Estimating optimal decision trees for treatment assignment: The case of K > 2 treatment alternatives.

For many problems in clinical practice, multiple treatment alternatives are available. Given data from a randomized controlled trial or an observational study, an important challenge is to estimate an optimal decision rule that specifies for each client the most effective treatment alternative, given his or her pattern of pretreatment characteristics. In the present paper we will look for such a rule within the insightful family of classification trees. Unfortunately, however, there is dearth of readily accessible software tools for optimal decision tree estimation in the case of more than two treatment alternatives. Moreover, this primary tree estimation problem is also cursed with two secondary problems: a structural missingness in typical studies on treatment evaluation (because every individual is assigned to a single treatment alternative only), and a major issue of replicability. In this paper we propose solutions for both the primary and the secondary problems at stake. We evaluate the proposed solution in a simulation study, and illustrate with an application on the search for an optimal tree-based treatment regime in a randomized controlled trial on K = 3 different types of aftercare for younger women with early-stage breast cancer. We conclude by arguing that the proposed solutions may have relevance for several other classification problems inside and outside the domain of optimal treatment assignment.

Effectiveness of Human-Supported and Self-Help eHealth Lifestyle Interventions for Patients With Cardiometabolic Risk Factors: A Meta-Analysis.

OBJECTIVE: eHealth is a useful tool to deliver lifestyle interventions for patients with cardiometabolic diseases. However, there are inconsistent findings about whether these eHealth interventions should be supported by a human professional, or whether self-help interventions are equally effective. METHODS: Databases were searched between January 1995 and October 2021 for randomized controlled trials on cardiometabolic diseases (cardiovascular disease, chronic kidney disease, type 1 and 2 diabetes mellitus) and eHealth lifestyle interventions. A multilevel meta-analysis was used to pool clinical and behavioral health outcomes. Moderator analyses assessed the effect of intervention type (self-help versus human-supported), dose of human support (minor versus major part of intervention), and delivery mode of human support (remote versus blended). One hundred seven articles fulfilled eligibility criteria and 102 unique ( N = 20,781) studies were included. RESULTS: The analysis showed a positive effect of eHealth lifestyle interventions on clinical and behavioral health outcomes ( p < .001). However, these effects were not moderated by intervention type ( p = .169), dose ( p = .698), or delivery mode of human support ( p = .557). CONCLUSIONS: This shows that self-help eHealth interventions are equally effective as human-supported ones in improving health outcomes among patients with cardiometabolic disease. Future studies could investigate whether higher-quality eHealth interventions compensate for a lack of human support.Meta-analysis registration: PROSPERO CRD42021269263 .

eHealth to Improve Psychological Functioning and Self-Management of People With Chronic Kidney Disease: A Randomized Controlled Trial.

OBJECTIVE: Psychological distress is common among patients with chronic kidney disease and can interfere with disease self-management. We assessed the effectiveness of the personalized E-GOAL electronic health care pathway with screening and cognitive-behavioral therapy including self-management support, aimed to treat psychological distress and facilitate self-management among people with chronic kidney disease not on dialysis ( N = 121). METHODS: Primary outcome of the open two-arm parallel randomized controlled trial in four Dutch hospitals was psychological distress at posttest directly after the intervention and at 3-month follow-up. Secondary outcomes were physical and mental health-related quality of life, self-efficacy, chronic disease self-management, and personalized outcomes, that is, perceived progress compared with the previous time point on functioning (e.g., mood or social functioning) and self-management (e.g., dietary or medication adherence) outcomes that were prioritized by each individual. RESULTS: Linear mixed-effects analyses showed no significant time-by-group interaction effects for psychological distress, health-related quality of life, self-efficacy, and chronic condition self-management, whereas analyses of covariance showed significantly more perceived progress in the intervention group at posttest on personally prioritized areas of functioning ( b = 0.46, 95% confidence interval = 0.07-0.85) and self-management ( b = 0.55, 95% confidence interval = 0.16-0.95), with Cohen d values of 0.46 and 0.54 (medium effects), respectively. Effects on personalized outcomes were maintained at follow-up. CONCLUSIONS: Compared with regular care only, the electronic health intervention did not reduce psychological distress, whereas personalized outcomes did improve significantly after intervention. Future studies could consider personalized outcomes that reflect individually relevant areas and treatment goals, matching person-tailored treatments. TRIAL REGISTRATION: Registered at the Netherlands Trial Register with study number NTR7555 ( https://trialsearch.who.int/Trial2.aspx?TrialID=NTR7555 ).

Refined moderation analysis with categorical outcomes in precision medicine.

Moderation analysis is an integral part of precision medicine research. Concerning moderation analysis with categorical outcomes, we start with an interesting observation, which shows that heterogeneous treatment effects could be equivalently estimated via a role exchange between the outcome and the treatment variable in logistic regression models. Hence two estimators of moderating effects can be obtained. We then established the joint asymptotic normality for the two estimators, on which basis refined inference can be made for moderation analysis. The improved precision is helpful in addressing the lack-of-power problem that is common in search of moderators. The above-mentioned results hold for both experimental and observational data. We investigate the proposed method by simulation and provide an illustration with data from a randomized trial on wart treatment.

Moderators of the effect of guided online self-help for people with HIV and depressive symptoms.

The goal of the study was to investigate moderators of intervention effect of a guided Internet-based self-help cognitive behavioral intervention for people with HIV and depressive symptoms. This study was part of a randomized controlled trial where the intervention was found to be effective in reducing depressive symptoms, compared to an attention-only control group. Demographic characteristics (e.g., age), HIV characteristics (e.g., duration of HIV), and psychological characteristics (e.g., coping self-efficacy) were investigated as potential moderators of intervention effect. In 2015, 188 people with HIV and depressive symptoms were included in the study: 97 were randomized to the intervention group and 91 to the control group. Two moderators of intervention effect were found: coping self-efficacy and baseline depression severity. Participants with low coping self-efficacy and baseline depression severity improved more in the intervention group than in the control group, and participants with high coping self-efficacy and baseline depression severity improved in both groups. The results indicate that the intervention may be provided to all people with HIV and depressive symptoms. It may be especially important for people with HIV and low coping self-efficacy to start with the intervention since they show less improvement in the control group with only attention. Trial registration: Nederlands Trialregister NTR5407, September 11, 2015.

Multiple moderator meta-analysis using the R-package Meta-CART.

In meta-analysis, heterogeneity often exists between studies. Knowledge about study features (i.e., moderators) that can explain the heterogeneity in effect sizes can be useful for researchers to assess the effectiveness of existing interventions and design new potentially effective interventions. When there are multiple moderators, they may amplify or attenuate each other's effect on treatment effectiveness. However, in most meta-analysis studies, interaction effects are neglected due to the lack of appropriate methods. The method meta-CART was recently proposed to identify interactions between multiple moderators. The analysis result is a tree model in which the studies are partitioned into more homogeneous subgroups by combinations of moderators. This paper describes the R-package metacart, which provides user-friendly functions to conduct meta-CART analyses in R. This package can fit both fixed- and random-effects meta-CART, and can handle dichotomous, categorical, ordinal and continuous moderators. In addition, a new look ahead procedure is presented. The application of the package is illustrated step-by-step using diverse examples.

The Benefits of Psychosocial Interventions for Mental Health in People Living with HIV: A Systematic Review and Meta-analysis.

In this systematic review and meta-analysis we investigated the effectiveness of different psychosocial treatments for people living with HIV (PLWH) and mental health problems. Additionally, characteristics that may influence the effectiveness of a treatment (e.g., treatment duration) were studied. PubMed, PsycINFO and Embase were searched for randomized controlled trials on psychosocial interventions for PLWH. Depression, anxiety, quality of life, and psychological well-being were investigated as treatment outcome measures. Sixty-two studies were included in the meta-analysis. It was found that psychosocial interventions for PLWH had a small positive effect on mental health (g = 0.19, 95% CI [0.13, 0.25]). Furthermore, there was evidence for publication bias. Six characteristics influenced the effectiveness of a treatment for depression. For example, larger effects were found for studies with psychologists as treatment providers. To conclude, this systematic review and meta-analysis suggests that psychosocial interventions have a beneficial effect for PLWH with mental health problems.

Interventions to Promote Healthy Eating, Physical Activity and Smoking in Low-Income Groups: a Systematic Review with Meta-Analysis of Behavior Change Techniques and Delivery/Context.

PURPOSE: Healthy eating, physical activity and smoking interventions for low-income groups may have small, positive effects. Identifying effective intervention components could guide intervention development. This study investigated which content and delivery components of interventions were associated with increased healthy behavior in randomised controlled trials (RCTs) for low-income adults. METHOD: Data from a review showing intervention effects in 35 RCTs containing 45 interventions with 17,000 participants were analysed to assess associations with behavior change techniques (BCTs) and delivery/context components from the template for intervention description and replication (TIDieR) checklist. The associations of 46 BCTs and 14 delivery/context components with behavior change (measures of healthy eating, physical activity and smoking cessation) were examined using random effects subgroup meta-analyses. Synergistic effects of components were examined using classification and regression trees (meta-CART) analyses based on both fixed and random effects assumptions. RESULTS: For healthy eating, self-monitoring, delivery through personal contact, and targeting multiple behaviors were associated with increased effectiveness. Providing feedback, information about emotional consequences, or using prompts and cues were associated with reduced effectiveness. In synergistic analyses, interventions were most effective without feedback, or with self-monitoring excluding feedback. More effective physical activity interventions included behavioral practice/rehearsal or instruction, focussed solely on physical activity or took place in home/community settings. Information about antecedents was associated with reduced effectiveness. In synergistic analyses, interventions were most effective in home/community settings with instruction. No associations were identified for smoking. CONCLUSION: This study identified BCTs and delivery/context components, individually and synergistically, linked to increased and reduced effectiveness of healthy eating and physical activity interventions. The identified components should be subject to further experimental study to help inform the development effective behavior change interventions for low-income groups to reduce health inequalities.

Training staff to promote self-management in people with intellectual disabilities.

BACKGROUND: People with intellectual disabilities have increasing difficulties managing their daily affairs. This study examined the effectiveness of a staff training, which teaches staff to promote self-management in people with intellectual disabilities. METHOD: Effectiveness was assessed with questionnaires addressing clients' (n = 26) independence and self-reliance, support needs and challenging behaviour, using a pre-posttest control group design. Additionally, focus groups were conducted with trained staff members 6 months after the training. RESULTS: In the long term, the intervention group showed a significant increase in independence and self-reliance, in contrast to the comparison group. No effect was found on support needs and challenging behaviour. Trained staff members reported limited benefits of the training, but had noticed changes in their attitude and method of working afterwards. CONCLUSIONS: Further self-management research is required to investigate how independence and self-reliance can be promoted more effectively in this population. Future trainings should carefully consider their content, format, and implementation.

Physical activity and relaxation in the work setting to reduce the need for recovery: what works for whom?

BACKGROUND: To recover from work stress, a worksite health program aimed at improving physical activity and relaxation may be valuable. However, not every program is effective for all participants, as would be expected within a "one size fits all" approach. The effectiveness of how the program is delivered may differ across individuals. The aim of this study was to identify subgroups for whom one intervention may be better suited than another by using a new method called QUalitative INteraction Trees (QUINT). METHODS: Data were used from the "Be Active & Relax" study, in which 329 office workers participated. Two delivery modes of a worksite health program were given, a social environmental intervention (group motivational interviewing delivered by team leaders) and a physical environmental intervention (environmental modifications). The main outcome was change in Need for Recovery (NFR) from baseline to 12 month follow-up. The QUINT method was used to identify subgroups that benefitted more from either type of delivery mode, by incorporating moderator variables concerning sociodemographic, health, home, and work-related characteristics of the participants. RESULTS: The mean improvement in NFR of younger office workers in the social environmental intervention group was significantly higher than younger office workers who did not receive the social environmental intervention (10.52; 95 % CI: 4.12, 16.92). Furthermore, the mean improvement in NFR of older office workers in the social environmental intervention group was significantly lower than older office workers who did not receive the social environmental intervention ( -10.65; 95 % CI: -19.35, -1.96). The results for the physical environmental intervention indicated that the mean improvement in NFR of office workers (regardless of age) who worked fewer hours overtime was significantly higher when they had received the physical environmental intervention than when they had not received this type of intervention (7.40; 95 % CI: 0.99, 13.81). Finally, for office workers who worked more hours overtime there was no effect of the physical environmental intervention. CONCLUSIONS: The results suggest that a social environmental intervention might be more beneficial for younger workers, and a physical environmental intervention might be more beneficial for employees with a few hours overtime to reduce the NFR. TRIAL REGISTRATION: NTR2553.

Modeling Caries Experience: Advantages of the Use of the Hurdle Model.

In dental epidemiology, the decayed (D), missing (M), and filled (F) teeth or surfaces index (DFM index) is a frequently used measure. The DMF index is characterized by a strongly positive skewed distribution with a large stack of zero counts for those individuals without caries experience. Therefore, standard generalized linear models often lead to a poor fit. The hurdle regression model is a highly suitable class to model a DMF index, but its use is subordinated. We aim to overcome the gap between the suitability of the hurdle model to fit DMF indices and the frequency of its use in caries research. A theoretical introduction to the hurdle model is provided, and an extensive comparison with the zero-inflated model is given. Using an illustrative data example, both types of models are compared, with a special focus on interpretation of their parameters. Accompanying R code and example data are provided as online supplementary material.

Failure to meet language milestones at two years of age is predictive of specific language impairment.

AIM: This study established predictive properties of single language milestones for specific language impairment (SLI) after the age of four, as these had not previously been reported in the literature. METHODS: In this nested case-control study, children attending special needs schools for severe speech and language difficulties were matched with children attending mainstream schools. Data covering the ages of zero to four years were retrieved from well-child care clinics and the outcomes of 23 language milestones in the Dutch Developmental Instrument were analysed. The predictive properties were expressed as positive likelihood ratios, sensitivity and specificity. RESULTS: We included 253 pairs of children with and without SLI, aged from four to 11 years. The mean age was eight years and three months, and 77% were boys. From the age of 18 months, cases and controls differed significantly on all milestones (p < 0.01). After 24 months, the language milestones had positive likelihood ratios that ranging from 6 to 108. In general, language milestones had a high specificity (range 77-100%), but the sensitivity was relatively low (range 0-68%). CONCLUSION: Failure to meet language milestones from the age of 24 months was predictive of SLI, but the use of separate milestones had limited value due to low sensitivity.

Which Combinations of Techniques and Modes of Delivery in Internet-Based Interventions Effectively Change Health Behavior? A Meta-Analysis.

BACKGROUND: Many online interventions designed to promote health behaviors combine multiple behavior change techniques (BCTs), adopt different modes of delivery (MoD) (eg, text messages), and range in how usable they are. Research is therefore needed to examine the impact of these features on the effectiveness of online interventions. OBJECTIVE: This study applies Classification and Regression Trees (CART) analysis to meta-analytic data, in order to identify synergistic effects of BCTs, MoDs, and usability factors. METHODS: We analyzed data from Webb et al. This review included effect sizes from 52 online interventions targeting a variety of health behaviors and coded the use of 40 BCTs and 11 MoDs. Our research also developed a taxonomy for coding the usability of interventions. Meta-CART analyses were performed using the BCTs and MoDs as predictors and using treatment success (ie, effect size) as the outcome. RESULTS: Factors related to usability of the interventions influenced their efficacy. Specifically, subgroup analyses indicated that more efficient interventions (interventions that take little time to understand and use) are more likely to be effective than less efficient interventions. Meta-CART identified one synergistic effect: Interventions that included barrier identification/ problem solving and provided rewards for behavior change reported an average effect size that was smaller (g=0.23, 95% CI 0.08-0.44) than interventions that used other combinations of techniques (g=0.43, 95% CI 0.27-0.59). No synergistic effects were found for MoDs or for MoDs combined with BCTs. CONCLUSIONS: Interventions that take little time to understand and use were more effective than those that require more time. Few specific combinations of BCTs that contribute to the effectiveness of online interventions were found. Furthermore, no synergistic effects between BCTs and MoDs were found, even though MoDs had strong effects when analyzed univariately in the original study.

Quint: An R package for the identification of subgroups of clients who differ in which treatment alternative is best for them.

In the analysis of randomized controlled trials (RCTs), treatment effect heterogeneity often occurs, implying differences across (subgroups of) clients in treatment efficacy. This phenomenon is typically referred to as treatment-subgroup interactions. The identification of subgroups of clients, defined in terms of pretreatment characteristics that are involved in a treatment-subgroup interaction, is a methodologically challenging task, especially when many characteristics are available that may interact with treatment and when no comprehensive a priori hypotheses on relevant subgroups are available. A special type of treatment-subgroup interaction occurs if the ranking of treatment alternatives in terms of efficacy differs across subgroups of clients (e.g., for one subgroup treatment A is better than B and for another subgroup treatment B is better than A). These are called qualitative treatment-subgroup interactions and are most important for optimal treatment assignment. The method QUINT (Qualitative INteraction Trees) was recently proposed to induce subgroups involved in such interactions from RCT data. The result of an analysis with QUINT is a binary tree from which treatment assignment criteria can be derived. The implementation of this method, the R package quint, is the topic of this paper. The analysis process is described step-by-step using data from the Breast Cancer Recovery Project, showing the reader all functions included in the package. The output is explained and given a substantive interpretation. Furthermore, an overview is given of the tuning parameters involved in the analysis, along with possible motivational concerns associated with choice alternatives that are available to the user.

QUINT: A tool to detect qualitative treatment-subgroup interactions in randomized controlled trials.

OBJECTIVE: The detection of subgroups involved in qualitative treatment-subgroup interactions (i.e., for one subgroup of clients treatment A outperforms treatment B, whereas for another the reverse holds true) is crucial for personalized health. In typical Randomized Controlled Trials (RCTs), the combination of a lack of a priori hypotheses and a large number of possible moderators leaves current methods insufficient to detect subgroups involved in such interactions. A recently developed method, QUalitative INteraction Trees (QUINT), offers a solution. However, the paper in which QUINT has been introduced is not easily accessible for non-methodologists. In this paper, we want to review the conceptual basis of QUINT in a nontechnical way, and illustrate its relevance for psychological applications. METHOD: We present a concise introduction into QUINT along with a summary of available evidence on its performance. Subsequently, we subject RCT data on the effect of motivational interviewing in a treatment for substance abuse disorders to a reanalysis with QUINT. As outcome variables, we focus on measures of retention and substance use. RESULTS: A qualitative treatment-subgroup interaction is found for retention. By contrast, no qualitative interaction is detected for substance use. CONCLUSIONS: QUINT may lead to insightful and well-interpretable results with straightforward implications for personalized treatment assignment.

Students' beliefs and behaviour regarding low-calorie beverages, sweets or snacks: are they affected by lessons on healthy food and by changes to school vending machines?

OBJECTIVE: To study the effects of school lessons about healthy food on adolescents' self-reported beliefs and behaviour regarding the purchase and consumption of soft drinks, water and extra foods, including sweets and snacks. The lessons were combined with the introduction of lower-calorie foods, food labelling and price reductions in school vending machines. DESIGN: A cluster-randomized controlled design was used to allocate schools to an experimental group (i.e. lessons and changes to school vending machines) and a control group (i.e. 'care as usual'). Questionnaires were used pre-test and post-test to assess students' self-reported purchase of extra products and their knowledge and beliefs regarding the consumption of low-calorie products. SETTING: Secondary schools in the Netherlands. SUBJECTS: Twelve schools participated in the experimental group (303 students) and fourteen in the control group (311 students). The students' mean age was 13.6 years, 71.5% were of native Dutch origin and mean BMI was 18.9 kg/m(2). RESULTS: At post-test, the experimental group knew significantly more about healthy food than the control group. Fewer students in the experimental group (43%) than in the control group (56%) reported bringing soft drinks from home. There was no significant effect on attitude, social norm, perceived behavioural control and intention regarding the consumption of low-calorie extra products. CONCLUSIONS: The intervention had limited effects on students' knowledge and self-reported behaviour, and no effect on their beliefs regarding low-calorie beverages, sweets or snacks. We recommend a combined educational and environmental intervention of longer duration and engaging parents. More research into the effects of such interventions is needed.

Cultural, social and intrapersonal factors associated with clusters of co-occurring health-related behaviours among adolescents.

BACKGROUND: Adverse health-related behaviours (HRBs) have been shown to co-occur in adolescents. Evidence lacks on factors associated with these co-occurring HRBs. The Theory of Triadic Influence (TTI) offers a route to categorize these determinants according to type (social, cultural and intrapersonal) and distance in the causal pathway (ultimate or distal). Our aims were to identify cultural, social and intrapersonal factors associated with co-occurring HRBs and to assess the relative importance of ultimate and distal factors for each cluster of co-occurring HRBs. METHODS: Respondents concerned a random sample of 898 adolescents aged 12-18 years, stratified by age, sex and educational level of head of household. Data were collected via face-to-face computer-assisted interviewing and internet questionnaires. Analyses were performed for young (12-15 years) and late (16-18 years) adolescents regarding two and three clusters of HRB, respectively. RESULTS: For each cluster of HRBs (e.g. smoking, delinquency), associated factors were found. These accounted for 27 to 57% of the total variance per cluster. Factors came in particular from the intrapersonal stream of the TTI at the ultimate level and the social stream at the distal level. Associations were strongest for parenting practices, risk behaviours of friends and parents and self-control. CONCLUSION: Results of this study confirm that it is possible to identify a selection of cultural, social and intrapersonal factors associated with co-occurring HRBs among adolescents.

Effects of transfer-oriented curriculum on multiple behaviors in the Netherlands.

Many school health promotion curricula address a single health behavior, without paying attention to potential learning effects in associated behavioral domains. We developed an innovative curriculum about smoking and safe sex that also focused on promoting students' transfer of knowledge, skills and attitudes to other domains. In a quasi-experimental study involving 1107 students (Grades 7 and 8) in the Netherlands, the curriculum was compared with regular lessons about smoking and safe sex. The central research questions were to what extent the transfer-oriented curriculum: (i) had effects on psychosocial determinants and behaviors in the domains of smoking and safe sex, (ii) had effects on determinants and behaviors in three domains about which no lessons were taught (consumption of alcohol, fruit and breakfast). Multi-level analyses showed that the answer to both questions is positive. The results indicate that a transfer approach may have surplus value over the classic domain-specific approach and warrant further elaboration in the future.

Qualitative interaction trees: a tool to identify qualitative treatment-subgroup interactions.

When two alternative treatments (A and B) are available, some subgroup of patients may display a better outcome with treatment A than with B, whereas for another subgroup, the reverse may be true. If this is the case, a qualitative (i.e., disordinal) treatment-subgroup interaction is present. Such interactions imply that some subgroups of patients should be treated differently and are therefore most relevant for personalized medicine. In case of data from randomized clinical trials with many patient characteristics that could interact with treatment in a complex way, a suitable statistical approach to detect qualitative treatment-subgroup interactions is not yet available. As a way out, in the present paper, we propose a new method for this purpose, called QUalitative INteraction Trees (QUINT). QUINT results in a binary tree that subdivides the patients into terminal nodes on the basis of patient characteristics; these nodes are further assigned to one of three classes: a first for which A is better than B, a second for which B is better than A, and an optional third for which type of treatment makes no difference. Results of QUINT on simulated data showed satisfactory performance, with regard to optimization and recovery. Results of an application to real data suggested that, compared with other approaches, QUINT provided a more pronounced picture of the qualitative interactions that are present in the data.

A primer on the use of cluster analysis or factor analysis to assess co-occurrence of risk behaviors.

OBJECTIVE: The aim of this paper is to provide a guideline to a universal understanding of the analysis of co-occurrence of risk behaviors. The use of cluster analysis and factor analysis was clarified. METHOD: A theoretical introduction to cluster analysis and factor analysis and examples from literature were provided. A representative sample (N=4395) of the Dutch population, aged 16-40 and participating from fall 2005 to spring 2006, was used to illustrate the use of both techniques in assessing the co-occurrence of risk behaviors. RESULTS: Whereas cluster analysis techniques serve to focus on particular clusters of individuals showing the same behavioral pattern, factor analysis techniques are used to assess possible groups of interrelated health-risk behaviors that can be explained by an unknown common source. Choice between the techniques partly depends on the research question and the aim of the research, and has different implications for inferences and policy. CONCLUSION: By integrating theory and results from an illustrative example, a guideline has been provided that contributes towards a systematic approach in the assessment of co-occurrence of risk behaviors. Following this guideline, a better comparison between outcomes from various studies is expected, leading to improved effectiveness of multiple behavior change interventions.