Handling an uncertain control group event risk in non-inferiority trials: non-inferiority frontiers and the power-stabilising transformation.

BACKGROUND: Non-inferiority trials are increasingly used to evaluate new treatments that are expected to have secondary advantages over standard of care, but similar efficacy on the primary outcome. When designing a non-inferiority trial with a binary primary outcome, the choice of effect measure for the non-inferiority margin (e.g. risk ratio or risk difference) has an important effect on sample size calculations; furthermore, if the control event risk observed is markedly different from that assumed, the trial can quickly lose power or the results become difficult to interpret. METHODS: We propose a new way of designing non-inferiority trials to overcome the issues raised by unexpected control event risks. Our proposal involves using clinical judgement to specify a 'non-inferiority frontier', i.e. a curve defining the most appropriate non-inferiority margin for each possible value of control event risk. Existing trials implicitly use frontiers defined by a fixed risk ratio or a fixed risk difference. We discuss their limitations and propose a fixed arcsine difference frontier, using the power-stabilising transformation for binary outcomes, which may better represent clinical judgement. We propose and compare three ways of designing a trial using this frontier: testing and reporting on the arcsine scale; testing on the arcsine scale but reporting on the risk difference or risk ratio scale; and modifying the margin on the risk difference or risk ratio scale after observing the control event risk according to the power-stabilising frontier. RESULTS: Testing and reporting on the arcsine scale leads to results which are challenging to interpret clinically. For small values of control event risk, testing on the arcsine scale and reporting results on the risk difference scale produces confidence intervals at a higher level than the nominal one or non-inferiority margins that are slightly smaller than those back-calculated from the power-stabilising frontier alone. However, working on the arcsine scale generally requires a larger sample size compared to the risk difference scale. Therefore, working on the risk difference scale, modifying the margin after observing the control event risk, might be preferable, as it requires a smaller sample size. However, this approach tends to slightly inflate type I error rate; a solution is to use a slightly lower significance level for testing, although this modestly reduces power. When working on the risk ratio scale instead, the same approach based on the modification of the margin leads to power levels above the nominal one, maintaining type I error under control. CONCLUSIONS: Our proposed methods of designing non-inferiority trials using power-stabilising non-inferiority frontiers make trial design more resilient to unexpected values of the control event risk, at the only cost of requiring somewhat larger sample sizes when the goal is to report results on the risk difference scale.

A randomized trial comparing concise and standard consent forms in the START trial.

BACKGROUND: Improving the effectiveness and efficiency of research informed consent is a high priority. Some express concern about longer, more complex, written consent forms creating barriers to participant understanding. A recent meta-analysis concluded that randomized comparisons were needed. METHODS: We conducted a cluster-randomized non-inferiority comparison of a standard versus concise consent form within a multinational trial studying the timing of starting antiretroviral therapy in HIV+ adults (START). Interested sites were randomized to standard or concise consent forms for all individuals signing START consent. Participants completed a survey measuring comprehension of study information and satisfaction with the consent process. Site personnel reported usual site consent practices. The primary outcome was comprehension of the purpose of randomization (pre-specified 7.5% non-inferiority margin). RESULTS: 77 sites (2429 participants) were randomly allocated to use standard consent and 77 sites (2000 participants) concise consent, for an evaluable cohort of 4229. Site and participant characteristics were similar for the two groups. The concise consent was non-inferior to the standard consent on comprehension of randomization (80.2% versus 82%, site adjusted difference: 0.75% (95% CI -3.8%, +5.2%)); and the two groups did not differ significantly on total comprehension score, satisfaction, or voluntariness (p>0.1). Certain independent factors, such as education, influenced comprehension and satisfaction but not differences between consent groups. CONCLUSIONS: An easier to read, more concise consent form neither hindered nor improved comprehension of study information nor satisfaction with the consent process among a large number of participants. This supports continued efforts to make consent forms more efficient. TRIAL REGISTRATION: Informed consent substudy was registered as part of START study in clinicaltrials.gov #NCT00867048, and EudraCT # 2008-006439-12.

Cost effectiveness analysis of clinically driven versus routine laboratory monitoring of antiretroviral therapy in Uganda and Zimbabwe.

BACKGROUND: Despite funding constraints for treatment programmes in Africa, the costs and economic consequences of routine laboratory monitoring for efficacy and toxicity of antiretroviral therapy (ART) have rarely been evaluated. METHODS: Cost-effectiveness analysis was conducted in the DART trial (ISRCTN13968779). Adults in Uganda/Zimbabwe starting ART were randomised to clinically-driven monitoring (CDM) or laboratory and clinical monitoring (LCM); individual patient data on healthcare resource utilisation and outcomes were valued with primary economic costs and utilities. Total costs of first/second-line ART, routine 12-weekly CD4 and biochemistry/haematology tests, additional diagnostic investigations, clinic visits, concomitant medications and hospitalisations were considered from the public healthcare sector perspective. A Markov model was used to extrapolate costs and benefits 20 years beyond the trial. RESULTS: 3316 (1660LCM;1656CDM) symptomatic, immunosuppressed ART-naive adults (median (IQR) age 37 (32,42); CD4 86 (31,139) cells/mm(3)) were followed for median 4.9 years. LCM had a mean 0.112 year (41 days) survival benefit at an additional mean cost of $765 [95%CI:685,845], translating into an adjusted incremental cost of $7386 [3277,dominated] per life-year gained and $7793 [4442,39179] per quality-adjusted life year gained. Routine toxicity tests were prominent cost-drivers and had no benefit. With 12-weekly CD4 monitoring from year 2 on ART, low-cost second-line ART, but without toxicity monitoring, CD4 test costs need to fall below $3.78 to become cost-effective (<3xper-capita GDP, following WHO benchmarks). CD4 monitoring at current costs as undertaken in DART was not cost-effective in the long-term. CONCLUSIONS: There is no rationale for routine toxicity monitoring, which did not affect outcomes and was costly. Even though beneficial, there is little justification for routine 12-weekly CD4 monitoring of ART at current test costs in low-income African countries. CD4 monitoring, restricted to the second year on ART onwards, could be cost-effective with lower cost second-line therapy and development of a cheaper, ideally point-of-care, CD4 test.

Parametric randomization-based methods for correcting for treatment changes in the assessment of the causal effect of treatment.

We develop parametric maximum likelihood methods to adjust for treatment changes during follow-up in order to assess the causal effect of treatment in clinical trials with time-to-event outcomes. The accelerated failure time model of Robins and Tsiatis relates each observed event time to the underlying event time that would have been observed if the control treatment had been given throughout the trial. We introduce a bivariate parametric frailty model for time to treatment change and time to trial endpoint. Estimating equations which respect the randomization are constructed and compared to maximum likelihood methods in a simulation study. The Concorde trial of immediate versus deferred zidovudine in HIV infection is used as a motivating example and illustration of the methods.