Random effects model. AC indicates active website; ACE, active website with email; ACP, active website with telephone; ACT-E, acceptance and commitment therapy with extensive email support; ACT-M, acceptance and commitment therapy with minimal email support; C, control website; CP, control website with telephone; GI, group intervention; II, internet intervention; and SMD, standardized mean difference.
eAppendix. Search Strategies
eTable 1. Characteristics of 29 Randomized Clinical Trials Included to Prevent Anxiety
eTable 2. Risk of Bias of 29 Randomized Clinical Trials Included to Prevent Anxiety
eTable 3. Data Used to Calculate SM
eTable 4. Subgroup Analysis of Effectiveness of Psychological and/or Educational Interventions to Prevent Anxiety
eFigure 1. Funnel Plot
eFigure 2. Normal Probability Plot of Standardized Shrunken Residuals
Customize your JAMA Network experience by selecting one or more topics from the list below.
Moreno-Peral P, Conejo-Cerón S, Rubio-Valera M, et al. Effectiveness of Psychological and/or Educational Interventions in the Prevention of Anxiety: A Systematic Review, Meta-analysis, and Meta-regression. JAMA Psychiatry. 2017;74(10):1021–1029. doi:10.1001/jamapsychiatry.2017.2509
Are psychological and/or educational preventive interventions for anxiety effective in varied populations?
This systematic review and meta-analysis of 29 randomized clinical trials (36 comparisons) including 10 430 patients from 11 countries on 4 continents showed a small, but statistically significant, effect size of psychological and/or educational preventive interventions for anxiety. Sensitivity analyses and adjustment for publication bias confirmed that the results were robust.
Although the benefit of psychological and/or educational interventions in the prevention of anxiety is modest, the results suggest that psychological and/or educational preventive interventions for anxiety should be further developed and implemented.
To our knowledge, no systematic reviews or meta-analyses have been conducted to assess the effectiveness of preventive psychological and/or educational interventions for anxiety in varied populations.
To evaluate the effectiveness of preventive psychological and/or educational interventions for anxiety in varied population types.
A systematic review and meta-analysis was conducted based on literature searches of MEDLINE, PsycINFO, Web of Science, EMBASE, OpenGrey, Cochrane Central Register of Controlled Trials, and other sources from inception to March 7, 2017.
A search was performed of randomized clinical trials assessing the effectiveness of preventive psychological and/or educational interventions for anxiety in varying populations free of anxiety at baseline as measured using validated instruments. There was no setting or language restriction. Eligibility criteria assessment was conducted by 2 of us.
Data Extraction and Synthesis
Data extraction and assessment of risk of bias (Cochrane Collaboration’s tool) were performed by 2 of us. Pooled standardized mean differences (SMDs) were calculated using random-effect models. Heterogeneity was explored by random-effects meta-regression.
Main Outcomes and Measures
Incidence of new cases of anxiety disorders or reduction of anxiety symptoms as measured by validated instruments.
Of the 3273 abstracts reviewed, 131 were selected for full-text review, and 29 met the inclusion criteria, representing 10 430 patients from 11 countries on 4 continents. Meta-analysis calculations were based on 36 comparisons. The pooled SMD was −0.31 (95% CI, −0.40 to −0.21; P < .001) and heterogeneity was substantial (I2 = 61.1%; 95% CI, 44% to 73%). There was evidence of publication bias, but the effect size barely varied after adjustment (SMD, −0.27; 95% CI, −0.37 to −0.17; P < .001). Sensitivity analyses confirmed the robustness of effect size results. A meta-regression including 5 variables explained 99.6% of between-study variability, revealing an association between higher SMD, waiting list (comparator) (β = −0.33 [95% CI, −0.55 to −0.11]; P = .005) and a lower sample size (lg) (β = 0.15 [95% CI, 0.06 to 0.23]; P = .001). No association was observed with risk of bias, family physician providing intervention, and use of standardized interviews as outcomes.
Conclusions and Relevance
Psychological and/or educational interventions had a small but statistically significant benefit for anxiety prevention in all populations evaluated. Although more studies with larger samples and active comparators are needed, these findings suggest that anxiety prevention programs should be further developed and implemented.
The annual prevalence of anxiety disorders is 6.7% in the general population.1 The burden of disease in terms of years lived with disability attributable to anxiety disorders increased by 14.8% (relative increase) between 2005 and 2015, ranking ninth in the world and eighth in high-income countries.2 Among mental and substance use disorders, anxiety ranked second in the world.3 Although there are effective treatments for anxiety disorders,4-6 not all persons with anxiety receive the appropriate treatment,7 and cost-effectiveness studies suggest that treatment alone is not sufficient to eliminate the disease burden attributable to anxiety disorders.8 An additional way to reduce the burden of anxiety disorders is to lower the incidence of new cases, which can be achieved through prevention rather than treatment.
Previous systematic reviews or meta-analyses of the prevention of anxiety have been mainly undertaken in children and/or adolescents.9-13 In adults, studies have traditionally focused on specific anxiety disorders, such as posttraumatic stress disorder.14,15 One meta-analysis16 assessed the prevention of anxiety in the general population, although it was centered on specific cognitive-behavioral interventions. To the best of our knowledge, no systematic review or meta-analysis has been performed on the effectiveness of psychological and/or educational interventions in preventing anxiety in several types of populations. The aim of the present systematic review and meta-analysis was to evaluate the effectiveness of psychological and/or educational interventions in preventing anxiety in varied populations.
We followed PRISMA guidelines for reporting systematic reviews and meta-analyses.17 The protocol of this systematic review was previously registered at the International Prospective Register of Systematic Reviews. The Institute of Biomedical Research in Málaga, Málaga, Spain, approved the study.
We systematically searched 6 electronic databases, including PubMed, PsycINFO, EMBASE, Web of Science, OpenGrey (System for Information on Grey Literature in Europe), and CENTRAL (Cochrane Central Register of Controlled Trials) from inception to March 7, 2017. No date or language restrictions were imposed. This search strategy was complemented with hand searching of reference lists in articles and other reviews on this topic. In addition, experts in the field were contacted and asked to complete the list of selected publications. Databases were searched separately by 2 of us (P.M.-P. and S.C.-C.). The specific search strategies used are described in the eAppendix in the Supplement.
We selected randomized clinical trials (RCTs) because they are the standard for clinical trials.18 We focused on educational and/or psychological interventions. The former simply provide information about anxiety through lectures or fact sheets, whereas psychological interventions attempt to change how people think by using a variety of strategies (eg, cognitive behavioral or interpersonal therapy). Randomized clinical trials based on medication or physical interventions (eg, sports) were excluded. The comparators allowed were care-as-usual, no intervention, a waiting list for intervention, or attention control. To separate the effectiveness of prevention from that of therapies, baseline anxiety was required to have been discarded through standardized interviews (eg, Structured Clinical Interview for DSM Disorders), validated self-reports with standard cutoff points (eg, Beck Anxiety Inventory-II), or diagnosis by a mental health specialist. Outcomes included the incidence of new cases of any DSM-IV anxiety disorder and/or the reduction of anxiety symptoms. Outcomes were required to have been measured by standardized interviews or validated symptom scales. Posttraumatic stress disorder was excluded because, in this case, it is difficult to separate treatment from prevention.19 Participants could have any demographic characteristic (eg, age, sex), and all settings and languages were considered.
Titles and abstracts were reviewed independently by 2 of us (P.M.-P. and S.C.-C.), who used the abovementioned criteria to determine study eligibility. The full text of potentially relevant studies was reviewed for final inclusion. All discrepancies were resolved by consensus with 1 of us (M.R.-V. or A.F.). The degree of agreement between the initial reviewers was good (Cohen κ = 0.74; 95% CI, 0.60-0.89).20
Data extracted from each study were recorded in an evidence table. Two of us (P.M.-P. and S.C.-C.) collected data from primary studies. Any discrepancies were solved by consensus between the reviewers. When necessary information was not reported in the study, the authors of the original article were contacted for further details.
We used the Cochrane Collaboration risk of bias tool to assess the quality of the studies included.21 All of the studies included were assessed, and any disagreements were resolved by 2 of us (P.M.-P. and S.C.-C.). The level of agreement was good (intraclass correlation coefficient, 0.83; 95% CI, 0.77-0.87).22
When the outcome was differences in anxiety symptoms between intervention and control groups, means (SDs) for each arm were extracted. We then calculated standardized mean differences (SMDs) for each RCT by estimating the mean SMD at different follow-up times. The pooled SMDs for all RCTs and their 95% CIs were estimated. Negative SMDs represented an improvement in the reduction of anxiety symptoms in the intervention group. If only new cases of anxiety were reported (incidence of anxiety), CMA, version 3.0 (Comprehensive Meta-analysis Software Biostat Inc) was used to obtain the equivalent SMDs. Cohen23 suggested that an SMD of 0.2 is indicating a low effect; 0.5, a moderate effect; and 0.8, a large effect. We inflated the SEs of the nested comparisons in the same RCT following the suggestions of Cates.24 We selected the random-effects model for the study under the assumption that the studies included in the meta-analysis were performed in a variety of populations that may differ from each other.21
Statistical heterogeneity was evaluated using the I2 statistic, in which a value of 0% to 40% might indicate no important heterogeneity; 30% to 60%, moderate; 50% to 90%, substantial; and 75% to 100%, considerable.21 In addition, we calculated the Q statistic and its P value.
To detect publication bias, a funnel plot was examined by visual inspection and the Duval and Tweedie25 trim-and-fill procedure, which is a test of symmetry of the funnel plot. This procedure yields an adjusted pooled effect size after accounting for missing studies due to publication bias. We also performed Begg and Mazumdar rank correlation26 and the Egger test.27
Since the SMD could differ at varying follow-up times, we also conducted sensitivity analyses at first and last follow-ups. Sensitivity analyses also included fixed effects and Hedges g and excluded some RCTs from analysis (those that caused the greatest increase in heterogeneity and those that measured anxiety as a secondary outcome).
We used a mixed-effects model for subgroup analyses according to the type of prevention, type of outcome measure (symptoms scale vs standardized diagnostic interview), type of anxiety, country, population age, setting, comparator, professional delivering the intervention, intervention orientation, type of intervention, intervention format, number of sessions, follow-up, sample size, and risk of bias.
Meta-regression was performed to explain the between-trial heterogeneity observed. We verified normality of the quantitative variables that were included in the meta-regression by the skewness-kurtosis normality test,28 performing the pertinent transformations to get approximation to normality when it was necessary. We forced 2 quantitative variables—risk of bias and sample size—in the meta-regression models for adjustment. The former is related to the quality of the RCTs and the latter to publication bias. Of the remaining covariables considered for subgroup analysis, only 1 covariable was introduced in each new model. The final model was composed of the 2 forced quantitative variables and dummy covariables with a significance level of P < .15 that were not removed from the model due to collinearity. We used the Knapp and Hartung method29 to estimate SE and 95% CIs. We also used a Higgins and Thompson30 permutation test approach to calculate P values, taking into account multiplicity adjustment (Monte Carlo approach; 20 000 permutations). We used CMA, version 3.0 and Stata, version 14.2 (StataCorp) to perform analyses.
A total of 3273 abstracts were reviewed. Of these, 131 articles were included for full-text review, and 29 RCTs met the inclusion criteria of the meta-analysis (Figure 1).31-60
The characteristics of the 29 RCTs included are described in eTable 1 in the Supplement. Twenty studies were published in or after 2010, and only 3 RCTs were published before 2005. Seven RCTs were conducted in the United States.
A total of 10 430 patients were enrolled. Sample sizes ranged from 24 to 2998 (median, 165). A total of 9 RCTs included adults (from 18 to 65 years), 9 included children or adolescents, 4 were performed in older adults, 6 in adults and elderly individuals, and 1 in both adults and children. Settings included school or university in 12 RCTs.
Interventions were based on the principle of cognitive behavioral therapy in 25 RCTs, whereas 4 RCTs were based on other types of interventions (2 psychoeducational, 1 acceptance and commitment therapy, and 1 biopsychosocial). Interventions were delivered in individual format in 14 RCTs. Four RCTs included interventions with a guided self-help format (computerized). The number of sessions ranged from 1 to 12 (median, 8). The comparator used was care-as-usual in 13 RCTs. Interventions were conducted by a mental health specialist in 13 RCTs.
Follow-up periods ranged from 7 weeks to 60 months (median, 12 months). The duration of follow-up exceeded 12 months in 8 RCTs.
Indicated, selective, and universal prevention were evaluated in 11, 10, and 8 RCTs, respectively. With respect to outcomes, 10 RCTs measured the reduction in anxiety symptoms and 10 determined the incidence of anxiety disorders; 9 RCTs measured both reduction of symptoms and incidence.
The risk of bias for each study is reported in eTable 2 in the Supplement. Eight RCTs had a low risk (≤4 points), 9 had a moderate risk (5-6 points), and 12 had a high risk (≥7 points) of bias.
Meta-analysis calculations were based on 36 comparisons performed in 29 RCTs (eTable 3 in the Supplement). The pooled SMD was −0.31 (95% CI, −0.40 to −0.21; P < .001), and the equivalent pooled odds ratio (OR) was 0.57 (95% CI, 0.48 to 0.68; P < .001)—a 43% reduction in the incidence of anxiety—with substantial heterogeneity (I2 = 61.1%; Q35 = 90.13; P < .001). This finding means that psychological and/or educational preventive interventions for anxiety had a small and statistically significant effect on anxiety prevention. Figure 2 shows the forest plot for the overall and individual effect sizes.
Results of the Egger (intercept, −1.30; 95% CI, −2.25 to −0.34; P = .01) and Begg and Mazumdar (z = −2.40; P = .02) tests indicated the presence of publication bias. The SMDs adjusted for publication bias according to the Duval and Tweedie trim-and-fill procedure barely decreased (SMD, −0.27; 95% CI, −0.37 to −0.17; P < .001). The funnel plot is shown in eFigure 1 in the Supplement.
Sensitivity analyses are reported in Table 1. The pooled SMDs changed little with the first or last evaluation, the fixed-effect model, or the Hedges g test when the RCT 41 that most increased heterogeneity or the RCTs in which anxiety was a secondary outcome33,41,43 were excluded.
Subgroup analyses revealed differences in the effectiveness of psychological and/or educational interventions depending on the outcome measure, comparator, provider, sample size, risk of bias, and follow-up (eTable 4 in the Supplement).
Meta-regression is reported in Table 2. The skewness-kurtosis normality test was not statistically significant for logarithmic transformation of the sample size (χ22 = 4.11; P = .13) and risk of bias (χ22 = 1.99; P = .37) variables. As many as 85.3% of total variance was attributable to within-study variability, and the remaining 14.7% was attributable to between-study variability. In total, 99.6% of between-study variance was explained by the 5 variables included in the meta-regression model (F5,30 = 9.31; P < .001). There was a statistically significant association between higher SMD when the comparator was waiting list (comparator) (β = −0.33 [95% CI, −0.55 to −0.11]; P = .005) and a lower sample size (lg) (β = 0.15 [95% CI, 0.06 to 0.23]; P = .001). There was no association between SMD and risk of bias, type of outcome measure (standardized interview), and family physician (caregiver). Analysis of residuals showed a good fit of the meta-regression model (eFigure 2 in the Supplement).
We found that psychological and/or educational interventions are effective in the prevention of anxiety. The overall effect size was small but statistically significant. This result was derived from 29 RCTs (36 comparisons) that included 10 430 patients from 11 countries on 4 continents. Sensitivity analyses and adjustment for publication bias demonstrated that the overall effect size was robust. Heterogeneity was substantial and explained by a model of meta-regression including 5 variables, 2 of which had a statistically significant association with effect size (waiting list and sample size), whereas the other 3 variables did not (risk of bias, family physician as caregiver, and standardized interview as outcome).
To the best of our knowledge, this is the first systematic review and meta-analysis to examine the effectiveness of psychological and/or educational interventions in the prevention of anxiety in varied types of populations. Our meta-analysis included a reasonable number of RCTs representing a large population of individuals with different characteristics and from diverse settings. In addition, this study involved a wide spectrum of interventions (any psychological and/or educational intervention) for most types of anxiety disorders (except posttraumatic stress disorder) and implemented by a variety of professionals in different settings. These aspects give the study a wide scope, which supports its external validity. We used multiple complementary electronic databases with supplementary hand searching. Thus, the variety of databases utilized, combined with the broad range of search terms, contributed to a highly sensitive search. In addition, the strict inclusion criteria, analyzing only RCTs with a study population free of anxiety at baseline, allowed us to distinguish prevention effectiveness from treatment effectiveness. Study selection, data extraction, and risk of bias assessment were performed by trained and independent reviewers, with good interobserver reliability. We performed sensitivity analyses and adjustment for publication bias, which support the robustness of the pooled SMDs. Finally, the meta-regression model explained heterogeneity and enabled adjustment for confounding biases and multiple comparisons.
Our meta-analysis has several limitations. First, although most of the studies included had low to moderate risk of bias (17 RCTs), 12 had a high risk of bias. Subgroup analysis showed that the studies with a lower risk of bias had a tendency to report a smaller effect size (SMD, −0.15; 95% CI, −0.23 to −0.07) (eTable 4 in the Supplement); nevertheless, after adjusting for confounding biases, the meta-regression model showed that the risk of bias was not significant. Second, the duration of follow-up exceeded 12 months in only 8 RCTs, which tended to report a lower effect size (SMD, 0.15; 95% CI, 0.01-0.29). Although follow-up duration was not significant when adjustment for sample size and risk of bias was performed, firm conclusions about long-term effectiveness cannot be drawn from our study. Third, the interventions implemented in most of the RCTs included in the meta-analysis were aimed at preventing more than 1 specific anxiety disorder or used nonspecific anxiety symptoms as outcomes; therefore, no inferences can be made about any specific anxiety disorder. Fourth, reduction of anxiety symptoms (measured by scales) was the only outcome of 10 RCTs; although the reliability and validity of scales are widely accepted, standardized diagnostic interviews generally have greater validity. Nevertheless, the reduction of anxiety symptoms is also useful as an outcome because it has a positive and relevant effect on quality of life and cost.41,61 Another related aspect is that RCTs in which standardized diagnostic interviews were conducted tended to report a lower effect size (SMD, −0.18; 95% CI, −0.30 to −0.06) (eTable 4 in the Supplement). Again, adjustment for confounding biases and multiple comparisons in meta-regression discarded any statistical significance. Fifth, evidence of publication bias was found, which means that it is likely that some RCTs with nonsignificant results have not been published. Nevertheless, this limitation does not seem relevant because the effect size scarcely decreased after adjustment for publication bias. Sixth, we cannot establish the superiority of one intervention over another (eg, cognitive behavioral therapy vs acceptance and commitment therapy) because it was not within the scope of the study and we excluded the few RCTs in which these types of comparisons were made. Seventh, it is probable that the effect size of the anxiety prevention that we obtained was smaller since it was increased by the use of waiting list comparator and small samples. Finally, in some categories in specific subgroup analysis, the number of RCTs or comparisons was low; in these cases (eg, type of anxiety), the lack of statistical power prevents firm conclusions.
The overall effect size obtained was small. Other meta-analyses on anxiety prevention report similar SMDs (range, 0.13-0.32),11-13,16 and the same occurs with meta-analyses on the prevention of depression.62-64
We found statistically significant associations only for sample size and waiting list as comparators when adjustment for confounding biases and multiplicity testing were performed in meta-regression. A recent meta-analysis on the effectiveness of cognitive behavioral therapy for anxiety disorders and major depression showed that responses were large when the control condition was waiting list but small to moderate when it was care-as-usual or pill placebo.65 Similar results were found in a network meta-analysis on the effectiveness of treatments for depression. The authors suggest that waiting list can be a “nocebo” and may introduce negative psychological expectations in the sense of waiting for the desired active treatment, whereas patients allocated to nonintervention or usual care may more actively seek other treatments, either by themselves or by others, for their ailment.66 In a recent meta-analysis, subgroup analysis of the prevention of anxiety in young people revealed that the studies in which waiting list was used as a comparator tended to report a significantly higher effect size; nevertheless, after adjustment for confounding factors in meta-regression, this tendency disappeared.11 Zalta16 found no statistically significant differences between active and nonactive comparators in a meta-analysis, although waiting list was categorized within the nonactive category and no adjustment for confounding factors was made.
Regarding sample size, effect sizes were smaller in the RCTs with larger sample sizes after adjustment for confounding biases and multiple comparisons were performed. This effect could be due to publication bias, as evidenced in our meta-analysis, although its influence on the overall effect size was minimal. Studies with small samples are more likely to obtain negative results and not be published; however, when they yield statistically significant results, their effect sizes are higher.
Effect size tended to decrease when the outcomes were assessed using standardized diagnostic interviews, although no statistically significant associations were found in meta-regression. When RCTs had small samples, they were more likely to determine statistically significant effects on anxiety prevention if symptom scales (difference of means) were utilized compared with structured interviews (comparison of proportions). Furthermore, there was a negative correlation between risk of bias and standardized interview as outcomes (ρ = −0.27). Therefore, the loss of statistical significance between effect size and standardized interview can be explained in part by adjustment for confounding factors with the introduction of the variables sample size and risk of bias in the meta-regression model. Adjustment for multiple comparisons also contributed to the loss of significance.
Regarding the professional who delivered the intervention (caregiver), effect sizes tended to increase when the caregiver was a family physician, but this effect was attenuated after adjustments in meta-regression. The SMDs for family physician and mental health specialist were similar, but the latter was removed from meta-regression because of collinearity. Nevertheless, interventions were led by a family physician in only 2 RCTs and results should therefore be interpreted with caution.
Subgroup analysis revealed statistically significant associations between low risk of bias with a lower effect size and follow-up time (SMDs higher in RCTs <6 months). However, this effect disappeared when adjustment for confounding bias was performed. No relevant differences were observed when the first (SMD, 0.30) and last (SMD, 0.33) evaluations were included as outcomes for sensitivity analysis. Other meta-analyses assessing studies on anxiety and depression11,16,62,63 report that effect size seems to diminish over time. Yet, a recent meta-analysis of studies on the prevention of depression found no association between effect size and follow-up time.64 Regarding the risk of bias, a previous meta-analysis did not find any statistically significant association between risk of bias and effect size.13
A tendency was observed for SMDs to be higher in selective prevention, followed by indicated and universal prevention, although differences were not statistically significant even when adjustment for confounding bias was not performed. Some previous meta-analyses have not found any association between type of prevention and effect size11,12,16; however, 1 meta-analysis found that selective and indicated interventions had a greater effect size than universal prevention,13 whereas another meta-analysis reported that universal prevention was more effective.10
As in the rest of the meta-analyses concerning the prevention of anxiety,10-13,16 most of the interventions included in our study had a cognitive-behavioral orientation (31 comparisons; SMD, 0.25); the remaining 6 comparisons had other orientations and tended to have a greater response (SMD, 0.52), although it was not statistically significant. Therefore, no conclusions can be drawn about orientation.
The pooled effect size obtained for prevention of anxiety is modest compared with the sizes observed in treatments for anxiety disorders.65 Although the preventive fraction that derived from the pooled effect size that we obtained (OR, 0.57)— 43% reduction in the incidence of anxiety— was higher than those found in other meta-analyses,11-13,62-64 future studies should strive to further develop and test new prevention interventions with greater effect sizes. Yet, from the perspective of public health, small effects on prevention could have a high impact, thereby improving quality of life and reducing costs. By way of illustration, prevention programs reaching a large population through extensive school31 and primary care programs33 or information and communication technologies can have a large cumulative effect.67 From this point of view, our results suggest that programs for the prevention of anxiety should be implemented. Nevertheless, there are some aspects that are not well understood and should be further investigated. There are few studies that assess the cost-effectiveness and cost-utility of programs for the prevention of anxiety.61,68 In addition, long-term RCTs with larger samples and low risk of bias comparing different programs and strategies for the prevention of anxiety are needed.
Accepted for Publication: June 29, 2017.
Corresponding Author: Patricia Moreno-Peral, PhD, Institute of Biomedical Research in Málaga, C/Sevilla 23, 3ª Planta, 29009 Málaga, Spain (email@example.com).
Published Online: September 6, 2017. doi:10.1001/jamapsychiatry.2017.2509
Author Contributions: Drs Moreno-Peral and Bellón had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Moreno-Peral, Conejo-Cerón, Rubio-Valera, Fernández, Luna, Bellón.
Acquisition, analysis, or interpretation of data: Moreno-Peral, Conejo-Cerón, Rubio-Valera, Fernández, Navas-Campaña, Rodríguez-Morejón, Motrico, Rigabert, Martín-Pérez, Rodríguez-Bayón, Ballesta-Rodríguez, Luciano, Bellón.
Drafting of the manuscript: Moreno-Peral, Bellón.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Moreno-Peral, Conejo-Cerón, Luna, Bellón.
Obtained funding: Bellón.
Administrative, technical, or material support: Moreno-Peral.
Study supervision: Rubio-Valera, Fernández, Luna, Bellón.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by the Spanish Ministry of Health, the Institute of Health Carlos III, and European Regional Development Fund Una manera de hacer Europa grant PI12/02755 and the Andalusian Council of Health grant 0583/2012 as well as the Prevention and Health Promotion Research Network (redIAPP) RD16/0007, PRISMA group RD16/0007/0012, and SAMSERAP group RD16/0007/0010. Dr Luciano has a Miguel Servet research contract from the Institute of Health Carlos III (CP14/00087).
Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: The Institute of Health Carlos III, the European Regional Development Fund, the Andalusian Council of Health, the Institute of Biomedical Research in Málaga, and the Prevention and redIAPP provided logistical support.
Create a personal account or sign in to: