of repeated cognitive assessments detect treatment effects more efficiently than models of time to progression.

Methods

Multivariate continuous data are simulated from a Bayesian joint mixed-effects model fit to data from the Alzheimer’s Disease Neuroimaging Initiative. Simulated progression events are algorithmically derived from the continuous assessments using a random forest model fit to the same data.

Results

We find that power is approximately doubled with models of repeated continuous outcomes compared with the time-to-progression analysis. The simulations also demonstrate that a plausible informative missing data pattern can induce a bias that inflates treatment effects, yet 5% type I error is maintained.

Discussion

Given the relative inefficiency of time to progression, it should be avoided as a primary analysis approach in clinical trials of preclinical Alzheimer’s disease.

A note on the draft International Council for Harmonisation guidance on estimands and sensitivity analysis

In the second half of 2014, the Steering Committee of the International Council for Harmonisation endorsed the formation of an expert working group to develop an addendum to the International Council for Harmonisation E9 guideline ( Statistical Principles for Clinical Trials). The addendum was to focus on two clinical trial topics: estimands and sensitivity analysis. A draft of the addendum, referred to as E9/R1, was developed by the expert working group and made available for public comments across the International Council for Harmonisation regions in the second half of 2017. A structured framework for clinical trial design and analysis proposed in the draft addendum are briefly described, including four key inputs for developing objective-driven estimands and strategies for tackling one of the inputs (‘intercurrent events’). The proposed framework aligns each clinical trial objective with the corresponding statistical target of estimation (estimand), trial design and data to be collected, main method of estimation/inference, and sensitivity analysis to pressure test key analytic assumption(s) in the main analysis. A case study from the diabetes therapeutic area illustrates how the framework can be implemented in practice. International Council for Harmonisation E9/R1 is expected to enable better planning, conduct, analysis, and interpretation of randomised clinical trials. This will facilitate improvements in new drug applications and strengthen understanding of decision making by regulatory authorities and advisory committees.

Strategies for composite estimands in confirmatory clinical trials: Examples from trials in nasal polyps and steroid reduction

The draft addendum to the ICH E9 regulatory guideline asks for explicit definition of the treatment effect to be estimated in clinical trials. The draft guideline also introduces the concept of intercurrent events to describe events that occur post-randomisation that may affect efficacy assessment. Composite estimands allow incorporation of intercurrent events in the definition of the endpoint. A common example of an intercurrent event is discontinuation of randomised treatment and use of a composite strategy would assess treatment effect based on a variable that combines the outcome variable of interest with discontinuation of randomised treatment. Use of a composite estimand may avoid the need for imputation which would be required by a treatment policy estimand. The draft guideline gives the example of a binary approach for specifying a composite estimand. When the variable is measured on a non-binary scale, other options are available where the intercurrent event is given an extreme unfavourable value, for example comparison of median values or analysis based on categories of response. This paper reviews approaches to deriving a composite estimand and contrasts the use of this estimand to the treatment policy estimand. The benefits of using each strategy are discussed and examples of the use of the different approaches are given for a clinical trial in nasal polyposis and a steroid reduction trial in severe asthma.

Strategic Shift of Statistical Review on Data Quality Assessment for New Drug Applications in China

Data quality is critical for clinical trials to obtain robust conclusions about drug safety and efficacy evaluation. Effective data quality evaluation has been one of the major obstacles to new drug approvals in China, which hinders innovation in drug discovery and development ultimately. To improve the data quality submitted for regulatory drug approval, the China Food and Drug Administration (CFDA) has issued serial official announcements and industry guidelines regarding improvement of the clinical trial data integrity and quality since 2015. These announcements and follow-up measures are shaping up the entire pharmaceutical industry in China. While data quality is being strongly emphasized more than ever at the trial conduction phase, it is still an open question about how to assess data quality effectively at the review stage. Thus, this article describes the authors' standpoints to assess the quality and integrity of submitted clinical data via statistical review methods including advanced risk-based approaches, which may bring significant impact to new drug applications and motivate sustainable development of innovative medicines in China.