Last Observation Carried Forward Research Articles

4-year results from the RAPID-PsA phase 3 randomised placebo-controlled trial of certolizumab pegol in psoriatic arthritis

ObjectiveTo report the efficacy, patient-reported, radiographic and safety outcomes of 4 years’ certolizumab pegol (CZP) treatment in patients with psoriatic arthritis (PsA).MethodsRAPID-PsA (NCT01087788) was double-blind and placebo-controlled to Week 24,...

Survival analysis with time-dependent covariates subject to missing data or measurement error: Multiple Imputation for Joint Modeling (MIJM)

Modern epidemiological studies collect data on time-varying individual-specific characteristics, such as body mass index and blood pressure. Incorporation of such time-dependent covariates in time-to-event models is of great interest, but raises some challenges. Of specific concern are measurement error, and the non-synchronous updating of covariates across individuals, due for example to missing data. It is well known that in the presence of either of these issues the last observation carried forward (LOCF) approach traditionally used leads to bias. Joint models of longitudinal and time-to-event outcomes, developed recently, address these complexities by specifying a model for the joint distribution of all processes and are commonly fitted by maximum likelihood or Bayesian approaches. However, the adequate specification of the full joint distribution can be a challenging modeling task, especially with multiple longitudinal markers. In fact, most available software packages are unable to handle more than one marker and offer a restricted choice of survival models. We propose a two-stage approach, Multiple Imputation for Joint Modeling (MIJM), to incorporate multiple time-dependent continuous covariates in the semi-parametric Cox and additive hazard models. Assuming a primary focus on the time-to-event model, the MIJM approach handles the joint distribution of the markers using multiple imputation by chained equations, a computationally convenient procedure that is widely available in mainstream statistical software. We developed an R package "survtd" that allows MIJM and other approaches in this manuscript to be applied easily, with just one call to its main function. A simulation study showed that MIJM performs well across a wide range of scenarios in terms of bias and coverage probability, particularly compared with LOCF, simpler two-stage approaches, and a Bayesian joint model. The Framingham Heart Study is used to illustrate the approach.

ADALIMUMAB FOR ULCERATIVE COLITIS: RESULTS OF A BRAZILIAN MULTICENTER OBSERVATIONAL STUDY.

Adalimumab is a monoclonal antibody, tumor necrosis factor-alpha (TNFα) inhibitor that has efficacy for inducing and maintaining remission in moderate-to-severe ulcerative colitis. Real world studies with adalimumab in Latin American ulcerative colitis patients are scarce. To assess the clinical remission rates in induction and maintenance with adalimumab therapy in ulcerative colitis. Observational, multicenter and retrospective study on a case series of patients with moderate-to-severe ulcerative colitis under adalimumab therapy. The variables analyzed were: demographic data, previous infliximab status, concomitant drugs, the Montreal Classification, disease activity (Mayo score) at weeks 0, 8, 26 and 52, or until the last follow-up. Clinical remission was defined as a partial Mayo score ≤2 and Last observation carried forward (LOCF) and Non responder imputation (NRI) analysis were used. Thirty-six patients were included in the study. With LOCF analysis, remission rates at weeks 8, 26 e 52 were of 41.7%, 47.2% and 47.2%, respectively. With NRI analysis, remission rates at weeks 8, 26 and 52 were of 41.7%, 41.7% and 27.8%, respectively. Adalimumab was effective in the treatment of moderate-to-severe ulcerative colitis. Clinical remission was observed in approximately 40% of the patients at weeks 8 and 26, and in almost a quarter of the patients after 1 year of follow up.

Post-hoc analysis of randomised, placebo-controlled, double-blind study (MCI186-19) of edaravone (MCI-186) in amyotrophic lateral sclerosis

Post-hoc analyses of the ALS Functional Rating Scale-Revised (ALSFRS-R) score data, the primary endpoint in the 24-week double-blind placebo-controlled study of edaravone (MCI186-19, NCT01492686), were performed to confirm statistical robustness of the result. The previously reported original analysis had used a last observation carried forward (LOCF) method and also excluded patients with fewer than three completed treatment cycles. The post-hoc sensitivity analyses used different statistical methods as follows: 1) including all patients regardless of treatment cycles received (ALL LOCF); 2) a mixed model for repeated measurements (MMRM) analysis; and 3) the Combined Assessment of Function and Survival (CAFS) endpoint. Findings were consistent with the original primary analysis in showing superiority of edaravone over placebo. We also investigated the distribution of change in ALSFRS-R total score across all patients in the study as well as which ALSFRS-R items and domains may have contributed to the overall efficacy findings. The distribution of changes in ALSFRS-R total score from baseline to the end of cycle 6 (ALL LOCF) shifted in favour of edaravone compared to placebo. Edaravone was descriptively favoured for each ALSFRS-R item and each of the four ALSFRS-R domains at the end of cycle 6 (ALL LOCF), suggesting a generalised effect of edaravone in slowing functional decline across all anatomical regions. The effect of edaravone appeared to be similar in patients with bulbar onset and limb onset. Together, these observations would be consistent with its putative neuroprotective effects against the development of oxidative damage unspecific to anatomical regions.

Safety and efficacy of edaravone in well defined patients with amyotrophic lateral sclerosis: a randomised, double-blind, placebo-controlled trial

In a previous phase 3 study in patients with amyotrophic lateral sclerosis (ALS), edaravone did not show a significant difference in the Revised ALS Functional Rating Scale (ALSFRS-R) score compared with placebo. Post-hoc analysis of these data revealed that patients in an early stage with definite or probable diagnosis of ALS, defined by the revised El Escorial criteria, who met a select set of inclusion criteria showed a greater magnitude of effect than did the full study population. We aimed to substantiate this post-hoc result and assess safety and efficacy of edaravone in a phase 3 trial that focused on patients with early stage ALS who met the post-hoc analysis inclusion criteria. In this phase 3, randomised, double-blind, parallel-group study, patients aged 20-75 years with ALS of grade 1 or 2 in the Japan ALS Severity Classification, scores of at least 2 points on all 12 items of ALSFRS-R, forced vital capacity of 80% or more, definite or probable ALS according to the revised El Escorial criteria, and disease duration of 2 years or less were recruited from 31 hospitals in Japan. Eligible patients also had a decrease of 1-4 points in the ALSFRS-R score during a 12-week observation period before randomisation. Patients meeting all criteria were then randomly assigned 1:1 to receive 60 mg intravenous edaravone or intravenous saline placebo for 6 cycles (4 weeks per cycle with 2 weeks on, 2 weeks off) for a total treatment duration of 24 weeks. In cycle 1, the study drug or placebo was administered once per day for 14 days within a 14 day period, followed by the drug-free period. In cycle 2 and thereafter, the study drug or placebo was administered for 10 days within a 14 day period, followed by a 2 week drug-free period. Participants and investigators, including those assessing outcomes, were masked to treatment allocation. The primary efficacy outcome was the change in ALSFRS-R score from the baseline to 24 weeks (or at discontinuation if this was after the third cycle) after randomisation. The primary outcome was assessed in all patients who had received at least one treatment infusion, had at least one assessment post-baseline, and reached the end of cycle 3. For patients with missing values at the end of cycle 6, data were imputed by the last observation carried forward (LOCF) method, provided the patients had completed at least cycle 3. Safety was assessed in all patients who had received at least one treatment infusion and had at least one assessment post-baseline. This trial is registered with ClinicalTrials.gov, NCT01492686. Between Nov 28, 2011, and Sept 3, 2014, we screened 213 patients, and enrolled 192 as potential participants. Of these, 137 patients completed the observation period: 69 were randomly assigned to receive edaravone, and 68 were randomly assigned to receive placebo. 68 patients taking edaravone and 66 taking placebo were included in the primary efficacy analysis. For the primary outcome, the change in ALSFRS-R score was -5·01 (SE 0·64) in the edavarone group and -7·50 (0·66) in the placebo group. The least-squares mean difference between groups was 2·49 (SE 0·76, 95% CI 0·99-3·98; p=0·0013) in favour of edaravone. Treatment-emergent adverse events were reported in 58 (84%) patients receiving edaravone and 57 (84%) patients receiving placebo. 11 (16%) patients taking edaravone and 16 (24%) taking placebo had serious adverse events, and one (1%) patient receiving edaravone and four (6%) patients receiving placebo had adverse events (one dysphagia in edaravone group and one dyspnoea, two respiratory disorder, and one rash in the placebo group) that led to withdrawal. Edaravone showed efficacy in a small subset of people with ALS who met criteria identified in post-hoc analysis of a previous phase 3 study, showing a significantly smaller decline of ALSFRS-R score compared with placebo. There is no indication that edaravone might be effective in a wider population of patients with ALS who do not meet the criteria. Mitsubishi Tanabe Pharma Corporation.

SA114. Lurasidone for the Treatment of Adolescent Patients With Schizophrenia: Effect on PANSS Subscales

Background: Lurasidone is an atypical antipsychotic agent that has demonstrated efficacy in the treatment of adults with schizophrenia in the dose range of 40–160 mg/d. The objective of this randomized, double-blind, placebo-controlled study was to evaluate the acute efficacy and safety of lurasidone in adolescent patients with schizophrenia. Methods: Adolescents (13–17 years old) diagnosed with schizophrenia were randomly assigned to 6 weeks of double-blind treatment with fixed-dose lurasidone 40 mg/d, 80 mg/d, or placebo. Changes from baseline to Week 6 in PANSS total and subscale (positive, negative, general psychopathology, excitability) scores were evaluated using mixed-model, repeated-measures analysis. Treatment response was defined as ≥20% reduction from baseline in PANSS total score at Week 6 and was analyzed using a logistic regression model with last observation carried forward (LOCF) approach. Results: A total of 326 patients (mean age, 15.4 years) were randomized and received lurasidone 40 mg/d (N = 108), 80 mg/d (N = 106), or placebo (N = 112). The PANSS total score at Week 6 demonstrated a placebo-adjusted, least-squares (LS) mean improvement of −8.0 (P < .001; effect size [ES] = 0.51) for the 40 mg/d group and −7.7 (P < .001; ES = 0.48) for the 80 mg/d group. Placebo-adjusted LS mean change for lurasidone 40 mg/d and 80 mg/d, respectively, was −3.2 (P < .001; ES = 0.62) and −3.2 (P < .001; ES = 0.60) on the PANSS positive subscale, −1.7 (P = .011; ES = 0.41) and −1.6 (P = .022; ES = 0.35) on the PANSS negative subscale, −2.8 (P = .012; ES = 0.38) and −2.8 (P = 0.011; ES = 0.37) on the PANSS general psychopathology subscale, and −1.1 (P = .016; ES = 0.36) and −1.8 (P < .001; ES = 0.53) on the PANSS excitability subscale. The proportion of treatment responders at Week 6 (LOCF) was 42.0% in the placebo group, 63.9% in the lurasidone 40 mg/d group (P < .001; NNT = 5), and 65.1% in the lurasidone 80 mg/d group (P < .001; NNT = 5). The most common adverse events (incidence ≥5% in either lurasidone group and at least twice the rate of placebo) for lurasidone 40 mg/d, 80 mg/d, and placebo, respectively, were nausea (12.7%, 14.4%, and 2.7%), somnolence (9.1%, 11.5%, and 5.4%), akathisia (9.1%, 8.7%, and 1.8%), vomiting (8.2%, 6.7%, and 1.8%), and sedation (5.5%, 1.9%, and 1.8%). The rate of study discontinuation due to adverse events was higher in placebo-treated versus lurasidone-treated patients (8.0% vs. 3.7%, respectively). Conclusion: In this double-blind, placebo-controlled study of adolescent patients with schizophrenia, lurasidone (40 mg/d and 80 mg/d) demonstrated statistically significant efficacy and clinically meaningful improvement across a range of symptoms associated with this disorder. Lurasidone treatment was generally safe and well tolerated in this study. Sponsored by Sunovion Pharmaceuticals Inc. ClinicalTrials.gov identifier: {type:clinical-trial,attrs:{text:NCT01911429,term_id:NCT01911429}}NCT01911429

A Simple Correction to Completer Analyses and Improvement on Baseline Observation Carried Forward

Replacing missing data using the baseline observation carried forward (BOCF) technique is known to be fraught with problems. Despite recommendations to the contrary, BOCF and the related last observation carried forward (LOCF) continue to be used in some fields of research. We first show the use of BOCF in testing for a change in a single sample is essentially equivalent to what results from a completer analysis. Next, we derive a simple method based only on summary statistics for adjusting inference from a completer analysis in situations where the estimand of the completer analysis is expected to be close to that of the full analysis set. For those with missing follow-up data, the method assumes a mean change of zero from baseline, with variance equivalent to that estimated from the patients who complete; in other words, it is essentially BOCF analysis with plausible variance inflation. The extension to two samples is considered and a similar adjustment of the completer analysis is presented. The method presented here can be used to reanalyze results from publications that treat missing data inadequately, to adjust for missing data in power calculations and as a quick way of checking the plausibility of an appropriate treatment of missing data. These practical tools are meant to complement, rather than compete with, established methods for dealing with missing data. Numerical simulations are used to elucidate some properties of the proposed method and, finally, we apply the method to data from a recent trial which evaluated weight loss programs.

A dynamic Bayesian network approach to forecast short-term urban rail passenger flows with incomplete data

Abstract We propose an approach to forecast the short-term passenger flows of the urban rail network of Paris. Based on dynamic Bayesian networks, this approach is designed to perform even in case of incomplete data. The structure of the model is built so that the flows are predicted from their spatiotemporal neighbourhood, while the local conditional distributions are described as linear Gaussians. A first application carried out on a single station highlights the need to incorporate information on the transport service. In the presence of missing data, we perform the structural expectation-maximization (EM) algorithm to learn both the structure and the parameters of the model. Short-term forecasting is conducted by inference via the bootstrap filter. Finally, we apply the model to an entire Paris metro line, using on-board counting, ticket validation and transport service data. The overall forecasting results outperform the historical average and last observation carried forward (LOCF) methods. They also evidence the key role of the transport service in the modeling.

Imputation Methods for Longitudinal Data: A Comparative Study

Longitudinal studies play an important role in scientific researches. The defining characteristic of the longitudinal studies is that observations are collected from each subject repeatedly over time, or under different conditions. Missing values are common in longitudinal studies. The presence of missing values is always a fundamental challenge since it produces potential bias, even in well controlled conditions. Three different missing data mechanisms are defined; missing completely at random (MCAR), missing at random (MAR) and missing not at random (MNAR). Several imputation methods have been developed in literature to handle missing values in longitudinal data. The most commonly used imputation methods include complete case analysis (CCA), mean imputation (Mean), last observation carried forward (LOCF), hot deck (HOT), regression imputation (Regress), K-nearest neighbor (KNN), The expectation maximization (EM) algorithm, and multiple imputation (MI). In this article, a comparative study is conducted to investigate the efficiency of these eight imputation methods under different missing data mechanisms. The comparison is conducted through simulation study. It is concluded that the MI method is the most effective method as it has the least standard errors. The EM algorithm has the largest relative bias. The different methods are also compared via real data application.

P.3.d.044 - Effects of aripiprazole once-monthly and paliperidone palmitate in patients with schizophrenia stratified by disease severity: a post-hoc analysis of QUALIFY

Naturalistic effectiveness trials of atypical antipsychotics are needed to provide broader information on efficacy, safety, and tolerability in patients with schizophrenia treated in a community practice setting.In this 26-week, open-label, multicentre study, patients with schizophrenia requiring a switch in antipsychotic medication because current medication was not well tolerated and/or clinical symptoms were not well controlled were randomized to receive aripiprazole or an atypical antipsychotic standard of care (SOC) treatment (i.e., olanzapine, quetiapine, or risperidone based on the investigator's judgment of the optimal treatment for the individual patient and the patient's prior response to antipsychotic medication). The primary objective was to compare the effectiveness of a 26-week treatment of aripiprazole versus SOC, as measured by the Investigator Assessment Questionnaire (IAQ) total score at Week 26 last observation carried forward (LOCF) (primary endpoint), a validated measure that monitors relief or worsening of 10 key symptoms associated with the psychopathology of schizophrenia and side effects of antipsychotic treatment. Secondary objectives were to further assess effectiveness using the Clinical Global Impression – Global Improvement (CGI-I) and Clinical Global Impression – Severity of Illness scale, to assess time to treatment discontinuation, patient preference of medication, quality of life, and the tolerability of aripiprazole compared with SOC.Aripiprazole treatment (n = 268) resulted in significantly better effectiveness than SOC treatment (n = 254; P < 0.001; Week 26 LOCF) as evidenced by the IAQ total score beginning at Week 4 (the first assessment point) and sustained through Week 26. A similar relationship was demonstrated among patients who completed the study (observed cases analysis); aripiprazole was associated with significantly better effectiveness at all time points with a greater differential effect from SOC over time. Patients treated with aripiprazole also demonstrated significantly greater improvements on the CGI-I scale (responder rate, P = 0.009 at Week 26 LOCF), as well as on quality of life (Quality of Life scale total score; P < 0.001 at Week 26). Furthermore, a significantly higher proportion of patients receiving aripiprazole rated their study medication as “much better” on the Preference of Medication Questionnaire (POM) scale than their pre-study medication compared with SOC patients (P < 0.001; Week 26). Time to treatment discontinuation and rates of discontinuation due to adverse events were similar in both treatment groups. The incidence of patients with one or more extrapyramidal symptom (e.g., akathisia, dystonia, parkinsonian events, and residual events) was higher in patients receiving aripiprazole compared with patients treated with SOC (13.5% vs. 5.6%); however, a higher proportion of patients in the SOC-treated group had clinically significant weight gain (21.2% vs. 7.3% for aripiprazole) and potentially clinically relevant elevated fasting levels of total cholesterol, low-density lipoprotein cholesterol, triglycerides, and serum prolactin compared with patients receiving aripiprazole.Aripiprazole is an effective atypical antipsychotic for the treatment of schizophrenia, demonstrating better effectiveness than SOC agents used in this study in patients for whom a switch in antipsychotic medication was warranted.

Pregabalin for pain in fibromyalgia in adults.

Pregabalin for pain in fibromyalgia in adults.

Efficacy and safety of canagliflozin in combination with insulin: a double-blind, randomized, placebo-controlled study in Japanese patients with type 2 diabetes mellitus.

BackgroundCombination therapy with canagliflozin and insulin was investigated in a prescribed substudy of the canagliflozin Cardiovascular Assessment Study (CANVAS); however, it was not evaluated in Japanese patients with type 2 diabetes mellitus (T2DM). Since the usage profile of insulin therapy and pathologic features of Japanese patients differ from those of Caucasian patients, we determined the clinical benefit of such a combination therapy in Japanese patients.MethodsPatients who had inadequate glycemic control despite insulin, diet and exercise therapies were randomized into placebo (n = 70) and canagliflozin 100 mg (n = 76) groups that were administered once daily in addition to their prior insulin therapy in this double-blind, placebo-controlled study. The primary endpoint was the change in glycated hemoglobin (HbA1c) levels from the baseline to week 16.ResultsThere was a statistically significant decrease in HbA1c levels from the baseline in the canagliflozin group (−0.97 ± 0.08 %) compared with the placebo group (0.13 ± 0.08 %) at week 16 [last observation carried forward (LOCF)]. The decrease in HbA1c levels in the canagliflozin group was independent of the insulin regimen (premixed, long-acting and long-acting plus rapid- or short-acting). Compared with the placebo group, canagliflozin significantly decreased fasting plasma glucose levels (−34.1 ± 4.8 vs −1.4 ± 5.0 mg/dL) and body weights (−2.13 ± 0.25 vs 0.24 ± 0.26 %), and significantly increased HDL cholesterol (3.3 ± 1.0 vs −0.5 ± 1.0 mg/dL) and HOMA2- %B (10.15 ± 1.37 vs 0.88 ± 1.42 %). The overall incidence of adverse events was similar between the two groups. The incidence and incidence per subject-year exposure of hypoglycemia (hypoglycemic symptoms and/or decreased blood glucose) were slightly higher in the canagliflozin group (40.0 % and 7.97) than in the placebo group (29.6 % and 4.51). However, hypoglycemic events in both groups were mild in severity and dose-reduction of insulin by <10 % from the baseline following hypoglycemic events decreased the incidence per subject-year exposure in the canagliflozin group. The incidence of hypoglycemia between the groups did not differ according to the insulin regimen.ConclusionCanagliflozin in combination with insulin was effective in improving glycemic control and reducing body weight and well tolerated by Japanese patients with T2DM.Trial Registration ClinicalTrials.gov identifier: NCT02220920Electronic supplementary materialThe online version of this article (doi:10.1186/s12933-016-0407-4) contains supplementary material, which is available to authorized users.

Missing Data in Alcohol Clinical Trials with Binary Outcomes.

Missing data are common in alcohol clinical trials for both continuous and binary end points. Approaches to handle missing data have been explored for continuous outcomes, yet no studies have compared missing data approaches for binary outcomes (e.g., abstinence, no heavy drinking days). This study compares approaches to modeling binary outcomes with missing data in the COMBINE study. We included participants in the COMBINE study who had complete drinking data during treatment and who were assigned to active medication or placebo conditions (N=1,146). Using simulation methods, missing data were introduced under common scenarios with varying sample sizes and amounts of missing data. Logistic regression was used to estimate the effect of naltrexone (vs. placebo) in predicting any drinking and any heavy drinking outcomes at the end of treatment using 4 analytic approaches: complete case analysis (CCA), last observation carried forward (LOCF), the worst case scenario (WCS) of missing equals any drinking or heavy drinking, and multiple imputation (MI). In separate analyses, these approaches were compared when drinking data were manually deleted for those participants who discontinued treatment but continued to provide drinking data. WCS produced the greatest amount of bias in treatment effect estimates. MI usually yielded less biased estimates than WCS and CCA in the simulated data and performed considerably better than LOCF when estimating treatment effects among individuals who discontinued treatment. Missing data can introduce bias in treatment effect estimates in alcohol clinical trials. Researchers should utilize modern missing data methods, including MI, and avoid WCS and CCA when analyzing binary alcohol clinical trial outcomes.

Relationships Between Functional Outcomes and Symptomatic Improvement in Atomoxetine-Treated Adult Patients with Attention-Deficit/Hyperactivity Disorder: Post Hoc Analysis of an Integrated Database.

Atomoxetine treatment is associated with improvements in functional outcomes in patients with attention-deficit/hyperactivity disorder (ADHD), although relationships between improvements in these outcomes and reductions in ADHD symptoms have not been comprehensively investigated in adults. The aim of this study was to assess relationships between functional outcomes and ADHD symptoms (primary objective), and to assess time courses of changes in functional outcomes from baseline to weeks 10 and 24 (secondary objective). We analyzed data pooled from seven Eli Lilly-sponsored placebo-controlled trials of atomoxetine in adults with ADHD that had Conners' Adult ADHD Rating Scales-Investigator Rated: Screening Version (CAARS-Inv:SV) total scores and functional outcome data at baseline and at week 10. Two trials also had these data at week 24. Patients were included in these pooled analyses if they had a CAARS-Inv:SV total score at baseline and at one or more post-baseline visits at weeks 10 or 24, or had post-baseline scores that would allow missing scores at weeks 10 or 24 to be imputed. To address the primary objective, changes in functional outcomes during treatment with atomoxetine versus placebo were assessed using last observation carried forward (LOCF) analysis of covariance (ANCOVA) and mixed-effects model repeated measures (MMRM) analysis, and correlations between score changes in CAARS-Inv:SV total and functional outcomes were assessed using Spearman's rank correlation coefficient (r) at weeks 10 and 24. The secondary objective was addressed using MMRM. At baseline, patients generally had moderately severe or worse ADHD symptoms (based on CAARS-Inv:SV total scores) and impaired functional outcomes (based on Adult ADHD Quality-of-Life [AAQoL], Behavior Rating Inventory of Executive Function-Adult Version [BRIEF-A], Sheehan Disability Scale [SDS], and 36-item Short-Form Health Survey [SF-36] scores). These baseline characteristics were comparable in the atomoxetine and placebo groups. For atomoxetine versus placebo, statistically significant improvements were detected in AAQoL total and subscores at weeks 10 and 24, and in BRIEF-A Self-Report scores at week 10, but not in BRIEF-A Informant Report or SDS scores at week 10 (no BRIEF-A or SDS data were available at week 24), and not in SF-36 at weeks 10 or 24. All functional improvements were gradual. During treatment with atomoxetine, there were moderate correlations between reductions in CAARS-Inv:SV total scores and increases in AAQoL total and subscores at weeks 10 and 24 (r range -0.58 to -0.39; n=394-545), and also with reductions in BRIEF-A Self-Report at week 10 (r=0.49; n=256). With placebo, moderate correlations were also found between reductions in CAARS-Inv:SV total scores and increases in AAQoL total and subscores at weeks 10 and 24 (r range -0.56 to -0.28; n=321-542), and with reductions in BRIEF-A Self-Report at week 10 (r=0.49; n=271). However, correlations between changes in CAARS-Inv:SV and BRIEF-A Informant at week 10 were low for atomoxetine-treated patients (r=0.25; n=65), moderate with placebo (r=0.42; n=72), and there were low/no correlations between changes in CAARS-Inv:SV and functional outcome rating scales that are not specific to ADHD; that is, for atomoxetine-treated patients, SDS total r=0.19 (n=32 at week 10) and SF-36 r range -0.20 to -0.01 (n=51 at week 10, n=183 at week 24). Atomoxetine-treated adult patients experienced improvements in functional outcomes (AAQoL and BRIEF-A Self-Report) that correlated with reductions in ADHD symptoms. Although atomoxetine improved both the ADHD symptoms and functional outcomes, the correlation between symptoms and functional outcomes was low to moderate, suggesting that they measure overlapping but different aspects of the disorder. Hence, clinicians should assess not just ADHD symptoms, but also the functional impairments.

Long-term safety and effectiveness of lurasidone in schizophrenia: a 22-month, open-label extension study.

To evaluate the safety and effectiveness of lurasidone in the long-term treatment of patients with schizophrenia. Patients who completed a 6-week, double-blind (DB), placebo-controlled trial continued in a 22-month, open-label (OL) study during which they received once-daily, flexible-doses of lurasidone, 40-120 mg. Change in the Positive and Negative Syndrome Scale (PANSS) was analyzed using both observed case (OC) and last observation carried forward (LOCF) analyses. Of the 251 patients who entered the OL extension, 51.4% completed 6 months, 36.7% completed 12 months, and 26.7% completed 22 months of OL treatment. Treatment with lurasidone was associated with a mean change from DB baseline, in weight of +0.4 kg at Month 12 (n=99), and +0.8 kg at Month 24 (n=67; OC analyses). Median change from DB baseline to Month 12 and Month 24, respectively, was -1.0 and -9.0 mg/dL for total cholesterol; 0.0 and -1.0 mg/dL for LDL; +1.0 and -11.0 mg/dL for triglycerides; and 0.0 and +0.1/% for HbA1c (OC analyses). The mean PANSS total score was 96.5 at DB baseline and 69.5 at OL baseline. The mean change from DB baseline in the PANSS total score at Month 24 was -43.6 (OC) and -28.4 (LOCF). Thirty-seven patients (14.7%) discontinued due to an adverse event (AE) during OL treatment. Three AEs occurred in ≥10% of patients: schizophrenia (12.4%), akathisia (10.8%), and somnolence (10.8%); and 19.2% reported at least one movement disorder-related AE. Discontinuations due to AEs occurred in 14.8% of patients. In this 22-month, open-label extension study, treatment with lurasidone was associated with minimal effects on weight, glucose, lipids, and prolactin. Patients demonstrated sustained improvement in the PANSS total score for up to 24 months of lurasidone treatment.

Methods for Missing Data Handling in Randomized Clinical Trials With Nonnormal Endpoints With Application to a Phase III Clinical Trial

ABSTRACTIn randomized clinical trials, when the endpoint is the change from baseline at the last scheduled visit, various parametric, semiparametric, and nonparametric methods have been developed to handle the possible missing data due to dropouts. Although the last observation carried forward (LOCF) and the mixed-effects model for repeated measures (MMRM) have been extensively compared and widely used, they may lead to biased results when the required distributional or missing mechanism assumptions are not satisfied. Nonparametric missing data handling methods including the last rank carried forward (LRCF) and mean rank imputation (MRI) relax the underlying distributional assumptions; however, conditions for them to be valid have been investigated to a very limited extent. This article rigorously derives asymptotic properties of the MRI method and proves its validity to test the primary endpoint under certain mild distributional and missing mechanism assumptions. The test-based estimator for the location difference between the treatment and the control groups is also derived when the randomized clinical trial has two arms under a location shift assumption. The investigated methods are applied to an illustrative phase III clinical trial. Simulation studies based on the empirical distributions from the illustrative clinical trial and additional intensive simulation studies, based on various prespecified distributions and missing mechanisms, were conducted to compare the MRI method with selected traditional methods including LOCF, MMRM, and LRCF and they confirmed the better performance of the MRI method in terms of the Type I error rate control and the power under certain mild conditions.

On last observation carried forward and asynchronous longitudinal regression analysis

In many longitudinal studies, the covariates and response are often intermittently observed at irregular, mismatched and subject-specific times. Last observation carried forward (LOCF) is one of the most commonly used methods to deal with such data when covariates and response are observed asynchronously. However, this can lead to considerable bias. In this paper, we propose a weighted LOCF estimation using asynchronous longitudinal data for the generalized linear model. We further generalize this approach to utilize previously observed covariates in addition to the most recent observation. In comparison to earlier methods, the current methods are valid under weaker assumptions on the covariate process and allow informative observation times which may depend on response even conditional on covariates. Extensive simulation studies provide numerical support for the theoretical findings. Data from an HIV study is used to illustrate our methodology.

The Impact of Sparse Follow-up on Marginal Structural Models for Time-to-Event Data.

The impact of risk factors on the amount of time taken to reach an endpoint is a common parameter of interest. Hazard ratios are often estimated using a discrete-time approximation, which works well when the by-interval event rate is low. However, if the intervals are made more frequent than the observation times, missing values will arise. We investigated common analytical approaches, including available-case (AC) analysis, last observation carried forward (LOCF), and multiple imputation (MI), in a setting where time-dependent covariates also act as mediators. We generated complete data to obtain monthly information for all individuals, and from the complete data, we selected "observed" data by assuming that follow-up visits occurred every 6 months. MI proved superior to LOCF and AC analyses when only data on confounding variables were missing; AC analysis also performed well when data for additional variables were missing completely at random. We applied the 3 approaches to data from the Canadian HIV-Hepatitis C Co-infection Cohort Study (2003-2014) to estimate the association of alcohol abuse with liver fibrosis. The AC and LOCF estimates were larger but less precise than those obtained from the analysis that employed MI.

Clinical trial methodology and drug-placebo differences.

As Khan and Brown (1) correctly note, the magnitude of the placebo response in antidepressant trials has increased over the years. But it is not only the placebo response that has increased. So too has the response to antidepressants, a fact that has been widely ignored. In the Walsh et al meta-analysis (2), the correlation between the placebo response and year of publication was r = .45; that between the response to selective serotonin reuptake inhibitors (SSRIs) and year of publication was r = .47, and the difference between the two remained relatively constant. What might account for the finding that the response to both placebo and antidepressants has increased over the years? One thing that it points to is that the drug effect and the placebo effect are probably additive. That is, the response to antidepressants comprises the effect of the drug and the response to placebo, so that when the placebo effect increases, so too does the response to the drug (3). What, then, is responsible for the increase in antidepressant and placebo responses over the years? It cannot be due to decreased baseline severity in more recent trials, because pharmaceutical companies abandoned including mildly and moderately depressed patients in efficacy trials after finding that these patients did not benefit beyond placebo (4). A more likely explanation is that marketing has led to increased public perception that antidepressants are effective, thus enhancing the placebo effect, and because the placebo effect is a component of the drug response, the latter also increased. As Khan and Brown note, “if you lower the risk of exposure to placebo, then the apparent therapeutic effect with the antidepressants and placebo is greater”. Once again, these data suggest additivity. Increasing expectancy of getting a drug rather than a placebo increases the response to the drug and the placebo (5–7). According to Khan and Brown, the last observation carried forward (LOCF) method acts to minimize antidepressant-placebo differences. My colleagues and I made the same assumption when conducting our first meta-analysis of the trial data submitted to the U.S. Food and Drug Administration (FDA) (8). However, the data proved us wrong. LOCF analyses indicated greater drug-placebo differences than did the observed cases method, in which dropouts were excluded from the analyses. Caution is needed when drawing conclusions from data indicating that higher baseline scores are associated with greater improvement in both drug and placebo arms. This is exactly what would be expected based on the statistical artifact of regression toward the mean. It is a very substantial effect that is often ignored. Elsewhere I have shown that when difference scores between two random variables are correlated with the score on one of them, the resulting correlation is about r =. 70 (9). This is the chance standard against which one might judge an association between baseline severity and improvement scores. Finally, a note on the “quick and easy approval” of vilazodone by the FDA is needed. Trovis Pharmaceuticals submitted seven clinical trials to the FDA. The first five showed negative results. However, in one of these, the company noted a non-significant trend toward superiority of vilazodone over placebo on the Montgomery-Asberg Depression Rating Scale (MADRS), but not on the Hamilton Rating Scale for Depression (HAM-D), which had been used in previous antidepressant approvals as the primary outcome for assessing efficacy. Trovis Pharmaceuticals was then allowed to designate the MADRS instead of the HAM-D as the primary outcome measure for what were subsequently considered the two pivotal trials. The difference between vilazodone and placebo on the HAM-D improvement was only 1.69 points on the two “pivotal” trials and, considering all seven trials, it was only 1.01. The FDA approved labeling states that “the efficacy of VIIBRYD was established in two 8-week, randomized, double-blind, placebo-controlled trials”, but makes no mention of the five negative trials. In an internal memo dated May 4, 1998, P. Leber, writing in his capacity as Director of the FDA Division of Neuropharmacological Drug Products, stated his opinion that “labeling that selectively describes positive studies and excludes mention of negative ones can be viewed as being potentially ‘false and misleading’” (10), p. 11). Despite these minor qualifications, the points made by Khan and Brown are well taken. Drug-placebo differences are small in efficacy trials, and most of the response to antidepressants seems due to expectancy.

How Much Missing May Be Too Much in Clinical Trials of Neuropharmacological Drug Products

Missing data due to dropout of patients in randomized clinical trials may be potential sources of bias in the efficacy evaluation of investigational drugs. The biases may be severe when the presence of missing data is substantial. Over the last two decades, extensive research works on dealing with missing data in clinical trials were conducted by both the academicians and pharmaceutical industry statisticians. Majority of the research findings supported the use of Mixed Model Repeated Measure (MMRM) and Multiple Imputation (MI) ANCOVA analyses for efficacy evaluation of study drugs instead of using the conventional last observation carried forward (LOCF) approach to deal with missing at random (MAR) data in drug development research. However, there is a little evidence of exploring the impact of increases in the prevalence of missing data on biases in the statistical inferences obtained from MMRM and MI ANCOVA analyses. In this paper, an attempt is made to explore the impact of increases in the prevalence of missing data on the biases drawn from the above two approaches based on an extensive simulation study. Keywords: ANCOVA, MAR, Missing data, MMRM analysis, Multiple imputation, LOCF.