Testing mediation: The endogeneity problem and the solution

A mediator is a dependent variable, m (e.g., charisma), that is thought to channel the effect of an independent variable, x (e.g., receiving training or not), on another dependent variable (e.g., subordinate satisfaction), y. In experimental settings x is manipulated-subjects are randomized to treatment-to isolate the causal effect of x on other variables. If m is not or cannot be manipulated, which is often the case, its causal effect on other variables cannot be determined; thus, standard mediation tests cannot inform policy or practice. I will show how an econometric procedure, called instrumental-variable estimation, can examine mediation in such cases.

Training medical students to conduct motivational interviewing: a randomized controlled trial.

OBJECTIVE: To examine the effectiveness of motivational interviewing (MI) training among medical students. METHODS: All students (n=131) (year 5) at Lausanne Medical School, Switzerland were randomized into an experimental or a control group. After a training in basic communication skills (control condition), an 8-h MI training was completed by 84.8% students in the exprimental group. One week later, students in both groups were invited to meet with two standardized patients. MI skills were coded by blinded research assistants using the Motivational Interviewing Treatment Integrity 3.0. RESULTS: Superior MI performance was shown for trained versus control students, as demonstrated by higher scores for "Empathy" [p<0.001] and "MI Spirit" [p<0.001]. Scores were similar between groups for "Direction", indicating that students in both groups invited the patient to talk about behavior change. Behavior counts assessment demonstrated better performance in MI in trained versus untrained students regarding occurences of MI-adherent behavior [p<0.001], MI non-adherent behavior [p<0.001], Closed questions [p<0.001], Open questions [p=0.001], simple reflections [p=0.03], and Complex reflections [p<0.001]. Occurrences were similar between groups regarding "Giving information". CONCLUSION: An 8-h training workshop was associated with improved MI performance. PRACTICE IMPLICATIONS: These findings lend support for the implementation of MI training in medical schools.

Increased activation in Broca's area after cognitive remediation in schizophrenia.

Functional magnetic resonance imaging (fMRI) was used to measure changes in cerebral activity in patients with schizophrenia after participation in the Cognitive Remediation Program for Schizophrenia and other related disorders (RECOS). As RECOS therapists make use of problem-solving and verbal mediation techniques, known to be beneficial in the rehabilitation of dysexecutive syndromes, we expected an increased activation of frontal areas after remediation. Executive functioning and cerebral activation during a covert verbal fluency task were measured in eight patients with schizophrenia before (T1) and after (T2) 14 weeks of RECOS therapy. The same measures were recorded in eight patients with schizophrenia who did not participate in RECOS at the same intervals of time (TAU group). Increased activation in Broca's area, as well as improvements in performance of executive/frontal tasks, was observed after cognitive training. Metacognitive techniques of verbalization are hypothesized to be the main factor underlying the brain changes observed in the present study.

Effect of training-sample size and classification difficulty on the accuracy of genomic predictors.

Introduction: As part of the MicroArray Quality Control (MAQC)-II project, this analysis examines how the choice of univariate feature-selection methods and classification algorithms may influence the performance of genomic predictors under varying degrees of prediction difficulty represented by three clinically relevant endpoints. Methods: We used gene-expression data from 230 breast cancers (grouped into training and independent validation sets), and we examined 40 predictors (five univariate feature-selection methods combined with eight different classifiers) for each of the three endpoints. Their classification performance was estimated on the training set by using two different resampling methods and compared with the accuracy observed in the independent validation set. Results: A ranking of the three classification problems was obtained, and the performance of 120 models was estimated and assessed on an independent validation set. The bootstrapping estimates were closer to the validation performance than were the cross-validation estimates. The required sample size for each endpoint was estimated, and both gene-level and pathway-level analyses were performed on the obtained models. Conclusions: We showed that genomic predictor accuracy is determined largely by an interplay between sample size and classification difficulty. Variations on univariate feature-selection methods and choice of classification algorithm have only a modest impact on predictor performance, and several statistically equally good predictors can be developed for any given classification problem.

Effect of intermittent hypoxic training on HIF gene expression in human skeletal muscle and leukocytes

Intermittent hypoxic exposure with exercise training is based on the assumption that brief exposure to hypoxia is sufficient to induce beneficial muscular adaptations mediated via hypoxia-inducible transcription factors (HIF). We previously demonstrated (Mounier et al. Med Sci Sports Exerc 38:1410-1417, 2006) that leukocytes respond to hypoxia with a marked inter-individual variability in HIF-1alpha mRNA. This study compared the effects of 3 weeks of intermittent hypoxic training on hif gene expression in both skeletal muscle and leukocytes. Male endurance athletes (n = 19) were divided into an Intermittent Hypoxic Exposure group (IHE) and a Normoxic Training group (NT) with each group following a similar 3-week exercise training program. After training, the amount of HIF-1alpha mRNA in muscle decreased only in IHE group (-24.7%, P < 0.05) whereas it remained unchanged in leukocytes in both groups. The levels of vEGF(121) and vEGF(165) mRNA in skeletal muscle increased significantly after training only in the NT group (+82.5%, P < 0.05 for vEGF(121); +41.2%, P < 0.05 for vEGF(165)). In leukocytes, only the IHE group showed a significant change in vEGF(165) (-28.2%, P < 0.05). The significant decrease in HIF-1alpha mRNA in skeletal muscle after hypoxic training suggests that transcriptional and post-transcriptional regulations of the hif-1alpha gene are different in muscle and leukocytes.

Effects of Intermittent Training on Anaerobic Performance and MCT Transporters in Athletes.

This study examined the effects of intermittent hypoxic training (IHT) on skeletal muscle monocarboxylate lactate transporter (MCT) expression and anaerobic performance in trained athletes. Cyclists were assigned to two interventions, either normoxic (N; n = 8; 150 mmHg PIO2) or hypoxic (H; n = 10; ∼3000 m, 100 mmHg PIO2) over a three week training (5×1 h-1h30.week-1) period. Prior to and after training, an incremental exercise test to exhaustion (EXT) was performed in normoxia together with a 2 min time trial (TT). Biopsy samples from the vastus lateralis were analyzed for MCT1 and MCT4 using immuno-blotting techniques. The peak power output (PPO) increased (p<0.05) after training (7.2% and 6.6% for N and H, respectively), but VO2max showed no significant change. The average power output in the TT improved significantly (7.3% and 6.4% for N and H, respectively). No differences were found in MCT1 and MCT4 protein content, before and after the training in either the N or H group. These results indicate there are no additional benefits of IHT when compared to similar normoxic training. Hence, the addition of the hypoxic stimulus on anaerobic performance or MCT expression after a three-week training period is ineffective.

Local heating of human skin causes hyperemia without mediation by muscarinic cholinergic receptors or prostanoids.

RESUME Les changements locaux de la température à la surface de la peau humaine ont une influence importante sur sa perfusion. La chaleur augmente localement le flux sanguin cutané, mais les mécanismes et les médiateurs de cette réponse (réponse thermique d'hyperémie) sont incomplètement élucidés. Dans la présente étude, nous avons examiné la relation possible entre la réponse thermique d'hyperémie, les récepteurs cholinergiques muscariniques et la production des prostaglandines vasodilatatrices. Chez 13 sujets de sexe masculin en bonne santé âgés entre 20 et 30 ans, une chambre métallique (contenant de l'eau) dont la température peut être contrôlée, a été placée sur la face palmaire de leur avant-bras et utilisée pour augmenter la température de surface de 34 à 41°C. L'hyperémie cutanée consécutive a été enregistrée par l'intermédiaire d'un scanner laser-Doppler. Dans une expérience, chacun des 8 sujets a reçu un bolus i.v. de glycopyrolate (agent antimuscarinique) (4 µg/kg) lors d'une visite et de NaCl 0,9% lors de l'autre visite. La réponse thermique d'hyperémie a été déterminée dans l'heure suivant les injections. Les glycopyrolate a efficacement empêché la vasodilation des micro-vaisseaux cutanés induite par iontophorèse d'acétylcholine mais n'a pas influencé la réponse thermique d'hyperémie. Dans une deuxième expérience entreprise avec 5 autres sujets 1 g d'aspirine (inhibiteur de la cyclooxygénase) administrée oralement a totalement supprimé la vasodilatation induite dans la peau par le courant anodique, sans modifier la réponse thermique d'hyperémie. La présente étude confirme l'absence de stimulation des récepteurs muscariniques et la production de prostaglandines vaso-dilatatrices dans la vasodilatation induite chez l'homme par réchauffement local de la peau de l'avant-bras. ABSTRACT Local changes in surface temperature have a powerful influence on the perfusion of human skin. Heating increases local skin blood flow (SkBF), but the mechanisms and mediators of this response (thermal hyperemia response) are incompletely elucidated. In the present study, we examined the possible dependence of the thermal hyperemia response on stimulation of muscarinic cholinergic receptors and on production of vasodilator prostanoids. In 13 male healthy subjects aged 20 - 30 years, a temperature- controlled chamber was positioned on the volar face of one forearm and used to raise surface temperature from 34to41°C. The time-course of the resulting thermal hyperemia response was recorded with a laser-Doppler imager. In one experiment, each of 8 subjects received an i.v. bolus of the antimuscarinic agent glycopyrrolate (4µg/kg) on one visit and saline on the other. The thermal hyperemia response was determined within the hour following the injections. Glycopyrrolate effectively inhibited the skin vasodilation induced by iontophoresis of acetylcholine, but did not influence the thermal hyperemia response. In a second experiment conducted in 5 other subjects, 1 gram of the cyclooxygenase inhibitor aspirin administered orally totally abolished the vasodilation induced in the skin by anodal current, but also failed to modify the thermal hyperemia response. The present study excludes the stimulation of muscarinic receptors and the production of vasodilator prostaglandins as essential and nonredundant mechanisms for the vasodilation induced by local heating in human forearm skin.

Same wording, distinct concepts? Testing differences between expectancies and motives in a mediation model of alcohol outcomes.

Per definition, alcohol expectancies (after alcohol I expect X), and drinking motives (I drink to achieve X) are conceptually distinct constructs. Theorists have argued that motives mediate the association between expectancies and drinking outcomes. Yet, given the use of different instruments, do these constructs remain distinct when assessment items are matched? The present study tested to what extent motives mediated the link between expectancies and alcohol outcomes when identical items were used, first as expectancies and then as motives. A linear structural equation model was estimated based on a national representative sample of 5,779 alcohol-using students in Switzerland (mean age = 15.2 years). The results showed that expectancies explained up to 38% of the variance in motives. Together with motives, they explained up to 48% of the variance in alcohol outcomes (volume, 5+ drinking, and problems). In 10 of 12 outcomes, there was a significant mediated effect that was often higher than the direct expectancy effect. For coping, the expectancy effect was close to zero, indicating the strongest form of mediation. In only one case (conformity and 5+ drinking), there was a direct expectancy effect but no mediation. To conclude, the study demonstrates that motives are distinct from expectancies even when identical items are used. Motives are more proximally related to different alcohol outcomes, often mediating the effects of expectancies. Consequently, the effectiveness of interventions, particularly those aimed at coping drinkers, should be improved through a shift in focus from expectancies to drinking motives.