Subgroups from regression trees with adjustment for prognostic effects and postselection inference.

Identification of subgroups with differential treatment effects in randomized trials is attracting much attention. Many methods use regression tree algorithms. This article addresses 2 important questions arising from the subgroups: how to ensure that treatment effects in subgroups are not confounded with effects of prognostic variables and how to determine the statistical significance of treatment effects in the subgroups. We address the first question by selectively including linear prognostic effects in the subgroups in a regression tree model. The second question is more difficult because it falls within the subject of postselection inference. We use a bootstrap technique to calibrate normal-theory t intervals so that their expected coverage probability, averaged over all the subgroups in a fitted model, approximates the desired confidence level. It can also provide simultaneous confidence intervals for all subgroups. The first solution is implemented in the GUIDE algorithm and is applicable to data with missing covariate values, 2 or more treatment arms, and outcomes subject to right censoring. Bootstrap calibration is applicable to any subgroup identification method; it is not restricted to regression tree models. Two real examples are used for illustration: a diabetes trial where the outcomes are completely observed but some covariate values are missing and a breast cancer trial where the outcome is right censored.

Identification of subgroups with differential treatment effects for longitudinal and multiresponse variables.

We describe and evaluate a regression tree algorithm for finding subgroups with differential treatments effects in randomized trials with multivariate outcomes. The data may contain missing values in the outcomes and covariates, and the treatment variable is not limited to two levels. Simulation results show that the regression tree models have unbiased variable selection and the estimates of subgroup treatment effects are approximately unbiased. A bootstrap calibration technique is proposed for constructing confidence intervals for the treatment effects. The method is illustrated with data from a longitudinal study comparing two diabetes drugs and a mammography screening trial comparing two treatments and a control. Copyright © 2016 John Wiley & Sons, Ltd.

A regression tree approach to identifying subgroups with differential treatment effects.

In the fight against hard-to-treat diseases such as cancer, it is often difficult to discover new treatments that benefit all subjects. For regulatory agency approval, it is more practical to identify subgroups of subjects for whom the treatment has an enhanced effect. Regression trees are natural for this task because they partition the data space. We briefly review existing regression tree algorithms. Then, we introduce three new ones that are practically free of selection bias and are applicable to data from randomized trials with two or more treatments, censored response variables, and missing values in the predictor variables. The algorithms extend the generalized unbiased interaction detection and estimation (GUIDE) approach by using three key ideas: (i) treatment as a linear predictor, (ii) chi-squared tests to detect residual patterns and lack of fit, and (iii) proportional hazards modeling via Poisson regression. Importance scores with thresholds for identifying influential variables are obtained as by-products. A bootstrap technique is used to construct confidence intervals for the treatment effects in each node. The methods are compared using real and simulated data.

A machine learning analysis of correlates of mortality among patients hospitalized with COVID-19.

It is vital to determine how patient characteristics that precede COVID-19 illness relate to COVID-19 mortality. This is a retrospective cohort study of patients hospitalized with COVID-19 across 21 healthcare systems in the US. All patients (N = 145,944) had COVID-19 diagnoses and/or positive PCR tests and completed their hospital stays from February 1, 2020 through January 31, 2022. Machine learning analyses revealed that age, hypertension, insurance status, and healthcare system (hospital site) were especially predictive of mortality across the full sample. However, multiple variables were especially predictive in subgroups of patients. The nested effects of risk factors such as age, hypertension, vaccination, site, and race accounted for large differences in mortality likelihood with rates ranging from about 2-30%. Subgroups of patients are at heightened risk of COVID-19 mortality due to combinations of preadmission risk factors; a finding of potential relevance to outreach and preventive actions.

Should all smokers use combination smoking cessation pharmacotherapy? Using novel analytic methods to detect differential treatment effects over 8 weeks of pharmacotherapy.

INTRODUCTION: Combination pharmacotherapy for smoking cessation has been shown to be more effective than monotherapy in meta-analyses. We address the question of whether combination pharmacotherapy should be used routinely with smokers or if some types of smokers show little or no benefit from combination pharmacotherapy versus monotherapy. METHODS: Two smoking cessation trials were conducted using the same assessments and medications (bupropion, nicotine lozenge, nicotine patch, bupropion + lozenge, and patch + lozenge). Participants were smokers presenting either to primary care clinics in southeastern Wisconsin for medical treatment (Effectiveness trial, N = 1,346) or volunteering for smoking cessation treatment at smoking cessation clinics in Madison and Milwaukee, WI (Efficacy trial, N = 1,504). For each trial, decision tree analyses identified variables predicting outcome from combination pharmacotherapy versus monotherapy at the end of treatment (smoking 8 weeks after the target quit day). RESULTS: All smokers tended to benefit from combination pharmacotherapy except those low in nicotine dependence (longer latency to smoke in the morning as per item 1 of the Fagerström Test of Nicotine Dependence) who also lived with a spouse or partner who smoked. CONCLUSIONS: Combination pharmacotherapy was generally more effective than monotherapy among smokers, but one group of smokers, those who were low in nicotine dependence and who lived with a smoking spouse, did not show greater benefit from using combination pharmacotherapy. Use of monotherapy with these smokers might be justified considering the expense and side effects of combination pharmacotherapy.

Using decision tree analysis to identify risk factors for relapse to smoking.

This research used classification tree analysis and logistic regression models to identify risk factors related to short- and long-term abstinence. Baseline and cessation outcome data from two smoking cessation trials, conducted from 2001 to 2002 in two Midwestern urban areas, were analyzed. There were 928 participants (53.1% women, 81.8% White) with complete data. Both analyses suggest that relapse risk is produced by interactions of risk factors and that early and late cessation outcomes reflect different vulnerability factors. The results illustrate the dynamic nature of relapse risk and suggest the importance of efficient modeling of interactions in relapse prediction.

Machine learning models of tobacco susceptibility and current use among adolescents from 97 countries in the Global Youth Tobacco Survey, 2013-2017.

Adolescents are particularly vulnerable to tobacco initiation and escalation. Identifying factors associated with adolescent tobacco susceptibility and use can guide tobacco prevention efforts. Novel machine learning (ML) approaches efficiently identify interactive relations among factors of tobacco risks and identify high-risk subpopulations that may benefit from targeted prevention interventions. Nationally representative cross-sectional 2013-2017 Global Youth Tobacco Survey (GYTS) data from 97 countries (28 high-income and 69 low-and middle-income countries) from 342,481 adolescents aged 13-15 years (weighted N = 52,817,455) were analyzed using ML regression tree models, accounting for sampling weights. Predictors included demographics (sex, age), geography (region, country-income), and self-reported exposure to tobacco marketing, secondhand smoke, and tobacco control policies. 11.9% (95% CI 11.1%-12.6%) of tobacco-naïve adolescents were susceptible to tobacco use and 11.7% (11.0%-12.5%) of adolescents reported using any tobacco product (cigarettes, other smoked tobacco, smokeless tobacco) in the past 30 days. Regression tree models found that exposure or receptivity to tobacco industry promotions and secondhand smoke exposure predicted increased risks of susceptibility and use, while support for smoke-free air policies predicted decreased risks of tobacco susceptibility and use. Anti-tobacco school education and health warning messages on product packs predicted susceptibility or use, but their protective effects were not evident across all adolescent subgroups. Sex, region, and country-income moderated the effects of tobacco promotion and control factors on susceptibility or use, showing higher rates of susceptibility and use in males and high-income countries, Africa and the Americas (susceptibility), and Europe and Southeast Asia (use). Tobacco policy-related factors robustly predicted both tobacco susceptibility and use in global adolescents, and interacted with adolescent characteristics and other environments in complex ways that stratified adolescents based on their tobacco risk. These findings emphasize the importance of efficient ML modeling of interactions in tobacco risk prediction and suggest a role for targeted prevention strategies for high-risk adolescents.

Gender, race, and education differences in abstinence rates among participants in two randomized smoking cessation trials.

INTRODUCTION: Smoking is the leading preventable cause of morbidity and mortality in the United States, but this burden is not distributed equally among smokers. Women, Blacks, and people with low socioeconomic status are especially vulnerable to the health risks of smoking and are less likely to quit. METHODS: This research examined cessation rates and treatment response among 2,850 participants (57.2% women, 11.7% Blacks, and 9.0% with less than a high school education) from two large cessation trials evaluating: nicotine patch, nicotine lozenge, bupropion, bupropion + lozenge, and nicotine patch + lozenge. RESULTS: Results revealed that women, Blacks, and smokers with less education were less likely to quit smoking successfully than men, Whites, and smokers with more education, respectively. Women did not appear to benefit more from bupropion than from nicotine replacement therapy, but women and smokers with less education benefited more from combination pharmacotherapy than from monotherapy. DISCUSSION: Women, Blacks, and smokers with less education are at elevated risk for cessation failure, and research is needed to understand this risk and develop pharmacological and psychosocial interventions to improve their long-term cessation rates.

Predictors of adherence to nicotine replacement therapy: Machine learning evidence that perceived need predicts medication use.

BACKGROUND: Nonadherence to smoking cessation medication is a frequent problem. Identifying pre-quit predictors of nonadherence may help explain nonadherence and suggest tailored interventions to address it. AIMS: Identify and characterize subgroups of smokers based on adherence to nicotine replacement therapy (NRT). METHOD: Secondary classification tree analyses of data from a 2-arm randomized controlled trial of Recommended Usual Care (R-UC, n = 315) versus Abstinence-Optimized Treatment (A-OT, n = 308) were conducted. R-UC comprised 8 weeks of nicotine patch plus brief counseling whereas A-OT comprised 3 weeks of pre-quit mini-lozenges, 26 weeks of nicotine patch plus mini-lozenges, 11 counseling contacts, and 7-11 automated reminders to use medication. Analyses identified subgroups of smokers highly adherent to nicotine patch use in both treatment conditions, and identified subgroups of A-OT participants highly adherent to mini-lozenges. RESULTS: Varied facets of nicotine dependence predicted adherence across treatment conditions 4 weeks post-quit and between 4- and 16-weeks post-quit in A-OT, with greater baseline dependence and greater smoking trigger exposure and reactivity predicting greater medication use. Greater quitting motivation and confidence, and believing that stop smoking medication was safe and easy to use were associated with greater adherence. CONCLUSION: Adherence was especially high in those who were more dependent and more exposed to smoking triggers. Quitting motivation and confidence predicted greater adherence, while negative beliefs about medication safety and acceptability predicted worse adherence. Results suggest that adherent use of medication may reflect a rational appraisal of the likelihood that one will need medication and will benefit from it.

Implementing Clinical Research Using Factorial Designs: A Primer.

Factorial experiments have rarely been used in the development or evaluation of clinical interventions. However, factorial designs offer advantages over randomized controlled trial designs, the latter being much more frequently used in such research. Factorial designs are highly efficient (permitting evaluation of multiple intervention components with good statistical power) and present the opportunity to detect interactions amongst intervention components. Such advantages have led methodologists to advocate for the greater use of factorial designs in research on clinical interventions (Collins, Dziak, & Li, 2009). However, researchers considering the use of such designs in clinical research face a series of choices that have consequential implications for the interpretability and value of the experimental results. These choices include: whether to use a factorial design, selection of the number and type of factors to include, how to address the compatibility of the different factors included, whether and how to avoid confounds between the type and number of interventions a participant receives, and how to interpret interactions. The use of factorial designs in clinical intervention research poses choices that differ from those typically considered in randomized clinical trial designs. However, the great information yield of the former encourages clinical researchers' increased and careful execution of such designs.

Toward precision smoking cessation treatment I: Moderator results from a factorial experiment.

BACKGROUND: The development of tobacco use treatments that are effective for all smokers is critical to improving clinical and public health. The Multiphase Optimization Strategy (MOST) uses highly efficient factorial experiments to evaluate multiple intervention components for possible inclusion in an optimized tobacco use treatment. Factorial experiments permit analyses of the influence of patient characteristics on main and interaction effects of multiple, relatively discrete, intervention components. This study examined whether person-factor and smoking characteristics moderated the main or interactive effects of intervention components on 26-week self-reported abstinence rates. METHODS: This fractional factorial experiment evaluated six smoking cessation intervention components among primary care patients (N=637): Prequit Nicotine Patch vs. None, Prequit Nicotine Gum vs. None, Preparation Counseling vs. None, Intensive Cessation In-Person Counseling vs. Minimal, Intensive Cessation Telephone Counseling vs. Minimal, and 16 vs. 8 Weeks of Combination Nicotine Replacement Therapy (NRT; nicotine patch+nicotine gum). RESULTS: Both psychiatric history and smoking heaviness moderated intervention component effects. In comparison with participants with no self-reported history of a psychiatric disorder, those with a positive history showed better response to 16- vs. 8-weeks of combination NRT, but a poorer response to counseling interventions. Also, in contrast to light smokers, heavier smokers showed a poorer response to counseling interventions. CONCLUSIONS: Heavy smokers and those with psychiatric histories demonstrated a differential response to intervention components. This research illustrates the use of factorial designs to examine the interactions between person characteristics and relatively discrete intervention components. Future research is needed to replicate these findings.

Comparative effectiveness of intervention components for producing long-term abstinence from smoking: a factorial screening experiment.

AIMS: To identify promising intervention components that help smokers attain and maintain abstinence during a quit attempt. DESIGN: A 2 × 2 × 2 × 2 × 2 randomized factorial experiment. SETTING: Eleven primary care clinics in Wisconsin, USA. PARTICIPANTS: A total of 544 smokers (59% women, 86% white) recruited during primary care visits and motivated to quit. INTERVENTIONS: Five intervention components designed to help smokers attain and maintain abstinence: (1) extended medication (26 versus 8 weeks of nicotine patch + nicotine gum); (2) maintenance (phone) counseling versus none; (3) medication adherence counseling versus none; (4) automated (medication) adherence calls versus none; and (5) electronic medication monitoring with feedback and counseling versus electronic medication monitoring alone. MEASUREMENTS: The primary outcome was 7-day self-reported point-prevalence abstinence 1 year after the target quit day. FINDINGS: Only extended medication produced a main effect. Twenty-six versus 8 weeks of medication improved point-prevalence abstinence rates (43 versus 34% at 6 months; 34 versus 27% at 1 year; P = 0.01 for both). There were four interaction effects at 1 year, showing that an intervention component's effectiveness depended upon the components with which it was combined. CONCLUSIONS: Twenty-six weeks of nicotine patch + nicotine gum (versus 8 weeks) and maintenance counseling provided by phone are promising intervention components for the cessation and maintenance phases of smoking treatment.

Comparative effectiveness of motivation phase intervention components for use with smokers unwilling to quit: a factorial screening experiment.

AIMS: To screen promising intervention components designed to reduce smoking and promote abstinence in smokers initially unwilling to quit. DESIGN: A balanced, four-factor, randomized factorial experiment. SETTING: Eleven primary care clinics in southern Wisconsin, USA. PARTICIPANTS: A total of 517 adult smokers (63.4% women, 91.1% white) recruited during primary care visits who were willing to reduce their smoking but not quit. INTERVENTIONS: Four factors contrasted intervention components designed to reduce smoking and promote abstinence: (1) nicotine patch versus none; (2) nicotine gum versus none; (3) motivational interviewing (MI) versus none; and (4) behavioral reduction counseling (BR) versus none. Participants could request cessation treatment at any point during the study. MEASUREMENTS: The primary outcome was percentage change in cigarettes smoked per day at 26 weeks post-study enrollment; the secondary outcomes were percentage change at 12 weeks and point-prevalence abstinence at 12 and 26 weeks post-study enrollment. FINDINGS: There were few main effects, but a significant four-way interaction at 26 weeks post-study enrollment (P = 0.01, ß = 0.12) revealed relatively large smoking reductions by two component combinations: nicotine gum combined with BR and BR combined with MI. Further, BR improved 12-week abstinence rates (P = 0.04), and nicotine gum, when used without MI, increased 26-week abstinence after a subsequent aided quit attempt (P = 0.01). CONCLUSIONS: Motivation-phase nicotine gum and behavioral reduction counseling are promising intervention components for smokers who are initially unwilling to quit.

Identifying effective intervention components for smoking cessation: a factorial screening experiment.

AIMS: To identify promising intervention components intended to help smokers to attain and maintain abstinence in their quit smoking attempts. DESIGN: A fully crossed, six-factor randomized fractional factorial experiment. SETTING: Eleven primary care clinics in southern Wisconsin, USA. PARTICIPANTS: A total of 637 adult smokers (55% women, 88% white) motivated to quit smoking who visited primary care clinics. INTERVENTIONS: Six intervention components designed to prepare smokers to quit, and achieve and maintain abstinence (i.e. for the preparation, cessation and maintenance phases of smoking treatment): (1) preparation nicotine patch versus none; (2) preparation nicotine gum versus none; (3) preparation counseling versus none; (4) intensive cessation in-person counseling versus minimal; (5) intensive cessation telephone counseling versus minimal; and (6) 16 versus 8 weeks of combination nicotine replacement therapy (nicotine patch + nicotine gum). MEASUREMENTS: Seven-day self-reported point-prevalence abstinence at 16 weeks. FINDINGS: Preparation counseling significantly improved week 16 abstinence rates (P = .04), while both forms of preparation nicotine replacement therapy interacted synergistically with intensive cessation in-person counseling (P < 0.05). Conversely, intensive cessation phone counseling and intensive cessation in-person counseling interacted antagonistically (P < 0.05)-these components produced higher abstinence rates by themselves than in combination. CONCLUSIONS: Preparation counseling and the combination of intensive cessation in-person counseling with preparation nicotine gum or patch are promising intervention components for smoking and should be evaluated as an integrated treatment package.

Enhancing the effectiveness of smoking treatment research: conceptual bases and progress.

BACKGROUND AND AIMS: A chronic care strategy could potentially enhance the reach and effectiveness of smoking treatment by providing effective interventions for all smokers, including those who are initially unwilling to quit. This paper describes the conceptual bases of a National Cancer Institute-funded research program designed to develop an optimized, comprehensive, chronic care smoking treatment. METHODS: This research is grounded in three methodological approaches: (1) the Phase-Based Model, which guides the selection of intervention components to be experimentally evaluated for the different phases of smoking treatment (motivation, preparation, cessation, and maintenance); (2) the Multiphase Optimization Strategy (MOST), which guides the screening of intervention components via efficient experimental designs and, ultimately, the assembly of promising components into an optimized treatment package; and (3) pragmatic research methods, such as electronic health record recruitment, that facilitate the efficient translation of research findings into clinical practice. Using this foundation and working in primary care clinics, we conducted three factorial experiments (reported in three accompanying papers) to screen 15 motivation, preparation, cessation and maintenance phase intervention components for possible inclusion in a chronic care smoking treatment program. RESULTS: This research identified intervention components with relatively strong evidence of effectiveness at particular phases of smoking treatment and it demonstrated the efficiency of the MOST approach in terms both of the number of intervention components tested and of the richness of the information yielded. CONCLUSIONS: A new, synthesized research approach efficiently evaluates multiple intervention components to identify promising components for every phase of smoking treatment. Many intervention components interact with one another, supporting the use of factorial experiments in smoking treatment development.

Identification of drug-specific pathways based on gene expression data: application to drug induced lung injury.

Identification of signaling pathways that are functional in a specific biological context is a major challenge in systems biology, and could be instrumental to the study of complex diseases and various aspects of drug discovery. Recent approaches have attempted to combine gene expression data with prior knowledge of protein connectivity in the form of a PPI network, and employ computational methods to identify subsets of the protein-protein-interaction (PPI) network that are functional, based on the data at hand. However, the use of undirected networks limits the mechanistic insight that can be drawn, since it does not allow for following mechanistically signal transduction from one node to the next. To address this important issue, we used a directed, signaling network as a scaffold to represent protein connectivity, and implemented an Integer Linear Programming (ILP) formulation to model the rules of signal transduction from one node to the next in the network. We then optimized the structure of the network to best fit the gene expression data at hand. We illustrated the utility of ILP modeling with a case study of drug induced lung injury. We identified the modes of action of 200 lung toxic drugs based on their gene expression profiles and, subsequently, merged the drug specific pathways to construct a signaling network that captured the mechanisms underlying Drug Induced Lung Disease (DILD). We further demonstrated the predictive power and biological relevance of the DILD network by applying it to identify drugs with relevant pharmacological mechanisms for treating lung injury.

Smoking cessation and the risk of diabetes mellitus and impaired fasting glucose: three-year outcomes after a quit attempt.

UNLABELLED: Weight gain after smoking cessation may increase diabetes mellitus and impaired fasting glucose (IFG) risk. This study evaluated associations between smoking cessation and continued smoking with incident diabetes and IFG three years after a quit attempt. The 1504 smokers (58% female) were mean (standard deviation) 44.7 (11.1) years old and smoked 21.4 (8.9) cigarettes/day. Of 914 participants with year 3 data, the 238 abstainers had greater weight gain, increase in waist circumference, and increase in fasting glucose levels than the 676 continuing smokers (p ≤ 0.008). In univariate analyses, Year 3 abstinence was associated with incident diabetes (OR = 2.60, 95% CI 1.44-4.67, p = .002; 4.3% absolute excess) and IFG (OR = 2.43, 95% CI 1.74-3.41, p<0.0001; 15.6% absolute excess). In multivariate analyses, incident diabetes was associated independently with older age (p = 0.0002), higher baseline body weight (p = 0.021), weight gain (p = 0.023), baseline smoking rate (p = 0.008), baseline IFG (p<0.0001), and baseline hemoglobin A1C (all p<0.0001). Smoking more at baseline predicted incident diabetes among eventual abstainers (p<0.0001); weighing more at baseline predicted incident diabetes among continuing smokers (p = 0.0004). Quitting smoking is associated with increased diabetes and IFG risk. Independent risk factors include older age, baseline body weight, baseline glycemic status, and heavier pre-quit smoking. These findings may help target smokers for interventions to prevent dysglycemia. TRIAL REGISTRATION: Clinicaltrials.gov NCT00332644.

Are tobacco dependence and withdrawal related amongst heavy smokers? Relevance to conceptualizations of dependence.

Measured tobacco dependence is typically only modestly related to tobacco withdrawal severity among regular smokers making a quit attempt. The weak association between dependence and withdrawal is notable because it conflicts with core theories of dependence and because both measures predict cessation outcomes, suggesting they both index a common dependence construct. This study used data from a smoking cessation comparative effectiveness trial (N = 1504) to characterize relations of tobacco dependence with craving and negative affect withdrawal symptoms using multiple dependence measures and analytic methods to detect both additive and interactive effects and to determine whether withdrawal meaningfully mediates the influence of dependence on smoking cessation. We conclude: (a) Although univariate analyses suggest dependence and withdrawal measures are only modestly interrelated, more powerful analytic techniques show they are, in fact, meaningfully related and their shared variance is associated with cessation likelihood; (b) there are clear differences between craving and negative affective withdrawal symptoms, with the former more related to smoking heaviness and the latter related to trait measures of negative affect; moreover, craving more strongly mediates dependence effects on cessation; and (c) both craving and negative affect withdrawal symptoms are strongly related to a pattern of regular smoking that is sensitive to the passage of time and powerfully affected by smoking cues. These findings support models that accord an important role for associative processes and withdrawal symptoms, especially craving, in drug dependence. The findings also support the use of withdrawal variables as criteria for the evaluation of dependence measures.

Tobacco withdrawal components and their relations with cessation success.

RATIONALE: Tobacco withdrawal is a key factor in smoking relapse, but important questions about the withdrawal phenomenon remain. OBJECTIVES: This research was intended to provide information about two core components of withdrawal (negative affect and craving): (1) how various withdrawal symptom profile dimensions (e.g., mean level, volatility, extreme values) differ between negative affect and craving; and (2) how these dimensions relate to cessation outcome. METHODS: Adult smokers (N = 1,504) in a double-blind randomized placebo-controlled smoking cessation trial provided real-time withdrawal symptom data four times per day for 4 weeks (2 weeks pre-quit and 2 weeks post-quit) via palmtop computers. Cessation outcome was biochemically confirmed 8-week point-prevalence abstinence. RESULTS: Examination of craving and negative affect dimensions following a cessation attempt revealed that craving symptoms differed from negative affect symptoms, with higher means, greater variability, and a greater incidence of extreme peaks. Regression analyses revealed that abstinence was associated with lower mean levels of both craving and negative affect and fewer incidences of extreme craving peaks. In a multivariate model, the increase in mean craving and negative affect scores each uniquely predicted relapse. CONCLUSIONS: Real-time reports revealed different patterns of abstinence-related negative affect and craving and that dimensions of both predict cessation outcome, suggesting that negative affect and craving dimensions each has motivational significance. This underscores the complexity of withdrawal as a determinant of relapse and the need to measure its distinct components and dimensions.