Methods These methods are extended to include multi-stage adaptive designs, with a solution given for a two-stage design. The methods are applied to two examples.

Results As demonstrated by the two examples, substantial increases in the expected net gain (ENG) can be realized by using multi-stage adaptive designs based on expected value of information methods. In addition, the expected sample size and total cost may be reduced.

Limitations Exact solutions have been provided for the two-stage design. Solutions for higher-order designs may prove to be prohibitively complex and approximate solutions may be required.

Conclusions The use of multi-stage adaptive designs for randomized clinical trials based on expected value of sample information methods leads to substantial gains in the ENG and reductions in the expected sample size and total cost. Clinical Trials 2008; 5: 289-300. http://ctj.sagepub.com

Purpose We present a graphical method to summarize a network of randomized comparisons and to examine the incoherence of the network, without making any distributional assumptions.

Methods At each treatment-pair level, the inverse variance method is used to pool results from multiple studies. We consider the magnitude of pairwise treatment contrasts as a measure of pairwise dissimilarity. We summarize a network of randomized comparisons as a dissimilarity matrix, and then apply weighted multidimensional scaling to the dissimilarity matrix. The weights are chosen according to the inverse variance method. We show that, with this weighting scheme, 1D multidimensional scaling configuration is closely related to a fixed effect model. Therefore, our interest is to explore a departure from 1D constraint.

Results Two-dimensional multidimensional scaling configuration is useful to explore the incoherence of the network. Our method is illustrated with two published datasets.

Limitations The weighting scheme in our multidimensional scaling setting is chosen to be optimal for independent treatment pairs. Pairwise differences within a multi-arm trial are correlated to one another and intrinsically coherent. Thus our weighting scheme may not apply to data with large numbers of multi-arm trials.

Conclusions Multidimensional scaling provides a useful tool for investigators to visualize the network of randomized comparisons and to assess incoherence of the network. Clinical Trials 2008; 5: 301—307. http://ctj.sagepub.com

Purpose Describe the steps used to independently re-evaluate this trial.

Methods NIH appointed an independent multidisciplinary committee that gained access to the original data. We undertook analyses of t-PA efficacy accounting for this imbalance, as well as analyses to identify subgroups that experienced additional harm or benefit from t-PA. Analyses of time from stroke onset to treatment (OTT), blood pressure, and intracerebral hemorrhage are given as illustrations.

Results Despite subgroup imbalances in baseline stroke severity, when t-PA was administered to acute ischemic stroke patients according to study protocol, there was a statistically significant and clinically important benefit of t-PA treatment resulting in a higher likelihood of having a favorable clinical outcome at 3 months. Moreover, we were unable to identify subgroups of patients between which t-PA treatment effect differed, albeit these analyses had low power. These data failed to support the NINDS investigators' conclusion that effect of t-PA therapy diminished with increasing values of OTT within the protocol-specified 3 h time limit. In addition, the blood pressure measurements were highly variable and inconsistently determined so as to be too unreliable for inclusion in analysis.

Conclusion With new NIH requirements for data-sharing, the frequency of re-analysis of clinical trial data may increase substantially. This re-evaluation provides a blueprint for future re-evaluations of other trials. These best practices include re-analysis of the study data, after suitable replication, by an independent multidisciplinary committee, including a skilled statistical programmer analyst. Primary investigators should address significant errors determined in such re-analyses. Clinical Trials 2008; 5: 308—315. http://ctj.sagepub.com

Purpose We present the randomization scheme used in the ZAMSTAR trial of tuberculosis control interventions in Southern Africa.

Methods We used stratification and restriction to randomize 24 clusters (16 Zambian, 8 South African) into four intervention groups in a 2 x 2 factorial design. Stratification was by country and tuberculous infection prevalence and restriction by tuberculous infection prevalence, HIV prevalence, urban/rural, social context, and geographical location. Balance was defined in terms of covariate-specific tolerance thresholds for the measure of imbalance. For binary (0/1) covariates we defined imbalance = max(S_i) - min(S_i), where, S_i was the number of 1s in group i = 1,2,3,4. For continuous covariates we defined imbalance = (max(M_i) - min(M_i))/ min(M_i ), where, M _i was the average in group i = 1,2,3,4.

We used simulation to estimate the restriction factor (proportion of unacceptable allocations) both for individual covariates and overall. Simulation was also used to investigate the validity of the restricted randomization design, with the use of the validity matrix, by monitoring the probability that any given pair of clusters is allocated to the same intervention group.

Results There were 3 657 930 400 possible ways of allocating the 24 clusters to the four groups after stratification. With a combined restriction factor of 0.998 this still left 7 million acceptable allocations. The final allocation was selected at a public ceremony from a randomly-generated list of acceptable allocations. The design of the allocation process was observed to be valid.

Limitations The restricted randomization scheme significantly decreased the total number of available allocations of clusters into intervention groups.

Conclusion Our restricted randomization was successful in that it achieved good balance while preserving the impartiality and validity of the trial. Clinical Trials 2008; 5: 316—327. http://ctj.sagepub.com

Purpose To identify predictors of attrition during two early phases, i.e., from consent to screening (Phase-1), and from screening to intake interview (Phase-2) in two clinical trials.

Methods Cancer patients undergoing chemotherapy were asked to enroll in one of two clinical trials. In each trial the benefits of a cognitive behavioral intervention were compared with a psycho-educational intervention to assist patients to manage cancer and treatment-related symptoms. Following consent patients were screened for their symptoms' severity to determine their eligibility.

Results Of the 885 consenters 785 completed screening and of the 782 eligible for participation, 713 completed intake interview. In the first phase, longer delays between consent and first contact attempt, lower levels of patient education, minority race, and prolonged duration of screening increased the likelihood of dropping out with a significantly stronger effect on minorities than white patients. In the second phase, low education, being a minority, longer screening delays, and impact of symptom severity on enjoyment of life significantly increased probability of attrition.

Limitations Participant reported causes of attrition were not modeled; however, exclusion of patients who died during the time period of this research meant that most patients leaving the study made a conscious decision to do so.

Conclusions To assure preservation of external validity, the time between consent and randomization into the arms of a trial must be held to a minimum. Delays between contacts and run in time, that may include screening patients to assure they will benefit from a trial, must be balanced against rates of attrition. Compressing intervals between contacts is particularly important to retain minorities. Clinical Trials 2008; 5: 328—335. http://ctjsagepub.com

Purpose To describe factors associated with successful recruitment, and identify predictors of continued retention of pregnant women attending a community health center into a randomized controlled clinical trial to prevent early childhood caries.

Methods The Mothers and Youth Access (MAYA) Trial recruited women in the second trimester of pregnancy. At baseline, consenting women completed an oral health questionnaire and received a dental exam and oral health counseling. Four months postpartum, women returned with their babies for randomization with follow up at 9-, 12-, 18-, 24-, 30-, and 36-month postpartum visits. To assess predictors of retention, data about respondents' demographics, and oral health-related knowledge, attitudes, and behaviors were obtained by questionnaire and analyzed by logistic and discrete time-to-event regression analyses.

Results Of 556 predominantly Mexican-American women recruited at baseline, 195 (35%) were excluded after baseline for not meeting inclusion criteria; 361 (65%) continued to randomization. Factors such as race/ethnicity, annual household income, household composition, oral health-related knowledge and behaviors significantly related to retention until randomization. In multivariable models, women reporting a higher annual household income were less likely to be lost to attrition before randomization (odds ratio = 0.73, 95% confidence interval (CI) 0.60—0.89); while Mexican/Mexican-American women were less likely to be lost beyond randomization (hazard ratio = 0.53, 95% CI 0.26—1.08).

Limitations Factors not measured at baseline may have been important in predicting attrition. The MAYA Trial is expected to finish by November 2008; therefore, complete results for total retention may differ from those reported here.

Conclusions Recruitment and retention efforts for pregnant Hispanic women should place heavy emphasis on culture as ethnicity remained the only borderline significant predictor in postrandomization retention. Clinical Trials 2008; 5: 336—346. http://ctj.sagepub.com

Purpose The Hip Impact Protection Project (HIP PRO) was designed to investigate the efficacy and safety of hip protectors in an elderly nursing home population. This paper describes the innovative clustered matched-pair research design used in HIP PRO to overcome the inherent limitations of clustered randomization.

Methods Three clinical centers recruited 37 nursing homes to participate in HIP PRO. They were randomized so that the participating residents in that home received hip protectors for either the right or left hip. Informed consent was obtained from either the resident or the resident's responsible party. The target sample size was 580 residents with replacement if they dropped out, had a hip fracture, or died. One of the advantages of the HIP PRO study design was that each resident was his/her own case and control, eliminating imbalances, and there was no confusion over which residents wore pads (or on which hip).

Limitations Generalizability of the findings may be limited. Adherence was higher in this study than in other studies because of: (1) the use of a run-in period, (2) staff incentives, and (3) the frequency of adherence assessments. The use of a single pad is not analogous to pad use in the real world and may have caused unanticipated changes in behavior. Fall assessment was not feasible, limiting the ability to analyze fractures as a function of falls. Finally, hip protector designs continue to evolve so that the results generated using this pad may not be applicable to other pad designs. However, information about factors related to adherence will be useful for future studies.

Conclusions The clustered matched-pair study design avoided the major problem with previous cluster-randomized investigations of this question — unbalanced risk factors between the experimental group and the control group. Because each resident served as his/her own control, the effects of unbalanced risk factors on treatment effect were virtually eliminated. In addition, the use of frequent adherence assessments allowed us to study the effect of various demographic and environmental factors on adherence, which was vital for the assessment of efficacy. Clinical Trials 2008; 5: 347—355. http://ctj.sagepub.com

Purpose To describe ethical, methodological and legal problems of RCTs on coercive interventions and to suggest possible solutions.

Methods Literature research on possible study designs, ethical considerations and legal regulations was conducted in PubMed.

Results Corresponding to the procedures in emergency medicine informed consent can be obtained after the intervention when the patients are capable again. Informed consent refers only to participation in an interview and utilization of data. Randomization can be ethically approved, if exclusion criteria for randomization are defined. A comprehensive cohort study seems to be the most practicable study design. As primary outcome variable an assessment of subjective experiences of the patients' restrictions to human rights. Clinical Trials 2008; 5: 356—363. http://ctjsagepub.com

Methods: We propose an outcome-based adaptive randomization trial design for patients with advanced stage non-small cell lung cancer. All patients have baseline biopsy samples taken for biomarker assessment prior to randomization to treatments. The primary endpoint of this study is the disease control rate at 8 weeks after randomization. The Bayesian probit model is used to characterize the disease control rate. Patients are adaptively randomized to one of four treatments with the randomization rate based on the updated disease control rate from the accumulated data in the trial. For each biomarker profile, high-performing treatments have higher randomization rates, and vice versa. An early stopping rule is implemented to suspend low-performing treatments from randomization.

Results: Based on extensive simulation studies, with a total of 200 evaluable patients, our trial has desirable operating characteristics to: (1) identify effective agents with a high probability; (2) suspend ineffective agents; and (3) treat more patients with effective agents that correspond to their biomarker profiles. Our trial design continues to update and refine the estimates as the trial progresses.

Limitations: This biomarker-based trial requires biopsible tumors and a two-week turn around time for biomarker profiling before randomization. Additionally, in order to learn from the interim data and adjust the randomization rate accordingly, the outcome-based adaptive randomization design is applicable only for trials when the endpoint can be assessed in a relative short period of time.

Conclusion: Bayesian adaptive randomization trial design is a smart, novel, and ethical design. In conjunction with an early stopping rule, it can be used to efficiently identify effective agents, eliminate ineffective ones, and match effective treatments with patients' biomarker profiles. The proposed design is suitable for the development of targeted therapies and provides a rational design for personalized medicine. Clinical Trials 2008; 5: 181—193. http://ctj.sagepub.com

Purpose: In this article, we focus our attention on the trial-level surrogacy and propose alternative procedures to evaluate such surrogacy measure, which do not pre-specify the type of association. A promising correction based on cross-validation is investigated. As well as the construction of confidence intervals for this measure.

Methods: In order to avoid making assumption about the type of relationship between the treatment effects and its distribution, a collection of alternative methods, based on regression trees, bagging, random forests, and support vector machines, combined with bootstrap-based confidence interval and, should one wish, in conjunction with a cross-validation based correction, will be proposed and applied. We apply the various strategies to data from three clinical studies: in opthalmology, in advanced colorectal cancer, and in schizophrenia.

Results: The results obtained for the three case studies are compared; they indicate that using random forest or bagging models produces larger estimated values for the surrogacy measure, which are in general stabler and the confidence interval narrower than linear regression and support vector regression. For the advanced colorectal cancer studies, we even found the trial-level surrogacy is considerably different from what has been reported.

Limitations: In general the alternative methods are more computationally demanding, and specially the calculation of the confidence intervals, require more computational time that the delta-method counterpart.

Conclusions: First, more flexible modeling techniques can be used, allowing for other type of association. Second, when no cross-validation-based correction is applied, overly optimistic trial-level surrogacy estimates will be found, thus cross-validation is highly recommendable. Third, the use of the delta method to calculate confidence intervals is not recommendable since it makes assumptions valid only in very large samples. It may also produce range-violating limits. We therefore recommend alternatives: bootstrap methods in general. Also, the information-theoretic approach produces comparable results with the bagging and random forest approaches, when cross-validation correction is applied. It is also important to observe that, even for the case in which the linear model might be a good option too, bagging methods perform well too, and their confidence intervals were more narrow. Clinical Trials 2008; 5: 194—208. http://ctj.sagepub.com

Purpose: The performance of the parametric and nonparametric methods is compared to that of using a binomial endpoint based on dichotomizing the event time at a clinically relevant landmark. As sufficient details on the various methods and related designs for phase II clinical design with survival endpoints are provided, this article should also serve as a comparative reference on these three methods for designing phase II studies in cancer with time-to-event endpoints.

Methods: The statistical performance, by virtue of considering the type I and II error rates, of the three methods is compared by carrying out a comprehensive simulation study.

Results: The parametric method may fail to control the type I error rate if the underlying survival distribution is not Exponential, while the nonparametric method may fail to control the type I error rate as the sample sizes for phase II studies are typically small. Both of these methods may be sensitive to the distribution of the censoring variable.

Limitations: The results provided in this article are mostly discussed in the framework of specific examples and by using specific implementations of the tests. As such the results may not be universally generalizable. The recommended method has some drawbacks if patients are censored before the landmark of interest.

Conclusion: A method that should be considered for the purpose of the statistical design and decision rule for phase II studies in cancer is the employment of a binomial endpoint based on dichotomizing the event time at a clinically relevant landmark. Clinical Trials 2008; 5: 209—221. http://ctj.sagepub.com

Purpose: To review and develop imputation methods for missing outcome data in meta-analysis of clinical trials with binary outcomes.

Methods: We review some common strategies, such as simple imputation of positive or negative outcomes, and develop a general approach involving `informative missingness odds ratios' (IMORs). We describe several choices for weighting studies in the meta-analysis, and illustrate methods using a meta-analysis of trials of haloperidol for schizophrenia.

Results: IMORs describe the relationship between the unknown risk among missing participants and the known risk among observed participants. They are allowed to differ between treatment groups and across trials. Application of IMORs and other methods to the haloperidol trials reveals the overall conclusion to be robust to different assumptions about the missing data.

Limitations: The methods are based on summary data from each trial (number of observed positive outcomes, number of observed negative outcomes and number of missing outcomes) for each intervention group. This limits the options for analysis, and greater flexibility would be available with individual participant data.

Conclusions: We propose that available reasons for missingness be used to determine appropriate IMORs. We also recommend a strategy for undertaking sensitivity analyses, in which the IMORs are varied over plausible ranges. Clinical Trials 2008; 5: 225—239. http://ctj.sagepub.com

Study Design and Setting: A total of 161 808 postmenopausal women aged 50—79 were enrolled at 40 clinical centers in the U.S. in 1993—1998 and followed prospectively. At baseline, prevalent medication treated diabetes was defined as a self-report of physician diagnosis and treatment with insulin or oral antidiabetic drugs. During followup, incident treated diabetes was defined as a self-report of a new physician diagnosis of diabetes treated with insulin or oral drugs. Diabetes self-reports were compared with medication inventories and fasting glucose levels at baseline and during follow-up.

Results: At baseline, self-reported treated diabetes was concordant with the medication inventory in 79% of clinical trial, and 77% of observational study participants. Self-reported incident treated diabetes was concordant with the medication inventory in 78% between baseline and Year 1 in the clinical trials, in 62% between Year 1 and Year 3 in the clinical trials, and in 72% between baseline and Year 3 in the observational study. Over similar periods, 99.9% of those who did not report treated diabetes had no oral antidiabetic drugs or insulin in the medication inventory. At baseline, about 3% not reporting diabetes had fasting glucose >126 mg/dl, and 88% of these subjects subsequently reported treated diabetes during 6.9 years of follow-up.

Limitations: Incident self-reported diabetes treated by lifestyle alone was not determined in WHI. Medication inventories may have been incomplete and fasting glucose may have been lowered by treatment; therefore, concordance with self-reported treatment or fasting glucose ≥ 126 may have been underestimated.

Conclusion: In the WHI, self-reported prevalent and incident diabetes was consistent with medication inventories, and a high proportion of those with undiagnosed diabetes subsequently reported diabetes treatment. Self-reports of `treated diabetes' are sufficiently accurate to allow use in epidemiologic studies. Clinical Trials 2008; 5: 240—247. http://ctj.sagepub.com

Purpose: The purpose of this article is to demonstrate that a Cox proportional hazard model with a hazard ratio parameter is equivalent to a Cox proportional hazard model with a parameter equal to the probability that a patient given one treatment will have an event earlier than if the same patient were given a different treatment. This probability will subsequently be referred to as . Clinically interesting differences between the treatment arms are easier for researchers to quantify in terms of in situations where they have a difficult time with the hazard ratio, allowing better communication between the statistician and the researcher.

Methods: The problem and its solution are demonstrated mathematically. The utility of the Cox proportional hazard model in terms of is illustrated through a Lymphoma clinical trial example.

Results: The Cox proportional hazard model with parameter is shown to be equivalent to the Cox proportional hazard model with a hazard ratio parameter . A table of typical hazard ratios is presented with their equivalent values. In the appendix the mathematical derivations are developed and an unbiased estimate of is provided using Gehan's [2] generalization of the Wilcoxon statistic.

Limitations: The equivalence of the Cox proportional hazard model in terms of the probability and the hazard ratio is established only for continuous failure times with a single binary covariate. Conditions under which approximate equivalence holds with multiple covariates are discussed in the Appendix.

Conclusions: The probability provides a natural parameterization for the Cox proportional hazard model, affords a tool to conceptualize treatment differences, and provides a method to improve communication between statisticians and researchers. Clinical Trials 2008; 5: 248—252. http://ctj.sagepub.com

Purpose: This report describes how those challenges were met in the study ``Depression: The Search for Treatment Relevant Phenotypes'', being carried out at the University of Pittsburgh and the University of Pisa, Italy.

Methods: Implementing the study required the investigators to address methodological and practical challenges related to the different requirements of the two Institutional Review Boards (IRBs), psychotherapy training, independent evaluator training, patient recruitment, development of common tools for data entry, quality control and generation of weekly reports of patient progress as well as establishing a similar clinical and research framework in two countries with substantially different health care systems.

Results: By having bilingual investigators and staff members who spent time at one another's sites, making use of frequent conference-call staff meetings and being flexible within the bounds of the sometimes contradictory requirements of the IRBs, the investigators were able to meet the human subjects protection requirements of both institutions, surmount language barriers to consistent therapist and evaluator training and develop common tools for study management. As a result, recruitment goals were met at both sites and retention rates were high. One instance of inconsistent implementation of the protocol was corrected within the first year.

Limitations: This study was conducted in two Western cultures by researchers with long-standing collaboration. Our findings may not be generalizable to other countries or research settings.

Conclusions: The implementation of a cross-national protocol and the adoption and maintenance of common procedures is possible when investigators are aware of the challenges this may present and are proactive in trying to address them. Clinical Trials 2008; 5: 253—261. http://ctj.sagepub.com

Purpose:/Methods We recently completed an NIH-funded phase II trial that utilized a presurgical model in testing the comparative effects of flaxseed supplementation and/or dietary fat restriction on the biology and biomarkers associated with prostatic carcinoma. Herein, we report the rationale for our original design, discuss modifications in strategy, and relay experiences in implementing this trial related to the following topics: (1) subject accrual; (2) subject retention; (3) intervention delivery; and (4) retrieval and completion rates regarding the collection of paraffin-embedded and fresh frozen prostate tissue, blood, urine, ejaculate, anthropometric measures and survey data.

Results: This trial achieved its accrual target, i.e., a racially-representative (70% white, 30% minority) sample of 161 participants, low rates of attrition (7%); and collection rates that exceeded 90% for almost all biospecimens and survey data. While the experience gained from pilot studies was invaluable in designing this trial, the complexity introduced by the collection of several biospecimens, inclusion of a team of pathologists (to provide validated readings), and shifts in practice patterns related to prostatectomy, made it necessary to revise our protocol; lessons from our experiences are offered within this article.

Conclusions: While our experience specifically relates to the implementation of a presurgical model-based trial in prostate cancer aimed at testing flaxseed-supplemented and fat-restricted diets, many of the lessons learned have broad application to trials that utilize a presurgical model or dietary modification within various cancer populations. Clinical Trials 2008; 5: 262—272. http://ctj.sagepub.com

Purpose: To develop a randomization scheme for an HIV prevention trial of a community-level intervention that would be credible to the communities involved while maintaining the scientific integrity of the intervention trial at a rural site in Zimbabwe.

Methods: Project staff developed strong partnerships with community stakeholders and embedded randomization into the trial's community preparedness processes. Local idioms were used to explain the concept, purpose, and mechanics of randomization. Actual allocation of communities to intervention or control status took place at a public lottery conducted by local chiefs.

Results: The Project obtained the endorsement of its randomization of eight rural communities by local political stakeholders and community members.

Limitations: This case study may not generalize to other settings.

Conclusions: By developing strong community partnerships, and communicating randomization through local idioms, community based intervention trials conducted in resource-poor environments can successfully mitigate risks inherent in randomizing communities to control status. Clinical Trials 2008; 5: 273—276. http://ctj.sagepub.com

Purpose Our goal is to develop an efficient and flexible design that possesses desirable statistical properties.

Methods A flexible design based on Bayesian predictive probability and the minimax criterion is constructed. A three-dimensional search algorithm is implemented to determine the design parameters.

Results The new design controls type I and type II error rates, and allows continuous monitoring of the trial outcome. Consequently, under the null hypothesis when the experimental treatment is not efficacious, the design is more efficient in stopping the trial earlier, which results in a smaller expected sample size. Exact computation and simulation studies demonstrate that the predictive probability design possesses good operating characteristics.

Limitations The predictive probability design is more computationally intensive than two- or three-stage designs. Similar to all designs with early stopping due to futility, the resulting estimate of treatment efficacy may be biased.

Conclusions The predictive probability design is efficient and remains robust in controlling type I and type II error rates when the trial conduct deviates from the original design. It is more adaptable than traditional multi-stage designs in evaluating the study outcome, hence, it is easier to implement. S-PLUS/R programs are provided to assist the study design. Clinical Trials 2008; 5: 93—106. http://ctj.sagepub.com

Purpose To illustrate how treatment- and profile-specific survival estimates are obtained from the Cox method, and can be displayed in a compact form.

Methods We derive treatment- and profile-specific survival probabilities from the estimated survival function for the `reference' profile. Data from the Systolic Hypertension in the Elderly Program study serve as an illustration.

Results Two different formats, tabular and nomogram-based, allow the entire set of estimated treatment- and profile-specific survival probabilities to be reported.

Limitations Estimates are limited to the profiles within the covariate-space spanned by the trial, and depend on the correctness of the model.

Conclusion Treatment- and profile-specific survival estimates are practice-relevant, almost never reported, estimable from the Cox model, and easy to report in a compact form. Clinical Trials 2008; 5: 107—115. http://ctj.sagepub.com

Purpose To provide a cumulative summary of the clinical trial evidence on rosiglitazone along with a sensitivity analysis of different methods to estimate the combined effect. Methods A previous meta-analysis (N Engl J Med 2007; 356: 2457—2471) was updated to include event rates of myocardial infarction and death due to cardiovascular causes from the recent reports of the RECORD, DREAM and ADOPT trials. Odds ratios (OR) with their confidence intervals were calculated for all outcomes using the Mantel—Haenszel method with Robins—Breslow—Greenland variance estimation and a fixed effects model. Sensitivity analysis was performed, using different methods for estimating the combined effect and using different continuity corrections for studies with zero events in one or both arms.

Results Rosiglitazone was associated with an increased risk of myocardial infarction (OR, 1.29; CI: 1.01—1.66; p = 0.05) but not death due to cardiovascular causes (OR, 1.12; CI: 0.80—1.55; p = 0.58). Pooled analysis of the ADOPT, RECORD, and DREAM trials did not reach statistical significance for either myocardial infarction (OR, 1.29; CI: 0.95—1.74; p = 0.12) or death due to cardiovascular causes (OR, 0.90; CI: 0.61—1.33; p = 0.67). Based on these three trials, rosiglitazone was associated with a clear increase in the risk of heart failure (OR, 2.17; CI: 1.49—3.17; p<0.0001). Despite minor discrepancies, different calculation methods demonstrated an increased risk of myocardial infarction for rosiglitazone treated patients. There was no evidence of an association between rosiglitazone and death due to cardiovascular causes regardless of the calculation method used. The increased risk of heart failure conferred by rosiglitazone treatment was consistently demonstrated across different calculation methods.

Limitations Trials with short-term follow-up and trials not specifically designed to evaluate cardiovascular outcomes were included in this meta-analysis and patient-level data where not available.

Conclusions Rosiglitazone appears to be associated with an increased risk of myocardial infarction and heart failure, but not death due to cardiovascular causes. When a meta-analysis of rare events is contemplated, a thorough sensitivity analysis using different methods to combine studies and an evaluation of different continuity corrections should be undertaken. When possible, an individual patient data meta-analysis should be performed, allowing time-to-event analysis and the identification of patient subgroups at an increased risk of adverse outcomes. Clinical Trials 2008; 5: 116—120. http://ctj.sagepub.com

Purpose We present statistical issues related to the estimation of individual component effects in trials of geriatric conditions using factorial and standardly-tailored designs. The medical community is interested in second stage translational research involving the transfer of results from a randomized clinical trial to a community setting. Before such research is undertaken, main effects and synergistic and or antagonistic interactions between them should be identified. Knowledge of the relative strength and direction of the effects of the individual components and their interactions facilitates the successful transfer of clinically significant findings and may potentially reduce the number of interventional components needed. Therefore the current inability of the standardly-tailored design to provide unbiased estimates of individual interventional components is a serious limitation in their applicability to second stage translational research.

Methods We discuss estimation of individual component effects from the family of factorial designs and this limitation for standardly-tailored designs. We use the phrase `factorial designs' to describe full-factorial designs and their derivatives including the fractional factorial, partial factorial, incomplete factorial and modified reciprocal designs. We suggest two potential directions for designing multicomponent interventions to facilitate unbiased estimates of individual interventional components.

Results Full factorial designs and their variants are the most common multicomponent trial design described in the literature and differ meaningfully from standardly-tailored designs. Factorial and standardly-tailored designs result in similar estimates of net effect with different levels of precision. Unbiased estimation of individual component effects from a standardly-tailored design will require new methodology.

Limitations Although clinically relevant in geriatrics, previous applications of standardly-tailored designs have not provided unbiased estimates of the effects of individual interventional components.

Discussion Future directions to estimate individual component effects from standardly-tailored designs include applying D-optimal designs and creating independent linear combinations of risk factors analogous to factor analysis.

Conclusion Methods are needed to extract unbiased estimates of the effects of individual interventional components from standardly-tailored designs. Clinical Trials 2008; 5: 121—130. http://ctj.sagepub.com

Purpose The purpose of this survey was to describe the experience with and attitudes of SRG members toward the inclusion guidelines and to identify characteristics of respondents that predict their attitudes towards the policy. Methods A survey was sent to 746 SRG members. 425 SRG members responded and univariate and bivariate statistical analysis conducted.

Results The results of the survey identify one clear measure of success regarding the implementation of the NIH guidelines; SRG members indicate the guidelines are in part responsible for their attention to the inclusion of women, minorities and children in clinical research. In addition, SRG members believe that gender and race are important factors when assessing the diversity of study samples and that the current NIH guidelines are adequate for encouraging their inclusion. As a proxy measure of success, SRG members believe that PIs responsible for protocols reviewed by their study group are generally compliant with the inclusion guidelines. Limitation At least one potential limitation of this study is that while an effort was made to assure confidentiality, because the project was funded by the NIH, respondents may have been less critical of the guidelines than they would have been if the study was funded by non-NIH funds.

Conclusion Future research ought to explore whether IRB members and NIH program officers find PIs to be compliant as their projects get underway. In addition, more research ought to be conducted to assess the downstream effects of this important social policy. Clinical Trials 2008; 5: 140—146. http://ctj.sagepub.com

Purpose To understand the evolution of CABs and community partnerships at international research sites conducting HIV prevention trials.

Methods Three research sites of the HIV Prevention Trials Network (HPTN) were selected to include geographical representation and diverse populations at risk for HIV/AIDS — in Lima, Peru; Chitungwiza, Zimbabwe; and Chiang Mai, Thailand. Data collection included review of secondary data, including academic publications and site-specific progress reports; observations at the research sites; face-to-face interviews with CAB members, research staff, and other key informants; and focus groups with study participants. Rapid assessment techniques were used for data analysis.

Results Two of the three CABs developed new strategies for community representation in response to new studies. All three CABs expanded their original function and became advocates for broader community interests beyond HIV prevention. The participation and input of community representatives, in response to critical incidents that occurred at the sites over the past five years, helped to solidify partnerships between researchers and communities.

Limitations Rapid Assessment is an exploratory methodology designed to provide an understanding of a situation based on the integration of multiple data sources, collected within a short period of time, without a formal examination of transcribed and coded data. Case studies, as a method, are meant to draw out what can be learned from a single case but are not, in the scientific sense, generalizable.

Conclusions In developing countries, CABs can be dynamic entities that enhance the HIV research process, assist in responding to issues involving research ethics, and prepare communities for HIV research. Clinical Trials 2008; 5: 147—156. http://ctj.sagepub.com

Purpose To discuss the reasoning and evidence supporting the principle of confidentiality of interim data with emphasis on the setting where the interim data include complete results on a short-term outcome.

Methods We review the reasons why wider access to interim data can increase the risk of false positive or false negative conclusions and discuss the types of harm which can occur. We provide illustrations and insights from recent experiences and discuss the level of consensus in the research community.

Results The arguments in favor of early release of interim data include the need to provide reliable data in a timely manner to patients and physicians, the potential to increase the enthusiasm of trial investigators, and to restore equipoise. However interim data, even where these include complete results on a short-term outcome measure, provide an unreliable and biased assessment of the overall benefit-to-risk profile of the trial treatments. Pre-judgment based on over-interpretation of such interim data can affect recruitment, treatment delivery, and follow-up, risking the ability of the trial to achieve its goals.

Conclusions In order to preserve the integrity of a trial and safeguard the interests of patients, interim data, including complete data on short-term outcomes, should remain confidential to the DMC and the statistical group responsible for preparing interim reports until the trial has achieved its primary objectives. Clinical Trials 2008; 5: 157—167. http://ctj.sagepub.com

Purpose To show how g-estimation based on randomization status is a natural generalization of ITT null testing to estimating efficacy from trials with important discontinuation or noncompliance.

Methods We contrast with an analysis of the effect of a tiotropium inhaler on the occurrence of chronic obstructive pulmonary disease (COPD) events in a six-month double-blind placebo-controlled trial of 1829 patients with good but imperfect compliance.

Results The covariate-adjusted point estimates, 95% confidence limits (CL), and null P-values comparing expected COPD event times in placebo versus tiotropium patients were: ITT, 1.21, CL = 1.02, 1.43, P = 0.027; on-treatment, 1.27, CL = 1.06, 1.52, P = 0.009; per-protocol, 1.36, CL = 1.13, 1.63, P = 0.001; and g-estimation, 1.31, CL = 1.03,1.72, P = 0.027. Thus g-estimation preserved the ITT test of the null, but exhibited more uncertainty about the size of the tiotropium effect than the other methods. In particular, it allowed for a much larger potential effect than did ITT analysis, but produced a much larger null P than exhibited by per-protocol analysis. Limitations Like ITT analysis, g-estimation requires all patients be followed to the end of the trial protocol, regardless of whether they comply with the protocol. Like on-treatment and per-protocol analyses, it also requires accurate compliance information be recorded.

Conclusion G-estimation should become a standard procedure for the analysis of trials with noncompliance. Software to do so is available in major packages, and the procedure is easily coded for other packages. Clinical Trials 2008; 5: 5—13. http://ctj.sagepub.com

Purpose The purpose of this study is to explore the possibility of incorporating an intermediate endpoint into interim futility testing of Phase III trials.

Methods Using a simple two-stage exponential survival model based on recent Southwest Oncology Group Phase III studies in several disease settings, we perform a series of simulation studies. Survival data are simulated under both the null and alternative hypotheses, and analyzed using overall survival, progression-free survival, and a composite endpoint for futility testing.

Results In all disease settings examined here, when survival data were simulated under the null hypothesis, the probability of stopping a trial early for futility was substantially increased by incorporating PFS into interim futility analyses. When testing for futility with the composite endpoint, average patient accrual was reduced by 6—11% in a wide variety of disease settings. In the study scenario with the longest survival, the savings in study duration was more dramatic than in patient resources. When data were simulated under the alternative hypothesis, this procedure resulted in a negligible loss of power.

Limitations The properties of this procedure outside of the context of cancer clinical trials and/or using a different intermediate endpoint are not examined. These results also do not address its performance in the context of less conservative stopping rules.

Conclusions Interim futility monitoring of Phase III trials using a suitable intermediate endpoint may substantially increase the probability of stopping early for futility when there is no treatment effect. These simulation studies suggest that this would lead to meaningful reductions in study duration and patient resources in many disease settings, with no substantial loss of power for the primary test of efficacy. Future work is needed to explore in detail whether modifications to the stopping rules used in this procedure may yield greater savings without compromising power. Clinical Trials 2008; 5: 14—22. http://ctj.sagepub.com

Purpose We sought to examine the effectiveness of blinding secondary stroke prevention trials with a warfarin treatment arm in which the blinding system incorporates use of placebo warfarin dose modification schedules for patients in the placebo warfarin arm.

Methods We examined treatment assignment guesses of 569 patients or their next of kin as well as study coordinators and principal neurologists at the clinical sites in a multicenter, randomized, double-dummy, double-blinded clinical trial of warfarin and aspirin using dose adjustment schedules for management of placebo warfarin. Results Overall, the crude rates of correct responses are 60% for patient/proxy, 66% for study coordinator, and 56% for principal neurologist. Several indices were used to assess the consistency of guesses with what would be expected if the guessing were done completely at random, and all measures indicate adequate blinding. Limitations Comparison to other trials using warfarin is difficult due to limited data and differences in assessment of blinding. However, results compared favorably to one existing trial.

Conclusions Placebo warfarin dose adjustment schedules can protect blinding adequately in trials involving warfarin. Clinical Trials 2008; 5: 23—30. http://ctj.sagepub.com

Purpose and Methods A specifically designed web-based data collection system, the Asthma Agents System, was used to collect daily data from participant children at school. These data were utilized to examine the intervention's effectiveness in reducing the frequency of asthma exacerbations. This study examines the Asthma Agents System's effect on the frequency of missing data. Data collection methods are discussed in detail, as well as the processes for retrieving missing data.

Results For the 290 children randomized, 97% of the daily data expected were available. Of the outcome data retrieved via the Asthma Agents System, 5% of those expected were missing during the period examined.

Limitations Challenges encountered in this study include issues regarding the use of technology in urban school settings, transfer of data between study sites, and availability of data during school breaks.

Conclusions Use of the Asthma Agents System resulted in lower rates of missing data than rates reported elsewhere in the literature. Clinical Trials 2008; 5: 31—37. http://ctj.sagepub.com

Complex regulations result in multiple ethics approvals for a multi-center study, unnecessary complexity in the study protocol, delays in securing regulatory approval, and cumbersome regulatory procedures, even for drugs widely used in clinical practice. The type of detailed safety monitoring currently needed in trials of new drugs is being applied indiscriminately to all studies including a simpler and basic level of monitoring that constitutes good practice in most trials could be agreed on, with that level being exceeded only in specific instances. More evidence about the pros and cons of alternative approaches to data quality monitoring would help inform this process. Complex procedures in the form of multiple-page consent forms, overzealous monitoring of side effects and adverse events, source data verification, and over-restrictive approaches to protocol amendments, can impede, rather than facilitate, trial objectives. Finally, further education on the nuances and functions of randomisation would facilitate trial conduct, and reduce the need for burdensome complexity. A radical re-evaluation of existing trial guidelines is needed, based on a clear understanding of the important principles of randomized trials, with the objective of eliminating unnecessary documentation and reporting without sacrificing validity or safety. Researchers should encourage public debate about how best to strike the balance between regulation and cost. Clinical Trials 2008; 5: 40—48. http://ctj.sagepub.com

Methods We reviewed the literature concerning central adjudication and documented the experience of adjudication in several clinical trials. Since definitions for nonfatal events are generally heterogeneous and subjective, one reason for a central process of adjudication is to assist in assuring systematic application of the definition used in the trial. In open-label trials, assuring that the adjudication is done blinded to treatment assignment may provide protection against differential misclassification. Regulatory authorities, including the FDA, derive confidence in the validity of results when central adjudication is performed. The clinical community has become accustomed to a certain amount of adjudication and may criticize trials that lack adjudication.

Limitations It is difficult to document the value of adjudication in trials that have reported adjudicated and nonadjudicated event rates and related treatment effects. Making rationale decisions about when and how to adjudicate is hampered by the lack of published study of when and how central adjudication is helpful to improve the quality and validity of trials and at what cost.

Conclusions Adjudication has not been shown to improve the ability to determine treatment effects. Thus, adjudication may be overly complex and overused in many large simple trials. The appropriate role of central adjudication — which trials, which outcomes, what methods — deserves scrutiny and further study. Clinical Trials 2008; 5: 56—60. http://ctj.sagepub.com

In particular, unnecessarily restrictive interpretation of the law may be a serious impediment to identification of potential participants for a trial, access to records to confirm events, continued follow-up of patients after the trial has been concluded, and secondary use of the trial data for purposes not directly related to the original purpose of the study. These obstacles could be overcome by better informing patients of the uses of records for medical research purposes, by using informed consent procedures that explain the nature of the research and the uses of the data, and by the use of identifiers, such as social security numbers that allow central follow-up.

The clinical trial research community needs to ensure that the substantial benefits of large-scale randomized trials are explained both to the public and to those responsible for introducing legislation. The negative impact of privacy legislation on the use of personal health information and on conducting large studies needs to be understood and minimized. Clinical Trials 2008; 5: 70—74. http://ctj.sagepub.com

Objective To assess the value of practices commonly employed in the conduct of large-scale clinical trials, and to identify areas where costs could be reduced without compromising scientific validity.

Methods In the qualitative phase of the study, an expert panel recommended potential modifications of mega-trial designs and operations in order to maximize their value (cost versus scientific benefit tradeoff). In the quantitative phase, a mega-trial economic model was used to assess the financial implications of these recommendations. Our initial chronic disease trial design included 20,000 patients randomized at 1000 sites. Each site was assigned 24 monitoring visits and a $10,000 per patient site payment. The case report form (CRF) was 60 pages long, and trial duration was assumed to be 48 months.

Results The total costs of the initial trial design were $421 million ($US 2007). Following the expert panel's recommendations, we varied study duration, CRF length, number of sites, electronic data capture (EDC), and site management components to determine their individual and combined effects upon total trial costs. The use of EDC and modified site management practices were associated with significant reductions in total trial costs. When reductions in all five trial components were combined in a streamlined pharmaceutical industry design, a 59% reduction in total trial costs resulted. When we assumed an even more streamlined trial design than has typically been considered for regulatory submissions in the past, there was a 90% reduction in total trial costs.

Conclusion Our results suggest that it is possible to reduce substantially the cost of large-scale clinical trials without compromising the scientific validity of their results. If implemented, our recommendations could free billions of dollars annually for additional clinical studies. Research in the setting of clinical trials should be conducted to refine these findings. Clinical Trials 2008; 5: 75—84. http:// ctj.sagepub.com