Search

Aims To illustrate a machine learning-based approach for identifying and investigating moderators of alcohol use intervention effects in aggregate-data meta-analysis. Methods We illustrated the machine learning technique of random forest modeling using data from an ongoing meta-analysis of brief substance use interventions implemented in general healthcare settings. A subset of 40 trials testing brief alcohol interventions (BAIs) was used; these trials provided 344 estimates of post-intervention effects on participants' alcohol use as well as data on 20 potential moderators of intervention effects. These candidate moderators included characteristics of trial methodology and implementation, intervention design and participant samples. Results The best-fitting random forest model identified 10 important moderators from the pool of 20 candidate moderators. Meta-regression utilizing the selected moderators found that inclusion of prescriptive advice in a BAI session significantly moderated BAI effects on alcohol use. Observed effects were also significantly moderated by several methodological characteristics of trials, including the type of comparison group used, the overall level of attrition and the strategy used to address missing data. In a meta-regression model that included all candidate moderators, fewer coefficients were found to be significant, indicating that the use of a preliminary data reduction technique to identify only important moderators for inclusion in final analyses may have yielded improved statistical power to detect moderation. Conclusions Machine learning methods can be valuable tools for clarifying the influence of trial, intervention and sample characteristics on alcohol use intervention effects, in particular when numerous candidate moderators are available. [ABSTRACT FROM AUTHOR]

Background: : Neonatal vitamin A supplementation (NVAS) is an intervention hypothesized to reduce infant morbidity and mortality. The objective of this study was to assess the efficacy of neonatal vitamin A supplementation in reducing infant morbidity and mortality and assess potential sources of heterogeneity of the effect of NVAS.Methods: : We completed an individually randomized, double-blind, placebo-controlled trial in Tanzania. Infants were randomized within 3 days of birth to a single dose of vitamin A (50 000 IU) or placebo. We assessed infants at 1 and 3 days after supplementation, as well as 1, 3, 6 and 12 months after supplementation. We included all live births in the analysis and used relative risks (RR) and 95% confidence intervals (CI) to assess the risks of mortality and hospitalization by 12 months. We used general estimating equations to assess the incidence of morbidities during infancy.Results: : A total of 31 999 infants were enrolled in the study between August 2010 and March 2013. At 12 months, vitamin A did not reduce all-cause infant mortality (RR 1.04; 95% CI 0.92-1.16), nor affect hospitalization (RR 1.09; 95% CI 0.97-1.22) or all-cause morbidity (RR 1.00; 95% CI 0.96-1.05). Postpartum maternal vitamin A supplementation modified the effect of neonatal vitamin A supplementation on mortality at 12 months ( P -value, test for interaction = 0.04). Among infants born to women who received a mega-dose of vitamin A after delivery, NVAS appeared to increase the risk of death (RR 1.12; 95% CI 0.98-1.29), whereas the risk of death among infants born to women who did not receive a mega-dose was reduced (RR 0.86; 95% CI 0.70-1.06). We noted no modification of the effect of NVAS by infant gender, birthweight or maternal HIV status.Conclusion: : NVAS did not affect the risk of death or incidence of common childhood morbidities. However, this study sheds light on potential sources of heterogeneity of the effect of neonatal vitamin A supplementation which should be further examined in a pooled analysis of all NVAS trials. [ABSTRACT FROM AUTHOR]