Dataset Information

Analysis of multicenter clinical trials with very low event rates.

ABSTRACT:

Introduction

In a five-arm randomized clinical trial (RCT) with stratified randomization across 54 sites, we encountered low primary outcome event proportions, resulting in multiple sites with zero events either overall or in one or more study arms. In this paper, we systematically evaluated different statistical methods of accounting for center in settings with low outcome event proportions.

Methods

We conducted a simulation study and a reanalysis of a completed RCT to compare five popular methods of estimating an odds ratio for multicenter trials with stratified randomization by center: (i) no center adjustment, (ii) random intercept model, (iii) Mantel-Haenszel model, (iv) generalized estimating equation (GEE) with an exchangeable correlation structure, and (v) GEE with small sample correction (GEE-small sample correction). We varied the number of total participants (200, 500, 1000, 5000), number of centers (5, 50, 100), control group outcome percentage (2%, 5%, 10%), true odds ratio (1, >?1), intra-class correlation coefficient (ICC) (0.025, 0.075), and distribution of participants across the centers (balanced, skewed).

Results

Mantel-Haenszel methods generally performed poorly in terms of power and bias and led to the exclusion of participants from the analysis because some centers had no events. Failure to account for center in the analysis generally led to lower power and type I error rates than other methods, particularly with ICC?=?0.075. GEE had an inflated type I error rate except in some settings with a large number of centers. GEE-small sample correction maintained the type I error rate at the nominal level but suffered from reduced power and convergence issues in some settings when the number of centers was small. Random intercept models generally performed well in most scenarios, except with a low event rate (i.e., 2% scenario) and small total sample size (n???500), when all methods had issues.

Discussion

Random intercept models generally performed best across most scenarios. GEE-small sample correction performed well when the number of centers was large. We do not recommend the use of Mantel-Haenszel, GEE, or models that do not account for center. When the expected event rate is low, we suggest that the statistical analysis plan specify an alternative method in the case of non-convergence of the primary method.

SUBMITTER: Kim J

PROVIDER: S-EPMC7654615 | biostudies-literature | 2020 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Analysis of multicenter clinical trials with very low event rates.

Kim Jiyu J Troxel Andrea B AB Halpern Scott D SD Volpp Kevin G KG Kahan Brennan C BC Morris Tim P TP Harhay Michael O MO

Trials 20201109 1

<h4>Introduction</h4>In a five-arm randomized clinical trial (RCT) with stratified randomization across 54 sites, we encountered low primary outcome event proportions, resulting in multiple sites with zero events either overall or in one or more study arms. In this paper, we systematically evaluated different statistical methods of accounting for center in settings with low outcome event proportions.<h4>Methods</h4>We conducted a simulation study and a reanalysis of a completed RCT to compare fi ...[more]

PMID: 33168073

Similar Datasets

Project description:IntroductionPatient adherence to therapy in clinical practice is often low, and the difference between efficacy measured in clinical trials and effectiveness in clinical practice is probably a function of discontinuation of therapy because of lack of efficacy or because of unmanageable or intolerable adverse events. Discontinuation is frequently measured in clinical trials but is not usually described in detail in published reports, often because of limitations in the size of publications. By contrast, company clinical trial reports include much more detail.MethodsWe examined company clinical trial reports of trials involving etoricoxib in four musculoskeletal conditions: osteoarthritis, rheumatoid arthritis, chronic low back pain and ankylosing spondylitis. Information was available from 18 randomized trials (10,143 patients) lasting 4 to 12 weeks (one 4 weeks, three 6 weeks, one 8 weeks and seven 12 weeks) and from three trials with a mean duration of about 80 weeks (34,695 patients). These clinical trial reports contain over 73,000 pages of information.ResultsOver 12 weeks, lack of efficacy and adverse event discontinuations were similar between osteoarthritis, rheumatoid arthritis and back pain, with lack of efficacy discontinuation rates some three times higher than for adverse events. All-cause and lack of efficacy discontinuations were lower with etoricoxib (all doses combined) and traditional nonselective nonsteroidal anti-inflammatory drugs (NSAIDs) than with placebo, although NSAIDs produced higher rates of clinical adverse events and gastrointestinal discontinuations than did placebo. Etoricoxib had fewer discontinuations than NSAIDs for lack of efficacy, clinical adverse events, and laboratory and gastrointestinal adverse events, but with more discontinuations because of hypertension and oedema. Comparison with two similar meta-analyses of other cyclo-oxygenase-2 selective inhibitors (more than 80,000 patients in total) revealed consistency between analyses.ConclusionExamining discontinuation data from clinical trials, even when the numbers of patients are very large, does not necessarily predict what will happen in the real world, where clinical effectiveness may differ from clinical efficacy assessed in trials. Data from these analyses appears to agree with findings from real world practice.

Project description:BackgroundRepresentativeness of antihypertensive drug trials is uncertain, as many trials recruit few or no older people. Some trials specifically recruit older participants to address this. Here, we assess the representativeness of trials focusing on older people by comparing the rates of serious adverse events in these trials with the rates in trials of a general adult population (ie, standard trials), and comparing these findings to the rate of hospitalisations and deaths in people with hypertension starting a similar treatment in routine clinical practice.MethodsFor this observational study, we identified randomised controlled trials (phase 2/3, 3, or 4) of renin-angiotensin-aldosterone system (RAAS) drugs for hypertension registered from 1999 onwards with ClinicalTrials.gov. Serious adverse events are routinely included in trial reports and are predominantly accounted for by all-cause hospitalisations and deaths. We compared serious adverse event rates in older-people trials (minimum inclusion age ≥60 years) and standard trials (minimum inclusion age <60 years) using Poisson regression models adjusted for trial characteristics (drug type, comparison type, phase, and outcome type). We identified a community cohort of 56 036 adults with hypertension commencing similar drugs to obtain an expected rate of emergency or urgent hospitalisations or deaths, and compared this rate to observed serious adverse event rates in each trial, adjusted for age and sex. For standard trials and for older-people trials, we calculated the standardised ratio of the expected to the observed rate of serious adverse events using Poisson regression models.FindingsWe included 110 trials, of which 11 (10%) were older-people trials and 99 (90%) were standard trials. Older-people trials had a higher rate of serious adverse events than did standard trials (median events per person per year 0·18 [IQR 0·12-0·29] vs 0·11 [0·08-0·18]; adjusted incidence rate ratio 1·76 [95% CI 1·01-3·03]). The hospitalisation and death rate in the community for those taking RAAS antihypertensives was much greater than the rate of serious adverse events reported in standard trials (standardised ratio [SR] 4·23, 95% CI 3·51-5·09) and older-people trials (4·76, 2·89-7·86), adjusting for age and sex. The magnitude of risk increase for serious adverse events in community patients taking RAAS did not differ when comparing older-people and standard trials (ratio of SRs 1·13, 95% CI 0·66-1·92).InterpretationTrials report substantially fewer serious adverse events than expected from rates of hospitalisations and deaths among similar-aged people receiving equivalent treatments in the community. Serious adverse event rates might be a useful metric to assess trial representativeness. Clinicians should be cautious when applying trial recommendations to older people, even when trials focus on older participants.FundingWellcome Trust, Medical Research Council.

Dataset Information

Analysis of multicenter clinical trials with very low event rates.

Introduction

Methods

Results

Discussion

Publications

Analysis of multicenter clinical trials with very low event rates.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets