Search the Library
NOTE: This is a new search platform (as of May 2026). If you do a search and don’t get the results you were expecting, please email us at ctnlib@uw.edu to let us know? (If possible, please share your exact search strategy. Thank you!)
Enter keywords and hit Enter (or click the magnifying glass) to search. You can then also select document type or subject/topic to narrow results further (or just use those for searching without a keyword). Results display below this search form.
Document types
Subjects
- CTN-#### format for protocols (CTN-0001, e.g.)
- “exact phrase” (if phrase is not found, it will return results that contain all terms
- word1 NOT word2
- word1 word2 (finds both words)
- Click title to access full-text
- “Show details” reveals abstract & other info
- Checkboxes select items for copy/pasting or printing
- Need help getting a copy of a journal article?
Email ctnlib@uw.edu
Search results
Generalized linear mixed models (GLMM) are commonly used to analyze clustered data, but when the number of clusters is small to moderate, standard statistical tests may produce elevated type I error rates.
Small-sample corrections have been proposed for continuous or binary outcomes without covariate adjustment. However, appropriate tests to use for count outcomes or under covariate-adjusted models remains unknown.
An important setting in which this issue arises is in cluster-randomized trials (CRTs). Because many CRTs have just a few clusters (e.g., clinics or health systems), covariate adjustment is particularly critical to address potential chance imbalance and/or low power (e.g., adjustment following stratified randomization or for the baseline value of the outcome).
In this study, researchers conducted simulations to evaluate GLMM-based tests of the treatment effect that account for the small (10) or moderate (20) number of clusters under a parallel-group CRT setting across scenarios of covariate adjustment (including adjustment for one or more person-level or cluster-level covariates) for both binary and count outcomes. They illustrated their methods with an application to the CTN PRimary Care Opioid Use Disorders Treatment (PROUD) trial, CTN-0074, a parallel two-group cluster-randomized implementation trial randomizing 12 clinics across 6 health care systems.
The authors find that when the intraclass correlation is non-negligible (= 0.01) and the number of covariates is small (= 2), likelihood ratio tests with a between-within denominator degree of freedom have type I error rates close to the nominal level. When the number of covariates is moderate (= 5), across our simulation scenarios, the relative performance of the tests varied considerably and no method performed uniformly well. Therefore, they recommend adjusting for no more than a few covariates and using likelihood ratio tests with a between-within denominator degree of freedom.
Related protocols: CTN-0074
Understanding the mechanisms of action of interventions is a major general goal of scientific inquiry. The collection of statistical methods that use data to achieve this goal is referred to as mediation analysis. Natural direct and indirect effects provide a definition of mediation that matches scientific intuition, but they are not identified in the presence of time-varying confounding. Interventional effects have been proposed as a solution to this problem, but existing estimation methods are limited to assuming simple (e.g., linear) and unrealistic relations between the mediators, treatments, and confounders. We present an identification result for interventional effects in a general longitudinal data structure that allows flexibility in the specification of treatment-outcome, treatment-mediator, and mediator-outcome relationships. Identification is achieved under the standard no-unmeasured-confounders and positivity assumptions.
In this article, we study semi-parametric efficiency theory for the functional identifying the mediation parameter, including the non-parametric efficiency bound, and was used to propose non-parametrically efficient estimators. Implementation of our estimators only relies on the availability of regression algorithms, and the estimators in a general framework that allows the analyst to use arbitrary regression machinery were developed. The estimators are doubly robust, sqrt(n)-consistent, asymptotically Gaussian, under slow convergence rates for the regression algorithms used. This allows the use of flexible machine learning for regression while permitting uncertainty quantification through confidence intervals and p-values. A free and open-source R package implementing the methods is available on GitHub. The proposed estimator to a motivating example from a trial of two medications for opioid-use disorder was applied (CTN-0051, the X:BOT study), where we estimate the extent to which differences between the two treatments on risk of opioid use are mediated by craving symptoms.
Related protocols: CTN-0051
Causal mediation analysis has historically been limited in two important ways: (i) a focus has traditionally been placed on binary exposures and static interventions and (ii) direct and indirect effect decompositions have been pursued that are only identifiable in the absence of intermediate confounders affected by exposure. We present a theoretical study of an (in)direct effect decomposition of the population intervention effect, defined by stochastic interventions jointly applied to the exposure and mediators. In contrast to existing proposals, our causal effects can be evaluated regardless of whether an exposure is categorical or continuous and remain well-defined even in the presence of intermediate confounders affected by exposure. Our (in)direct effects are identifiable without a restrictive assumption on cross-world counterfactual independencies, allowing for substantive conclusions drawn from them to be validated in randomized controlled trials.
Beyond the novel effects introduced, we provide a careful study of nonparametric efficiency theory relevant for the construction of flexible, multiply robust estimators of our (in)direct effects, while avoiding undue restrictions induced by assuming parametric models of nuisance parameter functionals. To complement our nonparametric estimation strategy, we introduce inferential techniques for constructing confidence intervals and hypothesis tests, and discuss open-source software, the medshiftR package, implementing the proposed methodology. Application of our (in)direct effects and their nonparametric estimators is illustrated using data from a comparative effectiveness trial examining the direct and indirect effects of pharmacological therapeutics on relapse to opioid use disorder (CTN-0051, the X:BOT trial).
Related protocols: CTN-0051
The HIV/AIDS epidemic remains a major public health concern since the 1980s; untreated HIV infection has numerous consequences on quality of life. To optimize patients’ health outcomes and to reduce HIV transmission, this study, using data from CTN-0049 and CTN-0064, focused on vulnerable populations of people living with HIV (PLWH) and compared different predictive strategies for viral suppression using longitudinal or repeated measures.
The four methods of predicting viral suppression are (1) including the repeated measures of each feature as predictors, (2) utilizing only the initial (baseline) value of the feature as predictor, (3) using the last observed value as the predictors and (4) using a growth curve estimated from the features to create individual-specific prediction of growth curves as features. These models were compared using Synthetic Random Forests (SRF).
The SRF models predicted HIV viral suppression in CTN-0064 with an accuracy rate as high as 70%. The person-specific trajectories (Model 4) had the best predictive performance of the approaches. Not surprisingly, among the other models, those with characteristics from closer time-points produced better model fit than using baseline aspects only.
Conclusions: The model with person-specific trajectories had the best predictive power as compared to other models. The findings from this study are valuable, since they provide evidence that incorporating not just levels of predictors but also their change over time improves predictive performance of our models. Using person-specific intercepts and slopes provides a novel and useful approach to creating predictive models using repeated measurements. It also suggests the possibility of incorporating these types of modeling efforts into ongoing clinical monitoring using medical records.
Related protocols: CTN-0049, CTN-0064
In randomized controlled trials (RCTs), a common strategy to increase power to detect a treatment effect is adjustment for baseline covariates. However, adjustment with partly missing covariates, where complete cases are only used, is inefficient. This paper considers different alternatives in trials with discrete-time survival data, where subjects are measured in discrete-time intervals while they may experience an event at any point in time. The results of a Monte Carlo simulation study, as well as a case study of randomized trials in smokers with attention deficit hyperactivity disorder (CTN-0029), indicated that single and multiple imputation methods outperform the other methods and increase precision in estimating the treatment effect. Missing indicator method, which uses a dummy variable in the statistical model to indicate whether the value for that variable is missing and sets the same value to all missing values, is comparable to imputation methods. Nevertheless, the power level to detect the treatment effect based on missing indicator method is marginally lower than the imputation methods, particularly when the missingness depends on the outcome.
In conclusion, complete case analysis is wasteful and drops the power level to a large degree, resulting in an undetectable treatment effect. Also, it can introduce bias in the estimate of the treatment effects when the missingness mechanism depends on the outcome variable. This method is therefore invalid and the authors do not recommend it. Instead, it appears that imputation of partly missing (baseline) covariates should be preferred in the analysis of discrete-time survival data.
Related protocols: CTN-0029
HIV testing is the foundation for consolidated HIV treatment and prevention. This study aimed to discover the most relevant variables for predicting HIV testing uptake among substance users in substance use disorder treatment programs by applying random forest (RF), a robust multivariate statistical learning method. It also provides a descriptive introduction to this method for those who are unfamiliar with it. This secondary analysis used data from the NIDA Clinical Trials Network HIV testing and counseling study (CTN-0032). A total of 1281 HIV-negative or status unknown participants from 12 U.S. community-based substance use disorder treatment programs were included and were randomized into three HIV testing and counseling groups. The a priori primary outcomes was self-reported receipt of HIV test results. Classification accuracy of RF was compared to logistic regression, a standard statistical approach for binary outcomes. Variable importance measures for the RF model were used to select the most relevant variables. RF based models produced much higher classification accuracy than those based on logistic regression. Treatment group is the most important predictor among all covariates, with a variable importance index of 12.9%. RF variable importance revealed that several types of condomless sex behaviors, condom use self-efficacy and attitudes towards condom use, and level of depression are the most important predictors of receipt of HIV testing results. There is a non-linear negative relationship between count of condomless sex acts and the receipt of HIV testing.
Conclusions: RF seems promising in discovering important factors related to HIV testing uptake among large numbers of predictors and should be encouraged in future HIV prevention and treatment research and intervention program evaluations.
Related protocols: CTN-0032
The purpose of this study was to estimate how results would have varied if a substance abuse clinical trial had been conducted with nationally representative adults with substance use and with representative adults receiving substance use treatment. Results were analyzed from NIDA Clinical Trials Network protocol CTN-0044, a multisite clinical trial comparing the effectiveness of the Therapeutic Education System to treatment as usual for outpatient addiction treatment (n = 507). Patients were recruited between June 2010 and August 2011. Abstinence was the primary outcome. The general population sample and general population-treated samples were derived from Wave 1 of the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) (n = 43,093). Propensity scores provided a standardized measure of the difference between clinical trial participants and the 2 NESARC samples. The clinical trial was reanalyzed by reweighting the sample with propensity scores derived from the 2 samples to obtain generalizable estimates of treatment effects.
Before the clinical trial sample was reweighted, the odds ratio (OR) of response to Therapeutic Education System versus treatment as usual in the trial was 1.62 (95% CI, 1.12-2.35). After the sample was reweighted to be representative of the 2 NESARC groups, ORs were 1.33 (95% CI, 0.34-5.26) for the representative sample with any substance use and 1.64 (95% CI, 0.82-3.27) for the representative treated sample. The effect size of the original study was statistically significant; the estimate effect size for the nationally representative sample was not. This does not necessarily mean that the Therapeutic Education System is not efficacious for the treatment of substance use disorders. Instead, the width of the confidence intervals reflects increased uncertainty associated with extrapolating the results of the clinical trial sample to broader populations.
Conclusions: Applying propensity score weighting to clinical trial results provides a method for estimating the population generalizability of clinical trial findings that relies on effect moderators observed in the study sample and population. Broader confidence intervals in the reweighted samples do not necessarily indicate lack of efficacy of the Therapeutic Education System but rather greater uncertainty concerning effectiveness in general population samples.
Related protocols: CTN-0044
Traditional approaches to subgroup analyses that test each moderating factor as a separate hypothesis can lead to erroneous conclusions due to the problems of multiple comparisons, model misspecification, and multicollinearity. This study aimed to demonstrated a novel, systematic approach to subgroup analyses that avoids these pitfalls. A Best Approximating Model (BAM) approach that identifies multiple moderators and estimates their simultaneous impact on treatment effect sizes was applied to a randomized, controlled, 11-week, double-blind efficacy trial on smoking cessation of adult smokers with attention-deficit/hyperactivity disorder (ADHD), randomized to either OROS-methylphenidate (n=127) or placebo (n=128) and treated with nicotine patch (National Drug Abuse Treatment Clinical Trials Network protocol CTN-0029). Binary outcomes measures were prolonged smoking abstinence and point prevalence smoking abstinence.
Although the original clinical trial data analysis showed no treatment effect on smoking cessation, the BAM analysis showed significant subgroup effects for the primary outcome of prolonged smoking abstinence: (1) lifetime history of substance use disorders, and (2) more severe ADHD symptoms. A significant subgroup effect was also shown for the secondary outcome of point prevalence smoking abstinence — age 18-29 years.
Conclusions: The BAM analysis resulted in different conclusions about subgroup effects compared to a hypothesis-driven approach. These divergent findings underscore the need for investigators to consider more advanced statistical methods to better analyze subgroup effect sizes. By examining moderator independence and avoiding multiple testing, BAMs have the potential to better identify and explain how treatment effects vary across subgroups in heterogeneous patient populations, thus providing better guidance to more effectively match individual patients with specific treatments.
Related protocols: CTN-0029
This secondary analysis of data from National Drug Abuse Treatment Clinical Trials Network protocol CTN-0003 (“Suboxone (Buprenorphine/Naloxone) Taper: A Comparison of Taper Schedules”) compared three missing data strategies: 1) Latent growth model that assumes the data are missing at random (MAR), 2) Diggle-Kenward missing not at random (MNAR) model where dropout is a function of previous/concurrent urinalysis (UA) submissions, and 3) Wu-Carroll MNAR model where dropout is a function of the growth factors. CTN-0003 examined a 7-day versus 28-day taper for buprenorphine/naloxone to see which taper schedule reduced the likelihood of submitting an opioid-positive UA during treatment.
The MAR model showed a significant effect (B=-0.45, p <0.05) of trial arm on the opioid-positive UA slope (i.e., 28-day taper participants were less likely to submit a positive UA over time) with a small effect size (d=0.20). The MNAR Diggle-Kenward model demonstrated a significant (B=-0.64, p<0.01) effect of trial arm on the slope with a large effect size (d=0.82). The MNAR Wu-Carroll model evidenced a significant (B=-0.41, p<0.05) effect of trial arm on the UA slope that was relatively small (d=0.31).
Conclusions: This performance comparison of three missing data strategies (latent growth model, Diggle-Kenward selection model, Wu-Carrol selection model) on sample data indicates a need for increased use of sensitivity analyses in clinical trial research. Given the potential sensitivity of the trial arm effect to missing data assumptions, it is critical for researchers to consider whether the assumptions associated with each model are defensible.
In randomized controlled trials (RCTs), the most compelling need is to determine whether the treatment condition was more effective than the control. However, it is generally recognized that not all participants in the treatment group of most clinical trials benefit equally. While subgroup analyses are often used to compare treatment effectiveness across pre-determined subgroups categorized by patient characteristics, methods to empirically identify naturally occurring clusters of persons who benefit most from the treatment group have rarely been implemented. This article provides a modeling framework to accomplish this important task.
Utilizing information about individuals from the treatment group who had poor outcomes, the present study proposes an a priori clustering strategy that classifies the individuals with initially good outcomes in the treatment group into: (a) group GE (good outcome, effective), the latent subgroup of individuals for whom the treatment is likely to be effective and (b) group GI (good outcome, ineffective), the latent subgroup of individuals for whom the treatment is not likely to be effective. The method is illustrated through a reanalysis of a publicly available data set from the National Institute on Drug Abuse’s National Drug Abuse Treatment Clinical Trials Network (protocol CTN-0004). That study examined the effectiveness of motivational enhancement therapy from 461 outpatients with substance use disorder problems. As a diagnostic means utilizing out-of-sample forecasting performance, the present study compared the relapse rates during the long-term follow-up period for the two subgroups. As expected, group GI, composed of individuals for whom the treatment was hypothesized to be ineffective, had a significantly higher relapse rate than group GE (63% vs. 27%).
Conclusions: The proposed method, LGEM, identified latent subgroups GE and GI, and the comparison between the two groups revealed several significantly different and informative characteristics even though both subgroups had good outcomes during the immediate post-therapy period. LGEM has potential as a means of further exploring reasons why individuals respond to treatment conditions, regardless of which treatment arm they are exposed to, and can be implemented after the trial is completed, without need for a pre-specified design and can be used by any type of RCT in a variety of topic areas.
Related protocols: CTN-0004
Given that treatment dropout among stimulant abusers has been found in prior research to be associated with relapse and continued substance use, identifying variables that best predict treatment completion for particular subgroups among stimulant abusers may aid clinicians in targeting dropout prevention strategies. The purpose of this study was to explore the selection of predictor variables in the evaluation of drug treatment completion using an ensemble approach with classification trees. The basic methodology is reviewed and the subagging procedure of random subsampling is applied. Among 234 individuals with stimulant use disorders randomized to a 12-step facilitative intervention shown to increase stimulant use abstinence (National Drug Abuse Treatment Clinical Trials Network study CTN-0031, “STAGE-12”), 67.52% were classified as treatment completers. A total of 122 baseline variables were used to identify factors associated with completion. The number of types of self-help activity involvement prior to treatment was the predominant predictor. Other effective predictors included better coping self-efficacy for substance use in high-risk situations, more days of prior meeting attendance, greater acceptance of the disease model, higher confidence for not resuming use following discharge, lower ASI Drug and Alcohol composite scores, negative urine screens for cocaine or marijuana, and fewer employment problems.
Conclusions: The application of an ensemble subsampling regressions tree method utilizes the fact that classification trees are unstable but, on average, produce an improved prediction of the completion of drug abuse treatment. The results support the notion that there are early indicators of treatment completion that may allow for modification of approaches more tailored to fitting the needs of individuals and potentially provide more successful treatment engagement and improved outcomes. Given these results, in addition to considering mostly static variables like race, gender, or marital status, researchers should attend to the selection of more dynamic variables, such as confidence and self-efficacy, that may have stronger implications in the development of treatment interventions.
Related protocols: CTN-0031
In case multiple treatment alternatives are available for some medical problem, the detection of treatment-subgroup interactions (i.e., relative treatment effectiveness varying over subgroups of persons) is of key importance for personalized medicine and the development of optimal treatment strategies. Randomized clinical trials (RCTs) often go without clear a priori hypotheses on the subgroups involved in treatment-subgroup interactions, and with a large number of pre-treatment characteristics in the data. In such situations, relevant subgroups (defined in terms of pre-treatment characteristics) are to be induced during the actual data analysis. This comes down to a problem of cluster analysis, with the goal of this analysis being to find clusters of persons that are involved in meaningful treatment-person cluster interactions. For such a cluster analysis, five recently proposed methods can be used, all being of a recursive partitioning type. However, these five methods have been developed almost independently, and the relations between them are not yet understood.
This paper aims to close that gap. It starts by outlining the basic principles behind each method, and by illustrating it with an application on a data set from an RCT in the National Drug Abuse Treatment Clinical Trials Network that evaluated two treatment strategies for substance abuse problems (CTN-0005, “Motivational Interviewing (MI) to Improve Treatment Engagement and Outcome in Subjects Seeking Treatment for Substance Abuse”). Next, it presents a comparison of the methods, focusing on major similarities and differences. The discussion concludes with practical advice for end users with regard to the selection of a suitable method, and with an important challenge for future research in this area.
Related protocols: CTN-0005
Overdispersion and structural zeros are two major manifestations of departure from the Poisson assumption when modeling count responses using Poisson log-linear regression. As noted in a large body of literature, ignoring such departures could yield bias and lead to wrong conclusions. Different approaches have been developed to tackle these two major problems. This paper reviews available methods for dealing with overdispersion and structural zeroes within a longitudinal data setting and proposes a distribution-free modeling approach to address the limitations of these methods by utilizing a new class of functional response methods.
This approach is illustrated first with simulated data, and then with real study data from the National Drug Abuse Treatment Clinical Trials Network protocol CTN-0018 (“Reducing HIV/STD Risk Behaviors : A Research Study for Men in Drug Abuse Treatment”). The examples demonstrate that the proposed approach works well for longitudinal data under both complete and missing data settings, as well as for samples with a sample size as small as 50.
Related protocols: CTN-0018
This study examined the impact of contingency management (CM) on stimulant use heterogeneity across two 12-week clinical trials, National Drug Abuse Treatment Clinical Trials Network protocols CTN-0006 and CTN-0007. The hypothesis was that CM effects on stimulant use would differ across multiple sub-groups of patients with distinct trajectories of use throughout the treatment period. The outcome of positive stimulant urine analysis (UA+) was measured two times per week for 12 weeks. Growth mixture modeling was used to estimate multiple latent class solutions (classes 1 through 6). The best fitting, clinically interpretable model was the 3-class linear model (BIC=7624). The model produced the following classes: Class 1 (21% of sample) = low probability (35%) of UA+ at baseline, steep decline in UA+ submissions during treatment. Class 2 (38%) = moderate probability of UA+ at baseline (42%), moderate decline in UA+ submissions over time caused by CM. Class 3 (41%) = high probability of UA+ at baseline (65%), increase in UA+ submissions over time and no effect of CM.
Conclusions: Identifying sub-groups may help explain heterogeneity in substance use trajectories and identify characteristics that could inform treatment nonresponse (e.g. Class 3). Such models could also assist with identifying segments of the stimulant use population who could benefit from ancillary services in order to more effectively impact abstinence.
Related protocols: CTN-0006, CTN-0007
This two-hour webinar, produced by the National Drug Abuse Treatment Clinical Trials Network (CTN) Clinical Coordinating Center for CTN members and the public, features a plain-English description of the intuition behind basic statistical concepts used in clinical trials. Its content requires no statistical background and aims to bridge the communication gap between researchers and biostatisticians. It does not teach how to perform statistical tasks; there are no formulas and no proofs. Instead, it explains why these statistical tasks are performed and what they mean once they are performed.
This webinar is intended for CTN members and the public, especially non-statisticians with experience in clinical trials who seek a better understanding of statistical concepts encountered throughout the cycle of a clinical trial.
Note: The webinar included the showing of two short videos, which you can only hear the sound for in this recording (the picture is blank). If you wish to view the short films, they can both be found on YouTube:
First clip:
http://www.youtube.com/watch?v=AukrpAoAYY0
Second clip:
http://www.youtube.com/watch?v=AukrpAoAYY0
Presented by Paul G. Wakim, PhD (NIDA Center for the Clinical Trials Network) and Abigail G. Matthews, PhD (CTN Data & Statistics Center, EMMES).